Improving linkage into HIV care among adults in Blantyre, Malawi

Thesis submitted in accordance with the requirements of the University of Liverpool for the degree of Doctor in Philosophy by

Peter MacPherson

March 2013

2 Preface | My role

I, Peter MacPherson, confirm that the work presented in this thesis is my own. Where information has been derived from other sources, I confirm that this has been indicated in the thesis.

My Role

I conducted the background literature review and systematic literature review of studies reporting on the linkage into HIV care. Chigomezgo Munthali and I conducted the systematic review of the diagnostic accuracy of the WHO clinical staging system; I performed the meta-analysis. I designed the study protocols and questionnaires for the prospective cohort study, qualitative study, cluster randomised trial and diagnostic accuracy studies with input from Liz Corbett, David Lalloo, Bertie Squire and Gillian

Mann. I wrote all study standard operating procedures, except for the standard operating procedures, for laboratory measurement of CD4 count, which were written by

Aaron Mdolo. I recruited and trained the Research Assistants and Study Nurses who conducted the prospective cohort study, qualitative study and the home initiation of HIV care and ascertainment of facility outcome components of the cluster randomised trial.

Liz Corbett’s HitTB study team recruited and trained Field Workers who performed mapping, household enumeration and the demographic census for the cluster- randomised trial. Community Counsellors who provided home-based HIV self-testing

(HIVST), recorded uptake of HIVST and reporting of HIVST results were recruited, trained and supervised by the HitTB study team. Aaron Mdolo and his laboratory team performed laboratory analysis of CD4 counts. Data management was performed by myself and the HitTB study team, supervised by Augustine Choko and Deus Thindwa.

Data cleaning was performed by myself with support from the HitTB study team. I wrote the statistical analysis plan for the cluster-randomised trial, with input from Liz Corbett and Emily Webb. I analysed the data for the prospective cohort study. Emily Webb

3 Preface | My role wrote do-files for analysis of the baseline census for the HitTB study and I modified these to analyse the baseline census for the home initiation of HIV care trial. I wrote programmes to analyse the data for the endpoint analysis of the cluster randomised trial, with support from Emily Webb and Richard Hayes. The interview guides for the qualitative study were designed by me, with input from Liz Corbett, Nic Desmond,

Eleanor MacPherson and Daniel Mwale. Daniel Mwale conducted qualitative interviews.

I designed the coding framework and socio-ecological model. I analysed qualitative data, with input from Liz Corbett, Eleanor MacPherson and Daniel Mwale.

4 Preface | Table of contents

Table of Contents

My Role ...... 3 Table of Contents ...... 5 List of Tables ...... 12 List of Figures ...... 14 List of Acronyms ...... 16 List of Supporting Manuscripts ...... 18 Abstract ...... 20 Acknowledgements ...... 23 1. Introduction ...... 25 1.1. Background ...... 25

1.2. Project starting points ...... 27

1.3. Aims and objectives ...... 28

1.4. Outline of thesis ...... 29

2. Literature review ...... 31 2.1. Human Immunodeficiency Virus Infection ...... 31 2.1.1. Origins of the HIV pandemic ...... 31 2.1.2. Modes of transmission of HIV infection and prevention ...... 33 2.1.3. Risk factors for HIV transmission and prevention strategies ...... 35 2.1.4. Pathogenesis and natural history of HIV infection ...... 41 2.1.5. Global and sub-Saharan African of HIV ...... 45 2.1.6. HIV epidemiology in Malawi ...... 47

2.2. HIV testing and Counselling ...... 50 2.2.1. Objectives of HIV testing and counseling ...... 50 2.2.2. HIV testing technologies ...... 52 2.2.3. Uptake of HIV testing and counselling ...... 53 2.2.4. Models of HIV testing and counselling ...... 55 2.2.5. Facility-based approaches to HTC ...... 56 2.2.6. Community-based approaches to HTC ...... 62 2.2.7. HIV self-testing ...... 65

5 Preface | Table of contents

2.2.8. Which models of HTC are recommended? ...... 71 2.2.9. HIV testing in Malawi ...... 72

2.3. Treatment of HIV infection ...... 74 2.3.1. Antiretroviral therapy for treatment of HIV infection ...... 74

2.4. Assessment of eligibility for antiretroviral therapy ...... 80 2.4.1. When to start antiretroviral therapy? ...... 80 2.4.2. Measurement of CD4 count to determine ART eligibility ...... 82 2.4.3. Clinical staging assessments for ART eligibility ...... 83 2.4.4. Alternative approaches to assessing ART eligibility ...... 90

2.5. The public health approach to scale-up of ART delivery in low income countries with generalised HIV epidemics ...... 93

2.6. The National ART programme in Malawi ...... 100

2.7. Linkage between HIV diagnosis and initiation of ART ...... 104 2.7.1. The HIV care cascade ...... 104 2.7.2. Defining linkage into care ...... 107 2.7.3. Linkage and retention between HIV diagnosis and initiation of ART ...... 111 2.7.4. Summary estimates for pre-treatment patient loss to care ...... 119 2.7.5. Reasons for pre-treatment loss to follow-up ...... 123 2.7.6. Interventions to improve linkage and retention between HIV diagnosis and initiation of ART ...... 127

2.8. Development of a conceptual framework to understand pre-treatment loss to care in the HIV care cascade ...... 132 2.8.1. Summary of evidence regarding magnitude, reasons for, consequences and potential areas for intervention to improve pre-treatment loss to care .. 134

2.9. Measurement of household poverty status ...... 136 2.9.1. Poverty, illness and health ...... 136 2.9.2. Measuring poverty in resource limited settings – the proxy means test method ...... 137

2.10. Malawi ...... 140 2.10.1. Geopolitical and socioeconomic indicators ...... 140

3. Methods ...... 143 3.1. Study location and population ...... 143

6 Preface | Table of contents

3.1.1. Health facilities available to study population ...... 145

3.2. Delineation of study clusters and enumeration of population ...... 146 3.2.1. Background to trial structure ...... 146 3.2.2. Demarcation and enumeration of study clusters ...... 147

3.3. Cluster population census ...... 150

3.4. Classification of cluster residency status of facility attendees (“map-book” system) ...... 151 3.4.1. Background and need for cluster categorisation system ...... 151 3.4.2. Cluster categorisation method 1 (extraction of address locations from routinely captured data in clinic registers) ...... 152 3.4.3. Cluster categorisation method 2 (“map book” system) ...... 152 3.4.4. Validation of map book system ...... 154 3.4.5. Results: Cluster categorisation using description of physical location of residence ...... 155 3.4.6. Results: Cluster categorisation using the map book system ...... 156 3.4.7. Summary ...... 157

3.5. HIV testing and counselling (HTC) ...... 157 3.5.1. HTC in primary health care centres ...... 157 3.5.2. HIV self-testing through community counsellors ...... 158

3.6. Key informant system for recording cluster resident deaths ...... 160

3.7. Laboratory methods ...... 162 3.7.1. CD4 count measurement ...... 162

3.8. Measurement of poverty status ...... 162

3.9. Ethical issues in studies ...... 163

4. Uptake of HIV testing and counselling and determinants of pre-treatment loss to follow-up among primary clinic attendees: a prospective cohort study ...... 169 4.1. Introduction ...... 169

4.2. Methods ...... 170 4.2.1. Study design ...... 170 4.2.2. Study sites ...... 171 4.2.3. HIV care pathway in Malawian National Programme ...... 171 4.2.4. Participant recruitment and assessment ...... 172

7 Preface | Table of contents

4.2.5. Ascertainment of outcomes on the HIV care pathway during follow-up .. 174 4.2.6. Laboratory methods ...... 176 4.2.7. Sample size calculations ...... 176 4.2.8. Data analysis ...... 177 4.2.9. Ethical Approval ...... 178

4.3. Results ...... 178 4.3.1. Clinic attendances during study period ...... 178 4.3.2. HTC during clinic attendances ...... 179 4.3.3. HIV testing outcomes ...... 180 4.3.4. Baseline characteristics of cohort participants ...... 180 4.3.5. Awareness and readiness for ART ...... 183 4.3.6. Factors associated with same-day referral for ART ...... 183 4.3.7. Evaluation of patient flow through HIV care pathway ...... 185 4.3.8. Initiation of ART during cohort follow-up ...... 188 4.3.9. Factors associated with ART initiation ...... 190 4.3.10. Retention in pre-ART care for participants who did not initiate ART during follow-up ...... 192 4.3.11. Cohort mortality and factors associated with death ...... 192

4.4. Summary of findings ...... 193

4.5. Limitations of Study ...... 195

4.6. Findings related to other studies ...... 197

4.7. Implications for future studies ...... 198

4.8. Conclusions ...... 200

5. Understanding reasons for loss to HIV care at primary health care level: a qualitative study with patients and health workers ...... 201 5.1. Introduction ...... 201

5.2. Methods ...... 201 5.2.1. Study design ...... 201 5.2.2. Rationale for qualitative study ...... 202 5.2.3. Qualitative methods ...... 202 5.2.4. Study participants ...... 203 5.2.5. Ethical considerations ...... 207

8 Preface | Table of contents

5.3. Results ...... 207 5.3.1. Baseline Characteristics ...... 207 5.3.2. Conceptualising determinants of patient loss to care and delay ...... 208 5.3.3. Linkage to ART: 1) HIV testing and counselling ...... 208 5.3.4. Linkage to ART: 2) Eligibility assessments ...... 214 5.3.5. Linkage to ART: 3) Pre-ART care ...... 216

5.4. Summary of findings ...... 221

5.5. Limitations ...... 223

5.6. Finding in relation to other studies ...... 227

5.7. Implications for future studies ...... 230

5.8. Conclusions ...... 232

6. Home initiation of HIV care following self-testing: a cluster randomised trial in urban Blantyre ...... 235 6.1. Introduction ...... 235

6.2. Methods ...... 236 6.2.1. Study design ...... 236 6.2.2. Participants and study setting ...... 236 6.2.3. Interventions ...... 237 6.2.4. Sample Size ...... 239 6.2.5. Randomisation and masking ...... 240 6.2.6. Statistical analysis ...... 241 6.2.7. Ethical approval ...... 242

6.3. Results ...... 242 6.3.1. Baseline characteristics of clusters ...... 242

6.4. Primary outcome ...... 246

6.5. Secondary outcomes ...... 248

6.6. Additional analyses ...... 250 6.6.1. Estimated impact of interventions on ART coverage among PLHIV and ART- eligible residents ...... 252 6.6.2. Loss from ART over six months ...... 254 6.6.3. Adherence to ART ...... 258 6.6.4. Mortality in residents aged 16 to 49 years ...... 258

9 Preface | Table of contents

6.7. Summary of findings ...... 259

6.8. Limitations ...... 263

6.9. Findings relating to other studies ...... 265

6.10. Implications for future studies ...... 267

6.11. Conclusions ...... 268

7. Evaluation of a novel community health workers ART eligibility assessment tool for use in community settings in Blantyre, Malawi ...... 269 7.1. Introduction ...... 269

7.2. Methods ...... 270 7.2.1. Type of study ...... 270 7.2.2. Study site and population ...... 270 7.2.3. Study procedures ...... 271 7.2.4. WHO clinical staging assessment and measurement of CD4 count ...... 274 7.2.5. Statistical methods ...... 275 7.2.6. Ethical considerations ...... 276

7.3. Results ...... 276 7.3.1. Baseline Characteristics ...... 276 7.3.2. Outcomes of ART eligibility assessments ...... 280 7.3.3. Performance of CHW tool in identifying CD4<350 cells/mm3 ...... 280 7.3.4. Characteristics associated with performance of community health workers screening tool ...... 282 7.3.5. Secondary analysis ...... 284

7.4. Summary of findings ...... 286

7.5. Limitations ...... 287

7.6. Findings relating to other studies ...... 291

7.7. Implications for future studies ...... 291

7.8. Conclusions ...... 292

8. Summary, recommendations and conclusions ...... 295 8.1. Summary of rationale, objectives and key findings ...... 295

8.2. Recommendations and future directions ...... 297 8.2.1. Achieving universal HIV testing and counselling ...... 297

10 Preface | Table of contents

8.2.2. Achieving universal HIV treatment ...... 302

8.3. Conclusion ...... 306

References ...... 307 Appendix 1: Research ethics committee approvals for cohort study and qualitative study ...... 341 Appendix 2: Research ethics committee approval for cluster-randomised trial ...... 343

11 Preface | List of tables

List of Tables

Table 2.1: WHO-recommended application of HTC models in different epidemiological situations ...... 72 Table 2.2: Drugs approved by US Food and Drug Administration (and currently marketed) for the treatment of HIV, 2013 ...... 75 Table 2.3: WHO Clinical Staging System (2007) for adults with confirmed HIV infection 85 Table 2.4: Summary estimates and meta-analysis of studies comparing WHO clinical stage 3/4 with CD4 ≤200 cells/mm3 ...... 87 Table 2.5: Summary estimates and meta-analysis of studies comparing WHO clinical stage 3/4 with CD4≤350 cells/mm3 ...... 88 Table 2.6: Ten guiding principles of the WHO "3 by 5" initiative ...... 95 Table 2.7: Studies reporting on pre-treatment loss to care in the HIV care cascade in sub- Saharan Africa, 2006-2011 ...... 112 Table 2.8: Systematic review and meta-analysis of studies reporting on loss throughout the HIV care cascade ...... 120 Table 2.9: Studies reporting risk of death in individuals with pre-treatment loss to follow- up ...... 124 Table 2.10: Evaluated interventions to improve linkage into HIV care ...... 127 Table 2.11: Urban model proxy means test for poverty in Malawi ...... 140 Table 4.1: Sample size estimation and 95% confidence intervals for ART initiation ...... 177 Table 4.2: Clinic attendances by department and sex during study recruitment period 179 Table 4.3: Baseline characteristics of cohort study participants ...... 182 Table 4.4: Univariate and multivariate associations with direct referral to initiate ART 184 Table 4.5: Univariate and multivariate associations with ART initiation ...... 191 Table 4.6: Univariate and multivariate associations with death ...... 193 Table 5.1: Characteristics of participants recruited to qualitative study ...... 207 Table 6.1: Baseline cluster, household and individual characteristics ...... 245 Table 6.2: Primary and secondary trial endpoints ...... 247 Table 6.3: Comparison of home and facility ART initiators by trial arm ...... 251 Table 6.4: Individual-level risk factors for loss to retention from ART ...... 257 Table 6.5: Primary cause of death in 44 cluster residents aged 16-49 during 6 months of interventions ...... 259 Table 7.1: Baseline characteristics of study participants ...... 279 Table 7.2: Accuracy of novel community health worker ART eligibility screening tool for identifying CD4<350 cells/mm3 ...... 281

12 Preface | List of tables

Table 7.3: Regression model fit to participant data for community health worker tool and WHO clinical staging assessment using sensitivity as dependent variable ...... 283 Table 7.4: Regression model fit to participant data for community health worker tool and WHO clinical staging assessment using specificity as dependent variable ...... 284 Table 7.5: Accuracy of novel community health worker ART eligibility assessment tool for identifying CD4 ≤500 cells/mm3 ...... 286

13 Preface | List of figures

List of Figures

Figure 2.2: Structure of HIV ...... 42 Figure 2.3: HIV viral load and CD4+ count during course of untreated infection ...... 43 Figure 2.4: Prevalence of HIV infection, 2011 ...... 46 Figure 2.5: HIV prevalence by sex and age group, Malawi, 2004 and 2010 ...... 48 Figure 2.6: Universal HTC for universal prevention and universal treatment (HTC+) ...... 51 Figure 2.7: Typology of approaches to HIV testing and counselling ...... 56 Figure 2.8: Acceptance of HTC in community-based approaches ...... 63 Figure 2.9: Spectrum of approaches to HIV self-testing ...... 66 Figure 2.10: OraQuick In Home HIV Test Kit ...... 68 Figure 2.11: Hierarchical summary receiver operator characteristic (HSROC) plot for accuracy of WHO stage 3/4 disease in identifying CD4≤200 cells/mm3 ...... 89 Figure 2.12: Hierarchical summary receiver operator characteristic (HSROC) plots for accuracy of WHO clinical stage 3/4 in identifying CD4≤350 cells/mm3 ...... 89 Figure 2.13: Evolution of ART eligibility criteria for adults in children in Malawian National programme ...... 101 Figure 2.14: Number of ART initiators by quarter in the Malawian National Programme 2004, Q2 to 2012, Q4 ...... 104 Figure 2.15: The HIV care cascade ...... 106 Figure 2.16: Effect of time-delineated outcome definition on ascertainment bias ...... 118 Figure 2.17: Euler diagram showing studies included in 3 systematic reviews of linkage to ART (n=51) ...... 121 Figure 2.18: Socio-ecological conceptual framework of progression through HIV care pathway ...... 134 Figure 2.19: Geopolitical map of Malawi ...... 141 Figure 3.1: Study wards and health facilities in Blantyre, Malawi ...... 144 Figure 3.2: Schemata of design of HitTB and CONDA-YAPA cluster randomised trials .. 147 Figure 3.3: Overall study area showing boundaries of 28 HitTB clusters ...... 149 Figure 3.4: Example cluster boundary showing residencies and annotated landmarks 153 Figure 3.5: Map book system for categorisation of cluster of residence of ART initiators at study clinics ...... 154 Figure 4.1: HIV care-seeking steps for patients in Malawian National HIV Programme under 3rd Edition of Guidelines for Use of Antiretroviral Therapy, 2008 .... 172 Figure 4.2: Schema of cohort study procedures ...... 175 Figure 4.3: Clinic attendance, uptake of HTC and HIV prevalence by sex and pregnancy status ...... 180 Figure 4.4: STROBE flowchart of participant outcomes ...... 186

14 Preface | List of figures

Figure 4.5: Completion of HIV care pathway steps if WHO staged on same day as HIV diagnosis ...... 188 Figure 4.6: Kaplan-Meier plots of initiation of ART by (a) sex and pregnancy status; and (b) self-rated general health ...... 189 Figure 4.7: Loss to HIV care in primary clinic attendees, Blantyre ...... 195 Figure 6.1: Study clusters and allocation ...... 237 Figure 6.2: Sample sizes required to detect difference in per-capita ART initiation with power=0.8 ...... 240 Figure 6.3: CONSORT flowchart of outcomes (modified for cluster randomised design) ...... 243 Figure 6.4: Cluster resident adult ART initiations during 6 months of HIVST availability 248 Figure 6.5: Uptake of HIV self-testing, by study arm, age group and sex ...... 249 Figure 6.6: Home and facility ART initiations by trial arm ...... 250 Figure 6.7: Proportion of ART initiators who were men ...... 252 Figure 6.8: Estimated impact on ART coverage over 6 months ...... 253 Figure 6.9: Retention on ART 6 months after initiation ...... 255 Figure 7.1: Community health worker ART eligibility assessment tool ...... 273 Figure 7.2: Participant flow in diagnostic accuracy study ...... 277 Figure 7.3: Receiver Operator Characteristic (ROC) curves comparing community health worker tool and WHO clinical staging system with CD4 count ≤350 cells/mm3 for assessment of ART eligibility, stratified by sex ...... 282

15 Preface | List of acronyms

List of Acronyms

ACTG AIDS Clinic Trials Group AIDS Acquired immunodeficiency syndrome ANC Antenatal clinic AUC Area under the curve ART Antiretroviral therapy CD4 CD4+ lymphocyte count CDC United States Centers for Disease Control CHW Community health worker CMV Cytomegalovirus COMREC College of Medicine of Malawi Research Ethics Committee CPT Cotrimoxazole prophylactic therapy CROI Conference on Retroviruses and Opportunistic Infections CRT Cluster randomised trial CI Confidence interval DBS Dried blood spot DHS Demographic and health survey DNA Deoxyribonucleic acid DV Domestic violence EIA Enzyme immunoassay FCO Facility initiation of care only arm FDA US Food and Drug Administration GPS Global positioning satellites HAART Highly active antiretroviral therapy HCT HIV testing and counselling HIV Human immunodeficiency virus HIVST HIV self-testing HR Hazard ratio IPT Isoniazid preventive therapy IPV Intimate partner violence IQR Inter-quartile range LR Likelihood ratio LSHTM London School of Hygiene and Tropical Medicine LSTM Liverpool School of Tropical Medicine MDHS Malawi Demographic and Health Survey MLW Malawi-Liverpool-Wellcome Trust Clinical Research Programme MSM Men who have sex with men MTCT Mother-to-child transmission of HIV OHC Optional home initiation of care arm OR Odds ratio PCP Pneumocystis carinii pneumonia PITC Provider-initiated testing and counselling PLWH People living with HIV

16 Preface | List of acronyms

PMTCT Prevention of mother-to-child transmission of HIV PrEP Pre-exposure prophylaxis RNA Ribonucleic acid ROC Receiver-operator characteristic RR Risk ratio SD Standard deviation SIV Simian immunodeficiency virus STD Sexually transmitted disease TB UNAIDS United Nations Joint Programme on HIV/AIDS WHO World Health Organization

17 Preface | List of supporting manuscripts

List of Supporting Manuscripts

1. MacPherson P, Lalloo DG, Choko AT, van Oosterhout JJ, Thindwa D, Webb EL,

Squire SB, Makombe SD, Hayes RJ, Corbett EL. Home Assessment and Initiation of ART

Following HIV Self-Testing: a Cluster-Randomised Trial to Improve Linkage to ART in

Blantyre, Malawi. 20th Conference on Retroviruses and Opportunistic Infections, Atlanta,

USA. March 2013. Abstract 95LB.

2. MacPherson P, Lalloo DG, Choko AT, Mann GH, Squire SB, Manda E, Makombe SD,

Desmond N, Corbett, EL. Understanding reasons for failure of linkage to ART in primary

health centres in Blantyre, Malawi. A prospective cohort study. 19th Conference on

Retroviruses and Opportunistic Infections (CROI) 2012, Seattle, USA, March 2012.

3. Sloan DJ, van Oosterhout JJ, Malisita K, Phiri EM, Lalloo DG, O’Hare B, MacPherson

P. Evidence of improving antiretroviral therapy treatment delays: an analysis of eight

years of programmatic outcomes in Blantyre, Malawi. BMC Public Health. 2013 13:490.

4. MacPherson P, Lalloo DG, Thindwa D, Webb EL, Squire SB, Chipungu GA, Desmond

N, Makombe SD, Taegtmeyer M, Choko AT, Corbett EL. A novel community health

worker tool outperforms WHO clinical staging for assessment of antiretroviral therapy

eligibility in a resource-limited setting. Journal of Acquired Immune Deficiency

Syndromes. 2014. 65(2): 74-78

5. MacPherson P, Houben RJGM, Glynn JR, Corbett EL, Kranzer K. Pre-treatment

loss-to-follow-up among TB patients in high burden and low-to-middle income

countries: a systematic review. Bulletin of World Health Organization 2014. 92: 126-

138

6. MacPherson P, MacPherson EE, Mwale D, Squire SB, Makombe SD, Corbett EL,

18 Preface | List of supporting manuscripts

Lalloo, DG, Desmond N. Barriers and Facilitators to linkage to ART in primary care: a

qualitative study of patients and providers in Blantyre, Malawi. Journal of the

International AIDS Society. 2012. 15(2): 18020.

7. MacPherson P, Corbett EL, Makombe SD, van Oosterhout JJ, Manda E, Choko AT,

Thindwa D, Squire SB, Mann, GH, Lalloo DG. Determinants and consequences of failure

of linkage to antiretroviral therapy at primary care level in Blantyre, Malawi: a

prospective cohort study. PLoS One. 2012. 7(9): e44794.

8. MacPherson P, Choko AT, Webb EL, Thindwa D, Squire SB, Sambakunsi R, van

Oosterhout JJ, Chunda T, Chavula K, Makombe SD, Lalloo DG, Corbett, EL.

Development and validation of a GPS-based map book system for categorizing cluster

residency status of community members living in high-density urban slums in Blantyre,

Malawi. American Journal of Epidemiology. 2013 177(10): 1143-7

9. MacPherson P, Lalloo DG, Choko AT, Mann GH, Squire SB, Mwale D, Manda E,

Makombe SD, Desmond N, Heyderman RS, Corbett EL. Suboptimal patterns of provider

initiated HIV testing and counselling, ART eligibility assessments and referral in primary

clinic attendees in Blantyre, Malawi. Tropical Medicine & International Health 2012.

17(4): 507-517.

10. MacPherson P, Webb EL, Choko AT, Desmond N, Chavula K, Napierala

Mavedzenge S, Makombe SD, Chunda T, Squire SB, Corbett EL. Stigmatising attitudes

among people offered home-based HIV testing and counselling in Blantyre, Malawi:

construction and analysis of a stigma scale. PLOS One 2011. 6(10): e26814.

19 Preface | Abstract

Abstract

This thesis is concerned with understanding the patient flow from diagnosis of HIV infection to initiation of antiretroviral therapy (ART). Untreated HIV-infected individuals have a high risk of progression to AIDS and death, and of transmitting infection to others. Interventions to ensure prompt linkage into HIV care and ART could have substantial individual and public health benefits.

At the turn of the millennium, very few HIV-infected individuals in sub-Saharan Africa had access to lifesaving ART. Since then, national HIV care programmes, supported by international funding, have driven impressive achievements in scaling up ART delivery, with over 9 million people having initiated ART by the end of 2012. Despite these achievements, the majority of HIV infected adults in sub-Saharan Africa remain unaware of their HIV status, meaning that they do not have the opportunity to access ART.

Additionally, concerning reports have arisen from a number of HIV care programmes about high rates of patient drop-out and death between diagnosis of HIV and ART initiation. Together, these factors could significantly hinder efforts to achieve universal knowledge of HIV status and coverage of ART.

The uptake of HIV testing, and magnitude of and reasons for drop-out of care between

HIV diagnosis and initiation of ART were investigated in a prospective cohort study and linked qualitative study at primary care level in Blantyre, Malawi. The main findings were of extremely low uptake of provider-initiated HIV testing and counselling (completed on only 13% of adult facility attendances) and high rates of patient loss to care before ART initiation. Difficulties in completing ART eligibility assessments (WHO clinical staging assessments and measurement of CD4 count) were the major barrier to successful

20 Preface | Abstract initiation of ART, with over-busy clinics, rushed providers, as well as high patient care- seeking expenses being significant contributory factors. Other sub-Saharan African countries that have implemented the public health approaches to HIV care delivery are likely to face similar problems.

To attempt to improve uptake of HTC and linkage into ART care, two novel interventions were investigated. A community-based cluster-randomised trial compared two approaches to improve linkage into HIV care: facility-based initiation of HIV care following home HIV self-testing (HIVST), or optional home initiation of HIV care following

HIVST. Uptake of ART, completion of HIVST, reporting of positive HIVST results and retention on and adherence to ART were the outcomes of interest. Over 6-months of

HIVST availability, there was a highly significant 3-fold increase in the proportion of the adult population initiating ART, and a doubling of the proportion of adults reporting positive HIVST results to community counsellors where home initiation of HIV care was available, indicating increased willingness to access home-based HIV care.

Having identified substantial problems with the use of the WHO clinical staging system for identifying ART eligibility, the accuracy of a brief novel community health worker

(CHW) ART eligibility assessment tool was compared against a gold standard of CD4 count. The CWH tool significantly outperformed the WHO clinical staging system in identifying CD4 count of <350 cells/mm3 in terms of sensitivity, positive predictive value, negative predictive value and area under the receiver operator characteristic curve.

Nevertheless, overall performance of the CHW tool was still suboptimal with nearly half of ART eligible participants missed, and was worse when compared against the new

WHO-recommended ART eligibility threshold of CD4<500 cell/mm3.

21 Preface | Abstract

In conclusion, suboptimal rates of facility-based HTC and subsequent linkage into care are potential major stumbling blocks in efforts to achieve universal access to ART.

Current ART eligibility tools are either not widely available (CD4 count measurement), or are insensitive, overly-complex and time-consuming (WHO clinical staging system).

Home initiation of HIV care following community-based HIVST overcomes these barriers, with high uptake of HIVST achieved over short time-frame and substantial and significantly increased population-level rates of ART initiation. In an era where “test-and- treat” is increasingly being seen as a strategy to impact upon the HIV epidemic, interventions that improve uptake of testing and linkage into care, such as home HIVST and initiation of HIV care, will be required.

22 Preface | Acknowledgements

Acknowledgements

I would like to thank everyone who has helped with the development, writing and research of this thesis. My supervisors, Liz Corbett, David Lalloo and Bertie Squire have been a wonderful source of academic and personnel support and encouragement throughout and I am privileged to have had the chance to work with them.

I am indebted to the patients, families and communities who gave their time and support to participate in these studies and who have undoubtedly helped lead improvements in care for people living with HIV in Malawi and also further afield. I am extremely grateful to the Blantyre District Health Office (in particular Mr. Eddie Manda) and to colleagues at the HIV Department of the Ministry of Health of Malawi, especially

Dr Simon Makombe, Dr Frank Chimbwadira and Dr Andreas Jahn for lending such unswerving support to the studies and providing logistical and operational support.

The research team was composed of a number of highly dedicated and skilled individuals, without whom the studies would not have gone as smoothly as they did.

Data collection was led by Jospeh Msimko, Wezi Mukaka, Joseph Phiri, Myamiko

Kwabwili and Eleanor Sanga. Both study nurses, Gift Radge and Jofrisi Jofrisi, displayed the epitome of professionalism in caring for study participants. Daniel Mwale performed the qualitative interviews and Hendrina Kalliani undertook translation. From Liz

Corbett’s Hit-TB Study, a large number of individuals provided expert support and skills, including Deus Thindwa, Augustine Choko, Vinjeru Shonga, Rodrick Sambauksani, Treza

Chunda, Kondwani Chavula, Mervis Chima, Aaron Mdolo (for supervising laboratory activities) and Lucky Ngwira. I am grateful to MLW (particularly Prof Rob Heyderman) and the College of Medicine of Malawi for hosting the studies and for Dr Nicola

23 Preface | Acknowledgements

Desmond for providing expert advice in qualitative methods and analysis. Emily Webb and Richard Hayes provided expert statistical advice for which I am very grateful.

Finally, I would like to thank my wife, Eleanor whose support, love and keen wisdom makes this thesis worthwhile. Our son Callan was born during the completion of this thesis and inspired many sleep-deprived writing sessions!

24 Chapter 1| Introduction

1. Introduction

1.1. Background

After more than 30 years, the human immunodeficiency virus (HIV) epidemic has had a devastating impact for people living in sub-Saharan Africa. Approximately 24 million people living in the region are infected with HIV, equating to two-thirds of the global epidemic1. In Southern Africa, the worst affected region, adult HIV prevalence is above

15% in almost all countries1. Life expectancy rates in many countries in sub-Saharan

Africa have plummeted, predominately affecting adults in working age groups, a major barrier to achieving sustained development in the region2.

Despite these stark figures, recent epidemiological data suggest shows that substantial progress is now being made in interrupting the epidemic in the worst affected countries.

A 25% decrease in the incidence rate of new HIV infections has been reported from sub-

Saharan Africa, with a number of countries, including Malawi, achieving declines of up to

50%1. Moreover, there has been a 32% decline in deaths due to HIV in the region between 2005 and 20111.

Combination antiretroviral therapy (ART) delivery has been central to these achievements. Early ART initiation has been shown to dramatically reduce incidence of

HIV-associated infections such as tuberculosis, and the risk of death and transmission of

HIV to others. Globally in 2012, 9.7 million HIV-infected individuals were taking ART, a thirty-fold increase since 20033. This public health success has been predominately been driven by rapid scale up in sub-Saharan Africa, where 7.5 million individuals are now taking ART. For the first time ever, the majority of people in need of ART in low- and middle-income countries (54%) are now taking treatment 3.

25 Chapter 1 | Introduction

Most countries in sub-Saharan Africa are on-track to achieve universal coverage of ART

(>80% of individuals in need) by 2015, with data from a cohort study in South Africa showing that HIV incidence fell by 17% for every 10% increase in the number of the population taking ART4. The scaling up of ART in low- and middle-income countries is estimated to have averted 4.2 million deaths, and a further 3.0 million deaths and 3.5 million new infections could be averted between 2012 and 2025 as a consequence of

WHO’s recommended raising the CD4 count threshold to <500 cell/mm3 in 20133.

Malawi has led the way in developing the public health approach to delivery of ART, with the national HIV programme supporting a limited number of effective ART regimens, minimal laboratory investigations, task-shifting to lower-cadre healthworkers, decentralisation of care to primary health care clinics and a unified reporting system to allow accurate and timely audit 5.

Despite these impressive achievements, a number of key issues could potentially threaten efforts to achieve WHO’s goal of 15 million HIV infected individuals taking ART by 2015. Particularly concerning are the continued low rates of annual HIV testing and counselling (HTC) among individuals living high prevalence countries 6 and continued late entry into care with advanced immunosuppression: over one quarter of HIV- infected individuals in low-income settings initiate ART with a CD4 count of <100 cells/mm3 3. Moreover, an increasing number of studies from African programmes are reporting that patients show a high risk of loss to follow-up between diagnosis of HIV and initiation of ART, undermining the value of HTC and contributing to adverse individual and public health outcomes.

26 Chapter 1| Introduction

Systematic reviews of studies reporting on linkage into HIV care in sub-Saharan Africa have found high rates of patient loss prior to ART initiation, but caution that insufficient studies have been conducted to accurately quantify the magnitude or associated risk factors. With moves towards universal test-and-treat (UTT) approaches to HIV care, achieving universal access to ART for people living with HIV (PLWH) will require novel combined approaches that act at all steps of the HIV care pathway to ensure high rates of annual uptake of HTC, with subsequent prompt and complete linkage into HIV care and ART initiation.

1.2. Project starting points

The initial stimulus for the current work was the clinical observation that high proportions of HIV-infected adults in sub-Saharan Africa were continuing to present for

ART initiation with advanced immunosuppression, and with consequent high rates of pre- and post-ART initiation mortality, despite rapid scale-up of ART delivery. At the beginning of this project, a small number of studies from sub-Saharan Africa had been published, suggesting that the magnitude of patient loss between diagnosis of HIV and initiation of ART was substantial. However, there was little understanding of the true magnitude or risk factors, or of interventions that could improve linkage into care. A mathematical modelling study published in The Lancet in 2009 showed that universal uptake of HTC and immediate initiation of ART (“universal test and treat”) could have rapid impact on HIV incidence and mortality in the high prevalence countries of sub-

Saharan Africa7.

Together, these initial observations suggested that, if confirmed, poor rates of linkage into HIV care could have a substantial individual and public health consequences, hindering efforts to turn the tide of the HIV epidemic. A successful intervention that

27 Chapter 1 | Introduction improved linkage into HIV care would be beneficial in both achieving the target of universal access to ART, as well as serving as one model for delivery of “test and treat” care for HIV.

Malawi, as well as being one of the poorest countries in the world, has been among the countries hardest hit by the HIV epidemic. Nevertheless, with limited resources, the national HIV care programme has been held up as a success story, developing and rapidly scaling up the public health approach to ART delivery now endorsed by WHO with support from international funding organizations. Investigation of rates of linkage into care and interventions to improve linkage in a model public health system such as

Malawi’s would provide robust evidence that could inform policymakers and practitioners across the region.

1.3. Aims and objectives

The overall aims of this thesis were to investigate the patient flow from HIV testing to initiation of antiretroviral therapy at primary care level in Malawi, a country with a generalised HIV epidemic, and to investigate strategies for improving uptake of HTC and linkage into HIV care.

The specific objectives were to investigate:

1. the proportion of adult primary clinic attendees who completed HTC

2. the flow of patients throughout the HIV care pathway, allowing identification of

key points where patient loss occurred.

3. the proportion of newly diagnosed HIV-positive adults who successfully initiated

ART and identify factors associated with ART initiation.

28 Chapter 1| Introduction

4. rates of mortality in the immediate post-HIV diagnosis period and identify

factors associated with death

5. patient, provider, socio-cultural and health system factors associated with

patient loss before ART initiation

6. the impact of a strategy of home-assessment and initiation of ART on uptake of

ART and HTC, disclosure of HIV positive results and on retention in ART and

adherence at 6 months.

7. the accuracy of a novel, brief ART eligibility screening tool conducted by

community health workers (CHWs) compared first to the WHO clinical staging

system, and assessed against a gold standard of CD4 count.

1.4. Outline of thesis

In Chapter 1, the research starting points, aims and objectives and the thesis plan are provided. In Chapter 2, a review of published literature relating to the epidemiology of

HIV in sub-Saharan Africa, approaches to HIV testing and ART delivery, and tools for assessment of ART eligibility is presented. Studies describing linkage from HIV diagnosis to initiation of ART are also reviewed. Methods that are common to more than one

Chapter are described in Chapter 3. The methods specific to individual studies are described in the respective chapters in which the study is described.

Chapter 4 refers to a prospective cohort study that was conducted in two primary care clinics in Blantyre, Malawi and describes the uptake of HTC and progression through the

HIV care pathway of individuals who were diagnosed HIV positive. Chapter 5 describes reasons for patient loss to care prior to ART initiation identified in qualitative interviews with cohort study participants and with clinic healthworkers.

29 Chapter 1 | Introduction

Chapter 6 describes a community-based cluster randomised trial of an intervention to impact upon poor rates of linkage to ART conducted among 16,660 adult residents of 14 urban neighbourhoods in north west Blantyre. Chapter 7 refers to a diagnostic accuracy study in which a novel CHW ART eligibility assessment tool was evaluated against the currently used assessment tool (the WHO clinical staging system), using the CD4 count thresholds as the referent for ART eligibility.

The main conclusions, recommendations for further research and potential implementation and intervention strategies are discussed in Chapter 8.

30 Chapter 2 | Literature review

2. Literature review

2.1. Human Immunodeficiency Virus Infection

2.1.1. Origins of the HIV pandemic

The evolutionary history of the human immunodeficiency virus (HIV) has had a major bearing on the epidemiology of the global pandemic. Multiple, independent zoonotic transmissions of simian immunodeficiency virus (SIV) from non-human primates to humans have occurred, predominately in West and Central Africa8. Over forty species of non-human primates are known to harbour species-specific SIV strains and zoonotic transmission of these viruses has generated numerous lineages of HIV infection in humans9, 10. The major lineages of HIV found in humans are classified within HIV type 1

(HIV-1) groups M, N, O and P; and HIV type 2 (HIV-2) groups A to H8, 11. HIV-1, group M, which is responsible for the current global pandemic11 originated from independent transmissions of SIVcpz to humans from the chimpanzee Pan troglodytes troglodytes12,

13.

HIV-2 is a much less common cause of infection in humans than HIV-1 and remains predominantly confined to West Africa14. The non-human primate source of HIV-2 infection was identified in 1989 from a sooty mangabey (Cercocebus atys) infected with

SIVsm15, 16, considerably earlier than the source of HIV-1 infection was found.

Transmission episodes of SIV from non-human primates to humans are likely to have occurred by the blood-borne route during hunting and butchering of primates for bushmeat9, and when monkeys are captured and kept as pets10. The earliest direct evidence of HIV infection in humans comes from stored serum and lymph node samples from Kinshasa, Democratic Republic of Congo in 195917, 18. The time to most recent

31 Chapter 2 | Literature review common ancestor (tMRCA) of HIV lineages is the elapsed time since two or more groups diverged from a single common viral ancestor and has been estimated using established molecular genetic techniques and mathematical modelling19. The oldest HIV-1 lineage in humans appears to be HIV-1 group M, with cross-species transmission estimated to have first occurred around 1853 (95% confidence interval 1799-1904)20.

Since these earliest transmission episodes, lineages of HIV have undergone dramatic diversification, predominately as a consequence of the large capacity in retroviruses for genetic variability and rapid evolution. HIV experiences extremely high rates of mutation and recombination during replication of the enzyme reverse transcriptase21, resulting in rapid generation of populations of genetically diverse virions within infected individuals.

With over 1010 virions produced each day22, infected individuals can have viral sequences that differ by up to 10%23, resulting in numerous circulating subtypes and recombinants. The highest genetic diversity of HIV is found in central Africa (where the global pandemic originated)8, with more homogenous distribution seen in other global region. The regional differences in HIV phylogeny, whilst highlighting the more mature epidemics within central Africa, are not thought to be the main factor in explaining inter- regional variation in rates of HIV transmission and prevalence8.

In 1981, the US Centers for Diseases Control (CDC) reported on a case series of five young adult homosexual men who were treated for pneumocystis carinii pneumonia

(PCP – now renamed Pneumocystis jiroveci pneumonia), mucosal candidal infection and cytomegalovirus infection between 1980 and 1981 in Los Angeles24. Although one of the five patients was noted to have leukopaenia, no evidence of malignancy was found in any of the men. By the time of publication, two of the patients had died. Over the subsequent months, further case series were reported of homosexual men from New

32 Chapter 2 | Literature review

York City and California25-27 who had evidence of clustering by sexual contact and who had unusual illnesses usually associated with immune deficiency, including: PCP;

Kaposi’s sarcoma; invasive CMV; cerebral toxoplasmosis; severe recurrent herpes simples virus infection; invasive candidiasis and cryptococcal meningitis. The spectrum of illnesses identified led researchers to suspect an acquired immunodeficiency syndrome, a hypothesis which was confirmed in 1983 by the identification of a novel T- lymphotropic retrovirus from affected patients28, 29. By this time, increasing numbers of cases of AIDS (as the acquired immunodeficiency syndrome had been named) were being identified in the US among homosexual men, as well as people without a history of homosexual contact: sex workers, recipients of blood donations (including patients with haemophilia) and migrants from the Caribbean (especially Haiti)30.

In 1984, the first cases of AIDS were reported from Zaire (now Democratic Republic of

Congo), and in a foreshadowing of the devastating impact of AIDS in sub-Saharan Africa,

Dr Jonathan Mann, the Centers for Disease Control’s (CDC) Epidemiological Officer based in Kinshasa, reported to the 1st International AIDS Conference that the incidence of AIDS was 10 to 15 times higher in Zaire than in the USA30.

2.1.2. Modes of transmission of HIV infection and prevention

Modes of transmission, with a focus on transmission in sub-Saharan Africa

HIV can be transmitted from person-to-person through a number of body fluids, including genital secretions and blood, and perinatally, either in-utero, during delivery or through breast milk31. The predominant mode of transmission globally is heterosexual transmission, accounting for an estimated 85% of all transmission events32. This is especially true in Southern Africa, the epicentre of the HIV pandemic, where data from

Rwanda and Zambia shows that heterosexual transmission between members of long-

33 Chapter 2 | Literature review term sero-discordant partnerships (where one member of the couple is HIV-positive and the other is HIV-negative) is the commonest mode of HIV transmission in the region, with between 55% and 93% of new infections attributable to this mode33. The risk of transmission in sero-discordant partnerships appears to be equally high regardless of whether the man or women is the infected partner34, 35. The prevalence of HIV sero- discordance has been found to range between 2% in Rwanda36 and 13% in Zimbabwe37 and Lesotho37. The likelihood that a HIV-positive member of a couple has a HIV-negative spouse lies between 45% and 75%33.

HIV transmission arising from multiple concurrent sexual partnership networks (where a person has more than one sex partner during the same time period) was previously thought to partially explain the very high prevalence of HIV in many countries in sub-

Saharan Africa38, 39. Mathematical modelling, based on epidemiological data showing that both men and women living with sub-Saharan Africa were more likely to have multiple long-terms concurrent partnerships than men and women in Thailand and the

USA40, 41, suggested that a 10-fold increase in epidemic size could be expected in areas where multiple concurrent partnerships were common compared to where they were not39. However, empirical data from demographic and health surveys (DHS) in South

Africa42 has shown that the role of multiple concurrent partnerships may have been overestimated. Increased mean number of lifetime sexual partners was independently associated with HIV acquisition, but high prevalence of multiple concurrencies in the local community was not associated with risk of acquisition42.

Since the early days of the HIV epidemic in central Africa, a high prevalence of HIV has been described among female sex workers and their clients43, 44. Especially high rates of infection have been found concentrated around areas where men undertake migrant

34 Chapter 2 | Literature review labour (such as mining areas45 and along trucking routes46). However, surveillance suggests that transmission between female sex workers and their clients is likely to be a minor contributor to HIV infection in sub-Saharan Africa32.

Although homosexual relationships are not legally recognised in most countries in

Southern Africa (the exception being South Africa)47, sexual transmission of HIV between men who have sex with men (MSM) is likely to be under-recognised as a mode of transmission in the region48. Although uncommon in the region, outside sub-Saharan

Africa, transmission as a result of needle sharing during injecting drug use is a more frequent mode of transmission32, 49.

Mother-to-child transmission of HIV (MTCT) can occur in-utero, intra-partum and, after delivery, through breast milk50. Whereas MTCT has virtually been eliminated in high income countries1, it remains a disappointingly common cause of HIV infection in infants in the sub-Saharan Africa, contributing to 330,000 new infections in 2011 alone1. In the absence of any interventions, the risk of MTCT in sub-Saharan Africa is estimated to be

35% for breastfeeding mothers, and 20% for non-breast feeding mothers51. In non- breastfeeding HIV infected mothers, 20% of HIV transmission occurs before 36 weeks of gestation in utero, whilst the remaining 80% occurs from 36 weeks to delivery52. In women who practice prolonged breastfeeding (18 to 24 months), as common in many parts of sub-Saharan Africa, 40% of infant infections are attributable to transmission during breast feeding52.

2.1.3. Risk factors for HIV transmission and prevention strategies

The infectivity of HIV can be estimated as the risk of transmission per heterosexual coital act53. The Rakai Project in Uganda followed HIV sero-discordant couples in the era

35 Chapter 2 | Literature review before treatment for HIV became widely available and showed that the probability of transmission per coital act was 0.11% (95% CI 0.08%–0.15%), with no statistically significant difference in rate of transmission by sex of the index case within the couple53.

A number of factors have been shown to increase the risk of transmission during heterosexual contact including: the HIV viral load of the infected partner54; the presence of genital tract infections, including other sexually transmitted infections; sexual contact that is violent or causes genital tract damage, whether the male partner is circumcised; and whether condoms are used during intercourse. Interventions to reduce HIV transmission are based on a firm understanding of the modes and risk factors for infection, and are described below.

The HIV viral load (the concentration of HIV viral RNA found per millilitre of blood) is the strongest predictor of HIV transmission during heterosexual intercourse55, for MTCT56, 57 and for transmission between MSM58. The HIV viral load in blood is strongly correlated with the HIV viral load in genital secretions and so can be used as a proxy measure of infectivity59. Studies from Rakai, Uganda show that that there is a significant dose- response relationship between blood HIV viral load and risk of heterosexual transmission. After adjusting for other factors associated with HIV transmission, for every tenfold increase in HIV blood viral load, the risk of transmission to an uninfected partner is independently increased by a factor of 2.45 (95% CI: 1.85-3.26)54. No HIV transmissions occurred with a blood viral load of less than 1500 copies/ml54.

Reductions in the HIV viral load (both as a result of result of an infected person entering the asymptomatic latent phase of infection – see Chapter 2.1.4 – and as a consequence of treatment with antiretroviral therapy) are associated with reduction in the risk of

36 Chapter 2 | Literature review heterosexual transmission60. Evidence from ecological studies61, surveillance data62 and from mathematical modelling63 shows that interventions to reduce in the so-called community viral load (the mean viral load of a population of people) could lead to substantial reductions in HIV incidence. Of note, in a well-characterised DHS site in

KwaZulu Natal, South Africa, a cohort of 16,667 HIV-negative adults were followed up between 2004 and 2010, a period when ART was being rapidly rolled-out in the district4.

After adjusting for other factors known to be associated with HIV infection, the risk of

HIV acquisition significantly declined with increasing local ART coverage (the proportion of HIV-infected ART eligible individuals who were taking treatment); individuals living in communities where mean coverage of ART was between 30 and 40% were 38% less likely to acquire HIV than individuals living in communities where ART coverage was

<10%4.

At the individual level, three broad approaches have been taken to reduce the risk of

HIV transmission using antiretroviral therapy (ART): early initiation of ART for HIV infected individuals to rapidly reduce the viral load; pre-exposure prophylaxis (PrEP - where ART is given to individuals at risk of infection); and post-exposure prophylaxis

(where ART is given to HIV-negative individuals following an exposure event such as unprotected sex, or occupational exposure within a health care setting).

In a multisite individually randomised controlled trial (RCT) of immediate ART initiation following HIV diagnosis compared to delayed initiation until national ART eligibility criteria were met (CD4 count <350 cells/mm3), Cohen et al found a 96% reduction in HIV transmission to uninfected partners in the group initiated onto ART immediately after diagnosis60.

37 Chapter 2 | Literature review

Pre-exposure prophylaxis has been examined among a number of placebo controlled randomised trials in different populations, including in the HIV-negative partners of HIV- infected MSM in San Francisco64 (where a 44% reduction in HIV incidence was found); in sero-discordant couples in Kenya and Uganda65 (where a 75% reduction in incidence was found); and heterosexual couples at risk of infection in Botswana66 (where a 63% reduction in incidence was found). In contrast the FEMPrEP study, in which HIV-negative women were randomised to receive either ART PrEP or placebo, no protective effect was observed67, although adherence to PrEP was noted to be suboptimal in this study.

HIV infection can also be prevented with a vaginal microbicide gel containing tenofovir disoproxil fumarate (TDF) immediately before and up to 12 hours after sex. In an RCT in

KwaZulu-Natal, South Africa, women using the microbicide gel were 39% less likely to become infected than women using a placebo gel68.

The foreskin of the penis has a high concentration of Langerhan’s cells, dendritic cells,

CD4 cells and macrophages, making it a vulnerable site for HIV infection69. Before 2005, over 80 observations studies suggested that circumcised men were at lower risk of HIV acquisition compared to uncircumcised men. A systematic review of these studies showed an overall protective effect of circumcision of 44%70. Since then, three randomised controlled trials have investigated male circumcision as a HIV prevention tool in countries in sub-Saharan Africa71-73, finding significant protective efficacy ranging between 51% and 60%. Male circumcision is now recommended by WHO as a key HIV prevention strategy for high prevalence countries74.

Infection with sexually transmitted diseases (STDs) such as herpes simplex virus, chlamydia and gonorrhoea increases genital shedding of HIV75, increasing risk of HIV

38 Chapter 2 | Literature review transmission during intercourse76. In the first study to demonstrate the effect of a preventative intervention on HIV incidence in a general population, a cluster randomised trial of improved case management of patients presenting with symptoms of STDs to study clinics within intervention clusters was conducted in Mwanza, Tanzania. Analysis of primary trial endpoint showed a 42% reduction in HIV incidence over two years (risk ratio: 0.58, 95% CI: 0.42-0.79), with reductions in HIV incidence observed among both men and women77. However, three subsequent RCTs all demonstrated no overall significant effect of STD treatment on HIV transmission78.

In a cluster-randomised trial conducted in Rakai, Uganda79, the effect of 10-monthly mass community STD treatment with antibiotics was compared to standard STD management. The authors reported that, after three rounds of mass STD treatment, the adjusted rate ratio for HIV incidence compared between groups was 0.97 (95% CI: 0.81-

1.20), indicating no effect.

In Masaka, Uganda80, 18 communities were randomly allocated into three groups: behavioural interventions alone; behavioural interventions and syndromic management of STDs; and standard of care. Comparison of the rate of HIV infection between the behavioural intervention and syndromic STD management group and the standard of care group gave a rate ratio of 1.00 (95% CI: 0.63-1.60), indicating no effect.

In Zimbabwe81, 12 communities were randomly allocated to receive either a programme consisting of community-based peer education, free condom distribution, income generation programmes and improved syndromic management of STDs in facilities, or standard of care. The overall incidence rate ratio for HIV was 1.3 (95% CI: 0.92-1.80),

39 Chapter 2 | Literature review showing no effect. However, there was a statistically significant reduction in HIV incidence among men.

Meta-analysis combining the four studies shows no evidence of effectiveness of improved STD management on HIV incidence (pooled risk ratio: 0.93, 95% CI: 0.67-

1.30)78. A number of possible reasons for the difference between the Mwanza trial

(which did show an effect) and the other three trials (that did not show an effect) exist.

In particular, the baseline prevalence of STDs was substantially higher in Mwanza compared to the other sites. Additionally, authors have speculated that where HIV incidence is rapidly rising (as was the case in Mwanza), a greater proportion of partnerships are likely to be sero-discordant and having an STD will increase the risk of acquiring HIV within a discordant relationship78. In contrast, where epidemics have stabilised (such as in the other three study sites), a greater proportion of relationships will be concordant, meaning that treatment of STDs will have little impact on HIV transmission. Finally, broadly different strategies were tested within the four trials. It may have been that the strengthened facility-level management strategy in Mwanza was able to identify individuals at high risk of HIV as opposed to the other community- based strategies. Although improved management of STDs does not appear to reduce risk of HIV, better control of STDs is an important public health goal in its own right.

Condoms have been used for centuries as contraceptives, and more recently to prevent infection with STDs82. There are two main types of condoms: the male condom, and the female condom82. There have been no randomised control trials assessing the effectiveness of either male or female condoms in preventing HIV infection83. However, systematic review and meta-analysis of thirteen observational studies of male condom use in heterosexual couples found an estimated effectiveness of 80% when used

40 Chapter 2 | Literature review consistently83. There are insufficient data available on the effectiveness of female condoms in preventing HIV infection84. Major barriers to the widespread usage of condoms for HIV prevention in sub-Saharan Africa are limited availability, low acceptability and desire for fertility85.

Traumatic sexual encounters, including rape86, increase the risk of HIV transmission, with genital abrasions and lacerations facilitating viral infection87. In recent years, additional evidence has emerged linking domestic violence (DV) and intimate partner violence (IPV) with risk of HIV infection88, 89. In South Africa, women who reported more than one episode of IPV during follow-up as part of a cluster randomised trial had an adjusted incidence rate ratio of HIV acquisition of 1.51 (95% CI: 1.04-2.21) compared to women who reported one or no episodes88. The population attributable fraction for IPV in this cohort was 11.9%, showing the important substantial contribution that IPV makes to HIV infection. In addition to genital tract trauma, women who experience IPV are thought to be at risk of HIV because of multiple interrelated factors mediated by gender inequalities including: having more risky partners (i.e. their partners may be more likely to have more sexual partners, HIV and STDs); reduced ability to negotiate safe sex; more frequent occurrence of sex whilst intoxicated; and a higher likelihood of being involved in sex work88. Two cluster randomised trials that investigated structural interventions to improve gender equity and reduce HIV risk showed no effect on HIV incidence, but did find reductions in occurrence of IPV90, 91.

2.1.4. Pathogenesis and natural history of HIV infection

HIV is a lentivirus that is comprised of a central capsid surrounded by a lipid membrane protein (ENV) studded with a cross-membrane glycoprotein gp41 bound to gp120

(Figure 2.2)28, 29, 92. Within the capsid (which contains the viral p24 protein), there are

41 Chapter 2 | Literature review two copies of single-stranded RNA that code for enzymes that are essential for viral replication: reverse transcriptase and integrase92. Replication of HIV occurs within host

CD4+ T lymphocyte cells, and is dependent on viral cell entry93.

Figure 2.1: Structure of HIV

Source: US National Institute of Health

In the immediate period following HIV entry into local tissue (for example in the genital tract), a so-called eclipse phase occurs, lasting approximately 10 days when HIV virus cannot be detected in blood or lymphoid tissue94. Subsequently, HIV disseminates to multiple organs and is found in extremely high concentrations (up to 106 copies/ml) in peripheral blood95. During this acute infection viraemic period, there is massive release of pro-inflammatory cytokines and chemokines96. Individuals in this phase may experience an acute retroviral syndrome with rash, night sweats, fatigue, myalgia and lymphadenopathy. Antibodies to HIV become detectable by sensitive enzyme

42 Chapter 2 | Literature review immunoassays (EIA) from about day 1395 (known as “seroconversion”). Previously, the period between infection and seroconversion was known as the “window period”.

Following seroconversion, there is a precipitous drop in the blood HIV viral load as the patient enters the phase of clinical latency (Figure 2.3). Additionally, there is a temporary rapid decline in the CD4+ lymphocyte count as a consequence of massive on- going viral replication within these cells, and their subsequent destruction97.

Figure 2.2: HIV viral load and CD4+ count during course of untreated infection

Source: after Panteleo et al, NEJM, 199397

During the phase of clinical latency, the viral load establishes at a “set-point”, with considerable heterogeneity in set-points observed between infected individuals98.

Individuals with higher viral set-points have a more rapid course of infection99 and are more likely to transmit HIV to others54, leading some authors to hypothesis that the viral set point represents evolutionary adaption of the viral populations to maximise reproduction before the host is killed100. The phase of clinical latency, which may last up to 8 years and longer, is characterised by immune activation with inflammation in a

43 Chapter 2 | Literature review number of organs101, progressive decline in CD4+ lymphocytes102 and increased susceptibility to opportunistic infections and malignancies103.

A number of studies have described the strong association between absolute CD4+ lymphocyte count and risk of opportunistic infections. In particular, a cohort study in

Cape Town, South Africa104, that prospectively followed 2086 HIV-infected adults, found a mean decline in CD4 count of 47 cells/mm3 per year, similar to that reported from developed countries105. Tuberculosis (incidence rate: 5.7 per 100 person years), and oral candidiasis (15.2 per 100 person-years) were the only opportunistic infections that occurred commonly with a CD4 count of between 200 and 350 cells/mm3 and were the only opportunistic infections that occurred with a CD4 count of greater than 500 cells/mm3. Chronic diarrhoea, wasting, tuberculosis and invasive candidiasis became more common in the 50 to 200 cell/mm3 stratum. At CD4 counts of less than 50 cells/mm3, many opportunistic infections were common (>1 per 100 person-years).

These data, and meta-analysis of clinical trials showing strongly reduced risks of severe bacterial infections, fungal infections and death with co-trimoxazole prophylaxis (CPT)106,

107, and reduced incidence of pulmonary TB with isoniazid preventative therapy (IPT)108, were instrumental in informing the development of guidelines for the recommended use of chemoprophylaxis for HIV infected individuals in high HIV prevalence settings109,

110.

Although more than 38 cohort studies representing 13,000 HIV-infected people have described the natural history of untreated HIV in industrialised countries111, only one such study has been conducted in Africa112. Measurement of time to first AIDS-defining episode in Africa has been limited by diagnostic capabilities111. In the Ugandan study, median time to death was 9.8 years overall, approximately one year lower than the

44 Chapter 2 | Literature review median survival time for people who seroconverted between 15 and 24 years in developed countries112. However, careful review of these data have suggested that an appreciable difference in time from seroconversion to AIDS does not in fact exist111, rather it is survival after the first AIDS-defining diagnosis that is curtailed in resource- limited settings.

2.1.5. Global and sub-Saharan African epidemiology of HIV

At the end of 2011, there were 34 million people living with HIV compared to 26 million in 1999 – a 30% increase1. There are large disparities in the global burden of HIV infection (Figure 2.4); the region with the largest number of infected people is sub-

Saharan Africa, where 23.5 million HIV-infected people live (69% of the global total), despite having only 11% of the global population. The ten countries with the highest adult (15-49 years) prevalence of HIV infection in 2011 are all found in sub-Saharan

Africa, with prevalence ranging from 6.5% in Uganda to 25.9% in Swaziland1.

45 Chapter 2 | Literature review

Figure 2.3: Prevalence of HIV infection, 2011

Source: UNAIDS1

In 2011, there were an estimated 2.5 million new HIV infections, 22% lower than 10 years previously1. Twelve of the high HIV prevalence countries in sub-Saharan Africa are estimated to have reduced the number of new infections by greater than 50% compared to 2001, with ten other countries estimated to have a 25% decline in new infections.

Only Guinea Bissau had an increase in incidence rate of HIV infection in the 47 sub-

Saharan African countries where data are available. These impressive declines are likely to reflect a combination of factors, including the increasing availability of antiretroviral therapy (ART – see Chapter 2.3), the impact of preventative strategies, behavioural responses and the maturation of the epidemic1.

Substantial reductions in the rate of mother-to-child transmission of HIV globally and in sub-Saharan Africa have also been observed. Between 2003 and 2011, there was an estimated 43% reduction (from 560,000 to 330,000) in new infections among children

46 Chapter 2 | Literature review younger than 15 years, and a 26% reduction in HIV-related deaths among children (from

120,000 to 90,000)1.

However, enormous challenges remain. In 2011 1.3 million people in sub-Saharan Africa were estimated to have died of HIV-related illness, 72% of the total global burden of deaths attributable to the HIV epidemic1. Additionally, the limited data available suggest that there may be large unrecognised epidemics among MSM and injecting drug users in many African countries1.

2.1.6. HIV epidemiology in Malawi

The most accurate and up-to-date estimates of HIV prevalence in Malawi are provided in the National Statistics Office’s 2010 Malawi Demographic and Health Survey (2010

MDHS)113. As a previous Demographic and Health Survey was conducted in 2004 (2004

MDHS), it is possible to estimate national and population-based trend data for HIV prevalence.

The MDHS used the 2008 Malawi Population and Housing Census (MPHC) as a sampling frame. Within the MPHC, individuals were identified at a number of levels, including the regional level (Northern, Central and Southern Regions), the District-level (28 Districts in total), and the Enumeration Area-level. To obtain a representative population for the

MDHS, a stratified two stage random sampling process was undertaken, using the list of census-enumerated households as the sampling frame. In total, 27,345 households were selected for interview in the 2010 MDSH. From these households, a sub-sample comprising of one third of households was randomly sampled to participate in HIV testing. During HIV testing, all members of selected households aged 15-49 years old

47 Chapter 2 | Literature review were asked to provide anonymous finger-prick dried blood samples (DBS) for HIV testing.

In total, 87% of household members age 15-49 underwent DBS HIV testing, with higher completion among members of rural households (88%) compared to urban households

(84%) and in women (90%) compared to men (84%)113. There were similar patterns of completion by age group, educational status and household wealth quintile113.

The overall HIV prevalence in 2010 in Malawi was 10.6%, with the highest prevalence seen in people in the 35-39 years age group (24%)113 – Figure 2.5. The HIV prevalence was higher among women aged 15-49 years (12.9%) compared to men of the same age

(8.1%)113.

Figure 2.4: HIV prevalence by sex and age group, Malawi, 2004 and 2010

Men 2010 25 Women 2010 Men 2004 20 Women 2004

15

10 HIV Prevalence (%)

5

0 15-49 20-24 25-29 30-34 35-39 40-44 45-49 Age Group

Source: 2010 MDHS113

48 Chapter 2 | Literature review

Comparing HIV-prevalence by sex and age group between 2004 and 2010, there has been a shift in the age distribution of infection, with a trend towards higher prevalence in older age groups for both men and women over time113. This is indicative of an epidemic maturing in the presence of availability of life-prolonging treatment (ART), resulting in a cohort of people who were infected with HIV at a younger age (in their twenties) surviving into their thirties.

There is pronounced geographical variation in the HIV epidemic in Malawi, with a strong

North-South gradient. Adult HIV prevalence is 15% in the Southern Province compared to 8% in the Central Province and 7% in the Northern Province113. Urban residents (and especially urban women) had a substantially higher HIV prevalence than rural residents

(17.4% vs. 8.9% respectively)113. There was no discernible trend in HIV prevalence by educational attainment, but a clear trend existed in household wealth quintile, with individuals in the lowest quintile (7.5%) having a substantially lower prevalence than individuals in the highest wealth quintile (15.3%)113.

Both partners were HIV-negative in 85% of cohabiting couples tested, both were HIV- positive in 6% and 9% of couples were HIV sero-discordant113. Risk factors for being in a sero-discordant relationship included increasing age group, increasing educational level and urban residency113.

Data on HIV prevalence among pregnant women in Malawi is available from two sources: the 2010 MDHS, and sentinel surveillance among all pregnant women attending a sample of clinics providing antenatal care between 1994 and 2010114. In the 2010

49 Chapter 2 | Literature review

MDHS, the HIV prevalence among pregnant women was 13% (compared to 9% in non- pregnant women)113, whereas in antenatal surveillance data, HIV prevalence was 10.6%.

2.2. HIV testing and Counselling

2.2.1. Objectives of HIV testing and counseling

HIV testing and counselling (HTC) is the entry point to treatment and prevention services. In 2012, recognising the importance of equity in access to HTC, WHO released guidance stating that HTC should reach all areas and populations that could potentially receive benefit: so-called “Universal HTC for universal prevention and treatment”, or

“HTC+”115. The range of services that can be accessed through HTC are shown in Figure

2.6 and include ART for individual benefit, ART to prevent transmission of HIV to others

(treatment as prevention [“TaSP”], prevention of mother to child transmission

[“PMTCT”], pre-exposure prophylaxis [“PrEP”], and microbicides), and other prevention services (male circumcision, behavioural interventions and harm reduction interventions). It is clear that, to be able to access the benefits of these services, people infected with HIV, or at-risk of infection with HIV, must receive HTC and subsequently link into services.

