Global Health Initiatives Building a healthier world

UN Sustainable Development Goal 3 Good Health and Wellbeing Global Health Initiatives Building a healthier world

Special global health article collection celebrating the 40th anniversary of the Declaration of Alma-Ata

Foreword 1 Achieving universal health coverage in an era of emerging global health threats Kieran Walsh, Lalitha Bhagavatheeswaran, Mitali Wroczynski, Elisa Roma

Research report 2 Socioeconomic inequalities in child vaccination in low/middle-income countries: what accounts for the differences? From Journal of Epidemiology and Community Health Mohammad Hajizadeh

RESEARCH 9 Risk factors and risk factor cascades for communicable disease outbreaks in complex ­humanitarian emergencies: a qualitative systematic review From BMJ Global Health Charlotte Christiane Hammer, Julii Brainard, Paul R Hunter 19 Evaluation of a programme for ‘Rapid Assessment of Febrile Travelers’ (RAFT): a clinic-based ­quality improvement initiative From BMJ Open Farah Jazuli, Terence Lynd, Jordan Mah, Michael Klowak, Dale Jechel, Stefanie Klowak, Howard Ovens, Sam Sabbah, Andrea K Boggild

Short research report 26 Prevalence and factors associated with the use of antibiotics in non-bloody diarrhoea in children under 5 years of age in sub-Saharan Africa From Archives of Disease in Childhood Asa Auta, Brian O Ogbonna, Emmanuel O Adewuyi, Davies Adeloye, Barry Strickland-Hodge

Short report 30 Infectious disease outbreaks: how online clinical decision support could help From BMJ Simulation & Technology Enhanced Learning Kieran Walsh

Analysis AND EDITORIALS 33 Emerging and re-emerging infectious disease threats in South Asia: status, vulnerability, ­preparedness, and outlook From The BMJ Ramanan Laxminarayan, Manish Kakkar, Peter Horby, Gathsaurie Neelika Malavige, Buddha Basnyat 38 Health workers are vital to sustainable development goals and universal health coverage From The BMJ Lara Fairall, Eric Bateman 40 Implementing One Health as an integrated approach to health in Rwanda From BMJ Global Health Thierry Nyatanyi, Michael Wilkes, Haley McDermott, Serge Nzietchueng, Isidore Gafarasi, Antoine Mudakikwa, Jean Felix Kinani, Joseph Rukelibuga, Jared Omolo, Denise Mupfasoni, Adeline Kabeja, Jose Nyamusore, ­Julius Nziza, Jean Leonard Hakizimana, Julius Kamugisha, Richard Nkunda, Robert Kibuuka, Etienne Rugigana, Paul Farmer, Philip Cotton, Agnes Binagwaho 45 Building resilient health systems: a proposal for a resilience index From The BMJ Margaret E Kruk, Emilia J Ling, Asaf Bitton, Melani Cammett, Karen Cavanaugh, Mickey Chopra, Fadi el-Jardali, Rose Jallah Macauley, Mwihaki Kimura Muraguri, Shiro Konuma, Robert Marten, Frederick Martineau, ­Michael Myers, Kumanan Rasanathan, Enrique Ruelas, Agnès Soucat, Anung Sugihantono, Heiko Warnken

C ommentary 51 Productive disruption: opportunities and challenges for innovation in infectious disease ­surveillance From The BMJ Caroline O. Buckee, Maria I E Cardenas, June Corpuz, Arpita Ghosh, Farhana Haque, Jahirul Karim, Ayesha S. Mahmud, Richard J Maude, Keitly Mensah, Nkengafac Villyen Motaze, Maria Nabaggala, Charlotte Jessica Eland Metcalf, Sedera Aurélien Mioramalala, Frank Mubiru, Corey M. Peak, Santanu Pramanik, Jean Marius Rakotondramanga, Eric Remera, Ipsita Sinha, Siv Sovannaroth, Andrew J Tatem, Win Zaw

©BMJ Publishing Group Ltd 2018 All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any other means, electronic, mechanical, photocopying, recoding, or otherwise, without prior permission, in writing, of BMJ.UN Sustainable Development Goal 3 Good Health and Wellbeing Foreword

Achieving universal health coverage in an era of emerging global health threats

he Alma-Ata Declaration of 1978 was a pivotal moment becoming pandemics.10 More specifically, a health system which in public health, stating primary health care would be provides comprehensive primary healthcare and integrates all essential to obtaining the goal of “Health for All” by the relevant interventions can help respond to these emerging threats year 2000.1 There was much enthusiasm around the as well as various coinciding health challenges, such as the rise in world to work towards “Health for All” through primary non-communicable diseases.9 Thealth care. But soon after this, the goal of comprehensive primary As we approach the 40th anniversary of the Declaration of Alma- health care was criticised as being idealistic, too broad and having Ata, we renew our own commitment to advancing healthcare an unrealistic timetable.2 This idea of comprehensive primary health worldwide by sharing knowledge and expertise to improve care, as promoted by the Declaration, was quickly overshadowed experiences and outcomes for patients and populations. Our by the idea of “selective primary health care” just a short time later. global health initiatives focus on health systems strengthening Despite criticism that selective primary health care neglects the through workforce development, research capacity building, root causes of disease, significant progress was made in tackling the and knowledge translation. In this supplement, we are delighted prevailing causes of morbidity and mortality in children.3 By 1989, to present articles which we hope will advance the global health the World Health Organization stated that over half of children in discourse and confirm our commitment to the vision of a healthier Low and Middle Income countries were receiving immunisation world. each year. In 2015, world leaders met at the United Nations in New Contributing authors: York, to commit to the Sustainable Development Goals (SDGs). The Lalitha Bhagavatheeswaran, BMJ Clinical Outreach and Engagement Manager; SDGs address interconnected global challenges related to poverty, Mitali Wroczynski, BMJ Head of Strategic Partnerships; Dr Elisa Roma, BMJ Partnerships and Programmes Manager inequality, climate, environmental degradation and justice.4 In Competing interests: The authors work for BMJ, which produces a range of edu- order to make progress towards SDG Goal 3 (ensure healthy lives cational and clinical decision support resources in infectious and non-infectious and promote well-being for all at all ages), countries must ensure diseases. that everyone has access to high quality, safe and resilient health References: For full references, see page 55 services. An integrated, people-centred approach is fundamental to health systems that can respond to the various health challenges and emerging threats of today’s world.5 These include urbanization, the double burden of communicable diseases and noncommunicable diseases, aging populations, rising health care costs, antimicrobial resistance, and new and emerging infectious disease outbreaks.5 Antimicrobial resistance is one of the most challenging public health threats of our time. In 2015, the WHO launched a Global Action Plan on antimicrobial resistance to address the overuse and misuse of antimicrobials.6 The emergence and spread of pathogens resistant to antimicrobials is occurring worldwide, compromising our ability to treat infectious diseases. Furthermore, there have been 15 serious zoonotic or vector-borne global outbreaks, both viral (e.g., Hanta, Ebola, Marburg, H5N1 avian influenza) and bacterial (e.g., E coli, plague, and anthrax) over the past two decades.7 Epidemics of the 21st century have the ability to spread faster and further across the national borders, making all countries vulnerable and damaging their social and economic environment.8 The influenza pandemic of 2009 reached all continents in less than nine weeks. The West African Ebola outbreak was critical in showing us how connected we are globally and brought attention to the importance of a One Health approach. One Health is the “collaborative effort of multiple disciplines—working locally, nationally, and globally—to attain optimal health for people, animals and [the] environment through policy, research, education, and practice”.9 Strong and resilient health systems can help countries better Dr Kieran Walsh detect and respond to diseases and prevent outbreaks from Clinical Director, BMJ