50 Chapter 2 | Literature review

Figure 2.5: Universal HTC for universal prevention and universal treatment (HTC+)

HTC

ART intervenons Other HIV services

ART TaSP Male circumcision PMTCT Behavioural prevenon PrEP Harm reducon Microbicides

Source: WHO115 ART: antiretroviral therapy TaSP: treatment as prevention PMTCT: prevention of mother to child transmission PrEP: pre-exposure prophylaxis

HTC should follow a rights-based approach. Mandatory HIV testing (for example mandatory testing of prisoners) is never appropriate and healthcare providers, family members or others should not coerce individuals into testing for HIV. To reinforce this commitment to a rights-based approach to HTC, WHO115 emphasises 5 guiding principles for HTC, known as the “5Cs”:

• Consent. HTC should be performed with informed consent and individuals

should have opportunity to refuse testing

51 Chapter 2 | Literature review

• Confidential. What is discussed during HTC and the result of the test should not

be disclosed to anyone else without the express consent of the person being

tested

• Counselling. HIV testing should be accompanied by appropriate and high quality

counselling

• Correct test result. HTC providers should strive to provide high-quality testing

services and quality assurance procedures should be in place

• Connections to care. HTC providers should strive to achieve timely and complete

linkage into care and prevention for individuals receiving HTC

2.2.2. HIV testing technologies

HIV tests may either be performed in laboratories on whole blood, serum, plasma (from dried blood spot samples), urine or oral fluid samples, or at the point of care using rapid diagnostic kits (blood or oral fluid based)116. Most HIV diagnostic tests screen for antibodies to HIV, using enzyme linked immunosorbent assays (ELISA). First generation

HIV ELISA tests were introduced in 1985 and current 4th generation assays allow the detection of antibodies to HIV, as well as the presence of the p24 antigen, allowing earlier detection of HIV116. The US Food and Drug Administration (FDA) lists twelve assays which are approved for detection of antibodies (and in some cases p24 antigen) to HIV117. However, a much larger number are produced and marketed worldwide. HIV

RNA may also be detected using qualitative polymerase chain reaction (PCR) and

Western blot technologies in laboratories116.

Recognising that performing HIV testing in laboratories is not practical or feasible for many health systems, and may act as a significant barrier to individuals testing, WHO has issued guidance recommending that in high HIV prevalence populations (prevalence

52 Chapter 2 | Literature review greater than 5%), rapid diagnostic tests should be the first choice115. Guidelines are designed to maximise accuracy and convenience, while minimising cost115.

In these settings, WHO recommends that all specimens are first tested with one approved assay and those that are non-reactive are considered HIV negative and advised to retest within 3-6 months115. Specimens that are reactive on the first test should be retested using a different second assay, and if both are reactive, the result should be reported as HIV-positive115. Where results are discrepant, both assays should be repeated, as this may eliminate discrepant results that are due to technical error115. If repeat assays are concordant (i.e. both non-reactive or reactive), the result can be reported HIV-negative or HIV-positive respectively115. Where results remain discrepant, testing should be done with a third assay, which, if non-reactive, the result can be reported as HIV-negative115. If the third test is positive, an inconclusive result is given and the individual should be advised to repeat the testing algorithm in 14 days115. This algorithm results in an overall positive predictive value of 99%115. In lower HIV prevalence settings, alterative algorithms are recommended115.

2.2.3. Uptake of HIV testing and counselling

Recognising that increasing uptake of HTC is an essential first step in improving access to

HIV treatment and prevention services, UNAIDS has set a goal that 80% of HIV-positive people worldwide should know their status by 2015118. WHO also recommends that individuals living in countries with generalised HIV epidemics should undergo HTC at least every 6-12 months119.

However, knowledge of HIV status falls below these targets, despite improvements over the last four years. In 2009, an estimated 38% of HIV-infected people were aware of

53 Chapter 2 | Literature review their status120. Data collected by WHO showed that in 2012, in 124 countries, 118 million people received HTC, a 12% increase from 20113. Nevertheless, in most low and middle income countries surveyed, the majority of adults have never been tested for HIV and so will be unaware of their status and unable to access the benefits of treatment and prevention. Men have significantly lower uptake of HTC than women in nearly all countries, likely related women receiving HIV testing during antenatal care3.

More detailed regional data comes from AIDS indicator surveys performed as part of demographic and health surveys conducted in a number of sites in sub-Saharan Africa countries. Data on 15 AIDS indicator surveys conducted between 2009 and 2012 show that a median of 20.0% of women and 20.5% of men report having undergone HTC in the last 12 months and that 49.1% of women and 62.8% of women have never been tested6.

Uptake of HTC among pregnant women is also suboptimal, with 40% of pregnant women in low and middle income countries receiving HTC during pregnancy during

2012, an increase from 26% in 20093. However, there is substantial regional variation, with higher rates of testing in countries of sub-Saharan Africa. For example, Kenya,

Malawi, South Africa, Mozambique Botswana, Namibia, Swaziland and Zambia all achieved greater than 80% coverage of HTC during pregnancy in 20123. Uptake of HTC among individuals in key populations and adolescents is extremely low in all global regions3.

These data show that there is still considerable way to go until the goal of universal HTC for universal access to treatment and prevention will be achieved. Moreover, mathematical models based on prospectively collected data show that a substantial

54 Chapter 2 | Literature review proportion of HIV transmission events (for example up to 40% in Lilongwe, Malawi) are attributable to individuals in the acute HIV infection stage when viral loads are high121.

This reinforces that fact that a high proportion of individuals living in high HIV prevalence countries should undergo regular HTC. In addition to identifying previously undiagnosed HIV-positive individuals in the latent stage of infection (who may contribute relative little to transmission but will require treatment for individual benefit), this strategy could also increase rates of diagnosis of HIV during the acute phase of infection, allowing behavioural and TaSP interventions to prevent transmission, as well as identifying HIV-negative individuals who can access prevention services122.

2.2.4. Models of HIV testing and counselling

Since the earliest days of the epidemic, a wide range of models of HTC have been implemented in various settings across the world. To better understand how best to improve HTC services, WHO has developed a typology of the various different models of

HTC that are currently being implemented (Figure 2.7).

55 Chapter 2 | Literature review

Figure 2.6: Typology of approaches to HIV testing and counselling

Facility-based Community-based HTC HTC

Clinical settings Other settings Home-based door-to-door

ANC VCT Home-based index TB Drop in for key Outreach for populations key Outpatients populations

Events Other clinical (e.g. STI clinics) Workplace

Schools

Source: World Health Organization115

The various recognised models for HTC are grouped under two broad approaches: facility-based HTC and community-based HTC.

2.2.5. Facility-based approaches to HTC

In facility-based HTC approaches, individuals undergo HTC at health facilities. HTC may be provider initiated (PITC), where a health worker recommends HTC to someone attending a health facility, or client-initiated, where individuals attend facilities and request HTC119. Facility-based HTC may take place in a wide variety of clinical and non- clinical settings, including antenatal clinics, TB clinics, general outpatients departments, inpatient wards, STI (and other specialist outpatient clinics), as well as at standalone

56 Chapter 2 | Literature review testing sites (voluntary counselling and testing [VCT]) and drop centres for members of key populations (e.g. drug users drop in centres)115.

In the early response to the epidemic in sub-Saharan Africa, two approaches to facility- based HTC prevailed: VCT and diagnostic testing for people with signs, symptoms or medical conditions related to HIV.

Voluntary counselling and testing (VCT)

VCT is defined by an intervention initiated by persons who wish to learn their HIV status and uptake is dependent upon individuals recognising that they may be at risk of HIV and seeking out testing123, 124. VCT centres may be based at traditional health facilities

(e.g. primary health care centres, district hospitals), or in outreach sites in the community. In the late 1990s and early 2000s, as governments increasingly recognised the devastating impact that HIV epidemics were having in sub-Saharan Africa, rapid scale-up of delivery of VCT services was achieved in a number of countries. For example, in Kenya in 2000, only 3 sites offered VCT125. With support from the Ministry of Health and a media campaign, by 2005, there were more than 680 sites nationwide offering

VCT, with over half a million tests performed in that year alone.125 Overall in sub-

Saharan Africa by 2005, an estimated 16.5 million people per year were accessing client initiated HTC services in countries with generalised epidemics126.

The rationale behind the VCT approach was that individuals would wish to learn their

HIV status in order to take actions (e.g. reducing number of sexual partners, using condoms) to prevent themselves becoming infected if they tested negative, or to prevent transmission of infection to others if they tested HIV positive. Additionally, when ART became available, VCT was the access point to receive treatment for

57 Chapter 2 | Literature review individual benefit, as well as for prevention of mother to-child-transmission of HIV. A number of additional services were incorporated into VCT sites, including: screening and treatment for STDs, behavioural and educational interventions to reduce risk (including couples counselling), condom distribution, TB screening and home-based care for people with debilitating illness.

A small number of clinical trials have examined the impact of availability of VCT on uptake of testing. In a cluster randomised trial in Harare, Zimbabwe127, factories were randomised to receive either access to VCT at their workplace, or given vouchers that workers could use to access VCT at a site outside the workplace. In factories where VCT was available in the workplace 50% of workers underwent VCT, with uptake of HTC 12- fold higher than factories where workers had to travel off-site to request a test. In

Zambia, participants resident in clusters randomly allocated to receive VCT at a location of their own choice (home, clinic, or any other private place) were 5-times more likely to undergo HTC than residents of clusters allocated to availability of clinic-based testing only128. Again, in the control arm, uptake of VCT was low at only 12%.

A number of studies have evaluated the impact of VCT on HIV risk behaviour. A recent carefully conducted cohort study using DHS data from Kwa-Zulu Natal in South Africa found that among youths aged 15 to 24 years, HTC was associated with reduced risk of

HIV acquisition (adjusted hazard ratio: 0.59, 95% CI: 0.45-0.78)129. However, these data contrast with the majority of other studies that have examined the effect of VCT on HIV risk behaviour. Data presented at the 20th Conference on Retroviruses and Opportunistic

Infections from the ongoing large cluster-randomised trial Project Accept study showed no community-level effect on the risk of HIV acquisition from HTC exposure130, similar to

58 Chapter 2 | Literature review the results of a Cochrane systematic review and meta-analysis, which found no significant impact of VCT on rates of condom use or unprotected sex131.

In the large number of non-randomised studies and programmatic evaluations of VCT interventions, a number of key factors have been identified that have limited the widespread uptake of VCT. Failure to acknowledge the possibility of HIV-infection and the need for testing has been identified as a common reason for non-attendance in studies in Tanzania132 and Ethiopia133. Completion of HTC is as low as one-third where results are not provided on the same day as testing is undertaken and patients are asked to make a second visit to collect results (e.g. where testing is laboratory-based rather than point of care)134, 135. Location of VCT at convenient sites in facilities, or in the community, tends to increase uptake134, 136. Major barriers to attending for HTC include fear of being observed by others, cost of travelling to health centres, lack of confidentiality (particularly if providers are also members of the local community) and lack of ongoing care once testing is complete134, 136. Fear of the consequences of a positive HIV test is a significant reason for failure to attend for VCT134. This can include fear of ill health and death, in addition to fear of social consequences such as abandonment by partners, violence, loss of employment and stigmatisation and ostracism by communities134.

Provider initiated testing and counselling (PITC)

Recognising that because of the many barriers identified above, facility-based VCT approaches were having limited impact on population rates of uptake of HTC and on of knowledge of HIV status, leading to suboptimal opportunities to access ART, in 2007

WHO issued updated guidance for facility-based approaches to HTC. In this guidance,

WHO recognised that health facilities represented a key point of contact for people who

59 Chapter 2 | Literature review were HIV infected and in need of prevention, treatment, care and support, but that the majority of health facility encounters resulted in a missed opportunity to offer HTC. For example studies from the USA137, 138, the UK139 and Uganda140 found evidence of frequent repeated health facility attendance with missed opportunities to complete

HTC.

In response, WHO recommended an “opt-out” approach to HTC within health facilities, known as provider initiated HTC (PITC). In the PITC approach, health workers recommend HTC to individuals attending health facility, regardless of their reason for attending and with limited pre-test counselling in line with information recommendations in WHO and UNAIDS policy. PITC was recommended:

1. For all facility attendees, irrespective of epidemic setting, who’s clinical

presentation might result from underlying HIV infection

2. As a standard part of medical care for all patients attending health facilities in

generalised epidemics (HIV prevalence consistently greater than 1% in pregnant

women)

3. More selectively in concentrated (HIV prevalence consistently greater than 5% in

at least one key population, but below 1% in pregnant women) and low-level

epidemics (HIV prevalence less than 5% in any defined sub-population), with

individuals in at-risk groups and key populations prioritised.

PITC should be implemented in line with the 5C principles. Importantly, WHO notes that verbal communication is normally adequate for the purpose of obtaining informed consent. Pre-test counselling should be brief and designed to provide clients with information as to the reason for HIV testing, the benefits of HIV testing, services available following a positive or negative HIV test and an opportunity to decline testing.

60 Chapter 2 | Literature review

Certain groups who may be at increased risk of coercive testing or adverse outcomes of

HIV testing, including prisoners and people in abusive relationships, should receive more detailed counselling. Results-based post-test counselling should be provided to all individuals undergoing PITC, focused on ensuring clients understand their test results and a discussion of the implications of the test result and actions that should be undertaken.

Rapid scale up of PITC with low rates of refusal have been demonstrated in facilities in

Zambia141, 142, Uganda143, Ethiopia144, Botswana145, South Africa146 and among TB patients147-149, STD clinic attendees150 and antenatal clinic attendees151, 152. A systematic review of studies reporting on outcomes of programmes offering PITC in sub-Saharan

Africa identified 44 studies that showed acceptance of PITC ranging from 31% among outpatients in South Africa to 99% in hospital in-patients in Uganda153. Acceptance of

PITC was noted to be highest among women, people in the youngest age groups and among TB patients.

PITC clearly provides an important and substantial contribution to coverage of HIV testing in the countries of sub-Saharan Africa. PITC has increased uptake of HTC in clinical settings, particularly antenatal care154. However, given the continued suboptimal population rates of coverage of yearly HIV testing, it could be said that facility-based

PITC is necessary, but not sufficient, to achieve the high uptake required. There are number of groups who may find it difficult to benefit from a wholly facility-based PITC approach. In many settings women have greater access to HTC services due to their more frequent contact with health services; men, children and adolescents, rural and key populations vulnerable to HIV infection often have less access to facility-based services, and, even where services are available, these groups face barriers to access or

61 Chapter 2 | Literature review do not find services acceptable115. Individuals may also be reluctant to attend health facilities for fear of stigma and discrimination155, or because of competing time and financial demands with work156.

In all epidemics, complementary approaches to facility-based PITC that can reach a greater proportion of the population, especially currently underserved groups are required. In sub-Saharan Africa, this should include approaches to increase coverage of annual HTC among men and non-pregnant women.

2.2.6. Community-based approaches to HTC

Community-based HTC has been defined as HTC happening outside of health facilities. A systematic review in 2013157 identified eleven different models of community-based

HTC, including:

• Door-to-door testing (systematically offering HTC to household members in a

catchment area)

• Mobile testing for the general population (offering HTC via mobile services in

areas visited by the public, e.g. shopping centres, transport hubs)

• Index testing (offering HTC to household members of people with HIV and

people who may have been exposed to HIV such as spouses and sexual contacts)

• Mobile testing for MSM

• Mobile testing for people who inject drugs

• Mobile testing for female sex workers

• Mobile testing for adolescents

• Self-testing

• Workplace HTC

62 Chapter 2 | Literature review

• Church-based HTC

• School-based HTC

In the systematic review, Suthar and colleagues identified 61 studies including 713,632 participants that reported on acceptance of community-based HTC approaches157. In pooled analysis, acceptance of HTC was high across all approaches (Figure 2.8), ranging from 88% of participants offered index testing to 62% of participants offered school- based testing. Percentage of clients who were men ranged from 45% for index testing to

67% for workplace testing.

Figure 2.7: Acceptance of HTC in community-based approaches

100% 90% 80% 70% 60% 50% 40%

Acceptance (95% CI) 30% 20% 10% 0%

Source: Suthar et at 2013157

Community-based HTC was also able to reach a high proportion of first time HIV testers with a pooled average of 62% of all individuals receiving community-based HTC testing for the first time157. The HIV positivity rate was significantly lower in individuals testing

63 Chapter 2 | Literature review through community based sites than in facility-based sites (relative risk of positivity:

0.59, 95% CI: 0.37–0.96) and the proportion of individuals diagnosed with CD4 count

>350 cells/mm3 was significantly higher in community-based testers compared to facility-based testers (relative risk: 1.42, 95% CI 1.16–1.74)157. Importantly, rates of linkage into care following community-based HTC were high, with 80% of all HIV-positive individuals completing CD4 count measurement, and 73% of eligible participants initiating ART157. Accuracy of home-based HTC performed by lay healthworkers has also been found to be high158.

Despite the advantages of community-based HTC in terms of high rates of uptake,

(particularly among men and other hard-to-reach groups), high proportion of first-time testers and reasonably good linkage into care, a number of concerns remain.

In particular, although uptake has been noted to be high, there are little available data on the characteristics of population who are not reached by community-based HTC and their risk factors for HIV. It may be that individuals refusing community-based HTC may be at higher risk of HIV than individuals accepting testing. Moreover, the effect of community-based HTC on HIV incidence and infectivity (individual and community HIV viral load) is unknown, although a series of large cluster-randomised trials are currently investigating these endpoints in sub-Saharan Africa159, 160.

There are few available data on the effect of community-based HIV testing on HIV risk behaviour. A cluster randomised trial in South Africa, compared HIV prevalence between communities offered door-to-door HTC compared to communities with facility-based

HTC alone. A 55% reduction in multiple partnership and a 45% reduction in casual sexual partnerships was seen in communities where door-to-door HTC was available.

64 Chapter 2 | Literature review

Few studies to date have reported on repeat HIV testing, high levels of which mathematical modelling shows will be important in reversing epidemics121, 122. Where repeat testing has been evaluated, this has been part of demographic and health surveys and the high levels of repeat testing (up to 85%) may not reflect uptake of repeat testing seen under routine programmatic conditions161. No studies have reported on harm as a consequence of community-based HTC157.

Finally, community-based HTC approaches can be resource intensive and may not be sustainable once support provided during clinical trials is withdrawn. There is little evidence of the relative cost-effectiveness of different community-based HTC approaches (and in comparison to facility-based approaches) to guide policy-makers decisions for implementation. In a study from Uganda, the cost per HIV positive person diagnosed were found to be highest for standalone VCT sites ($19) compared to facility- based HTC ($12), index testing ($14) and door-to-door HTC ($8)162.

2.2.7. HIV self-testing

Self-testing as a screening tool to identify previously unrecognised conditions or risk factors for disease is already widely supported by health systems throughout the world.

For example, home pregnancy testing163, home chlamydia screening164, breast self- examination165 and home measurement of blood glucose166 and blood pressure167 are effectively and widely performed to a high degree of accuracy. Self-testing facilitates earlier identification of the condition of interest, prompts treatment seeking and may encourage behaviour modifications to reduce risk of future complications of disease.

65 Chapter 2 | Literature review

In a symposium hosted by WHO in April 2013, a consensus definition of HIV self testing

(HIVST) was developed:

“[HIV self-testing occurs when] the test is collected, performed and interpreted in private by the individual who wants to know about their HIV status.”168

A wide spectrum of approaches to HIV self-testing can be implemented, dependent on the degree of supervision by health workers. At one end of the scale, individuals can complete the self-testing process without any involvement of health workers

(unsupervised HIVST), for example where HIVST kits are obtained from vending machines our over the internet or by post. At the other end of the spectrum, individuals may self-test for HIV in the presence of a healthworker in a facility (Supervised HIVST) –

Figure 2.9.

Figure 2.8: Spectrum of approaches to HIV self-testing

Vending Community Community health Supervised in machine availability (e.g. worker supervision facilities pharmacies)

Unsupervised Supervised HIVST HIVST

Internet/postal Community health Unsupervised in worker distribution facilities

66 Chapter 2 | Literature review

Self-testing for HIV was first identified as a potential strategy in 1998, by Kevin De Cock, who noted that:

“While caution is clearly required, the reticence around the concept of self testing contrasts with modern approaches to self diagnosis or screening for other conditions such as breast self examination, home pregnancy testing, and melanoma awareness.”139

However, until very recently, few attempts had been made to evaluate the feasibility, acceptability, accuracy and outcomes of HIV self-testing. A WHO evaluation in 2011 noted that informal HIV self-testing (commonly by self-performing finger-prick rapid diagnostic kit testing) was noted to be common among health workers in sub-Saharan

Africa (with 41% of randomly sampled healthworkers in Mozambique reporting having self-tested for HIV)169. Whilst highlighting a potentially acceptable approach to expanding uptake of HIVST, this report raised concerns that individuals may be testing themselves for HIV without receiving adequate counselling and without sufficient measures in place to support HIV care and prevention afterward169. Moreover, the accuracy of HIVST performed in this manner was unknown169.

In July 2012, the US Food and Drug Administration reported results of an evaluation of the OraQuick in home HIV test kit (OraSure Technologies), a manually-performed oral fluid kit for detection of antibodies to HIV 1 and 2 (Figure 2.10).

67 Chapter 2 | Literature review

Figure 2.9: OraQuick In Home HIV Test Kit

Results of Phase 1 studies in laboratory samples showed that OraQuick in home test kit correctly identified 834/840 HIV-positive samples (sensitivity 99.3%) and correctly detected 3674/3682 (specificity 99.8%) HIV-negative samples170. In Phase 2 studies,

1031 individuals (514 known HIV-positive and 499 unknown) were recruited to perform the test. In this evaluation, sensitivity was found to be 97.8% and specificity 99.8%170. In

Phase 3 studies (in hands of intended users) in high and low prevalence settings, 5562 individuals were recruited to evaluate the kit. The sensitivity was found to be 91.7% and specificity to be 100.0%170. No serious adverse events were recorded170. The sensitivity in the phase 3 study was found to be below the FDA recommended minimum of 95%170.

68 Chapter 2 | Literature review

However, because of evidence of public health benefit in reducing undiagnosed HIV in the USA, in July 2013, the FDA granted approval for home use of the OraQuick in home test kit, meaning that consumers could buy the kit over the counter or over the internet170. A telephone hotline is available for individuals performing the home self- test. Evaluations of programmatic outcomes are awaited.

In sub-Saharan Africa, a HIV self-testing feasibility study171 was conducted in Blantyre,

Malawi in 2010 as part of the HitTB cluster-randomised trial described in Chapter 3.

Adult members of households and community groups in urban Blantyre were randomly selected and invited to undergo HIVST (using a Thai-assembled version of OraQuick, which does not contain the supplementary instruction booklet and test stand) followed by confirmatory fingerprick testing. 260/283 (92%) of individuals approached agreed to participate, with all opting to self-test171. HIVST was highly accurate (99.2%) and acceptable, with HIVST being the preferred option for subsequent testing. 10% of participants made minor procedural errors during self-testing. The majority of participants (90%) stated that they would value post-test counselling after HIVST171. The results of this feasibility study directly informed the design of the HitTB cluster randomised trial, as well as the nested cluster randomised trial described in Chapter 6.

Publication of the Blantyre feasibility study in 2011 and the US FDA approval of the

OraQuick in home HIV test kit for self-use stimulated considerable interest in HIV self- testing as a strategy to improve population coverage of HIV testing. A systematic review in 2013 identified 21 studies including 12,401 participants that examined the feasibility and acceptability of supervised and unsupervised HIVST172. Only one study from sub-

Saharan Africa (Choko et al’s study from Blantyre, Malawi171) was identified. Overall accuracy was noted to be high, with supervised HIVST (range: 97.4%-97.5%) having

69 Chapter 2 | Literature review higher accuracy than unsupervised HIVST (92.9% to 100%)172. No study reported on linkage into care following HIVST and adverse events were noted to be rare172.

In Kenya, a feasibility study of HIVST was conducted among health workers, although results remain unpublished173. In this study, 89% of 765 healthworkers accepted an oral swab HIV test kit (Calypte® AwareTM oral test), of whom, 96% tested within one week173.

Accuracy or outcomes were not assessed. Kenya is to date the only country in sub-

Saharan Africa to specifically approve HIVST in national HIV guidelines168.

Concerns have been raised that unsupported HIVST may result in increased risk of self- harm, suicide, coercive testing or interpartner violence. This has not been found in the systematic review of studies performed to data, however, greater surveillance data under routine conditions in resource-limited settings is required.

Despite this rapidly evolving evidence base, a considerable number of questions remain unanswered, including:

• What is the impact of HIV self-testing on

o Increasing population coverage of HIV testing (and repeat testing)?

o HIV incidence?

• What is the rate of serious adverse events (e.g. suicide, interpartner violence)

following HIVST?

• What are rates of linkage into HIV care following HIVST?

• What is the accuracy of HIVST when used under programmatic conditions in

resource limited settings?

70 Chapter 2 | Literature review

• What are the optimal models for implementation of HIVST in different

epidemiological situations?

2.2.8. Which models of HTC are recommended?

In the section above, an overview of the various models of HTC available has been presented. However, it is important to recognise that different models may be suitable for different epidemics and in different health systems. Moreover, more than one model may be implemented simultaneously to achieve maximal population access to HTC, with the central objective of increasing awareness of HIV status, improving uptake of treatment and prevention interventions and reducing mortality, morbidity and transmission.

WHO has summarised current recommendations for implementing HTC models in different epidemiological situations (Table 2.1). However, there remains a critical lack of evidence on the cost-effectiveness and impact of individual approaches, and approaches in combination to guide policymakers planning to implement HTC programmes with limited budgets.

71 Chapter 2 | Literature review

Table 2.1: WHO-recommended application of HTC models in different epidemiological situations Recommendations for Recommendations for HTC model concentrated or low-level generalized epidemic epidemic Facility-based Clinical settings Cost-effectiveness needs ANC assessment PITC in these services, which TB, STI clinics, OST, NSP PITC in every health-care contact serve key populations PITC in prisons and other closed General and other specific clinics institutions

Other facilities—e.g.: Stand-alone VCT For those self-identifying as at Located to serve key populations risk who find these facilities more Drop-in centres suitable than clinics

Community-based To key populations and high To key populations and remote Outreach and mobile prevalence geographic areas areas Where other HTC is lacking or Home-based door-to-door women cannot leave home To households of those known or To households of those known or Home-based index suspected to have HIV or TB suspected to have HIV or TB For general population; can reach Events men, youth (e.g. sports events) Can address general population Campaigns or specific audiences—e.g., couples, men, young people For working people and students Workplaces, schools whose hours do not allow clinic visits OST: oral substitution programmes for injecting drug users NSP: Needle substitution programmes for injecting drug users Source: WHO 2012115

2.2.9. HIV testing in Malawi

Guidelines for HTC in Malawi are laid out in the Ministry of Health’s Guidelines for

Clinical Management of HIV in Children and Adults 2010174. The Guidelines recognise that HIV services should be offered as part of the essential health package (EHP) provided through the health system. Provider initiated HTC is strongly recommended for all patients attending health facilitates, regardless of reason for attendance, and couples

72 Chapter 2 | Literature review

HTC is encouraged. The guidelines recognise that a number of other models of HTC

(including door-to-door and client initiated HTC at facilities and at stand-alone sites) remain important in offering choice to people about how they test and continue to support these approaches.

HIV testing in Malawi should be performed by Ministry of Health accredited HTC counsellors174. Testing protocols follow WHO recommendations and consist of a serial testing approach with a second confirmatory test only if the first test is positive174. In the majority of cases, HIV testing is performed using finger prick rapid testing, but other modes (oral fluid, laboratory EIA on venous blood) are acceptable where available. All individuals undergoing HTC have the demographic details and test results recorded in the standard Ministry of Health HTC register174. Quarterly aggregate data from every

HTC site is sent to the Ministry of Health for monitoring and reporting of data174.

In the Malawi Demographic and Health Survey (DHS) 2010, respondents were asked about their prior HIV testing behaviour113. In total, 73% of women and 52% of men stated that they had ever previously tested for HIV, with consistently higher rates of reported previous testing in older individuals, those from urban areas, those who were married or previously been married and those in higher wealth brackets. Further, 79% of women who had been pregnant in the previous two years reported that they had received HTC during their pregnancy.

Monitoring data from the National HIV Programme supports these self-reported data175.

There has been a rapid scale-up of HTC in Malawi and at the end of December 2012, 825 static sites and 534 outreach sites had been accredited to provide HTC175. Between

January 2008 and December 2012, a cumulative number of 4,276,462 unique Malawians

73 Chapter 2 | Literature review received HIV testing and counselling (HTC) for the first time175. In the last quarter of

2012, 33,718 /378,560 (19%) of people who received HTC through the national programme tested HIV positive175. Uptake of HTC was substantially higher among women (67% of testers) than men (33% of testers), with 48% of female testers being pregnant at the time of HTC175. Excluding pregnant women, the proportions of male

(49%) and non-pregnant female testers (51%) were similar175, suggesting gender balance in uptake.

2.3. Treatment of HIV infection

2.3.1. Antiretroviral therapy for treatment of HIV infection

Following the identification of HIV as the cause of AIDS in the early 1980s, concerted efforts were made by the international scientific community and pharmaceutical industry to develop effective and tolerable drugs to treat the infection. Understanding of the structure and replicative pathway of HIV has been critical to the development of the approximately thirty antiretroviral therapy (ART) drugs in seven classes that are currently used in clinical practice to treat HIV infection (Table 2.2).

74 Chapter 2 | Literature review

Table 2.2: Drugs approved by US Food and Drug Administration (and currently marketed) for the treatment of HIV, 2013 Year of FDA Class Generic drug name approval Nucleoside reverse transcriptase inhibitor Zidovudine 1987 (NRTIs) Didanosine 1991 Zalcitabine 1992 Stavudine 1994 Lamivudine 1995 Abacavir 1998 Emtracitabine 2003

Nucleotide reverse transcriptase inhibitors Tenofavir disoproxil fumarate 2001 (NtRIs)

Non-nucleoside reverse transcriptase Nevirapine 1996 inhibitor (NNRTIs) Delviradine 1997 Efavirenz 1998 Entravirine 2008 Rilpirivine 2011

Protease inhibitors (PIs) Saquinavir 1995 Ritonavir 1996 Indinavir 1996 Nelfinavir 1997 Amprenavir 1998 Lopinavir + ritonavir 2000 Tipranavir 2005 Darunavir 2006 Atazanavir 2003 Fosamprenavir 2003 Integrase inhibitors (INIs) Raltegravir 2007 Dolutegravir 2013

Fusion Inhibitors (FIs) Enfuvirtide 2003

Entry Inhibitors - CCR5 co-receptor Maraviroc 2007 antagonist

Fixed dose drug combinations Lamivudine + zidovudine 1997 Abacavir, lamivudine + zidovudine 2000 Abacavir + lamivudine 2004 Tenofovir + emtricitabine 2004 Tenofovir, emtircitabine + efavirenz 2006 Emtricitabine, rilpirivine + tenofovir 2011 Elvitegravir, cobicistat, emtricitabine, 2012 tenofovir disoproxil fumarate Source: US FDA. http://www.fda.gov/ForConsumers/byAudience/ForPatientAdvocates/HIVandAIDSActivities/ucm 118915.htm

75 Chapter 2 | Literature review

In 1985, the Burroughs-Wellcome Trust Company and US National Cancer Institute undertook an exploratory trial of zidovudine (AZT), a nucleoside reverse transcriptase inhibitor previously investigated in mice as a potential treatment for human leukaemia, in patients with AIDS-defining illnesses176. The treated individuals demonstrated resolution of opportunistic infections (such as fungal nail infections), a mean weight gain of 2.2kg and culture of peripheral blood monocytes became negative for HIV176. Further clinical studies of the related compounds zalcitabine and didanosine in the late 1980s demonstrated similar effects177. Following further placebo-controlled trials of AZT, demonstrating significantly improved survival advantage, as well as improvements in clinical, immunological and virological responses178, 179, the US Food and Drug

Administration grant approval to AZT for treatment of HIV.

However, once AZT monotherapy for HIV was introduced into clinical practice, a number of limitations became apparent, chief of which were the rapid accumulation of resistant viral clones177 and short-lived clinical improvements. A series of clinical trials reported in

1996 that patients taking treatment regimens containing at least three antiretroviral therapy agents (including the newly discovered protease inhibitors) had sustained virological suppression, immune reconstitution and substantial declines in AIDS events and death180, 181. ART regimens containing at least three drugs from at least two classes has since been standard practice for the treatment of HIV.

In the seventeen years that combination antiretroviral therapy (ART) has been available, numerous clinical trials and epidemiological studies have demonstrated the effectiveness of various combinations of ART treatment on clinical endpoints such as survival182, 183 incidence of opportunistic infections184 and quality of life185, as well as on

76 Chapter 2 | Literature review non-clinical markers including virological suppression183 and reconstitution of immune function as measured by CD4 count186.

In the United States, HIV-related mortality has been complied yearly by the CDC from death certificate reports. From a peak of 17 HIV-related deaths per 100,000 population in 1995, there was a sharp 3.5-fold decline in mortality rate with the introduction of ART to approximately 5 per 100,000 by 1998187. Similar dramatic declines in HIV-associated mortality were recorded in European Union counties, although varying case definitions and incomplete reporting from a number of countries hinders comparison188. The impact of the introduction of ART on mortality in the high prevalence countries of sub-Saharan will be considered in Section 2.3.2.

In industrialised countries, the introduction of ART also had impressive effects on the incidence of opportunistic infections and cancers189. Clinical and microbiological resolution of diarrhoea caused by cryptosporidiosis and microsporidiosis was correlated with reductions in HIV viral load following ART initiation190, and stabilisation of retinitis caused by cytomegalovirus allowed discontinuation of maintenance therapy191. Further, regression of Kaposi’s sarcoma lesions192 and improved radiological findings in patients with progressive multifocal leucoencephalopathy193 were associated with declines in HIV viral load. In more recent years, epidemiological studies have shown that in high income countries, ART can decrease the risk of tuberculosis disease by 54%194 and surveillance data from France and the Netherlands show rapid declines in the incidence of cryptococcal meningitis following the introduction of ART195, 196.

Sustained suppression of the circulating HIV viral load to levels that are undetectable by conventional laboratory PCR techniques (usually 50 copies/ml) is associated with fewer

77 Chapter 2 | Literature review treatment failures197, 198 and improved survival. Achievement of a sustained suppressed viral load is then a primary treatment aim of HIV therapy and individuals in whom viral suppression is not achieved should be rapidly changed to second line therapy containing three new drugs, at least one of which is from a class not previously used199, 200. Data from over 3,000 ART initiators in three European cohort studies showed that 85% of patients achieve viral suppression on first line therapy201. Modern ART regimens that contain Tenofovir disproxil fumarate (TDF), emtricitabine (FTC) and efavirenz (EFV) in a single-pill once-daily dosing schedules (Atripla®) have a significantly greater chance of achieving sustained suppression of viral load202, 203, and have fewer side effects and a lower incidence of acquired resistance204 than regimens containing more complex dosing schedules. For reasons of ease of adherence205, efficacy, cost-effectiveness and tolerability, Atripla® is now recommended as the optimal first line treatment for HIV in patients without previously ART exposure in US and UK HIV treatment guidelines199, 200.

Consistent adherence to ART, with greater than 95% of doses taken at the correct time is a key predictor of treatment success206, 207. However, considerable controversy exists about how adherence should be most reliably measured208. Current strategies used in clinical care and in research include interview to obtain patient-reported adherence, pharmacy refill monitoring, pill counts and electronic pill-boxes that record the frequency and timing of tablet removal from packets208. The AIDS Clinical Trial Group

(ACTG) has developed and validated a brief adherence questionnaire administrated by health workers that shows strong correlation with adherence measured by pill counts and electronic pill-boxes209. The ACTG Adherence Questionnaire is now widely used in research to record adherence to ART. Meta-analysis of 65 studies containing data from

15,351 patients found that patients with self-reported adherence of less than 95% had a

78 Chapter 2 | Literature review pooled odds of 2.31 for virological failure compared to patients with adherence of greater than 95%210.

Side effects of ART drugs (and particularly the multi-drug toxicities associated with combination ART) are recognised as important barriers to optimal adherence and achievement of viral suppression. Patients taking older NRTI’s such as zidovudine, stavudine and didanosine had between a 1 and 7% risk of developing pancreatitis, a 0.8 to 2.7% risk of lactic acidosis and a prevalence of lipodystrophy syndrome of between 20 to 70%211. NNRTI’s (such as nevirapine and efavirenz) can cause hypersensitivity reactions (including skin rashes [10%] and Steven-Johnson syndrome [0.1-0.3%]) and hepatitis in 2.5-11%211. Protease inhibitors are associated with hypersensitivity skin reactions (7%), diarrhoea, nausea and dyslipidaemia211, 212, increasing the risk of cardiovascular events213. Despite considerably fewer adverse drug reactions with newer fixed-drug combination regimens such as Atripla®, approximately 20% of patients will switch to an alternative regimen because of side effects (mainly EFV-associated central nervous system symptoms including dizziness)214.

The ENCORE1 Study is a non-inferiority trial comparing lower dose efavirenz (400mg once daily) to standard dose (600mg once daily) in combination with tenofavir and emtricitibine215. Analysis of the primary trial outcome found that ART regimens containing the reduced dose of EFV were non-inferior in terms of proportion of participants achieving viral load <200 copies/ml at week 42 (reduced dose=94.1%, standard dose=92.2%, difference: 1.85%, 95% CI: -2.1-5.8). Additionally, rates of side effects (predominately central nervous system) and stopping treatment were significantly lower with the reduced dosing schedule. These data suggest that a lower dose of efavirenz could be given as a part of first line ART regimens, reducing treatment

79 Chapter 2 | Literature review associated complications and improving patient acceptability without any loss of efficacy.

These findings may be of particular relevance in Malawi, which has recently (September

2012) made efavirenz-containing ART regimens routinely available through the national

ART programme. Meta-analysis shows that approximately 6% of individuals will experience efavirenz-related central nervous system side-effects216, with higher rates from studies in sub-Saharan Africa. Poor nutrition leading to lower weight and higher plasma efavirenz levels could explain higher rates of toxicity in individuals receiving efavirenz217. Presence of the CYP2B6 516G-->T(*6) variant genotype (which is more common in populations from sub-Saharan Africa compared to European populations218,

219) results in poor metabolism of efavirenz and increased plasma half-time, and has been associated with increased risk of drug toxicities219.

2.4. Assessment of eligibility for antiretroviral therapy

2.4.1. When to start antiretroviral therapy?

The availability of highly effective ART drugs means that HIV infection is now considered to be a chronic illness that can be managed by patients taking regular medication on an out-patient basis. Estimates drawn from cohorts of HIV-infected adults in a number of

European countries show that individuals initiating ART at age 20 years in 2003-05 had a mean life expectancy of a further 43 years220.

When ART first became available, clinicians and patients had to carefully consider when to initiate ART, balancing the risk of accumulation of serious side effects and limited availability of second-line and salvage therapy regimens with risk of acquiring

80 Chapter 2 | Literature review opportunistic infections. Today, a considerable body of evidence exists to guide when

ART should be initiated, with the general consensus being that early initiation of ART has important benefits, including reduced mortality, opportunistic infections, and transmission of HIV to others.

Evidence from the SMART Study suggests that prolonged periods of immunosuppression, either before ART initiation, or during interruptions of treatment, lead to uncontrolled replication of HIV and activation of inflammatory pathways, which are associated with increased risk of death221.

Prior to the publication of the landmark HPTN052 randomised controlled trial, which showed a 96% decline in risk of acquisition of HIV infection by the HIV-uninfected partner in sero-discordant couples with immediate initiation of ART60, several observational cohort studies and two randomised controlled trials had examined the prognosis of ART initiators by CD4 count strata at initiation of treatment. Meta-analysis of data from 18 cohort studies in Europe showed that the adverse effect of deferring initiation of ART on mortality and AIDS increased with decreasing CD4 count threshold, with results suggesting that a CD4 count of 350 cells/mm3 should be the minimum threshold for initiation222. Further meta-analysis of two randomised controlled trials223,

224 that compared the outcomes of individuals initiating ART at different CD4 strata found that individuals who initiated ART early (>350 cells/mm3) had a significantly lower risk of death and incidence of tuberculosis than individuals with delayed initiation (≤200 cells/ mm3)225. As older ART regimens are used in a number of countries in sub-Saharan

Africa, and clinic follow-up may be less intensive than in industrialised countries, further trials and observational studies are required to determine the optimal CD4 cut-off for initiation in these settings. Cost effectiveness analysis and mathematical modelling data

81 Chapter 2 | Literature review show that early initiation of ART at higher CD4 count thresholds would rapidly become cost effective226-228.

Based on this body of evidence, WHO issued guidelines in 2010 recommending that all

HIV-infected individuals with a CD4 count of ≤350 cells/mm3 be initiated onto ART229.

Many national HIV programmes in sub-Saharan Africa, including Malawi174, adopted these recommendations. In July 2013, WHO updated recommendations to expand ART eligibility to HIV-infected individuals with a CD4 count of ≤500 cells/mm3230.

2.4.2. Measurement of CD4 count to determine ART eligibility

The CD4 count is as a marker of immune function in HIV infection231 and is recommended by WHO as the most appropriate means of determining immunological stage of HIV disease and in ascertaining when ART should be initiated in settings with appropriate clinical laboratory infrastructure232.

Measurement of CD4 count can be particularly challenging for HIV-infected patients in sub-Saharan Africa, and often requires multiple visits to tertiary hospitals to complete measurement with substantial out-of-pocket expenditure233. A cohort study from South

Africa showed that only 31% of newly diagnosed HIV-positive individuals had completed

CD4 count measurement by 12-months234 and in Mozambique 23% had not completed

CD4 count measurement by 6-months235.

In recent years, a portable point of care CD4 count analyser (PIMA, Alere Inc. Waltham,

Maryland, USA) has been developed that can be used in primary health care centres and provide CD4 count results within 20 minutes. A number of other companies have now introduced similar systems. In a before-and-after study design field evaluation in

82 Chapter 2 | Literature review

Mozambique, the introduction of the PIMA system led to a 57% decrease in the proportion of HIV-positive patients who were lost to follow-up before ART initiation236.

Similar improvements in retention and delays to initiation were seen in studies in South

Africa237, 238.

2.4.3. Clinical staging assessments for ART eligibility

In the majority of countries in sub-Saharan Africa, CD4 count measurement using laboratory-based platforms is not feasible and point of care CD4 count measurement has not yet been introduced233. Therefore, ART eligibility cannot be determined by identifying whether HIV-infected individuals have a CD4 count of less than the WHO- or nationally-recommended CD4 count threshold. Instead, clinical staging assessments are used where assessment of physical signs and symptoms and diagnosis of opportunistic infections and cancers are used to categorise individuals to a stage that can be used to monitor disease progression and identify when ART should be initiated according to risk of illness and death. The WHO Clinical Staging and Disease Classification System (“WHO clinical staging system”) can be used in resource-limited settings and is currently widely used in most HIV programmes in sub-Saharan Africa.

The WHO clinical staging system arose out of a need for highly specific means of diagnosing AIDS in resource-limited settings in the early days of the epidemic, when rapid testing for HIV was not available. A provisional case definition was developed in

Zaire in 1986239 before being approved by WHO at a meeting in Bangui, Central African

Republic in the same year240, 241. The case definition was evaluated in hospitalised in- patients and rural community members in Zaire and in-patients and in outpatients in

Uganda, showing of sensitivity of between 50-59% and a specificity of 78-90%239, 242, 243 for diagnosis of HIV.

83 Chapter 2 | Literature review

In 1990, WHO published a proposal for a staging system for HIV that could be used to

“improve clinical management of patients; establish reliable prognoses; help in designing and evaluating drug and vaccine trials and perform studies on pathogenesis and natural history of HIV infection”244. Four stages were included in the system, comprising of: 1.

Asymptomatic/persistent generalised lymphadenopathy; 2. Early disease; 3. Moderate disease; and 4. Severe disease (AIDS). Although WHO stated in the description of the staging system that a major advantage was that classification of stage could be made using by clinical assessment facilitating use in resource-limited settings, a number of stage-defining conditions (particularly in stage 4) require laboratory diagnosis of an organism (for example: cryptosporidiosis or cytomegalovirus)245. The WHO clinical staging system was revised in 2005246 and again in 2007247. Table 2.3 shows current stages and stage-defining conditions included in the 2007 WHO clinical staging system.

84 Chapter 2 | Literature review

Table 2.3: WHO Clinical Staging System (2007) for adults with confirmed HIV infection

Stage WHO clinical stage-defining conditions 1. Asymptomatic Persistent generalized lymphadenopathy

2. Moderate unexplained weight loss (<10% of presumed or measured body weight) Recurrent respiratory tract infections (sinusitis, tonsillitis, otitis media and pharyngitis) Herpes zoster Angular cheilitis Recurrent oral ulceration Papular pruritic eruption Seborrhoeic dermatitis Fungal nail infections

3. Unexplained severe weight loss (>10% of presumed or measured body weight) Unexplained chronic diarrhoea for longer than one month Unexplained persistent fever (above 37.6°C intermittent or constant, for longer than one month) Persistent oral candidiasis Oral hairy leukoplakia Pulmonary tuberculosis (current) Severe bacterial infections (such as pneumonia, empyema, pyomyositis, bone or joint infection, meningitis or bacteraemia) Acute necrotizing ulcerative stomatitis, gingivitis or periodontitis Unexplained anaemia (<8 g/dl), neutropaenia (<0.5 × 109 per litre) or chronic thrombocytopaenia (<50 × 109 per litre)

4. HIV wasting syndrome Pneumocystis pneumonia Recurrent severe bacterial pneumonia Chronic herpes simplex infection (orolabial, genital or anorectal of more than one month’s duration or visceral at any site) Oesophageal candidiasis (or candidiasis of trachea, bronchi or lungs) Extrapulmonary tuberculosis Kaposi’s sarcoma Cytomegalovirus infection (retinitis or infection of other organs) Central nervous system toxoplasmosis HIV encephalopathy Extrapulmonary cryptococcosis including meningitis Disseminated non-tuberculous mycobacterial infection Progressive multifocal leukoencephalopathy Chronic cryptosporidiosis (with diarrhoea) Chronic isosporiasis Disseminated mycosis (coccidiomycosis or histoplasmosis) Recurrent non-typhoidal Salmonella bacteraemia Lymphoma (cerebral or B-cell non-Hodgkin) or other solid HIV-associated tumours Invasive cervical carcinoma Atypical disseminated leishmaniasis Symptomatic HIV-associated nephropathy or symptomatic HIV-associated cardiomyopathy

85 Chapter 2 | Literature review

In a number of prospective studies both in developed countries and in sub-Saharan

Africa, the prognostic value of the WHO clinical staging system has been clearly demonstrated. In pre-ART era studies from Uganda248, 249, there was a clear relationship between advanced WHO clinical stage and risk of death, with 73% of individuals in stage

1 alive at 6-years compared to 6% of individuals in stage 4. WHO clinical stage is also a good predictor of prognosis following initiation of ART, with higher rates of mortality and lost and loss-to-follow-up in individuals in more advanced stages250, 251.

Because of the limited capacity for assessment of immune status by measurement of

CD4 count in much of sub-Saharan Africa, most national programmes recommend that

WHO clinical stage be used for screening for eligibility for ART. For example, in Malawi,

HIV-infected patients in WHO stage 3 or 4 are eligible for ART initiation174. WHO also recommends that, where CD4 count measurement is not possible, HIV-positive individuals in WHO stage 3 or 4 should be initiated on to ART229.

However there are concerns that WHO clinical staging has low sensitivity compared to

CD4 count when used as an eligibility assessment tool, potentially resulting in delayed initiation of ART until immunosuppression is advanced252. Tables 2.4 and 2.5 summarise studies identified in a systematic literature review (conducted by Chigomezgo Munthali,

MPH student at LSTM, and myself) that have reported on the diagnostic performance of

WHO clinical stage 3 or 4 disease in identifying ART eligibility thresholds of CD4 count

≤200 cells/mm3, ≤350 cells/mm3 and ≤500 cells/mm3. Weighted summary estimates were obtained by fitting a two-level mixed logistic regression model, with independent binomial distributions for the true positives and true negatives conditional on the sensitivity and specificity in each study, and a bivariate normal model for the logit transforms of sensitivity and specificity between studies using a method described by

86 Chapter 2 | Literature review

Harbord et al253. Hierarchical summary receiver operator characteristic (HSROC) curves were plotted (Figures 2.11 and 2.12).

Table 2.4: Summary estimates and meta-analysis of studies comparing WHO clinical stage 3/4 with CD4 ≤200 cells/mm3

Cadre of HIV-positive healthworker Sensitivity Specificity Study Country N population performing (95% CI) (95% CI) WHO staging Boniphace Hospital 72% 57% Tanzania NR 331 2011254 outpatients (65%-78%) (49%-67%) Cameroon, Cote d’Ivoire, Kenya, Mozambique, Antenatal 27% 93% Carter 2010255 Rwanda, South clinic Clinicians 6036 (24%-29%) (92%-94%) Africa, Uganda, attendees Zambia, and Thailand Hospital 56% 62% Fox 2010256 South Africa NR 802 outpatients (52%-61%) (56%-67%) French Hospital 71% 67% Uganda NR 210 1999257 outpatients (59%-80%) (58%-76%) Inpatients 89% 54% Ilovi 2011258 Kenya and NR 152 (81%-95%) (41%-66%) outpatients Primary clinic Clinicians 52% 68% Jaffar 2008259 Uganda 4302 attendees (Doctors) (50%-54%) (66%-70%) Kagaayi Community 51% 88% Uganda Clinicians 1221 2007260 members (46-56%) (85%-90%) Clinicians McGrath Community 20% 94% Malawi (Clinical 120 2007261 members (7%-41%) (87%-98%) Officer) Morpeth Hospital 75% 35% Tanzania NR 202 2007262 outpatients (66%-83%) (26%-46%) Pre-ART clinic 83% 52% Tassie 2004263 Malawi NR 206 attendees (74%-90%) (42%-62%) Torpey Pre-ART clinic 70% 59% Ghana Clinicians 5784 2009264 attendees (69-72%) (57%-61%) Weighted 62% 70% 20,197 Summary (48%-74%) (56%-81%) Cochran’s Q statistic for heterogeneity – sensitivity: Q=914.26, p<0.001; specificity Q=1439.43, p<0.001 NR: not reported Clinician may mean either a doctor, a nurse or clinical officer. Where subcategory not specified, healthworkers from more than one cadre performed staging

87 Chapter 2 | Literature review

Table 2.5: Summary estimates and meta-analysis of studies comparing WHO clinical stage 3/4 with CD4≤350 cells/mm3

Cadre of HIV-positive healthworker Sensitivity Specificity Study Country N population performing (95% CI) (95% CI) WHO staging Baveewo Pre-ART clinic 49% 87% Uganda Clinicians 395 2011265 attendees (43%-55%) (79%-92%) Cameroon, Cote d’Ivoire, Kenya, Mozambique, Ante-natal 255 18% 95% Carter 2010 Rwanda, South clinic Clinicians 6036 (17%-20%) (94%-95%) Africa, Uganda, attendees Zambia, and Thailand Inpatients 82% 63% Ilovi 2011258 Kenya and NR 152 (73%-88%) (46%-78%) outpatients Pre-ART clinic Clinicians 48% 74% Jaffar 2008259 Uganda 4302 attendees (Doctors) (46%-50%) (72%-77%) 100% McGrath Community Clinical 22% Malawi 120 (95%- 2007261 members Officers (11%-35%) 100%) Torpey Pre-ART clinic 65% 71% Ghana Clinicians 5784 2009264 attendees (64%-67%) (68%-74%) Weighted 45% 85% 16,789 Summary (26%-66%) (69-93%) Cochran’s Q statistic for heterogeneity – sensitivity: Q=1607.31, p<0.001; specificity Q=896.70, p<0.001 NR: not reported Clinician may mean either a doctor, a nurse or clinical officer. Where subcategory not specified, healthworkers from more than one cadre in this cadre performed staging

88 Chapter 2 | Literature review

Figure 2.10: Hierarchical summary receiver operator characteristic (HSROC) plot for accuracy of WHO stage 3/4 disease in identifying CD4≤200 cells/mm3

Figure 2.11: Hierarchical summary receiver operator characteristic (HSROC) plots for accuracy of WHO clinical stage 3/4 in identifying CD4≤350 cells/mm3

89 Chapter 2 | Literature review

A considerable degree of heterogeneity was found between studies and so summary estimates should be treated with caution. At the ART eligibility threshold of CD4 ≤200 cells/mm3, the sensitivity and specificity of WHO clinical 3 or 4 disease were 60% (95%

CI: 45%-73%) and 73% (95% CI: 60%-83%) respectively. Analysis at the higher ART eligibility threshold of CD4 ≤350 cells/mm3 the sensitivity and specificity were 45% (95%

CI: 26%-66%) and 85% (95% CI: 69%-93%) respectively. Only one study was identified that examined the ART eligibility threshold of CD4 ≤500cells/mm3, which found a sensitivity and specificity of 14% (95% CI: 13%-15%) and 95% (95% CI: 94%-96%) respectively.

These analyses show that a substantial number of patients in need of ART are likely to be missed when WHO clinical staging alone is used as an ART eligibility assessment tool, with worsening performance as CD4 count thresholds are increased. Until point of care

CD4 count measurement is readily available in remote and rural sites, and at primary health care centres and antenatal clinics, alternative, simplified and accurate ART eligibility assessment tools are required.

2.4.4. Alternative approaches to assessing ART eligibility

A number of alternative ART eligibility assessment tools have been studied to attempt to overcome the low sensitivity of WHO clinical staging and the lack of widely available access to CD4 count measurement. In Uganda, Miiro et al examined the performance of low body mass index (BMI <18.5 kg/m2) and anaemia (haemoglobin of <11 g/dl and <12 g/dl in women and men respectively) in identifying CD4 count of ≤200 cells/mm3 266. Low

BMI had a sensitivity of 23% and a specificity of 89%, and anaemia had a sensitivity of

47% and a specificity of 76%. Given the progressive natural history of HIV infection, the sensitivity of both measurement of BMI and anaemia of can be expected to decrease as

90 Chapter 2 | Literature review the CD4 count threshold is raised to ≤500 cells/mm3. Measurement of haemoglobin does not overcome the need for availability of either point of care measurement equipment or a centralised laboratory system.

In Karonga, rural Malawi, a simplified check-list version of the WHO clinical staging system consisting of 19 items was administered to randomly sampled community- members by a nurse as part of a demographic and health survey261. Using independent

WHO clinical staging assessment as stage 3 or 4 performed by nurse as the gold standard, the field check list had a sensitivity of 38% using CD4≤250 cells/mm3 as the gold standard (ART eligibility criteria in in Malawi at the time). The authors concluded that the field check list was not an appropriate ART eligibility screening tool as reliance on it would have missed two-thirds of HIV-positive patients eligible for ART under guidelines at the time.

One simple approach that has not been evaluated as an ART eligibility assessment tool is patient’s self-rated general health. Self-rated general health belongs to a family of self- rated morbidity measures267 and involves recording individuals’ response to the question “How would you rate your general health?” on a five-item Likert scale

(“Excellent”, “very good”, “good”, “fair”, “poor”)268-272. Variations in the wording of the question and in the number of response-items have been used273. Conceptually, assessment of self-rated health allows patients to give an “internal” view of their health status based on their own perceptions, whereas purely external assessments based on the observations of clinicians (such as WHO clinic staging assessments) may be inadequately responsive to patient’s experiences of their illness274. The self-rated general health question has been shown to have satisfactory reliability and validity in

Tanzania when compared to more detailed quality of life questionnaires275.

91 Chapter 2 | Literature review

In a meta-analysis of longitudinal population studies that included 779,266 participants from 9 countries and where 51,368 deaths occurred, there was a statistically significant relationship between increasing risk of death and worse self-rated general health: individuals with poor, fair and good self-rated general health had an adjusted rate ratios of death of 1.92 (95% CI: 1.64-2.25), 1.44 (95% CI: 1.21-1.72) and 1.23 (95% CI: 1.09-

1.39) respectively compared to individuals who self-rated their health as excellent276.

In participants of three cohort studies of HIV-infected adults in the United States, worse self-rated general health was independently associated with age, lower CD4 counts and

ART treatment failure277. In a separate study, individuals with worse self-rated general health had poorer adherence to ART278. Few studies have investigated the predictive value of self-rated general health on health-related outcomes among people living with

HIV in sub-Saharan Africa. In a study of factory workers in Harare, Zimbabwe, HIV- positive individuals had significantly worse self-rated general health than HIV-negative individuals279. No association was found between self-rated general health and likelihood of successfully collection of CD4 count results in a prospective cohort study in

South Africa280.

In summary, self-rated general health offers a potentially simple means of assessing ART eligibility. Studies to assess the diagnostic accuracy of this scale against both the WHO clinical staging system and the current WHO CD4 cut-off are required.

92 Chapter 2 | Literature review

2.5. The public health approach to scale-up of ART delivery in low income countries with generalised HIV epidemics

In high income countries, people living with HIV are almost universally treated by expert

HIV clinicians with ART regimens specifically tailored to their prior drug exposures199, 200, likely resistance patterns, and even to genotype to avoid the occurrence of hypersensitivity reactions281. ART drugs are almost always given in fixed dose combinations to improve adherence: the majority of newly diagnosed HIV-positive individuals are initiated onto Atripla®, a single pill that contains three antiretroviral drugs and is required to be taken once daily199, 200. A wide range of regimens are available from all ARV classes should a change of first line therapy be required. Intensive monitoring and follow-up is undertaken with frequent measurement of immunological (CD4 count) and virological markers (HIV viral load) used to aid clinicians in guiding treatment decisions282. Additionally, clinicians are able to draw upon a wide network of health professionals to support treatment and manage complications, including clinical psychologists, social workers and occupational therapists. In short, patients diagnosed with HIV in high-income countries receive individualised care.

In contrast, in the low-income countries of sub-Saharan Africa that have the highest prevalence of HIV, chronic shortages of health care providers and major limitations in available resources for health care have meant that adoption of the individualised approach to ART delivery has not been feasible283.

Prior to 2003, when less than 2% of African HIV-infected individuals had initiated ART284, there was considerable opposition to expansion of ART programmes in sub-Saharan

Africa285. Whilst recognising that devastatingly high rates of mortality and mobility were attributable to HIV in these countries, a number of key policymakers and politicians from

93 Chapter 2 | Literature review developed countries raised concerns that increased delivery of ART would be unaffordable and unsustainable286. Moreover, they argued that the limited clinical supervision should ART be introduced could lead to widespread failures of adherence, leading to rapidly escalating resistance: so-called “ART anarchy”287.

However, at the 13th International AIDS Conference held in Durban, South Africa – the first conference to be held in a country with a generalised HIV epidemic – activists (most notably from the Treatment Action Campaign), scientists and African leaders highlighted the devastating effects of HIV on individuals and societies in sub-Saharan Africa and called for urgent action to increase availability of ART in the countries most affected by the epidemic288. In 2001, the Global Fund to Fight AIDS, Tuberculosis and Malaria was established, the first multinational organisation fund to dedicate significant financial resources through public-private partners to countries wishing to scale-up delivery of

ART289. This was followed in 2003 by the launch of The U.S. President’s Emergency Plan for AIDS Relief (PEPFAR) which committed US$18.8 billion over five years, initially to be to be used to support the establishment of infrastructure and purchase of ART drugs to treat 2 million people in 15 high-focus countries by 2008290. Concurrently, the World

Bank committed US$500 million to HIV-treatment and prevention programmes through its Multi-Country AIDS Programme291.

In 2003, WHO launched the “3 by 5” initiative, with the target of providing 3 million people living in low- and middle-income countries with ART by 2005292. WHO requested all member states to commit to ten guiding principles, shown in Table 2.6292.

94 Chapter 2 | Literature review

Table 2.6: Ten guiding principles of the WHO "3 by 5" initiative

10 Guiding principles of the “3 by 5” initiative

Urgency

The centrality of people living with HIV

Life-long care

Country ownership

Human rights

Partnership and plurality

Complementarity

Learning, innovation and sharing

Ethical standards

Equity

Accountability Source: WHO 2003292

This “3 by 5” initiative was supported by global advocacy and legal action to reduce the price of ART drugs, cumulating in a number of pharmaceutical countries based in India,

Thailand, Brazil and South Africa being permitted to produce and sell ART drugs at 5% of the market value at which they were being sold in the USA, making ART affordable at approximately US$50 per person per year285.

WHO recognized that because of resource limitations, scaling-up of ART could not be achieved in low-income countries using the individualized model of care practiced in high-income, countries. Instead, building upon the DOTS model for TB control293 and treatment and heavily influence by the establishment and rapid scale up of a successful

ART programme in Malawi294, WHO recommended a public health approach to ART delivery. The components of the public health approach to ART delivery include284: standardized ART treatment regimens and formularies with a limited number of drugs; simplified clinical decision-making with a very limited number of treatment-related

95 Chapter 2 | Literature review actions (when to start, substitution for toxicity, switching treatment for failure, and when to stop for end-of life care); standardized monitoring with a limited number of tools (CD4 count if available prior to ART initiation and regular WHO clinical staging); district- and national-level reporting systems; decentralized delivery of care from hospitals to primary health care centres; and task shifting from clinicians to nurses and clinical officers. The public health approach to ART delivery results in vertically implemented programmes, where funding, management and clinical care are provided as stand-alone activities and with little integration with other existing clinic activities and health programmes such as tuberculosis care, maternal and child health and general outpatient care295. Considerable debate is ongoing over whether the benefits of this vertical structure (greater expertise, improved outcomes, simplified logistics, better accountability, ease of monitoring and evaluation) outweigh the potential detriments

(poor linkage between TB and HIV care, an emphasis on HIV care at the expense of other important health issues, reduced chances of sustainability, responsive to diseases and not to patients)295. In short, current consensus seems to show that when health systems and clinic structures are weak, vertically implemented HIV care programmes have a greater likelihood of success than programmes where HIV care is integrated with other clinic activities.

The public health approach to ART delivery, implemented in most of the low- and middle-income countries of sub-Saharan Africa, has been tremendously successful in achieving rapid scale-up of ART delivery. Between 2004 and 2011, there was a 20-fold increase in the number of HIV-infected people in low-and middle-income countries receiving ART, with the most rapid gains seen between 2010 and 2011, when the proportion of HIV infected individuals receiving ART increased by 21%. Three quarters of the 9.7 million people who have initiated ART globally come from sub-Saharan Africa3.

96 Chapter 2 | Literature review

Outcomes for individuals who initiate ART in public health treatment programmes in sub-Saharan Africa have also been impressive. Since 2003, a large number of cohort studies and evaluations of routinely-collected programmatic data from countries in sub-

Saharan Africa have reported on the proportion of ART initiators who are retained on treatment, who die following initiation and who achieve immune reconstitution and virological suppression. Standardisation of reporting of these outcomes has meant that data can be compared between programmes and countries. A meta-analysis of 33 patient cohorts containing 74,289 ART initiators in 13 countries found that a weighted mean of 80%, 75% and 62% of ART initiators were retained in care at 6-months, 12- months and 24-months following initiation respectively, with drop-out from care and mortality the largest contributors to attrition251. Approximately 40% of patients who drop-out of ART care are found to have died296. Interventions such as having nominated treatment supporters297, and shifting care of stable ART patients to nurses298, 299, have been shown to be effective in reducing drop-out.

A separate meta-analysis of 18 cohort studies containing data on almost 40,000 ART initiators in 2008 found that mortality at 12-months following ART initiation ranged between 8% and 26%, with mortality particularly high in the immediate post-ART initiation period and strongly associated with advanced immunosuppression at treatment initiation250. Further analysis of pooled cohort data shows that late entry to

HIV care remains a significant problem, with mean CD4 counts at initiation of ART in sub-

Saharan African programmes ranging between 43 and 147 cells/mm3 300, considerably lower than found in developed country settings. For example, in the Swiss HIV cohort, mean initiation CD4 count was 240 cells/mm3 301.

97 Chapter 2 | Literature review

Adherence to ART has also been well documented in a number of cohorts of ART initiators in sub-Saharan Africa. The majority of studies have used one or more of a small number of validated tools for assessing adherence, including the ACTG Adherence

Questionnaire (based on self-reported adherence)209, pill counts, estimation from pharmacy refill slips and electronic pill boxes that tally when opened208. Meta-analysis of

28 studies from North American and 27 studies from sub-Saharan Africa showed a pooled mean estimate of 55% adherence among North Americans compared to 77% in

Africans302. Long term follow-up of ART initiators in South Africa has demonstrated that patients with good early adherence were significantly more likely to achieve sustained virological suppression than those with poor initial adherence303. A number of interventions have been shown to improve adherence to ART in low-resource settings, including patient support and education304, counselling interventions305 and mobile phone text message reminders306.

Fewer studies have reported on immunological and virological outcomes of ART initiators in sub-Saharan Africa, predominantly because of expense and lack of laboratory capacity to measure CD4 count and HIV viral load in most national programmes. In a large cohort study from Cape Town, South Africa, where the median

CD4 count at ART initiation was 101 cells/mm3, impressive rates of immune reconstitution were achieved, with less than 10% of patients having a CD4 count of less than 200 cells/mm3 by 24 months307, comparing favourably to studies in resource-rich settings308. Similar results have been found in studies from Uganda309, 310,

Mozambique311 and Botswana312. Meta-analysis shows that initiation of ART at higher

CD4 counts in low-resource settings results in a lower incidence of new opportunistic infections225, predominately as a result of less person-time exposed to CD4 counts of less than 200 cells/mm3.

98 Chapter 2 | Literature review

The key predictor of treatment success with ART is achievement of virological suppression, defined by the US Department of Health and Human Services as a confirmed HIV RNA viral load of below the limit of detection (usually <50 copies/ml)313.

Virological failure is defined as failure to achieve or sustain a HIV viral load below 200 copies/ml. Because of continued improvements in lowering the limit of detection of HIV

RNA, clinical trials and cohort studies reporting on virological outcomes of ART treated patients in sub-Saharan Africa have used varying definitions for virological suppression and failure. Two systematic review and meta-analyses of studies from sub-Saharan

African countries found that between 69% and 75% of ART initiators retained in care at

24 weeks achieved virological suppression314, 315. A comparison of ART initiators from urban slums in Cape Town and Switzerland found similar rates of virological suppression but higher mortality in Cape Town301, emphasising the substantial contribution of late

ART initiation with advanced immunosuppression to adverse prognosis. A separate meta-analysis found that 89% of patients with virological failure had acquired resistance to one or more antiretroviral drugs316.

In summary, the public health approach to ART delivery has achieved impressive rapid scale-up of ART delivery in most countries severely affected by the HIV pandemic and in the face of substantial limitation in available health sector resources. Adherence, immunological and virological outcomes appear to be as good, or better, than those achieved in patients treated under the individualised care model in the developed world. However, substantial concerns remain about the high proportion of individuals initiating ART with advanced immunosuppression and the high rate of immediate post-

ART initiation mortality and patient loss from care.

99 Chapter 2 | Literature review

2.6. The National ART programme in Malawi

The Malawian National HIV care programme has been held-up as a success story in sub-

Saharan Africa, being one of the first programmes to achieve rapid scale-up of treatment delivery in the face of extremely limited resources5, 283, 294. With only 3,998 healthcare workers (doctors, clinical officers and nurses) to serve a population of approximately 13 million in 2006317, Malawi pioneered the public health approach to ART delivery, modelled on the principles of the well-performing national TB programme294.

Malawian national guidelines for ART have evolved in concert with WHO recommendations (Figure 2.13). From commencement of the national programme in

2003 until 2006, HIV-infected adults (16 years and older) who were in WHO stage 3 or 4, had a CD4 count of ≤200 cell/mm3 or a total lymphocyte count of <1200/mm3 were eligible for ART initiation318. From 2006 to 2008, the CD4 count criterion for adults was raised to ≤250 cells/mm3. From 2008 to September 2011 (the period during which the cohort study described in Chapter 4 was carried out), the WHO stage 3 or 3 CD4 count

≤250 cells/mm3 criteria were kept in place, but the total lymphocyte count criterion was removed. From September 2011, the CD4 count eligibility threshold was raised to ≤350 cells/mm3 and pregnant and breastfeeding women became universally eligible for ART

(“Option B+”).

100 Chapter 2 | Literature review

Figure 2.12: Evolution of ART eligibility criteria for adults in children in Malawian National programme

2003 2006 2008 2011

Adults & children >5 Adults >15 years: Adults >15 years: years CD4 <200 cells/mm3 or CD4 <250 cells/mm3 or 3 WHO 3/4 or TLC <1200/ WHO 3/4 or TLC <1200/ CD4<350 cells/mm or mm3 mm3 Children <12 months: WHO 3/4 or pregnant/ breaseeding Children >18 months: Children >18 months: Universally eligible Children 2-5 years: CD4<15% or WHO stage CD4<15% or WHO stage irrepsecve of CD4 3/4 3/4 count/WHO stage CD4<25% or CD4<750 cells/mm3 or WHO 3/4 Children <18 months: Children <18 months: Children <24 months CD4<20% or WHO 3/4 CD4<25% or WHO 3/4 Universally elgibile

TLC: Total lymphocyte count Source: Sloan et al, BMC Public Health318

By the end of the first quarter of 2012, 810 health facilities in Malawi had been approved as sites for provision of HIV care, including ART319. Fifty-four (7%) HIV care clinics had a functioning CD4 count machine on site. Since the beginning of the scale up in 2003, 580,409 people have initiated ART in Malawi, 37% of whom were male and 63% of whom were female319. Children made up 9% of ART initiators. Until updated guidelines were introduced in September 2011, the Malawian National ART programme had two ART regimens (first line – consisting of D4T, 3TC and NVP – and second-line), with individual substitutions available in the case of toxicities. In the 2011 guidelines

101 Chapter 2 | Literature review individuals who are pregnant (“Option B+”), breastfeeding or being treated for TB have access to a tenofovir-containing first-line regimen (TDF, 3TC and EFV)174.

A large increase in the number of people initiated onto ART was seen in Q3 and Q4 of

2011319 (Figure 2.14), coinciding with the expanded CD4 criteria and introduction of

“Option B+” in Malawi174, 320, 321. The introduction of Option B+ has been hailed as a success by some researchers, policy-makers and international organisations, recognising the potential for large maternal health gains and reductions in mother to child transmission of HIV322. However, concerning national-level data showing high rates of loss to follow-up at 6-months of pregnant women initiated onto ART through the Option

B+ programme have been recently published323. Seventeen per cent of 21,939 pregnant women initiated onto ART were lost to follow-up, with the highest risk of loss occurring during the first 3 months after ART initiation. Moreover, pregnant women who initiated

ART through the Option B+ programme were 5 times more likely to be loss to follow-up than non-pregnant women initiated onto ART with CD4 count of <350 cells/mm3 or in

WHO clinical stage 3 or 4.

There are a number of potential reasons for these concerning findings, all of which require further investigation. Firstly, under the Option B+ approach, ART may be initiated from within the antenatal clinic. Antenatal clinic staff may have little experience or training with supporting patients’ ART initiation and treatment. Secondly, after delivery, women will transfer into the routine HIV care system. It is possible that systems for linkage into care between clinics are weak, making it difficult to access continuation treatment. Thirdly, pregnant, women may appreciate the value of taking ART to prevent

HIV transmission of HIV to their baby. However, they may not perceive themselves to be ill, and may mistakenly believe that ART can be discontinued following delivery.

102 Chapter 2 | Literature review

Fourthly, following delivery, pregnant women may move to reside with family members in other Districts and not successfully transfer care to an alternative HIV care clinic.

Fifthly, there was considerable variation between clinics with ART initiators at larger clinics faring worst, suggesting that some facility HIV care recording and reporting systems may be suboptimal. Finally mortality rates among individuals lost to follow up from ART programmes are known to be high324. It is possible that the high rate of loss to follow-up among pregnant women reflects the high maternal mortality rates in Malawi.

A total 347,983 ART initiators were alive on treatment on the 31st of March 2012, meaning that overall retention on ART since programme commencement was 73%319. A further 50,503 (11%) had died, 77,645 (16%) were lost to follow-up and 1699 <1%) had stopped ART319. Rates of retention on ART have improved consecutively, quarter-on- quarter between 2004 and 2012319. Additionally, the proportion of ART initiators in WHO stage 3 and 4 have declined from 24% in 2004 to 6% in 2012, with early ART mortality rates contemporaneously declining319. However, the 12-month survival rate (proportion of ART initiators alive and taking treatment) was 80% for adults initiating ART in 2012319, below WHO’s target of 85%.

103 Chapter 2 | Literature review

Figure 2.13: Number of ART initiators by quarter in the Malawian National Programme 2004, Q2 to 2012, Q4

3 40000 CD4≤350 cells/mm Option B+ 35000

30000

25000

20000

ART iniaons 15000

10000

5000

0

Source: Integrated HIV Program Report January – March 2012319

2.7. Linkage between HIV diagnosis and initiation of ART

2.7.1. The HIV care cascade

For individuals who are infected with HIV, a number of care-seeking steps are required to be undertaken before ART treatment can be initiated. This pathway is known as the

HIV care cascade (Figure 2.15) and is based upon a model of care-seeking developed by

Piot in the 1970s which described the various diagnostic and treatment states individuals infected with TB could inhabit (undiagnosed, diagnosed but untreated, partially treated, fully treated)325.

Conceptualisation of the HIV care cascade has been hailed as an important development in understanding the performance of HIV care programmes326, as it allows programme

104 Chapter 2 | Literature review managers and researchers to identify weaknesses that may be hindering opportunities to achieve universal access to ART, and achieve maximal impact upon mortality and HIV transmission327. In progressing through the HIV care cascade, HIV-infected individuals may fail to complete HTC (or receive a false negative result), fail to complete ART eligibility assessment (CD4 count or WHO clinical staging assessment), fail to remain in care until ART is initiated, or drop out of care after ART has been initiated327. Important to note is the fact that individuals who drop out of the HIV care cascade at any point may “recycle” back into the pathway at a later date327. Additionally a number of studies have emphasised the high rate of mortality found amongst individuals who drop out of the HIV care cascade, with one meta-analysis of 10 studies estimating that 10.8% of ART eligible individuals who drop out of care are found to have died, with mortality rates ranging between 1.6 to 53.2 per 100 person-years328.

105 Chapter 2 | Literature review

Figure 2.14: The HIV care cascade

106 Chapter 2 | Literature review

2.7.2. Defining linkage into care

Studies summarising and evaluating patient flow throughout the HIV care pathway have been hampered by lack of clear definitions. Three systematic reviews of patient outcomes in the period between HIV testing and initiation of ART have been published328-330. However, all three systematic reviews use differing definitions of linkage outcomes, reflecting the varied definitions used in the studies included within the reviews.

In the systematic review conducted by Rosen et al330, the authors’ objective was to examine “retention in care between HIV testing and treatment”. Three stages of care were identified, with “linkage” defined by the completion of each stage. Thus linkage into care was assessed as: 1) the proportion of HIV-positive study participants who were

“staged” (in fact, defined by patient referral by a healthworker for “further HIV care”

[not defined in review] following completion of CD4 count measurement); 2) the proportion of ART-ineligible (on healthworker assessment of CD4 count result) HIV- positive individuals enrolled in “pre-ART care” who were subsequently determined to by

ART eligible (by healthworker assessment of a subsequent CD4 count); and 3) the proportion of individuals assessed as ART eligible on healthworker review of CD4 count who subsequently initiated ART.

Although the definitions of stages and of linkage between stages is clear, as defined, this model may be insufficiently detailed to capture all aspects of the care-seeking process for individuals in the HIV care pathway. In particular, in many countries where CD4 count is not available or is not widely available (such as Malawi), the WHO clinical staging system is primarily used to define ART eligibility. Additionally, individuals in WHO stage 3 or 4 are eligible for ART initiation under WHO guidelines331. Thus, individuals may be ART

107 Chapter 2 | Literature review eligible without having ever completed a CD4 count measurement. This means that, as defined in this systematic review, these individuals would not have completed stage 1, and would not contribute to the starting population for stage 2. Given that there are a number of studies included from countries without widespread access to CD4 count testing (e.g. Malawi) in this systematic review, it is not clear how these data were handled in estimating rates of linkage and loss to care.

In the systematic review conducted by Mugglin et al328, the authors’ objective was to determine the extent of patient loss to programme between diagnosis of HIV and initiation of ART. Data are presented against three “linkage” periods, although clear and detailed descriptions of these periods are not provided. Stage 1 is defined as “from HIV testing to CD4 count measurement”; Stage 2 is defined as “CD4 testing to ART eligibility”; and Stage 3 is defined as “ART eligibility to ART start”.

These definitions are insufficiently precise to allow meaningful comparison across studies. For example, “CD4 testing” could mean: referral by a healthworker for CD4 count measurement; attendance to have blood taken; completion of blood draw; collection of CD4 count results; return to the health facility to discuss result with a healthworker; or a healthworker referral for further care on the basis of CD4 count result. With such imprecisely defined outcomes, the proportions of individuals completing each stage (or linking between steps) could vary considerably depending on how individual studies included within the review classify outcomes and how the reviewers handle data. Additionally, as in the review by Rosen et al it is not clear how individuals classified as ART eligible by WHO clinical staging (but who may not have had

CD4 count measured) are handled when determining rates of linkage.

108 Chapter 2 | Literature review

In the systematic review conducted by Kranzer et al332, five stages of care are identified: the HIV testing period (although no studies were identified for this period and instead the authors present the proportion of individuals within various populations who are aware of their HIV status, estimated from population surveys); the ART eligibility assessment period; the period between identification as non-ART eligibility and subsequent ART eligibility; the period between ART eligibility and initiation of ART; and following ART initiation. The authors clearly highlight some limitations of these definitions, noting that studies included in the review use differing definitions of

“assessment for ART eligibility”, “ART eligible” and “pre ART care”. They pragmatically use available study definitions to derive summary estimates for the proportion of study participants completing each stage.

Given that there are no published standard definitions for “linkage into HIV care” relevant to public health HIV care programmes in low resources settings (nor for some composites parts of the HIV care pathway), for the studies conducted in this thesis, I have developed a set of definitions against which patient outcomes along the HIV care pathway can by assessed:

• Stage 1: From attendance for HIV testing to receipt of HIV-positive test result

• Stage 2: From receipt of HIV positive test result to completion of ART eligibility

assessment (defined according to National HIV guidelines at the time of the

study: i.e. either classified as being in WHO stage 3 or 4 by a healthworker, or

completion CD4 count measurement [reassessment by healthworker with CD4

count result and having CD4 count <250 cells/mm3 [for the cohort study

described in Chapter 4])

109 Chapter 2 | Literature review

• Stage 3: From classification as ART ineligible (i.e. WHO stage 1 or 2, or CD4 count

>250 cells/mm3 in the cohort study described in Chapter 4) to a subsequent

healthworker assessment determining that they are ART eligible (as defined in

Stage 2)

• Stage 4: From ART eligibility (as defined in Stage 2 above) to initiation of ART

• Stage 5: Continued retention in ART care following initiation

In the literature review and in the prospective cohort study described in Chapter 4, the proportions of participants successfully completing the first 4 of these stages of the HIV care pathway are described (as the purpose of the thesis was to investigate patient outcomes between HIV testing and ART initiation).

Recognising that maximal individual and public health benefit is gained from the greatest proportion of the HIV-infected population completing all stages of the HIV care pathway and continuing to take ART, in description of results of the cluster-randomised trial in Chapter 6 and in discussion of public health and policy implications of thesis findings (Chapter 8), “linkage into HIV care” is defined by an HIV-positive individual completing all five stages of the HIV care pathway.

There are a number of limitations to these definitions. Firstly varying definitions of national ART eligibility criteria and available HIV care services between countries and over times means that no one definition will comprehensively address all facets of the

HIV care pathway for a particular time-period or country. However, I have intentionally developed definitions that will be relevant for the national HIV care programme in

Malawi (where studies were conducted) and that that will be sufficiently generalizable to allow comparison between different countries and periods. Secondly, within these

110 Chapter 2 | Literature review definitions it is extremely difficult to account for individuals who recycle into the HIV care pathway (i.e. who drop-out of care at one point, but subsequently re-enter care at a later date – shown in Figure 2.15 above). This is an issue that numerous investigators have struggled to address332, without any clear resolution. When discussing study findings relating to linkage into care or policy and practice implications, I have highlighted this limitation.

2.7.3. Linkage and retention between HIV diagnosis and initiation of ART

Patient loss from the HIV care cascade in national HIV care programmes following the public health approach to ART scale up was first identified as a potential major problem in a 2005 study from Cape Town, South Africa333. In this study, 19% of HIV-positive individuals registered with an HIV care clinic did not initiate ART, with high rates of pre- treatment mortality (35.6 deaths/100 person-years; 95% CI, 25.5– 43.0 deaths/100 person-years).

Between 2005 and 2011 (when data collection for this thesis commenced), a total of 51 studies234, 235, 255, 332-379 providing data on patient loss between HTC and ART initiation were published from 17 countries in sub-Saharan Africa (20 studies from South Africa, 8 from Uganda, 7 from Kenya, 5 from Malawi, 2 from Ethiopia, 3 multicounty studies and one each from Cameroon, Rwanda, Mozambique, Tanzania, Swaziland and the Gambia) -

Table 2.7.

111 Chapter 2 | Literature review

Table 2.7: Studies reporting on pre-treatment loss to care in the HIV care cascade in sub-Saharan Africa, 2006-2011

Stage of Year Study Time-delineated cascade Study Country Population published period study endpoint reported on HIV-positive adults enrolled 2002- Lawn333 2005 South Africa ND 3, 4 in a urban HIV 2005 care clinic Newly diagnosed HIV 2002- Lawn351 2006 South Africa ND 2, 3, 4 care clinic 2005 attendees Cryptococcal 2004- Parks363 2006 Uganda meningitis ND 2, 3, 4 2006 prevention trial Proportion Hospital registration at Wazenyze369 2006 Uganda inpatients 2004 ART clinic within 2, 3, 4 offered HTC 18-months of diagnosis TB patients in 2003- Zachariah372 2006 Malawi one rural ND 1, 2, 4 2004 district ART eligible HIV-infected Feucht340 2007 South Africa children 2004 ND 3, 4 attending a ART clinic ART eligible 2004- Karcher346 2007 Kenya adults enrolled ND 3, 4 2005 at an ART clinic Attendees at a District Hospital Nakanjako359 2007 Uganda 2004 ND 1, 2, 3 medical emergency unit Adult Attendees at a District Waxman371 2007 Kenya Hospital 2006 ND 1, 2 medical emergency unit HIV diagnosed 2002- Kaplan345 2008 South Africa women referred ND 3, 4 2007 to an ART clinic Women participating in Not Luseno354 2008 South Africa ND 1, 2 HIV prevention stated trial, urban HIV diagnosed 2004- Amuron335 2009 Uganda adults referred ND 2, 3 2007 to an ART clinic Adults 2001- April336 2009 South Africa attending urban ND 1, 2, 4 2006 HIV clinic ART eligible Bassett338 2009 South Africa adults at an 2006 ND 3, 4 urban HIV clinic ART eligible 2006- Loubiere373 2009 Cameroon ND 4 adults at 14 2007

112 Chapter 2 | Literature review

central and 13 district hospitals A: proportion enrolled in care 30 days after diagnosis; B: proportion Adults enrolled who had attending 2 2004- Micek235 2009 Mozambique CD4 count drawn 2, 3, 4 urban HIV 2005 within 30 days; C: clinics proportion eligible on CD4 who initiated ART within 90 days Adults attending 500 2005- Nsingaye361 2009 Tanzania static and ND 2, 3 2008 mobile VCT sites Nigeria, Employees of van der Burundi, beer brewing 2001- 2011 ND 3 Borght378 Rwanda, company at 16 2006 DRC, Congo sites National HIV 2005- Assefa337 2010 Ethiopia care ND 1, 2, 4 2008 programme Newly Proportion diagnosed HIV 2006- initiating ART 6- Bassett234 2010 South Africa positive adults 2, 3, 4 2008 months after at a district diagnosis hospital Cameroon, Cote d’Ivoire, Kenya, HIV positive Proportion Mozambique, pregnant receiving CD4 2003- Carter255 2010 Rwanda, women count result 2 2008 South Africa, attending within 91 days of Uganda, antenatal clinics diagnosis Zambia, Thailand HIV-positive pregnant 2006- Gareta341 2010 Malawi ND 2, 3, 4 women at 2008 tertiary hospital ART eligible 342 adults at a 2009- Geng 2010 Uganda ND 3, 4 district ART 2010 clinic New adult Proportion attendees at 36 344 2004- attending follow- Ingle 2010 South Africa HIV care 2, 3, 4 2007 up appointment facilities in one within 6-months province Random sample of HIV-positive Proportion pregnant attending for women, STD 2004- Kranzer332 2010 South Africa CD4 count 2, 3, 4 clinic attendees, 2009 measurement VCT attendees within 6-months and all TB/HIV coinfected

113 Chapter 2 | Literature review

patients from one urban clinic and hospital Adults Proportion attending an 2008- receiving CD4 Larson349 2010 South Africa 2 urban tertiary 2009 count results by HIV clinic 6-months Proportion Adults attending for attending an 2007- Larson350 2010 South Africa follow-up 3 urban tertiary 2008 appointment by HIV clinic 1 year Proportion Newly having blood diagnosed HIV 2006- drawn for CD4 Losina353 2010 South Africa positive adults 1, 2 2007 count at a district measurement by hospital 6-months ART eligible adult attendees 2005- McGrath355 2010 Malawi ND 3, 4 at a district- 2006 level ART clinic Adult HIV care clinic attendees 2001- McGuire356 2010 Malawi at 1 district ND 3 2007 hospital and 10 rural clinics Adult attendees at 1 HIV clinic at 2003- Mulissa357 2010 Ethiopia ND 3, 4 a district 2008 hospital HIV infected Proportion adults with TB initiating ART or other 2006- Murphy374 2010 South Africa within 6-months 4 opportunistic 2007 of hospital infections at a discharge district hospital HIV diagnosed 358 adults at one Not Naidoo 2010 South Africa ND 2 urban hospital- stated based VCT clinic Attendees at Ochieng- 2001- 2010 Kenya national HIV ND 3, 4 Ooko375 2007 care clinics ART eligible Nunu362 2010 Swaziland adults at 1 2009 ND 3 district hospital Adult HIV- positive HIV 2008- Tayler-Smith366 2010 Malawi clinic attendees ND 2, 3, 4 2009 at one district hospital Adult attendees at mobile testing sites, a van Schaik368 2010 South Africa 2008 ND 1, 2 primary clinic and a district hospital Participants in a home-based 2008- Amolloh334 2011 Kenya ND 1 HCT 2009 programme Braustein339 2011 Rwanda HIV-diagnosed 2007- ND 1, 2, 4

114 Chapter 2 | Literature review

female sex 2008 workers Sample of all HIV positive 2008- Govindasamy343 2011 South Africa adults who ND 2, 3 2009 tested at a mobile clinic HIV-diagnosed Proportion adult attendees 2005- initiating ART Kohler347 2011 Kenya at one district 2, 3, 4 2007 within 1 year of hospital HIV clinic registration clinic Adults attending for family, Konde-Lule348 2011 Uganda planning, VCT 2009 ND 1, 2 and antenatal services at 3 primary clinics Proportion ART ineligible completing a adults in one Lessells352 2011 South Africa 2007 repeat CD4 count 3 district HIV care measurement programme within 1 year HIV diagnosed adults in a 2005- Nakigozi360 2011 Uganda ND 3 community 2008 cohort Proportion HIV/TB initiating ART coinfected 2008- Pepper376 2011 South Africa within 6-months 3, 4 patients from 2009 of TB treatment one urban clinic commencement HIV diagnosed attendees of Scott377 2011 South Africa 133 primary 2010 ND 3 level urban clinics ART eligible 2005- Tayler-Smith365 2011 Malawi patients in 1 ND 3 2008 rural district HIV/TB coinfected 364 2004- Tayler-Smith 2011 Kenya patients from ND 3, 4 2008 three urban clinics ART eligible 2004- Togun367 2011 The Gambia adults at 1 ND 3, 4 2009 urban HIV clinic District hospital Proportion inpatients registering at Wanyenze380 2011 Uganda participating in 2004 ART clinic within 1, 3, 4 a trial of HCT 18 months of delivery diagnosis ART eligible Malawi, 2004- Zachariah379 2011 adults in HIV ND 3, 4 Kenya 2008 care clinics ND: Not defined

115 Chapter 2 | Literature review

Of note, from these 51 studies, no study followed individuals from attendance for HTC throughout the HV care cascade to ART initiation, or reported patient loss at all stages of care. Only six studies followed patients from confirmation of HIV diagnosis to initiation of ART234, 235, 332, 344, 347, 366 (3 from South Africa, 1 each from Malawi, Mozambique and

Kenya), an important paucity of information. This fact has been emphasised as a major hindrance to understanding the true extent of patient loss throughout the cascade and in developing strategies to overcome the barriers to successful linkage to ART both at the individual and programme level327.

The majority of studies reported on HIV care programmes from single clinic sites (n=26) with the remainder recruiting participants nationally or from more than one district- level facility (n=14), from clinical trial participants (n=4), from home or mobile HTC sites

(n=6) or from a workplace programme (n=1). Only 8/51 (16%) of studies included participants recruited from primary health care facilities, where the majority of HIV care in sub-Saharan Africa is delivered.

Reflecting the evolution of WHO and national ART guidelines over time, studies used varying definitions for ART eligibility, including CD4 ≤200 cells/mm3, CD4≤250 cells/mm3,

CD4≤350 cells/mm3, WHO clinical stage 3 or 4, and two studies did not define ART eligibility.

The definition of a time delineated study endpoint is an essential component of cohort studies (whether retrospective or prospective), as this ensures that every individual recruited has opportunity to be followed-up for the same length of time, thus reducing the potential for ascertainment bias. The implications of differing lengths of follow-up being available to individual study participants is illustrated in the hypothetical studies in

116 Chapter 2 | Literature review

Figure 2.16. In studies where outcomes are measured over a relatively short period of time, the consequences of not defining a time-delineated outcome can be clearly seen.

In the upper panel (i), both Patient A and Patient B undergo follow-up for 3-months and so can experience the outcome of interest for an equal period of time. However, in the lower panel (ii), where follow-up was censored at a specific calendar date rather than on than on length of patient follow-up time, Patient A and Patient B are exposed to differing follow-up periods, meaning that Patient A has longer exposure, and greater opportunity to experience the outcome of interest.

117 Chapter 2 | Literature review

Figure 2.15: Effect of time-delineated outcome definition on ascertainment bias

i) Study with time-delineated endpoint defined

A recruited B recruited Total follow-up

Recruitment period A: 3 months B: 3 months

A completes B completes follow-up follow-up

Follow-up period

Month 0 Month 3 Month 6 Month 9

ii) Study without time-delineated endpoint defined

A recruited B recruited Total follow-up

Recruitment period A: 9 months B: 3 months

A and B complete follow-up

Follow-up period

Month 0 Month 3 Month 6 Month 9

In the 51 cohort studies reporting outcomes of individuals in the HIV care cascade, the period during which individuals were followed-up (and could reach study endpoints) was

118 Chapter 2 | Literature review not defined in the majority of studies, with only 14 studies providing a time-delineated period of follow-up with a defined clinical endpoint for all recruited individuals234, 235, 255,

332, 344, 347, 349, 350, 352, 353, 369, 370, 374, 376. Follow-up periods ranged from as short as 8 weeks353, to as long as 13 months352. In the studies that defined a time-delineated endpoint, definitions included: attendance at follow-up appointment within a defined period of time; having had blood drawn for CD4 count measurement by a defined period of time; receipt of CD4 count results by a defined period of time; or initiation of ART by a defined period of time (Table 2.7). Only one study provided time-delineated endpoints for every stage of the HIV care cascade in which patients were assessed235.

Loss to follow-up was also defined differently between studies. Most studies specified that individuals who did not return to the clinic within a certain time period (ranging from 30 days to 13 months) or who missed a certain number of clinic appointments

(ranging from 1-3) were classified as lost to follow-up.

2.7.4. Summary estimates for pre-treatment patient loss to care

To date, three systematic reviews have attempted to summarise the available data relating to linkage and retention through the HIV care cascade from HTC to initiation of

ART in resource-limited settings328-330. The findings of these systematic reviews are summarised in Table 2.8.

119 Chapter 2 | Literature review

Table 2.8: Systematic review and meta-analysis of studies reporting on loss throughout the HIV care cascade

Completed ART Retained in Initiated ART Number of Completed HTC eligibility pre-ART care (if Study (if eligible) studies (Stage 1) assessment ART ineligible) (Stage 4) (Stage 2) (Stage 3)

Rosen330 28 n/s 59% 46% 68% Kranzer329 37 39%* 57% 45% 66% Mugglin328 29 n/s 72% 46% 63%

*Figure taken from WHO estimate, not from meta-analysis of selected studies n/s: not stated

Of note, the three systematic reviews that examined loss to care between HIV diagnosis and initiation of ART (Rosen et al330, Kranzer et al329 and Mugglin et al328) all examined the period between 2001 and 2011 and with similar inclusion criteria, but each included a different set of studies (in total 51 studies were included in the three systematic reviews, with Rosen et al including 28 of these, Kranzer et al, including 37 of these and

Mugglin et al including 29 of these) – Figure 2.17. All three systematic reviews obtained similar summary estimates for retention in care during Stage 2 (range: 57-72%), Stage 3

(range 45-46%) and Stage 4 (63-68%).

120 Chapter 2 | Literature review

Figure 2.16: Euler diagram showing studies included in 3 systematic reviews of linkage to ART (n=51)

Potential reasons for the differences in numbers of studies included in each of the three systematic reviews include: differing inclusion criteria; slightly differing time periods covered; different databases searched; and differing search strategies.

In their study protocol, Kranzer et al excluded studies that would not be generalizable to general adult populations of community members in high HIV prevalence settings.

During the systematic search, two such studies were included: one that was conducted among sex workers, and one that was conducted among individuals being assessed at a mobile clinic.

Whilst all 3 reviews searched the PubMed database, Mugglin et al additionally searched the EMBASE database. Rosen et al additionally searched the ISI Web of Knowledge

Database and African Indicus Medicus. Kranzer et al and Rosen et al additionally hand

121 Chapter 2 | Literature review searched the abstracts of the Conferences of Retroviruses and Opportunistic Infections and the International AIDS Society Conferences.

There were differences in the time period searched: Mugglin et al searched through to

9th August 2011, Rosen et al searched through to 5th January 2011 and Kranzer et al searched through to June 2011. The smaller number of studies identified by Rosen et al could be explained by the early search cut off time point.

There may have also been differences in the search strategies used. Comprehensive search strategies are not published (either in the main manuscript body or in supplementary material) for Kranzer et al. (I was able to obtain the search strategy from the corresponding author, but it is insufficiently detailed to allow replications of study results). Rosen et al provide a search strategy in supplementary material, but it is insufficiently detailed to allow replication of results. Only Mugglin et al provide a comprehensive and replicable search strategy.

Although no one study assessed linkage to care through all steps of the HIV care pathway, the systematic review by Rosen et al used the available data to estimate that approximately 17% (90% CI: 7%-32%) of HIV diagnosed individuals in sub-Saharan Africa will be retained in care continuously until ART is initiated330. However, Kranzer at al note that the approach taken by Rosen et al to estimate cumulative retention (multiplication of the individual proportions at each stage) would lead to an erroneously low estimate of retention in care, as this does not take into account individuals who “recycle” back into care following drop-out329.

122 Chapter 2 | Literature review

In short, despite a large number of studies being published over a relatively short period of time, problems with lack of clear time-delineated outcomes and failure to consider patient “recycling” back into care has meant that we have no clear understanding of the true magnitude of patient loss throughout the HIV care cascade. Neither do we have a clear understanding of the magnitude of patient loss from the routine primary health care system, where the majority of HIV care is now delivered. This is a major gap in understanding the performance of HIV care programmes in sub-Saharan Africa. Greater understanding of the patient flow could help improve programme design, individual patient outcomes and the development of interventions to improve linkage to care, with the ultimate aims of reducing population mortality and transmission of HIV.

2.7.5. Reasons for pre-treatment loss to follow-up

Mortality is a potential major reason for drop out from the HIV care pathway. In the systematic review performed by Mugglin et al328, 5/29 studies were identified as presenting data on mortality for patients who dropped out of the HIV care pathway344,

351, 357, 367, 378. Review of the full 51 studies that reported on pre-treatment loss-to-follow- up identified a further 12 studies reporting on mortality outcomes of individuals who default from the HIV care cascade234, 335, 338, 343, 352, 355, 356, 365, 366, 370, 374, 376 (Table 2.9)

123 Chapter 2 | Literature review

Table 2.9: Studies reporting risk of death in individuals with pre-treatment loss to follow-up Number of Number patients Number Method of % pre- Completeness of with pre- pre- Study ascertainment treatment of tracing patients treatment treatment of mortality mortality recruited loss-to- deaths follow-up Amuron ND ND 4321 637 184 4.3% (2009)335 Bassett Telephone ND 1474 391 89 6.0% (2010)234 tracing Bassett Telephone 61% 501* 82 28 5.6% (2009)338 tracing Govindasamy Telephone 381 29.2% 192 58 2 1.0% (2011) tracing Linkage to Ingle South African ND 44844 13098 9232 20.6% (2010)344 National Deaths Register Lawn Community- ND 1235 191 56 4.5% (2006)382 based tracing Demographic Lessells and health 100% 930§ 498 21 2.3% (2011)352 survey Demographic McGrath and health 65.0% 659 93 35 5.3% (2010)355 survey McGuire Community- 71.6% 19626 1747 317 1.6% (2010)356 based tracing Mullisa Community- 357 18.6% 2191 651 102 4.6% (2010) based tracing Murphy 374 ND 91.8% 49 18 14 28.6% (2010) Electronic Pepper 376 hospital 90% 111 21 15 13.5% (2011) records Tayler-Smith ND ND 427 82 17 4.0% (2011)364 Tayler-Smith 366 ND ND 1078 825 1 0.1% (2010) Togun Community- 53.5% 790* 254 118 14.9% (2011)367 based tracing van der Borght ND ND 431 49 16 3.7% (2009)378 Waynenze ND 99% 500 123 118 23.6% (2011)380 *Only ART eligible individuals traced for outcomes 4Only lost to follow-up patients with demographic and health survey records assessed for outcomes ND: Not described

Tracing of individuals who had defaulted from the HIV care cascade was poorly performed by the majority of studies, with only 4/17 studies (2 of which were

124 Chapter 2 | Literature review demographic and health surveys) successfully tracing more than 85% of defaulting patients to ascertain outcomes and 6/17 not describing methods by which tracing was performed. The risk of death varied widely between studies, from 0.1% to 28.6% of recruited patients. These reflect data from studies reporting mortality among patients lost to follow-up from ART treatment programmes383 and patients awaiting384 and receiving TB treatment385, where tracing was also poorly performed, leading to considerable uncertainty around mortality rates.

In a meta-analysis performed by Mugglin et al328, men, individuals from poorer socioeconomic groups and individuals with lower CD4 counts were at increased risk of default from the HIV care cascade. In contrast, older age and less advanced clinical stage were associated with retention in care. Insufficient data was found to investigate factors associated with mortality in individuals with pre-treatment loss to follow-up.

A systematic review of risk factors, barriers and facilitators for linkage into ART386 found that the most common reasons identified for loss-to-care in quantitative analysis were transport costs, difficulties accessing clinics, stigma, fear of disclosure and clinic related factors including staff shortages, long waiting times and unwelcoming environments.

Male sex, younger age and competing demands with work requirements were noted to be barriers to ART initiation in the majority of studies reviewed.

A small number of qualitative studies have examined factors contributing to drop-out from the HIV care cascade prior to ART initiation, including a multi-site study from urban

Nigeria and Tanzania387, and studies from rural Tanzania155, 361, rural Malawi (Karonga

District)388, urban Zambia389 and urban Kenya390. All studies undertook in-depth interviews with patients seeking care throughout the HIV care cascade, with two studies

125 Chapter 2 | Literature review additionally conducting focus group discussions or in-depth interviews with health workers involved in providing HIV care services.

Across the qualitative studies, patients identified a number of common themes as barriers to completing steps on the HIV care cascade and initiating ART. These included: cost (including transport cost and user fees); distance from clinics (which often meant individuals had to miss work or schooling); severity of disease and debility; dissatisfaction with crowded clinics and rushed health workers; fear of stigmatization from other community members, family and health workers (including fear of loss of respect, as attending an HIV care clinic is tantamount to openly declaring one's HIV- positive status); fear of the side-effects of ART; perceived negative impacts of having to take treatment (e.g. avoiding alcohol and sex); perceived negative marital consequences, including disharmony, break-up or reduced chances of marrying; and particularly among men, fear of loss of masculinity and status within the family and community.

Factors important in facilitating access to ART were less commonly reported and included: perceived improvements in health with treatment; having a "treatment supporter" who would accompany the patient to the clinic; and, particularly among women, a sense of responsibility for the children within the household. Similar factors have been reported as being important barriers and facilitators to tuberculosis treatment in resource-limited settings391. However, the fact that HIV remains a highly stigmatized disease in many societies, accompanied by the taboo surrounding diseases that are sexually transmitted, may mean that stigma plays an important role as a barrier to HIV treatment.

126 Chapter 2 | Literature review

2.7.6. Interventions to improve linkage and retention between HIV diagnosis and initiation of ART

A small number of studies have examined strategies to improve linkage between diagnosis of HIV and initiation of ART in resource-limited settings, with the majority of studies reporting non-randomised evaluations of interventions. Studies that have evaluated interventions to improve linkage in to HIV care are shown in Table 2.10. These studies were identified by undertaking a comprehensive search of PubMed and Google

Scholar at any time before 1st April 2013, using a pre-defined search strategy that included the search terms “HIV”, “antiretroviral”, “linkage”, “initiation” and “access”.

Additionally, abstracts of the Conference on Retroviruses and Opportunistic Infections and the International AIDS Society Conference were searched by hand. Because of the very small number of randomised controlled trials identified, the search was expanded beyond sub-Saharan Africa to encompass all low- and middle-income countries.

Table 2.10: Evaluated interventions to improve linkage into HIV care Year of Intervention Evaluated Author Study Design Reported outcomes publication assessed populations 55 health facilities in Increased number of Ethiopia (30 Description of ART initiation Assefa392 2012 Task-shifting physician- programme between 2006 and staffed and 25 implementation 2009 reported nurse/officer staffed clinics) Estimated 95% ART 1 hospital and coverage by 2006, 12 clinics in Decentralisati Description of although no Eastern Cape, Bedelu393 2007 on and task- programme derivation of South Africa. shifting implementation denominator or Population of eligibility criteria 150,000 given Thyolo District, Decentralisati Before-and- 5-fold increase in Bemelmans Malawi. 2010 on and task- after HCT, 2-fold increase 394 Population of shifting comparison in ART initiation 587,455

127 Chapter 2 | Literature review

Community One rural HTC, POC CD4 Ugandan Description of 57% of HIV-positive Chamie395 2012 measurement community. programme linked to ART within and transport Population of implementation 3-months stipend 6,000 One rural health Description of 5376 individuals Decentralisati zone in Lesotho. Cohen396 2009 programme initiated ART over 3 on Population of implementation years 200,000 One urban Immediate primary care provision of clinic in RR (95% CI) for ART CD4 results Johannesburg, initiation in 3-months vs. 1-week Single-blind, South Africa. for immediate Faal397 2011 delayed randomised 3- Newly delivery of results vs. delivery of arm study diagnosed HIV- standard care: 2.1 CD4 result vs. positive adults (1.39-3.17) standard of with CD4<215 care cell/mm3 Task-shifting (nurse initiation 31 primary 3712/5390 (69%) in [intervention] clinics in Free intervention arm vs. vs. physician State, South Cluster 2418/3862 (63%) in Fairall299 2011 initiation Africa. HIV- randomised control arm initiated [control]) - positive adults trial ART (RR: 1.24, 0.88- ART initiation with CD4 <350 1.73) secondary cells/mm3 planned endpoint Estimated 57% of One rural "people in need of district, ART" initiated ART, Before-and- Decentralisati KwaZulu Natal, although no Fredlund398 2007 after on South Africa. derivation of comparison Population of denominator or 95,000 eligibility criteria given One province of Recipients of "expert "Expert Kenya. 808 patient" visit more patients" newly Description of likely to link to ART home visits Hatcher399 2012 diagnosed HIV- programme (men: HR: 1.35, 0.97- (procedures positive adults implementation 1.87; women: HR: not (62% assessed 1.20, 1.00-1.43) than described) for outcomes) non-recipients Four primary Prospective 94/437 of enrolled health care non- POC CD4 recipients clinics in 2 POC CD4 randomised initiated ART within provinces of Jani236 2011 count cohort study 60 days compared to Mozambique. measurement with before- 57/492 of laboratory 1021 adults and-after CD4 recipients (OR: enrolled in HIV comparison 2.05, 1.42-2.96) care

128 Chapter 2 | Literature review

"ART ineligible patients" twice as One urban HIV likely to be lost-to- Free clinic in Nairobi, Before-and- follow-up before Kohler347 2011 cotrimoxazole Kenya. 5854 after than those after the provision adults enrolled comparison introduction of free in the clinic cotrimoxazole (aHR: 2.64, 1.95-3.57) HIV-positive Mobile HTC individuals offered units in Gauteng POC CD4 more likely Province, South Retrospective POC CD4 to attend follow-up Africa. Newly non- Larson237 2012 count visit (117/197, 59%) diagnosed randomised measurement than those not adults (63% cohort study offered (52/122, assessed for 47%, RR: 1.25, 1.00- outcomes) 1.57) Decentralisati on, task- 7 rural primary Before-and- Rapid increase in HTC Mukherjee4 shifting, 2006 care clinics in after uptake and ART 00 removal of Haiti comparison initiations reported use fees and integration One township in 183/214 (86%) of ART Cape Town, eligible pregnant Integration of South Africa. Before-and- women initiated ART ART care into Newly HIV- Myer238 2013 after after integration antenatal diagnosed HIV- comparison compared to 58/271 clinics positive (21%) before pregnant integration women One rural ward in Tanzania. HIV 68% of HIV diagnosed Patient positive adults attended follow-up carried Prospective Nsigaye361 2009 diagnosed at appointment as referral card cohort study rural VCT sites. measured by referral system Population at card collection risk not defined Rapid increase in HTC National HIV Before-and- uptake and ART care Sherr401 2009 Task-shifting after initiations following programme in comparison training of "tecnicos" Mozambique cadre Patients who Reduction in received pre-ART clinic visits One rural Retrospective adherence and removal district in non- counselling had Siedner402 2012 of ART Uganda. Adults randomised significantly longer adherence initiating ART cohort study delays to ART counselling initiation (21 vs. 42 requirement days - p=0.04)

129 Chapter 2 | Literature review

Incentivised 27/60 (45%) in the attendance Participants in a incentivised arm vs. with vouchers drug users Randomised 16/60 (27%) in the Solomon403 2012 for programme in controlled trial control arm initiated completing Chennai, India ART by 12-months targets (aHR: 2.93, 1.39-6.20)

One rural Community district in South 88% of ART eligible van HTC and POC Africa. Prospective 2013 individuals initiated Rooyen404 CD4 Recipients of cohort study at 3-months measurement home-based HTC

Only per-protocol analysis presented. 13% of non-deceased recipients of inpatient HCT One district compared to 18% of Wanyenze3 Randomised 2011 Inpatient HTC hospital in non-deceased 80 controlled trial Uganda recipients of outpatient VCT initiated ART by 6- months No effect size or confidence interval presented No evidence of 15 clinics increased uptake of offering task- HIV care services in On site shifting for HIV Cluster individuals attending Zwarenstei outreach 2011 care in the Free randomised intervention n405 nurse State of South trial (3048/5793, 53%) vs. education Africa. Adult control (2187/4343, clinic attendees 50%) clinics (OR: 1.19, 0.51-2.77) aHR: adjusted hazard ratio RR: risk ratio CI: confidence interval POC: point of care

Four randomised trials of interventions to improve linkage to ART were identified: a cluster randomised trial from South Africa299, and three individually randomised trials from Uganda370, India403 and South Africa397.

The South African cluster randomised trial compared rates of viral suppression in primary clinics where HIV care was task-shifted to trained nurses to those achieved in

130 Chapter 2 | Literature review clinics where usual physician-led care was provided. Linkage to ART was evaluated as a pre-planned secondary outcome and there was no significant difference in treatment uptake between arms (intervention arm: 3712/5390 [69%]), control arm: 2418/3862

[63%], risk ratio: 1.24, 95% confidence interval [CI]: 0.88-1.73).

In 2004, the Ugandan trial compared HIV care outcomes in hospital inpatients randomised to receive in-patient HIV testing and counselling (HCT) compared to referral for post discharge HTC. In per-protocol analysis (data from intention to treat analysis was not presented), 11/98 (13.1%) of HIV-diagnosed individuals in the in-patient HTC arm initiated ART compared to 9/55 (16.4%) HIV-diagnosed individuals in the out-patient

HTC arm (no effect size or confidence interval presented).

The 2012 Indian trial examined rates of ART initiation in injecting drug users randomised to receive cash-redeemable voucher incentives or standard care. Significantly more individuals in the incentivised arm initiated ART (27/60 [45.0%]) compared to the standard care arm (16/60 [26.7%], adjusted hazard ratio: 2.93, 95% CI: 1.39-6.20).

In Johannesburg, South Africa, newly HIV-diagnosed adults were randomised to receive immediate CD4 count results (intervention arm) or were asked to return in seven days to collect CD4 count results (standard care arm). A greater number of participants in the immediate results provision arm (28/124 [22.5%]) initiated ART compared to the standard care arm (9/112 [8.0%], risk ratio: 2.1, 95% CI: 1.39-3.17).

A number of non-randomised studies were also identified, which evaluated the effect of point of care CD4 count measurement236, 406, decentralisation393, 394, 398, integration of HIV care into primary clinics407, 408, “expert patients”399 and patient-carried referral cards361

131 Chapter 2 | Literature review on ART initiation. No studies examined the effectiveness of HIVST or home initiation of

HIV care on linkage to ART.

A cluster-randomised non-inferiority trial from Uganda showed that over 42 months, home delivery of ART (provided by nurses during visits) following clinic initiation was feasible and non-inferior to clinic provision in terms of rates of viral suppression compared to continuation of ART care delivered at facilities409. Mortality rates were similar between groups and a nested costing exercise (formal cost effectiveness analysis was not performed) showed that patient costs were substantially higher for participants in the facility-based care are than in the home-based care arm, mainly attributable to transport costs.

2.8. Development of a conceptual framework to understand pre- treatment loss to care in the HIV care cascade

A number of models have been developed by psychologists and health researchers to attempt to explain individuals’ attitudes and behaviour towards health, illness and care- seeking. Broadly, these models can be divided into intrapersonal theories (which emphasise factors within individuals that influence behaviour, including knowledge, attitudes, experience, beliefs and perceptions), interpersonal theories (which additionally assume that interactions with others influences behaviour), and community- level theories (which in addition to intra- and inter-personal factors, consider the role of social systems including communities, institutions, norms and legislation on behaviour)410. As evidenced by the qualitative studies conducted among individuals receiving pre-ART care described above, although intra- and inter-personal factors were noted to be important in determining care-seeking behaviour (e.g. with regards to stigma and perceived self-respect), external factors (including the structure of health

132 Chapter 2 | Literature review services, quality of care available and the economic environment) were noted by participants in every study to be major contributors to likelihood of successful progression through the HIV care pathway. Therefore, to conceptualise barriers and facilitators to linkage to ART, a community-level theory was drawn upon.

The socio-economic model has been developed in recent years and uses both internal and external factors to explain care-seeking behaviour. The internal and external factors that underpin behaviour are structured into levels, expanding from the intra- and inter- personal levels, through to the institutional, community and societal levels. The basic underlying principle of the socio-ecological model is that there is interplay between factors at different levels that impact upon behaviour, and so modifications of factors at one level result in changes at all other levels.

Drawing upon the existing qualitative and quantitative literature that reports on care- seeking behaviour for patients in the HIV care cascade, and studies describing care- seeking for other diseases including tuberculosis, a socio-ecological conceptual framework was constructed (Figure 2.18) to describe the interplay of factors that could act as barriers and facilitators to individual patient progression through the HIV care cascade. Drawing as it does upon existing health behaviour theory and existing knowledge of factors known to be important in influencing progression through the HIV care pathway, this conceptual framework allows an understanding of the factors that may be important to patients seeking care for HIV in Blantyre, and in identifying potential modifiable areas that could be incorporated into a novel and effective intervention to improve linkage to ART.

133 Chapter 2 | Literature review

Figure 2.17: Socio-ecological conceptual framework of progression through HIV care pathway

Structural Factors

• Policy (e.g. national HIV/ART guidelines, international guidelines) • Health systems (including financing, structure) • Socio-economic environment (including macro-economic context, external donor funding)

Social and cultural factors Individual factors Programme/health facility factors • Socio-cultural • Knowledge, attitudes environment and beliefs of: • Accessibility o Gender norms o HIV o Costs o Attitudes to HIV, o ART o Distance stigma & discrimination • Perceived benefits of: • Availability o HIV testing o Drugs • Networks o ART o Equipment o Family, friends & o Providers work • Perceived constraints o Church to: • Quality of care o HIV testing o Effectiveness & • Micro-economic context o ART timeliness o Poverty o Patient-centeredness o Equity

Linkage into HIV care

2.8.1. Summary of evidence regarding magnitude, reasons for, consequences and potential areas for intervention to improve pre-treatment loss to care

In summary, despite a large number of studies published over a relatively short time period (6 years), there are still substantial knowledge gaps in our understanding of patient flow through the HIV care cascade in sub-Saharan Africa. No study has followed individuals through the complete HIV care cascade, from HTC to initiation of ART.

Despite the large majority of HIV care being provided at primary care level in sub-

134 Chapter 2 | Literature review

Saharan Africa, only 8 studies have evaluated linkage to ART among HIV-positive primary clinic attendees. The considerable variability in the quality of studies reported, the populations recruited, the clinic settings, definitions of ART eligibility, and the suboptimal rates of patient follow-up are a substantial barrier to accurate understanding of the magnitude of pre-treatment loss-to-follow-up, and in developing potential interventions. From the few studies that have presented disaggregated data, there are suggestions that certain groups of HIV-infected individuals (especially younger men with competing demands from work) could be at increased risk of pre-treatment loss to follow-up. Moreover, the data available suggests that pre-treatment loss to follow-up could have significant consequences at the individual level, with some studies reporting high rates of mortality among individuals who have dropped out of care.

In national HIV care programmes such as Malawi, which has been hailed as a success story in expanding access to ART using the public health approach, as well as other countries in sub-Saharan Africa, impressive numbers of individuals initiating ART may be concealing extensive pre-treatment loss, hindering efforts to achieve universal coverage of ART with the ultimate aim of reducing HIV transmission and population mortality.

Only three randomised controlled trials in sub-Saharan Africa have tested strategies to improve linkage to ART, with the most methodologically robust study comparing linkage as a secondary outcome in South African clinics with nurses who had received intensive

HIV care training with clinics where nurses had not received additional training. No randomised trials have evaluated interventions that aim to improve linkage at all steps of the HIV care pathway, addressing factors noted to be important barriers to retention in care by the small number of qualitative studies conducted to date. Additionally, no

135 Chapter 2 | Literature review randomised controlled trials have examined the effectiveness of HIVST or home initiation of HIV care on linkage into HIV care.

Therefore, a considerable and urgent imperative exists to better understand the magnitude, reasons for and consequence of pre-treatment loss to follow-up in the HIV care cascade in national programmes following the public health approach to ART delivery. Gaining an in-depth understanding of the patient flow will aid understanding of how best to develop effective interventions that could have substantial impact on improving patient-important outcomes.

2.9. Measurement of household poverty status

2.9.1. Poverty, illness and health

Poverty has long been recognised to be an important determinant of poor health.

WHO’s Commission on the Social Determinants of Health reported in 2008 that life chances vary considerably within countries, with the poorest people having the highest burden of illness and premature death411. Moreover, there is a strong direct relationship between socioeconomic position and illness412.

Throughout history infectious diseases have been strongly associated with poverty.

Whether manifest through the extremely high rates of infection with typhus in poor peasants in the 19th century identified by Virchow, or the recent cholera epidemic among individuals living in the slums and refugee camps of Haiti, the poorest in society bear the brunt of communicable diseases413. The institutionalised factors that amplify the risk of infectious diseases among the poor – including substandard housing, water and sanitation, poor nutrition, poor access to healthcare services, the disproportionate

136 Chapter 2 | Literature review consequences of natural disasters and war and lack of political power – have been collectively termed “structural violence”414.

Whereas the majority of countries organise their health services according to the principal of universal coverage (everyone can access the same range of services according to need), in practice consideration disparities are found in the ability of individuals from different socioeconomic positions to access healthcare412. For example, within and across countries, individuals in the lowest socioeconomic groups are substantially less likely to have accessed a range of antenatal and child health services compared to individuals in the highest socioeconomic group412. Sir Michael Marmot,

Chair of the Commission on the Social Determinants of Health has emphasised the ethical and public health imperative to address these health and illness disparities:

“If systematic differences in health for different groups of people are avoidable by reasonable action, their existence is, quite simply, unfair. We call this imbalance health inequity.”412

2.9.2. Measuring poverty in resource limited settings – the proxy means test method

One area of investigation in this thesis was the relationship between socioeconomic status and an HIV-infected individual’s ability to access treatment. Thus a measure of poverty that could be used to assess HIV-infected individuals’ socioeconomic status was required.

137 Chapter 2 | Literature review

Defining poverty is complex as a number of different domains contribute to the construct and consequences of low-socioeconomic status415. The World Bank defines poverty as:

“…whether households or individuals have enough resources or abilities today to meet their needs; inequality in the distribution of income, consumption or other attributes across the population; and vulnerability, defined here as the probability or risk today of being in poverty – or falling deeper into poverty -- in the future.”415

An important concept to note arising from this definition is that inequality in individuals’ or households’ distribution of income, consumption or vulnerability means that their relative (and not just their absolute) position in society is important415.

There are two main monetary measures of poverty: income-based measures and consumption-based measures415. The World Bank emphasises that consumption-based measures are more useful than income-based measures for two reasons: firstly, in poor economies, income may fluctuate considerable and be difficult to estimate precisely; and secondly, consumption is a better reflection of actual standards of living and ability to meet basic needs such as nutrition, health, housing and security415. The use of a monetary-based measure of poverty allows the construction of a “poverty line”, which is a cut-off line that separates the poor from non-poor. In the case of consumption-based measures, the poverty line separates individuals who are able to need their basic needs from those who are not. Poverty lines may be absolute (anyone below a certain level of consumption is defined as “poor”) or relative (defined in relation to the overall distribution of consumption in the population, e.g. 50% of the a country’s mean consumption)415.

138 Chapter 2 | Literature review

Integrated household surveys (IHSs) are population-based surveys that collect data on the social and economic profile of households416. A large number of items on income, education, asset ownership and expenditure are collected from a representative sample of households. Data from IHSs are used to inform the public and policymakers about the socioeconomic profile of households within the country. As such, they are useful source of information for quantifying poverty within a population. IHSs have been conducted in

Malawi between 1997-1998 and 2004-2005.

As a large number of comprehensive socioeconomic variables are collected by IHSs, it would be impractical to collect the same set of indicators from individuals participating in medical research to allow assessment of their poverty status. Instead, the proxy means test modelling approach has been advocated417. Here researchers collect a smaller number of socioeconomic indicators from study participants that are correlated with the consumption-level of households in the study population. An individual’s response to each of this smaller set of indicators within the proxy means test regression model allows construction of a measure of consumption, by which individuals can be divided into poverty quintiles, or classified against the national poverty line417.

Payongayong and Colleagues used data from the 1997-1998 Malawi IHS to develop a proxy means test for Malawi. With the use of ordinary least squares regression, they were able to construct a proxy means for individuals living in urban areas (“urban model”) that was strongly correlated (r2=0.60) with the consumption-based poverty measurement derived from analysis of the full IHS data. Nine variables were included in the final urban model proxy means test and are listed in Table 2.11.

139 Chapter 2 | Literature review

Table 2.11: Urban model proxy means test for poverty in Malawi Dependent variable: (ln) household poverty indicator coefficient t-statistic Household owns a fridge 0.518 6.25*** Household size (number of residents) -0.306 8.96*** Household size squared 0.016 5.26*** Age of head of household 0.005 2.22** Education level of household head 0.151 7.04*** Number of salaried household members 0.061 1.87* Household owns a car or motor cycle 0.704 6.30*** Household get lighting from electricity or gas 0.280 5.00*** Household owns a bed 0.247 4.11*** Household is in Blantyre City -0.037 -0.42 Household is in Zomba Municipality -0.347 3.45*** Household is in Mzuzu City -0.205 -1.66 Constant 2.347 16.84*** * significant at P<0.10; ** significant at P<0.05; *** significant at P<0.01

The validity of this Malawian urban model proxy means test has been evaluated using participatory wealth ranking methods418. The proxy means model has also been evaluated among patients accessing treatment for tuberculosis in Lilongwe, Malawi. The proxy means test had good agreement (86%), sensitivity (46%) and specificity (95%) when compared to the gold standard of the IHS set of indicators419. However, the data used to construct the proxy means test is now old. Despite repeated efforts, I was unable to obtain the raw datasets from the updated Malawi HIS conducted in

2004/2005 from the National Statistical Office. Therefore, the measurement of poverty used in this thesis may not be optimal.

2.10. Malawi

2.10.1. Geopolitical and socioeconomic indicators

The Republic of Malawi is a landlocked country situated in Central Africa and is bordered by Zambia, Mozambique and Tanzania (Figure 2.19). Lake Malawi, formed by the East

African Rift Valley, makes up approximately 12% of the surface area of the country and feeds into the Zambezi River via the Shire River.

140 Chapter 2 | Literature review

Figure 2.18: Geopolitical map of Malawi

Source: United Nations Dag Hammarskjöld Library

Archaeological evidence suggests that the area now known as Malawi was first inhabited by migratory Bantu tribes in the 10th century. Following David Livingstone’s exploration in 1878, Great Britain claimed the region and established the Nyasaland Protectorate.

141 Chapter 2 | Literature review

European settlers were given land for coffee and sugar plantations, usually forcing

Africans to work under conditions of indentured labour. Dr Hastings Kamuzu Banda led the campaign for decolonisation in the 1950s and, following negotiations in London, elections were held in 1961. Banda’s Malawi Congress Party (MCP) won over 90% of the popular vote and Banda was appointed Prime Minister in 1963. Despite progress in infrastructure development, Banda’s rule became increasingly autocratic and he was succeeded by Presidents Bakili Muluzi in 1994 and Bingu wa Mutharika in 2004.

Mutharika’s presidency was marked by substantial gains in food security and access to health services. During the course of this PhD Fellowship, Mutharika died suddenly in

Office in April 2012 and was succeeded by the Vice President, Joyce Banda.

142 Chapter 3 | Methods

3. Methods

This Chapter describes methods that were common to all, or most, of the studies described in the following chapters of the thesis.

3.1. Study location and population

The four studies described in this thesis were all based in three wards (Ndirande,

Chilomoni and Likhabula) of the northern suburbs of Blantyre, Malawi (Figure 3.1).

These wards were selected as, in common with other densely-populated urban and peri- urban centres in Southern Africa, they had a high burden of undiagnosed HIV420. A study that randomly sampled household residents of these wards in 2010 found adult HIV- prevalence to be 18.5%171.

Chilomoni ward is located towards the north west of Blantyre, on the foothills of Mount

Michuri. Chilomoni has little access to public services such as piped water or sanitation and subsistence farming is common. Ndirande ward, located in the northeast of

Blantyre, is more densely populated than Chilomoni and supports a large amount of formal and informal trading. Likhabula ward is situated between Chilomoni and

Ndirande, directly to the north of Blantyre city centre. Similar to Ndirande, Likhabula is a very densely populated urban slum with little public amenities.

143 Chapter 3 | Methods

Figure 3.1: Study wards and health facilities in Blantyre, Malawi

Malawi

Blantyre

Blantyre

Likhabula

Chilomoni

Ndirande

Ndirande Chilomoni Health Centre Health Centre

QECH

QECH: Queen Elizabeth Central Hospital

144 Chapter 3 | Methods

3.1.1. Health facilities available to study population

There are two primary health care facilities located within the study wards: Ndirande

Health Centre and Chilomoni Health Centre (Figure 3.1). Together, these clinics are the main public providers of HTC and ART in the primary health care system in the northwest of Blantyre. Both clinics offer a range of primary care services, including antenatal care and delivery (including assisted delivery, but not caesarean section), under-5 years clinics, treatment for acute illness and minor trauma and laboratory microscopy services for diagnosis and treatment of TB and malaria.

PITC and comprehensive HIV care are offered according to Malawian national HIV guidelines174. This includes: management of opportunistic infections; chemoprophylaxis with cotrimoxazole and isoniazid preventive therapy (IPT); syndromic management of

STDs; diagnosis and treatment of tuberculosis; contraceptive services and initiation of

ART.

There are two main cadres of healthworkers who provide HIV care within the primary healthcare system: counsellors and nurses. Counsellors offer and undertake HTC and nurses undertake ART eligibility assessments and initiate ART. At the commencement of these studies (January 2011) in the two clinics a combined total of 5720 patients had been initiated onto ART since treatment was made available in 2003421.

In the Malawian primary health care system, standardised Ministry of Health clinic attendance registers are used in each department (general outpatients, under-fives clinic, antenatal clinic, TB clinic and HIV/ART clinics). Monthly aggregate data are compiled by each facility and reported to the Ministry of Health via the District Health

Office.

145 Chapter 3 | Methods

Both primary health care centres refer patients requiring tertiary-level care to Queen

Elizabeth Central Hospital (QECH), which is located in the city centre of Blantyre (Figure

3.1). In addition to providing specialist medical, surgical and obstetric care, QECH has a very large HIV care clinic where over 15,000 HIV-infected individuals had initiated ART between 2003 and the commencement of these studies421.

3.2. Delineation of study clusters and enumeration of population

3.2.1. Background to trial structure

The cluster-randomised trial described in Chapter 6 (CONDA-YAPA Study) was nested within a larger cluster randomised trial (HitTB, Principal Investigator: Liz Corbett).

The HitTB study investigated the effectiveness of two strategies of TB control: active case finding for TB alone (control arm: 14 clusters); or active case finding for TB with an intensive package of TB and HIV prevention including distribution of isoniazid preventive therapy, offer of community-based HIVST with prompt linkage to facility-based HIV care and treatment for HIV-positive individuals (intervention arm: 14 clusters). The HitTB

Study primary trial endpoint was a between study arm comparison of adult TB case notification rates.

The CONDA-YAPA study was a second-stage randomisation of the 14 clusters allocated to the intervention arm in the HitTB Study (Figure 3.2). Adult residents of all 14 clusters in the CONDA-YAPA study had access to active case finding for TB, isoniazid preventive therapy, community-based HIVST and linkage to facility-based HIV care and treatment.

146 Chapter 3 | Methods

Figure 3.2: Schemata of design of HitTB and CONDA-YAPA cluster randomised trials

28 clusters (HitTB Study)

Randomised

HitTB intervention HitTB control arm arm

14 study clusters 14 study clusters

Randomised

7 clusters 7 clusters (CONDA-YAPA intervention (CONDA-YAPA control arm) Arm)

3.2.2. Demarcation and enumeration of study clusters

Based on sample size calculations for the primary endpoint comparing TB case notification rates, the HitTB cluster-randomised trial had a total sample size of n=28 clusters, with a required mean adult (≥16 years) cluster population of 1200. Clusters were based on existing community health worker (CHW) catchment areas. CHWs are health professionals employed by the Ministry of Health with the responsibility of providing health promotion, basic diagnosis and treatment to defined geographical catchment areas.

As part of the HitTB Study, research assistants and CHWs undertook circumferential walks of CHW catchment areas with global positioning satellite (GPS) readings (eTrex

147 Chapter 3 | Methods

Legend® HCx, Garmin International Inc, Olathe, USA) to demarcate the preliminary cluster boundaries. Research assistants visited each household within the cluster and enumerated household members. Cluster boundaries were revised following the completion of enumeration to ensure that each cluster had approximately 1200 adult residents and that, as far as possible, cluster boundaries followed natural delineators such as roads and rivers. The final cluster boundaries were imported into Google Earth

Pro (Google, Mountain View, USA) and overlaid onto satellite maps of the study area

(Geo Eye-1, Eurimage, E-GEOS, Rome, Italy). A map showing the boundaries of all 28 clusters included in the HitTB Study is shown in Figure 3.3.

148 Chapter 3 | Methods

Figure 3.3: Overall study area showing boundaries of 28 HitTB clusters 2 3 Ndirande HealthCentre 4 1 9 7 8 6 5 11 Ndirande 10 12 QECH 21 22 20 14 Likhabula 19 18 13 15 16 17 23 Chilomoni HealthCentre 25 26 24 27 Chilomoni 28

149 Chapter 3 | Methods

3.3. Cluster population census

A census of cluster residents was undertaken between April and July 2011 to estimate the denominator population for the HitTB Study

Adult cluster residents were selected to be enumerators using a community participatory approach, with the eligibility criteria being that they should be literate and reside within the cluster that they would enumerate. In total, four enumerators were selected per cluster.

Study enumerators visited all dwellings (defined as a physical structure in which one or more households may reside) within each cluster during the census period and recorded a set of GPS coordinates. At each dwelling, they recorded the number of households

(defined as “…a person living alone or a group of people living together who share meals or other essentials for living; and may be related or unrelated”) residing within the dwelling and undertook a detailed census questionnaire with the head of each household, or the most senior member of the household if the head of the household was not available.

Study census forms incorporated questions used in the 2011 Malawi Demographic and

Health Survey113 and on publically available forms used in the South African National

Census of 2011, as well as questions used in a previous cluster-randomised trial conducted in urban Harare, Zimbabwe422. Census questionnaires recorded the number of adult (≥16 years of age) and child residents in each household (with a “resident” defined as someone who sleeps most nights in the dwelling), disaggregated by age and sex. Additional data were recorded on: household access to water, sanitation, lighting

150 Chapter 3 | Methods and heating; household asset ownership; level of education and literacy of household head; and migration and mortality among household members in the preceding year.

Completed census record forms were entered into a Microsoft Access® database using an optical character recognition system and 5% of households were randomly sampled for re-enumeration for quality assurance. Following finalisation of cluster boundaries (see

Chapter 3.2.2), the census database was limited to include only members of households resident within the final cluster boundaries. Cluster-, household- and individual-level characteristics of this study population are presented in Chapter 6.

3.4. Classification of cluster residency status of facility attendees (“map-book” system)

3.4.1. Background and need for cluster categorisation system

In both the HitTB and CONDA-YAPA studies, a substantial methodological challenge arose from the need to categorise individuals initiating TB treatment and ART at the two primary health care centres (Ndirande and Chilomoni) or Queen Elizabeth Central

Hospital as cluster residents (yes/no) and to individual cluster level (cluster 1 to 28). As study endpoints (TB case notification and ART initiation) were recorded at sites distant from individuals’ place of residence a system was required that could accurately and reliably categorise cluster residency status for a large number of clinic attendees and be used practically by research assistants working in busy health facility settings.

Previous community-based studies with the same need to identify cluster residents outside the household have used a number of methods. Examples have included distribution of “referral cards” during cluster enumeration embedded with trial-relevant demographic details that residents are requested to bring to facilities whenever

151 Chapter 3 | Methods attending for care423, “key informant” systems where cluster residents report study endpoint events to the study team who then carry out confirmatory home visits424 and biometric systems, where residents fingerprints are captured during enumeration and used for identification upon clinic presentation425. However, these can be resource intensive and subject to failure. For these studies, cost, the large number of people initiating ART and antituberculosis treatment in each clinic and the need for confidentiality made all of these approaches not feasible.