BMJ Global Health Initiatives 1 RESEARCH

Socioeconomic inequalities in child vaccination in low/middle-­ income countries: what accounts for the differences? Mohammad Hajizadeh

School of Health Administration, ABSTRACT status group were less likely to receive all the four Faculty of Health, Dalhousie Background core vaccines than their lower socioeconomic University, Halifax, Nova Scotia, Socioeconomic inequalities in child vaccination status counterparts. Meta-regression analyses Canada continue to be a global public health concern. suggested that, across countries, the concentration Correspondence to: Dr Mohammad Hajizadeh This study aimed to measure and identify factors of antenatal care visits among wealthier mothers School of Health Administration, associated with socioeconomic inequalities in full was positively associated with the concentration Faculty of Health, Dalhousie immunisation coverage against the four core vaccine- of vaccination coverage among wealthier children University, Halifax, NS B3H 4R2, Canada; m. hajizadeh@ dal. ca preventable diseases (ie, bacille Calmette-Guérin, (coefficient=0.606, 95% CI 0.301 to 0.911). Additional material is published diphtheria-tetanus-pertussis (three doses), polio Conclusions online only. To view please visit (three doses) and measles vaccines) in 46 low/ Pro-rich distribution of child vaccination in most the journal online (http://dx.doi. middle-income countries. org/10.1136/jech-2017- low/middle-income countries remains an important 210296). Methods public health policy concern. Policies aimed to Cite this as: Hajizadeh M. J The most recent nationally representative samples of improve antenatal care visits among mothers in lower Epidemiol Community Health children (aged 10–59 months, n=372 499) collected socioeconomic groups may mitigate socioeconomic 2018;72:719–725. through the Demographic Health Surveys were used inequalities in vaccination coverage in low/middle- doi:10.​1136/​jech-​2017-​210296 to measure vaccination rates. The concentration index income countries. Received: 28 November 2017 (C) was used to quantify socioeconomic inequalities Revised: 14 March 2018 in vaccination coverage. Furthermore, meta-regression Accepted: 15 March 2018 Introduction analyses were used to determine factors affecting Republished with permission In spite of the remarkable global improvement in socioeconomic inequalities in vaccination coverage from J Epidemiol Community routine vaccination coverage worldwide, the overall Health 2018;72:719–725. across countries. improvement is not equally distributed among the Results countries.1 Furthermore, a growing body of evidence Results suggested that immunisation coverage suggests that there exists a social gradient in child was pro-rich in most countries (median C=0.161, vaccination within countries. For example, children IQR 0.131). Gambia (C=−0.146, 95% CI −0.223 whose parents are well-educated, wealthy or living in to −0.069), Namibia (C=−0.093, 95% CI −0.145 urban areas have a higher probability to be immunised to −0.041) and Kyrgyz Republic (C=−0.227, 95% against vaccine-preventable diseases.2 3 Social CI −0.304 to −0.15) were the only countries where inequalities in vaccination uptake can potentially children who belong to higher socioeconomic hinder the global efforts to reduce the burden of disease in low/middle-income countries because children from What is already known on this subject socioeconomically disadvantaged backgrounds are generally at increased risk of contracting infectious There have been several international initiatives set out to improve child diseases.4 immunisation rates in low/middle-income countries over the last four decades. Incomplete or complete lack of vaccination remains These initiatives led to a significant increase in child vaccination coverage the cause of millions of preventable child deaths globally, which, in turn, reduced child mortality and morbidity due to infectious each year in low/middle-income countries; there diseases in several countries. Despite the remarkable global improvement is, therefore, a considerable amount of literature in child vaccination coverage, the improvement is not equally distributed devoted to factors that impact vaccination uptake. among the countries. There is also a growing body of evidence suggesting Although several studies (eg, refs 5–9) suggest that socioeconomic inequalities in child vaccination within low/middle-income socioeconomic inequalities are still a major barrier to countries. child immunisation in low/middle-income countries, What this study adds few studies (eg, ref 10) aimed to quantify socioeconomic inequalities in vaccination coverage using a summary Although the current studies indicated socioeconomic gradients in child measure of inequality such as the concentration index. vaccination in low/middle-income countries, few studies have summarised the Measuring and monitoring of socioeconomic-related magnitude of socioeconomic inequalities across countries. There are even fewer inequalities in child vaccination in low/middle-income studies that have examined the determinants of socioeconomic inequalities in countries play a major role in evaluating progress child vaccination across countries. This study provides a comprehensive analysis towards the targets made by national and international of socioeconomic inequalities in vaccination coverage in 46 low/middle-income programmes to improve child vaccination coverage. countries. The results suggested that pro-rich distribution of child vaccination Identifying factors that explain the concentration in most low/middle-income countries remains an important public health of incomplete child vaccination among lower policy concern. Policies aimed at improving antenatal care among mothers in socioeconomic status (SES) children can also provide lower socioeconomic groups may mitigate socioeconomic inequalities in child valuable information to introduce effective strategies vaccination in low/middle-income countries. and policies to address such inequalities.