3.4.2. Cluster categorisation method 1 (extraction of address locations from routinely captured data in clinic registers)

We initially investigated whether cluster residency status could be reliably ascertained from a written description of the household location that is routinely collected by facility staff from patients who start ART in Malawi and is recorded in ART registers. Research assistants extracted descriptions of household location of all patients who initiated ART from clinic registers at Ndirande Health Centre, Chilomoni Health Centre and Queen

Elizabeth Central Hospital during August 2011 and independently categorised each ART initiator as either “cluster resident” (binary and to cluster level), “out of cluster resident” or “uncertain” based on their knowledge of cluster boundaries.

3.4.3. Cluster categorisation method 2 (“map book” system)

For the second method, a “map books” system was developed. Map books contained a front cover of a high-resolution satellite image of the overall study area and cluster boundaries, annotated with local landmarks (Figure 3.3). Subsequent pages showed each cluster boundary with sufficient resolution to identify cluster dwellings (Figure 3.4).

Important local landmarks such as health centres, churches, chiefs’ residences, social meeting places and natural features were also indicated on each page.

152 Chapter 3 | Methods

Figure 3.4: Example cluster boundary showing residencies and annotated landmarks

The research assistant and the ART initiator together located the initiator’s place of residence (Figure 3.5). An ART initiator’s residence was defined as the place where they usually ate their meals and slept. Research assistants were asked to record the time taken to categorise the cluster residency status of each ART initiator with a stopwatch.

153 Chapter 3 | Methods

Figure 3.5: Map book system for categorisation of cluster of residence of ART initiators at study clinics

ART initiator invited to identify their place of residence following instruction by research assistant

ART initiator locates their residential area using the front page of cluster map book

“Out of cluster resident” if from non-study suburb

Research assistant chooses most likely page of map.

ART initiator uses roads, rivers and annotated local landmarks to identify residence (if necessary with assistance of research assistant)

If unable to locate dwelling, categorised as “uncertain”

Categorised as: • “In-cluster resident” if dwelling is inside a Home visit with GPS coordinates cluster boundary from dwelling o Cluster number (1-28) recorded

• “Out of cluster resident”

3.4.4. Validation of map book system

A random sample of 10% of ART initiators and a further sample of 10% of ART initiators classified as “in-cluster residents” were drawn from the study database, weighted by clinic of ART initiation and excluding ART initiators with “uncertain” cluster residency status (who were accompanied home to obtain GPS coordinates). A research assistant

154 Chapter 3 | Methods blinded to map book cluster categorisation then took a set of GPS coordinates from each

ART initiator’s household during a home visit.

Inter-observer agreement between the two research assistants and agreement between cluster categorisation using the “map-book” system against GPS cluster categorisation were assessed using the kappa statistic. Bias corrected 95% confidence intervals for the kappa statistic were calculated using bootstrapped estimates with 1000 repetitions426.

We calculated a sample size of 40 to test whether an obtained reliability of 0.95 exceeded a reliability of 0.90 given 2 raters, a 1-tailed alpha of 0.05 and power of 0.80 using the method described by Walter et al427. Random sampling and statistical analysis was done using Stata 11.2 (College Station, Texas, USA).

3.4.5. Results: Cluster categorisation using description of physical location of residence

During August 2011, the physical locations of residence of 129 adult ART initiators were extracted from the ART initiation registers at the three facilities (100% capture). There was a high proportion of cases in which the research assistants were unable to categorise ART initiator’s cluster residency status: the first research assistant was unable to categorise 41/129 (31.8%) of ART initiators; the second research assistant was unable to categorise 77/129 (59.7%). Overall inter-rater agreement between the two research assistants was 61.5%, with a kappa statistic of 0.51 (bias corrected 95% confidence interval [CI]: 0.41-0.62). With ART initiators unable to be categorised removed from the sample, the agreement and kappa statistic were 88.6% and 0.87 (bias corrected 95% CI:

0.77-0.97) respectively.

155 Chapter 3 | Methods

3.4.6. Results: Cluster categorisation using the map book system

There were 202 ART initiations between 20th December and 27th January 2012 at the three facilities. Using the map book allocation system in the ART clinics, research assistants assigned 48 (23.8%) of ART initiators to a study cluster, 147 (72.8%) to outside study clusters and 7 (3.4%) were allocated to “uncertain” cluster status and received

GPS determination of “final cluster status”.

The seven ART initiations with “uncertain” cluster status were removed from the database leaving 195 who were included in the sampling frame for GPS validation.

Following clinic-weighted random sampling, 20 ART initiators were selected. There was complete agreement (100%) between map books cluster categorisation and GPS cluster categorisation with a kappa statistic of 1.00 (bias corrected 95% CI: 1.00-1.00, p<0.0001).

A further 20 ART initiators categorised as “in-cluster” residents by the “map-book” system were randomly selected (weighted by clinic) from the study database. There was one “in-cluster resident” incorrectly categorised as an “out of cluster resident” by the map book system, due to very close proximity. This gave an agreement and kappa statistic of 95.0% and 0.96 (bias corrected 95% CI: 0.84-1.00). The overall agreement between map book categorisation and GPS categorisation was 97.5% with a kappa statistic of 0.97 (bias corrected 95% CI: 0.90-1.00).

The median time to categorise ART initiators cluster residency status using the map book system was 27 seconds, (interquartile range: 20-34 seconds).

156 Chapter 3 | Methods

3.4.7. Summary

This study showed that cluster residency status can be rapidly and reliably ascertained at facilities distant from community members’ place of residence using a simple and non- expensive map book system. Cluster status could be rapidly ascertained in all but a small proportion (3%) of ART initiators, making home visits necessary only for few uncertain cases. The clusters in the study were located in densely populated urban slums with no formal address system and high levels of illiteracy. Nevertheless, ART initiators (with the assistance of research assistants) at health facilities were able to accurately determine their cluster residency status using the map book system.

3.5. HIV testing and counselling (HTC)

3.5.1. HTC in primary health care centres

In the cohort study described in Chapter 4 based at Ndirande and Chilomoni Health

Centres, HIV testing and counselling (HTC) was offered as a routine clinical service.

Ministry of Health-trained counsellors provided HTC for clinic attendees who were referred for HTC during routine clinic attendance. Study participants were recruited following the completion of all HTC procedures to avoid selection bias in the population of HIV-positive individuals. According to Ministry of Health and WHO guidelines, all clinic attendees, regardless of risk factors, should receive provider-initiated HTC. This included individuals attending the general outpatients department, antenatal departments and

TB departments.

Healthworkers performing HTC in the clinics followed testing algorithms recommended in national guidelines428 (serial finger-prick blood testing using Determine 1/2, Abbott,

Tokyo, Japan, and, if positive, Uni-Gold HIV, Trinity Biotech, Dublin, Ireland, with SD

157 Chapter 3 | Methods

Bioline 1/2, Standard Diagnostics, Korea was used as the arbiter if the first two were discordant).

3.5.2. HIV self-testing through community counsellors

All fourteen study clusters that formed the intervention arm of the HitTB study and the starting cohort of the CONDA-YAPA Study received access to community-based HIVST with oral mouth swab kits (OraQuick®, OraSure Technologies, Bethlehem, USA; assembled in Thailand). Although community-based HTC was available for two years in each cluster, the CONDA-YAPA study only examined the first six months of HIVST availability.

In each cluster, participants could request an HIV oral self-test kit from one of two HTC- trained community counsellors. Community counsellors were selected by a community- led participatory approach and underwent HTC training provided by the Ministry of

Health, in addition to further study training in the use of oral HIV test tests. Community counsellors promoted the availability of HIVST in their cluster by leaflet distribution and community awareness activities. Participants could attend the community counsellor’s place of residence to request HIVST between 8am and 5pm from Mondays to Saturdays.

Community counsellors additionally went door-to-door in their cluster, promoting the availability of HIVST.

Adult cluster residents who requested HIVST from the community counsellor were assessed for study eligibility (inclusion criteria: age ≥ 16 years and cluster resident). If eligible, they were provided with pre-test counselling by the community counsellor and asked if they wanted to undergo HIVST at home. If they agreed to home HIVST, they

158 Chapter 3 | Methods were required to undertake a competency test, demonstrating that they were able to complete the procedural steps of HIVST and could interpret results correctly. If participants did not want to undergo HIVST at home, they were offered the option of supervised HIVST in the presence of the community counsellor or standard HCT with finger-prick blood sampling (serial testing using Determine 1/2™, Alere, Waltham, USA and Uni-Gold™ Recombigen® HIV, Trinity Biotech, Bray, Ireland). If participants failed the competency test, they were offered standard HTC using finger prick testing. The community counsellors promoted couples HTC.

Participants who were eligible for HIVST were provided with an OraQuick® test kit and given a detailed demonstration of how to use the kit and interpret the results by the community counsellor. Written informed consent (or witnessed thumbprint if illiterate) was obtained from all adults undergoing home HIVST. They were also provided with a brief questionnaire about their experiences of HIVST that they could complete after completing testing procedures.

Participants who agreed to HIVST were asked to test in the privacy of their own house, but to return the used test kit to the community counsellor after testing (with community counsellors making a follow-up home visit if the test kit had not been return within 7 days). Self-testing participants were advised that they were not required to disclose their HIVST results to the community counsellor, but if they wished to do so, they would receive targeted post-test counselling including supported access to local health facilities for HIV care. Participants who did not wish to disclose their status to the community counsellor could place their used test kit (and a self-completed questionnaire) into a sealed envelope and place them into a locked ballot box at the community counsellor’s residence. They also received generic post-test counselling,

159 Chapter 3 | Methods including instructions for further action if their test was negative or positive. Uptake of

HIVST and reporting of positive HIVST results was recorded by the community counsellors in HTC registers.

All participants were given a self-referral card that they could use to access the local health facilities for further routine HIV-related care, including confirmatory HIV testing,

ART eligibility assessment and initiation or HIV prevention advice. Study nurses based within each health facilities were available to assist participants attending the clinics and to integrate them into the routine care system.

3.6. Key informant system for recording cluster resident deaths

An exploratory trial endpoint of both the HitTB Study and the cluster-randomised trial described in Chapter 6 of this thesis was a comparison of adult non-traumatic, HIV- associated mortality rates. At the time of the study, Malawi did not have a system of vital registration, therefore an alternative method of recording deaths and ascertaining the likely cause of death was required. To achieve this aim, a key informant system was developed based upon the health and demographic surveillance activities used at the

Karonga Prevention Study429 in northern Malawi.

To record cluster mortality events, four key informants (cluster residents) were selected by the population using a participatory approach for each cluster. Each key informant was responsible for recording all deaths in one-quarter of the cluster residents.

Meetings were held in the cluster and the presence and function of key informants publicised. Upon receiving notification of a death within their cluster, key informants attended the household to confirm the death by interviewing a surviving family member.

160 Chapter 3 | Methods

As soon as possible following confirmation of death, a study nurse attended the household and conducted a verbal autopsy with the deceased person’s primary caregiver. A verbal autopsy is a structured questionnaire that asks the primary caregiver to describe the symptoms and signs displayed in the period before the individual’s death and allows trained healthworkers to determine a likely cause of death. Study nurses also reviewed any available medical records. Verbal autopsies are widely used in demographic and health surveys where there is no vital registration systems. In this study we used verbal autopsy questions from the Karonga Health and Demographic

Survey and from the World Health Organizations’ toolkit. Study nurses undertaking verbal autopsies were also asked to determine the likelihood that the deceased individual was infected with HIV, regardless of the likely cause of death or whether HIV had been diagnosed ante-mortem.

To standardise analysis of population mortality data, a computer programme specifically designed for analysing verbal autopsy data was used (InterVA 3.0). This programme uses

Bayesian probabilistic models to determine the most likely cause of death given the response to a number of verbal autopsy questions, removing observer bias and subjectivity in analysis of causes of death. The InterVA system is widely used in demographic and health surveys and has been shown to have good correlation with physician assessment of cause of death in studies in Ethiopia430 and South Africa431.

161 Chapter 3 | Methods

3.7. Laboratory methods

3.7.1. CD4 count measurement

In the CONDA-YAPA cluster randomised trial described in Chapter 6 and the evaluation of a novel ART eligibility assessment tool described in Chapter 7, CD4 lymphocyte counts were measured for individuals receiving home assessment for ART eligibility.

Study nurses drew 5ml of venous blood with the aid of a tourniquet into an

Ethylenediaminetetraacetic acid (EDTA)-containing tube. Blood samples were place in a cool box and transported by motorbike to the HIV and TB research laboratory at the

College of Medicine of Malawi where they were analysed within six hours. CD4 count analysis was performed using the Partec Cyflow® SL-3 platform (Partec, Görlitz,

Germany).

3.8. Measurement of poverty status

Measurement of study participants’ poverty status was undertaken using the proxy means test methods described in Chapter 2.9. Although it would have been ideal to develop a new proxy means test for Malawi using the 2004-2005 IHS data, despite numerous attempts we were unable to obtain the IHS datasets from the National

Statistics Office of Malawi.

Participants were asked the nine urban model proxy means test questions to allow them to be wealth ranked against the Malawian IHS reference standards and classified as poor or non-poor. This approach was used in the census enumeration of cluster residents

(Chapter 3.3), the cohort study of newly diagnosed HIV-positive adults in primary health

162 Chapter 3 | Methods care centres (Chapter 4) and the cluster randomised trial of home assessment and initiation of HIV care (Chapter 6).

3.9. Ethical issues in studies

Throughout the history of the response to the HIV epidemic, numerous ethical dilemmas have arisen, which have challenged clinicians, researchers and policymakers to reflect on how their actions can impact upon some of the most vulnerable members of society, particular in developing countries432. Key areas of ethical contention have centred on the struggle to obtain affordable access to ART in the poorest countries that have been worst affected by HIV epidemics and implementation of international travel restrictions for individuals infected with HIV. Other important debates continue around whether it is ethically acceptable to adopt more utilitarian approaches to HIV testing to control the epidemic. For example some ethicists have argued that efforts to achieve universal coverage of HTC during pregnancy could subvert women’s right to autonomy, with implementation of some testing PITC programmes potentially giving insufficient opportunity for women to make informed decision to opt in or out of testing433.

In designing and undertaking the studies in this thesis, a number of ethical issues arose that required careful consideration to minimise the risk of harm to participants and the wider study communities in Blantyre. Findings from studies also raised important ethical questions relating to how clinical research can best inform clinical practice, maximise the health of the population and ensure that the rights of individuals populations are protected.

In the prospective cohort study described in Chapter 4, I set out to investigate rates uptake of HTC and linkage into HIV care under routine programme conditions in primary

163 Chapter 3 | Methods health care centres in Blantyre. In designing this study, I decided to not offer additional research-supported services to improve uptake of HTC or access to HIV care in the study clinics. This decision was taken to allow collection of accurate and generalizable epidemiological data and identification of risk factors for failure of linkage into care.

Offering additional interventions (such as making study staff available for HIV counselling or testing, providing on-site measurement of CD4 counts or encouraging participants to complete care-seeking steps) would have undoubtedly influenced participant behaviour and study outcomes, making it more difficult to describe key health system failures and develop interventions to improve linkage into HIV care.

Additionally, it was important in the cohort study to achieve low rates of loss to follow- up among participant groups that other studies have found difficult to effectively retain in care. Recognising that many participants would likely not return to clinics, undertaking home tracing visits would be important to minimise loss to follow-up.

However, this could potentially compromise the privacy of participants who had been recently diagnosed with HIV.

These decisions raised a number of ethical questions. In particular, was it justifiable to not offer participants additional research-supported services above and beyond what was being provided through the routine health system, especially given the large individual and public health benefits gained through knowledge of HIV status and accessing care? There is an additional clear gap between services that are recommended in WHO and Malawi National HIV care guidelines, and what was provided through the primary health care clinics during the study. Would it be justified to provide additional research-supported services for the duration of the study, then remove these services once the study was complete (with the understanding that study resources were not infinite and could not support services indefinitely)?

164 Chapter 3 | Methods

There are often competing ethical principles and frameworks to be considered in clinical and public health research434, with individuals, community and the wider population frequently having different needs and requirements. Ultimately, in consultation with supervisors, other researchers and providers in the study clinics, I made a decision to conduct an observational cohort study, without offering any additional study-supported interventions. The ethical reasons for this were based upon a consequentialist public health ethical platform434: whilst the health of individual study participants would not be adversely affected by their participation in the study (in comparison to what would have been expected through non-participation given the resources available in the routine health system), a substantially greater number of individuals in the study clinics, in

Blantyre and in other regions in Malawi and in Africa could be expected to receive wider benefit from the study. Critically, this ethical approach is dependent on study findings being widely disseminated and resulting in improved patient care and policies.

Although a utilitarian approach was adopted participants right to be protected from harm and treated with respect was protected. Moreover, from a justice perceptive, individuals (and study clinics) participating in the study were not unduly favoured by receiving additional unsustainable resources compared to non-participants. The

Declaration of Helsinki435 states that underrepresented groups (such as people infected with HIV, women and pregnant women) should be provided with appropriate access to participation in research. In this study, such underrepresented groups were eligible to participate.

In considering the issue of undertaking home tracing to ascertain outcomes of study participants, considerable ethical dilemmas were raised. Liberal ethical theories hold that individuals have rights to privacy and to be protected from harm at the expense of

165 Chapter 3 | Methods others (although in some cases, such as vaccination or quarantine, these rights are overridden)434. However, failing to achieve high rates of follow-up could compromise the integrity of the study and invalidate the results, negating any potential benefits from the study to the wider population. A pragmatic decision was ultimately taken that would protect the rights of participants, whilst maximising the opportunities for public health benefit. Following recruitment, participants were offered a choice of options for home- based follow-up in the event that they did not return to the study clinic: no home- follow-up; telephone tracing only; or home visits after an appointment was made by telephone. During the study, all participants gave written informed consent to home visits after a telephone appointment, suggesting that this issue was possibly not as problematic as we first supposed.

The cluster-randomised trial described in Chapter 6 raised particular ethical issues.

Concerns over privacy and confidentiality were not as prominent as in the cohort study as community members could make an informed decision to request HIV self-testing and home assessment and initiation of care through the community counsellors, and no incentives were offered to participants to take up home-based services. Additionally, study nurses were careful to maintain privacy during home visits: timings of visits were arranged beforehand by telephone and nurses did not wear any identifying uniforms or logos that might suggest they were providing HIV-related services.

The design of cluster-randomised trials are well-recognised to provide particular ethical challenges436. In this study, although individual-level consent for participation in the trial was sought from individuals undergoing HIVST and ART initiation, consent for participation was not sought from all community members during enumeration. We felt

166 Chapter 3 | Methods this was ethically justified as they would not receive any study interventions (if they did not opt to request HIV self-testing) and so would not be exposed to any potential harm.

A key area of debate centred on how results of the cluster randomised trial should inform policy and practice, and how participants in the control arm wider population could access study interventions following completion of the study. The Declaration of

Helsinki states that researchers and host country governments should make provisions for post-trial access for participants who need an intervention identified as beneficial in the trial435. Following the completion of the cluster-randomised trial, access to home initiation of HIV care (the intervention being assessed) was expanded to all 14 clusters receiving access to HIV self-testing for the remaining duration of the parent trial. Thus participants in the intervention arm and control arm received 24 months and 18 months of access to home initiation of HIV care respectively. This duration of availability should be sufficient to allow all study participants to access home initiation of HIV care if they wish.

Determining how the trial intervention could be made available to the wider HIV- positive community in Malawi and in other countries raises greater challenges.

Throughout the research process, we developed strong links with the HIV Department of the Ministry of Health of Malawi, who were very supportive of the study. Nevertheless, even if trial interventions are shown to be cost-effective, Malawi (and other countries) may not have capacity or economic resources to implement them immediately. Working with international organisations (such as WHO) to promote adoption of HIV self-testing and home initiation of care within guidelines and policy, and to better direct international sources of funding (e.g. PEPFAR) towards such home-based interventions

167 Chapter 3 | Methods may achieve wider implementation. It is encouraging to note that trial results were highlighted in the 2013 UNAIDS Report on the Global Epidemic437.

168 Chapter 4 | Prospective cohort study

4. Uptake of HIV testing and counselling and determinants of pre-treatment loss to follow-up among primary clinic attendees: a prospective cohort study

4.1. Introduction

As discussed in Chapter 2, HTC is the entry point to comprehensive care for HIV, including ART. However, despite massive expansion in availability of HIV testing services, uptake of HTC and knowledge of HIV status among individuals in sub-Saharan Africa remains disappointingly low, hindering efforts to achieve universal coverage of ART and prevention services.

WHO guidelines recommend a policy of provider initiated HTC (PITC) for all individuals who attend health facilities in countries with generalised HIV epidemics and Malawi recommends PITC for all facility attendees in the National Integrated HIV Care guidelines. Although high rates of uptake of PITC have been achieved as part of well- resourced research studies, there are few data that describe uptake when PITC is operationalized as part of routine care services. In primary care, where the majority of health services and HIV care are delivered in Malawi, a number of factors, including resource limitations, could hinder delivery and uptake of PITC.

Poor rates of linkage from HIV diagnosis to initiation of ART are also potentially a major barrier to universal coverage of ART, and the reasons for failure are poorly understood.

Previous cohort studies have suggested that there may be substantial barriers to ART initiation, with high rates of patient loss at all stages of the HIV care pathway330.

However, the majority of studies investigating barriers to ART have followed-up patients from within district- and tertiary-level health facilities and there is a pressing need to

169 Chapter 4 | Prospective cohort study determine the magnitude of and risk factors for patient loss to care at primary health care level. This could help in the development of interventions to improve rates of linkage into HIV care.

This Chapter presents data from a prospective cohort study assessing the uptake of PITC, and progression through the HIV care pathway, among primary clinic attendees in

Blantyre, Malawi. The objectives of the study were to:

a) Estimate the proportion of primary clinic attendees who underwent PITC

b) Describe patient flow through the HIV care pathway under programmatic

conditions for individuals who had been diagnosed HIV-positive

c) Estimate the proportion of clinic attendees diagnosed HIV-positive who

completed ART eligibility assessments and initiated ART.

d) Investigate risk factors for patient loss to care

e) Investigate mortality rates and associated risk factors in the pre-ART period.

This cohort study was selected as the initial study for this thesis (in addition to the qualitative study described in Chapter 5) to provide accurate and generalizable epidemiological data on uptake of HTC, rates of linkage to ART and risk factors for drop- out of care to inform interventions aimed at improving rates of linkage into HIV care.

4.2. Methods

4.2.1. Study design

A prospective cohort study of adults attending primary health care centres and undergoing HTC in Blantyre, Malawi.

170 Chapter 4 | Prospective cohort study

4.2.2. Study sites

Two primary health care facilities (Ndirande Health Centre and Chilomoni Health Centre) were selected as study sites (for descriptions of study clinics, see Chapter 3.1.1).

4.2.3. HIV care pathway in Malawian National Programme

At the time of this study (January to September 2011), the HIV care pathway in the

Malawian National Programme was laid out in the 3rd Edition of the Guidelines for use of

Antiretroviral Therapy in Malawi428 (described in detail in Chapter 2.6). Figure 4.1 shows care-seeking steps required for HIV-positive diagnosed participants in the national HIV programme, and ART eligibility criteria at the time of the study.

CD4 count measurement was not available on-site at either of the primary health care centres during the study. Instead, patients referred for CD4 count measurement were required to make at least two visits to Queen Elizabeth Central Hospital (QECH): the first to have a venous blood sample taken, and the second to collect results, which they would then take back to the assessing clinician at the primary health care centre.

171 Chapter 4 | Prospective cohort study

Figure 4.1: HIV care-seeking steps for patients in Malawian National HIV Programme under 3rd Edition of Guidelines for Use of Antiretroviral Therapy, 2008

HIV testing and counselling

Receives HIV- positive diagnosis

WHO stage 3 or 4 “Direct referral” Clinician performs WHO staging

WHO stage 1 or 2 Not eligible ART eligible

Referred for CD4 Referred for ART count

Blood taken for Pre-ART CD4 count education

Collect CD4 count ART initiation results >250 cell/mm3 ≤250 cell/mm3

4.2.4. Participant recruitment and assessment

In the 3-month period between January 2011 and April 2011, study Research Assistants extracted anonymised aggregate data on all clinic attendees (aged over 15 years old) from both clinic registers. This was to allow estimation of the number of clinic attendees to be used as a denominator in assessing uptake of HTC, and included data on number of attendees and sex and age distribution of attendees, disaggregated by clinic department.

172 Chapter 4 | Prospective cohort study

Research Assistants were stationed at HTC clinics in Ndirande and Chilomoni health centres and undertook exit interviews with all adult (>15 years old) clinic attendees who received HTC. Interviews were undertaken in a private room following the completion of all HIV testing procedures. Individuals who met study eligibility criteria (>15 years old and confirmed HIV infection on HTC) were invited to participate in the cohort study.

Because of clinic space constraints, a maximum of six participants per clinic per day were recruited. Individuals were excluded if they did not consent to participate, declined to disclose the result of their HIV status or required hospitalisation. Individuals who had previously tested positive for HIV, but had returned for a further test were eligible for participation. Once consent for participation had been obtained, Research Assistants confirmed reported HIV status by inspection of patient-carried health cards and clinic

HTC registers.

Research Assistants performed a baseline assessment that consisted of a socio- demographic questionnaire, extraction of clinical data from patient carried cards, measurement of height and weight and performed WHO clinical staging assessment using a checklist (this was done as WHO clinical staging was not always completed by clinic staff). In order to minimise risk of inadvertent disclosure of HIV status during follow-up, participants were invited to give consent for follow-up with two possible options: clinic-based follow-up only; or clinic-based follow-up with telephone- and home-based tracing in the event of non-return by the end of the follow-up period.

Following completion of all study procedures, participants were instructed to follow instructions given to them by clinic staff for on-going HIV care. Participants who reported not having received clear instructions were referred back to clinic staff for further instruction.

173 Chapter 4 | Prospective cohort study

4.2.5. Ascertainment of outcomes on the HIV care pathway during follow- up

Participants recruited to the cohort study underwent follow-up assessment at 3-months and 6-months following diagnosis of HIV. A schema of study procedures is shown in

Figure 4.2. Research Assistants identified study participants returning to the clinic for routine care by inspecting patient carried cards (health passports) of all individuals attending the general outpatient, and HIV/ART clinics, and by reviewing clinic registers to identify planned appointments. During clinic follow-up visits, Research Assistants assessed the steps completed on the HIV care-seeking pathway using a questionnaire and confirmed these by inspecting patient-carried records and HIV clinic registers.

174 Chapter 4 | Prospective cohort study

Figure 4.2: Schema of cohort study procedures

Primary clinic attendance (≥16 years)

Not offered HTC by routine clinic system

Offer of HTC by routine clinic system

Did not complete HTC in routine clinic system

Completion of HTC by routine clinic system

Study exit interview and consent to participate

Declined to participate in study

Baseline study assessment

Routine 3-month clinic Failure to attend clinic by 3 assessment & study follow- months: 3 home-tracing up interview attempts

Failure to attend clinic by 6 Routine 6-month clinic months: 3 home-tracing assessment & study follow- attempts up interview If not traced: lost-to-follow-up

For participants who did not return to the clinic by the end of each follow-up period, an initial telephone tracing call was made to ascertain vital status and to arrange a home tracing visit for participants who had consented to this. At home tracing visits, Research

175 Chapter 4 | Prospective cohort study

Assistants undertook a questionnaire and reviewed patient-carried records to ascertain clinical care steps completed. If the participant had died, household members were asked to provide the date of death and the participant’s patient-carried record for review. Participants who could not be traced after 3 attempts at telephone- and home- based tracing at both 3-months and 6-months were classified as lost-to-follow-up.

In this study, “retained in pre-ART care” was defined as: currently taking co-trimoxazole or having collected co-trimoxazole in the last 3-months; or having attended an ART support group meeting or pre-ART education classes in past 3-months; and not having died or been lost to follow-up. Participants who transferred care to another clinic, but who were successfully traced to home were defined as retained in care.

4.2.6. Laboratory methods

Clinic counsellors undertook HIV testing as part of routine clinical care and participants were recruited following the completion of all testing procedures. HIV testing algorithms followed national guidelines428 (serial testing using Determine 1/2, Abbott, Tokyo, Japan, and, if positive, Uni-Gold HIV, Trinity Biotech, Dublin, Ireland, with SD Bioline 1/2,

Standard Diagnostics, Korea if the first two were discordant).

4.2.7. Sample size calculations

Based upon clinic register records from the 6-months prior to the commencement to the study, we estimated that a combined total of approximately 620 adults undergo HTC and 160 are diagnosed HIV-positive each month in the two study clinics. Previous studies in Malawi suggest that approximately 50% of people testing HIV-positive in facilities meet eligibility criteria for ART261. We anticipated that, over a two-month recruitment

176 Chapter 4 | Prospective cohort study period, approximately 320 HIV-positive adults would be invited to participate in the study. Assuming that 85% of those eligible agreed to participate in the study, we estimated a sample size of 272 for this cohort study. From previous studies, we anticipated that approximately 40% of ART eligible participants would initiate ART during 6 months of follow-up. Table 4.1 below shows estimates of confidence intervals around the primary outcome of initiation of ART under different study scenarios.

Table 4.1: Sample size estimation and 95% confidence intervals for ART initiation

Proportion lost to follow- Proportion eligible for Proportion initiating ART Sample size up ART (95% CI)

272 5% 30% 40% (29.1 – 50.9%)

272 15% 30% 40% (28.4 – 51.6%)

272 5% 40% 40% (30.5 – 49.5%)

272 15% 40% 50% (39.8 – 60.2%)

272 5% 40% 50% (40.3 – 60.0%)

4.2.8. Data analysis

Uptake of PITC was expressed as a proportion of total adult clinic attendances and stratified by study clinic, department, sex and pregnancy status. Cohort participant demographic and clinical characteristics were stratified by sex and expressed as proportions, medians or means as appropriate and compared using chi square tests for categorical data and Wilcoxon rank sum tests for non-normally distributed continuous variables. Men and non-pregnant women were combined into one group as they both received HTC through the outpatient HTC clinic and had similar clinical characteristics, whereas pregnant women received HTC at the antenatal clinic and were diagnosed at an earlier stage of HIV infection.

177 Chapter 4 | Prospective cohort study

For the primary outcome of uptake of ART, time-at-risk in the study was estimated as the period between date of HIV diagnosis and date of ART initiation, or completion of 6- months follow-up or death in the event of non-initiation of ART. Data on participants who were lost-to-follow-up were right-censored at the midpoint between their last follow-up assessment and the date of their next expected follow-up visit. A Cox proportional hazard regression model was constructed to examine associations with initiation of ART. In constructing the multivariate regression model, variables significant at the p<0.10 level were included using a forward step-wise procedure. Proportional hazard assumptions were checked by inspecting Schoenfield residual plots. A Poisson regression model was constructed to describe mortality rates and investigate factors associated with death. Statistical analysis was undertaken using Stata v11.2 (Statacorp,

College Station, Texas).

4.2.9. Ethical Approval

Ethical approval for this study was obtained from the research ethics committees of the

College of Medicine of Malawi and Liverpool School of Tropical Medicine. All participants provided written informed consent to participate (or thumbprint witnessed by someone independent from the study team where illiterate).

4.3. Results

4.3.1. Clinic attendances during study period

There were 18,021 adult (16 years or older) attendance episodes at the two study clinics between January and April 2011 (Table 4.2). The majority of attendance episodes were by women (12,424/18,021 [68.9%]) with only 5597/18021 (31.1%) made by men. This

178 Chapter 4 | Prospective cohort study pattern remained after excluding attendances made by pregnant women (non-pregnant women-visits: 9625/15222 [63.2%], male-visits: 5597/15222 [36.8%]).

Table 4.2: Clinic attendances by department and sex during study recruitment period Chilomoni Health Centre Ndirande Health Centre

6348 11,673 Women Men Women Men

4357 (68.6%) 1991 (31.4%) 8067 (69.1%) 3606 (30.9%) General outpatients 2967 (68.1%) 1718 (86.3%) 6028 (74.7%) 3364 (93.3%) department

TB clinic 13 (0.3%) 15 (0.8%) 13 (0.2%) 26 (0.7%)

Antenatal clinic 460 (10.6%) n/a n/a 1019 (12.6%) n/a n/a HIV testing and 917 (21.1%) 258 (13.0%) 1007 (12.5%) 216 (6.0%) counselling clinic

4.3.2. HTC during clinic attendances

HTC was completed on 2398 of 18021 (13.3%) clinic attendances. The highest rate of

HTC was seen among pregnant women (1320/2799 attendances, 47.2%) but was substantially lower among men (474/5597 attendances, 8.5%) and non-pregnant women

(604/9625 attendances, 6.3%). Individuals attending Chilomoni were significantly more likely to receive HTC than at Ndirande (1175/6348 [18.5%] vs. 1223/11673 [10.5%], p<0.001). A higher proportion of male attendees at Chilomoni underwent HTC compared to Ndirande (258/1991 [13.0%] vs. 216/3606 [6.0%]. p<0.001) - Figure 4.3.

179 Chapter 4 | Prospective cohort study

Figure 4.3: Clinic attendance, uptake of HTC and HIV prevalence by sex and pregnancy status

Proporon of non-pregnant 70 63 aendees Proporon of aendees 60 tested for HIV HIV prevalence if tested

50 47

40 37

30 Percent (%) 24 21 20 15

8 10 6

0 Female (Non-Pregnant) Male Female (Pregnant)

4.3.3. HIV testing outcomes

444 (18.5%) of the 2398 attendees who received HTC were HIV-positive. HIV prevalence was 19.2% (208/1175) in Ndirande compared to 17.7% (235/1223) in Chilomoni

(p=0.340). HIV prevalence was lowest in pregnant women (199/1320 [15.1%]), followed by men (99/474 [20.9%]) and non-pregnant women (146/604 [24.2%]; p<0.001). HIV prevalence was significantly higher in men tested at Ndirande (55/216 [25.5%]) than at

Chilomoni (44/258 [17.0%], p=0.025).

4.3.4. Baseline characteristics of cohort participants

280 participants (76.9% of those eligible) were recruited into the prospective cohort study (80 of the total of 444 individuals with confirmed HIV infection were not eligible

180 Chapter 4 | Prospective cohort study because six participants had already been recruited from each clinic on that day and 84 patients did not consent to participation).

Seventy-five per cent of participants were women (211/280), of whom 57.3% (121/211) were pregnant (Table 4.3). Pregnant women were more likely to be younger (median 25 years, interquartile range [IQR]: 21-30 vs. 31 years, IQR: 27-37; p<0.001), unemployed

(86/121 [71.1%] vs. 71/159 [44.7%]; p<0.001), to have previously tested for HIV (70/121

[57.9%] vs. 59/159 [37.1%]; p=0.001), to be married (110/121 [90.9%] vs. 106/159

[66.7%]; p<0.001) and to be from the wealthiest household quartile (26/121 [27.7%] vs.

32/159 [23.7%]; p=0.014) than men and non-pregnant women.

Men and non-pregnant women appeared to receive HTC at a more advanced stage of illness than pregnant women demonstrated by differences in self-rated general health

(30/159 [18.9%] vs. 2/121 [1.7%]; p<0.001) and study assessed WHO clinical stage 3 or 4 disease (129/159 [81.8%] vs. 45/121 [37.2%]; p<0.001) – Table 4.3. Median CD4 count among participants who completed CD4 count testing was significantly higher in pregnant women (373 cell/mm3, IQR: 236-527) than in men and non-pregnant women

(240 cells/mm3, IQR: 143-376); p=0.039.

181 Chapter 4 | Prospective cohort study

Table 4.3: Baseline characteristics of cohort study participants Men and non- Pregnant Characteristic pregnant % % P-value women women Total 159 100 121 100 Age (years) – median (IQR) 31 27-37 25 21-30 <0.001 Marital status Married 106 66.7 110 90.9 <0.001 Divorced 24 15.1 2 1.7 Never married 18 11.3 8 6.6 Widowed 11 6.9 1 0.8 Employment status Not in formal employment 71 44.7 86 71.1 <0.001 In formal employment 88 55.3 35 28.9 Literacy Able to read a newspaper 129 81.1 98 81.0 0.976 Illiterate 30 18.9 23 19.0 Wealth quartile Poorest quartile 42 31.1 14 14.9 0.014 Next poorest quartile 35 25.9 23 24.5 Next wealthiest quartile 26 19.3 31 33.0 Wealthiest quartile 32 23.7 26 27.7 Previously tested for HIV Yes 59 37.1 70 57.9 0.001 No 100 62.9 51 42.1 Tuberculosis history Current 6 3.8 0 0.0 0.092 Previous 9 5.7 6 5.0 Never 144 90.6 115 95.0 Self-rated general health Excellent 21 13.2 48 39.7 <0.001 Good 38 23.9 51 42.1 Fair 70 44.0 20 16.5 Poor 30 18.9 2 1.7 WHO clinical stage* Stage 1 or 2 30 18.9 76 62.8 <0.001 Stage 3 or 4 129 81.1 45 37.2 Received same day routine WHO 89 56.0 45 37.2 0.002 staging ART eligibility at 6-months♯ ART eligible 136 85.5 47 38.8 Not ART eligible 7 4.4 12 9.9 <0.001 Not fully assessed 16 10.1 62 51.2 CD4 count (median, IQR) cells/mm3 § 240 143-376 373 236-527 0.039 *Study assessed WHO stage: participants were independently staged by research assistants. ♯ART eligibility defined as meeting national ART criteria (CD4<250 cells/mm3 or WHO stage 3 or 4) on routine clinic assessments §National guidelines recommend CD4 count measurement for HIV-infected individuals in stage 1 or 2 only. CD4 count was performed through the routine clinic system for 99/280 participants

182 Chapter 4 | Prospective cohort study

4.3.5. Awareness and readiness for ART

Participants had high awareness of HIV/AIDS, with nearly half (128/280, 45.7%) knowing at least one family or household member who had been diagnosed HIV-positive.

Similarly, 31.4% (88/280) knew of at least one family member who had taken ART and

28.2% (79/280) knew at least one family or household member who had died or had been suspected to have died of HIV/AIDS. Readiness to take ART was high, with 98.5%

(267/271) agreeing that they would take ART if referred by a health worker and 87.1%

(243/279) agreeing that they felt ready to start taking ART.

4.3.6. Factors associated with same-day referral for ART

108 participants (38.6%) were referred on the same-day as HIV diagnosis to initiate ART by clinic staff, 153 (54.6%) were referred for further ART eligibility assessment (CD4 count) and 19 (6.8%) reported having received no clear instructions from clinic staff.

Factors significantly associated with same-day referral for ART initiation on univariate analysis (Table 4.4) included male sex (OR 2.6, 95% CI: 1.53-4.67), worse self-reported general health (p trend <0.001) and being in WHO Stage 3 or 4 (OR: 2.37, 95% CI: 1.40-

4.01). Pregnant women were significantly less likely to be directly referred for ART (OR

0.22, 95% CI: 0.13-0.38).

183 Chapter 4 | Prospective cohort study

Table 4.4: Univariate and multivariate associations with direct referral to initiate ART

Odds Ratio (95% CI) Adj. Odds Ratio (95% CI) P-value

Male sex 2.68 (1.53-4.67) 0.87 (0.43-1.75) 0.697 Pregnant (if female) 0.22 (0.13-0.38) 0.29 (0.13-0.63) 0.002 Age – years (continuous) 1.06 (1.03-1.10) 1.03 (1.00-1.07) 0.042 BMI (kg/m2 - continuous) 0.85 (0.79-0.92) 0.95 (0.86-1.04) 0.057 Illiterate 1.28 (0.77-2.04) In formal employment 1.91 (1.17-3.11)§ WHO stage± Stage 1 or 2 1 1 0.997 Stage 3 or 4 2.37 (1.40-4.01) 1.00 (0.50-2.01) Self reported general health Excellent 1 1 Good 1.90 (0.91-3.96) 2.43 (0.98-5.98) 0.036 Fair 3.44 (1.68-7.05) 2.62 (1.15-5.93) Poor 10.04 (3.81-26.45) 4.55 (1.41-14.60) Clinic where HTC performed Chilomoni Health Centre 1 1 0.001 Ndirande Health Centre 1.61 (0.99-2.62) 2.82 (1.54-5.17) First time tested for HIV No 1 Yes 1.26 (0.77-2.04) Household food insecurity* No 1 Yes 1.06 (0.59-1.91) ± WHO staging assessment as performed by study research assistants *Defined as skipping a meal in the last 2-weeks to ensure food for children BMI: body mass index

On multivariate analysis, self-reported “poor” (adjusted OR: 4.55, 95% CI: 1.41-14.60) and “fair” (adjusted OR: 2.62, 95% CI: 1.15-5.93) general health remained significantly associated with direct referral for ART, whereas pregnancy (adjusted OR: 0.29, 95% CI:

0.13-0.63) was associated with referral for further ART eligibility assessment (CD4 count). Being in WHO stage 3 or 4 was nolonger independently associated with direct referral for ART (adjusted OR: 1.00, 95% CI: 1.00-2.01), nor was male sex (adjusted OR:

0.87, 95% CI: 0.43-1.75). The loss of association for WHO clinical stage 3 or 4 is likely a consequence of collinearity between male sex and advanced stage. Attendees at

Ndirande were more likely to be referred directly for ART initiation compared to those at

Chilomoni (adjusted OR: 2.82, 95% CI: 1.54-5.17).

184 Chapter 4 | Prospective cohort study

4.3.7. Evaluation of patient flow through HIV care pathway

Following HIV diagnosis, health workers completed WHO clinical staging on the same day for 134 (47.8%) participants with the remainder asked to return to the clinic on an alternative day when a clinician was available, or were not given clear instructions

(Figure 4.4). By 6-months, a further 45 (16.1%) participants had been WHO staged by health workers, meaning that 36.1% of participants were unstaged within six months of

HIV diagnosis. Overall, by 6-months of follow-up, 61/179 (34.1%) of those staged were assessed by health workers as being in WHO stage 3 or 4.

185 Chapter 4 | Prospective cohort study

Figure 4.4: STROBE flowchart of participant outcomes

186 Chapter 4 | Prospective cohort study

In total, 153/280 (54.6%) participants were referred by health workers for CD4 count measurement, of whom 99 (64.7%) reported having had blood taken and 81 (52.9%) received results. The median CD4 count was 294 cells/mm3 (interquartile range: 173-473 cells/mm3) and 40/81 (49.4%) had a CD4 count of <250 cells/mm3. Median CD4 count was lower in men and non-pregnant women (240 cells/mm3, IQR: 143-376) than in pregnant women (373 cells/mm3, IQR: 236-527, p=0.039).

By six months, 146/280 (52.1%) participants had completed ART eligibility assessments.

91/280 (32.5%) participants met National ART criteria by completion of six months follow-up (CD4<250 cells/mm3 or WHO stage 3 or 4), with 30 (10.7%) completing assessments but not ART eligible and 159 (56.8%) having not completed eligibility assessments.

Failure to complete ART eligibility assessment was highest in pregnant women (95/121,

78.5%) compared to men (27/69, 39.1%) and non-pregnant women (37/90, 41.1%, p<0.001). Participants who were not WHO staged by clinic staff on the same day as HIV diagnosis (32/146, 21.9%) were significantly less likely to complete ART eligibility assessments than participants who were WHO staged (74/134, 55.2%; p<0.001 – Figure

4.5).

187 Chapter 4 | Prospective cohort study

Figure 4.5: Completion of HIV care pathway steps if WHO staged on same day as HIV diagnosis

P<0.001 P<0.001

P=0.007 P=0.445

P=0.007

4.3.8. Initiation of ART during cohort follow-up

The 280 cohort participants were followed-up for a total of 82.6 patient-years. By 6- months, 87/280 (31.1%) had initiated ART, with 32 (11.4%) lost to follow-up (Figure 4.3).

ART was initiated in 25/89 (28.1%) of participants who were assessed as being in WHO stage 1 or 2 by study WHO staging, and in 27/45 (60.0%) of those who were WHO stage

3 or 4. Overall, participants who did not receive same-day WHO staging assessment were substantially less likely to initiate ART (35/146 [24.0%] vs. 52/134 [38.8%], p=0.007

– Figure 4.6). Twenty per cent (6/30) of participants who were assessed by clinicians as not ART eligible initiated ART.

Of the 91 participants who had met national ART eligibility criteria through routine clinic assessments over 6-months, 68 (74.7%) initiated ART. The median number of days to

ART initiation was 47 (interquartile range [IQR]: 21-74). Men (34 days, IQR: 19-57) had a

188 Chapter 4 | Prospective cohort study

shorter time to ART initiation than pregnant (median 56 days, IQR: 47-87), and non-

pregnant women (median 49 days, IQR: 21-85). This may have been because pregnant

women were more likely to be assessed in WHO stage 1 or 2 and to have a higher

median CD4 count (Table 4.3).

Kaplan-Meir plots (Figure 4.6) showed that probability of ART initiation was higher in

men and non-pregnant women compared to pregnant women (log rank test: p<0.001).

Additionally participants who had fair or poor self-rated general health had higher ART

initiation compared to participants who had self-rated excellent or good general health

(log rank test: p<0.001).

Figure 4.6: Kaplan-Meier plots of initiation of ART by (a) sex and pregnancy status; and (b) self-rated general health

a)

189 Chapter 4 | Prospective cohort study b)

4.3.9. Factors associated with ART initiation

On univariate analysis (Table 4.5), men (hazard ratio [HR]: 1.29, 95% CI: 0.81-2.04), older participants (HR: 1.05, 95% CI: 1.04-1.07), participants who had fair self-rated general health (HR: 8.48, 95% CI: 3.62-19.88) or poor self-rated general health (HR: 13.70, 95%

CI: 5.44-34.51), participants in WHO stage 3 or 4 (HR: 3.25, 95% CI: 1.89-5.60) and participants who received same-day WHO staging (HR: 1.66, 95% CI: 1.08-2.55) were more likely to initiate ART. Pregnant women (HR: 0.20, 95% CI: 0.11-0.40), participants who were assessed as not ART eligible (HR: 0.16, 95% CI: 0.07-0.36) and who didn’t complete ART eligibility assessments (HR: 0.07, 95% CI: 0.04-0.12) were less likely to initiate ART.

On multivariate adjusted analysis, failure to complete ART eligibility assessment

(adjusted HR: 0.11, 95% CI: 0.06-0.21) or assessment as not ART eligible (adjusted HR:

0.16, 95% CI: 0.07-0.36) were significantly associated with non-initiation of ART. Fair

(adjusted HR: 2.87, 95% CI: 1.11-7.40) or poor self-rated general health (adjusted HR:

3.16, 95% CI: 1.11-8.93) were associated with ART initiation. However, male sex

190 Chapter 4 | Prospective cohort study

(adjusted HR: 1.29, 95% 0.78-2.14), pregnancy in women (adjusted HR: 0.58, 95% CI:

0.28-1.21), and age (adjusted HR: 1.01, 95% CI: 0.99-1.04) were no longer associated with ART initiation.

Table 4.5: Univariate and multivariate associations with ART initiation

Univariate Multivariate hazard 95% CI p 95% CI p hazard ratio ratio Sex and pregnancy Female non-pregnant 1 <0.001 0.316 Male 1.29 0.81-2.04 1.29 0.78-2.14 Female pregnant 0.20 0.11-0.40 0.58 0.28-1.21 BMI (kg/m2) – continuous 0.83 0.77-0.89 <0.001 0.96 0.89-1.04 0.312 Age (years) - continuous 1.05 1.04-1.07 <0.001 1.01 0.99-1.04 0.341 Employment status Not in formal employment 1 In formal employment 1.25 0.81-1.90 0.306 Literacy Able to read a newspaper 1 Illiterate 0.81 0.46-1.43 0.459 Wealth quartile Poorest quartile 1 0.372 Next poorest quartile 1.01 0.55-1.83 Next wealthiest quartile 0.75 0.39-1.43 Wealthiest quartile 0.61 0.31-1.20 Self-reported general health Excellent 1 <0.001 1 0.104 Good 2.21 0.86-5.64 1.70 0.65-4.49 Fair 8.48 3.62- 2.87 1.11-7.40 Poor 13.70 19.885.44- 3.16 1.11-8.93 Study assessed WHO clinical 34.51 stage Stage 1 or 2 1 1 Stage 3 or 4 3.25 1.89-5.60 <0.001 1.00 0.54-1.85 0.992 Health facility Chilomoni Health Centre 1 1 Ndirande Health Centre 0.66 0.42-1.02 0.056 0.94 0.46-1.95 0.876 Received same-day WHO 1.66 1.08-2.55 0.019 0.72 0.36-1.43 0.349 staging Eligibility assessment at 6 months ART eligible 1 <0.001 1 <0.001 Not ART eligible 0.16 0.07-0.36 0.19 0.08-0.44 Not fully assessed 0.07 0.04-0.12 0.11 0.06-0.21

191 Chapter 4 | Prospective cohort study

4.3.10. Retention in pre-ART care for participants who did not initiate ART during follow-up

At 6-months follow-up, 193 participants had not initiated ART. 138 (72%) of these had not been retained in pre-ART care, with 3 (1.6%) deaths recorded. There was a similar risk of failure of retention in pre-ART care between men (24/35, 68.6%), non-pregnant women (35/50, 70.0%) and pregnant women (76/108, 70.4%). Of the 55 participants who were retained in pre-ART care, 28 (50.9%) reported that they had collected cotrimoxazole within the last 3-months. There were no differences in rates of retention in pre-ART care between participants who were staged on the same day as HIV diagnosis (21/82, 25.6%) and those who were not (34/111, 30.6%; p=0.445).

4.3.11. Cohort mortality and factors associated with death

By 6-months, 15/280 (5.4%) participants had died (mortality rate 14.8 per 100 person- years, 95% CI: 8.9-24.6 per 100 person-years). Eleven of fifteen deaths (73.3%) occurred following ART initiation. The median time from HIV diagnosis to death in participants who initiated ART was 62.5 days (IQR: 48-121).

On univariate analysis (Table 4.6), male sex (incidence rate ratio [IRR]: 1.52, 95% CI:

0.51-4.53), older age (IRR: 1.06, 95% CI: 1.02-1.11) worse self-rated general health (IRR:

4.60, 95% CI: 2.20-9.59), being in WHO clinical stage 3 or 4 (IRR: 3.96, 95% CI: 0.89-

17.55) were associated with death. Pregnant compared to non-pregnant women (IRR:

0.25, 95% CI: 0.05-1.23), higher body mass index (IRR: 0.70, 95% CI: 0.57-0.86) and failure to complete ART eligibility assessments (IRR: 0.38, 95% CI: 0.13-1.07) were associated with a lower risk of death. On multivariate analysis, only worse self-rated general health (adj. IRR: 3.27, 95% CI: 1.31-8.15) remained significantly associated with death.

192 Chapter 4 | Prospective cohort study

Table 4.6: Univariate and multivariate associations with death

Univariate Multivariate incidence 95% CI incidence rate 95% CI rate ratio ratio

Sex and pregnancy Female non-pregnant 1 1 Male 1.52 0.51-4.53 1.15 0.36-3.64 Female pregnant 0.25 0.05-1.23 1.04 0.17-6.37 BMI (kg/m2) – continuous 0.70 0.57-0.86 0.87 0.70-1.08 Age (years) - continuous 1.06 1.02-1.11 1.02 0.97-1.07 Employment status Not in formal employment 1 In formal employment 1.91 0.68-5.38 Literacy Able to read a newspaper 1 Illiterate 0.66 0.15-2.92 Wealth quartile Poorest quartile 1 Next poorest quartile 1.61 0.38-6.73 Next wealthiest quartile 0.65 0.11-3.92 Wealthiest quartile 0.32 0.03-3.09 Self-reported general health (per 1- 4.60 2.20-9.59 3.27 1.31-8.15 unit increase) Study assessed WHO clinical stage Stage 1 or 2 1 1 Stage 3 or 4 3.96 0.89-17.55 0.98 0.20-4.81 Health facility Chilomoni Health Centre 1 Ndirande Health Centre 0.77 0.27-2.16 Received same-day WHO staging 1.63 0.58-4.59 Eligibility assessment at 6 months ART eligible 1 1 Not ART eligible* ------Not fully assessed 0.38 0.13-1.07 1.11 0.37-3.35 * Study assessed WHO stage § Insufficient events to allow estimation CI: confidence interval

4.4. Summary of findings

There were two major novel findings in this study. Firstly, in contrast to the high uptake of HTC recorded when PITC has been introduced as part of clinical trials or observational

193 Chapter 4 | Prospective cohort study studies, completion of PITC by facility attendees was low when evaluated under routine programmatic conditions in the primary health care system (13%), with especially low rates achieved for men and non-pregnant female attendees. Secondly, whilst rates of linkage from HIV diagnosis to ART initiation were as poor as seen in previous cohort studies and in meta-analyses, identification of the low rates of completion of ART eligibility assessment (52% at 6-months) and the strong association between failure to complete ART eligibility assessment and patient loss to care are key to understanding reasons for suboptimal linkage into care and potential areas for intervention.

Critical bottlenecks in the HIV care pathway were in the completion of ART eligibility assessments with a high proportion of participants failing to complete WHO clinical staging and CD4 count measurement in a timely manner. This is likely to reflect the complex and time-consuming nature of the WHO clinical staging system assessment for healthworkers, and the difficulties patients experience in accessing CD4 count measurements, which had to be carried out at the city’s central hospital and required multiple visits. These factors will be investigated in detail from the patient and providers’ perspective in the qualitative study described in Chapter 5.

Linkage into ART was high (although still not complete) among individuals who were promptly assessed for ART eligibility and found to be ART eligible (75%) However, ART initiators experienced long treatment delays with a median of 47 days between HIV diagnosis and ART initiation. This is substantially longer than the 7 days recommended in updated Malawian National HIV Care Guidelines174, raising concerns that HIV infected individuals are exposed to unnecessary periods without ART, with attendant risks of opportunistic infections, death or transmission of HIV to others.

194 Chapter 4 | Prospective cohort study

Participants who did not meet ART eligibility criteria had extremely poor rates of retention in care. When compiled to show patient loss at each stage of the HIV care cascade (Figure 4.7), the large losses accrued at every step are starkly apparent, emphasising that considerable efforts are needed to retain patients in care until ART is initiated.

Figure 4.7: Loss to HIV care in primary clinic attendees, Blantyre

20000 100% 18000 16000 14000 12000 10000 8000 Number of adults 6000 4000 13% 19% 2000 52% 75% 75% 0 Aended Received Diagnosed Completed ART eligible Iniated ART clinic HTC HIV-posive ART eligibility assessment

Percentages are calculated from previous column

4.5. Limitations of Study

There were a number of limitations to this study. CD4 counts were not measured in this study, as this would have constituted an intervention in the care-seeking process and may have biased estimates of linkage into care upwards. Instead, practitioners working in the routine care system requested CD4 counts in accordance with Malawian national

HIV care guidelines. This meant that that true ART eligibility status on the basis of CD4 count could not be determined for all participants. The advantage of this approach is that participant flow through the HIV care pathway was observed without study

195 Chapter 4 | Prospective cohort study procedures influencing participant behaviour, giving an accurate and generalizable assessment of outcomes experience under routine conditions.

Only patients attending at two primary health care centres were studied, although these clinics are the two largest primary care providers of ART in Blantyre. Factors influencing linkage into care may be different in other clinics or health systems that are structured differently. A maximum of six participants per clinic per day were recruited due to time and space constraints. This may have resulted in some selection bias in participant recruited. Reasons for undergoing HTC may have influenced linkage to ART, but sufficient data on testing behaviour was not captured in this study to allow this to be explored. With rigorous telephone and home tracing, loss to follow-up was only 11% at

6-months, which is lower than the majority of other studies investigating pre-ART patients, but may conceal missed events such as initiation of ART or death. Mortality may be an important competing risk for pre-ART attrition and larger studies are required to give further insights into the extent of and factors associated with death.

It is possible that some participants who dropped out of care may “recycle” into the HIV care pathway at a later date, a phenomenon that has been reported from other studies in sub-Saharan Africa329. This may mean that, if participants were followed for a longer period of time, uptake of ART may be higher than reported here. Nevertheless, the high rates of early drop out raise concerns about the effectiveness of current HIV programme and health centre systems to ensure retention in care and unbroken linkage into ART. To complete assessment of HIV care outcomes, rates of virological suppression 6-months after ART initiation could have been measured. However, under national HIV care guidelines at the time of the study, HIV viral load measurement was not recommended and was not available through the routine care system.

196 Chapter 4 | Prospective cohort study

4.6. Findings related to other studies

Upwards of 50 studies have been published describing retention in care during various stages of the HIV care pathway in sub-Saharan. Nevertheless, no study published prior to this work had reported on patient loss at all steps from HIV diagnosis to ART initiation.

In this study, only 13% of facility attendees underwent provider-initiated HTC during the study period. This contrasts sharply with data from results of clinical trials and implementation research. In a systematic review of 44 studies reporting on outcomes of provider initiated HTC programmes, Roura et al153 found that uptake of provider initiated HTC was generally high (upto 99%,) with the highest rates achieved among hospitalised inpatients and TB diagnosed patients. Lower rates of uptake were reported from studies among outpatients and STD clinic attendees. In the three studies438-440 where provider initiated HTC was offered among outpatient populations, uptake ranged from 24% in Ethiopia439 to 53% in Zambia440 during the implementation of study PITC activities. However, perhaps reflecting the findings of this study, Chakaya found that in

Malawi, after study activities had finished, rates of uptake of PITC in TB patients nearly halved441. Operationalizing successful interventions may be challenging as considerably less staff, logistical support and financial resources may be available once study activities are completed442. The reasons for low uptake of provider-initiated HTC under programmatic conditions are an important and under-researched topic and will be considered in the qualitative study described in Chapter 5.

This study obtained remarkably similar estimates for retention in care through the stages in the HIV care pathway following HIV diagnosis as summary estimates in the three systematic reviews described in Chapter 2.7328-330. In total 52% of participants

197 Chapter 4 | Prospective cohort study completed ART eligibility assessments according to national guidelines (either CD4 count or WHO, or both if WHO stage 1 or 2) by 6-months. In the three systematic reviews, summary estimates for completion of ART eligibility assessments lay between 57% and

72%. However, it is important to note that the majority of studies included in the systematic reviews that assessed completion of ART eligibility were based in South

Africa, where access to CD4 on-site count measurement is more common than in

Blantyre. In the one previous study from Malawi (based in the rural Thyolo District) that reported ART eligibility assessments outcomes, 46% of HIV-diagnosed individuals in

WHO stage 1 or 2 completed CD4 count measurement366. This is similar to the 53% completion rate seen here in individuals referred for CD4 count measurement.

Overall, 75% of participants assessed as being ART eligible initiated treatment within 6- months, a slightly higher estimate than the 63-68% summary estimates obtained by the three systematic reviews, but lower than the 86% reported from a small study of 88 patients in Karonga District, Malawi355.

4.7. Implications for future studies

In this study, which was the first to prospectively follow-up clients though all steps of the

HIV care pathway from clinic attendance to ART initiation, a number of key stages for future study and intervention were identified. Factors that may contribute to clinic attendees’ reluctance to undergo HTC, and healthworker and health system barriers will be explored in the linked qualitative study in the next Chapter. Future operational research studies and clinical trials that compare different models of provision of provider initiated HTC in facilities will be required to refine current approaches and to determine how best to translate findings into routine clinical practice. Research should

198 Chapter 4 | Prospective cohort study be linked to policy and guidelines, promoting a greater commitment to expansion and resourcing of PITC to ensure that all clinic attendees are offered HIV testing

In conjunction with expansion of facility-based PITC, community-based interventions that encourage men and non-pregnant women to test for HIV at an earlier stage of illness and more frequently, such as regular door-to-door and workplace-based testing programmes are required. These could utilise novel testing technologies such as oral test kits to increase acceptability and ease of use. There have been very few clinical trials that have evaluated the effectiveness of these HTC approaches in improving uptake of testing or linkage to ART.

A rethinking of the current approach to assessment of ART eligibility is urgently required.

Reliance on the insensitive, time-consuming and complex WHO clinical staging system in the primary health care system is a major barrier to timely linkage to ART, propagating high-levels of patient loss from the HIV care pathway. Point of care CD4 count measurement systems offer an opportunity to rapidly assessment ART eligibility, but are expensive and are not yet widely available. Alternative clinical assessment tools that maximise sensitivity and specificity and can be rapidly used immediately following HIV diagnosis should be investigated. Evaluation of the accuracy of one such novel tool will be described in Chapter 7.

Novel approaches that promote greater linkage between the stages in the HIV care pathway are required, including same-day, same site HTC and eligibility assessment, with prompt and complete ART initiation. Studies that investigate interventions aimed at improving linkage will be vital as there are increasing moves towards implementing

199 Chapter 4 | Prospective cohort study

“test-and-treat” approaches to HIV delivery. The evaluation of one potential approach, home initiation of HIV care following self-testing, will be described in Chapter 8.

4.8. Conclusions

Significant failings in provision of provider-initiated HTC and timely completion of ART eligibility assessments have a major impact on primary clinic patients’ ability to link to antiretroviral therapy, hindering efforts to achieve universal knowledge of HIV status and coverage of ART. Rigorous evaluation of interventions aimed at increasing uptake of

HTC and linkage to ART are urgently required.

200 Chapter 5 | Qualitative study

5. Understanding reasons for loss to HIV care at primary health care level: a qualitative study with patients and health workers

5.1. Introduction

The prospective cohort study described in the previous Chapter identified suboptimal rates of completion of PITC and subsequent linkage into HIV care in the primary care clinics in Blantyre, and highlighted the importance of ART eligibility assessments as a major contributor to patient loss to care. These failings could seriously hinder efforts to achieve universal coverage of ART and reduce HIV-associated morbidity, mortality and transmission. Understanding the underlying reasons for low rates of completion of PITC and linkage into care could assist with identifying modifiable factors and development of the interventions.

This Chapter describes a qualitative study undertaken with patients and healthcare providers that was conducted to better understand the reasons for poor rates of completion of PITC and reasons for retention and loss to care before ART initiation. A socio-ecological model (described in Chapter 2.8) was drawn upon to conceptualise linkage from HIV diagnosis to ART initiation and to identify key areas for intervention.

5.2. Methods

5.2.1. Study design

A qualitative study nested within a prospective cohort study investigating patient flow from HIV diagnosis to initiation of ART at primary care level.

201 Chapter 5 | Qualitative study

5.2.2. Rationale for qualitative study

Qualitative methods offer opportunities to understand the perspectives and experiences of different groups of individuals, including both patients and providers443. In this study, qualitative methods were used to help understand the underlying pathways to, and reasons for, successful completion of HTC and linkage to ART.

5.2.3. Qualitative methods

A Malawian social science Research Assistant (Daniel Mwale) was recruited to undertake qualitative interviews and assist with analysis and interpretation of results. Daniel had grown-up in Blantyre, and was familiar with the study communities. He had completed a

Bachelors degree in social sciences at the University of Malawi and spoke English and

ChiChewa fluently. He had previously worked in social science research projects related to HIV in Blantyre. Daniel received intensive training prior to commencement of the study. I provided training sessions on the study objectives and procedures, including identification and recruitment of potential participants, informed consent procedures, completion of study forms and the use of audio-recording equipment. With support from the social science team in MLW, I provided training in conducting sensitive qualitative interviews for individuals infected with HIV, with particular attention paid to issues of confidentiality and stigma.

Prior to the commencement of the study, I developed a socio-ecological theoretical framework to inform the design of interview topics and guides, drawing upon existing literature regarding barriers and facilitators to HIV care in resource-limited settings (See

Chapter 2.18 for description of theoretical framework) 410. The socio-ecological model has advantages of including factors that are both internal and external to the patient to

202 Chapter 5 | Qualitative study explain health-related decision-making444. External factors are composed of the physical, social and political environment. The internal and external factors that underpin health behaviour can then be understood in terms of dynamic interactions between levels, including the individual-, community-, institutional-, and societal level444.

Together, Daniel and I (with additional input from other members of the qualitative research team in MLW) developed participant interview guides (with separate guides for patients and providers), drawing upon the theoretical framework. The interview guides incorporated themes on potential barriers and facilitators to linkage to ART. These themes were developed from reviews of the existing literature on care-seeking for HIV and other illnesses (such as tuberculosis – see Chapter 2.7). At a weekly meeting the interview guides were reviewed by Daniel and I and modified to incorporate emerging themes from the data. Interview guides were translated into Chichewa and back translated into English by a Chichewa speaker independent from the study.

5.2.4. Study participants

Between January and April 2011 newly diagnosed HIV-positive adults (16 years or older) at Ndirande and Chilomoni Health Centres were recruited and followed up for six months to assess their completion of steps on the HIV care pathway (Chapter 4). From

July to September 2011, Daniel purposively sampled participants from this group of patients and invited them to participate in the qualitative study. There were two groups of patients sampled: those who had successfully linked from HIV diagnosis to ART initiation during six months of follow-up, and those who had not. Working in the study clinics, Daniel identified potential participants by reviewing study files to determine whether ART had been initiated during 6-months follow-up.

203 Chapter 5 | Qualitative study

From previous studies in South Africa and Malawi353, 445, we anticipated that gender and age might be important factors determining progression through the HIV care pathway.

Therefore, we selected patients with a range of identities, including: men, women and pregnant women; and older and younger people. According to available resources, we pragmatically set a total number of participants for inclusion at 30 (15 ART initiators and

15 individuals who did not initiate ART).

We additionally interviewed healthworkers who provided HTC, pre-ART and ART services at the two clinics. Two cadres of health workers were recruited: counsellors (who undertook HTC and provided pre-treatment psychosocial support and education); and nurses, midwifes and clinical officers (who undertook ART eligibility assessments, and initiated ART). We set out to interview a minimum of five healthworkers from each clinic.