2 BMJ Global Health Initiatives RESEARCH

Using the most recent nationally representative used. Maternal reports may suffer from recall bias; samples of children collected through the Demographic however, previous studies (eg, ref 16) showed that Health Surveys (DHS) and the concentration index maternal recall reports are valid when we compare approach (which accounts for inequality across child vaccination coverage across populations. the whole socioeconomic distribution), this study Socioeconomic inequalities in vaccination coverage aimed to measure socioeconomic inequalities in full among children were measured using a constructed immunisation coverage against the four core vaccine- wealth index (WI) for each household in the DHS preventable diseases (ie, bacille Calmette-Guérin surveys as a measure of child SES. Using a principal (BCG), diphtheria-tetanus-pertussis (DTP, three doses), components analysis (PCA) technique the DHS polio (three doses) and measles vaccines, see online uses information on selected household’s assets to supplementary table 1) in 46 low/middle-income construct the WI.17 Based on the extant literature countries. Additionally, meta-regression analyses (eg, refs 5 6), the effects of inequalities in income were performed to examine whether inequalities in and other proximate determinants of vaccination proximate determinants of child vaccination were uptake (which have been consistently collected in associated with the extent of socioeconomic inequality all DHS surveys) on socioeconomic inequalities in in vaccination across countries. vaccination coverage were assessed. This included socioeconomic inequalities in mother’s education, mother’s age 19 and below, higher number of children Methods under 5 in the household (3 and above) and antenatal Data care (ANC) coverage (a minimum of four prenatal The data for this study were obtained from the care visits), and income inequality as measured by Standard DHS from 46 low/middle-income countries Gini index. As the DHS does not collect information collected through the MEASURE DHS project over the on expenditure or income, the World Bank’s World period between 2010 and 2015. The DHS surveys are Development Indicators and Global Development nationally representative cross-sectional household Finance data set18 was used to obtain Gini index for surveys with large sample sizes, typically between each country in the study. 5000 and 30 000 households.11 Using a multistage 12 sampling procedure, the DHS collects reliable Statistical analysis and comparable information on a variety of health- The statistical analysis involved two steps: (1) 13 related topics such as vaccination status of children calculating socioeconomic inequalities in vaccination in the household. High response rates and national coverage and (2) performing meta-regression 14 coverage are key advantages of the DHS. To facilitate analyses to determine factors affecting socioeconomic comparability of surveys across time and countries inequalities in vaccination coverage. a similar set of questions is used.13 Data collection methods, validation and reliability assessment are Measuring socioeconomic inequalities 13 described elsewhere. Immunisation records of Socioeconomic inequality in vaccination coverage children aged 59 months and younger were drawn in each country was measured using the (relative) from the DHS surveys. To allow each child a follow-up concentration (C) index. The C index is measured period to receive all the four routinely recommended regarding the (relative) concentration curve, which core vaccines: BCG, DTP (three doses), polio (three plots the cumulative share of vaccination coverage, doses) and measles vaccines (10 months as per the on its y-axis, against the cumulative share of the 15 WHO recommended immunisation schedules for children, ranked in ascending order of SES (WI), on the four routine vaccines, see online supplementary its x-axis. If all children, ranked by WI, received an table 1), 82 694 live births that occurred less than 10 equal proportion of vaccination coverage, the curve months before the survey interview were excluded would coincide with the 45° line (ie, the line of ‘perfect from the analysis. The final sample contained 372 499 equality’). The C index is described as twice the area live births aged 10–59 months in 46 low/middle- between the 45° line and the concentration curve. The income countries between 2010 and 2015. Online C index ranges from −1 to +1, with zero representing supplementary table 2 presents the survey years, ‘perfect equality’.19 If a health variable is concentrated sample size and gross national income (GNI) per capita among the rich, multiplying the C index by 75 will (current US$) for the sampled countries. give us the fraction of health variable that needs to be transferred from the wealthier half to the poorer half to Measures achieve ‘perfect equality’.20 The outcome variable in this study is binary variable The C index can be computed using the ‘convenient indicating whether a child received all the four regression’ approach as follows21: routinely recommended core vaccines. Immunisation status of children was collected using vaccination (1) record cards provided by mothers or caregivers ‍ ‍ during the survey interview. In the absence of where ‍ ‍ is child ‍‍’s vaccination coverage status, vaccination cards, mothers’ or caregiver’s verbal μ is the mean of vaccination coverage for the total reports of children’s immunisation coverage were sample, ‍ ‍ is the fractional rank of child ‍‍ in the

BMJ Global Health Initiatives 3 RESEARCH

Fig 1 | Child vaccination coverage (%) against the four core vaccine-preventable diseases in 46 low/middle-income countries

distribution (‍ ‍ and ‍ ‍ for the poorest and wealthiest because fixed effects meta-regression assumes that the child, correspondingly), and ‍ ‍ indicates the variance covariates included in the model can explain all the of fractional rank. The ordinary least squares (OLS) heterogeneity in the outcome variable and thus is not 26 estimate of ‍ ‍ represents the C index and its SE provides often recommended. The inverse variances of the C the SE for the C index.22 As the outcome variable in index for vaccination coverage were used as weights this study is binary, the C index was normalised23 to in all univariate and multivariate meta-regression summarise wealth-related inequalities in vaccination analyses. All analyses were performed in V.13 of the coverage by multiplying by ‍ ‍. To obtain STATA software package (StataCorp, College Station, estimates that are representative of children (aged 10– Texas). 59 months) living in each country sampling weights were applied in the calculation of the C index. 95% CIs Results were used to assess statistical significance of the C Child vaccination coverage 24 index. A method suggested by Altman and Bland was Online supplementary table 3 and figure 1 present used to examine the significance of differences in the the percentage of children (aged 10–59 months) who C at 95% CI. received all the four core vaccines in 46 low/middle- income countries. Overall, 55.5% of children in the Meta-regression analyses sampled countries were immunised against the diseases. Univariate and multivariate meta-regression analyses There was a significant difference across the countries were performed to assess the determinants of in the vaccination uptake; for example, in Nepal, socioeconomic inequalities in vaccination coverage. Honduras and Armenia more than 85% of children The C index in vaccination coverage for each country received all the four vaccines, whereas this figure was was used as the dependent variable and socioeconomic less than 35% in Mali, Nigeria and Ethiopia. As reported inequalities in proximate determinants of vaccination in online supplementary table 3, up to 58.3% of children uptake (measured by the C index) and income in low-income countries completed immunisation for inequality (measured by the Gini index) were used the four main diseases; these figures were 52.9% and as explanatory variables. The number of covariates 68% in the sampled lower middle-income and upper in the model satisfies the ‘rule of thumb’ of having at middle-income countries, respectively. least 5–10 observations per predictor in multivariable As reported in online supplementary table 3, the regression analysis.25 The random effects models vaccination rates were generally similar for males and were used in the meta-regression analyses. This is females in all countries, except Gambia, Tanzania,