Daniel conducted semi-structured interviews in ChiChewa using the interview guides.,

Interviews took place either in the health facility or in the participant’s residence. Daniel audio-recorded interviews and additionally recorded observations in a notepad. Audio recordings were transcribed into Chichewa scripts at a dedicated qualitative data centre at the Malawi-Liverpool-Wellcome Trust Clinical Research Programme by a transcriber. A translator who had not transcribed the original audio file undertook translation to

English. A sample of 5% of English transcripts were back-translated to Chichewa by a independent translator (who had not completed the original translation) and were reviewed for consistency of translation by Daniel. At weekly meetings and with reference to written notes, Daniel and I discussed emerging findings from the qualitative interviews and identified novel areas for exploration. We additionally modified interview guides when required and discussed whether saturation of data had been achieved.

204 Chapter 5 | Qualitative study

Analysis of qualitative data

I undertook analysis of the qualitative analysis data using a framework approach446. The framework approach to analysis was chosen as this deductive approach fitted well with the objectives of the study and allowed analysis to be grounded within the previously conducted qualitative research from other settings, the quantitative data obtained from the cohort study describe in Chapter 4 and the predesigned theoretical framework

(Chapter 2.8.1). Pope describes the framework approach as:

“Starting deductively from pre-set aims and objectives. The data collection tends to be more structured than would be the norm for much other qualitative research and the analytical process tends to be more explicit and more strongly informed by a priori reasoning.”447

In the framework approach, a series of steps are followed, including: familiarisation, identification of a thematic framework, indexing, charting and interpretation. In analysis of data from this study, these step were followed. Initially, as an ongoing process during and after data collection, Daniel and I reviewed transcripts to establish familiarity with the data. We compared qualitative texts with the theoretical framework to identify themes in the data that matched with those in the theoretical framework. Using this data, a detailed coding index of themes relevant to the data was developed within the

NVIVO-9 software platform (QRS, Melbourne, Australia). NVIVO-9 is essentially a relational database software programme that allows qualitative text to be mapped to user-defined themes and key terms. The index of themes was modified as data emerged from the qualitative interviews, and following weekly feedback discussions between

Daniel and me.

205 Chapter 5 | Qualitative study

Following completion of all qualitative interviews, I repeatedly read transcripts in full to gain familiarity with the data. I then coded initial samples of interview transcripts by relating data-items (i.e. words, phrases, sentences, paragraphs and interactions with the interviewer) to index themes within the coding framework in NVIVO-9. The results were compared with a sample of transcripts coded by a second investigator (EEM) to assess the validity of the coding index and to identity further themes for inclusion in the coding system. I then systematically reviewed all qualitative transcripts and coded data (i.e. words, phrases, sentences and interactions with the interviewer) against the final index of terms using NVIVO-9. Each data item could be indexed against one or more coding theme. For example a passage in which a participant discussed the difficulties they experienced in accessing clinics due to transport costs could be coded against index terms such as “poverty”, “transport”, “access” and “inconvenience”.

Analysis of qualitative data involved thematic charting to aid understanding and interpretation. Thematic charting involves iteratively sorting and organising data coded under index terms into an understandable and meaningful structure that honestly represents the experiences of study participants and best represents the data. This was done by grouping index themes (and their related data) under common subheadings and headings within NVIVO-9 and was built upon the theoretic framework developed at the onset of the study. This abstracted thematic chart was discussed with Daniel to ensure that my understanding of key themes truly represented the views and experiences of participants. I then summarised the data under key themes and headings and related data to the various stages of the HIV care pathway, using quotes from participants to illustrate findings and interpretation.

206 Chapter 5 | Qualitative study

5.2.5. Ethical considerations

The College of Medicine of Malawi Research Ethics Committee and Liverpool School of

Tropical Medicine granted ethical approval for the study. All participants gave written informed consent.

5.3. Results

5.3.1. Baseline Characteristics

Between May and September 2011, 30 newly diagnosed HIV positive adults were recruited to the qualitative study and participated in semi-structured interviews (15 from Chilomoni and 15 from Ndirande). Thirteen participants were men and 17 women, of whom 10 were pregnant. Participants’ ages ranged from 16 to 45 years. In total, 15 participants had successfully initiated ART during the 6-months follow-up, whilst the remaining 15 participants had not initiated ART (Table 5.1). Ten healthcare workers were also recruited to the study and participated in in-depth interviews. Of these, five were

Counsellors, three were Nurses and two were Clinical Officers.

Table 5.1: Characteristics of participants recruited to qualitative study Method Participant type Gender Age Range

6 Men 25-41 years Individuals who initiated ART 9 Women (6 ANC) 19-37 years

7 Men 21-45 years Individuals who did not initiate ART Semi-structured 8 Women (4 ANC) 18-31 years interviews 2 Men 27-33 years Nurses & Clinical Officers 3 Women 26-35 years

1 Man 25 years Counsellors 4 Women 23-41 years

ANC: antenatal clinic

207 Chapter 5 | Qualitative study

5.3.2. Conceptualising determinants of patient loss to care and delay

Thematic analysis identified key determinants of dropout and delay at the individual-, socio-cultural-, programmatic/policy- and structural levels. Participants who did not initiate ART experienced negative effects from factors at all of these levels, whereas participants successful in initiating treatment were frequently able to utilise factors from one or more of these levels to their benefit.

Failure to link into care was then multifactorial and negative factors at more than one level frequently interacted, having a multiplicative effect on risk of dropout. For example participants who were poor (and so less able to cope with the expense of clinic visits) and who were not able to draw upon support from networks of friends and families reported greater difficulty in staying in care than participants with only one of these factors. All patients and providers were influenced by the health system and HIV care policies, although individual behaviour in response to this environment varied considerably.

5.3.3. Linkage to ART: 1) HIV testing and counselling

As reported in the cohort study described in Chapter 4, only a small proportion (13%) of individuals attending primary health care centres received PITC, with particularly low rates of completion among men and non-pregnant women. Less than half of women- visits to antenatal clinics were accompanied by completion of PITC. Thus, despite WHO and Malawian national HIV care guidelines recommending that PITC be offered to all health facility attendees regardless of presenting complaint, for individuals attending primary health care centres, completion of PITC was the exception rather than the rule.

208 Chapter 5 | Qualitative study

In qualitative interviews, participants reinforced these findings, with most men and non- pregnant women stating that they were not offered PITC during clinic attendance, but instead had attended the facility under their own volition to request HTC. Thus for men and non-pregnant women, HIV testing in the primary health care centres operated under a traditional VCT model, rather than the recommended PITC model.

For men and non-pregnant women, the first step on the HIV care pathway was then making a proactive decision to undergo HTC. This decision-making process was strongly influenced on one side by awareness of ailing health and the potential benefits of ART treatment, and on the other by fear of the consequences of being diagnosed HIV- positive. Decision to attend the clinic and request HTC was made in private and not discussed with partners, family members or friends.

Men and non-pregnant women’s principal motivation to attend the clinic and request

HTC was a prolonged period of declining health that impacted upon ability to work or carry out household chores.

“I am one of the people who does hard jobs: I stay in the sun the whole day building

houses. From last year I was having stomach pains often, so I was just taking

painkillers. Then I reached a stage of having fever regularly. So I was working one

day and I felt weak and dizzy. I had heard that near my house there is VCT so I

decided I have to go for a test.” Male participant, Chilomoni, 39 years, initiated ART

Therefore HTC was seen as the first step towards regaining strength and avoiding an impending catastrophic social event, such as loss of employment or abandonment by a partner.

209 Chapter 5 | Qualitative study

Men worried that, with increasingly outwardly obvious signs of failing health, they would be perceived to be less masculine and less respectable, with the assumption made that they could not provide for their family and were not able to perform sexually.

“I was a person and I had a good looking body. Even when I was walking people

knew that aah that man that is passing there, he is really a man. […] So I was

examining myself and could see that my body was not all right. That is what made

me think that aah its better to go where I hear that they do some tests - maybe I

have a disease. That is why I mustered up boldness to go for testing.” Male

participant, Ndirande, 25 years, didn’t initiate ART

As seen in this quote, the man reported having previously been proud of his physical attractiveness and prowess. But, with increasingly obvious outwardly visible signs of HIV, he recognised he needed to seek help. His emphasis on male characteristics and his attractiveness is at odds with his presumed perceptions of people living with HIV as being weak, frail and non-masculine. He described the process of resolving to attend the clinic to request HTC as “mustering up boldness”, emphasising that HIV remained something shameful and to be feared.

The majority of men and non-pregnant women interviewed reported delaying attending the clinic and requesting HTC. Delay was strongly influenced by socio-cultural norms, particularly gender norms and conceptualisations of the consequences of HIV diagnosis.

Emphasising the private manner in which decision to test was made, many participants described this period in terms of a private struggle between their fears of the consequences of not testing (abandonment, poverty, illness, death) and their fears of

210 Chapter 5 | Qualitative study the consequences of HIV diagnosis (stigma, loss of status). Men and non-pregnant women reported being spurred into action to test once their health deteriorated to the point they felt others might recognise them to be HIV-positive. By doing so, they attempted to regain control over who had knowledge of their HIV-status and the manner in which it was disclosed.

Non-pregnant women cited fear of being identified as HIV-positive by other community members (and the accompanying loss of status – often expressed as anticipated stigma) as a common reason for delaying HTC. However, they also reported fear of being accused by their husbands of being unfaithful and so bringing HIV into the relationship.

They often had to seek permission from their husbands prior to attending for HTC.

“The problem with men is that if we tell our husbands ‘you have to attend for a test’,

they are rude. They say ‘Eeh. Why are you telling me that? You are the one who has

caught the disease and you have transmitted it to me.’” Female participant,

Ndirande, 32 years, initiated ART

This reflected men’s position of decision-making power within the household, but also men’s understanding that if their wife was diagnosed HIV-positive, this would mean that they too were HIV-positive. Thus men prohibited their wife from being tested as a means of controlling when they learned their own status.

The primary motivators towards HTC for non-pregnant women were, like for men, strongly influenced by prevalent gender norms within the family and community. In particular they expressed a desire to improve their ailing health to allow them to fulfil their household and family duties. They noted that if they did not take action to ensure

211 Chapter 5 | Qualitative study they were healthy, there would be no-one else to cook, look after the children or to provide petty cash for the household by selling produce at the market. They feared this would result in their husband leaving them for another woman who was a better provider.

“Have you seen men doing household chores like washing dishes [laughs]? In the

morning, when the husband wakes up, you have to give him water to bathe – and

he comes in the night! […] We have responsibility of taking care of the children so

we say ‘aah if I die, will my husband take care of my children?’ They have the

freedom to have another wife, taking the children to that other woman. That’s why

we say ‘let us go to the hospital!’” Female participant, Ndirande

In this quote, the women voices her acute fear of abandonment, raising concerns that her husband will leave her for another women who would be a better provider if she was unable to work through illness. The motivation for HTC was then not principally about her own health, but based on the perceived consequences of illness, which would impact heavily upon her and her children. There is also an tacit acceptance that her husband’s needs (including sexual needs) take priority over all else in the household.

Therefore, through seeking HTC, she has taken proactive steps to ensure that she is not abandoned.

Pregnant women on the other hand had a very different journey to HTC. Most of the pregnant women interviewed stated that they did not have symptoms that led them to suspect they might be HIV-infected. Instead, they reported that midwives offered HTC to them when attending for routine care at antenatal clinics (ANCs). The perception of the pregnant women interviewed was that there was little opportunity to refuse HTC as

212 Chapter 5 | Qualitative study healthworkers were seen as powerful, senior members of the community. There was also a perception that the government had decreed that all pregnant women should be tested for HIV, and this was interpreted to mean that there would be no opportunity to refuse testing.

“[At the antenatal clinic] everyone was told that she is supposed to do the test. They

test everyone for the disease of what do you call it, HIV? So when I was told that you

should have a test I didn’t worry about what was going to happen. I was already

knowing that it was there and so I knew when I went there it would not be a difficult

thing.” Pregnant women, Ndirande, 30 years, didn’t initiate ART

The pregnant women interviewed here highlights the fact that provider initiated HTC during antenatal care was normalised, with women using social networks to inform each other what happened in the clinic. PITC also seemed to be acceptable, with this women suggesting that HTC would “not be a difficult thing”. This routine offer of HTC at ANC is in sharp contrast to the experience of men, who noted that they had to “muster up boldness” to decide to attend for testing.

Counsellors and nurses working in the ANC held the view that HTC was mandatory for pregnant women and that the Ministry of Health had promoted this policy. They were supportive of the provider initiated HTC approach, reasoning that it would help prevent mother-to-child transmission of HIV.

“As of now, we give the first priority [for testing] to a woman who is pregnant. If the

woman is expectant, she is expecting a baby…we want to protect the unborn child

213 Chapter 5 | Qualitative study

so that they should be born free from the virus.” Counsellor, female, 23 years,

Ndirande.

However, clinicians working outwith the ANC in the general outpatients department, although recognising the value of HTC, took a much less proactive approach towards HIV testing: they limited their requests for HTC to individuals in whom they had a high clinical suspicion of HIV, meaning that PITC was not fully implemented outside of the

ANCs.

5.3.4. Linkage to ART: 2) Eligibility assessments

Following HIV-positive diagnosis, the counsellor who had performed the test referred the patient to a clinician for ART eligibility assessment. Participants reported that this system meant that they would start the day at the clinic by queuing to see the clinician for HTC or management of their presenting ailment (or at the antenatal clinic if pregnant), then queue to see the counsellor for HTC and finally (often late in the day), queue again to see the clinician for ART eligibility assessment. Eligibility assessment involved firstly the clinician performing WHO clinical staging and secondly referral for measurement of CD4 count for patients who were in WHO stage 1 or 2.

WHO staging was frequently not completed on the same day as diagnosis of HIV- infection. If it was late in the day, or the clinician had left the clinic (absenteeism was commonly reported by patients but not by health workers), patients were advised to return on another day. Participants frequently described making repeat visits to the clinic before being WHO staged, or dropping out of care at this point as they could no longer manage to keep returning to the clinic due to pressures of time, employment and expense.

214 Chapter 5 | Qualitative study

Even when the clinician was available to complete eligibility assessments, there were considerable problems with their completion of WHO clinical staging. Clinicians reported that the process of staging was difficult and time-consuming.

“Sometimes we have more than 80 clients, so out of that number, to do WHO

staging for everybody, it seems as if we are delaying others. So we just tell them in a

short-cut way and refer them for CD4 count to ensure that everybody should feel

that they have been helped in the right time.” Nurse, female, ANC, 35 years

If they suspected a complex clinical problem (that could signify a WHO stage 3 or 4 illness), they often found that the staging condition could not be confirmed due to lack of diagnostic laboratory or radiological services at the primary health care centre.

Instead of formally WHO staging patients according to guidelines, clinicians reported that they would often base their eligibility assessment on a superficial assessment of the patient’s wellbeing and to save time in busy clinics, would refer patients for CD4 count measurement to provide a “definitive” assessment of ART eligibility.

“You see that the person is weak when he is entering into the room and you already

know. Yes, we say that you shouldn’t judge a book by its cover, but somehow you are

able to see the way it is.” Clinician, male, Ndirande, 33 years

However, if patients were referred for CD4 count measurement to complete their ART eligibility assessment this was a particularly problematic, being a common point of dropout from the HIV care pathway. At the time of the study, CD4 counts could only be performed at the city’s central hospital and required at least two visits to complete

215 Chapter 5 | Qualitative study measurement (one to give blood and the second to collect the result). The CD4 count machine was frequently out of service and results were misplaced or lost, necessitating repeat visits. Each journey for CD4 measurement would require substantial expenditure, often borrowed from family members or friends and participants reported feeling guilty about borrowing money from others to complete CD4 count measurement.

“I went there [Central Hospital] and they said ‘The machines are not working’.

Another time, I went and reached the point that I did a blood test. After that they

told me ‘Come on Tuesday and hear your results’. On Tuesday I begged the

transport money from my boss. When they looked at the result they said ‘that blood

which was taken that time … the machine was not working. You will have to come

again’.” Male participant, Ndirande, 41 years initiated ART

5.3.5. Linkage to ART: 3) Pre-ART care

In contrast to the private, independent role taken when accessing HTC, participants described adopting a dependent care-seeking role during and after completion of ART eligibility assessments. In particular they found themselves relying on the financial, emotional and practical support of networks of family members (for example to provide money and childcare for clinic attendances), friends (especially members of their church congregations and friends who were also HIV-positive or who were already taking ART) and employers (to provide time off work and money).

“My friend who I am working with, she encourages me because she is taking the

treatment. She tells me ‘do you see how I am looking compared to the way I was

looking in the past? Is it the same? You have to start receiving the treatment so you

can look the way I am.’” Female participant, Chilomoni, 29 years initiated ART

216 Chapter 5 | Qualitative study

“My family and my relatives they all agree because they have seen other people

reach the point of being finished [died]. But when they started receiving this

treatment they recovered and their bodies become as it was before. So they are the

ones who encourage me the most.” Female participant, Chilomoni, 29 years initiated

ART

“Talking of the job, I can say that it helps me. I still go to work there and they helped

me [gave money] to go to the clinic so that I could receive the treatment that I

needed.” Male participant, Ndirande, 41 years, initiated ART

“That time we were living far from the clinic […]. But because of my wish to receive

the treatment, I wanted to attend. I was borrowing money from friends.

The process of adopting a dependent care-seeking role, necessitated that participants disclose their HIV status to partners, family members and friends from whom they required support. There was a range of responses and reactions to the need for disclosure that did not appear to be related to any social or demographic characteristics of participants. Some participants reported that they found this process easy and valued the support they received from others. Whereas some participants reported that they were not keen to disclose, but felt obliged to by the need to obtain money to attend the clinic for pre-ART care and to identify a treatment supporter.

If I thought they would give me money, I would have to ask three days before the

journey so that I would know if they would give me the money. Eeh these are big

217 Chapter 5 | Qualitative study

problems if you are far from the clinic.” Male participant, Ndirande, 31 years, didn’t

initiate ART.

This enforced dependent role extended to numerous facets of participants’ lives. In post-test counselling sessions, they were advised as to what food they should eat, to whom they should disclose their status (with most being advised to disclose to a family member who would act as their “treatment guardian”) and even how they should have sexual relationships. These were perceived as a set of rules that were imposed by healthworkers, which, if not strictly followed, would result in ART being withheld.

Participants, especially older adults, felt that these rules infantilised them. This perception was reinforced by the fact that they had to attend pre-ART lessons accompanied by a relative, bringing back memories of attending school as a child.

“The things they tell you not to do, like do not eat certain foods and drinks like beer…

and cigarettes - they say ‘do not smoke cigarettes’. Maybe you like drinking beer,

but they say that you should stop. You have to stop it.” Male participant, Ndirande.

Initiated ART

“So they gave me advice that I should avoid doing childish things like that. If I do, I

should use condoms. But we are not youths we are elderly people.” Male participant,

Ndirande. 45 years, didn’t initiate ART

In these two quotes, the two participants report feeling disempowered by the pre-ART counselling process and resent being told that they should eat healthily, not smoke or drink and avoid sex. Both men convey their resentment. Given that nearly all counsellors in the HIV care system are female, they would likely feel having to take such instruction

218 Chapter 5 | Qualitative study from women to be insulting. In the second quote, the man explicitly conveys his displeasure at being told what to do by someone younger than him. These interactions emphasise the prevailing patterns of power-relations in Malawian society, where men, and especially older men are held in the highest esteem. As the quotes highlight, the men take slight when they feel that their power is being usurped by someone perceived to be inferior to themselves. The lack of responsiveness to mens’ perceived needs and challenge to their authority is perhaps one reason why men feel uncomfortable attending facilities for HTC and HIV care.

All participants found the requirement to attend for pre-ART with a treatment supporter challenging. Despite guidelines stating that nomination of a treatment supporter was not essential for ART initiation, patients perceived this to be an absolute requirement imposed by health workers.

“If you do not bring your guardian when you are learning you are sent back. They

say ‘No. Go and get your guardian and bring her here.’ So, if I did not bring my

guardian here, I would not be given the treatment.” Female participant, pregnant.

Ndirande, 19 years, initiated ART

Participants found that their interactions with the health facilities and providers in the pre-ART period had a major bearing on their ability to link to ART, and their ability and willingness to remain in care if not yet ART eligible. Clinics were extremely busy and crowded, resulting in an unwelcoming environment and a disincentive to continued attendance. Providers were often perceived to be rushed and unapproachable and did not explain clearly what was provided in pre-ART care, or when and how they should initiate ART.

219 Chapter 5 | Qualitative study

“But I have done the test and I have been coming here at the clinic telling the doctor

that I am sick. The doctor just gives me panadol or aspirin, even LA [malaria

treatment]. You will just receive it and off you go home not knowing what to do.”

Female participant, Chilomoni, 30 years, didn’t initiate ART

For their part, health workers saw themselves as gatekeepers in the primary clinic system, determining who should and shouldn’t receive treatment according to national guidelines. ART was seen as a precious resource that required careful rationing to prevent misuse.

“On this point we make very strict procedures because these medicines are not

commonly found. We look at the client’s personal life and their problems and we see

that if we ignore these problems, we will kill the client [by giving them ART].” Nurse,

female, Ndirande, 26 years

Study participants who had been successful in initiating ART reported that they experienced little difficulty in linking from pre-ART care to ART initiation. This could be because they had more advanced HIV infection and were prioritised by clinicians for expedited initiation onto ART. In both clinics, clinicians reported that an informal

“monthly ART list” existed, limiting the number of patients who would be started on ART every month to avoid overburdening the ART clinic. Patients whom clinicians perceived to be particularly in need of ART (usually patients with obvious outward signs of advanced HIV infection) were prioritised for the list, and clinicians discussed having to

“squeeze” another patient onto the monthly list.

220 Chapter 5 | Qualitative study

“So we first see how these people look physically. Maybe that patient is very weak

and he has no power. So we say that it is good that he should start [ART] quickly and

he can be shifted [up the list.]” Nurse, female, Chilomoni, 26 years

Participants who did not initially meet eligibility criteria, or were not prioritised by clinicians for ART initiation, found it increasingly difficult to avoid dropping out of pre-

ART care. They were less likely to have advanced HIV infection and self-identify themselves as urgently needing treatment. They saw little benefit from the time consuming and expensive repeat clinic visits and pressures of work and family commitments became their main priority. Additionally, some participants reported that they felt that their continuous requests for support and money were having a negative impact on their relationships with family, friends and in the workplace.

5.4. Summary of findings

This qualitative study with patients and providers has identified a number of important and novel findings that help to understand the underlying reasons for suboptimal uptake of HTC and poor rates of linkage into HIV care in primary health cares. The main finding was that PITC was not fully implemented within the primary health care system and men and non-pregnant women sought HTC after prolonged periods of ill health, following a traditional VCT approach. This helps to explain continuing high rates of late presentation for ART and emphasises the fact that, in addition to fully resourcing supporting PITC in facilities, additional approaches that encourage earlier and more frequent HIV testing that are convenient, confidential and minimise stigma are urgently required.

221 Chapter 5 | Qualitative study

Participants and providers raised significant concerns about the suitability of the current

ART eligibility assessment process. The WHO clinical staging system was noted by practitioners to be overly-complex, time-consuming and dependent on access to diagnostic tests that were not available at primary health care level. This meant that instead of completing a thorough clinical staging assessment, healthworkers either didn’t complete staging or made a subjective assessment of ART eligibility based on outwards appearance, lending support to the quantitative data from the cohort study described in Chapter 4. For patients however, completion of CD4 count measurement was extremely problematic, requiring multiple, expensive facility visits and dependence on family members, friends and work colleagues. Given these factors, it is no surprise that rates of dropout were so high during the ART eligibility assessment stage and, if universal access to ART is to be achieved, alternative, client- and healthworker-friendly approaches must be developed and implemented.

Also important was the extent to which societal and individually-held gender norms influenced care seeking. For a number of years, it has been well recognised that fewer men than women presented for ART initiation and with more advanced immunosuppression, with consequent worse treatment outcomes. Whilst this study supports the finding that pregnant women are better able to access HTC through maternity services (being the only site in clinics where PITC was routinely implemented), potentially resulting in early entry into care, the reluctance of men to undergo HTC until signs of HIV were outwardly visible or impacted upon sexual prowess or capacity to work helps explain their delayed entry into care.

222 Chapter 5 | Qualitative study

5.5. Limitations

Key strengths of this study were the use of qualitative methods and the inclusion of both patients’ and health workers’ voices. Nevertheless, social desirability may have biased participant and health worker responses. In particular, we found that health workers were understandably reluctant to overtly criticize the HIV care programme, limiting our ability to draw conclusions about how the structure of the programme influenced the care they provided.

Although this study was conducted at two of the busiest primary care clinics in Malawi, it is possible that our findings may not be applicable to other settings that have differently structured health systems or socio-cultural environments. In particular, we noted that participants were reluctant to directly discuss their expectations for treatment and care following HIV diagnosis. The may reflect the Malawian context, where health workers are held in high esteem and their actions rarely questioned. Socio- cultural factors were important in determining linkages to ART. In other countries, with distinct socio-cultural traditions, the impact of support from net- works of family and friends may differ.

In this qualitative study, participants were purposively sampled from specific demographic groups (by age group and sex) and from individuals who had been successful and unsuccessful in initiating ART. This sampling strategy allowed in-depth inquiry of the experiences of a broad range of representative participants. However this strategy may have missed other important groups. Additionally, no participants aged over 45 years were included, limiting applicability of findings to this group.

223 Chapter 5 | Qualitative study

Individuals were only recruited during follow-up for the cohort study and so had already undertaken a number of care-seeking steps and overcame numerous barriers to care.

Identification of participants at earlier stages in the HIV care-seeking pathway (for example prior to HTC, during first clinic attendance for HTC) would have been challenging, but may have provided a wider view of the barriers and facilitators to accessing HTC and HIV care than the approach that was taken. Additionally, because participant identifiers used in the cohort study were not recorded during recruitment for the qualitative study, it was not possible to link data relating to participants in the qualitative study with participants in the cohort study. Therefore, we couldn’t determine which participants in the qualitative study were ART eligible and which were not. As such, some participants in the qualitative study may not have been ART eligible according to national guidelines. Availability of this data may have provided improved understanding and context to the reported barriers to ART.

Credibility in qualitative research refers to the process by which researchers seek to ensure that their study measures or tests what is actually intended. According to

Merriam, credibility deals with the question, “How congruent are the findings with reality?”448. Although findings from the study were not checked with study participants due to time constraints, a weekly meeting was held between myself and the Malawian research assistant, who was able use his knowledge of Malawian society and culture, as well as their own experiences of health-seeking within the Malawian health care system in Blantyre and to provide thorough, rich and deeper insights into participant accounts, frequently prompting new lines of enquiry and interpretations of participant accounts.

We were able to triangulate participant accounts of care-seeking with qualitative data obtained from in-depth interviews with healthworkers and with the findings of the

224 Chapter 5 | Qualitative study cohort study reported in Chapter 4. We were encouraged to find that our main findings were well supported by these sources, strongly emphasizing the credibility of the participant accounts. For example, a major finding of this study was that men and non- pregnant women sought HTC after a prolonged period of declining health and followed a more traditional VCT route to testing instead of receiving PITC. This was supported by quantitative data showing that uptake of PITC was extremely low among these groups and data from interviews with healthworkers in the general outpatient department, who described their extremely high workload and perception that only a high clinical suspicion of HIV should prompt referral for HTC.

Because I was not fluent in the ChiChewa language, I was not able to conduct interviews myself, potentially missing the opportunity to pick up on spoken, as well as unspoken and non-verbal communication during in-depth interviews. We attempted to mitigate this as far as possible by ensuring that the research assistant conducting interviews kept an interview diary in which he noted pertinent events and interpretations during interviews. This diary was reviewed in tandem with interview scripts during the weekly research meeting and triangulation of the data sources may have assisted with understanding the true underlying meanings in phrases. Nevertheless some important or sensitive information may have been missed.

A thorough approach to data capture (audio recorded interviews and use of research assistant interview notes) and transcription/translation (conducted within a dedicated qualitative research centre with experienced Malawian research assistants) was central to overcoming language and cultural barriers. This was supported by ensuring the

Malawian research assistant who conducted interviews was involved in the process of analysis.

225 Chapter 5 | Qualitative study

Criticality in qualitative research refers to the degree to which the research methods, analysis and interpretation are carried out in a reflexive manner, cognisant of the assumptions, privileges and knowledge background of the investigator that might influence the process. In Malawian society, doctors are held in high esteem and individuals may be reluctant to be outspoken about mistreatment or dissatisfaction.

Moreover, in Blantyre, because of long-established links with the College of Medicine of

Malawi and overseas university departments, including Liverpool School of Tropical

Medicine, research has been carried out in the study communities and clinic for at least

20 years. This means that individuals and the community in general may be familiar with the process of medical research, a cultural currency that is unlikely to be completely neutral. Exposure to medical research could for example promote, to a greater or lesser degree, a sense of altruism and trust in the medical and research professions, encouraging greater disclosure and candidness. Alternatively, this may breed distrust and dissatisfaction with research and health services in general, hindering efforts to achieve true or rich representations of care seeking experiences.

The research assistant who conducted participant interviews (Daniel) was born and grew up in in Blantyre and had completed under- and post-graduate training in social sciences. He was fluent in ChiChewa and English. Although he came from the middle- classes of Blantyre, he had considerable experiencing working in temporary employment in Ndirande township and had lived in Likhabula for a number of years. He was familiar to healthworkers in the study clinics, having worked on other qualitative studies previously. As such, he was trusted by participants and healthworkers and able to elicit candid and rich accounts.

226 Chapter 5 | Qualitative study

5.6. Finding in relation to other studies

A small number of previous studies have investigated barriers and facilitators to retention in the HIV care pathway in sub-Saharan Africa. A systematic review in 2012 summarised risk factors, barriers and facilitators to linkage into HIV care in 25 cohort or cross-sectional studies386. In all studies, factors influencing into care were determined by questionnaire or extraction of data from clinical record. The main factors impeding retention in the HIV care pathway were transport costs, distance from clinic, waiting times and stigma. Disclosure of HIV-positivity to a partner was associated with enrolment in HIV care in a number of studies. The authors conclude that, although a number of common barriers to retention in care were identified, many of these related to the clinic, HIV care programme and health system structure, interventions will likely be required to be tailored to local circumstances and to the needs of the local population.

A smaller number of qualitative studies have investigated barriers and facilitators to retention in the HIV care pathway. In a study among individuals who had disengaged from pre-ART and ART care (defined as missing at least one scheduled visit and had not returned to the clinic within three months) in Nigeria, Tanzania and Uganda387, participants highlighted costly and long travel distances, harsh treatment from staff and long waiting times as important barriers. These were all factors identified by participants in Blantyre. Participants in Nigeria, Tanzania and Uganda also reported experiencing at shame having missed appointments, leading to reluctance to return to the clinic for fear of being ostracised by healthworkers and other patients. Ware et al speculate that this anticipated negative reaction signifies a breakdown in the sense of connectedness between patients, community, healthworkers and health system. In Blantyre, although

227 Chapter 5 | Qualitative study evidence of anticipated negative feelings were not identified among individuals who had disengaged from care, participants’ concerns about being infantilised by clinic procedures run in a similar vein. These findings together support a conclusion that HIV care programmes and health systems in general are not responsive to the needs of patients, emphasising that as well changing procedures, shifts in individual, intuitional and programmatic culture are required.

Additional reasons for disengagement from care in Nigeria, Tanzania and Uganda were the competing demands made on participants’ time between attending the clinic and home- and work-life. Important work, cultural and family obligations (e.g. travelling for work, funerals) took precedence over clinic attendance. In Blantyre, family and cultural obligations were not specifically identified as barriers to retention in care; conversely, participants emphasised the emotional and financial support they received from their social networks. This difference between the studies may be explained by different cultural and social traditions in these countries, and the fact that participants in Blantyre were predominately young, urban dwellers, perhaps without the same family commitments.

In a study from Karonga District in Malawi388, participants registered at a rural ART clinic underwent in-depth interview. Similar to findings from Blantyre, participants

(particularly men) in Karonga reported delaying seeking HTC until HIV infection was advanced with outwardly visible signs. Men in Karonga also delayed seeking care for HIV and were reluctant to attend the clinic for fear of losing respect in the community. This differs slightly from findings in Blantyre, where men who had been successful in initiating ART reported receiving support from families and the workplace to continue accessing care. The difference may reflect the social context between the rural and

228 Chapter 5 | Qualitative study urban communities. Women (and particularly pregnant women) were keen to learn their

HIV status and access ART as a means of improving their health and maintaining their social standing within the household and community. This closely reflects findings from

Blantyre, where women were keen to regain health to avoid marital break-up.

A separate study from Kibera in Tanzania449, found that, similar to in Blantyre, individuals were motivated to access ART because they recognised the potential health gains would be beneficial to them. Likewise, they sought help from networks of friends and family (including people who were already taking ART) to provide financial, logistical and emotional support during the pre-ART period, lending support to the findings from

Blantyre. In Blantyre, in contrast to the highly dependent role adopted in the ART eligibility assessment and pre-ART periods, decision to attend the clinic and request HIV testing was made privately and, although validation of their decision was not sought from others before testing, behaviour was strongly influenced by the anticipated response from partners and other community members. Anticipated stigma and fear of ostracism are well-recognised barriers to uptake of HIV testing450 and challenging to overcome when designing HTC programmes451. Interventions to improve linkage into care should be designed with these issues in mind.

Participants in a study from rural Uganda identified transport costs as the major barrier to ART initiation452. Transport costs were also noted to be significant for participants in

Blantyre. A previous study from rural Malawi found that HIV-infected TB patients’ acceptance of ART was strongly influenced by their ability to afford transport to the hospital372. In a separate study, TB patients were reported to spend up to 574% of their monthly income during the pre-treatment period453. Together, these studies provide

229 Chapter 5 | Qualitative study strong evidence that the multiple facility visits required of patients in the pre-ART period is detrimental to efforts to achieve universal access to ART.

In a study from Kenya390, a predominant reason for failure to initiate ART was reported to be fear of side effects, an issue that was not spontaneously raised by participants in any interviews in Blantyre. It is not clear from these data why fear of side effects would be of major concern to participants in in Kenya, but not in Blantyre. It is possible that pre-ART education sessions may have differed in content and emphasis between the settings: the authors from Kenya note that patients received extensive counselling to ensure they ate well before taking ART to avoid side effects.

Fewer studies have examined healthcare workers perceptions of offering PITC and pre-

ART care. In a study from South Africa, TB nurses highlighted their chronic frustrations with lack of resources and stock-outs of test kits, as well as concerns that they had insufficient time available to offer and complete PITC for every patient454. Further studies, from a wider range of HIV care settings, are required to help explain healthworker-related reasons for disengagement from the HIV care pathway.

5.7. Implications for future studies

The findings obtained from participants and healthworkers in this qualitative study raise a number of important issues for improving uptake of HCT and linkage into HIV care.

Low rates of completion of PITC, especially among men and non-pregnant women, were noted in the cohort study described in Chapter 4. Analysis of qualitative data from these groups in this study provided important insights into the underlying reason for this phenomenon. In particular, delayed facility attendance for HTC until HIV was advanced with outwardly visible signs, was a consequence of fear of loss of status. Healthworkers

230 Chapter 5 | Qualitative study were complicit in this by only offering PITC to men and non-pregnant women in whom they had a high suspicion of HIV. Furthermore, the extremely busy clinics, unwelcoming environment and rushed healthworkers meant that facility attendance was not an appealing prospect for clients and contributed to participants’ and providers’ failure to complete HTC earlier.

These findings show that futher studies are required to investigate intervention and alternative implementation strategies for improving uptake of PITC in health facilities.

Additionally, they demonstrate the urgent need to investigate strategies to facilitate regular uptake of HTC outside of health facilities for all individuals living in high prevalence communities. Such services should be convenient to allow easy, yearly access, and targeted specifically towards groups underserved by current PITC, especially men and non-pregnant women. They should also be confidential and situated in locations where men both men, women and couples feel comfortable. Finally, there should be strong links to ongoing HIV care and prevention services after HTC has been completed.

The novel finding that HIV diagnosed individuals shifted from an active to passive care- seeking role following diagnosis of HIV raises a number of important areas for study. In particularly, similarly designed qualitative studies among patients seeking care for other diseases (e.g. tuberculosis, infective hepatitis) would worthwhile to determine whether this was a phenomenon associated with HIV infection, or related to facility and health systems factors. Additionally, although this was touched upon in this study, a greater and more-depth understanding through further qualitative studies of the personal, interpersonal and social drivers for this shift in roles adopted is required.

231 Chapter 5 | Qualitative study

The numerous barriers to retention in HIV care identified by participants and healthworkers, particularly during the ART eligibility assessment process emphasises the need to design and test interventions to improve linkage into HIV care. Given the concerns raised by patients about the challenges experienced in finding sufficient time and money to make repeated facility visits, and the concerns of healthworkers over their limited ability to complete WHO clinical staging assessments in a timely or accurate manner, a rethinking of the current pathway to treatment is required. As participants who were successful in initiating ART cited the assistance of networks of friends, families and work colleagues as major contributors, consideration should be given to interventions that actively seek the support of these groups.

5.8. Conclusions

The main finding from this study was that delays and drop out from the HIV care pathway in primary care were strongly determined by factors at multiple levels.

Individuals made private decisions to delay attending for HTC because of fear of the consequences of their HIV-positive status being revealed, but after diagnosis became rapidly dependent on social support networks to navigate the ART eligibility assessment and pre-ART care periods. Improving upon the currently low rates of linkage to ART330 will require a comprehensive rethinking of how HIV care is delivered.

In this study, poor rates of uptake of PITC and drop-out from the HIV care pathway were multifactorial, with lack of social networks, economic vulnerability and fear of the consequences of disclosure of HIV status major contributors. At the programme level, overly busy clinics, inappropriate ART eligibility screening tests and unintegrated HTC and ART programmes resulted in failure of linkage. To rapidly improve linkage, a rethink of the current HIV care model in primary care should be urgently considered and include

232 Chapter 5 | Qualitative study a guarantee of same-day, same-clinic eligibility assessments, greater utilisation of the support offered by peer-support groups and community health workers, and an integration of HTC and ART services at the policy and programmatic levels.

233 Chapter 5 | Qualitative study

234 Chapter 6 | Cluster randomised trial

6. Home initiation of HIV care following self-testing: a cluster randomised trial in urban Blantyre

6.1. Introduction

The quantitative and qualitative studies described in Chapters 4 and 5 showed extremely low rates of completion of PITC under routine conditions among primary care facility attendees in Blantyre. High rates of patient loss to care were also identified, with failure to complete ART eligibility assessments (CD4 count and WHO clinical staging assessment) being a major barrier to timely initiation of ART. Men and non-pregnant women fared worst, with extremely low completion of PITC. Additional factors identified by patients as barriers to linkage into HIV care included the cost and time required to make multiple facility visits, the unwelcoming clinic environment and perception that healthcare workers were rushed, and, for men, a fear of loss of masculinity. Taken together and reflecting studies from other sites in the region, these findings provide strong evidence that for both patients and providers the current HIV care pathway is overly complex, resulting in suboptimal rates of linkage into HIV care.

Mathematical modelling studies suggest that a strategy of “universal test and treat” for

HIV 229 could result in rapid decreases in both population mortality and HIV-incidence7.

With increasing moves towards a “universal test and treat” approach60 to HIV management, ensuring timely and complete linkage from HIV diagnosis to ART treatment will be of critical importance. However, few studies to date have rigorously evaluated interventions to improve linkage into HIV care.

This Chapter reports the results of a cluster-randomised trial that investigated two approaches to improving linkage into HIV care and ART in Blantyre. During the

235 Chapter 6 | Cluster randomised trial introduction of a community-level HIVST intervention as part of the HitTB Study, two strategies were compared: study-provided initiation of facility-based HIV care only, or the additional optional availability of home-based initiation of HIV care. The study intervention was specifically designed to address the barriers to ART identified in the previous cohort and qualitative studies. The HIV Unit of the Ministry of Health of Malawi and the Blantyre District Health Office were directly involved in the planning, design and implementation of the study to ensure that the study objectives met health needs of the population of Blantyre and that the interventions studied would be acceptable, affordable and sustainable for the Malawian health system, if found to be successful.

6.2. Methods

6.2.1. Study design

This was a cluster-randomised trial using restricted 1:1 randomisation of 14 community health worker (CHW) catchment areas into two arms. This was a second-stage randomisation of clusters allocated to the HIVST arm of the HitTB study: a cluster- randomised trial comparing health outcomes achieved under HIVST versus standard of care HTC (i.e. there were 14 vs. 14 clusters in the parent trial).

6.2.2. Participants and study setting

Fourteen clusters in three wards (five in Ndirande, six in Likhabula and three in

Chilomoni) in urban Blantyre (Figure 6.1) were defined using circumferential walks of

CHW catchment areas with GPS readings (eTrex Legend® HCx, Garmin International Inc,

Olathe, USA). All households were then enumerated, with subsequent adjustment of boundaries to obtain approximately 1200 adults (≥16 years) per cluster. A detailed description of the methods used for the definition of cluster boundaries and population

236 Chapter 6 | Cluster randomised trial census are given in Chapter 3.2.2. The adult HIV prevalence had previously been shown to be 18.5%171

Figure 6.1: Study clusters and allocation

6.2.3. Interventions

Two volunteer counsellors from each cluster, selected using participatory methods, were trained in HTC. Between February and November 2012, counsellors in all 14 clusters promoted the availability of HIVST and distributed oral HIV test kits (OraQuick

Advance® Rapid HIV-1/2 antibody test, OraSure Technologies, Bethlehem, USA; assembled in Thailand) to adult residents requesting testing (one kit per resident), with pre-test counselling and demonstration of kit use. All participants who were positive on self-testing could self-refer or be referred to study clinics in local health facilities for

237 Chapter 6 | Cluster randomised trial confirmatory HTC, provision of cotrimoxazole, TB screening (and isoniazid preventive therapy [IPT] if eligible), WHO staging, CD4 counts, and onward referral for ART initiation if eligible (CD4 count <350cells/mm3, WHO stage 3 or 4 or pregnant or breastfeeding).

Detailed description of the HIVST methods are provided in Chapter 3.5.2.

In the seven clusters allocated to the optional home initiation of care [OHC] arm, counsellors organised home-visits by one of two study nurses. Clients had to confide a positive HIVST result in order to request a home visit. Nurses visited each client twice

(first visit within three days and second visit within seven days of notification) to carry out confirmatory testing, WHO staging, CD4 count (venous blood sample drawn for laboratory testing), TB screening (with IPT if eligible), treatment literacy and provision of two weeks of HIV care medications, including ART if eligible. Clients were provided with completed ART registration cards and a follow-up appointment at their nearest HIV care clinic (Ndirande or Chilomoni).

All adults initiating ART at any of the three clinics serving the study population (Queen

Elizabeth Central Hospital, Ndirande or Chilomoni) were interviewed using printed satellite maps marked with cluster boundaries and local landmarks, followed if necessary by a home visit, to establish cluster residence for all 28 clusters in the parent study using the map book system described in Chapter 3.4.3. Data from clinic registers and treatment cards (including monthly clinic attendance, death, transfer out to other treatment sites and default from care [defined as missing two consecutive planned clinic appointments]) were extracted for six months in all identified study residents, without reference to arm, and regardless of mode of testing or entry into care.

238 Chapter 6 | Cluster randomised trial

Outcomes

The primary outcome compared between arms the cumulative incidence of ART initiation among all adult cluster residents (regardless of HIV status, ART eligibility, site of HTC or ART initiation) during the first six months of HIVST availability. The secondary outcomes compared the cumulative incidence of a) taking a HIVST kit, b) reporting a positive HIVST result to counsellors, and c) loss from retention by six months (with participants recorded as still taking ART or transferred out to another clinic classified as retained). Self-reported adherence was also assessed by questionnaire at three points following ART initiation (at two-to-four weeks, at three months and at six months) using the AIDS Clinical Trials Group (ACTG) adherence questionnaire.

In an additional preplanned analysis, non-traumatic HIV-related adult (aged 15-49 years) mortality rates were compared between study arms. Deaths were captured through a key informant system, confirmed through home visits. Data from verbal autopsies were used to define likely relationship to HIV, using the InterVA 3.0 system described in Chapter 3.6.

Laboratory methods

Blood samples for CD4 count measurement were collected from participants undergoing

ART eligibility assessments in the study clinics and at home. Samples were analysed using the Partec Cyflow® SL-3 platform (Partec, Görlitz, Germany) as described in

Chapter 3.7.1.

6.2.4. Sample Size

The primary outcome sample size calculation assumed an HIV prevalence of 18.5%171,

50% HIV-positive self-testing adults would report a positive result to counsellors and 5%

239 Chapter 6 | Cluster randomised trial of the adult population would initiate ART over six months in the facility initiation of care only [FCO] arm. A mean cluster population of 1200 adults in 14 clusters provided 80% power at a 5% level of significance to detect a risk ratio of 1.5 between arms of the adult population who initiated ART with a coefficient of variation k=0.20. Figure 6.2 shows sample sizes required to detect differences in per-capita initiation of ART at various levels of k.

Figure 6.2: Sample sizes required to detect difference in per-capita ART initiation with power=0.8

1.85 k=0.12 k=0.20 1.75 k=0.28 k=0.36 1.65

1.55

1.45

Relativearmsriskstudy between 1.35

1.25 0 5 10 15 20 25 30 35 40 Number of clusters required per study arm

6.2.5. Randomisation and masking

Clusters were randomised at a public meeting, following restriction of possible allocations to ensure that for each of the three wards, the difference in the number of clusters allocated to each arm was less than or equal to two. This provided 2100 unique cluster allocation patterns. At the randomisation meeting, community representatives

240 Chapter 6 | Cluster randomised trial drew coloured golf-balls (red and blue) from an opaque bag held above eye-level to select the distribution of arms, and arm allocation. Counsellors and residents were not masked to the intervention, but all subsequent data capture and management was carried out without reference to the intervention group until the final analysis.

6.2.6. Statistical analysis

The CONSORT guidelines (modified for cluster-randomised study designs) were followed455. For the primary and secondary endpoints, analysis used intention-to-treat, taking total number of adult residents as the cluster denominators (since HIV and ART eligibility status was not available). To estimate impact on ART coverage, we assumed an

HIV prevalence in study communities of 18.5%,456 ART coverage of 67% among HIV- positive individuals with CD4<350 cells/mm3 at the commencement of the study457, and

ART eligibility of 33% among undiagnosed HIV-positive individuals.

CONSORT guidelines were followed in reporting preplanned study outcomes455. The cluster-level proportions of residents who initiated ART, took an HIVST test kit and reported a positive HIVST result in each intervention group were compared using the t test, with the unadjusted risk ratio calculated as the ratio of cluster-averaged means using the method described by Hayes and Moulton458. Ninety-five per cent confidence intervals for risk ratios were calculated based on the t-distribution and using a Taylor approximation to estimate the standard error of the log risk ratio459. The need for adjusted analysis was pre-specified for covariates meeting definitions of imbalance; in variables where there was a 5% or greater difference, the outcome would be adjusted for the relevant variables. Analysis adjusted for baseline cluster mortality was conducted for the primary outcome using a two-step process. A logistic regression model including terms for the cluster-level mortality but no intervention term was first fitted. From this

241 Chapter 6 | Cluster randomised trial the predicted prevalence of the outcome for each cluster and the ratio between the observed and predicted prevalence was calculated (risk ratio-residuals). These cluster- specific residuals were then analysed using the same approach as described above for the cluster-specific proportions to obtain adjusted risk ratios, 95% confidence intervals and p-values.

A separate analysis was used to assess rates of loss to retention, with the denominator being adult residents who initiated ART. Cluster-level rates of loss from retention in each intervention group were compared using the t-test, with the unadjusted rate ratio calculated as the ratio of cluster-averaged means. To account for anticipated differences in characteristics of ART initiators in between intervention groups, analysis was adjusted for sex, age, pregnancy status, CD4 count strata and WHO stage. To investigate risk factors for loss to retention, a Cox proportional hazard model was constructed which included cluster of residence as a fixed effect term in the model.

6.2.7. Ethical approval

Approval for the study was granted by the research ethics committees of the College of

Medicine of Malawi (Blantyre, Malawi), Liverpool School of Tropical Medicine (Liverpool,

UK) and London School of Hygiene and Tropical Medicine (London, UK) – Appendix 2.

6.3. Results

6.3.1. Baseline characteristics of clusters

The 14 clusters included in the study had a combined adult population of 16660, with

8194 adults resident in 3213 households in the OHC arm and 8466 adults resident in

3397 households in the FCO arm (Figure 6.3).

242 Chapter 6 | Cluster randomised trial

Figure 6.3: CONSORT flowchart of outcomes (modified for cluster randomised design)

Randomised: 14 clusters

Allocated to optional home initiation of care Allocated to facility care only (FCO) arm: 7 (OHC) arm: 7 clusters clusters Received allocated interventions: 7 clusters Received allocated interventions: 7 clusters

Total adult population: 8194 Total adult population: 8466 Mean cluster adult population: 1179 Mean cluster adult population: 1209 Range of cluster adult population: 923-1416 Range of cluster adult population:1075-1276

Analysed from OHC arm: 7 clusters Analysed from FCO arm: 7 clusters

HIV self-tests: 5287 (64.9% of adult cluster HIV self-tests: 4433 (52.7% of adult cluster population) population)

Reporting of HIV-positive self-test results to Reporting of HIV-positive self-test results to community counsellor: 490 (6.0% of adult cluster community counsellor: 278 (3.3% of adult cluster population population

ART initiations: 181 (2.2% of adult cluster ART initiations: 63 (0.7% of adult cluster population), 116 home initiators population)

6-month outcomes 6-month outcomes

Retained on ART: 129 (1.57% of adult cluster Retained on ART: 48 (0.57% of adult cluster residents and 71.3% of ART initiators) residents and 76.2% of ART initiators)

Transferred out: 11 (0.13% of adult cluster Transferred out: 3 (0.03% of adult cluster residents and 6.1% of ART initiators) residents and 4.8% of ART initiators)

Died on ART: 5 (0.06% of adult cluster residents Died on ART: 1 (0.01% of adult cluster residents and 2.8% of ART initiators) and 1.6% of ART initiators)

Defaulted from ART: 47 (0.57% of adult cluster Defaulted from ART: 14 (0.17% of adult cluster residents and 26.0% of ART initiators) residents and 22.2% of ART initiators)

Self reported any missed ART doses: 19 (0.23% of Self reported any missed ART doses: 3 (0.03% of adult cluster residents and 11.6% of ART adult cluster residents and 5.0% of ART initiators) initiators)

Exploratory 6-month outcome Exploratory 6-month outcome

Non-traumatic, HIV-related deaths in adults aged Non-traumatic, HIV-related deaths in adults aged 16-49 years: 7/7522 (0.09%) 16-49 years: 10/7838 (0.13%)

243 Chapter 6 | Cluster randomised trial

Socio-demographic characteristics assessed in the baseline enumeration were well balanced between intervention groups (Table 6.1), apart from for reported household deaths in the previous year (OHC arm: 131/8194, 4.1%, FCO arm: 81/8466, 2.4%).

244 Chapter 6 | Cluster randomised trial

Table 6.1: Baseline cluster, household and individual characteristics

Optional home Facility care only initiation of care am arm arm Total number of clusters 7 7 Total number of households 3213 3397 Total number of adults (aged 16+ years) 8194 8466

Cluster level characteristics Number of adults per cluster (mean, range) 1171 (923-1416) 1209 (1075-1276) Population density per cluster, persons per m2 (mean, range) 0.016 (0.009-0.030) 0.024 (0.010-0.044)

Household level characteristics Number of adults (age 16+ years) per household (mean, SD) 2.55 (1.26) 2.48 (1.17) Number of children per household (mean, SD) 1.96 (1.58) 1.93 (1.54) Household wealth quintile 1 (poorest) 730 (23.3%) 806 (24.3%) 2 (poorer than average) 656 (21.0%) 703 (21.2%) 3 (average) 599 (19.2%) 696 (21.0%) 4 (wealthier than average) 602 (19.2%) 589 (17.8%) 5 (least poor) 540 (17.3%) 518 (15.6%)

Death in household in year preceding enumeration 131 (4.1%) 81 (2.4%)

Individual level characteristics (adults only) Age in years (mean, SD) 30 (11.8) 30 (11.4) Age group, years 16-20 1312 (16.1%) 1227 (14.5%) 20-29 3312 (40.5%) 3556 (42.1%) 30-39 2113 (25.9%) 2267 (26.8%) 40-49 785 (9.6%) 788 (9.3%) 50-59 363 (4.4%) 355 (4.2%) >=60 285 (3.5%) 255 (3.0%)

Male 4252 (52.5%) 4399 (52.6%) Marital status Married/cohabiting 5031 (62.8%) 5293 (64.4%) Never married 2441 (30.5%) 2403 (29.2%) Widowed/separated/divorced 535 (6.7%) 524 (6.4%) Ever lost a spouse to death 378 (4.7%) 373 (4.5%) Highest level of education No schooling 217 (2.7%) 237 (2.9%) Primary 3168 (39.7%) 3214 (38.7%) Secondary no MSCE 2389 (30.0%) 2863 (34.5%) Secondary with MSCE 1564 (19.6%) 1465 (17.6%) Higher 635 (8.0%) 520 (6.3%) * Number of households or individuals in optional home initiation of care (OHC) arm and facility initiation of care only (FCO) arm respectively, with missing values for: sex 73, 79; marital status

245 Chapter 6 | Cluster randomised trial

163, 228; education 197, 149; household wealth 86, 85. MSCE: Malawi secondary certificate of education

6.4. Primary outcome

Overall, 244 cluster resident adults initiated ART during the six month study period

(Figure 6.2). The cumulative incidence of ART initiation was significantly higher in the

OHC arm (181/8194, 2.2% of residents) compared to the FCO arm (63/8466, 0.7% of residents; risk ratio [RR]: 2.94, 95% confidence interval [CI]: 2.10-4.12), p<0.001 – Table

6.2. After adjusting for reported household mortality at baseline, the effect of availability of home initiation of HIV care remained highly significant (adjusted RR: 2.44,

95% CI: 1.61-3.68, p<0.001).

246 Chapter 6 | Cluster randomised trial

Table 6.2: Primary and secondary trial endpoints

247 Chapter 6 | Cluster randomised trial

The difference between arms in ART initiation was maintained throughout the analysis period (Figure 6.4), with a substantial early peak in the OHC arm, followed by an increasing difference between the two arms for months two to six.

Figure 6.4: Cluster resident adult ART initiations during 6 months of HIVST availability

0.60% OHC FCO 2.50%

OHC FCO

0.50% Cumulave proporon of adult residents iniang ART 2.00%

0.40%

1.50%

0.30%

1.00%

0.20% Proporon of adult residents iniang ART

0.50% 0.10%

0.00% 0.00% 1 2 3 4 5 6 Month

6.5. Secondary outcomes

During the 6-months availability, a total of 9720/16660 (58.3%) participants received

HIVST. This included 4174/8651 men (48.2%) and 5536/8009 women (69.1%), 107 (1.3%) of whom were pregnant at the time of HIVST. There were substantial differences in uptake of HIVST by age and sex (Figure 6.5). Across all age groups, a greater proportion

248 Chapter 6 | Cluster randomised trial

of women self-tested for HIV than men (test for trend: pt<0.0001). There was also a clear trend in uptake of self-testing by age group, with older participants less likely to test than younger residents. The highest uptake of self-testing was among 16 to 20 year old women, where 1256/1306 (96.2%) underwent self-testing. The lowest uptake of self- testing was in men in the 50 to 60 year age group, where only 124/422 (29.4%) self- tested.

Figure 6.5: Uptake of HIV self-testing, by study arm, age group and sex

Facility care only arm: 100% Women Facility care only arm: Men 90%

80% Oponal home iniaon of care arm: Women 70% Oponal home iniaon of care arm: Men 60%

50%

40%

30%

20% Percentage of residents sef-tesng 10%

0% 16-20 20-30 30-40 40-50 50-60 >=60 Age group

In both study arms, uptake of HIVST was highest in the first months of availability. In the

OHC arm, where 1362/8194 (16.6%) of participants underwent HIVST in the first month, compared to in the FCO arm, 1539/8466 (18.2%) of participants self-tested in the first month. Excluding the first month, uptake of HIVST uptake was higher in the OHC arm in every subsequent month.

249 Chapter 6 | Cluster randomised trial

Uptake of HIVST was higher in the OHC arm (5278/8194, 64.9%) than in the FCO arm

(4433/8466, 52.7%), although the difference did not reach statistical significance (RR:

1.23, 95% CI: 0.96-1.58) –Table 6.2. Residents in the OHC arm were, however, significantly more likely to confide a positive HIVST result, with 490/8194 (6.0%) of all

OHC arm adults confiding compared to 278/8466 (3.3%) of FCO arm residents (RR: 1.86,

95% CI: 1.16-2.97) – Table 6.2

6.6. Additional analyses

The availability of home initiation of care did not draw clients away from facilities, but instead substantially increased numbers initiating ART in the OHC arm. Over six months, in the OHC arm, 65/8194 (0.8%) initiated ART in health facilities and 116/8194 (1.4%) initiated ART at home. In comparison, in the FCO arm, all 63/8466 (0.7%) initiated ART at health facilities (Figure 6.6).

Figure 6.6: Home and facility ART initiations by trial arm

250 Chapter 6 | Cluster randomised trial

Without home initiation of HIV care, HIVST alone had no appreciable impact on improving uptake of ART: in the 14 clusters in the parent study without access to HIVST or home initiation of HIV care, 127/17796 (0.7% of adult population) initiated ART –

Figure 6.5.

Home ART initiators had significantly less advanced immunosuppression compared to facility initiators (Table 6.3): median CD4 count at ART initiation was highest among home initiators in the OHC arm (219 cells/mm3, interquartile range [IQR]: 135-305) compared to facility initiators in the FCO arm (187 cells/mm3, IQR: 100-256) and facility initiators in the OHC arm (154 cells/mm3, IQR: 116-249; p=0.042).

Table 6.3: Comparison of home and facility ART initiators by trial arm

OHC arm: FCO OHC arm: Home arm: Facility % or % or % or ART Facility P† initiators IQR IQR IQR initiators initiators (n=72) (n=109) (n=63) Men 28 38.9 49 45.0 22 34.9 0.409 Pregnant (if female) 13 29.6 2 3.3 14 34.2 <0.001 Age (years), median, 32 28-38 34 29-39 31 26-42 0.874 IQR Body mass index 2 20 18-22 20 18-22 19 22-24 0.039 (kg/m ), median, IQR

CD4 count (cells/mm3), 187 100-256 219 135-305 154 116-249 0.042 median, IQRX

CD4 count categoryX >350 cells/mm3 1 2.9 18 16.5 1 3.1 0.093 201-350 cells/ mm3 14 41.2 12 37.5 12 37.5 <=200 cells/ mm3 19 55.9 49 45.0 19 55.4 WHO clinical stage§ 1 or 2 46 64.8 83 76.2 49 80.3 0.098 3 or 4 25 35.2 26 23.8 12 19.7 OHC arm: optional home initiation of HIV care arm FCO arm: facility initiation of HIV care only arm X 69 missing values for CD4 count (OHC arm facility: 38; OHC arm home: 0; FCO arm facility: 31) § 3 missing values for WHO clinical stage (OHC arm facility: 1; OHC arm: home: 0; FCO arm facility: 2) † Chi-square test with 2 degrees of freedom for categorical data and Kruskal-Wallis test of equality for continuous data IQR: Interquartile range

251 Chapter 6 | Cluster randomised trial

There were markedly different trends when disaggregated by sex and site of initiation

(Figure 6.7). Participants who initiated ART in facilities in the FCO arm were least likely to be men (35%), a similar proportion to aggregated data from HIV care clinics in sub-

Saharan Africa (35%). Thirty-nine per cent of participants who initiated ART in facilities in the OHC arm were men and 45% of home initiators (all OHC arm) were men.

Figure 6.7: Proportion of ART initiators who were men

50% 45% 45% 39% 40% Mean proportion of 35% 35% male ART initiators in 15 SAA countries, 2000- * 30% 2012 (n=216,008)

25%

20%

15%

10%

5%

0% Control (facility FCO arm Intervenon (facility OHC arm Intervenon (home OHC arm (facility iniators) iniators) (Facility iniators) iniators) (home iniators) iniators)

* Druyts et al 2013460

6.6.1. Estimated impact of interventions on ART coverage among PLHIV and ART-eligible residents

Using the assumptions described in the Methods, we estimated that 12% of all HIV- positive, and 46% of all ART eligible adults not already taking ART in the OHC arm, and

4% of HIV-positive adults and 15% of ART eligible adults not already taking ART in the

FCO arm initiated ART over the six months of HIVST availability (Figure 6.8).

252 Chapter 6 | Cluster randomised trial

Figure 6.8: Estimated impact on ART coverage over 6 months

There are, however, considerable limitations to these estimates, hindering efforts to understand the impact of study interventions on population ART coverage. Baseline cluster-level estimates for ART coverage were not available, as an HIV-prevalence and

ART coverage survey was not planned until completion of the HitTB study. Instead, the

67% baseline coverage figure was taken from estimates provided by UNAIDS for 2012 country-level ART coverage, which were based on national surveillance data reported to

UNAIDS and included with the SPECTRUM epidemiological estimation tool1. Blantyre is a density-populated urban centre, with a higher HIV prevalence than other areas of

Malawi and with considerable in- and out-migration. It is possible therefore that the ART coverage estimate used in the methods is either an under- or over-estimate.

Additionally, only one study was identified that reported levels of ART eligibility among individuals with undiagnosed HIV in Malawi261. This study was based in Karonga District,

Northern Malawi and was conducted within an ongoing demographic and health

253 Chapter 6 | Cluster randomised trial surveillance survey. However, surveys have shown that coverage of HIV testing is substantially higher in the DHS site in Karonga compared to other Districts of Malawi, likely as a consequence of study activities429. A potential consequence is that with a greater proportion of HIV-infected individuals diagnosed, CD4 count profiles among undiagnosed adults may be biased upwards or downwards, influencing estimates of ART eligibility. In summary, these estimates of effect on ART coverage are at best approximate. It is anticipated that as more accurate study population estimates of ART coverage and eligibility become available through the HitTB HIV post-intervention survey

(due mid-2014), these estimates will be updated.

6.6.2. Loss from ART over six months

After six months, 129/181 participants (71.3%) who initiated ART in the OHC arm and 48 of 63 participants (76.2%) who initiated ART in the FCO arm were retained on ART.

(Figure 6.9) – log rank test p=0.432. There was a high risk of loss to ART care in the first month following ART initiation; with 17 ART initiators (9.4%, 7 of whom were home initiators) in the OHC arm and 6 (9.5%) ART initiators in the FCO arm failing to return to care following initiation.

254 Chapter 6 | Cluster randomised trial

Figure 6.9: Retention on ART 6 months after initiation

1.00

0.75

p=0.432 0.50

0.25 Probability of retention on ART OHC arm FCO arm 0.00 0 30 60 90 120 150 180 Days since ART initiation Number at risk OHC arm: 181 155 144 136 126 116 74 FCO arm: 63 56 52 51 50 44 26

7/116 (6%) were lost among home initiators in the OHC arm, 10/65 (15%) were lost in the clinic initiators in the OHC arm, and 6/63 (10%) were lost in the FCO arm.

In unadjusted analysis, the rate of loss to care was higher in ART initiators in the OHC arm (63.4 per 1000 person-months, 95% CI: 42.7-84.0) than in the FCO arm (53.5 per

1000 person-months, 95% CI: 23.7-83.4), although not significantly so (incidence rate ratio [IRR]: 1.18, 95% CI: 0.67-2.10) – Table 6.2. Adjusting for risk factors for loss at ART initiation had little impact (adjusted IRR: 1.18, 95% CI: 0.62-2.25).

In individual-level univariate analysis (Table 6.4), ART initiators who were in WHO clinical stage 3 or 4 were at increased risk of loss to retention (95.4 per 1000 person-months,

95% CI: 64.4-141.2) compared to ART initiators in stage 1 or 2 (48.5 per 1000 person-

255 Chapter 6 | Cluster randomised trial months, 95% CI: 35.7-65.9; hazard ratio: 1.90, 1.13-3.20). However in the multivariate model no risk factors remained associated with loss to retention.

256 Chapter 6 | Cluster randomised trial

Table 6.4: Individual-level risk factors for loss to retention from ART

257 Chapter 6 | Cluster randomised trial

6.6.3. Adherence to ART

201 ART initiators (82.4% of all ART initiations and 91.0% of initiators who returned for a subsequent clinic appointment) completed the ACTG self-reported adherence questionnaire at two-to-four weeks following ART initiation, 145 (59.4% of all ART initiators and 82.9% of individuals retained in care) completed the questionnaire at three months and 119 (48.8% of all ART initiators and 67.2% of individuals retained in care) completed questionnaires at six-months. Completion was lower than anticipated at the six month appointment due to the difficulty in identifying participants in busy ART clinics.

Overall, 19/164 (11.6%) of ART initiators in the OHC arm and 3/60 (5.0%) of ART initiators in the FCO arm with data available self-reported missing at least one dose of

ART in the past four days at any assessment point (p=0.142). 160/164 (97.6%) and 58/60

(96.7%) of individuals in the OHC arm and the FCO arm respectively self-reported mostly or always taking ART at the correct time over the last four days (p=0.714).

6.6.4. Mortality in residents aged 16 to 49 years

During the six months intervention period, there were a total of 44 deaths recorded in

15360 residents aged 16 to 49 years (OHC arm: 26/7838 [0.3%], FCO arm: 18/7522

0.2%), giving a crude overall mortality rate of 2.7 per 1000 (95% CI: 1.63-3.81). The crude mortality rates were 3.3 per 1000 (95% CI: 1.51-5.05) and 2.1 per 1000 (95% CI: 0.46-

3.87) in the OHC arm and the FCO arm respectively. Causes of death attributed by the verbal autopsy and InterVA probabilistic model system are shown in Table 6.5. The commonest primary cause of death (17/44, 38.6% overall) was HIV (OHC arm: 7/26,

258 Chapter 6 | Cluster randomised trial

26.9%, FCO arm: 10/18, 55.6%); giving a crude HIV-related mortality rate of 1.1 per

1000, 95% CI: 0.33-1.89. Tuberculosis (OHC arm: 6/26, 23.1%, FCO arm: 1/18, 5.6%) was the next most common. One suicide was recorded, with no relationship to HIVST.

Table 6.5: Primary cause of death in 44 cluster residents aged 16-49 during 6 months of interventions

Home initiation of Facility initiation of care Primary cause of death Total (%) care arm (%) arm (%) Chronic cardiac disease 0 (0.0%) 1 (5.6) 1 (2.3) HIV-related 7 (26.9) 10 (55.6) 17 (38.6) Liver disease 1 (3.8) 0 (0.0) 1 (2.3) Malignancy 1 (3.8) 1 (5.6) 2 (4.6) Maternity-related 3 (11.5) 0 (0.0) 3 (6.8) Meningitis 1 (3.8) 1 (5.6) 2 (4.6) Road traffic accident 0 (0.0) 1 (5.6) 1 (2.3) Stroke 1 (3.8) 0 (0.0) 1 (2.3) Suicide 1 (3.8) 0 (0.0) 1 (2.3) Tuberculosis 6 (23.1) 1 (5.6) 7 (15.9) Indeterminate 2 (7.7) 2 (11.1) 4 (9.0) No verbal autopsy 3 (11.5) 1 (5.6) 4 (9.1) performed

Total deaths 26 18 44

HIV-related mortality was lower in the OHC arm (0.90 per 1000 persons) than the FCO arm (1.32 per 1000 persons), although the difference was not statistically significant

(rate ratio: 0.69, 95% CI: 0.16-2.94).

6.7. Summary of findings

The main finding of this cluster-randomised trial comparing optional home initiation of

HIV care against best-practice facility-based initiation of HIV care in the context of availability of HIVST was that ART initiations were significantly and substantially increased in neighbourhoods with access to home initiation of care. Uptake of HIVST was high across all neighbourhoods: 58% of all adult residents chose to test within 6- months and a significantly greater proportion of participants reported positive results

259 Chapter 6 | Cluster randomised trial where home initiation of care was available. The initial benefit of increased population

ART initiation was not lost over time, with similar rates of retention and adherence over

6-months, resulting in greater population coverage of ART. This is the first study to investigate strategies for improving linkage to ART after home-based HIV testing (in this case HIVST). The offer of home initiation of HIV care shows high promise as a strategy to improve linkage where HIV testing is carried out at home.

Previous studies, reported in Chapters 4 and 5 of this thesis, have demonstrated that rates of completion of PITC and subsequent initiation of HIV care are suboptimal under routine facility conditions, with onerous ART eligibility assessments a major barrier.

Alternative approaches that increase the availability, accessibility and acceptability of

HIV testing to individuals living in high prevalence settings, and promote timely and complete linkage into HIV care are urgently needed, especially as we approach an era of

“universal test-and-treat”. In this trial, community availability of HIVST and optional home initiation of HIV care significantly improved upon alternative approaches at the populations level and reached an estimated 46% of ART eligible individuals (at the time of the study: CD4 count <350 cells/mm3) over only 6-months, providing evidence that this approach could help meet WHO’s target of achieving universal coverage of ART.

Evaluation over a longer time period will be required to determine whether the high rates of uptake of HIVST and increased population ART initiations are sustained.

We attribute the increased in population-level ART initiations to the fact that home initiation of HIV care removed many of the existing barriers to linkage to ART. We were reassured to see that the availability of home initiation of HIV care did not simply shift

ART initiations from health facilities to home: instead the rate of facility ART initiations

260 Chapter 6 | Cluster randomised trial was maintained, while home initiations provided additional contributions to population initiations.

Rates of ART initiation were similar in neighbourhoods where neither HIVST nor home initiation of HIV care interventions were available (as part of the larger HitTB trial), and in communities with access to HIVST and facility-based care only. This sends a strong message that community-based provision of HIVST alone may be insufficient to improve timely linkage; additional proactive steps to promote linkage into HIV care are required.

Home initiation of HIV care is one such approach that was effective in this community, however evaluation of alternative approaches that are responsive to the mode of HIVST delivery, as well as the context of locally available health services and geography will be required.

Despite the substantial and significant population-level impact on ART initiation and reporting of HIV positive results achieved in the intervention arm, the optional availability of home initiation of HIV care did not result in complete linkage into HIV care: even with the high level of home-, community- and facility-based services provided at all stages of the HIV care pathway, linkage into care remained below that required in

“test-and-treat” approaches, although only a short time period was assessed. Moreover, rates of linkage into care appear to be below that achieved in DOTS TB programmes, where a similar decentralised approach to provision of care has been taken384. Further research to understand participants’ perspectives of home-based HIVST and initiation of care are required to identify reasons for failure to initiate ART and where further gains in linkage can be achieved.

261 Chapter 6 | Cluster randomised trial

Although considerable mobilisation of community members and resources were required to implement home-based HIVST, in terms of resource requirements, the home initiation of HIV care intervention was a relatively minor component, with only two ART nurses providing home assessments using motorbikes for transport. Given that self- testing was implemented through neighbourhood volunteers and that HIVST took place within the home, in this study home initiation of HIV care was a suitable choice of intervention. However, the offer of home initiation of HIV care would be less appropriate following facility-based testing.

Perceived lack of confidentiality has been cited as a potential barrier to uptake when HIV interventions are offered in the home. However, drop-out due to stigma was extremely uncommon in a previous trial of home delivery of ART following facility initiation461. We overcame these potential difficulties by developing a partnership with the community

(through volunteer counsellors and community representative groups) and with participants undergoing HIVST, who received pre-test information about the potential advantages and disadvantages to home initiation of HIV care. Regular meetings were held with community stakeholders where concerns could be raised and addressed. No incentives (financial or otherwise) were provided to participants or providers.

Retention on ART over six months was worse in the home initiation arm, than in the facility initiation arm, although the differences were not statistically significant and overall numbers of initiators were small, limiting power to identify anything other than large differences. Pooled across study arms, 72.5% of ART initiators were retained on treatment at six months. This is below national HIV programme estimates for Malawi

(80% retained at 12-months175), although still within the range seen in other programmes where rates of attrition can reach 40% after a median of 10 months follow-

262 Chapter 6 | Cluster randomised trial up383. The study setting, with recruitment from urban slums, may have led to inclusion of a greater proportion of highly mobile individuals compared to other areas of the country462. The trend towards poorer outcomes among ART initiators in the OHC arm means that careful monitoring and treatment support should be provided for home ART initiators in future studies to avoid losing the initial population benefits of home ART initiation.

6.8. Limitations

Limitations of the study include the need to use all adult cluster residents (not people living with HIV or ART-eligible adults) as our denominator. This reflects the lack of available cluster-level HIV prevalence and CD4 count data. However, ART initiation remained highly significantly increased in the OHC arm after adjusting for household mortality in the previous year, which could have been indicative of differences in HIV- prevalence or ART coverage between study arms. To maintain privacy and confidentiality, individual HIVST participants were not followed as a cohort, and indeed confiding results was optional and not required. Therefore, we cannot estimate overall linkage into care following positive HIVST, nor examine any given step of the HIV care pathway. Total numbers of ART initiations were small, meaning that larger studies are required to fully evaluate the possible trends towards worse retention on ART in the home initiation arm. ART initiators were only followed up for a short period (six months) and default from ART was not followed-up by active tracing. The study was based in urban slums with a high prevalence of HIV and three nearby facilities providing ART. The acceptability and effectiveness of HIVST and home initiation of HIV care needs to be investigated in other populations, as social norms and the acceptability and effectiveness of routine services are likely to affect the relative impact of HIVST-based interventions.

263 Chapter 6 | Cluster randomised trial

A community-based key informant system was implemented for surveillance of mortality during the trial period. The study was underpowered to detect anything but any extremely large difference in HIV-related population mortality; hence this outcome was assessed as an exploratory outcome. There was intensive monitoring and quality assurance of the key informant system at regular cluster meetings. Nevertheless, it is possible that key informants may not have recorded a small number of deaths. This could have resulted in mortality rates being underestimated. Annual mortality rates in adults aged 15 to 44 years in the Karonga Demographic and Health Survey site in

Northern Malawi fell from 668 per 100,000 in 2004 to 452 per 100,000 in 2009463, similar to the annual mortality rate found in this study (414 per 100,000 per year). Using verbal autopsies (but not the InterVA system), in Karonga, HIV/AIDS was identified as the probable cause of cause in 42% of community members in 2004-2005 and 17% in 2008-

2009. In Blantyre, pooled between study arms, 38% of deaths were attributable to

HIV/AIDS, although numbers of events were considerably smaller in Blantyre

(Blantyre=44 deaths, Karonga 2008-2009=142 deaths). The authors attribute the decline in HIV-associated death in Karonga is to expanded coverage of ART during the study period. The HitTB study will report mortality rates from data collected over two years, providing more accurate estimates for Blantyre.

We were unable to complete verbal autopsies for a small number of participants. This could mean that HIV-related mortality rates could be underestimated. The probabilistic computer model used to determine causes of death (InterVA 3.0) was updated shortly after data collection for this study was completed (InterVA 4.0). The updated version of

InterVA, which purportedly has greater accuracy (although this has not yet been formally assessed) required collection of a small number of additional variables that

264 Chapter 6 | Cluster randomised trial were not collected by the study nurses who performed verbal autopsies. Therefore, the accuracy of the cause of death assessments could have been improved.

6.9. Findings relating to other studies

No previous randomised controlled trials have evaluated strategies to improve linkage into HIV care following home-based HIVST. This study therefore provides novel, timely and highly relevant data as policymakers and programme managers consider introducing

HIVST, as well as grapple with how best to improve linkage into HIV care following other home-based HTC strategies.

To date, only one peer-reviewed published study has described the implementation of

HIVST in sub-Saharan Africa456, which was part of the pilot study for the Hit TB Hard

Study. Outcomes of linkage into HIV care were not assessed in this study. An unpublished study from Kenya reported on healthworkers’ uptake of HIVST464, but likewise did not report on linkage into HIV care.

Four previous randomised controlled trials have evaluated strategies to improve linkage into HIV care in resource limited settings (two studies from South Africa, and one study each from Uganda and India – summarised in detail in Chapter 2.7.5). In the only rigorously conducted recent study, linkage into ART was evaluated as a pre-planned secondary outcome in a cluster randomised trial in primary health care clinics in South

Africa299. HIV diagnosed participants who were assessed in clinics allocated to receive intensive nurse-training and where task-shifting took place were no more likely to initiate ART than individuals assessed in clinics where these interventions were not available (intervention arm: 3712/5390 [69%]) initiated ART, control arm: 2418/3862

[63%] initiated ART, RR: 1.24, 95% CI: 0.88-1.73).

265 Chapter 6 | Cluster randomised trial

There were substantial differences between the trial described in this Chapter, and the trial conducted by Fairall et al. In the study by Fairall et al all, participants were recruited from individuals attending health care facilities for standard HTC, whereas in our trial,

HIVST was offered at home through community counsellors. Additionally, home initiation of HIV care (including ART) was offered in our trial intervention arm, whereas all HIV care services were provided at facilities in the South African study. Thus, results of the two trials are not directly comparable.

In a non-inferiority trial conducted in Uganda409, rates of virological suppression were compared between individuals who had initiated ART at health facilities and resided in clusters allocated to receive either facility-based continuation of ART, or home-based continuation of ART through nurse visits. This study differs substantially from ours in that in both arms, ART was initiated in health facilities and in the intervention arm, participants were followed-up and received continuation of ART at home. Nevertheless, it is useful to compare rates of retention on ART between the studies. Follow-up procedures were more intensive, with study nurses performing home visits for ART distribution in the intervention arm and home tracing for participants who were lost-to- care in both arms. Rates of patient loss to care were similar to those reported in our trial

(Arm 1: 22.2% and Arm 2: 26.0% after 6-months), but occurred over a longer period of follow-up. In the Ugandan trial, 183/857 (21.3%) in the intervention arm and 142/594

(23.8%) were retained on ART after a median of 28 months follow-up. In both trials (and common to all studies reporting on patient retention following ART initiation383), the majority of patient loss occurred early (within the first few months after ART initiation) with little patient loss thereafter. Should individuals in our trial have been followed up

266 Chapter 6 | Cluster randomised trial for a similar longer period of time, retention in care may not have substantially declined further.

In summary, there are no comparable randomised controlled trials in which interventions to improve linkage into HIV care from individuals undergoing home-based

HTC (or HIVST) were evaluated. Review of clinical trials registers identified a number of studies that are planned, or already underway, that could allow comparison of outcomes and insights into what sort of interventions are most effective in improving linkage into care465-473.

6.10. Implications for future studies

The optional availability of home initiation of HIV care had a significant impact on population ART initiations. However, to determine whether this intervention is cost- effective, formal economic evaluation will be required. This analysis is planned as part of the parent HitTB trial. Given the large effect size observed, and the relatively small additional resources required for home initiation of HIV care compared to HIVST alone

(additional two nurses and two motorbike), it is likely that the intervention will be a cost-effective addition to HIVST for improving uptake of ART.

As noted above, despite offering a comprehensive home-based service that addressed all steps in the HIV care pathway, over half of ART eligible adults remained did not initiate ART during the intervention period. Future studies should evaluate the underlying reasons for failure to link into care despite the availability of these interventions (potential reasons and strategies for interventions are discussed further in

Section 8.2).

267 Chapter 6 | Cluster randomised trial

The impact of home initiation of ART and HIVST on HIV incidence requires further evaluation, ideally as part of randomised controlled trials.

6.11. Conclusions

In conclusion, this study has shown that the optional availability of home initiation of

HIV care following HIVST had a substantial and significant effect on increasing population ART initiation. Achieving the high rates of HCT and ART coverage required to rapidly impact upon HIV incidence and mortality will require interventions that promote timely and complete linkage into HIV care. Home initiation of HIV care following HIVST is one approach that could provide substantial individual and public health benefit.

268 Chapter 7 | Evaluation of accuracy of CHW tool

7. Evaluation of a novel community health workers ART eligibility assessment tool for use in community settings in Blantyre, Malawi

7.1. Introduction

At the time of these studies, WHO guidelines recommended that HIV-infected adults with a CD4 count of less than 350 cells/mm3 be initiated onto antiretroviral therapy

(ART). These guidelines were revised in July 2013 to recommend that HIV-infected individuals with CD4 count ≤500 cells/mm3 should initiate ART. Availability of CD4 count measurement remains low in much of sub-Saharan Africa. For example, in Malawi only

12% of HIV care facilities have functional on-site capacity for CD4 count measurement175. Where facilities for CD4 count are not available, WHO recommends that individuals assessed as being in WHO clinical stage 3 or 4 should initiate ART.

Results of the prospective cohort study described in Chapter 4 and the qualitative study among healthworkers and patients described in Chapter 5 clearly demonstrated that

ART eligibility assessments are a common point of patient loss from the HIV care pathway, with the WHO clinical staging system perceived to be over-complex and time- consuming and CD4 counts difficult to access. Participants who remained unstaged by six months had a significantly lower uptake of ART compared to individuals who completed staging.

The WHO clinical staging system for HIV was introduced in 1990244, and was based on previous highly-specific WHO case definitions for AIDS that were used in the era before the availability of reliable and rapid HIV diagnostic tests239, 242. Although the WHO clinical staging system has good prognostic value in predicting death248 and response to ART250,

269 Chapter 7 | Evaluation of accuracy of CHW tool it was not originally designed to be used for as screening tool for ART eligibility. Previous studies255, 259, 264, 265 have shown that the sensitivity of nurse-performed WHO stage 3 or

4 assessment in identifying CD4 count of ≤350 cells/mm3 is low (see Chapter 2.4), potentially resulting in missed and delayed ART initiation474.

With task-shifting of HIV testing to community health workers (CHWs)475, decentralisation of HIV care to primary health care centres and moves towards earlier

HIV diagnosis and initiation of treatment, it is important that ART eligibility assessments are completed immediately following HIV diagnosis to maximise opportunities for linkage into ART. In this study, a novel ART eligibility assessment tool that could be used by CHWs was developed and its accuracy compared with that of the current WHO clinical staging system.

7.2. Methods

7.2.1. Type of study

This was an evaluation of the accuracy of a novel tool used by CHWs for determining the eligibility for ART of newly diagnosed HIV-positive adults in a community-setting according to WHO guidelines. It was nested within a cluster randomised trial investigating strategies to improve linkage into HIV care (Chapter 6).

7.2.2. Study site and population

This study was conducted in three high-density urban townships in the north west of

Blantyre, Malawi. Adult residents (≥16 years) of 14 clusters were provided with access to community-based HIV testing through trained Community Counsellors who were resident in each cluster. Community-based HIV self-testing was available for six months

270 Chapter 7 | Evaluation of accuracy of CHW tool in each cluster during January 2012 and November 2012 (Chapter 3.5). Eligible participants for this study were cluster residents who reported a positive HIVST result to community counsellors.

7.2.3. Study procedures

The CHW tool was designed based on WHO’s “Ask, Look, Classify, Act” approach incorporated within guidelines for the Integrated Management of Adult and Adolescent

Illness (IMAI)476. Questions were constructed for inclusion in the CHW tool from existing components of the WHO clinical staging system247; with additional validated questions that have previously assessed to be important predictors of ART eligibility (self-rated general health272, 477) and advanced immunosuppression (self-rated skin or hair change)478-480 . All questions included had to be easily understood and performed by

CHWs and not require further laboratory investigations. The CHW tool classifies HIV- positive participants into two groups: those who are ART eligible (analogous to WHO stage 3 or 4) and those who are not ART eligible (analogous to WHO stage 1 or 2).

Individuals who responded positively to any of the following items were classified as ART eligible:

• Reports fever or weight loss or diarrhoea for all of last month247

• Enough weight loss to affect fit of clothing247

• Sore mouth or swallowing247

• Treated for TB or any other infection (excluding malaria and “flu”) in the last

year247

• Self-rated change in skin or hair478-480

• Unable to carry out daily duties (work, go to school, do housework)247

271 Chapter 7 | Evaluation of accuracy of CHW tool

• Self-rated general health poor or fair477

The tool was translated into the Chichewa language and back-translated to English to check for consistency. It was pre-tested by Community Counsellors in a pilot study to ensure practicality and ease of use.

272 Chapter 7 | Evaluation of CHW tool

Figure 7.1: Community health worker ART eligibility assessment tool

273 Chapter 7 | Evaluation of CHW tool

7.2.4. WHO clinical staging assessment and measurement of CD4 count

Following completion of the CHW ART eligibility tool and post-test counselling, participants received a home visit from a study nurse within three days. Study nurses performed confirmatory HIV testing using finger-prick blood samples (serial testing using

Determine 1/2™, Alere, Waltham, USA and Uni-Gold™ Recombigen® HIV, Trinity Biotech,

Bray, Ireland).

Study nurses received training in HIV care provided by the Ministry of Health of Malawi that included instruction in the Revised (2007) WHO Clinical Staging System for Adults246.

Participants’ WHO clinical stage was recorded on a standardised case record form and study nurses were blinded to the result of the CHW tool. If participants were symptomatic of TB on a symptom screen481, two sputum samples were taken for smear and culture. If TB was confirmed, the participant’s WHO clinical stage was revised accordingly. Participants screening negative for TB were offered isoniazid preventive therapy. Participants’ self-rated general health was recorded on a 4-item Likert-type scale (“excellent”, “good”, “fair” or “poor”)270, 272.

Participants who met national ART eligibility criteria (CD4 count ≤350 cells/mm3 or WHO stage 3 or 4, or pregnant, or breastfeeding) received 2-weeks of ART at home using standard Ministry of Health recommended treatment regimens (Chapter 2.6). Following home initiation of ART, participants were referred to primary health care centres for continuing ART treatment.

274 Chapter 7 | Evaluation of CHW tool

7.2.5. Statistical methods

The Standards for Reporting of Diagnostic Accuracy Studies (STARD) guidelines were followed482. Baseline characteristics of participants were stratified by sex and compared using the t-test for continuous variables, and the chi-square test for categorical variables. For each of the ART eligibility assessment tools assessed (CHW tool and WHO clinical staging assessment), sensitivity, specificity, positive-predictive value and negative predictive value were calculated with CD4 count of ≤350 cells/mm3 as the referent. As

WHO has recently recommended expanding ART eligibility to include individuals with

CD4 ≤500 cells/mm3 230, analysis was repeated with this criteria as the referent.

Asymptotic 95% confidence intervals were calculated. Receiver operator characteristic

(ROC) curves were plotted and the areas under the curves (AUC) compared using

Bonferroni correction for multiple comparisons. To investigate the relative performance of the CHW tool and the WHO clinical staging system, sensitivity and specificity were modelled using generalised linear models with log-link functions, adjusting for clustering for multiple tests performed per participant, as well as age and sex.

Previous studies have shown the sensitivity and specificity of WHO clinical staging to detect a CD4 count of <350 cells/mm3 in resource-limited settings to be approximately

50% and 80% respectively (Chapter 2.4). Assuming that the prevalence of ART eligibility in HIV-positive community members disclosing their status to community counsellor was

50%, a sample size of 191 patients was required to estimate sensitivity and specificity with a precision of ±10% at the p=0.05 level483.

275 Chapter 7 | Evaluation of CHW tool

7.2.6. Ethical considerations

Ethical approval for the parent cluster randomised trial was obtained from Liverpool

School of Tropical Medicine, College of Medicine, University of Malawi and London

School of Hygiene and Tropical Medicine. All participants gave written informed consent

(or witnessed thumbprint where illiterate) to participate.

7.3. Results

7.3.1. Baseline Characteristics

Between February and November 2012, 213 adult cluster residents underwent self- testing for HIV and were referred for home assessment and initiation of ART (Figure 7.2).

Seven individuals declined assessment for eligibility. 198 of 206 individuals assessed for study eligibility consented to participate and were recruited to the study (96.1% participation).

276 Chapter 7 | Evaluation of CHW tool

Figure 7.2: Participant flow in diagnostic accuracy study

Men comprised 42% (84/198) of participants (Table 7.1) and, compared to women, were more likely to be older (mean 37 years vs. 31 years; p<0.001), literate (78/84 [94.0%] vs.

97/114 [85.8%]; p=0.069) and receive a regular salaried income (31/84 [40.3%] vs. 8/114

[8.2%]; p<0.001). A higher proportion of women reported being widowed, separated or divorced (36/114 [31.6%] vs. 14/84 [16.7%]; p=0.023) and having completed only primary-level education (75/114 [65.8%] vs. 39/84 [46.4%]; p=0.042) compared to men.

277 Chapter 7 | Evaluation of CHW tool

A greater proportion of women (85/114 [75.9%]) than men (46/84 [56.1%]; p=0.004) had previously tested for HIV. Women (24/114 [22.2%]) were also more likely to have undergone HTC in the past 12-months than men (12/84 [14.5%]; p=0.174), although this difference was not statistically significant. A substantial proportion of men (20/84

[23.8%]) and women (39/114 [34.5%]) reported that they had previously tested HIV- positive prior to undergoing community-based HTC.

278 Chapter 7 | Evaluation of CHW tool

Table 7.1: Baseline characteristics of study participants Men Women P-value† Total 84 114 Age in years (mean, SD) 37 8.6 31 8.3 <0.001 Pregnant (if female) n/a 2 1.8% Breastfeeding (if female) n/a 19 16.7% Body mass index in kg/m2 (mean, SD) 20 2.8 22 4.1 <0.001 Marital status Married/cohabiting 61 72.6% 73 64.0% 0.023 Never married 9 10.7% 5 4.4% Widowed/separated/divorced 14 16.7% 36 31.6% Ever lost a spouse to death 16 19.3% 26 22.8% 0.569 Able to read a one page letter or 78 94.0% 97 85.8% 0.069 Receives a regular salaried income 31 40.3% 8 8.2% <0.001 newspaper Highest level of education No schooling 3 3.6% 1 0.9% 0.042 Primary 39 46.4% 75 65.8% Secondary 40 47.6% 36 31.6% Higher 2 2.4% 2 1.8% Ever previously tested for HIV 46 56.1% 85 75.9% 0.004 Tested for HIV in preceding 12-months 12 14.5% 24 22.2% 0.174 Ever previously tested HIV-positive 20 23.8% 39 34.5% 0.105 Tested as a couple 25 29.8% 28 24.6% 0.414 Self-rated general health Excellent 8 9.5% 10 8.8% 0.556 Good 51 60.7% 75 66.4% Fair 22 26.2% 27 23.9% Poor 3 3.6% 1 0.9% WHO clinical stage Stage 1 or 2 65 77.4% 102 89.5% 0.021 Stage 3 or 4 19 22.6% 12 10.5% CD4 count in cells/mm3 (median, IQR) 262 148-373 374 195-512 0.002 CD4 count group >500 cell/mm3 11 13.1% 30 26.3% 0.015 >350-≤500 cells/mm3 15 17.9% 32 28.1% >200-≤350 cells/mm3 27 32.1% 23 20.2% >50-≤200 cells/mm3 30 35.7% 26 22.8% <50 cells/mm3 1 1.2% 3 2.6% CHW ART eligibility assessment tool ART eligible 28 33.7% 32 28.1% 0.394 Not ART eligible 55 66.3% 82 71.9% † T-test for comparison of means, rank sum test for comparison of medians and chi-square test for categorical data SD: standard deviation IQR: Interquartile range CHW: Community health worker

279 Chapter 7 | Evaluation of CHW tool

7.3.2. Outcomes of ART eligibility assessments

31/198 (15.6%) individuals were assessed to be in WHO stage 3 or 4 by nurses, with men significantly more likely to be in the more advanced WHO stage 3 or 4 than women

(19/84 [22.6%]) vs. 12/114 [10.5%] respectively; p=0.021). Men also had significantly lower median CD4 counts at assessment compared to women (262 cells/mm3

[interquartile range [IQR]: 148-373] vs. 374 cells/mm3 [IQR: 195-512], p=0.002). A total of 110/198 (55.6%) participants had a CD4 count below the Malawian national ART eligibility threshold of ≤350 cells/mm3.

7.3.3. Performance of CHW tool in identifying CD4<350 cells/mm3

One participant did not have the CHW tool completed by the community counsellor, leaving 197 participants who were analysed with complete data for outcome of CHW tool, CD4 count and WHO clinical staging assessment.

The sensitivity of the CHW tool (41.3%, 95% CI: 31.9% - 51.1%) was substantially higher than that of nurse-performed WHO clinical stage 3 or 4 assessment (19.1%, 95% CI:

12.2% - 27.7%) in identifying CD4 count ≤350 cells/mm3 (Table 7.2). The specificity of the

CHW tool was 83.0% (95% CI: 73.4% - 90.1%) compared to 88.6% (95% CI: 80.1% -

94.4%) for WHO clinical stage 3 or 4 assessment. The area under the curve (AUC) was significantly greater for the CHW tool (0.62, 95% CI: 0.56 – 0.68) compared to WHO clinical stage 3 or 4 assessment (0.54, 95% CI: 0.49 – 0.59; p= 0.017).

Overall, the positive predictive value (PPV) of the CHW tool was higher than the WHO clinical staging assessment (75.0%, 95% CI: 62.1% - 85.3% vs. 67.7%, 95% CI: 48.6% -

83.3%), although the 95% confidence intervals overlapped. Negative predictive values

280 Chapter 7 | Evaluation of CHW tool

(NPV) of both screening tools were low, with the NPV for the CHW tool (53.2%, 95% CI:

44.6% - 61.9%) higher than for the WHO clinical staging assessment (46.7%, 39.0% -

54.6%) and 95% confidence intervals overlapping.

Table 7.2: Accuracy of novel community health worker ART eligibility screening tool for identifying CD4<350 cells/mm3 vs. CD4 ≤ 350 Sensitivity Specificity PPV NPV AUC cells/mm3 (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) Overall

WHO clinical 19.1% 88.6% 67.7% 46.7% 0.54 stage 3 or 4 (12.2% - 27.7%) (80.1% - 94.4%) (48.6% - 83.3%) (39.0% - 54.6%) (0.49 - 0.59) 41.3% 83.0% 75.0% 53.2% 0.62 CHW tool (31.9% - 51.1%) (73.4% - 90.1%) (62.1 – 85.3%) (44.6% - 61.9%) (0.56 - 0.68)

Among men

WHO clinical 25.9% 84.6% 78.9% 33.8% 0.55 stage 3 or 4 (15.3% - 39.0%) (65.1% - 95.6%) (54.4% - 93.9%) (22.6% - 46.6%) (0.46 - 0.64) 45.6% 92.3% 92.9% 43.6% 0.69 CHW tool (32.4% - 59.3%) (74.9% - 99.1%) (76.5% - 99.1%) (30.3% - 57.7%) (0.61 - 0.77)

Among women

WHO clinical 11.5% 90.3% 50.0% 54.9% 0.51 stage 3 or 4 (4.4% - 23.4%) (80.1% - 96.4%) (21.1% - 78.9%) (44.7% - 64.8%) (0.45 - 0.57) 36.5% 79.0% 59.4% 59.8% 0.58 CHW tool (23.6% - 51.0%) (66.8% - 88.3%) (40.6% - 76.3%) (48.3% - 70.4% (0.49 - 0.66) PPV: Positive predictive value NPV: Negative predictive value AUC: Area under receiver operator curve CHW: community health worker

Unexpectedly, there were differences in test performance when results were stratified by gender (Figure 7.3). The sensitivity and specificity of the CHW tool were higher among men (45.6% and 92.3% respectively) than among women (36.5% and 79.0% respectively). For both men and women however, the PPV (92.9% vs. 78.9% and 59.4% vs. 50.0% respectively) and NPV (43.6% vs. 33.8% and 59.8% vs. 54.9% respectively) of the CHW tool was higher than that of WHO clinical staging. The largest AUC was observed for men identified as ART eligible by the CHW tool (AUC=0.69), which was

281 Chapter 7 | Evaluation of CHW tool significantly higher than for WHO stage 3 or 4 assessed men (AUC=0.55, p=0.016). For women, although the AUC for the CHW tool (AUC=0.58) was greater than for WHO stage

3 or 4 assessment (AUC=0.51), the difference was relatively small and did not reach statistical significance (p=0.104).

Figure 7.3: Receiver Operator Characteristic (ROC) curves comparing community health worker tool and WHO clinical staging system with CD4 count ≤350 cells/mm3 for assessment of ART eligibility, stratified by sex

100%

90%

80%

70%

60% Men CHW tool 50% Men WHO stage 40%

Sensitivity Women CHW 30% tool

Women WHO 20% stage

10%

0% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

1-Specificity

7.3.4. Characteristics associated with performance of community health workers screening tool

On unadjusted analysis, the CHW tool outperformed the WHO clinical staging system in ruling out ART eligibility (odds ratio ([OR] for sensitivity: 2.98, 95% CI: 1.74 – 5.07) –

282 Chapter 7 | Evaluation of CHW tool

Table 7.3. Although not statistically significant, the relative sensitivity was higher in men compared with women (OR: 1.75, 95% CI: 0.92 – 3.33). There was no evidence of effect modification on sensitivity of test performance by sex (OR for interaction: 0.54, 95% CI:

0.18 – 1.65). After adjusting for the effect of age and sex, the relative sensitivity of the

CHW tool remained significantly higher than for WHO clinical staging assessment (OR:

3.05, 95% CI: 1.78 – 5.25).

Table 7.3: Regression model fit to participant data for community health worker tool and WHO clinical staging assessment using sensitivity as dependent variable

Adjusted Relative relative sensitivity 95% CI 95% CI sensitivity (odds ratio) (odds ratio) Age 1.02 0.98 – 1.06 1.01 0.97 – 1.06 Sex Women ref ref Men 1.75 0.92 – 3.33 1.69 0.83 – 3.46 ART eligibility tool WHO stage 3 or 4 assessment ref ref CHW tool 2.98 1.74 – 5.07 3.05 1.78 – 5.25 Unadjusted interaction between ART eligibility assessment tool and sex: 0.54, 95% CI: 0.18-1.65

The tests also differed in their ability to rule in ART eligibility (Table 7.4). On unadjusted analysis the CHW tool had a higher relative specificity (OR: 1.60, 95% CI: 0.78 – 3.29), although this was not statistically significant. However, there was evidence of effect modification by sex, with men assessed as ART eligible by the CHW tool men having an odds ratio for relative specificity four-fold higher than for women who were assessed as being in WHO stage 3 or 4 (OR: 4.02, 95% CI: 0.62 – 28.23). After adjusting for the effect of age, sex and the interaction between sex and ART eligibility assessment tool, the relative specificity of the CHW tool was significantly higher than for WHO clinical staging assessment (OR: 2.49, 95% CI: 1.07 – 5.79).

283 Chapter 7 | Evaluation of CHW tool

Table 7.4: Regression model fit to participant data for community health worker tool and WHO clinical staging assessment using specificity as dependent variable

Adjusted Relative relative specificity 95% CI 95% CI specificity (odds ratio) (odds ratio)

Age 0.96 0.91 – 1.03 0.97 0.91 – 1.03 Sex Women ref ref Men 0.72 0.24 – 2.19 2.01 0.53 – 7.66 ART eligibility tool WHO stage 3 or 4 assessment ref ref CHW tool 1.60 0.78 – 3.29 2.49 1.07 – 5.79 Interactions* WHO stage 3 or 4 assessment in women ref ref WHO stage 3 or 4 assessment in men 1.70 0.43 – 6.64 2.01 0.53 – 7.66 CHW tool positive in women 2.47 1.06 – 5.76 2.49 1.07 – 5.79 CHW tool positive in men 4.02 0.62 – 28.23 5.02 0.80 – 31.51 *Unadjusted interaction between ART eligibility assessment tool and sex: 0.19, 95% CI: 0.03 – 1.08

7.3.5. Secondary analysis

WHO issued revised ART eligibility guidelines in July 2013. These guidelines recommended that HIV-positive individuals with a CD4 count of ≤500 cells/mm3 are eligible for ART initiation230. To assess the performance of the CWH eligibility assessment tool against these proposed revised criteria, the analyses were repeated, using CD4<500 cells/mm3 as the referent.

With ART eligibility defined by CD4 count ≤500 cells/mm3, 157/198 (79.3%) of participants would be eligible for ART initiation. A significantly higher proportion of men

(73/84, 86.9%) than women (84/114, 73.7%; p=0.023) would be ART eligible.

The sensitivity of the CHW tool (33.3%, 95% CI: 26.0% - 41.3%) remained low, although substantially and significantly higher than that of nurse-performed WHO clinical stage 3 or 4 assessment (15.9%, 95% CI: 10.6% - 22.6%) in identifying CD4 count ≤500 cells/mm3

284 Chapter 7 | Evaluation of CHW tool

(Table 7.5). The specificity of the CHW tool was 80.5% (95% CI: 65.1% - 91.2%) compared to 85.4% (95% CI: 70.8 – 94.4%) for WHO clinical stage 3 or 4 assessment. The area under the curve (AUC) was greater for the CHW tool (0.57, 95% CI: 0.50 – 0.64) compared to WHO clinical stage 3 or 4 assessment (0.51, 95% CI: 0.45 – 0.57), although this did not reach statistical significance (p=0.155). Although the PPVs of both the CHW tool and the WHO clinical staging assessment were high (86.7% and 80.6% respectively), the NPVs were both suboptimal (24.1% and 21.0% respectively).

285 Chapter 7 | Evaluation of CHW tool

Table 7.5: Accuracy of novel community health worker ART eligibility assessment tool for identifying CD4 ≤500 cells/mm3 vs. CD4 ≤ 500 Sensitivity Specificity PPV NPV AUC cells/mm3 (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) Overall

WHO clinical 15.9% 85.4% 80.6% 21.0% 0.51 stage 3 or 4 (10.6% - 22.6%) (70.8% - 94.4%) (62.5% - 92.5%) (15.1% - 27.9%) (0.44 - 0.57) 33.3% 80.5% 86.7% 24.1% 0.57 CWH tool (26.0% - 41.3%) (65.1% - 91.2%) (75.4% – 94.1%) (17.2% - 32.1%) (0.50 - 0.64)

Among men

WHO clinical 23.3% 81.8% 89.5% 13.8% 0.53 stage 3 or 4 (14.2% - 34.6%) (48.2% - 97.7%) (66.9% - 98.7%) (6.5% - 24.7%) (0.40 - 0.65) 38.9% 100% 100% 20.0% 0.69 CWH tool (27.6% - 51.1%) (71.5% - 100%) (87.7% - 100%) (10.4% - 33.0%) (0.64 - 0.75)

Among women

WHO clinical 9.5% 86.7% 66.7% 25.5% 0.48 stage 3 or 4 (4.2% - 17.5%) (69.3% - 96.2%) (34.9% - 90.1%) (17.4% - 35.1%) (0.41 - 0.55) 28.6% 73.3% 75.0% 26.8% 0.51 CHW tool (19.2% - 39.5%) (54.1% - 87.7%) (56.6% - 88.5%) (17.6% - 37.8%) (0.42 - 0.60) PPV: Positive predictive value NPV: Negative predictive value AUC: Area under receiver operator curve CHW: Community health worker

7.4. Summary of findings

The main finding from this study was that a novel brief CHW tool significantly outperformed the current WHO clinical staging system for identifying ART eligible individuals with CD4 count of less than 350 cells/mm3. The ease with which we were able to improve upon the current WHO staging system emphasises the need for improved tools that can be used by community and facility-based healthworkers to ensure “same-day, same-site” eligibility assessments and prompt linkage to ART.

Moreover, the low sensitivity and high risk of misclassification when using the standard

WHO clinical staging system in this and other studies255, 259, 264, 265 questions its appropriateness as an ART eligibility assessment tool, especially with increasing CD4 count thresholds222.

286 Chapter 7 | Evaluation of CHW tool

The CHW tool was found to be significantly more accurate than WHO clinical staging assessment for identification of CD4 count ≤350 cells/mm3 in terms of sensitivity, positive predictive value and negative predictive value. In contrast to the WHO clinical staging system, which, strictly followed, requires nurses or doctors to undertake a complex and time-consuming assessment477, the CHW tool was easily performed in the community and could be used in settings without sophisticated laboratories. The poor performance of nurse-performed WHO clinical staging system in this and other studies255, 259, 264, 265 is therefore not a factor of the capability of nurses. Instead it is a reflection of the inherent limitations of the WHO screening tool for identifying ART eligibility, a task for which it was never originally intended.

The CD4 count is the gold standard for determining ART eligibility, but access to laboratory measurement is problematic in most resource-limited settings353. Point of care CD4 count measurement has the potential to negate some of these issues and improve retention in pre-ART care236, but is not yet widely available. Where CD4 count is not available at the point of HIV diagnosis, improved eligibility assessment tools such as this CHW tool could allow an immediate decision on ART initiation to be made, facilitating “test-and-treat” delivery of ART

7.5. Limitations

Although the CHW outperformed the WHO clinical staging system in identifying HIV- positive individuals with CD4 count of ≤350 cells/mm3, the negative predictive value remained low at 53%. This means that, in individuals screened as ineligible by the CHW tool, would have a one-in-two chance of in fact being ART eligible. The negative

287 Chapter 7 | Evaluation of CHW tool predictive value when compared with CD count ≤500 cells/mm3 as the referent was lower. Because of the risks of both unnecessarily withholding ART, and committing individuals to take lifelong treatment (with concomitant risk of side effects and worsened adherence), the ideal screening test should have high positive and negative predictive values. Therefore, the CHW tool in its current form is unlikely to be the “ideal” screening test for assessing ART eligibility and further refinements to improve accuracy in this setting will be required.

As the WHO-recommend threshold for ART eligibility has been increased to CD4 ≤500 cells/mm3 230, the poor performance of the WHO clinical staging system will raise a substantial challenge in settings where CD4 count is not available. In this study, reflecting the small number of previous studies described in Chapter 2.4, accuracy of the

WHO clinical staging system was very poor (little better than random), with extremely low sensitivity and negative predicative values.

Although the area under the ROC curve was higher, the novel CHW tool assessed here did not significantly improve upon WHO clinical staging at this CD4 count threshold.

There is then a real risk that, in areas where CD4 count measurement is not easily and affordably available to all, a substantial proportion and absolute number of HIV diagnosed individuals will be unnecessarily delayed in initiating ART, limiting efforts to achieve coverage and reduced HIV-associated morbidity, mortality and transmission. An alternative approach, which is currently being considered by WHO for introduction from

2015 and is supported by clinic trial data showing a 96% reduction in HIV transmission, is initiation of ART immediately following HIV diagnosis, regardless of CD4 count or clinical stage: so-called “test-and-treat”. This approach has already been introduced into Malawi for pregnant women, known as Option B+. However, although universal test-and-treat is

288 Chapter 7 | Evaluation of CHW tool likely to be highly cost-effective, such an approach may be unaffordable for national HIV care programmes in resource-limited settings with generalised HIV epidemics and will require substantial donor support, as well as considerably improved uptake of regular

HTC.

The gold standard for comparison in this study (CD4 count threshold) was selected to align with WHO ART eligibility criteria. However, independent of CD4 count, advanced

WHO clinical stage (when staging is comprehensively performed) has strong prognostic value in assessing risk of disease progression and mortality in HIV-positive individuals.

Future studies may wish to evaluate the prognostic value of the CHW tool compared to

WHO clinical stage and CD4 count. Understanding the incremental value of the CHW in addition to WHO clinical stage as a predictor of ART eligibility measured by CD4 count would also be beneficial, but would require a larger sample size. As a greater number of participants in the HitTB trial received the CHW tool assessment, clinical staging and CD4 count measurement, these analyses may be possible in the future.

WHO clinical staging assessments were performed by nurses trained in HIV care in patients’ homes and with reference to checklists. However, nurses were not able to access radiographic or diagnostic laboratory services (beyond sputum smear microscopy and culture for tuberculosis). Therefore, WHO clinical staging was not performed as fully as would be done, for example, in tertiary hospital settings. The sensitivity of the WHO clinical staging system may have been improved with access to “optimal” diagnostic support. Nevertheless, as demonstrated by the quantitative and qualitative data presented in Chapters 4 and 5, this does not reflect how WHO clinical staging assessments are carried out in routine clinical practice in the primary health care system. In primary care clinics, where the majority of HIV care is now provided, there is

289 Chapter 7 | Evaluation of CHW tool little access to diagnostic laboratory or radiology services (beyond sputum smear microscopy) and healthworkers find the WHO clinical staging system overly complex and time-consuming. With huge demand for clinical services, they are unable to complete

WHO clinical staging at any level beyond a superficial assessment of outwardly obvious signs. WHO clinical staging assessment performed by nurses in this study was similar, or perhaps even superior to that provided in primary healthcare centres, meaning that the comparison between the CHW tool and WHO clinical staging assessment was valid and generalizable to real-life clinical practice in this setting.

The accuracy of the CHW tool was noted to be higher among men than women. This may reflect the relatively more advanced immunosuppression of men compared to women in the study, consistent with findings from HIV care programmes in sub-Saharan

Africa showing that men initiate ART later than women222. Women may have also been less likely to report symptoms included in the tool. Future, larger studies should examine the performance of CHWs tool among men and women across CD4 strata. The sex of the

CHW performing the tool may have influenced participant responses and could explain the observed gender-differences in accuracy. Sex of CHW was not captured, but is being captured as part of the HitTB study and will be analysed as part of future studies.

The CHW tool was designed pragmatically to combine ease of use by CHWs with questions that were hoped would have a high discriminatory power. Although not formally assessed by timing of completion and error rates, anecdotal evidence suggests that CHWs found the tool easy and quick to perform. Component parts of the tool were not captured and so we were unable to assess the contribution of individual items to the tool performance. As such it is possible that the number of items could be reduced, or the sensitivity and specificity improved beyond that reported here. Future research

290 Chapter 7 | Evaluation of CHW tool should attempt to refine the tool by examining the contribution of the included items to the test performance and evaluating any modifications.

7.6. Findings relating to other studies

A systematic review presented in Chapter 2.4.3 identified 15 studies that have evaluated the diagnostic accuracy of the WHO clinical staging system at CD4 count thresholds of

≤200 cells/mm3 (11 studies), ≤350 cells/mm3 (6 studies) and ≤500 cells/mm3 (1 study)

Meta-analysis found that the overall summary estimate for the sensitivity and specificity of WHO clinical stage 3 or 4 disease in identifying CD4 ≤350 cells/mm3 to be 45% and

85% respectively. Sensitivity was noted to be lower in the subset of studies conducted among participants recruited from communities and from antenatal clinics255, 261 where median CD4 counts were higher than in studies conducted among facility attendees and hospital inpatients. In this study, the sensitivity and specificity of WHO clinical stage 3 or

4 was similar to that seen in the community/ANC studies identified in the systematic review.

One previous study examined the accuracy of a CHW-performed abbreviated WHO clinical staging checklist in community members in Karonga District, Malawi. The abbreviated checklist failed to identify two-thirds of ART eligible participants (WHO stage 3 or 4 or CD4≤250 cells/mm3) and the authors recommended it not be used for

ART eligibility assessment. Although our CHW tool performed better than this, still nearly half of ART eligible participants would be missed.

7.7. Implications for future studies

Results from the study clearly demonstrate that, in the absence of widely available capacity for CD4 count measurement (including the only very limited availability of point

291 Chapter 7 | Evaluation of CHW tool of care CD4 count measurement) alternative approaches to screening for ART eligibility are urgently required. This is particularly important with increasing implementation of home- and community-based HTC, meaning that HIV-diagnosed individuals healthworkers will require an assessment that allows an immediate decision over whether to start ART, reducing the risk of dropout from care. The CHW tool described here is one such approach, but other tools should be developed and assessed. Future studies should also attempt to refine this CHW tool by determining the items that most efficiently contribute discriminatory information to the tool and refine the number of items (e.g. by using principal components analysis). Alternative items could also be evaluated.

Cohort studies or clinical trials could evaluate the impact of use of the CHW tool on retention in care and ART initiation. Additionally, evaluation of the accuracy of self- completed ART eligibility assessment following self-testing, and impact on linkage into

ART care could provide a novel approach to increasing access to treatment. Ultimately, clinical trials evaluating the impact of point of care CD4 count on linkage into ART for community-diagnosed HIV-positive individuals are required.

7.8. Conclusions

In conclusion, this study has shown that, a novel screening tool performed by CHWs outperformed the current WHO clinical staging system for identifying ART eligible community members following HIV self-testing. This could be incorporated into HTC counsellors’ post-test evaluations, allowing same-day, same-site HIV diagnosis and ART eligibility assessment. Although accuracy of the CHW tool is not optimal, and worsens at the now recommended higher CD4 count threshold of ≤500 cells/mm3, the ease with which we were able to improve upon the current WHO clinical staging system shows the

292 Chapter 7 | Evaluation of CHW tool urgent need for improved ART eligibility assessment tools. Alternative approaches, such as “universal test-and-treat” would bypass the need for ART eligibility assessments and minimise missed opportunities for ART initiation. However, the programmatic costs and long term individual impacts on side effects and retention are not known.

293 Chapter 7 | Evaluation of CHW tool

294 Chapter 8 | Summary, recommendations and conclusions

8. Summary, recommendations and conclusions

8.1. Summary of rationale, objectives and key findings

The potential preventative impact of widespread population coverage of ART on HIV transmission gained international attention with the publication of a mathematical modelling study in 20097. This study suggested that high population rates of uptake of

HTC followed by immediate ART initiation (“universal test and treat”) with subsequent high rates of retention in care and adherence to treatment could rapidly reduce epidemics towards elimination484. Mathematical models have been supported by clinical trial data showing a 96% reduction in HIV transmission in serodiscordant couples where

ART was initiated immediately60, and in a population cohort study, where higher levels of ART coverage were strongly associated with reduced risk of HIV incidence4.

These studies, as well as strong evidence of individual patient benefit in terms of reduced mortality and morbidity and improved quality of life have reinvigorated healthworkers and policymakers to work towards achieving the goal of universal access to HIV care and prevention. Enthusiasm has however been tempered by an increased recognition that the current low rates of HTC and high rates of patient loss before ART initiation seen in most countries of sub-Saharan Africa represent a major potential barrier to achieving universal coverage328-330. Prior to this thesis however, no studies in sub-Saharan Africa had estimated loss to care prior throughout all steps in the HIV care pathway, nor had undertaken a rigorous clinical trial to improve upon ART uptake.

The studies conducted in this thesis therefore set out to:

• Investigate the magnitude and reasons for loss to care prior to ART initiation in

Blantyre, Malawi and

295 Chapter 8 | Summary, recommendations and conclusions

• Develop and evaluate interventions to improve entry to into HIV care

A number of novel findings were identified. Uptake of provider initiated HTC was suboptimal under routine programmatic conditions, with only 13% of facility attendees testing, and men and non-pregnant women faring particularly poorly. This was contrary to previous studies that have shown higher rates of acceptance of PITC during research studies153. There was substantial patient loss to care at each step following HIV diagnosis, with only one-third of HIV-positive participants initiating ART by 6-months and with long delays to treatment initiation. Failure to complete ART eligibility assessments was associated with failure to initiate ART and was a common point of drop out. In a linked qualitative study, WHO clinical staging assessments for ART eligibility assessment were viewed by health workers as being overly complex and time-consuming and, for patients, access to CD4 count measurement through the city’s Central Hospital was expensive and subject to administrative and laboratory failures.

The intervention that was developed (home initiation of HIV care following self-testing) was designed to overcome the identified barriers to universal access to HTC and ART. In particular, the HIVST component facilitated convenient and confidential access to HTC in communities, evidenced by the high uptake (58%) of self-testing in both study arms over a short period of time (6-months). Home assessment and initiation of HIV care removed the need for multiple unnecessary facility visits, as well as streamlining the ART eligibility assessment process. The substantial and significant increase in population-level ART initiations where optional home initiation of HIV care was available shows the potential for this approach to improve linkage into care. However, the finding that, without additional intervention following diagnosis, provision of HIVST alone had no appreciable impact on ART uptake highlights the fact that HIVST interventions alone may be

296 Chapter 8 | Summary, recommendations and conclusions insufficient to increase ART uptake and additional interventions to facilitate linkage into care will be required.

Evaluation of a novel ART eligibility tool performed by community health workers showed that it outperformed the current WHO clinical staging system in terms of accuracy in identifying ART eligibility at CD4 <350 cells/mm3. This tool could be rapidly performed immediately following HIV diagnosis by community healthworkers with minimal training. This tool may be useful, particularly where access to CD4 count measurement is limited, and as HIVST and door-to-door HCT programmes are scaled up in the community. Nevertheless the negative predictive value of the CHW tool at the

CD4 count <350 cells/mm3 is low and accuracy worsens at the newly WHO- recommended ART eligibility cut off of CD4 count <500 cells/mm3 (although is still better than the WHO clinical staging system). In the absence of widely available point of care

CD4 count measurement, all current ART eligibility assessment tools are suboptimal, a situation that lends strength to current calls for a universal test and treat approach to management of HIV.

8.2. Recommendations and future directions

8.2.1. Achieving universal HIV testing and counselling

The findings of the prospective cohort study and qualitative study lend support to a growing body of evidence that facility-based provider initiated HTC is necessary, but not sufficient to achieve the high rates of yearly HTC required of HIV elimination models.

In counties with generalised HIV epidemics, a broad approach to improving uptake of

HTC is required. This approach should include efforts to strengthen exisiting

297 Chapter 8 | Summary, recommendations and conclusions recommendations for PITC, including examination of novel strategies to improve operationalisation. High rates of uptake of PITC have been achieved in antenatal clinics

(for example through Malawi’s Option B+ programme)485 and was acceptable to pregnant women participating in qualitative interviews in this thesis. Although workloads are higher in facility general outpatient clinics and inpatient departments, making HTC part of routine clinical pratice (with clear information for clients and ability to opt-out) is required. Patients should not be required to be referred to a second healthworker to have HTC completed, but should have HTC offered and completed by the first healthworker they are assessed by. There is clear evidence that, except under exceptional circumstances, extensive pre-test counselling is unnessary and adds little to subsequent retention or rates of linkage into care402.

Mouth swab HIV rapid diagnostic kits offer an exciting opportunity to improve patient flow and uptake of HTC in health facilities. Facility attendees could collect a rapid diagnostic kit at facility registration and self-test while waiting for clinical assessment, or test in the presence of healthworker if requested. Such a strategy would have the added benefit of allowing healthworkers and patients to make immediate treatment and prevention decisions rather than requiring attendance at the HTC clinic, or laboratory to provide a venous blood sample. Such strategies could be evaulated in contolled clinical trials against patient-important outcomes.

Only a small proportion of the general population will regularly attend health facilities486, and groups including men and key populations may be reluctant to attend due to cultural norms or fear of stigma and discrimination387. Therefore, accessible community-based HTC programmes run in parallel with faciliy-based HTC are required. A

298 Chapter 8 | Summary, recommendations and conclusions recent systematic review and meta-analysis emphasised the high rates of uptake of HTC that can be achieved with community-based approaches157.

To date, door-to-door HTC programmes may have been over-emphasised in the mix of community-based approaches, with the majority of data coming from studies of such programmes157. Although able to achieve high uptake, door-to-door HTC programmes are expensive and difficult to sustain and may not reach certain groups such as men

(who are more likely to be at work when home visits occur) and marginalised key populations487. Greater integration of door-to-door HTC with other community health programmes, such as active case finding for TB, vaccination campaigns and distribution of impregnated bednets could improve accesssbility and sustainability. Greater investigation of community-based HTC approaches that can reach men and key populations is required. Such approaches could include workplace-based HTC136, HTC campaigns at sports events (e.g. football matches) and outreach HTC at transport links

(e.g. taxi ranks).

Home-based HIVST, shown in the cluster randomised trial have high potential for achieving high coverage over a short time period, goes some way towards filling the current gaps of facility-based PITC and community, but uptake remained suboptimal in certain groups including men and older adults. Moreover, the impact on HIV incidence is not known, and neither is the risk of harm associated with HIVST (although this will be reported as part of the HitTB trial). Incentives (financial or promotional) targeted towards client groups and provided following completion of HIVST could promote increased social desireability for testing and improve uptake among clients. The Western

Cape Provincial Government in South Africa has introduced a lottery system, where individuals undergoing HTC at an approved site stood a chance of winning R10,000488.

299 Chapter 8 | Summary, recommendations and conclusions

High uptake of HTC was seen during the pilot period, with increased rates of testing among men compared to those seen in the provience in general489. A number of randomised controlled trials of various economic and non-financial incentives to promote uptake of HTC are currently planned. Incentivising community health workers distributing HISVT kits could also promote increased coverage among harder-to-reach groups, as has been shown in World Bank pilot study in Rwanda490.

Although WHO recommendations supported by mathematical modelling data recommend that individuals living in countries with generalised HIV epidemics undergo

HTC every six to twelve months120, too few studies have reported on uptake of repeat testing for HIV. The HitTB trial will report on uptake of HIVST over two years and will be able to provide an aggregate estimate for the proportion of the population who test each year. Further studies that evaluate interventions to promote regular uptake of HTC in terms of cost-effectiveness and acceptability are urgently required, and ideally should be compared against current “best practices” approaches to community-based HTC (e.g. door-to-door HTC and HIVST).

The universal test and treat strategy advocates regular (yearly), widespread HIV testing, followed by immediate ART initation, with the population effect on reduced HIV transmission premised on achieving reductions in average community HIV viral loads7.

However, recent mathematical modelling based on HIV incidence data from clinical trials undertaken in Lilongwe, Malawi have highlighted important limitations in this approach, with effectiveness critically dependent on the role of individuals acutely infected with

HIV, who have extremely high viral loads and to whom a substantial proportion of new infections can be attributed121. Individuals in the acute infection stage are rarely aware of their status and frequently do not seek healthcare. Even if they do attend a facility,

300 Chapter 8 | Summary, recommendations and conclusions they are often not tested for HIV, or may be non-reactive to HIV antibodies as they are in the window period.

Under generalised HIV epidemic conditions, HTC approaches that only identify and treat individuals chronically infected with HIV and provide optimal prevention interventions for individuals who test negative will reach HIV elimination, but this will be achieved over an extremely long time period and will required yearly HTC, prevention and treatment coverage rates of between 95-99%121. In contrast, where less than 95% coverage of HTC, prevention and treatment are achieved, identification of between 25% and 75% of individuals in the acute infection stage could achieve rapid reductions in incidence121. Thus, without near-complete coverage, interventions targeted solely at individuals chronically infected with HIV will be suboptimal unless complemented by interventions targeted at individuals in the acute infection stage.

These findings raise important questions for HIV testing programmes. In addition to promptly initiating treatment for individuals diagnosed HIV positive, greater attention needs to be paid to individuals who test HIV negative. Whilst considerable attention has been paid to retention in the HIV care cascade, no studies to date in sub-Saharan Africa have describe outcomes for individuals in the “HIV prevention cascade”. Understanding rates of uptake of HIV prevention interventions (such as voluntary male circumcision, condoms, pre-exposure prophylaxis, vaginal microbicides) following a negative HIV test will be important to inform interventions designed to improve uptake of these interventions in home, community and facility settings.

In the countries of sub-Saharan Africa with generalised HIV epidemics, where up to 2% of adults between 15 and 49 years old are newly infected with HIV each year1,

301 Chapter 8 | Summary, recommendations and conclusions evaluation of approaches to improve early diagnosis and treatment of HIV in the acute infection stage are required, ideally supported by mathematical modelling data. This could include more frequent (e.g. quarterly) home-testing or self-testing (perhaps supported by mobile and internet reminder systems), for individuals who test HIV negative. Fourth generation rapid HIV diagnostic kits that can identify the presence of the p24 antigen and allow diagnosis of HIV as soon as 10 days after infection are available and could be used if shown to be cost-effective. However, there are limited data available from studies in Malawi and Liverpool to show that the sensitivity of these kits may be suboptimal, leading to high rates of false positivity and excess requirement for confirmatory testing and patient concern491, 492. Assuming more accurate test fourth generation HIV rapid kits are developed, or exisiting technologies for point of care HIV viral load detection are evaluated as accurate as part of an acute HIV-focused diagnostic algorithm, these could be evaluted against against current HIV testing approaches with a trial endpoint of HIV incidence.

8.2.2. Achieving universal HIV treatment

The suboptimal rates of linkage into HIV care shown in the prospective cohort study in

Chapter 4, and supported by data from a large number of programmes in the region, emphasise that current approaches are insufficient to achieve the high rates of prompt and complete uptake of ART required to impact upon epidemics. This is in stark contrast to TB programmes, where meta-analysis shows that approximately 85% of individuals diagnosed with smear-positive TB in resource-limited settings will successfully start TB treatment493. The key differences between the TB care pathway and HIV care pathway is that HIV patients are required to undergo an eligibility assessment to determine whether treatment should be started.

302 Chapter 8 | Summary, recommendations and conclusions

As clearly demonstrated here, the eligibility assessment process is extremely problematic for patients, being a common point of drop out. Definitive randomised controlled trial results for the long term cost-effectiveness and harms of immediate ART initiation compared to deferred ART initiation until eligibility criteria are met, in sub-

Saharan African settings are not yet available (and may not be for a number of years to come). Nevertheless, the quantifiable and serious risks to individuals and populations of delayed or missed treatment initiation from setting strict ART eligibility criteria, compared to the hypothetical and marginally increased risk of small excess rates of drop-out from treatment, drug toxicity, treatment failure and development of resistance, makes a compelling argument for a policy of universal test and treat for HIV.

WHO has already moved towards such an approach in the 2013 consolidated HIV treatment guidelines331. In these new guidelines, individuals with CD4 count<500 cell/mm3, in WHO stage 3 or 4, with active TB, with HIV/hepatitis B co-infection, in sero- discordant couples, and pregnant and breastfeeding women and children are all eligible to start ART. If fully implemented, these guidelines would mean that an estimated 76% of HIV infected individuals worldwide would be eligible to start ART331.

The cost and difficulty of expanding and maintaining point of care CD4 count capacity from the current low coverage starting point in sub-Saharan Africa to allow same-day, same-site eligibility assessment for the small proportion of individuals not eligible for other reasons seems perverse. Discounting their current use as an ART eligibility assessment tool, CD4 counts add little to the management of HIV infected individuals

(beyond deciding on whether chemoprophylaxis for opportunistic infections should be prescribed). Thus, a strong argument exists for abolishing ART eligibility assessments and implementing universal test and treat.

303 Chapter 8 | Summary, recommendations and conclusions

In the cluster randomised trial reported in Chapter 6, ART initiations were significantly increased at the population-level, without appreciable worsening of retention in care over six months (although numbers were small). These results raise a number of areas for future research to guide strategies to achieve universal access to ART. A previous study in Uganda has shown home continuation of ART following clinic initiation to be non-inferior to clinic-delivered ART in terms of rates of viral suppression. Future studies could examine the effectiveness of a comprehensive home-delivered service: home testing, followed by home initiation of ART and home delivery of continuation treatment. Important outcomes to investigate would include impact on ART coverage, long-term rates of retention (and loss to mortality), viral suppression, acceptability and harm. Although at least three randomised-controlled trials of immediate vs. deferred

ART are underway in Africa (with HIV incidence the primary outcome), no trials are evaluating strategies to operationalize home-based universal test and treat approaches.

Although this trial focused on home initiation of HIV care following HIV self-testing, a considerable number of individuals are still diagnosed with HIV in health facilities.

Interventions to improve linkage from HIV diagnosed in facilities to treatment initiation and after are required. From the findings of the prospective cohort and qualitative studies, interventions that aimed to reduce the number of clinic visits would be desirable. Interventions could include same-day, same-site HIV diagnosis and ART initiation (after rapid screening for TB and other opportunistic infections).

The main arguments against universal test and treat relate to expense and health system capacity in the face of extremely limited resources. In a challenging financial climate, and where the cost savings from earlier diagnosis and treatment of HIV may not

304 Chapter 8 | Summary, recommendations and conclusions be realised in the health service until much later, Ministers of Health, policymakers, researchers, clinicians and patient groups need to advocate strongly from an individual and community rights perspective for the benefits of domestic, bilateral and international investment to achieve universal coverage of ART.

Urgent attention needs to be paid to the issue of facility overload and lack of health system capacity to expand treatment further whislt maintaining quality. Particularly concerning is the high rates of patient loss following ART initiation that continue to be seen in many programmes in the region383. Novel approaches to delivering HIV care, such as home initiaiton of ART following home- and community-based HTC, and possibly even home initiation followed by home continuation of ART, will be required to relieve some of the pressures on health facilities. Alternatively, community-based distribution for ART, for example from shops or pharmacies, or linked to other community-based health activities, could be more convient for patients and free-up clinic capacity.

Advances and expansion in coverage of mobile and internet technologies in Africa could be harnessed to support home- and community-based treatment delivery. For example, adherence reminders received via text message have been shown to be effective in improving adherence494, and could be expanded to allow remote diagnosis, triage and basic management, with support from facility-based practicioners.

In a relatively short period (under ten years), delivery of HIV care and treatment in low- and-middle income countries has been scaled up from virtually zero, to today where nearly 10 million people are regularly receiving ART. This is a major public health achievement. As we move into a new era of HIV care, with recent advances in HIV testing, prevention and treatment coupled to a drive to achieve universal coverage,

305 Chapter 8 | Summary, recommendations and conclusions novel public health responses will be require to meet the increased demands and uncertainty. Instead of concentrating focus on individuals self-presenting to health facilities with signs and symptoms of advanced HIV (who will remain important), it is imperative that we adopt a comprehensive home-, community- and facility-based response that addresses patient needs at all stages of the HIV care pathway. This will require innovative programme design and smart utilisation of existing resources and capabilities, and will be driven by people-living with HIV, their families and friends, policy-makers, practioners and researchers.

8.3. Conclusion

In summary, this thesis has found evidence of low rates of completion of PITC in primary health care centres in Blantyre and high rates of patient loss prior to ART initiation. A major barrier to linkage into HIV care was overly complex, time-consuming and inconvenient ART eligibility assessments. Taken together, these findings show that it will be extremely difficult to achieve the rapid and sustained reductions in HIV epidemics presaged in mathematical modelling studies without full implementation of existing facility- and community-based approaches to HTC and proactive interventions to improve uptake of ART. One approach examined here – home-based initiation of HIV care following HIVST – resulted in a substantial and significant population increases in

ART initiations and reporting of positive results. The combination of HIVST plus optional home initiation of could provide substantial individual and public health benefit and contribute to efforts to achieve universal access to HTC and ART.

306

References

1. UNAIDS. Report on the global AIDS epidemic. 2012 [cited; Available from: http://www.unaids.org/en/media/unaids/contentassets/documents/epidemiology/2012/gr2012/20121120 _UNAIDS_Global_Report_2012_en.pdf

2. Wang H, Dwyer-Lindgren L, Lofgren KT, Rajaratnam JK, Marcus JR, Levin-Rector A, Levitz CE, Lopez AD, Murray CJ. Age-specific and sex-specific mortality in 187 countries, 1970-2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet. 2012; 380(9859): 2071-94.

3. World Health Organization. Global Update on HIV Treatment 2013: Results, Impact and Opportunities. Genvea; 2013.

4. Tanser F, Barnighausen T, Grapsa E, Zaidi J, Newell ML. High coverage of ART associated with decline in risk of HIV acquisition in rural KwaZulu-Natal, South Africa. Science. 2013; 339(6122): 966-71.