4 BMJ Global Health Initiatives RESEARCH

Table 1 | Socioeconomic inequalities in child vaccination coverage in 46 low/middle-income countries The Concentration index Country Code Total Urban Rural Urban-rural Low-income countries (LIC) Bangladesh BD 0.248 (0.179 to 0.318) 0.209 (0.118 to 0.3) 0.262 (0.185 to 0.34) −0.054 (−0.174 to 0.066) Benin BJ 0.091 (0.052 to 0.131) 0.082 (0.021 to 0.144) 0.08 (0.033 to 0.126) 0.003 (−0.075 to 0.08) Burkina Faso BF 0.134 (0.084 to 0.184) 0.05 (−0.032 to 0.132) 0.149 (0.097 to 0.201) −0.099 (−0.196 to −0.001) Burundi BI 0.031 (−0.02 to 0.083) 0.048 (−0.173 to 0.269) 0.027 (−0.027 to 0.081) 0.021 (−0.206 to 0.248) Cambodia KH 0.296 (0.236 to 0.357) 0.304 (0.163 to 0.444) 0.264 (0.198 to 0.33) 0.04 (−0.115 to 0.195) Comoros KM 0.232 (0.158 to 0.306) 0.111 (−0.013 to 0.236) 0.28 (0.195 to 0.365) −0.169 (−0.319 to −0.018) Congo Democratic Republic CD 0.19 (0.147 to 0.234) 0.238 (0.172 to 0.304) 0.153 (0.1 to 0.205) 0.085 (0.001 to 0.169) Ethiopia ET 0.22 (0.159 to 0.281) 0.41 (0.316 to 0.505) 0.115 (0.053 to 0.178) 0.295 (0.181 to 0.408) Gambia GM −0.146 (−0.223 to −0.069) −0.071 (−0.166 to 0.024) 0.087 (0.021 to 0.153) −0.158 (−0.273 to −0.043) Guinea GN 0.122 (0.063 to 0.181) 0.041 (−0.055 to 0.137) 0.104 (0.047 to 0.161) −0.063 (−0.175 to 0.049) Haiti HT 0.046 (−0.008 to 0.099) 0.069 (−0.003 to 0.14) 0.076 (0.009 to 0.142) −0.007 (−0.105 to 0.091) Liberia LR 0.191 (0.126 to 0.256) 0.127 (0.045 to 0.209) 0.21 (0.156 to 0.263) −0.082 (−0.18 to 0.015) Malawi MW 0.029 (−0.011 to 0.068) −0.004 (−0.088 to 0.081) 0.086 (0.051 to 0.121) −0.09 (−0.182 to 0.001) Mali ML 0.129 (0.076 to 0.181) 0.039 (−0.032 to 0.109) 0.112 (0.049 to 0.174) −0.073 (−0.168 to 0.021) Mozambique MZ 0.197 (0.147 to 0.246) 0.018 (−0.077 to 0.113) 0.197 (0.145 to 0.25) −0.179 (−0.288 to −0.071) Nepal NP 0.227 (0.116 to 0.338) 0.359 (0.221 to 0.498) 0.219 (0.094 to 0.343) 0.141 (−0.046 to 0.327) Niger NE 0.253 (0.217 to 0.288) 0.094 (0.021 to 0.168) 0.212 (0.174 to 0.25) −0.117 (−0.2 to −0.034) Rwanda RW 0.167 (0.103 to 0.23) 0.147 (−0.036 to 0.331) 0.107 (0.042 to 0.172) 0.041 (−0.154 to 0.235) Sierra Leone SL 0.167 (0.103 to 0.23) 0.147 (−0.036 to 0.331) 0.107 (0.042 to 0.172) 0.041 (−0.154 to 0.235) Tajikistan TJ 0.167 (0.103 to 0.23) 0.147 (−0.036 to 0.331) 0.107 (0.042 to 0.172) 0.041 (−0.154 to 0.235) Tanzania TZ 0.167 (0.103 to 0.23) 0.147 (−0.036 to 0.331) 0.107 (0.042 to 0.172) 0.041 (−0.154 to 0.235) Togo TG 0.023 (−0.037 to 0.082) 0.156 (0.082 to 0.23) −0.07 (−0.132 to −0.008) 0.226 (0.129 to 0.323) Uganda UG 0.066 (0.02 to 0.112) 0.031 (−0.072 to 0.134) 0.034 (−0.016 to 0.083) −0.003 (−0.117 to 0.112) Zimbabwe ZW 0.171 (0.116 to 0.226) 0.134 (0.05 to 0.218) 0.15 (0.088 to 0.212) −0.016 (−0.12 to 0.088) Lower middle-income countries (lMIC) Armenia AM −0.017 (−0.144 to 0.11) −0.009 (−0.171 to 0.153) −0.129 (−0.288 to 0.031) 0.12 (−0.108 to 0.347) Cameroon CM 0.273 (0.217 to 0.329) 0.156 (0.096 to 0.216) 0.26 (0.188 to 0.331) −0.103 (−0.197 to −0.01) Congo Brazzaville CG 0.146 (0.088 to 0.204) 0.036 (−0.045 to 0.118) 0.147 (0.081 to 0.213) −0.111 (−0.215 to −0.006) Cote d'Ivoire CI 0.154 (0.094 to 0.214) 0.158 (0.076 to 0.24) 0.038 (−0.031 to 0.107) 0.12 (0.012 to 0.227) Egypt EG 0.096 (0.03 to 0.163) 0.079 (0.001 to 0.156) 0.218 (0.084 to 0.351) −0.139 (−0.293 to 0.015) Ghana GH 0.122 (0.063 to 0.181) 0.041 (−0.055 to 0.137) 0.104 (0.047 to 0.161) −0.063 (−0.175 to 0.049) Honduras HN −0.026 (−0.08 to 0.027) 0.078 (−0.011 to 0.166) 0.013 (−0.06 to 0.085) 0.065 (−0.05 to 0.18) Indonesia ID 0.263 (0.23 to 0.296) 0.182 (0.135 to 0.229) 0.282 (0.238 to 0.325) −0.1 (−0.163 to −0.036) Kenya KE 0.191 (0.126 to 0.256) 0.127 (0.045 to 0.209) 0.21 (0.156 to 0.263) −0.082 (−0.18 to 0.015) Kyrgyz Republic KG −0.227 (−0.304 to −0.15) −0.239 (−0.357 to −0.121) −0.167 (−0.264 to −0.07) −0.072 (−0.225 to 0.081) Nigeria NG 0.547 (0.517 to 0.577) 0.295 (0.244 to 0.346) 0.564 (0.507 to 0.62) −0.269 (−0.345 to −0.193) Pakistan PK 0.384 (0.326 to 0.442) 0.207 (0.127 to 0.288) 0.431 (0.368 to 0.494) −0.223 (−0.326 to −0.121) Philippines PH 0.167 (0.103 to 0.23) 0.147 (−0.036 to 0.331) 0.107 (0.042 to 0.172) 0.041 (−0.154 to 0.235) Senegal SN 0.167 (0.103 to 0.23) 0.147 (−0.036 to 0.331) 0.107 (0.042 to 0.172) 0.041 (−0.154 to 0.235) Yemen YE 0.34 (0.299 to 0.381) 0.192 (0.126 to 0.259) 0.245 (0.19 to 0.3) −0.053 (−0.14 to 0.033) Zambia ZM 0.15 (0.113 to 0.187) 0.175 (0.112 to 0.237) 0.088 (0.048 to 0.128) 0.087 (0.013 to 0.161) Upper middle-income countries (uMIC) Colombia CO 0.036 (0.005 to 0.067) 0.015 (−0.021 to 0.052) 0.12 (0.069 to 0.172) −0.105 (−0.168 to −0.042) Dominican Republic DO 0.096 (0.03 to 0.163) 0.079 (0.001 to 0.156) 0.218 (0.084 to 0.351) −0.139 (−0.293 to 0.015) Gabon GA 0.024 (−0.049 to 0.097) 0.009 (−0.074 to 0.092) 0.215 (0.08 to 0.35) −0.206 (−0.364 to −0.047) Jordan JO 0.148 (−0.023 to 0.32) 0.097 (−0.115 to 0.308) 0.3 (0.098 to 0.502) −0.204 (−0.496 to 0.089) Namibia NA −0.093 (−0.145 to −0.041) −0.029 (−0.102 to 0.044) 0.033 (−0.03 to 0.096) −0.062 (−0.159 to 0.035) Peru PE −0.009 (−0.053 to 0.035) 0.021 (−0.028 to 0.07) −0.121 (−0.209 to −0.034) 0.142 (0.042 to 0.243) Median (IQR) 0.161 (0.131; 0.066 to 0.197) 0.154 (0.141; 0.056 to 0.197) 0.163 (0.171; 0.027 to 0.198) 0.004 (0.049; −0.023 to 0.026) 95% CIs in parentheses.