5. Lowrance DW, Makombe S, Harries AD, Shiraishi RW, Hochgesang M, Aberle-Grasse J, Libamba E, Schouten E, Ellerbrock T, Kamoto K. A public health approach to rapid scale-up of antiretroviral treatment in Malawi during 2004-2006. J Acquir Immune Defic Syndr. 2008; 49(3): 287-93.

6. Staveteig S, Wang S, Head S, Bradley S, Nybro E. Demographic patterns of HIV testing uptake in sub-Saharan Africa: DHS comparative reports 30. Calverton, Maryland, USA: ICF International; 2013.

7. Granich RM, Gilks CF, Dye C, De Cock KM, Williams BG. Universal voluntary HIV testing with immediate antiretroviral therapy as a strategy for elimination of HIV transmission: a mathematical model. Lancet. 2009; 373(9657): 48-57.

8. Hemelaar J. The origin and diversity of the HIV-1 pandemic. Trends Mol Med. 2012; 18(3): 182-92.

9. Aghokeng AF, Ayouba A, Mpoudi-Ngole E, Loul S, Liegeois F, Delaporte E, Peeters M. Extensive survey on the prevalence and genetic diversity of SIVs in primate bushmeat provides insights into risks for potential new cross-species transmissions. Infect Genet Evol. 2010; 10(3): 386-96.

10. Hahn BH, Shaw GM, De Cock KM, Sharp PM. AIDS as a zoonosis: scientific and public health implications. Science. 2000; 287(5453): 607-14.

11. Robertson DL, Anderson JP, Bradac JA, Carr JK, Foley B, Funkhouser RK, Gao F, Hahn BH, Kalish ML, Kuiken C. HIV-1 nomenclature proposal. Science. 2000; 288(5463): 55.

12. Gao F, Bailes E, Robertson DL, Chen Y, Rodenburg CM, Michael SF, Cummins LB, Arthur LO, Peeters M, Shaw GM. Origin of HIV-1 in the chimpanzee Pan troglodytes troglodytes. Nature. 1999; 397(6718): 436- 40.

13. Keele BF, Van Heuverswyn F, Li Y, Bailes E, Takehisa J, Santiago ML, Bibollet-Ruche F, Chen Y, Wain LV, Liegeois F. Chimpanzee reservoirs of pandemic and nonpandemic HIV-1. Science. 2006; 313(5786): 523- 6.

14. Rowland-Jones SL, Whittle HC. Out of Africa: what can we learn from HIV-2 about protective immunity to HIV-1? Nat Immunol. 2007; 8(4): 329-31.

307

15. Clavel F, Guetard D, Brun-Vezinet F, Chamaret S, Rey MA, Santos-Ferreira MO, Laurent AG, Dauguet C, Katlama C, Rouzioux C. Isolation of a new human retrovirus from West African patients with AIDS. Science. 1986; 233(4761): 343-6.

16. Hirsch VM, Olmsted RA, Murphey-Corb M, Purcell RH, Johnson PR. An African primate lentivirus SIVsm closely related to HIV-2. Nature. 1989; 339: 389-92.

17. Worobey M, Gemmel M, Teuwen DE, Haselkorn T, Kunstman K, Bunce M, Muyembe JJ, Kabongo JMM, Kalengayi RM, Van Marck E. Direct evidence of extensive diversity of HIV-1 in Kinshasa by 1960. Nature. 2008; 455(7213): 661-4.

18. Zhu T, Korber BT, Nahmias AJ, Hooper E, Sharp PM, Ho DD. An African HIV-1 sequence from 1959 and implications for the origin of the epidemic. Nature. 1998; 391(6667): 594-7.

19. Wertheim JO, Worobey M. Dating the age of the SIV lineages that gave rise to HIV-1 and HIV-2. PLoS Comput Biol. 2009; 5(5): e1000377.

20. Korber B, Muldoon M, Theiler J, Gao F, Gupta R, Lapedes A, Hahn BH, Wolinsky S, Bhattacharya T. Timing the ancestor of the HIV-1 pandemic strains. Science. 2000; 288(5472): 1789-96.

21. Roberts J, Bebenek K, Kunkel T. The accuracy of reverse transcriptase from HIV-1. Science. 1988; 242(4882): 1171-3.

22. Perelson AS, Neumann AU, Markowitz M, Leonard JM, Ho DD. HIV-1 Dynamics in Vivo: Virion Clearance Rate, Infected Cell Life-Span, and Viral Generation Time. Science. 1996; 271(5255): 1582-6.

23. Korber B, Gaschen B, Yusim K, Thakallapally R, Kesmir C, Detours V. Evolutionary and immunological implications of contemporary HIV-1 variation. Br Med Bull. 2001; 58: 19-42.

24. Pneumocystis pneumonia--Los Angeles. MMWR Morb Mortal Wkly Rep. 1981; 30(21): 250-2.

25. Kaposi's sarcoma and Pneumocystis pneumonia among homosexual men--New York City and California. MMWR Morb Mortal Wkly Rep. 1981; 30(25): 305-8.

26. Siegal FP, Lopez C, Hammer GS, Brown AE, Kornfeld SJ, Gold J, Hassett J, Hirschman SZ, Cunningham-Rundles C, Adelsberg BR, et al. Severe acquired immunodeficiency in male homosexuals, manifested by chronic perianal ulcerative herpes simplex lesions. N Engl J Med. 1981; 305(24): 1439-44.

27. A cluster of Kaposi's sarcoma and Pneumocystis carinii pneumonia among homosexual male residents of Los Angeles and Orange Counties, California. MMWR Morb Mortal Wkly Rep. 1982; 31(23): 305- 7.

28. Barré-Sinoussi F, Chermann JC, Rey F, Nugeyre MT, Chamaret S, Gruest J, Dauguet C, Axler-Blin C, Vézinet-Brun F, Rouzioux C. Isolation of a T-lymphotropic retrovirus from a patient at risk for acquired immune deficiency syndrome (AIDS). Science. 1983; 220(4599): 868-71.

29. Gallo RC, Salahuddin SZ, Popovic M, Shearer GM, Kaplan M, Haynes BF, Palker TJ, Redfield R, Oleske J, Safai B. Frequent detection and isolation of cytopathic retroviruses (HTLV-III) from patients with AIDS and at risk for AIDS. Science. 1984; 224(4648): 500-3.

30. Curran JW, Jaffe HW. AIDS: the early years and CDC's response. MMWR Surveill Summ. 2011; 60 Suppl 4: 64-9.

308

31. Piot P, Colebunders R. Clinical manifestations and the natural history of HIV infection in adults. West J Med. 1987; 147(6): 709-12.

32. Piot P, Bartos M, Larson H, Zewdie D, Mane P. Coming to terms with complexity: a call to action for HIV prevention. Lancet. 2008; 372(9641): 845-59.

33. Dunkle KL, Stephenson R, Karita E, Chomba E, Kayitenkore K, Vwalika C, Greenberg L, Allen S. New heterosexually transmitted HIV infections in married or cohabiting couples in urban Zambia and Rwanda: an analysis of survey and clinical data. Lancet. 2008; 371(9631): 2183-91.

34. Guthrie BL, de Bruyn G, Farquhar C. HIV-1-discordant couples in sub-Saharan Africa: explanations and implications for high rates of discordancy. Curr HIV Res. 2007; 5(4): 416-29.

35. Hira SK, Nkowane BM, Kamanga J, Wadhawan D, Kavindele D, Macuacua R, Mpoko G, Malek M, Cruess DF, Perine PL. Epidemiology of human immunodeficiency virus in families in Lusaka, Zambia. J Acquir Immune Defic Syndr. 1990; 3(1): 83-6.

36. Institut National de la Statistique du Rwanda (INSR) ORC Macro. Rwanda Demographic and Health Survey 2005. 2006 [cited; Available from: http://www.measuredhs.com/

37. CSO., ORC., Macro. Zimbabwe Demographic and Health Survey 2005–06. 2007 [cited; Available from: www.measuredhs.com

38. Halperin DT, Epstein H. Concurrent sexual partnerships help to explain Africa's high HIV prevalence: implications for prevention. Lancet. 2004; 364(9428): 4-6.

39. Morris M, Kretzschmar M. Concurrent partnerships and the spread of HIV. AIDS. 1997; 11(5): 641- 8.

40. Mah TL, Halperin DT. Concurrent sexual partnerships and the HIV epidemics in Africa: evidence to move forward. AIDS Behav. 2010; 14(1): 11-6.

41. Morris M. A comparative study of concurrent sexual partnerships in the United States, Thailand and Uganda. American Sociology Association Annual Meeting; 2002; Anahien, California; 2002.

42. Tanser F, Barnighausen T, Hund L, Garnett GP, McGrath N, Newell ML. Effect of concurrent sexual partnerships on rate of new HIV infections in a high-prevalence, rural South African population: a cohort study. Lancet. 2011; 378(9787): 247-55.

43. Kreiss JK, Koech D, Plummer FA, Holmes KK, Lightfoote M, Piot P, Ronald AR, Ndinya-Achola JO, D'Costa LJ, Roberts P, et al. AIDS virus infection in Nairobi prostitutes. Spread of the epidemic to East Africa. N Engl J Med. 1986; 314(7): 414-8.

44. Van de Perre P, Clumeck N, Carael M, Nzabihimana E, Robert-Guroff M, De Mol P, Freyens P, Butzler JP, Gallo RC, Kanyamupira JB. Female prostitutes: a risk group for infection with human T-cell lymphotropic virus type III. Lancet. 1985; 2(8454): 524-7.

45. Corbett EL, Charalambous S, Moloi VM, Fielding K, Grant AD, Dye C, De Cock KM, Hayes RJ, Williams BG, Churchyard GJ. Human immunodeficiency virus and the prevalence of undiagnosed tuberculosis in African gold miners. Am J Respir Crit Care Med. 2004; 170(6): 673-9.

309

46. Bwayo J, Plummer F, Omari M, Mutere A, Moses S, Ndinya-Achola J, Velentgas P, Kreiss J. Human immunodeficiency virus infection in long-distance truck drivers in east Africa. Arch Intern Med. 1994; 154(12): 1391-6.

47. United Nations General Assembly Human Rights Council. Discriminatory laws and practices and acts of violence against individuals based on their sexual orientation and gender identity. Report of the United Nations High Commissioner for Human Rights. 2011 [cited; Available from: http://www2.ohchr.org/english/bodies/hrcouncil/docs/19session/A.HRC.19.41_English.pdf

48. Smith AD, Tapsoba P, Peshu N, Sanders EJ, Jaffe HW. Men who have sex with men and HIV/AIDS in sub-Saharan Africa. Lancet. 2009; 374(9687): 416-22.

49. Simon V, Ho DD, Abdool Karim Q. HIV/AIDS epidemiology, pathogenesis, prevention, and treatment. Lancet. 2006; 368(9534): 489-504.

50. Mepham SO, Bland RM, Newell ML. Prevention of mother-to-child transmission of HIV in resource- rich and -poor settings. BJOG. 2011; 118(2): 202-18.

51. Coutsoudis A, Kwaan L, Thomson M. Prevention of vertical transmission of HIV-1 in resource- limited settings. Expert Rev Anti Infect Ther. 2010; 8(10): 1163-75.

52. Kourtis AP, Lee FK, Abrams EJ, Jamieson DJ, Bulterys M. Mother-to-child transmission of HIV-1: timing and implications for prevention. The Lancet Infectious Diseases. 2006; 6(11): 726-32.

53. Gray RH, Wawer MJ, Brookmeyer R, Sewankambo NK, Serwadda D, Wabwire-Mangen F, Lutalo T, Li X, VanCott T, Quinn TC. Probability of HIV-1 transmission per coital act in monogamous, heterosexual, HIV-1-discordant couples in Rakai, Uganda. Lancet. 2001; 357(9263): 1149-53.

54. Quinn TC, Wawer MJ, Sewankambo N, Serwadda D, Li C, Wabwire-Mangen F, Meehan MO, Lutalo T, Gray RH. Viral Load and Heterosexual Transmission of Human Immunodeficiency Virus Type 1. N Engl J Med. 2000; 342(13): 921-9.

55. Royce RA, Sena A, Cates W, Jr., Cohen MS. Sexual transmission of HIV. N Engl J Med. 1997; 336(15): 1072-8.

56. John GC, Kreiss J. Mother-to-child transmission of human immunodeficiency virus type 1. Epidemiol Rev. 1996; 18(2): 149-57.

57. Siegfried N, van der Merwe L, Brocklehurst P, Sint TT. Antiretrovirals for reducing the risk of mother-to-child transmission of HIV infection. Cochrane Database of Systematic Reviews. 2011; (7).

58. Baggaley RF, White RG, Boily MC. HIV transmission risk through anal intercourse: systematic review, meta-analysis and implications for HIV prevention. Int J Epidemiol. 2010; 39(4): 1048-63.

59. Kalichman SC, Di Berto G, Eaton L. Human immunodeficiency virus viral load in blood plasma and semen: review and implications of empirical findings. Sex Transm Dis. 2008; 35(1): 55-60.

60. Cohen MS, Chen YQ, McCauley M, Gamble T, Hosseinipour MC, Kumarasamy N, Hakim JG, Kumwenda J, Grinsztejn B, Pilotto JH, Godbole SV, Mehendale S, Chariyalertsak S, Santos BR, Mayer KH, Hoffman IF, Eshleman SH, Piwowar-Manning E, Wang L, Makhema J, Mills LA, de Bruyn G, Sanne I, Eron J, Gallant J, Havlir D, Swindells S, Ribaudo H, Elharrar V, Burns D, Taha TE, Nielsen-Saines K, Celentano D, Essex M, Fleming TR. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med. 2011; 365(6): 493-505.

310

61. Castel AD, Befus M, Willis S, Griffin A, West T, Hader S, Greenberg AE. Use of the community viral load as a population-based biomarker of HIV burden. AIDS. 2012; 26(3): 345-53.

62. Das M, Chu PL, Santos GM, Scheer S, Vittinghoff E, McFarland W, Colfax GN. Decreases in community viral load are accompanied by reductions in new HIV infections in San Francisco. PLoS One. 2010; 5(6): e11068.

63. Estill J, Aubriere C, Egger M, Johnson L, Wood R, Garone D, Gsponer T, Wandeler G, Boulle A, Davies MA, Hallett T, Keiser O. Viral load monitoring of antiretroviral therapy, cohort viral load and HIV transmission in Southern Africa: A mathematical modelling analysis. AIDS. 2012; 26(11): 1403-13.

64. Grant RM, Lama JR, Anderson PL, McMahan V, Liu AY, Vargas L, Goicochea P, Casapia M, Guanira- Carranza JV, Ramirez-Cardich ME, Montoya-Herrera O, Fernandez T, Veloso VG, Buchbinder SP, Chariyalertsak S, Schechter M, Bekker LG, Mayer KH, Kallas EG, Amico KR, Mulligan K, Bushman LR, Hance RJ, Ganoza C, Defechereux P, Postle B, Wang F, McConnell JJ, Zheng JH, Lee J, Rooney JF, Jaffe HS, Martinez AI, Burns DN, Glidden DV. Preexposure chemoprophylaxis for HIV prevention in men who have sex with men. N Engl J Med. 2010; 363(27): 2587-99.

65. Baeten J, Donnell D, Ndase P, Mugo N, Mujugira A, Celum C, Team aPPS. ARV PrEP for HIV-1 Prevention among Heterosexual Men and Women, Abstract 29. 19th Conference on Retroviruses and Opportunistic Infections. Seattle, USA; 2012.

66. Thigpen C, Kebaabetswe PM, Smith DK, Segolodi TM, Soud FA, Chillag K, Chirwa KI, Kasonde K, Mutanhaurwa R, Henderson FL, Pathak S, R G, Rose CE, LA P, Group ftTS. Daily oral antiretroviral use for the prevention of HIV infection in heterosexually active young adults in Botswana: results from the TDF2 study. 6th IAS Conference on HIV pathogenesis, treatment and prevention. Rome, Italy; 2011.

67. Van Damme L, Corneli A, Ahmed A, Agot K, Lombaard J, Kapiga S, Grant R, Kashuba R, Crucitti T, Taylor D, Group aF-PS. The FEM-PrEP Trial of Emtricitabine/Tenofovir Disoproxil Fumarate (Truvada) among African Women. 19th Conference on Retroviruses and Opportunistic Infections. Seattle, USA; 2012.

68. Abdool Karim Q, Abdool Karim SS, Frohlich JA, Grobler AC, Baxter C, Mansoor LE, Kharsany ABM, Sibeko S, Mlisana KP, Omar Z, Gengiah TN, Maarschalk S, Arulappan N, Mlotshwa M, Morris L, Taylor D, CAPRISA 004 Trial Group. Effectiveness and safety of tenofovir gel, an antiretroviral microbicide, for the prevention of HIV infection in women. Science. 2010; 329(5996): 1168-74.

69. McCoombe SG, Short RV. Potential HIV-1 target cells in the human penis. AIDS. 2006; 20(11): 1491.

70. Weiss HA, Quigley MA, Hayes RJ. Male circumcision and risk of HIV infection in sub-Saharan Africa: a systematic review and meta-analysis. AIDS. 2000; 14(15): 2361-70.

71. Auvert B, Taljaard D, Lagarde E, Sobngwi-Tambekou J, Sitta R, Puren A. Randomized, controlled intervention trial of male circumcision for reduction of HIV infection risk: the ANRS 1265 Trial. PLoS Med. 2005; 2(11): e298.

72. Bailey RC, Moses S, Parker CB, Agot K, Maclean I, Krieger JN, Williams CFM, Campbell RT, Ndinya- Achola JO. Male circumcision for HIV prevention in young men in Kisumu, Kenya: a randomised controlled trial. Lancet. 2007; 369(9562): 643-56.

73. Gray RH, Kigozi G, Serwadda D, Makumbi F, Watya S, Nalugoda F, Kiwanuka N, Moulton LH, Chaudhary MA, Chen MZ, Sewankambo NK, Wabwire-Mangen F, Bacon MC, Williams CFM, Opendi P, Reynolds SJ, Laeyendecker O, Quinn TC, Wawer MJ. Male circumcision for HIV prevention in men in Rakai, Uganda: a randomised trial. Lancet. 2007; 369(9562): 657-66.

311

74. World Health Organization. Use of devices for adult male circumcision in public health HIV prevention programmes. 2012 [cited; Available from: http://whqlibdoc.who.int/hq/2012/WHO_HIV_2012.4_eng.pdf

75. Coombs RW, Reichelderfer PS, Landay AL. Recent observations on HIV type-1 infection in the genital tract of men and women. AIDS. 2003; 17(4): 455-80.

76. Galvin SR, Cohen MS. The role of sexually transmitted diseases in HIV transmission. Nat Rev Microbiol. 2004; 2(1): 33-42.

77. Grosskurth H, Mosha F, Todd J, Mwijarubi E, Klokke A, Senkoro K, Mayaud P, Changalucha J, Nicoll A, ka-Gina G, et al. Impact of improved treatment of sexually transmitted diseases on HIV infection in rural Tanzania: randomised controlled trial. Lancet. 1995; 346(8974): 530-6.

78. Ng BE, Butler LM, Horvath T, Rutherford GW. Population-based biomedical sexually transmitted infection control interventions for reducing HIV infection. Cochrane Database Syst Rev. 2011; (3): CD001220.

79. Wawer MJ, Sewankambo NK, Serwadda D, Quinn TC, Paxton LA, Kiwanuka N, Wabwire-Mangen F, Li C, Lutalo T, Nalugoda F, Gaydos CA, Moulton LH, Meehan MO, Ahmed S, Gray RH. Control of sexually transmitted diseases for AIDS prevention in Uganda: a randomised community trial. Rakai Project Study Group. Lancet. 1999; 353(9152): 525-35.

80. Kamali A, Quigley M, Nakiyingi J, Kinsman J, Kengeya-Kayondo J, Gopal R, Ojwiya A, Hughes P, Carpenter LM, Whitworth J. Syndromic management of sexually-transmitted infections and behaviour change interventions on transmission of HIV-1 in rural Uganda: a community randomised trial. Lancet. 2003; 361(9358): 645-52.

81. Gregson S, Adamson S, Papaya S, Mundondo J, Nyamukapa CA, Mason PR, Garnett GP, Chandiwana SK, Foster G, Anderson RM. Impact and process evaluation of integrated community and clinic- based HIV-1 control: a cluster-randomised trial in eastern Zimbabwe. PLoS Med. 2007; 4(3): e102.

82. Batar I, Sivin I. State-of-the-art of non-hormonal methods of contraception: I. Mechanical barrier contraception. Eur J Contracept Reprod Health Care. 2010; 15(2): 67-88.

83. Weller S, Davis-Betty K. Condom Effectiveness in reducing hetrosexual HIV transmission. Cochrane Database Syst Rev. 2009; 1(CD003255).

84. Gallo MF, Kilbourne-Brook M, Coffey PS. A review of the effectiveness and acceptability of the female condom for dual protection. Sexual Health. 2012; 9(1): 18-26.

85. Parker RG, Easton D, Klein CH. Structural barriers and facilitators in HIV prevention: a review of international research. AIDS. 2000; 14: S22.

86. Jewkes R, Dunkle K, Nduna M, Levin J, Jama N, Khuzwayo N, Koss M, Puren A, Duvvury N. Factors associated with HIV sero-status in young rural South African women: connections between intimate partner violence and HIV. Int J Epidemiol. 2006; 35(6): 1461-8.

87. Miller CJ, Shattock RJ. Target cells in vaginal HIV transmission. Microbes and Infection. 2003; 5(1): 59-67.

88. Jewkes RK, Dunkle K, Nduna M, Shai N. Intimate partner violence, relationship power inequity, and incidence of HIV infection in young women in South Africa: a cohort study. Lancet. 2010; 376(9734): 41-8.

312

89. Kim JC, Martin LJ, Denny L. Rape and HIV post-exposure prophylaxis: addressing the dual epidemics in South Africa. Reprod Health Matters. 2003; 11(22): 101-12.

90. Jewkes R, Nduna M, Levin J, Jama N, Dunkle K, Puren A, Duvvury N. Impact of stepping stones on incidence of HIV and HSV-2 and sexual behaviour in rural South Africa: cluster randomised controlled trial. BMJ. 2008; 337: a506.

91. Pronyk PM, Hargreaves JR, Kim JC, Morison LA, Phetla G, Watts C, Busza J, Porter JD. Effect of a structural intervention for the prevention of intimate-partner violence and HIV in rural South Africa: a cluster randomised trial. Lancet. 2006; 368(9551): 1973-83.

92. Weeks BS, Alcamo IE. AIDS : the biological basis. Sudbury, Mass.: Jones and Bartlett Publishers; 2010.

93. Wilen CB, Tilton JC, Doms RW. Molecular mechanisms of HIV entry. Adv Exp Med Biol. 2012; 726: 223-42.

94. Cohen MS, Gay CL, Busch MP, Hecht FM. The detection of acute HIV infection. J Infect Dis. 2010; 202 Suppl 2: S270-7.

95. Fiebig EW, Wright DJ, Rawal BD, Garrett PE, Schumacher RT, Peddada L, Heldebrant C, Smith R, Conrad A, Kleinman SH. Dynamics of HIV viremia and antibody seroconversion in plasma donors: implications for diagnosis and staging of primary HIV infection. AIDS. 2003; 17(13): 1871.

96. Stacey AR, Norris PJ, Qin L, Haygreen EA, Taylor E, Heitman J, Lebedeva M, DeCamp A, Li D, Grove D. Induction of a striking systemic cytokine cascade prior to peak viremia in acute human immunodeficiency virus type 1 infection, in contrast to more modest and delayed responses in acute hepatitis B and C virus infections. J Virol. 2009; 83(8): 3719-33.

97. Pantaleo G, Graziosi C, Fauci AS. The Immunopathogenesis of Human Immunodeficiency Virus Infection. N Engl J Med. 1993; 328(5): 327-35.

98. Mellors JW, Rinaldo Jr CR, Gupta P, White RM, Todd JA, Kingsley LA. Prognosis in HIV-1 infection predicted by the quantity of virus in plasma. Science. 1996; 272(5265): 1167-70.

99. de Wolf F, Spijkerman I, Schellekens PT, Langendam M, Kuiken C, Bakker M, Roos M, Coutinho R, Miedema F, Goudsmit J. AIDS prognosis based on HIV-1 RNA, CD4+ T-cell count and function: markers with reciprocal predictive value over time after seroconversion. AIDS. 1997; 11(15): 1799.

100. Fraser C, Hollingsworth TD, Chapman R, De Wolf F, Hanage WP. Variation in HIV-1 set-point viral load: epidemiological analysis and an evolutionary hypothesis. Proc Natl Acad Sci U S A. 2007; 104(44): 17441.

101. Pantaleo G, Graziosi C, Demarest JF, Butini L, Montroni M, Fox CH, Orenstein JM, Kotler DP, Fauci AS. HIV infection is active and progressive in lymphoid tissue during the clinically latent stage of disease. Nature. 1993; 362(6418): 355-8.

102. Sousa AE, Carneiro J, Meier-Schellersheim M, Grossman Z, Victorino RMM. CD4 T cell depletion is linked directly to immune activation in the pathogenesis of HIV-1 and HIV-2 but only indirectly to the viral load. J Immunol. 2002; 169(6): 3400-6.

103. Morgan D, Mahe C, Mayanja B, Whitworth JAG. Progression to symptomatic disease in people infected with HIV-1 in rural Uganda: prospective cohort study. BMJ. 2002; 324(7331): 193-6.

313

104. Holmes CB, Wood R, Badri M, Zilber S, Wang B, Maartens G, Zheng H, Lu Z, Freedberg KA, Losina E. CD4 Decline and Incidence of Opportunistic Infections in Cape Town, South Africa: Implications for Prophylaxis and Treatment. J Acquir Immune Defic Syndr. 2006; 42(4): 464-9.

105. Lang W, Perkins H, Anderson RE, Royce R, Jewell N, Winkelstein W, Jr. Patterns of T lymphocyte changes with human immunodeficiency virus infection: from seroconversion to the development of AIDS. J Acquir Immune Defic Syndr. 1989; 2(1): 63-9.

106. Grimwade K, Swingler. Cotrimoxazole prophylaxis for opportunistic infections in adults with HIV. Cochrane Database Syst Rev. 2003; (3): CD003108.

107. Suthar AB, Granich R, Mermin J, Van Rie A. Effect of cotrimoxazole on mortality in HIV-infected adults on antiretroviral therapy: a systematic review and meta-analysis. Bull World Health Organ. 2012; 90(2): 128C-38C.

108. Akolo C, Adetifa I, Shepperd S, Volmink J. Treatment of latent tuberculosis infection in HIV infected persons. Cochrane Database Syst Rev. 2010; (1): CD000171.

109. World Health Organization. Guidelines for intensified tuberculosis case-finding and isoniazid preventative therapy for people living with HIV in resource-constrained settings. Geneva; 2011.

110. World Health Organization. Guidelines on co-trimoxazole prophylaxis for HIV-related infections among children, adolescents and adults. 2006 [cited; Available from: http://www.who.int/hiv/pub/plhiv/ctx/en/index.html

111. Jaffar S, Grant AD, Whitworth J, Smith PG, Whittle H. The natural history of HIV-1 and HIV-2 infections in adults in Africa: a literature review. Bull World Health Organ. 2004; 82(6): 462-9.

112. Morgan D, Mahe C, Mayanja B, Okongo JM, Lubega R, Whitworth JAG. HIV-1 infection in rural Africa: is there a difference in median time to AIDS and survival compared with that in industrialized countries? AIDS. 2002; 16(4): 597.

113. National Statistics Office. Malawi Demographic and Health Survey. 2010 [cited; Available from: http://www.measuredhs.com/pubs/pdf/PR4/PR4.pdf

114. Malawi Government. Global AIDS Response Progress Report: Malawi Country Report for 2010 and 2011. 2012 [cited; Available from: http://www.unaids.org/en/dataanalysis/knowyourresponse/countryprogressreports/2012countries/ce_M W_Narrative_Report[1].pdf

115. World Health Organization. Service delivery approaches to HIV testing and counselling (HTC): a strategic HTC programme framework. Geneva; 2012.

116. Branson BM. The future of HIV testing. J Acquir Immune Defic Syndr. 2010; 55 Suppl 2: S102-5.

117. US Food and Drug Administration. Complete List of Donor Screening Assays for Infectious Agents and HIV Diagnostic Assays. Silver Spring, Maryland; 2013.

118. UNAIDS. Joint Action for Results: UNAIDS Outcome Framework. 2010 [cited; Available from: https://www.unaids.org/en/media/unaids/contentassets/documents/aboutunaids/programmecoordinating board/subcommittee/jc1713_joint_action_en.pdf

314

119. World Health Organization. Guidance on Provider-Initiated HIV Testing and Counselling in Health Facilities. 2007 [cited; Available from: http://www.who.int/hiv/pub/vct/pitc2007/en/

120. World Health Organization. Towards Univeral Access. Scaling up priority HIV/AIDS interventions in the health sector. Geneva; 2010.

121. Powers KA, Ghani AC, Miller WC, Hoffman IF, Pettifor AE, Kamanga G, Martinson FE, Cohen MS. The role of acute and early HIV infection in the spread of HIV and implications for transmission prevention strategies in Lilongwe, Malawi: a modelling study. Lancet. 2011; 378(9787): 256-68.

122. Cohen T, Corbett EL. Test and treat in HIV: success could depend on rapid detection. Lancet. 2011; 378(9787): 204-6.

123. De Zoysa I, Phillips KA, Kamenga MC, O'Reilly KR, Sweat MD, White RA, Grinstead OA, Coates TJ. Role of HIV counseling and testing in changing risk behavior in developing countries. AIDS. 1995; 9 Suppl A: S95-101.

124. VCT Efficacy Study Group. Efficacy of voluntary HIV-1 counselling and testing in individuals and couples in Kenya, Tanzania, and Trinidad: a randomised trial. The Voluntary HIV-1 Counseling and Testing Efficacy Study Group. Lancet. 2000; 356(9224): 103-12.

125. Marum E, Taegtmeyer M, Chebet K. Scale-up of voluntary HIV counseling and testing in Kenya. JAMA. 2006; 296(7): 859-62.

126. Denison JA, O'Reilly KR, Schmid GP, Kennedy CE, Sweat MD. HIV voluntary counseling and testing and behavioral risk reduction in developing countries: a meta-analysis, 1990--2005. AIDS Behav. 2008; 12(3): 363-73.

127. Corbett EL, Dauya E, Matambo R, Cheung YB, Makamure B, Bassett MT, Chandiwana S, Munyati S, Mason PR, Butterworth AE. Uptake of workplace HIV counselling and testing: a cluster-randomised trial in Zimbabwe. PLoS Med. 2006; 3(7): e238.

128. Fylkesnes K, Siziya S. A randomized trial on acceptability of voluntary HIV counselling and testing. Trop Med Int Health. 2004; 9(5): 566-72.

129. Rosenberg NE, Westreich D, Barnighausen T, Miller WC, Behets F, Maman S, Newell ML, Pettifor A. Assessing the effect of HIV counselling and testing on HIV acquisition among South African youth. AIDS. 2013.

130. Coates T, Eshleman SH, Chariyalertsak S, Chingono A, Gray G, Mbwambo J. Community-level reductions in estimated hiv incidence: HIV Prevention Trials Network 043, Project Accept. 20th Conference on Retroviruses and Opportunistic Infections. Atlanta, USA; 2013.

131. Fonner VA, Denison J, Kennedy CE, O'Reilly K, Sweat M. Voluntary counseling and testing (VCT) for changing HIV-related risk behavior in developing countries. Cochrane Database Syst Rev. 2012; 9: CD001224.

132. Killewo JZ, Kwesigabo G, Comoro C, Lugalla J, Mhalu FS, Biberfeld G, Wall S, Sandstrom A. Acceptability of voluntary HIV testing with counselling in a rural village in Kagera, Tanzania. AIDS Care. 1998; 10(4): 431-9.

133. Sahlu T, Kassa E, Agonafer T, Tsegaye A, Rinke de Wit T, Gebremariam H, Doorly R, Spijkerman I, Yeneneh H, Coutinho RA, Fontanet AL. Sexual behaviours, perception of risk of HIV infection, and factors associated with attending HIV post-test counselling in Ethiopia. AIDS. 1999; 13(10): 1263-72.

315

134. Obermeyer CM, Osborn M. The utilization of testing and counseling for HIV: a review of the social and behavioral evidence. Am J Public Health. 2007; 97(10): 1762-74.

135. Malonza IM, Richardson BA, Kreiss JK, Bwayo JJ, Stewart GC. The effect of rapid HIV-1 testing on uptake of perinatal HIV-1 interventions: a randomized clinical trial. AIDS. 2003; 17(1): 113-8.

136. Corbett EL, Dauya E, Matambo R, Cheung YB, Makamure B, Bassett MT, Chandiwana S, Munyati S, Mason PR, Butterworth AE, Godfrey-Faussett P, Hayes RJ. Uptake of workplace HIV counselling and testing: a cluster-randomised trial in Zimbabwe. PLoS Med. 2006; 3(7): e238.

137. Liddicoat RV, Horton NJ, Urban R, Maier E, Christiansen D, Samet JH. Assessing missed opportunities for HIV testing in medical settings. J Gen Intern Med. 2004; 19(4): 349-56.

138. Weintrob AC, Giner J, Menezes P, Patrick E, Benjamin DK, Jr., Lennox J, Pilcher CD, Eron JJ, Hicks CB. Infrequent diagnosis of primary human immunodeficiency virus infection: missed opportunities in acute care settings. Arch Intern Med. 2003; 163(17): 2097-100.

139. De Cock KM, Johnson AM. From exceptionalism to normalisation: a reappraisal of attitudes and practice around HIV testing. BMJ. 1998; 316(7127): 290-3.

140. Wanyenze R, Kamya M, Liechty CA, Ronald A, Guzman DJ, Wabwire-Mangen F, Mayanja-Kizza H, Bangsberg DR. HIV counseling and testing practices at an urban hospital in Kampala, Uganda. AIDS Behav. 2006; 10(4): 361-7.

141. Topp S, Chipukuma J, Chiko M, Wamulume C, Bolton-Moore C, Reid S. Opt-out provider-initiated HIV testing and counselling in primary care outpatient clinics in Zambia. Bull World Health Organ. 2011; 89: 328-225.

142. Silvestri DM, Modjarrad K, Blevins ML, Halale E, Vermund SH, McKinzie JP. A comparison of HIV detection rates using routine opt-out provider-initiated HIV testing and counseling versus a standard of care approach in a rural African setting. J Acquir Immune Defic Syndr. 2011; 56(1): e9-32.

143. Kiene SM, Bateganya M, Wanyenze R, Lule H, Nantaba H, Stein MD. Initial outcomes of provider- initiated routine HIV testing and counseling during outpatient care at a rural Ugandan hospital: risky sexual behavior, partner HIV testing, disclosure, and HIV care seeking. AIDS Patient Care STDS. 2010; 24(2): 117-26.

144. Deribew A, Negussu N, Kassahun W, Apers L, Colebunders R. Uptake of provider-initiated counselling and testing among tuberculosis suspects, Ethiopia. Int J Tuberc Lung Dis. 2010; 14(11): 1442-6.

145. Steen TW, Seipone K, Gomez Fde L, Anderson MG, Kejelepula M, Keapoletswe K, Moffat HJ. Two and a half years of routine HIV testing in Botswana. J Acquir Immune Defic Syndr. 2007; 44(4): 484-8.

146. Bassett IV, Giddy J, Nkera J, Wang B, Losina E, Lu Z, Freedberg KA, Walensky RP. Routine voluntary HIV testing in Durban, South Africa: the experience from an outpatient department. J Acquir Immune Defic Syndr. 2007; 46(2): 181-6.

147. Lawn SD, Fraenzel A, Kranzer K, Caldwell J, Bekker L-G, Wood R. Provider-initiated HIV testing increases access of patients with HIV-associated tuberculosis to antiretroviral treatment. S Afr Med J. 2011; 101(4): 258-62.

148. Wallrauch C, Heller T, Lessells R, Kekana M, Barnighausen T, Newell M-L. High uptake of HIV testing for tuberculosis patients in an integrated primary health care HIV/TB programme in rural KwaZulu- Natal. S Afr Med J. 2010; 100(3): 146-7.

316

149. Pope DS, Deluca AN, Kali P, Hausler H, Sheard C, Hoosain E, Chaudhary MA, Celentano DD, Chaisson RE. A cluster-randomized trial of provider-initiated (opt-out) HIV counseling and testing of tuberculosis patients in South Africa. J Acquir Immune Defic Syndr. 2008; 48(2): 190-5.

150. Kharsany ABM, Karim QA, Abdool Karim SS. Uptake of provider-initiated HIV testing and counseling among women attending an urban sexually transmitted disease clinic in South Africa - missed opportunities for early diagnosis of HIV infection. AIDS Care. 2010: 1-5.

151. Chandisarewa W, Stranix-Chibanda L, Chirapa E, Miller A, Simoyi M, Mahomva A, Maldonado Y, Shetty AK. Routine offer of antenatal HIV testing ("opt-out" approach) to prevent mother-to-child transmission of HIV in urban Zimbabwe. Bull World Health Organ. 2007; 85(11): 843-50.

152. Perez F, Zvandaziva C, Engelsmann B, Dabis F. Acceptability of routine HIV testing ("opt-out") in antenatal services in two rural districts of Zimbabwe. J Acquir Immune Defic Syndr. 2006; 41(4): 514-20.

153. Roura M, Watson-Jones D, Kahawita TM, Ferguson L, Ross DA. Provider-initiated testing and counselling programmes in sub-Saharan Africa: a systematic review of their operational implementation. AIDS. 2013; 27(4): 617-26.

154. Hensen B, Baggaley R, Wong VJ, Grabbe KL, Shaffer N, Lo YR, Hargreaves J. Universal voluntary HIV testing in antenatal care settings: a review of the contribution of provider-initiated testing & counselling. Trop Med Int Health. 2012; 17(1): 59-70.

155. Wringe A, Roura M, Urassa M, Busza J, Athanas V, Zaba B. Doubts, denial and divine intervention: understanding delayed attendance and poor retention rates at a HIV treatment programme in rural Tanzania. AIDS Care. 2009; 21(5): 632-7.

156. Farmer P, Leandre F, Mukherjee JS, Claude M, Nevil P, Smith-Fawzi MC, Koenig SP, Castro A, Becerra MC, Sachs J, Attaran A, Kim JY. Community-based approaches to HIV treatment in resource-poor settings. Lancet. 2001; 358(9279): 404-9.

157. Suthar AB, Ford N, Bachanas PJ, Wong VJ, Rajan JS, Saltzman AK, Ajose O, Fakoya AO, Granich RM, Negussie EK, Baggaley RC. Towards universal voluntary HIV testing and counselling: a systematic review and meta-analysis of community-based approaches. PLoS Med. 2013; 10(8): e1001496.

158. Molesworth AM, Ndhlovu R, Banda E, Saul J, Ngwira B, Glynn JR, Crampin AC, French N. High Accuracy of Home-Based Community Rapid HIV Testing in Rural Malawi. J Acquir Immune Defic Syndr. 2010; 55(5): 625-30.

159. Laeyendecker O, Piwowar-Manning E, Fiamma A, Kulich M, Donnell D, Bassuk D, Mullis CE, Chin C, Swanson P, Hackett J, Jr., Clarke W, Marzinke M, Szekeres G, Gray G, Richter L, Alexandre MW, Chariyalertsak S, Chingono A, Celentano DD, Morin SF, Sweat M, Coates T, Eshleman SH. Estimation of HIV incidence in a large, community-based, randomized clinical trial: NIMH project accept (HIV Prevention Trials Network 043). PLoS One. 2013; 8(7): e68349.

160. Cohen MS, Smith MK, Muessig KE, Hallett TB, Powers KA, Kashuba AD. Antiretroviral treatment of HIV-1 prevents transmission of HIV-1: where do we go from here? Lancet. 2013.

161. Helleringer S, Mkandawire J, Reniers G, Kalilani-Phiri L, Kohler HP. Should home-based HIV testing and counseling services be offered periodically in programs of ARV treatment as prevention? A case study in Likoma (Malawi). AIDS Behav. 2013; 17(6): 2100-8.

162. Menzies N, Abang B, Wanyenze R, Nuwaha F, Mugisha B, Coutinho A, Bunnell R, Mermin J, Blandford JM. The costs and effectiveness of four HIV counseling and testing strategies in Uganda. AIDS. 2009; 23(3): 395-401.

317

163. Bastian LA, Nanda K, Hasselblad V, Simel DL. Diagnostic efficiency of home pregnancy test kits. A meta-analysis. Arch Fam Med. 1998; 7(5): 465-9.

164. Jamil MS, Hocking JS, Bauer HM, Ali H, Wand H, Smith K, Walker J, Donovan B, Kaldor JM, Guy RJ. Home-based chlamydia and gonorrhoea screening: a systematic review of strategies and outcomes. BMC Public Health. 2013; 13: 189.

165. Nelson HD, Tyne K, Naik A, Bougatsos C, Chan B, Nygren P, Humphrey L. Screening for Breast Cancer: Systematic Evidence Review Update for the US Preventive Services Task Force. Rockville MD; 2009.

166. Langendam M, Luijf YM, Hooft L, Devries JH, Mudde AH, Scholten RJ. Continuous glucose monitoring systems for type 1 diabetes mellitus. Cochrane Database Syst Rev. 2012; 1: CD008101.

167. Uhlig K, Patel K, Ip S, Kitsios GD, Balk EM. Self-measured blood pressure monitoring in the management of hypertension: a systematic review and meta-analysis. Ann Intern Med. 2013; 159(3): 185- 94.

168. World Health Organization. Report on the first international symposium on self-testing for HIV. The legal, ethical, gender, human rights and public health implications of HIV self-testing scale up. Geneva; 2013.

169. World Health Organization. HIV self-testing among health workers: a review of the literature. 2011 [cited; Available from: http://whqlibdoc.who.int/publications/2011/9789241501033_eng.pdf

170. US Food and Drug Administration. OraQuick® In-Home HIV Test Summary of Safety and Effectiveness. Silver Spring, Maryland; 2013.

171. Choko AT, Desmond N, Webb EL, Chavula K, Napierala-Mavedzenge S, Gaydos CA, Makombe SD, Chunda T, Squire SB, French N, Mwapasa V, Corbett EL. The Uptake and Accuracy of Oral Kits for HIV Self- Testing in High HIV Prevalence Setting: A Cross-Sectional Feasibility Study in Blantyre, Malawi. PLoS Med. 2011; 8(10): e1001102.

172. Pant Pai N, Sharma J, Shivkumar S, Pillay S, Vadnais C, Joseph L, Dheda K, Peeling RW. Supervised and Unsupervised Self-Testing for HIV in High- and Low-Risk Populations: A Systematic Review. PLoS Med. 2013; 10(4): e1001414.

173. Kalibala S, Tun W, Muraah W, Cherutich P, Oweya E. ‘Knowing myself first’: feasibility of self- testing among health workers in Kenya. 2011 [cited; Available from: http://www.popcouncil.org/pdfs/2011HIV_KenyaHWSelfTesting.pdf

174. Ministry of Health. Clinical Management of HIV in Children and Adults. 2011 [cited; Available from: http://www.who.int/hiv/pub/guidelines/malawi_art.pdf

175. Ministry of Health. Integrated HIV programme report October to December 2012. 2012 [cited; Available from: http://www.hivunitmohmw.org/Main/AntiretroviralTherapy

176. Yarchoan R, Klecker RW, Weinhold KJ, Markham PD, Lyerly HK, Durack DT, Gelmann E, Lehrman SN, Blum RM, Barry DW, et al. Administration of 3'-azido-3'-deoxythymidine, an inhibitor of HTLV-III/LAV replication, to patients with AIDS or AIDS-related complex. Lancet. 1986; 1(8481): 575-80.

177. Broder S. The development of antiretroviral therapy and its impact on the HIV-1/AIDS pandemic. Antiviral Res. 2010; 85(1): 1-18.

318

178. Fischl MA, Richman DD, Grieco MH, Gottlieb MS, Volberding PA, Laskin OL, Leedom JM, Groopman JE, Mildvan D, Schooley RT, et al. The efficacy of azidothymidine (AZT) in the treatment of patients with AIDS and AIDS-related complex. A double-blind, placebo-controlled trial. N Engl J Med. 1987; 317(4): 185-91.

179. Parks WP, Parks ES, Fischl MA, Leuther MD, Allain JP, Nusinoff-Lehrman S, Barry DW, Makuch RW. HIV-1 inhibition by azidothymidine in a concurrently randomized placebo-controlled trail. J Acquir Immune Defic Syndr. 1988; 1(2): 125-30.

180. Gulick RM, Mellors JW, Havlir D, Eron JJ, Gonzalez C, McMahon D, Richman DD, Valentine FT, Jonas L, Meibohm A, Emini EA, Chodakewitz JA. Treatment with indinavir, zidovudine, and lamivudine in adults with human immunodeficiency virus infection and prior antiretroviral therapy. N Engl J Med. 1997; 337(11): 734-9.

181. Hammer SM, Katzenstein DA, Hughes MD, Gundacker H, Schooley RT, Haubrich RH, Henry WK, Lederman MM, Phair JP, Niu M, Hirsch MS, Merigan TC. A trial comparing nucleoside monotherapy with combination therapy in HIV-infected adults with CD4 cell counts from 200 to 500 per cubic millimeter. AIDS Clinical Trials Group Study 175 Study Team. N Engl J Med. 1996; 335(15): 1081-90.

182. Antiretroviral TCC. Life expectancy of individuals on combination antiretroviral therapy in high- income countries: a collaborative analysis of 14 cohort studies. Lancet. 2008; 372(9635): 293.

183. May M, Sterne JA, Sabin C, Costagliola D, Justice AC, Thiebaut R, Gill J, Phillips A, Reiss P, Hogg R, Ledergerber B, D'Arminio Monforte A, Schmeisser N, Staszewski S, Egger M, Antiretroviral Therapy Cohort C. Prognosis of HIV-1-infected patients up to 5 years after initiation of HAART: collaborative analysis of prospective studies. AIDS. 2007; 21(9): 1185-97.

184. Sterne JAC, Hernán MA, Ledergerber B, Tilling K, Weber R, Sendi P, Rickenbach M, Robins JM, Egger M, Swiss HIV Cohort Study. Long-term effectiveness of potent antiretroviral therapy in preventing AIDS and death: a prospective cohort study. Lancet. 2005; 366(9483): 378-84.

185. Tsevat J, Leonard AC, Szaflarski M, Sherman SN, Cotton S, Mrus JM, Feinberg J. Change in quality of life after being diagnosed with HIV: a multicenter longitudinal study. AIDS Patient Care STDS. 2009; 23(11): 931-7.

186. Kaufmann GR, Perrin L, Pantaleo G, Opravil M, Furrer H, Telenti A, Hirschel B, Ledergerber B, Vernazza P, Bernasconi E. CD4 T-lymphocyte recovery in individuals with advanced HIV-1 infection receiving potent antiretroviral therapy for 4 years: the Swiss HIV Cohort Study. Arch Intern Med. 2003; 163(18): 2187.

187. Centers for Disease Control and Prevention. Mortality slide series. 2012 [cited; Available from: http://www.cdc.gov/hiv/topics/surveillance/resources/slides/mortality/slides/mortality.pdf

188. World Health Organization Regional Office for Europe. HIV/AIDS in Europe: Moving from a death sentence to chronic disease management. 2006 [cited; Available from: http://www.euro.who.int/__data/assets/pdf_file/0010/78562/E87777.pdf

189. Sepkowitz KA. Effect of HAART on natural history of AIDS-related opportunistic disorders. Lancet. 1998; 351(9098): 228-30.

190. Carr A, Marriott D, Field A, Vasak E, Cooper DA. Treatment of HIV-1-associated microsporidiosis and cryptosporidiosis with combination antiretroviral therapy. Lancet. 1998; 351(9098): 256-61.

191. Whitcup SM, Fortin E, Nussenblatt RB, Polis MA, Muccioli C, Belfort R, Jr. Therapeutic effect of combination antiretroviral therapy on cytomegalovirus retinitis. JAMA. 1997; 277(19): 1519-20.

319

192. Murphy M, Armstrong D, Sepkowitz KA, Ahkami RN, Myskowski PL. Regression of AIDS-related Kaposi's sarcoma following treatment with an HIV-1 protease inhibitor. AIDS. 1997; 11(2): 261-2.

193. Elliot B, Aromin I, Gold R, Flanigan T, Mileno M. 2.5 year remission of AIDS-associated progressive multifocal leukoencephalopathy with combined antiretroviral therapy. Lancet. 1997; 349(9055): 850.

194. del Amo J, Moreno S, Bucher H, Furrer H, Logan R, Sterne J, Perez-Hoyos S, Jarrin I, Phillips A, Lodi S. Impact of Antiretroviral Therapy on Tuberculosis Incidence Among HIV-Positive Patients in High-Income Countries. Clin Infect Dis. 2012; 54(9): 1364-72.

195. Dromer F, Mathoulin-Pélissier S, Fontanet A, Ronin O, Dupont B, Lortholary O, Group obotFCS. Epidemiology of HIV-associated cryptococcosis in France (1985-2001): comparison of the pre- and post- HAART eras. AIDS. 2004; 18(3): 555-62.

196. van Elden LJ, Walenkamp AM, Lipovsky MM, Reiss P, Meis JF, de Marie S, Dankert J, Hoepelman AI. Declining number of patients with cryptococcosis in the Netherlands in the era of highly active antiretroviral therapy. AIDS. 2000; 14(17): 2787-8.

197. Raboud JM, Montaner JS, Conway B, Rae S, Reiss P, Vella S, Cooper D, Lange J, Harris M, Wainberg MA, Robinson P, Myers M, Hall D. Suppression of plasma viral load below 20 copies/ml is required to achieve a long-term response to therapy. AIDS. 1998; 12(13): 1619-24.

198. Pilcher CD, Miller WC, Beatty ZA, Eron JJ. Detectable HIV-1 RNA at levels below quantifiable limits by amplicor HIV-1 monitor is associated with virologic relapse on antiretroviral therapy. AIDS. 1999; 13(11): 1337-42.

199. British HIV Association. Guidelines for the treatment of HIV-1 positive adults with antiretroviral therapy. 2012 [cited; Available from: http://www.bhiva.org/TreatmentofHIV1_2012.aspx

200. Panel on Antiretroviral Guidelines for Adults and Adolescents - US Department of Health and Social Services. Guidelines for the use of antiretroviral agents in HIV-1 infected adults and adolescents; 2012.

201. Phillips AN, Staszewski S, Weber R, Kirk O, Francioli P, Miller V, Vernazza P, Lundgren JD, Ledergerber B. HIV viral load response to antiretroviral therapy according to the baseline CD4 cell count and viral load. JAMA. 2001; 286(20): 2560-7.

202. Gallant JE, DeJesus E, Arribas JR, Pozniak AL, Gazzard B, Campo RE, Lu B, McColl D, Chuck S, Enejosa J, Toole JJ, Cheng AK. Tenofovir DF, emtricitabine, and efavirenz vs. zidovudine, lamivudine, and efavirenz for HIV. N Engl J Med. 2006; 354(3): 251-60.

203. Post FA, Moyle GJ, Stellbrink HJ, Domingo P, Podzamczer D, Fisher M, Norden AG, Cavassini M, Rieger A, Khuong-Josses MA, Branco T, Pearce HC, Givens N, Vavro C, Lim ML. Randomized comparison of renal effects, efficacy, and safety with once-daily abacavir/lamivudine versus tenofovir/emtricitabine, administered with efavirenz, in antiretroviral-naive, HIV-1-infected adults: 48-week results from the ASSERT study. J Acquir Immune Defic Syndr. 2010; 55(1): 49-57.

204. von Wyl V, Yerly S, Boni J, Shah C, Cellerai C, Klimkait T, Battegay M, Bernasconi E, Cavassini M, Furrer H, Hirschel B, Vernazza PL, Ledergerber B, Gunthard HF. Incidence of HIV-1 drug resistance among antiretroviral treatment-naive individuals starting modern therapy combinations. Clin Infect Dis. 2012; 54(1): 131-40.

205. Parienti JJ, Bangsberg DR, Verdon R, Gardner EM. Better adherence with once-daily antiretroviral regimens: a meta-analysis. Clin Infect Dis. 2009; 48(4): 484-8.

320

206. Bangsberg DR, Perry S, Charlebois ED, Clark RA, Roberston M, Zolopa AR, Moss A. Non-adherence to highly active antiretroviral therapy predicts progression to AIDS. AIDS. 2001; 15(9): 1181-3.

207. Paterson DL, Swindells S, Mohr J, Brester M, Vergis EN, Squier C, Wagener MM, Singh N. Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Ann Intern Med. 2000; 133(1): 21-30.

208. Bangsberg DR. Preventing HIV antiretroviral resistance through better monitoring of treatment adherence. J Infect Dis. 2008; 197 Suppl 3: S272-8.

209. Reynolds NR, Sun J, Nagaraja HN, Gifford AL, Wu AW, Chesney MA. Optimizing measurement of self-reported adherence with the ACTG Adherence Questionnaire: a cross-protocol analysis. J Acquir Immune Defic Syndr. 2007; 46(4): 402-9.

210. Nieuwkerk PT, Oort FJ. Self-Reported Adherence to Antiretroviral Therapy for HIV-1 Infection and Virologic Treatment Response: A Meta-Analysis. J Acquir Immune Defic Syndr. 2005; 38(4): 445-8.

211. Calmy A, Hirschel B, Cooper DA, Carr A. A new era of antiretroviral drug toxicity. Antivir Ther. 2009; 14(2): 165-79.

212. Havlir DV, Currier JS. Complications of HIV disease and antiretroviral therapy. Top HIV Med. 2006; 14(1): 27-35.

213. Friis-Moller N, Reiss P, Sabin CA, Weber R, Monforte A, El-Sadr W, Thiebaut R, De Wit S, Kirk O, Fontas E, Law MG, Phillips A, Lundgren JD. Class of antiretroviral drugs and the risk of myocardial infarction. N Engl J Med. 2007; 356(17): 1723-35.

214. Scourfield A, Zheng J, Chinthapalli S, Waters L, Martin T, Mandalia S, Nelson M. Discontinuation of Atripla(R) as first-line therapy in HIV-1 infected individuals. AIDS. 2012.

215. ENCORE1 Study Group. Efficacy of 400 mg efavirenz versus standard 600 mg dose in HIV-infected, antiretroviral-naive adults (ENCORE1): a randomised, double-blind, placebo-controlled, non-inferiority trial. Lancet. 2014.

216. Shubber Z, Calmy A, Andrieux-Meyer I, Vitoria M, Renaud-Thery F, Shaffer N, Hargreaves S, Mills EJ, Ford N. Adverse events associated with nevirapine and efavirenz-based first-line antiretroviral therapy: a systematic review and meta-analysis. AIDS. 2013; 27(9): 1403-12.

217. Stohr W, Back D, Dunn D, Sabin C, Winston A, Gilson R, Pillay D, Hill T, Ainsworth J, Pozniak A, Leen C, Bansi L, Fisher M, Orkin C, Anderson J, Johnson M, Easterbrook P, Gibbons S, Khoo S. Factors influencing efavirenz and nevirapine plasma concentration: effect of ethnicity, weight and co-medication. Antivir Ther. 2008; 13(5): 675-85.

218. Nyakutira C, Roshammar D, Chigutsa E, Chonzi P, Ashton M, Nhachi C, Masimirembwa C. High prevalence of the CYP2B6 516G-->T(*6) variant and effect on the population pharmacokinetics of efavirenz in HIV/AIDS outpatients in Zimbabwe. Eur J Clin Pharmacol. 2008; 64(4): 357-65.

219. Haas DW, Ribaudo HJ, Kim RB, Tierney C, Wilkinson GR, Gulick RM, Clifford DB, Hulgan T, Marzolini C, Acosta EP. Pharmacogenetics of efavirenz and central nervous system side effects: an Adult AIDS Clinical Trials Group study. AIDS. 2004; 18(18): 2391-400.

220. Antiretroviral Therapy Cohort Collaboration. Life expectancy of individuals on combination antiretroviral therapy in high-income countries: a collaborative analysis of 14 cohort studies. Lancet. 2008; 372(9635): 293-9.

321

221. Kuller LH, Tracy R, Belloso W, De Wit S, Drummond F, Lane HC, Ledergerber B, Lundgren J, Neuhaus J, Nixon D, Paton NI, Neaton JD. Inflammatory and coagulation biomarkers and mortality in patients with HIV infection. PLoS Med. 2008; 5(10): e203.

222. When To Start Consortium, Sterne JAC, May M, Costagliola D, de Wolf F, Phillips AN, Harris R, Funk MJ, Geskus RB, Gill J, Dabis F, Miró JM, Justice AC, Ledergerber B, Fätkenheuer G, Hogg RS, Monforte ADA, Saag M, Smith C, Staszewski S, Egger M, Cole SR. Timing of initiation of antiretroviral therapy in AIDS-free HIV-1-infected patients: a collaborative analysis of 18 HIV cohort studies. Lancet. 2009; 373(9672): 1352-63.

223. El-Sadr WM, Lundgren JD, Neaton JD, Gordin F, Abrams D, Arduino RC, Babiker A, Burman W, Clumeck N, Cohen CJ, Cohn D, Cooper D, Darbyshire J, Emery S, Fatkenheuer G, Gazzard B, Grund B, Hoy J, Klingman K, Losso M, Markowitz N, Neuhaus J, Phillips A, Rappoport C. CD4+ count-guided interruption of antiretroviral treatment. N Engl J Med. 2006; 355(22): 2283-96.

224. Severe P, Juste MA, Ambroise A, Eliacin L, Marchand C, Apollon S, Edwards A, Bang H, Nicotera J, Godfrey C, Gulick RM, Johnson WD, Jr., Pape JW, Fitzgerald DW. Early versus standard antiretroviral therapy for HIV-infected adults in Haiti. N Engl J Med. 2010; 363(3): 257-65.

225. Siegfried N, Uthman OA, Rutherford GW. Optimal time for initiation of antiretroviral therapy in asymptomatic, HIV-infected, treatment-naive adults. Cochrane Database Syst Rev. 2010; (3): CD008272.

226. Granich R, Kahn JG, Bennett R, Holmes CB, Garg N, Serenata C, Sabin ML, Makhlouf-Obermeyer C, De Filippo Mack C, Williams P, Jones L, Smyth C, Kutch KA, Ying-Ru L, Vitoria M, Souteyrand Y, Crowley S, Korenromp EL, Williams BG. Expanding ART for treatment and prevention of HIV in South Africa: estimated cost and cost-effectiveness 2011-2050. PLoS One. 2012; 7(2): e30216.

227. Koenig SP, Bang H, Severe P, Jean Juste MA, Ambroise A, Edwards A, Hippolyte J, Fitzgerald DW, McGreevy J, Riviere C, Marcelin S, Secours R, Johnson WD, Pape JW, Schackman BR. Cost-effectiveness of early versus standard antiretroviral therapy in HIV-infected adults in Haiti. PLoS Med. 2011; 8(9): e1001095.

228. Walensky RP, Wood R, Ciaranello AL, Paltiel AD, Lorenzana SB, Anglaret X, Stoler AW, Freedberg KA, for the CEPAC-International Investigators. Scaling Up the 2010 World Health Organization HIV Treatment Guidelines in Resource-Limited Settings: A Model-Based Analysis. PLoS Med. 2010; 7(12): e1000382.

229. World Health Organization. Antiretroviral therapy for HIV infection in adults and adolescents: recommendations for a public health approach. 2010 [cited; Available from: http://whqlibdoc.who.int/publications/2010/9789241599764_eng.pdf

230. Hirnschall G, Harries AD, Easterbrook PJ, Doherty MC, Ball A. The next generation of the World Health Organization's global antiretroviral guidance. Journal of the International AIDS Society. 2013; 16: 18757.

231. Sterling TR, Chaisson RE, Moore RD. HIV-1 RNA, CD4 T-lymphocytes, and clinical response to highly active antiretroviral therapy. AIDS. 2001; 15(17): 2251-7.

232. World Health Organization. Technical Brief on CD4 Technologies. 2012 [cited; Available from: http://www.who.int/hiv/topics/treatment/tech_brief_20100501_en.pdf

233. Zachariah R, Reid SD, Chaillet P, Massaquoi M, Schouten EJ, Harries AD. Viewpoint: Why do we need a point-of-care CD4 test for low-income countries? Trop Med Int Health. 2011; 16(1): 37-41.

234. Bassett IV, Regan S, Chetty S, Giddy J, Uhler LM, Holst H, Ross D, Katz JN, Walensky RP, Freedberg KA, Losina E. Who starts antiretroviral therapy in Durban, South Africa?... not everyone who should. AIDS. 2010; 24 Suppl 1: S37-44.

322

235. Micek MA, Gimbel-Sherr K, Baptista AJ, Matediana E, Montoya P, Pfeiffer J, Melo A, Gimbel-Sherr S, Johnson W, Gloyd S. Loss to follow-up of adults in public HIV care systems in central Mozambique: identifying obstacles to treatment. J Acquir Immune Defic Syndr. 2009; 52(3): 397-405.

236. Jani IV, Sitoe NE, Alfai ER, Chongo PL, Quevedo JI, Rocha BM, Lehe JD, Peter TF. Effect of point-of- care CD4 cell count tests on retention of patients and rates of antiretroviral therapy initiation in primary health clinics: an observational cohort study. Lancet. 2011; 378(9802): 1572-9.

237. Larson B, Schnippel K, Ndibongo B, Xulu T, Brennan A, Long L, Fox MP, Rosen S. Rapid point-of-care CD4 testing at mobile HIV testing sites to increase linkage to care: An evaluation of a pilot program in South Africa. J Acquir Immune Defic Syndr. 2012.

238. Myer L, Zulliger R, Black S, Pienaar D, Bekker LG. Pilot programme for the rapid initiation of antiretroviral therapy in pregnancy in Cape Town, South Africa. AIDS Care. 2012; 24(8): 986-92.

239. Colebunders R, Mann JM, Francis H, Bila K, Izaley L, Kakonde N, Kabasele K, Ifoto L, Nzilambi N, Quinn TC. Evaluation of a clinical case-definition of acquired immunodeficiency syndrome in Africa. Lancet. 1987; 1(8531): 492-4.

240. World Health Organization. Workshop on AIDS in Central Africa. 1985 [cited; Available from: http://apps.who.int/iris/handle/10665/60154

241. World Health Organization. Acquired immuno- deficiency syndrome (AIDS). Weekly Epidemiological Report. 1986; 61: 69-73.

242. De Cock KM, Colebunders R, Francis H, Nzilambi N, Laga M, Ryder RW, Bondjobo M, McCormick JB, Piot P. Evaluation of the WHO clinical case definition for AIDS in rural Zaire. AIDS. 1988; 2(3): 219-21.

243. Widy-Wirski R, Berkley S, Downing R, Okware S, Recine U, Mugerwa R, Lwegaba A, Sempala S. Evaluation of the WHO clinical case definition for AIDS in Uganda. JAMA. 1988; 260(22): 3286-9.

244. Interim proposal for a WHO Staging System for HIV infection and Disease. Wkly Epidemiol Rec. 1990; 65(29): 221-4.

245. Gilks C. What use is a clinical case definition for AIDS in Africa. BMJ. 1990; 303(6811): 1189-90.

246. World Health Organization. Interim WHO clinical staging of HIV/AIDS and HIV/AIDS case definitions for surveillance. 2005 [cited; Available from: http://www.who.int/hiv/pub/guidelines/casedefinitions/en/index.html

247. World Health Organization. WHO case definitions for surveillance and revised clinical staging and immunological classification of HIV-related disease in adults and children. 2007 [cited; Available from: http://www.who.int/hiv/pub/guidelines/HIVstaging150307.pdf

248. Malamba SS, Morgan D, Clayton T, Mayanja B, Okongo M, Whitworth J. The prognostic value of the World Health Organisation staging system for HIV infection and disease in rural Uganda. AIDS. 1999; 13(18): 2555.

249. Sewankambo NK, Gray RH, Ahmad S, Serwadda D, Wabwire-Mangen F, Nalugoda F, Kiwanuka N, Lutalo T, Kigozi G, Li C. Mortality associated with HIV infection in rural Rakai District, Uganda. AIDS. 2000; 14(15): 2391.

323

250. Lawn SD, Harries AD, Anglaret X, Myer L, Wood R. Early mortality among adults accessing antiretroviral treatment programmes in sub-Saharan Africa. AIDS. 2008; 22(15): 1897-908.

251. Rosen S, Fox MP, Gill CJ. Patient retention in antiretroviral therapy programs in sub-Saharan Africa: a systematic review. PLoS Med. 2007; 4(10): e298.

252. Varughese JK, Rosenberg MG, Kim K. HIV in the tropics: staging in the resource-limited setting. Curr Opin Infect Dis. 2012.

253. Harbord RM, Deeks JJ, Egger M, Whiting P, Sterne J. A unification of models for meta-analysis of diagnostic accuracy studies. Biostatistics. 2007; 8: 239-51.

254. Boniphace I, Omari M, Susan Fred R, Ferdinand M, Marcel T. HIV/AIDS Clinical Manifestations and their Implication for Patient Clinical Staging in Resource Limited Settings in Tanzania. Open AIDS J. 2011; 5: 9-16.

255. Carter RJ, Dugan K, El-Sadr WM, Myer L, Otieno J, Pungpapong N, Toro PL, Abrams EJ, MTCT Plus Initiative. CD4+ cell count testing more effective than HIV disease clinical staging in identifying pregnant and postpartum women eligible for antiretroviral therapy in resource-limited settings. J Acquir Immune Defic Syndr. 2010; 55(3): 404-10.