Honduras and Gabon, where vaccination uptakes were middle-income countries. The results suggested that slightly higher among men compared with women. The immunisation coverage is pro-rich in the majority results also indicated that child vaccination was higher of countries (median C=0.161, IQR 0.131). Gambia in urban areas compared with rural areas in most (C=−0.146, 95% CI −0.223 to −0.069), Namibia countries. The vaccination rates were significantly (C=−0.093, 95% CI −0.145 to −0.041) and Kyrgyz higher in rural than urban areas in Gambia, Namibia, Republic (C=−0.227, 95% CI −0.304 to −0.15) were Kyrgyz Republic, Malawi and Honduras. the only countries where children who belong to higher SES were less immunised compared with lower Socioeconomic inequalities in child vaccination SES counterparts. As shown in figure 2, socioeconomic coverage inequalities in vaccination coverage were higher in Table 1 reports the magnitude of socioeconomic countries such as Nigeria, Pakistan, Yemen, Cambodia, inequalities in immunisation coverage in 46 low/ Cameroon and Indonesia compared with the other

BMJ Global Health Initiatives 5 RESEARCH

the concentration of mother’s educational level (coefficient=0.361, 95% CI 0.091 to 0.63) and ANC coverage (coefficient=0.505, 95% CI 0.248 to 0.762) among wealthier children was positively associated with socioeconomic inequalities in vaccination coverage. In multivariate analysis, only the concentration of ANC use among wealthier women was associated with socioeconomic inequalities in vaccination coverage; a 10% increase in the magnitude of the C index for ANC use was associated with a 6% increase in the value of the C index for vaccination coverage. Figure 3 demonstrates bubble plots of the correlations between the C index for mother’s ANC use and the C index for vaccination coverage.

Discussion and conclusions Fig 2 | Socioeconomic inequalities in child vaccination coverage and gross national This study, similar to previous studies in low/middle- income (GNI) per capital (current US$) in 46 low/middle-income countries income countries (eg, refs 5–8), found pro-rich inequalities in immunisation rate in most countries countries. The results did not suggest any association under study. Gambia, Namibia and Kyrgyz Republic between socioeconomic inequalities in immunisation were the only countries where children who belong coverage and (log) GNI per capita (r(44)=−0.020, to higher SES households were less likely to be fully p=0.89). immunised against the four diseases than their lower The C index indicated persistent pro-rich inequalities SES counterparts. Lower vaccination coverage among in vaccination coverage in urban (median C=0.154, IQR higher versus lower SES children was also observed 0.141) and rural areas (median C=0.163, IQR 0.171) in in more developed countries (eg, ref 27). Studies27–29 most of the sampled countries. Peru and Togo were the have suggested several factors that may be associated two countries with the pro-poor distribution of child with a decrease in child vaccination coverage among immunisation in rural areas. While in countries such higher SES population in more developed countries. as Nigeria, Pakistan, Gabon and Jordan socioeconomic The significant reductions in the incidences of inequalities were significantly higher in rural compared vaccine-preventable diseases in recent decades may with urban areas, in countries such as Ethiopia, have changed how parents perceive child vaccines. Togo, Peru and Nepal socioeconomic inequalities in In other words, parents may perceive fewer benefits vaccination uptake were greater in urban compared associated with vaccines than the risks of rare adverse with rural areas. events following immunisation. Some scientifically unfounded claims about vaccines that link some Factors affecting socioeconomic inequalities in vaccines to problems such as autism, multiple sclerosis, child vaccination coverage sudden infant death syndrome and other problems Table 2 reports the meta-regression analysis results. may have also altered the perception of the usefulness According to the univariate meta-regression results, of vaccines in some developed countries. Although these factors may explain pro-rich inequalities in child vaccinations in more developed countries, additional investigations are required to further elucidate this issue in low/middle-income countries. The finding from univariate meta-regression demonstrated that the concentration of educational attainment among mothers who belong to higher SES household was positively associated with the pro-rich distribution of child immunisation. The impact of unequal distribution of maternal education on inequality in child vaccination uptake can be explained through human, social, cultural and empowerment routes. The human capital advantage of maternal education in higher compared with lower SES households can lead to greater receptivity to public health messages aimed at boosting child immunisation rates among higher SES household because highly Fig 3 | Bubble plot of the association between the C indices for child vaccination and educated mothers have a greater understanding about 30 antenatal care coverage (at least four visits). The size of the bubbles in the figure is the benefits of immunisation. Educated mothers proportional to the weight of each observation in the meta-regression analysis also have larger social networks (social capital) which

6 BMJ Global Health Initiatives RESEARCH

Table 2 | Meta-regression analysis results The C for child vaccination Explanatory variables Univariate meta-regression Multivariate meta-regression Coefficient P values Adjusted R2 (%) Coefficient P values Gini index −0.299 0.215 1.64 0.153 0.523 (−0.778 to 0.18) (−0.326 to 0.631) The C for mother's educational level (year) 0.361 0.01 12.95 0.19 0.154 (0.091 to 0.63) (−0.074 to 0.453) The C for mother's age at birth—less than 19 0.078 0.731 −1.52 0.231 0.279 (−0.377 to 0.534) (−0.194 to 0.656) The C for number of children (less than 5 years old) 3 and above 0.129 0.553 −2.21 0.3 0.192 (−0.306 to 0.564) (−0.157 to 0.757) The C for antenatal care coverage (at least four visits) 0.505 0.000 26.06 0.606 0.000 (0.248 to 0.762) (0.301 to 0.911) Adjusted R2 (%) 33.72 n 46 95% CIs in parentheses.

provide good health behaviour-related information and (AC) index38 was used to compute and identify key where to find available healthcare services. The social determinants of absolute socioeconomic inequalities capital of well-educated mothers can result in higher in vaccination coverage across the sampled countries. concentration of immunisation rates among children More information about the AC index can be found from wealthier households. Moreover, well-educated elsewhere.38 39 The results were qualitatively very mothers generally have socially valued general skills similar to those obtained using the C index to measure that provide them with a higher social status. This socioeconomic inequalities (see online supplementary cultural capital can result in better communications tables 4 and 5 and supplementary figures 1 and 2). between mothers and medical providers and thus The results of multivariate meta-regression analysis enables wealthier households to use more healthcare using the AC index suggested that, across countries, such as immunisation services.31 Higher educational the concentration of ANC and educational attainment attainment (as a means for empowering women) among among wealthier mothers was statistically significantly wealthier mothers also assists them to play an active positively associated with the concentration of role in public and in their households which enables vaccination coverage among wealthier children. them to insist on better healthcare (eg, vaccination Furthermore, similar to the results obtained from coverage) for their newborn babies.32 multivariate meta-regression analysis using the C Results of both univariate and multivariate meta- index, the concentration of the presence of three or regression analyses suggested that, across countries, more siblings (aged <5 years) in the household among the concentration of ANC among wealthier mothers the poor households had a positive effect on absolute was statistically significantly positively associated socioeconomic inequality in immunisation coverage. with the concentration of vaccination coverage among This latter result can be explained by the fact that the wealthier children. Previous studies (eg, refs 33 34) concentration of larger numbers of children in poor have documented the positive impact of ANC visits on households encompasses a greater competing demand child vaccination coverage. The positive impacts of for mothers’ limited time and resources resulting in pro-rich inequalities in ANC visits on the concentration lower rates of immunisation among the poor children.40 of child vaccination coverage can be attributed to the This study is subject to some limitations. First, similar fact that higher compared with lower SES mothers to other studies using DHS data sets to determine have more opportunities to receive messages about factors associated with child vaccination,6 maternal the benefits of child vaccination that encourage them recall was used to identify the child’s vaccination to use vaccination services for their newborns because status in the absence of vaccination record card. they use more ANC services. ANC visits establish Although the validity of relying on maternal recall to communication and build trust between healthcare obtain child’s vaccination status has been confirmed,16 providers and mothers, which, in turn, may affect it would be ideal to use a written record to obtain this mothers’ immunisation-seeking behaviours.35 By information to avoid potential recall bias. Second, as increasing the interaction between health personnel the DHS collects vaccination coverage of still living and mother, ANC care visits can address some of the children at the time of the survey interview, this study reasons behind low rate of vaccine uptake, including indicates socioeconomic inequalities in vaccination lack of information about immunisation benefits, fear coverage among children who were alive at the time of of side effects and unfamiliarity with place and time of the survey. Third, this study examined socioeconomic immunisation.36 inequalities in child vaccination at the national level. Methodological overviews of measuring and Since there may be different associations between monitoring inequalities in health recommend factors at the subnational level in different countries, reporting both absolute and relative measures to the generalisability of the findings should not be ensure appropriate tracking of health inequalities.37 extended to population groups within the country. Thus, a modified absolute (generalised) concentration Finally, this study focused only on wealth-related