256. Fox MP, Sanne IM, Conradie F, Zeinecker J, Orrell C, Ive P, Rassool M, Dehlinger M, van der Horst C, McIntyre J, Wood R. Initiating patients on antiretroviral therapy at CD4 cell counts above 200 cells/mul is associated with improved treatment outcomes in South Africa. AIDS. 2010; 24(13): 2041-50.

257. French N, Mujugira A, Nakiyingi J, Mulder D, Janoff EN, Gilks CF. Immunologic and clinical stages in HIV-1-infected Ugandan adults are comparable and provide no evidence of rapid progression but poor survival with advanced disease. J Acquir Immune Defic Syndr. 1999; 22(5): 509-16.

258. Ilovi CS, Lule GN, Obel O, Irimu M. Correlation of who clinical staging with CD4 counts in adult HIV/AIDS patients at Kenyatta National Hospital, Nairobi. East Afr Med J. 2011; 88(2): 65-70.

259. Jaffar S, Birungi J, Grosskurth H, Amuron B, Namara G, Nabiryo C, Coutinho A. Use of WHO clinical stage for assessing patient eligibility to antiretroviral therapy in a routine health service setting in Jinja, Uganda. AIDS Res Ther. 2008; 5: 4.

260. Kagaayi J, Makumbi F, Nakigozi G, Wawer MJ, Gray RH, Serwadda D, Reynolds SJ. WHO HIV clinical staging or CD4 cell counts for antiretroviral therapy eligibility assessment? An evaluation in rural Rakai district, Uganda. AIDS. 2007; 21(9): 1208-10.

261. McGrath N, Kranzer K, Saul J, Crampin AC, Malema S, Kachiwanda L, Zaba B, Jahn A, Fine PE, Glynn JR. Estimating the need for antiretroviral treatment and an assessment of a simplified HIV/AIDS case definition in rural Malawi. AIDS. 2007; 21 Suppl 6: S105-13.

262. Morpeth SC, Crump JA, Shao HJ, Ramadhani HO, Kisenge PR, Moylan CA, Naggie S, Caram LB, Landman KZ, Sam NE, Itemba DK, Shao JF, Bartlett JA, Thielman NM. Predicting CD4 lymphocyte count <200 cells/mm(3) in an HIV type 1-infected African population. AIDS Res Hum Retroviruses. 2007; 23(10): 1230-6.

263. Tassie JM, Marquardt T, Damisoni H, Odhiambo OD, Mulemba M, Szumilin E, Legros D. Indirect markers to initiate highly active antiretroviral therapy in a rural African setting. AIDS. 2004; 18(8): 1226-8.

264. Torpey K, Lartey M, Amenyah R, Addo NA, Obeng-Baah J, Rahman Y, Suzuki C, Mukadi YD, Colebunders R. Initiating antiretroviral treatment in a resource-constrained setting: does clinical staging effectively identify patients in need? Int J STD AIDS. 2009; 20(6): 395-8.

324

265. Baveewo S, Ssali F, Karamagi C, Kalyango JN, Hahn JA, Ekoru K, Mugyenyi P, Katabira E. Validation of World Health Organisation HIV/AIDS Clinical Staging in Predicting Initiation of Antiretroviral Therapy and Clinical Predictors of Low CD4 Cell Count in Uganda. PLoS One. 2011; 6(5): e19089.

266. Miiro G, Nakubulwa S, Watera C, Munderi P, Floyd S, Grosskurth H. Evaluation of affordable screening markers to detect CD4+ T-cell counts below 200 cells/mul among HIV-1-infected Ugandan adults. Trop Med Int Health. 2010; 15(4): 396-404.

267. Murray CJL, Chen LC. Understanding morbidity change. The Population and Development Review. 1992: 481-503.

268. Idler EL, Benyamini Y. Self-rated health and mortality: a review of twenty-seven community studies. J Health Soc Behav. 1997; 38(1): 21-37.

269. Burström B, Fredlund P. Self rated health: Is it as good a predictor of subsequent mortality among adults in lower as well as in higher social classes? J Epidemiol Community Health. 2001; 55(11): 836-40.

270. Singh-Manoux A, Martikainen P, Ferrie J, Zins M, Marmot M, Goldberg M. What does self rated health measure? Results from the British Whitehall II and French Gazel cohort studies. J Epidemiol Community Health. 2006; 60(4): 364-72.

271. Brown AF, Ang A, Pebley AR. The relationship between neighborhood characteristics and self- rated health for adults with chronic conditions. Am J Public Health. 2007; 97(5): 926-32.

272. Singh-Manoux A, Dugravot A, Shipley MJ, Ferrie JE, Martikainen P, Goldberg M, Zins M, GAZEL Cohort. The association between self-rated health and mortality in different socioeconomic groups in the GAZEL cohort study. Int J Epidemiol. 2007; 36(6): 1222-8.

273. Streiner DL, Norman GR. Health measurement scales : a practical guide to their development and use. Oxford; New York: Oxford University Press; 2008.

274. Sen A. Health: perception versus observation. BMJ. 2002; 324(7342): 860-1.

275. Smide B, Whiting D, Mugusi F, Felten L, Wikblad K. Self-perceived health in urban diabetic patients in Tanzania. East Afr Med J. 1999; 76(2): 67-70.

276. DeSalvo KB, Bloser N, Reynolds K, He J, Muntner P. Mortality prediction with a single general self- rated health question. A meta-analysis. J Gen Intern Med. 2006; 21(3): 267-75.

277. Mrus JM, Schackman BR, Wu AW, Freedberg KA, Tsevat J, Yi MS, Zackin R. Variations in self-rated health among patients with HIV infection. Qual Life Res. 2006; 15(3): 503-14.

278. Cardarelli R, Weis S, Adams E, Radaford D, Vecino I, Munguia G, Johnson KL, Fulda KG. General health status and adherence to antiretroviral therapy. Journal of the International Association of Physicians in AIDS Care. 2008; 7(3): 123-9.

279. Munyati SS, Redzo N, Dauya E, Matambo R, Makamure B, Bandason T, Butterworth AE, Gwanzura L, Rusakaniko S, Mason PR, Corbett EL. Human immunodeficiency virus, smoking and self-rated health in Harare, Zimbabwe. Int J Tuberc Lung Dis. 2006; 10(11): 1279-85.

280. Losina E, Anglaret X, Yazdanpanah Y, Wang B, Toure S, Seage GR, N'Dri-Yoman T, Walensky RP, Dakoury-Dogbo N, Goldie SJ, Messou E, Weinstein MC, Deuffic-Burban S, Salamon R, Freedberg KA. Impact

325

of opportunistic diseases on chronic mortality in HIV-infected adults in Côte d'Ivoire. S Afr Med J. 2006; 96(6): 526-9.

281. Mallal S, Nolan D, Witt C, Masel G, Martin A, Moore C, Sayer D, Castley A, Mamotte C, Maxwell D. Association between presence of HLA-B* 5701, HLA-DR7, and HLA-DQ3 and hypersensitivity to HIV-1 reverse-transcriptase inhibitor abacavir. Lancet. 2002; 359(9308): 727-32.

282. Walker AS, Gibb DM. Monitoring of highly active antiretroviral therapy in HIV infection. Curr Opin Infect Dis. 2011; 24(1): 27-33.

283. Harries AD, Schouten EJ, Libamba E. Scaling up antiretroviral treatment in resource-poor settings. Lancet. 2006; 367(9525): 1870-2.

284. Gilks CF, Crowley S, Ekpini R, Gove S, Perriens J, Souteyrand Y, Sutherland D, Vitoria M, Guerma T, De Cock K. The WHO public-health approach to antiretroviral treatment against HIV in resource-limited settings. Lancet. 2006; 368(9534): 505-10.

285. Ford N, Calmy A, Mills EJ. The first decade of antiretroviral therapy in Africa. Global Health. 2011; 7: 33.

286. Marseille E, Hofmann PB, Kahn JG. HIV prevention before HAART in sub-Saharan Africa. Lancet. 2002; 359(9320): 1851-6.

287. Schwartlander B, Grubb I, Perriens J. The 10-year struggle to provide antiretroviral treatment to people with HIV in the developing world. Lancet. 2006; 368(9534): 541-6.

288. Abdool Karim SS, Abdool Karim Q. Breaking the silence, one year later: reflections on the Durban Conference. AIDS Clin Care. 2001; 13(7): 63-5, 70.

289. Malaria. TGFtFATa. The framework document of the Global Fund to Fight AIDS, Tuberculosis and Malaria. 2001 [cited; Available from: http://www.hivpolicy.org/Library/HPP001097.pdf

290. Bendavid E, Bhattacharya J. The President's Emergency Plan for AIDS Relief in Africa: An Evaluation of Outcomes. Ann Intern Med. 2009; 150(10): 688-95.

291. World Bank. The Africa Multi-Country AIDS Programme 2000-2006. Results of the World Bank's response to a development crisis. 2007 [cited; Available from: http://siteresources.worldbank.org/EXTAFRREGTOPHIVAIDS/Resources/717147- 1181768523896/complete.pdf

292. World Health Organization. Treat 3 million by 2005 initiative. 2003 [cited; Available from: http://www.who.int/3by5/publications/documents/en/3by5StrategyMakingItHappen.pdf

293. Harries AD, Nyangulu DS, Hargreaves NJ, Kaluwa O, Salaniponi FM. Preventing antiretroviral anarchy in sub-Saharan Africa. Lancet. 2001; 358(9279): 410-4.

294. Harries AD, Makombe SD, Libamba E, Schouten EJ. Why Did the Scale-up of HIV Treatment Work?: A Case Example From Malawi. J Acquir Immune Defic Syndr. 2011; 57 Suppl 2: S64-7.

295. World Health Organization, Atun RA, Bennett S, Duran A. When do vertical (stand-alone) programmes have a place in the health system? 2008 [cited; Available from: http://www.who.int/management/district/services/WhenDoVerticalProgrammesPlaceHealthSystems.pdf

326

296. Brinkhof MWG, Pujades-Rodriguez M, Egger M. Mortality of patients lost to follow-up in antiretroviral treatment programmes in resource-limited settings: systematic review and meta-analysis. PLoS One. 2009; 4(6): e5790.

297. Nachega JB, Chaisson RE, Goliath R, Efron A, Chaudhary MA, Ram M, Morroni C, Schoeman H, Knowlton AR, Maartens G. Randomized controlled trial of trained patient-nominated treatment supporters providing partial directly observed antiretroviral therapy. AIDS. 2010; 24(9): 1273-80.

298. Long L, Brennan A, Fox MP, Ndibongo B, Jaffray I, Sanne I, Rosen S. Treatment Outcomes and Cost- Effectiveness of Shifting Management of Stable ART Patients to Nurses in South Africa: An Observational Cohort. PLoS Med. 2011; 8(7): e1001055.

299. Fairall L, Bachmann MO, Lombard C, Timmerman V, Uebel K, Zwarenstein M, Boulle A, Georgeu D, Colvin CJ, Lewin S, Faris G, Cornick R, Draper B, Tshabalala M, Kotze E, van Vuuren C, Steyn D, Chapman R, Bateman E. Task shifting of antiretroviral treatment from doctors to primary-care nurses in South Africa (STRETCH): a pragmatic, parallel, cluster-randomised trial. Lancet. 2012; 380(9845): 889-98.

300. Lawn SD, Harries AD, Wood R. Strategies to reduce early morbidity and mortality in adults receiving antiretroviral therapy in resource-limited settings. Curr Opin HIV AIDS. 2010; 5(1): 18-26.

301. Keiser O, Orrell C, Egger M, Wood R, Brinkhof MWG, Furrer H, van Cutsem G, Ledergerber B, Boulle A, Swiss HIV Cohort Study (SHCS) and the International Epidemiologic Databases to Evaluate AIDS in Southern Africa (IeDEA-SA). Public-health and individual approaches to antiretroviral therapy: township South Africa and Switzerland compared. PLoS Med. 2008; 5(7): e148.

302. Mills EJ, Nachega JB, Buchan I, Orbinski J, Attaran A, Singh S, Rachlis B, Wu P, Cooper C, Thabane L, Wilson K, Guyatt GH, Bangsberg DR. Adherence to antiretroviral therapy in sub-Saharan Africa and North America: a meta-analysis. JAMA. 2006; 296(6): 679-90.

303. Ford N, Darder M, Spelman T, Maclean E, Mills E, Boulle A. Early adherence to antiretroviral medication as a predictor of long-term HIV virological suppression: five-year follow up of an observational cohort. PLoS One. 2010; 5(5): e10460.

304. Rueda S, Park-Wyllie LY, Bayoumi AM, Tynan AM, Antoniou TA, Rourke SB, Glazier RH. Patient support and education for promoting adherence to highly active antiretroviral therapy for HIV/AIDS. Cochrane Database Syst Rev. 2006; (3): CD001442.

305. Chung MH, Richardson BA, Tapia K, Benki-Nugent S, Kiarie JN, Simoni JM, Overbaugh J, Attwa M, John-Stewart GC. A randomized controlled trial comparing the effects of counseling and alarm device on HAART adherence and virologic outcomes. PLoS Med. 2011; 8(3): e1000422.

306. Horvath T, Azman H, Kennedy GE, Rutherford GW. Mobile phone text messaging for promoting adherence to antiretroviral therapy in patients with HIV infection. Cochrane Database Syst Rev. 2012; 3: CD009756.

307. Lawn SD, Little F, Bekker L-G, Kaplan R, Campbel E, Orrell C, Wood R. Changing mortality risk associated with CD4 cell response to antiretroviral therapy in South Africa. AIDS. 2009; 23(3): 335-42.

308. Battegay M, Nüesch R, Hirschel B, Kaufmann GR. Immunological recovery and antiretroviral therapy in HIV-1 infection. Lancet Infect Dis. 2006; 6(5): 280-7.

309. Weidle PJ, Malamba S, Mwebaze R, Sozi C, Rukundo G, Downing R, Hanson D, Ochola D, Mugyenyi P, Mermin J, Samb B, Lackritz E. Assessment of a pilot antiretroviral drug therapy programme in Uganda: patients' response, survival, and drug resistance. Lancet. 2002; 360(9326): 34-40.

327

310. Kipp W, Konde-Lule J, Saunders LD, Alibhai A, Houston S, Rubaale T, Senthilselvan A, Okech-Ojony J. Antiretroviral Treatment for HIV in Rural Uganda: Two-Year Treatment Outcomes of a Prospective Health Centre/Community-Based and Hospital-Based Cohort. PLoS One. 2012; 7(7): e40902.

311. Auld AF, Mbofana F, Shiraishi RW, Sanchez M, Alfredo C, Nelson LJ, Ellerbrock T. Four-year treatment outcomes of adult patients enrolled in Mozambique's rapidly expanding antiretroviral therapy program. PLoS One. 2011; 6(4): e18453.

312. Wester CW, Kim S, Bussmann H, Avalos A, Ndwapi N, Peter TF, Gaolathe T, Mujugira A, Busang L, Vanderwarker C, Cardiello P, Johnson O, Thior I, Mazonde P, Moffat H, Essex M, Marlink R. Initial response to highly active antiretroviral therapy in HIV-1C-infected adults in a public sector treatment program in Botswana. J Acquir Immune Defic Syndr. 2005; 40(3): 336-43.

313. US Department of Health and Human Services. Guidelines for the Use of Antiretroviral Agents in HIV-1-Infected Adults and Adolescents. 2011 [cited; Available from: http://aidsinfo.nih.gov/contentfiles/lvguidelines/adultandadolescentgl.pdf

314. Ivers LC, Kendrick D, Doucette K. Efficacy of antiretroviral therapy programs in resource-poor settings: a meta-analysis of the published literature. Clin Infect Dis. 2005; 41(2): 217-24.

315. Hammond R, Harry TC. Efficacy of antiretroviral therapy in Africa: effect on immunological and virological outcome measures -- a meta-analysis. Int J STD AIDS. 2008; 19(5): 291-6.

316. Gupta RK, Hill A, Sawyer AW, Cozzi-Lepri A, von Wyl V, Yerly S, Lima VD, Gunthard HF, Gilks C, Pillay D. Virological monitoring and resistance to first-line highly active antiretroviral therapy in adults infected with HIV-1 treated under WHO guidelines: a systematic review and meta-analysis. Lancet Infect Dis. 2009; 9(7): 409-17.

317. Makombe SD, Jahn A, Tweya H, Chuka S, Yu JK-L, Hochgesang M, Aberle-Grasse J, Pasulani O, Schouten EJ, Kamoto K, Harries AD. A national survey of the impact of rapid scale-up of antiretroviral therapy on health-care workers in Malawi: effects on human resources and survival. Bull World Health Organ. 2007; 85(11): 851-7.

318. Sloan DJ, van Oosterhout JJ, Malisita K, Phiri EM, Lalloo DG, O'Hare B, Macpherson P. Evidence of improving antiretroviral therapy treatment delays: an analysis of eight years of programmatic outcomes in Blantyre, Malawi. BMC Public Health. 2013; 13: 490.

319. Malawi. HAUMoHo. Integrated HIV Program Report January – March 2012. 2012 [cited; Available from: http://www.hivunitmohmw.org/Main/AntiretroviralTherapy

320. Schouten EJ, Jahn A, Midiani D, Makombe SD, Mnthambala A, Chirwa Z, Harries AD, van Oosterhout JJ, Meguid T, Ben-Smith A, Zachariah R, Lynen L, Zolfo M, Van Damme W, Gilks CF, Atun R, Shawa M, Chimbwandira F. Prevention of mother-to-child transmission of HIV and the health-related Millennium Development Goals: time for a public health approach. Lancet. 2011; 378(9787): 282-4.

321. Impact of an innovative approach to prevent mother-to-child transmission of HIV--Malawi, July 2011-September 2012. MMWR Morb Mortal Wkly Rep. 2013; 62(8): 148-51.

322. Goosby EP. Is Option B+ the best choice? Lancet. 2013; 381(9874): 1272.

323. Tenthani L, Haas AD, Tweya H, Jahn A, van Oosterhout JJ, Chimbwandira F, Chirwa Z, Ng'ambi W, Bakali A, Phiri S, Myer L, Valeri F, Zwahlen M, Wandeler G, Keiser O. Retention in care under universal antiretroviral therapy for HIV-infected pregnant and breastfeeding women ('Option B+') in Malawi. AIDS. 2014; 28(4): 589-98.

328

324. Yu JK-L, Chen SC-C, Wang K-Y, Chang C-S, Makombe SD, Schouten EJ, Harries AD. True outcomes for patients on antiretroviral therapy who are "lost to follow-up" in Malawi. Bull World Health Organ. 2007; 85(7): 550-4.

325. Waaler HT, Piot MA. Use of an epidemiological model for estimating the effectiveness of tuberculosis control measures. Sensitivity of the effectiveness of tuberculosis control measures to the social time preference. Bull World Health Organ. 1970; 43(1): 1-16.

326. McNairy ML, El-Sadr WM. The HIV Care Continuum. AIDS. 2012; 26(14): 1735-8.

327. Fox MP, Larson B, Rosen S. Defining retention and attrition in pre-antiretroviral HIV care: proposals based on experience in Africa. Trop Med Int Health. 2012; [Epub ahead of print].

328. Mugglin C, Estill J, Wandeler G, Bender N, Egger M, Gsponer T, Keiser O. Loss to programme between HIV diagnosis and initiation of antiretroviral therapy in sub-Saharan Africa: systematic review and meta-analysis. Trop Med Int Health. 2012; [Epub ahead of print].

329. Kranzer K, Govindasamy D, Ford N, Johnston V, Lawn SD. Quantifying and addressing losses along the continuum of care for people living with HIV infection in sub-Saharan Africa: a systematic review. Journal of the International AIDS Society. 2012; 15(2): 17383.

330. Rosen S, Fox MP. Retention in HIV Care between Testing and Treatment in Sub-Saharan Africa: A Systematic Review. PLoS Med. 2011; 8(7): e1001056.

331. World Health Organization. Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection. Geneva; 2013.

332. Kranzer K, Zeinecker J, Ginsberg P, Orrell C, Kalawe NN, Lawn SD, Bekker L-G, Wood R. Linkage to HIV Care and Antiretroviral Therapy in Cape Town, South Africa. PLoS One. 2010; 5(11): e13801.

333. Lawn SD, Myer L, Orrell C, Bekker L-G, Wood R. Early mortality among adults accessing a community-based antiretroviral service in South Africa: implications for programme design. AIDS. 2005; 19(18): 2141-8.

334. Amolloh M, Medley A, Owuor P, Audi B, Sewe M. Factors associated with early uptake of HIV care and treatment services after testing HIV-positive during home based testing and counseling (HBCT) in rural Western Kenya. 18th Conference on Retroviruses and Opportunistic Infections. Boston, USA; 2011.

335. Amuron B, Namara G, Birungi J, Nabiryo C, Levin J, Grosskurth H, Coutinho A, Jaffar S. Mortality and loss-to-follow-up during the pre-treatment period in an antiretroviral therapy programme under normal health service conditions in Uganda. BMC Public Health. 2009; 9: 290.

336. April MD, Walensky RP, Chang Y, Pitt J, Freedberg KA, Losina E, Paltiel AD, Wood R. HIV testing rates and outcomes in a South African community, 2001-2006: implications for expanded screening policies. J Acquir Immune Defic Syndr. 2009; 51(3): 310-6.

337. Assefa Y, Van Damme W, Mariam DH, Kloos H. Toward universal access to HIV counseling and testing and antiretroviral treatment in Ethiopia: looking beyond HIV testing and ART initiation. AIDS Patient Care STDS. 2010; 24(8): 521-5.

338. Bassett IV, Wang B, Chetty S, Mazibuko M, Bearnot B, Giddy J, Lu Z, Losina E, Walensky RP, Freedberg KA. Loss to care and death before antiretroviral therapy in Durban, South Africa. J Acquir Immune Defic Syndr. 2009; 51(2): 135-9.

329

339. Braunstein SL, Umulisa MM, Veldhuijzen NJ, Kestelyn E, Ingabire CM, Nyinawabega J, van de Wijgert JH, Nash D. HIV diagnosis, linkage to HIV care, and HIV risk behaviors among newly diagnosed HIV- positive female sex workers in Kigali, Rwanda. J Acquir Immune Defic Syndr. 2011; 57(4): e70-6.

340. Feucht UD, Kinzer M, Kruger M. Reasons for delay in initiation of antiretroviral therapy in a population of HIV-infected South African children. J Trop Pediatr. 2007; 53(6): 398-402.

341. Gareta D, Tewya H, Weigel R, Phiri S, Chiwoko J. Linking HIV-infected pregnant women to antiretroviral therapy: experience from Lilongwe, Malawi. XVIII International AIDS Conference. Vienna, Austria; 2010.

342. Geng EH, Bwana MB, Kabakyenga J, Muyindike W, Emenyonu NI, Musinguzi N, Mugyenyi P, Martin JN, Bangsberg DR. Diminishing Availability of Publicly Funded Slots for Antiretroviral Initiation among HIV- Infected ART-Eligible Patients in Uganda. PLoS One. 2010; 5(11): e14098.

343. Govindasamy D, van Schaik N, Kranzer K, Wood R, Mathews C, Bekker L-G. Linkage to HIV Care from a Mobile Testing Unit in South Africa by Different CD4 Count Strata. J Acquir Immune Defic Syndr. 2011; 58(3): 344-52.

344. Ingle SM, May M, Uebel K, Timmerman V, Kotze E, Bachmann M, Sterne JAC, Egger M, Fairall L, IeDEA-Southern Africa. Outcomes in patients waiting for antiretroviral treatment in the Free State Province, South Africa: prospective linkage study. AIDS. 2010; 24(17): 2717-25.

345. Kaplan R, Orrell C, Zwane E, Bekker LG, Wood R. Loss to follow-up and mortality among pregnant women referred to a community clinic for antiretroviral treatment. AIDS. 2008; 22(13): 1679-81.

346. Karcher H, Omondi A, Odera J, Kunz A, Harms G. Risk factors for treatment denial and loss to follow-up in an antiretroviral treatment cohort in Kenya. Trop Med Int Health. 2007; 12(5): 687-94.

347. Kohler PK, Chung MH, McGrath CJ, Benki-Nugent SF, Thiga JW, John-Stewart GC. Implementation of free cotrimoxazole prophylaxis improves clinic retention among antiretroviral therapy-ineligible clients in Kenya. AIDS. 2011; 25(13): 1657-61.

348. Konde-Lule J, Makumbi F, Pakker N, Muyinda A, Mubiru M, Cobelens FG. Effect of changing antiretroviral treatment eligibility criteria on patient load in Kampala, Uganda. AIDS Care. 2011; 23(1): 35- 41.

349. Larson BA, Brennan A, McNamara L, Long L, Rosen S, Sanne I, Fox MP. Lost opportunities to complete CD4+ lymphocyte testing among patients who tested positive for HIV in South Africa. Bull World Health Organ. 2010; 88(9): 675-80.

350. Larson BA, Brennan A, McNamara L, Long L, Rosen S, Sanne I, Fox MP. Early loss to follow up after enrolment in pre-ART care at a large public clinic in Johannesburg, South Africa. Trop Med Int Health. 2010; 15 Suppl 1: 43-7.

351. Lawn SD, Myer L, Harling G, Orrell C, Bekker LG, Wood R. Determinants of mortality and nondeath losses from an antiretroviral treatment service in South Africa: implications for program evaluation. Clin Infect Dis. 2006; 43(6): 770-6.

352. Lessells RJ, Mutevedzi PC, Cooke GS, Newell ML. Retention in HIV care for individuals not yet eligible for antiretroviral therapy: rural KwaZulu-Natal, South Africa. J Acquir Immune Defic Syndr. 2011; 56(3): e79-86.

330

353. Losina E, Bassett IV, Giddy J, Chetty S, Regan S, Walensky RP, Ross D, Scott CA, Uhler LM, Katz JN, Holst H, Freedberg KA. The "ART" of linkage: pre-treatment loss to care after HIV diagnosis at two PEPFAR sites in Durban, South Africa. PLoS One. 2010; 5(3): e9538.

354. Luseno W, Wechsberg W, Middlesteadt-Ellerson R, Gumula W. Linkages and barriers to care for high-risk South African women testing positive for HIV. XVII International AIDS Conference. Mexico City, Mexico; 2008.

355. McGrath N, Glynn JR, Saul J, Kranzer K, Jahn A, Mwaungulu F, Ngwira MH, Mvula H, Munthali F, Mwinuka V, Mwaungulu L, Fine PE, Crampin AC. What happens to ART-eligible patients who do not start ART? Dropout between screening and ART initiation: a cohort study in Karonga, Malawi. BMC Public Health. 2010; 10: 601.

356. McGuire M, Munyenyembe T, Szumilin E, Heinzelmann A, Le Paih M, Bouithy N, Pujades-Rodriguez M. Vital status of pre-ART and ART patients defaulting from care in rural Malawi. Trop Med Int Health. 2010; 15 Suppl 1: 55-62.

357. Mulissa Z, Jerene D, Lindtjorn B. Patients present earlier and survival has improved, but pre-ART attrition is high in a six-year HIV cohort data from Ethiopia. PLoS One. 2010; 5(10): e13268.

358. Naidoo N, Faal M, Venter W, Osih R. Patient retention - reasons why patients do or do not come back to care after HIV testing. XVIII International AIDS Conference. Vienna, Austria; 2010.

359. Nakanjako D, Kyabayinze DJ, Mayanja-Kizza H, Katabira E, Kamya MR. Eligibility for HIV/AIDS treatment among adults in a medical emergency setting at an urban hospital in Uganda. Afr Health Sci. 2007; 7(3): 124-8.

360. Nakigozi G, Makumbi F, Reynolds S, Galiwango R, Kagaayi J, Nalugoda F, Ssettuba A, Sekasanvu J, Musuuza J, Serwada D, Gray R, Wawer M. Non-enrollment for free community HIV care: findings from a population-based study in Rakai, Uganda. AIDS Care. 2011; 23(6): 764-70.

361. Nsigaye R, Wringe A, Roura M, Kalluvya S, Urassa M, Busza J, Zaba B. From HIV diagnosis to treatment: evaluation of a referral system to promote and monitor access to antiretroviral therapy in rural Tanzania. J Int AIDS Soc. 2009; 12(1): 31.

362. Nunu RP, Nkambule L, Kamiru H, Vandelanotte J, Preko P. Using phone follow-up system to understand barriers to ART initiation at Good Shepherd hospital in Swaziland. XVIII International AIDS Conference. Vienna, Austria; 2010.

363. Parks R, Namakoola I, Todd J, Kalanzi I, Hialaithe M. Barriers to rapid initiation of ART in a cohort of HIV positive Ugandan adults with CD4 counts less than 200. XVI International AIDS Conference. Toronto, Canada; 2006.

364. Tayler-Smith K, Zachariah R, Manzi M, Kizito W, Vandenbulcke A, Dunkley S, von Rege D, Reid T, Arnould L, Suleh A, Harries AD. Demographic characteristics and opportunistic diseases associated with attrition during preparation for antiretroviral therapy in primary health centres in Kibera, Kenya. Trop Med Int Health. 2011; 16(5): 579-84.

365. Tayler-Smith K, Zachariah R, Manzi M, Kizito W, Vandenbulcke A, Sitienei J, Chakaya J, Harries AD. Antiretroviral treatment uptake and attrition among HIV-positive patients with tuberculosis in Kibera, Kenya. Trop Med Int Health. 2011; 16(11): 1380-3.

366. Tayler-Smith K, Zachariah R, Massaquoi M, Manzi M, Pasulani O, van den Akker T, Bemelmans M, Bauernfeind A, Mwagomba B, Harries AD. Unacceptable attrition among WHO stages 1 and 2 patients in a

331

hospital-based setting in rural Malawi: can we retain such patients within the general health system? Trans R Soc Trop Med Hyg. 2010; 104(5): 313-9.

367. Togun T, Peterson I, Jaffar S, Oko F, Okomo U, Peterson K, Jaye A. Pre-treatment mortality and loss-to-follow-up in HIV-1, HIV-2 and HIV-1/HIV-2 dually infected patients eligible for antiretroviral therapy in The Gambia, West Africa. AIDS Res Ther. 2011; 8(1): 24.

368. van Schaik N, Kranzer K, Wood R, Bekker L-G. Earlier HIV diagnosis--are mobile services the answer? S Afr Med J. 2010; 100(10): 671-4.

369. Wanyenze RK, Bangsberg DR, Liechty CA, Nansubuga J, Kasakye H. Linkage to care and mortality at 18 month follow-up in a cohort of newly diagnosed HIV-positive inpatients in Mulago Hospital, Uganda. XVI International AIDS Conference. Toronto, Canada; 2006.

370. Wanyenze RK, Hahn JA, Liechty CA, Ragland K, Ronald A, Mayanja-Kizza H, Coates T, Kamya MR, Bangsberg DR. Linkage to HIV care and survival following inpatient HIV counseling and testing. AIDS Behav. 2011; 15(4): 751-60.

371. Waxman MJ, Kimaiyo S, Ongaro N, Wools-Kaloustian KK, Flanigan TP, Carter EJ. Initial outcomes of an emergency department rapid HIV testing program in western Kenya. AIDS Patient Care STDS. 2007; 21(12): 981-6.

372. Zachariah R, Harries AD, Manzi M, Gomani P, Teck R, Phillips M, Firmenich P. Acceptance of anti- retroviral therapy among patients infected with HIV and tuberculosis in rural Malawi is low and associated with cost of transport. PLoS One. 2006; 1: e121.

373. Loubiere S, Boyer S, Protopopescu C, Bonono CR, Abega SC, Spire B, Moatti JP. Decentralization of HIV care in Cameroon: increased access to antiretroviral treatment and associated persistent barriers. Health Policy. 2009; 92(2-3): 165-73.

374. Murphy RA, Sunpath H, Taha B, Kappagoda S, Maphasa KT, Kuritzkes DR, Smeaton L. Low uptake of antiretroviral therapy after admission with human immunodeficiency virus and tuberculosis in KwaZulu- Natal, South Africa. Int J Tuberc Lung Dis. 2010; 14(7): 903-8.

375. Ochieng-Ooko V, Ochieng D, Sidle JE, Holdsworth M, Wools-Kaloustian K, Siika AM, Yiannoutsos CT, Owiti M, Kimaiyo S, Braitstein P. Influence of gender on loss to follow-up in a large HIV treatment programme in western Kenya. Bull World Health Organ. 2010; 88(9): 681-8.

376. Pepper DJ, Marais S, Wilkinson RJ, Bhaijee F, De Azevedo V, Meintjes G. Barriers to initiation of antiretrovirals during antituberculosis therapy in Africa. PLoS One. 2011; 6(5): e19484.

377. Scott V, Zweigenthal V, Jennings K. Between HIV diagnosis and initiation of antiretroviral therapy: assessing the effectiveness of care for people living with HIV in the public primary care service in Cape Town, South Africa. Trop Med Int Health. 2011; 16(11): 1384-91.

378. Van der Borght SF, Clevenbergh P, Rijckborst H, Nsalou P, Onyia N, Lange JM, de Wit TF, Van der Loeff MF. Mortality and morbidity among HIV type-1-infected patients during the first 5 years of a multicountry HIV workplace programme in Africa. Antivir Ther. 2009; 14(1): 63-74.

379. Zachariah R, Tayler-Smith K, Manzi M, Massaquoi M, Mwagomba B, van Griensven J, van Engelgem I, Arnould L, Schouten EJ, Chimbwandira FM, Harries AD. Retention and attrition during the preparation phase and after start of antiretroviral treatment in Thyolo, Malawi, and Kibera, Kenya: implications for programmes? Trans R Soc Trop Med Hyg. 2011; 105(8): 421-30.

332

380. Wanyenze RK, Hahn JA, Liechty CA, Ragland K, Ronald A, Mayanja-Kizza H, Coates T, Kamya MR, Bangsberg DR. Linkage to HIV care and survival following inpatient HIV counseling and testing. AIDS Behav. 2011; 15(4): 751-60.

381. Govindasamy D, van Schaik N, Kranzer K, Wood R, Mathews C, Bekker LG. Linkage to HIV care from a mobile testing unit in South Africa by different CD4 count strata. J Acquir Immune Defic Syndr. 2011; 58(3): 344-52.

382. Lawn SD, Myer L, Harling G, Orrell C, Bekker L-G, Wood R. Determinants of mortality and nondeath losses from an antiretroviral treatment service in South Africa: implications for program evaluation. Clin Infect Dis. 2006; 43(6): 770-6.

383. Fox MP, Rosen S. Patient retention in antiretroviral therapy programs up to three years on treatment in sub-Saharan Africa, 2007-2009: systematic review. Trop Med Int Health. 2010; 15 Suppl 1: 1- 15.

384. Squire SB, Belaye AK, Kashoti A, Salaniponi FM, Mundy CJ, Theobald S, Kemp J. 'Lost' smear- positive pulmonary tuberculosis cases: where are they and why did we lose them? Int J Tuberc Lung Dis. 2005; 9(1): 25-31.

385. Waitt CJ, Squire SB. A systematic review of risk factors for death in adults during and after tuberculosis treatment. Int J Tuberc Lung Dis. 2011; 15(7): 871-85.

386. Govindasamy D, Ford N, Kranzer K. Risk factors, barriers and facilitators for linkage to art care in sub-saharan africa: a systematic review. AIDS. 2012; 26(16): 2069-67.

387. Ware NC, Wyatt MA, Geng EH, Kaaya SF, Agbaji OO, Muyindike WR, Chalamilla G, Agaba PA. Toward an understanding of disengagement from HIV treatment and care in sub-Saharan Africa: a qualitative study. PLoS Med. 2013; 10(1): e1001369; discussion e.

388. Parrott FR, Mwafulirwa C, Ngwira B, Nkhwazi S, Floyd S, Houben RM, Glynn JR, Crampin AC, French N. Combining qualitative and quantitative evidence to determine factors leading to late presentation for antiretroviral therapy in Malawi. PLoS One. 2011; 6(11): e27917.

389. Murray LK, Semrau K, McCurley E, Thea DM, Scott N, Mwiya M, Kankasa C, Bass J, Bolton P. Barriers to acceptance and adherence of antiretroviral therapy in urban Zambian women: a qualitative study. AIDS Care. 2009; 21(1): 78-86.

390. Unge C, Johansson A, Zachariah R, Some D, Van Engelgem I, Ekstrom AM. Reasons for unsatisfactory acceptance of antiretroviral treatment in the urban Kibera slum, Kenya. AIDS Care. 2008; 20(2): 146-9.

391. Mavhu W, Dauya E, Bandason T, Munyati S, Cowan FM, Hart G, Corbett EL, Chikovore J. Chronic cough and its association with TB-HIV co-infection: factors affecting help-seeking behaviour in Harare, Zimbabwe. Trop Med Int Health. 2010; 15(5): 574-9.

392. Assefa Y, Kiflie A, Tekle B, Mariam DH, Laga M, Van Damme W. Effectiveness and acceptability of delivery of antiretroviral treatment in health centres by health officers and nurses in Ethiopia. J Health Serv Res Policy. 2012; 17(1): 24-9.

393. Bedelu M, Ford N, Hilderbrand K, Reuter H. Implementing antiretroviral therapy in rural communities: the Lusikisiki model of decentralized HIV/AIDS care. J Infect Dis. 2007; 196 Suppl 3: S464-8.

333

394. Bemelmans M, Van Den Akker T, Ford N, Philips M, Zachariah R, Harries A, Schouten E, Hermann K, Mwagomba B, Massaquoi M. Providing universal access to antiretroviral therapy in Thyolo, Malawi through task shifting and decentralization of HIV/AIDS care. Trop Med Int Health. 2010.

395. Chamie G, Kwarisiima D, Kabami J, Clark T, Jain V, Black E, Geng E, Charlebois E, Kamya M, Havlir D. Outcomes in a Routine Linkage-to-care Strategy and an Enhanced Strategy with Accelerated ART Start; Community-based HIV Testing and Point-of-care CD4: Rural Uganda. 19th Conference on Retroviruses and Opportunistic Infections. Seatlle USA; 2012.

396. Cohen R, Lynch S, Bygrave H, Eggers E, Vlahakis N, Hilderbrand K, Knight L, Pillay P, Saranchuk P, Goemaere E, Makakole L, Ford N. Antiretroviral treatment outcomes from a nurse-driven, community- supported HIV/AIDS treatment programme in rural Lesotho: observational cohort assessment at two years. Journal of the International AIDS Society. 2009; 12: 23.

397. Faal M, Naidoo N, Glencross DK, Venter WD, Osih R. Providing immediate CD4 count results at HIV testing improves ART initiation. J Acquir Immune Defic Syndr. 2011; 58(3): e54-9.

398. Fredlund VG, Nash J. How far should they walk? Increasing antiretroviral therapy access in a rural community in northern KwaZulu-Natal, South Africa. J Infect Dis. 2007; 196 Suppl 3: S469-73.

399. Hatcher AM, Turan JM, Leslie HH, Kanya LW, Kwena Z, Johnson MO, Shade SB, Bukusi EA, Doyen A, Cohen CR. Predictors of linkage to care following community-based HIV counseling and testing in rural Kenya. AIDS Behav. 2012; 16(5): 1295-307.

400. Mukherjee JS, Ivers L, Leandre F, Farmer P, Behforouz H. Antiretroviral therapy in resource-poor settings. Decreasing barriers to access and promoting adherence. J Acquir Immune Defic Syndr. 2006; 43 Suppl 1: S123-6.

401. Sherr K, Pfeiffer J, Mussa A, Vio F, Gimbel S, Micek M, Gloyd S. The role of nonphysician clinicians in the rapid expansion of HIV care in Mozambique. J Acquir Immune Defic Syndr. 2009; 52 Suppl 1: S20-3.

402. Siedner MJ, Lankowski A, Haberer JE, Kembabazi A, Emenyonu N, Tsai AC, Muzoora C, Geng E, Martin JN, Bangsberg DR. Rethinking the "pre" in pre-therapy counseling: no benefit of additional visits prior to therapy on adherence or viremia in Ugandans initiating ARVs. PLoS One. 2012; 7(6): e39894.

403. Solomon S, Srikrishnan S, Vasudevan C, Hari Ramesh D, Anand S, Kumar M, Mehta S, Solomon S, Lucas G. The Impact of Voucher Incentives on Linkage to Care and ART Initiation among Drug Users: Chennai, India. 19th Conference on Retroviruses and Opportunistic Infections. Seattle, USA; 2012.

404. van Rooyen H, Barnabas R, Tumwesigye E, Humphries H, Tumwebaze H, Jospeh P, Krows M, Murnane P, Baeten J, Celum C. Characteristics and Testing Uptake of Heterosexual Couples Identified through Home-based Testing and Facilitated Referral: South Africa and Uganda. 20th Conference on Retroviruses and Opportunistic Infections. Atlanta, USA; 2013.

405. Zwarenstein M, Fairall LR, Lombard C, Mayers P, Bheekie A, English RG, Lewin S, Bachmann MO, Bateman E. Outreach education for integration of HIV/AIDS care, antiretroviral treatment, and tuberculosis care in primary care clinics in South Africa: PALSA PLUS pragmatic cluster randomised trial. BMJ. 2011; 342: d2022.

406. Larson BA, Schnippel K, Ndibongo B, Xulu T, Brennan A, Long L, Fox MP, Rosen S. Rapid point-of- care CD4 testing at mobile HIV testing sites to increase linkage to care: an evaluation of a pilot program in South Africa. J Acquir Immune Defic Syndr. 2012; 61(2): e13-7.

334

407. Myer L, Manuelli V, Abrams E, McIntyre J, Bekker L-G. Optimization of ART Initiation in Pregnancy through Linkage of Services vs Integration of ART into Antenatal Care. 20th Conference on Retroviruses and Opportunistic Infections. Atlanta, USA; 2013.

408. Schull MJ, Cornick R, Thompson S, Faris G, Fairall L, Burciul B, Sodhi S, Draper B, Joshua M, Mondiwa M, Banda H, Kathyola D, Bateman E, Zwarenstein M. From PALSA PLUS to PALM PLUS: adapting and developing a South African guideline and training intervention to better integrate HIV/AIDS care with primary care in rural health centers in Malawi. Implementation science : IS. 2011; 6: 82.

409. Jaffar S, Amuron B, Foster S, Birungi J, Levin J, Namara G. Rates of virological failure in patients treated in a home-based versus a facility-based HIV-care model in Jinja, southeast Uganda: a cluster- randomised equivalence trial. Lancet. 2009; 374(9707): 2080-9.

410. Hayden J. Introduction to health behavior theory. Sudbury, Mass.: Jones and Bartlett; 2009.

411. World Health Organization Commission on the Social Determinants of Health. Closing the gap in a generation: Health equity through action on the social determinants of health. 2011 [cited; Available from: http://www.who.int/social_determinants/thecommission/finalreport/en/index.html

412. Marmot M, Friel S, Bell R, Houweling TA, Taylor S. Closing the gap in a generation: health equity through action on the social determinants of health. Lancet. 2008; 372(9650): 1661-9.

413. Alsan MM, Westerhaus M, Herce M, Nakashima K, Farmer PE. Poverty, global health, and infectious disease: lessons from haiti and rwanda. Infect Dis Clin North Am. 2011; 25(3): 611-22.

414. Farmer P. Infections and inequalities : the modern plagues. Berkeley, Calif.: University of California Press; 2001.

415. World Bank. A sourcebook for poverty reduction strategies. 2003 [cited; Available from: http://web.worldbank.org/WBSITE/EXTERNAL/TOPICS/EXTPOVERTY/EXTPRS/0,,contentMDK:22404376~pag ePK:210058~piPK:210062~theSitePK:384201~isCURL:Y,00.html

416. National Statistics Office of Malawi. Integrated Household Survey. 1998 [cited; Available from: http://www.nso.malawi.net/index.php?option=com_content&view=article&id=74%3Asummary-report- 1998-integrated-household-survey&catid=2&Itemid=3

417. Payongayong E, Benson T, Ahmed A, Kanyanda S, Chakanza C, Mwanza P, Chilopa K, Banda N, Mphonda A. Simple household poverty assessment models for Malawi. Proxy means tests from the 1997-98 Malawi Integrated Household Survey. 2002 [cited; Available from: http://www.ifpri.org/sites/default/files/publications/malawiatlas.pdf

418. Mann GH. To what extent can the rural poor access free tuberculosis services in Malawi (Doctoral Thesis): Liverpool School of Tropical Medicine; 2008.

419. Simwaka B, Benson T, Salaniponi F, Theobald S, Squire SB, Kemp J. Developing a socio-economic measure to monitor access to tuberculosis services in urban Lilongwe, Malawi. Int J Tuberc Lung Dis. 2007; 11(1): 65-71.

420. UNAIDS. World AIDS Day Report. 2011 [cited; Available from: http://www.unaids.org/en/media/unaids/contentassets/documents/unaidspublication/2011/JC2216_World AIDSday_report_2011_en.pdf

421. Ministry of Health. Quarterly HIV Programme Report 2011, Q2; 2011.

335

422. Corbett EL, Bandason T, Duong T, Dauya E, Makamure B, Churchyard GJ, Williams BG, Munyati SS, Butterworth AE, Mason PR, Mungofa S, Hayes RJ. Comparison of two active case-finding strategies for community-based diagnosis of symptomatic smear-positive tuberculosis and control of infectious tuberculosis in Harare, Zimbabwe (DETECTB): a cluster-randomised trial. Lancet. 2010; 376(9748): 1244-53.

423. Shekalaghe SA, Drakeley C, van den Bosch S, ter Braak R, van den Bijllaardt W, Mwanziva C, Semvua S, Masokoto A, Mosha F, Teelen K, Hermsen R, Okell L, Gosling R, Sauerwein R, Bousema T. A cluster-randomized trial of mass drug administration with a gametocytocidal drug combination to interrupt malaria transmission in a low endemic area in Tanzania. Malar J. 2011; 10: 247.

424. Floyd S, Molesworth A, Dube A, Banda E, Jahn A, Mwafulirwa C, Ngwira B, Branson K, Crampin AC, Zaba B, Glynn JR, French N. Population-level reduction in adult mortality after extension of free anti- retroviral therapy provision into rural areas in northern Malawi. PLoS One. 2010; 5(10): e13499.

425. Serwaa-Bonsu A, Herbst AJ, Reniers G, Ijaa W, Clark B, Kabudula C, Sankoh O. First experiences in the implementation of biometric technology to link data from Health and Demographic Surveillance Systems with health facility data. Glob Health Action. 2010; 3.

426. Reichenheim ME. Confidence intervals for the kappa statistic. Stata J. 2004; 4(4): 421.

427. Walter SD, Eliasziw M, Donner A. Sample size and optimal designs for reliability studies. Stat Med. 1998; 17(1): 101-10.

428. Ministry of Health. Guidelines for the use of antiretroviral therapy in Malawi. Third Edition. 2008 April [cited; Available from: http://www.hivunitmohmw.org/uploads/Main/Malawi%20Integrated%20Guidelines%20for%20Clinical%20 Management%20of%20HIV%202011%20First%20Edition.pdf

429. Crampin AC, Dube A, Mboma S, Price A, Chihana M, Jahn A, Baschieri A, Molesworth A, Mwaiyeghele E, Branson K, Floyd S, McGrath N, Fine PE, French N, Glynn JR, Zaba B. Profile: the Karonga Health and Demographic Surveillance System. Int J Epidemiol. 2012; 41(3): 676-85.

430. Fantahun M, Fottrell E, Berhane Y, Wall S, Hogberg U, Byass P. Assessing a new approach to verbal autopsy interpretation in a rural Ethiopian community: the InterVA model. Bull World Health Organ. 2006; 84(3): 204-10.

431. Byass P, Kahn K, Fottrell E, Collinson MA, Tollman SM. Moving from data on deaths to public health policy in Agincourt, South Africa: approaches to analysing and understanding verbal autopsy findings. PLoS Med. 2010; 7(8): e1000325.

432. Faden R, Kass N. HIV research, ethics and the developing world. Am J Public Health. 1998; 88: 560- 3.

433. Gruskin S, Ahmed S, Ferguson L. Provider-initiated HIV testing and counseling in health facilities-- what does this mean for the health and human rights of pregnant women? Dev World Bioeth. 2008; 8(1): 23- 32.

434. Roberts M, Reich M. Ethical analysis in public health. Lancet. 2002; 359(9311): 1055-9.

435. World Medical Association. Decleration of Helsinki - Ethical principles for medical research involving human subjects. 64th WMA General Assembly, Fotaleza, Brazil; 2013.

436. Edwards S, Braunholtz D, Lilford R, Stevens A. Ethical issues in the design and conduct of cluster randomised controlled trials. BMJ. 1999; 318(7195): 1407.

336

437. UNAIDS. Global Report on the HIV Epidemic. Gevena; 2013.

438. Bassett IV, Walensky RP. Integrating HIV screening into routine health care in resource-limited settings. Clin Infect Dis. 2010; 50 Suppl 3: S77-84.

439. Fetene NW, Feleke AD. Missed opportunities for earlier HIV testing and diagnosis at the health facilities of Dessie town, North East Ethiopia. BMC Public Health. 2010; 10: 362.

440. Topp SM, Chipukuma JM, Giganti M, Mwango LK, Chiko LM, Tambatamba-Chapula B, Wamulume CS, Reid S. Strengthening health systems at facility-level: feasibility of integrating antiretroviral therapy into primary health care services in lusaka, zambia. PLoS One. 2010; 5(7): e11522.

441. Chakaya JM, Mansoer JR, Scano F, Wambua N, L'Herminez R, Odhiambo J, Mohamed I, Kangangi J, Ombeka V, Akeche G, Adala S, Gitau S, Maina J, Kibias S, Langat B, Abdille N, Wako I, Kimuu P, Sitienei J. National scale-up of HIV testing and provision of HIV care to tuberculosis patients in Kenya. Int J Tuberc Lung Dis. 2008; 12(4): 424-9.

442. Zachariah R, Ford N, Maher D, Bissell K, Van den Bergh R, van den Boogaard W, Reid T, Castro KG, Draguez B, von Schreeb J, Chakaya J, Atun R, Lienhardt C, Enarson DA, Harries AD. Is operational research delivering the goods? The journey to success in low-income countries. Lancet Infect Dis. 2012; 12(5): 415-21.

443. Flick U. An introduction to qualitative research. Los Angeles, USA: Sage Publications; 2011.

444. Glanz K, Rimer BK, Lewis FM. Health behavior and health education : theory, research, and practice. San Francisco: Jossey-Bass; 2002.

445. Zachariah R, Van Engelgem I, Massaquoi M, Kocholla L, Manzi M, Suleh A, Phillips M, Borgdorff M. Payment for antiretroviral drugs is associated with a higher rate of patients lost to follow-up than those offered free-of-charge therapy in Nairobi, Kenya. Trans R Soc Trop Med Hyg. 2008; 102(3): 288-93.

446. Ritchie J, Lewis J. Qualitative research practice : a guide for social science students and researchers. London; Thousand Oaks, Calif.: Sage Publications; 2003.

447. Pope C, Ziebland S, Mays N. Qualitative research in health care: Analysing qualitative data. BMJ: British Medical Journal. 2000; 320(7227): 114.

448. Merriam SB. Qualitative research: A guide to design and implementation: John Wiley & Sons; 2009.

449. Roura M, Busza J, Wringe A, Mbata D, Urassa M, Zaba B. Barriers to sustaining antiretroviral treatment in Kisesa, Tanzania: a follow-up study to understand attrition from the antiretroviral program. AIDS Patient Care STDS. 2009; 23(3): 203-10.

450. MacPherson P, Webb EL, Choko AT, Desmond N, Chavula K, Napierala Mavedzenge S, Makombe SD, Chunda T, Squire SB, Corbett EL. Stigmatising attitudes among people offered home-based HIV testing and counselling in Blantyre, Malawi: construction and analysis of a stigma scale. PLoS One. 2011; 6(10): e26814.

451. Brown L, Macintyre K, Trujillo L. Interventions to reduce HIV/AIDS stigma: What have we learned? AIDS Educ Prev. 2003; 15(1): 49-69.

337

452. Tuller DM, Bangsberg DR, Senkungu J, Ware NC, Emenyonu N, Weiser SD. Transportation costs impede sustained adherence and access to HAART in a clinic population in southwestern Uganda: a qualitative study. AIDS Behav. 2010; 14(4): 778-84.

453. Kemp J, Mann GH, Nhlema Simwaka B, Salaniponi FL, Squire SB. Can Malawi's poor afford free tuberculosis sevices? Patient and household costs associated with a tuberculosis diagnosis in Lilongwe. Bull World Health Organ. 2007; 85(8): 580-5.

454. Pope DS, Atkins S, DeLuca AN, Hausler H, Hoosain E, Celentano DD, Chaisson RE. South African TB nurses' experiences of provider-initiated HIV counseling and testing in the Eastern Cape Province: a qualitative study. AIDS Care. 2010; 22(2): 238-45.

455. Campbell MK, Piaggio G, Elbourne DR, Altman DG. Consort 2010 statement: extension to cluster randomised trials. BMJ. 2012; 345: e5661.

456. Choko AT, Desmond N, Webb EL, Chavula K, Napierala-Mavedzenge S, Gaydos CA, Makombe SD, Chunda T, Squire SB, French N, Mwapasa V, Corbett EL. The uptake and accuracy of oral kits for HIV self- testing in high HIV prevalence setting: a cross-sectional feasibility study in Blantyre, Malawi. PLoS Med. 2011; 8(10): e1001102.

457. UNAIDS. Report on the global AIDS epidemic. Geneva; 2012.

458. Hayes RJ, Bennett S. Simple sample size calculation for cluster-randomized trials. Int J Epidemiol. 1999; 28(2): 319-26.

459. Kendall MG. Kendall's advanced theory of statistics. London; New York: Edward Arnold ; Halsted Press; 1994. p. 351-2.

460. Druyts E, Dybul M, Kanters S, Nachega J, Birungi J, Ford N, Thorlund K, Negin J, Lester R, Yaya S, Mills EJ. Male sex and the risk of mortality among individuals enrolled in antiretroviral therapy programs in Africa: a systematic review and meta-analysis. AIDS. 2013; 27(3): 417-25.

461. Jaffar S, Amuron B, Foster S, Birungi J, Levin J, Namara G, Nabiryo C, Ndembi N, Kyomuhangi R, Opio A, Bunnell R, Tappero JW, Mermin J, Coutinho A, Grosskurth H. Rates of virological failure in patients treated in a home-based versus a facility-based HIV-care model in Jinja, southeast Uganda: a cluster- randomised equivalence trial. Lancet. 2009; 374(9707): 2080-9.

462. National Statistics Office Malawi. Malawi Population and Housing Census 2008. Zomba, Malawi; 2008.

463. Chihana M, Floyd S, Molesworth A, Crampin AC, Kayuni N, Price A, Zaba B, Jahn A, Mvula H, Dube A, Ngwira B, Glynn JR, French N. Adult mortality and probable cause of death in rural northern Malawi in the era of HIV treatment. Trop Med Int Health. 2012; 17(8): e74-83.

464. Kalibala s, Tun W, Muraah W, Cherutich P, Oweya E, Oluoch P. Knowing myself first’: Feasibility of self-testing among health workers in Kenya. Population Council, Nairobi; 2011.

465. Linkage and Retention: A Randomized Trial to Optimize HIV/TB Care in South Africa (Sizanani) - NCT01188941. clinicaltrials.gov; In progress.

466. HIV Counseling and Testing and Linkage to Care in Uganda - NCT00648232. clinicaltrials.gov; In progress.

338

467. Testing and Linkage to Care for Injecting Drug Users in Kenya (TLC-IDU Kenya) - NCT01557998. In progress.

468. Multidisciplinary Services to Enhance HIV Testing and Linkage to Care Among MSM - NCT01637324. clinicaltrials.gov; In Progress.

469. Home-based Partner Education and Testing (HOPE) Study - NCT01784783. clinicaltrials.gov; In progress.

470. Assisted Partner Notification to Augment HIV Treatment and Prevention in Kenya (APS) - NCT01616420. clinicaltrials.gov; In progress.

471. Positive Change Agents Program-Tanzania (Evaluation) - NCT01693458. clinicaltrials.gov; In progress.

472. Impact of Immediate Versus WHO Recommendations Guided ART Initiation on HIV Incidence - Feasibility Phase (TasP) - NCT01509508. clinicaltrials.gov; In progress.

473. Rapid Initiation of Antiretroviral Therapy to Promote Early HIV/AIDS Treatment in South Africa (RapIT Study) - NCT01710397. clinicaltrials.gov; In progress.

474. Athan E, O'Brien DP, Legood R. Cost-effectiveness of routine and low-cost CD4 T-cell count compared with WHO clinical staging of HIV to guide initiation of antiretroviral therapy in resource-limited settings. AIDS. 2010; 24(12): 1887-95.

475. Zachariah R, Ford N, Philips M, Lynch S, Massaquoi M, Janssens V, Harries AD. Task shifting in HIV/AIDS: opportunities, challenges and proposed actions for sub-Saharan Africa. Trans R Soc Trop Med Hyg. 2009; 103(6): 549-58.

476. World Health Organization. Integrated management of adolescent and adult illness. Chronic HIV care with ARV therapy and prevention. Interim guidelines for health workers at health centres or district hospital outpatients clinics. 2007 [cited; Available from: http://www.who.int/3by5/publications/documents/imai/en/

477. Macpherson P, Lalloo DG, Choko AT, Mann GH, Squire SB, Mwale D, Manda E, Makombe SD, Desmond N, Heyderman R, Corbett EL. Suboptimal patterns of provider initiated HIV testing and counselling, antiretroviral therapy eligibility assessment and referral in primary health clinic attendees in Blantyre, Malawi. Trop Med Int Health. 2012; 17(4): 507-17.

478. Ferrand RA, Weiss HA, Nathoo K, Ndhlovu CE, Mungofa S, Munyati S, Bandason T, Gibb DM, Corbett EL. A primary care level algorithm for identifying HIV-infected adolescents in populations at high risk through mother-to-child transmission. Trop Med Int Health. 2010; 16(3): 349-55.

479. Lewis DK, Callaghan M, Phiri K, Chipwete J, Kublin JG, Borgstein E, Zijlstra EE. Prevalence and indicators of HIV and AIDS among adults admitted to medical and surgical wards in Blantyre, Malawi. Trans R Soc Trop Med Hyg. 2003; 97(1): 91-6.

480. Cedeno-Laurent F, Gomez-Flores M, Mendez N, Ancer-Rodriguez J, Bryant JL, Gaspari AA, Trujillo JR. New insights into HIV-1-primary skin disorders. Journal of the International AIDS Society. 2011; 14: 5.

481. Getahun H, Kittikraisak W, Heilig CM, Corbett EL, Ayles H, Cain KP, Grant AD, Churchyard GJ, Kimerling M, Shah S, Lawn SD, Wood R, Maartens G, Granich R, Date AA, Varma JK. Development of a standardized screening rule for tuberculosis in people living with HIV in resource-constrained settings: individual participant data meta-analysis of observational studies. PLoS Med. 2011; 8(1): e1000391.

339

482. The STARD Group. STARD Checklist - Standards for the reporting of diagnostic accuracy studies; 2008.

483. Obuchowski NA. Sample size calculations in studies of test accuracy. Stat Methods Med Res. 1998; 7(4): 371-92.

484. Ford N, Hirnschall G. Modelling the Strategic Use of Antiretroviral Therapy for the Treatment and Prevention of HIV. PLoS Med. 2013; 10(10): e1001535.

485. US Centers for Disease Control and Prevention. Impact of an innovative approach to prevent mother-to-child transmission of HIV--Malawi, July 2011-September 2012. MMWR Morb Mortal Wkly Rep. 2013; 62(8): 148-51.

486. Tanser F, Gijsbertsen B, Herbst K. Modelling and understanding primary health care accessibility and utilization in rural South Africa: an exploration using a geographical information system. Soc Sci Med. 2006; 63(3): 691-705.

487. Sabapathy K, Van den Bergh R, Fidler S, Hayes R, Ford N. Uptake of home-based voluntary HIV testing in sub-Saharan Africa: a systematic review and meta-analysis. PLoS Med. 2012; 9(12): e1001351.

488. Western Cape Provincial Government. Premier Zille Draws Cash Prize Winners of HIV Testing Campaign Lottery. Cape Town, South Africa; 2013.

489. Western Cape HIV test lottery winners announced. The Times (South Africa). 2013.

490. de Walque D, Gertler P, Bautista-Arredondo S, Kwan A, Vermeersch C, de Dieu Bizimana J, Bingawaho A, Condo J, The World Bank. Using Provider Performance Incentives to Increase HIV Testing and Counseling Services in Rwanda. Washington DC; 2013.

491. Taegtmeyer M, MacPherson P, Jones K, Hopkins M, Moorcroft J, Lalloo DG, Chawla A. Programmatic evaluation of a combined antigen and antibody test for rapid HIV diagnosis in a community and sexual health clinic screening programme. PLoS One. 2011; 6(11): e28019.

492. Rosenberg NE, Kamanga G, Phiri S, Nsona D, Pettifor A, Rutstein SE, Kamwendo D, Hoffman IF, Keating M, Brown LB, Ndalama B, Fiscus SA, Congdon S, Cohen MS, Miller WC. Detection of acute HIV infection: a field evaluation of the determine(R) HIV-1/2 Ag/Ab combo test. J Infect Dis. 2012; 205(4): 528- 34.

493. MacPherson P, Houben RM, Corbett EL, Glynn JR, Kranzer K. Pre-treatment loss to follow-up among TB patients in low-to-middle income and high burden countries: a systematic review and meta- analysis. Bull World Health Organ. 2013; In press.

494. Chang LW, Kagaayi J, Arem H, Nakigozi G, Ssempijja V, Serwadda D, Quinn TC, Gray RH, Bollinger RC, Reynolds SJ. Impact of a mHealth intervention for peer health workers on AIDS care in rural Uganda: a mixed methods evaluation of a cluster-randomized trial. AIDS Behav. 2011; 15(8): 1776-84.

340

Appendix 1: Research ethics committee approvals for cohort study and qualitative study

341

DrPeter MacPherson WellcomeTrust Clinical Research Fellow LST) LIVERPOOLSCHOOL MRes/PhDCandidate OF TROPICALMEDICINE WellcomeTrust Tropical Centre BlockE,70 Pembroke Place PembrokePlace, Universityof Liverpool Liverpool,L3 5QA, UK Tel:.44(0)1 51 705 31 00 Liverpool Fax:*44(0)151705 33/0

1693GF www.liv.ac.uk/lstm

Friday,08October 2010

DearDr Peter MacPherson

Re:Research Protocol (10.65) The IDEA Study: ldentifying Delays in EquitableAccess to ART

Thankyou for yourletter dated 27 September 20L0 responding to the pointsraised by the Research EthicsCommittee. The protocol now has formal ethical approval from the Chairof LSTMResearch EthicsCommittee.

Theapproval is for a fixedperiod of threeyears, renewable annually thereafter. The committee may suspendor withdrawethical approval at anytime if appropriate'

Approvalis conditional upon:

r Submissionof ethical approval from other ethics commiitees'

o Notificationof allamendments to the protocolfor approvalbefore implementation.

Notificationof whenthe projectactually starts.

Provisionof anannual update to theCommittee. Failure to do socould result in suspension of the studywithout further notice.

Reportingof allsevere unexpected Adverse Events to the Committee

Reportingof newinformation relevant to patientsafety to the Committee

Provisionof DataMonitoring Committee reports (if applicable) to the Committee

Failureto complywith these requirements will result in withdrawal of approval.The Committee wouldalso like to of the finalreport o studyis completed.

cc. ProfDavid Lalloo

Researchingand educating to savelives r\,\--

A CompanyLimited by Guarantee.Registered Number 83405, England and Wales.Registered Charity Number 222655.

342

Appendix 2: Research ethics committee approval for cluster-randomised trial

Peter MacPherson Liverpool School of Tropical Medicine Pembroke Place Liverpool L3 5QA

Tuesday, 17 May 2011

Dear Peter MacPherson

Re: Research Protocol (11.63) Home-based assessment and initiation of ART: A cluster-randomised trial in urban communities in Blantyre, Malawi

Thank you for your letter dated 3 May 2011 responding to the points raised by the Research Ethics Committee. The protocol now has formal ethical approval from the Chair of LSTM Research Ethics Committee.

The approval is for a fixed period of three years, renewable annually thereafter. The committee may suspend or withdraw ethical approval at any time if appropriate.

Approval is conditional upon:

Submission of ethical approval from other ethics committees.

Notification of all amendments to the protocol for approval before implementation.

Notification of when the project actually starts.

Provision of an annual update to the Committee. Failure to do so could result in suspension of the study without further notice.

Reporting of all severe unexpected Adverse Events to the Committee

Reporting of new information relevant to patient safety to the Committee

Provision of Data Monitoring Committee reports (if applicable) to the Committee

Failure to comply with these requirements will result in withdrawal of approval. The Committee would also like to receive copies of the final report once the study is completed.

Yours sincerely

Dr Anja Terlouw Acting - Chair, Research Ethics Committee

cc: David Lalloo

343

344

LONDON SCHOOL OF HYGIENE & TROPICAL MEDICINE

ETHICS COMMITTEE

APPROVAL FORM

Application number: 6050

Name of Principal Investigator Dr Liz Corbett

Faculty Infectious and Tropical Diseases

Head of Faculty Professor Simon Croft

Title: Home-based assessment and initiation of ART: A cluster- randomised trial in urban communities in Blantyre, Malawi

This application is approved by the Committee.

Chair of the Ethics Committee ......

Date ...... 22 September 2011

Approval is dependent on local ethical approval having been received.

Any subsequent changes to the application must be submitted to the Committee via an E2 amendment form.

345

346