BMJ Global Health Initiatives 7 RESEARCH

inequalities in child vaccination coverage and did 13 Rutstein SO, Rojas G. Guide to DHS Statistics: Demographic and Health Surveys Methodology. Calverton, NY: U.S. Agency for not examine other social inequalities in vaccination International Development: ORC Macro, 2006. coverage (eg, race/ethnic-related inequalities) that 14 Corsi DJ, Neuman M, Finlay JE, et al. Demographic and health surveys: may be present in the sampled countries. a profile. Int J Epidemiol 2012;41:1602–13. 15 World Health Organization. WHO recommendations for routine Caveat considered, this study suggested that pro- immunization - summary tables. 2016. http://www.​who.​int/​ rich distribution of child vaccination in most low/ immunization/policy/​ Immunization_​ routine_​ table2.​ pdf?​ ua=1​ (accessed 8 Sep 2016). middle-income countries remains an important 16 AbdelSalam HHM, Sokal MM. Accuracy of Parental Reporting of public health policy concern and policies aimed to Immunization. Clin Pediatr 2004;43:83–5. improve ANC among mothers in lower socioeconomic 17 Rutstein SO, Johnson K. The DHS wealth index. DHS Comparative Reports No. 6. ORC Macro, MEASURE DHS 2004. groups may mitigate socioeconomic inequalities in 18 World Bank. World development indicators database and global vaccination coverage in low/middle-income countries. development finance. Washington DC, 2016. http://​databank.​ worldbank.org/​ data/​ home.​ aspx.​ (accessed 5 Jul 2016). MH gratefully acknowledges MEASURE DHS for their permission to 19 World Bank. Quantitative techniques for health equity analysis: use the Demographic Health Surveys (DHS) data. MH thanks two concentration curve: World Bank, 2017. anonymous reviewers and the associate editor for their helpful 20 Koolman X, van Doorslaer E. On the interpretation of a concentration comments and suggestions. MH also acknowledges the comments index of inequality. Health Econ 2004;13:649–56. from participants in the Canadian Immunization Research Network 21 Kakwani N, Wagstaff A, van Doorslaer E. Socioeconomic inequalities and the Canadian Center for Vaccinology (CIRN/CCfV) Education Series in health: measurement, computation, and statistical inference. at Dalhousie University. J Econom 1997;77:87–103. 22 O’Donnell O, van Doorslaer E, Wagstaff A, et al. Analyzing health Contributors: MH is the single author of the paper. equity using household survey data - a guide to techniques and their Funding: The author has not declared a specific grant for this research implementation. Geneva: The World Bank, 2008. from any funding agency in the public, commercial or not-for-profit 23 Wagstaff A. The bounds of the concentration index when the variable sectors. of interest is binary, with an application to immunization inequality. Health Econ 2005;14:429–32. Competing interests: None declared. 24 Altman DG, Bland JM. Interaction revisited: the difference between Patient consent: Detail has been removed from this case two estimates. BMJ 2003;326:219. description/these case descriptions to ensure anonymity. The editors 25 Muller KE, Regression FBA. and ANOVA: an integrated approach using and reviewers have seen the detailed information available and SAS software. SAS Institute 2002. are satisfied that the information backs up the case the authors are 26 Thompson SG, Sharp SJ. Explaining heterogeneity in meta-analysis: a making. comparison of methods. Stat Med 1999;18:2693–708. 27 Barata RB, Ribeiro MC, de Moraes JC, et al. Socioeconomic Ethics approval: This study was not subject to ethical approval inequalities and vaccination coverage: results of an immunisation because it did not involve data concerned with human participants. coverage survey in 27 Brazilian capitals, 2007-2008. J Epidemiol The study used secondary data from Demographic Health Surveys Community Health 2012;66:934–41. (DHS, http://www.​dhsprogram.​com/). The DHS surveys are fully 28 Bardenheier B, Yusuf H, Schwartz B, et al. Are parental vaccine available upon request without restriction. safety concerns associated with receipt of measles-mumps-rubella, diphtheria and tetanus toxoids with acellular pertussis, or hepatitis B Provenance and peer review: Not commissioned; externally peer vaccines by children? Arch Pediatr Adolesc Med 2004;158:569–75. reviewed. 29 Matsumura T, Nakayama T, Okamoto S, et al. Measles vaccine © Article author(s) (or their employer(s) unless otherwise stated in coverage and factors related to uncompleted vaccination among the text of the article) 2018. All rights reserved. No commercial use is 18-month-old and 36-month-old children in Kyoto, Japan. BMC permitted unless otherwise expressly granted. Public Health 2005;5:59. 30 Glewwe P. Why does mother’s schooling raise child health in 1 World Bank. Health nutrition and population statistics. Washington, developing countries? Evidence from Morocco. J Hum Resour DC, 2018. https://datacatalog.​ worldbank.​ org/​ dataset/​ health-​ ​ 1999;34:124–59. nutrition-​and-​population-​statistics. (accessed 16 Feb 2018). 31 Gittelsohn J, Bentley ME, Pelto PJ. Listening to women talk about 2 Glatman-Freedman A, Nichols K. The effect of social determinants on their health - Issues and evidences from India. New Delhi: South Asia immunization programs. Hum Vaccin Immunother 2012;8:293–301. Books, 1994. 3 Berkley S, Chan M, Elias C, et al. Global vaccine action plan 32 Vikram K, Vanneman R, Desai S. Linkages between maternal 2011-2020: World Health Organization, Decades of Vaccines education and childhood immunization in India. Soc Sci Med Collaboration, 2012. 2012;75:331–9. 4 World Bank Working Group on Child Poverty. Better health for poor 33 Konstantyner T, Taddei JA, Rodrigues LC. Risk factors for incomplete children: a special report. Geneva, switzerland, 2002. vaccination in children less than 18 months of age attending 5 Shrivastwa N, Gillespie BW, Kolenic GE, et al. Predictors of the nurseries of day-care centres in Sao Paulo, Brazil. Vaccine Vaccination in India for Children Aged 12-36 Months. Am J Prev Med 2011;29:9298–302. 2015;49:S435–44. 34 Mohamud AN, Feleke A, Worku W, et al. Immunization coverage 6 Bondy JN, Thind A, Koval JJ, et al. Identifying the determinants of 12-23 months old children and associated factors in Jigjiga of childhood immunization in the Philippines. Vaccine District, Somali National Regional State, Ethiopia. BMC Public Health 2009;27:169–75. 2014;14:865. 7 Zere E, Kirigia JM, Duale S, et al. Inequities in maternal and child 35 Rowe R, Calnan M. Trust relations in health care—the new agenda. health outcomes and interventions in Ghana. BMC Public Health Eur J Public Health 2006;16:4–6. 2012;12:252. 36 Dixit P, Dwivedi LK, Ram F. Strategies to improve child immunization 8 Branco FL, Pereira TM, Delfino BM, et al. Socioeconomic inequalities via antenatal care visits in India: a propensity score matching are still a barrier to full child vaccine coverage in the Brazilian analysis. PLoS One 2013;8:e66175. Amazon: a cross-sectional study in Assis Brasil, Acre, Brazil. Int J 37 King NB, Harper S, Young ME. Use of relative and absolute effect Equity Health 2014;13:118. measures in reporting health inequalities: structured review. BMJ 9 Arsenault C, Harper S, Nandi A, et al. Monitoring equity in vaccination 2012;345:e5774. coverage: a systematic analysis of demographic and health surveys 38 Erreygers G. Correcting the concentration index. J Health Econ from 45 Gavi-supported countries. Vaccine 2017;35:951–9. 2009;28:504–15. 10 Lauridsen J, Pradhan J. Socio-economic inequality of immunization 39 O’Donnell O, O’Neill S, Van Ourti T, et al. conindex: Estimation of coverage in India. Health Econ Rev 2011;1:11. concentration indices. Stata J 2016;16:112. 11 The DHS Program. DHS Overview. 2016 http://dhsprogram.​ ​com/​ 40 Danis K, Georgakopoulou T, Stavrou T, et al. Socioeconomic factors What-We-​ Do/​ Survey-​ Types/​ DHS.​ cfm​ (accessed 17 Jan 2016). play a more important role in childhood vaccination coverage than 12 Demographic and Health Survey. Sampling manual. DHS-III. parental perceptions: a cross-sectional study in Greece. Vaccine Calverton, 1996. 2010;28:1861–9.

8 BMJ Global Health Initiatives RESEARCH

Risk factors and risk factor cascades for communicable disease outbreaks in complex humanitarian emergencies: a qualitative systematic review Charlotte Christiane Hammer, Julii Brainard, Paul R Hunter

Norwich Medical School, ABSTRACT cascades that are triggered by individual risk factors or University of East Anglia Background clusters of risk factors. Faculty of Medicine and Health Communicable diseases are a major concern Sciences, Norwich, UK Conclusions during complex humanitarian emergencies (CHEs). Correspondence to: CHEs pose a significant threat to public health. More Charlotte Christiane Hammer Descriptions of risk factors for outbreaks are often rigorous research on the risk of disease outbreaks c.​ hammer@​ uea.​ ac.​ uk​ non-specific and not easily generalisable to similar in CHEs is needed, from a practitioner and from an Additional material is published situations. This review attempts to capture relevant academic point of view. online only. To view please visit evidence and explore whether it is possible to better the journal online. generalise the role of risk factors and risk factor Cite this as: Hammer CC, Introduction Brainard J, Hunter PR Risk cascades these factors may form. Complex humanitarian emergencies (CHEs1) pose a factors and risk factor Methods cascades for communicable A systematic search of the key databases and significant threat to public health, often in settings disease outbreaks in complex that were already deprived before the disruptive event humanitarian emergencies: websites was conducted. Search terms included a qualitative systematic terms for CHEs (United Nations Office for the or events. While CHEs generally affect the health of review. BMJ Glob Health Coordination of Humanitarian Affairs definition) the affected population negatively, they especially 2018;3:e000647. doi:10.1136/ exacerbate the risk of communicable diseases bmjgh-2017-000647 and terms for communicable diseases. Due to the types of evidence found, a thematic synthesis was including diarrhoeal diseases, acute respiratory Handling editor Soumitra diseases, measles, meningitis, tuberculosis, HIV, viral Bhuyan conducted. haemorrhagic fevers, hepatitis E, trypanosomiasis and Results Received: 16 November 2017 leishmaniosis.2 3 Priorities that need to be addressed in Revised: 4 June 2018 26 articles met inclusion criteria. Key risk factors Accepted: 5 June 2018 include crowded conditions, forced displacement, a complex emergency include rapid assessment of the health status of the affected population, mass measles Republished with permission poor quality shelter, poor water, sanitation and from BMJ Glob Health hygiene, lack of healthcare facilities and lack of vaccination, implementation of water and sanitation 2018;3:e000647. adequate surveillance. Most identified risk factors measures, food supply and nutrition programmes, do not relate to specific diseases, or are specific to site planning, provision of shelter, non-food items a group of diseases such as diarrhoeal diseases and and basic medical services, control and prevention not to a particular disease within that group. Risk of communicable diseases and potential epidemics, factors are often listed in general terms but are poorly surveillance and alert, mobilisation of community evidenced, not contextualised and not considered health workers, and coordination with national and 3 with respect to interaction effects in individual international agencies. Several of these interventions publications. The high level of the inter-relatedness of rightly target communicable diseases, as during risk factors became evident, demonstrating risk factor complex emergencies up to three quarters of excess deaths are attributable to infections.4 Key questions While research in this field is growing, there is inadequate understanding of the risk factors associated What is already known? with communicable diseases in these situations.5 There Complex humanitarian emergencies pose significant risks to human health and is a strong need for a better evidence and understanding communicable diseases are one of the most pressing concerns during a complex of the risk of communicable diseases in CHEs to inform humanitarian emergency. control strategies and emergency surveillance, both Complex humanitarian emergencies exacerbate many important risk factors for of which are based on risk assessments that currently outbreaks of communicable diseases. lack a common risk framework. We conducted the first (to our knowledge) systematic review on risk factors What are the new findings? for communicable diseases in complex humanitarian While not necessarily triggering different risk factors than other emergencies, emergencies. complex humanitarian emergencies trigger more risk factor cascades with interactive CHEs, for our purposes, are defined as crises in a feedback loops and provide a conductive environment for communicable diseases. region or area in which no local coping capacity can What do the new findings imply? handle the situation due to a complete breakdown of state authority. The problems in complex emergencies Humanitarian interventions need to be aware of a wide variety of possible risk are diverse and a multiagency international response factors and to identify those most likely to trigger risk factor cascades. is necessary to address the situation. They usually While mass population displacement triggers most other risk factors in complex result from extensive inter-state or intra-state armed humanitarian emergencies, more research is also needed on entrapment crises, conflict, leading to ‘(e)xtensive loss of life, massive which become more likely with the changing nature of conflict. displacement of population, widespread damage

BMJ Global Health Initiatives 9 RESEARCH

to societies and economies’; ‘Need for large-scale, 2015 West Africa Ebola outbreak, the Plague outbreak multi-faceted humanitarian assistance’; ‘Hindrance in Madagascar, tsunamis,8 tropical storms and other or prevention of humanitarian assistance by political disasters associated with a natural hazard are not and military constraints’; ‘Significant security risks classified as CHEs under the UNOCHA definition and 1 for humanitarian relief workers in some areas’. Any therefore not eligible for inclusion in this systematic such situation requires a multifaceted international review. response, usually led by the United Nations (UN). No We only included emergencies after 1990 and complex emergency would be adequately addressed publications published on or after 1 January 1994. by the activation of only one of the humanitarian These dates were chosen to exclude emergencies clusters. In fact, in most complex emergencies, most before 1990, which were mainly influenced by the if not all clusters would be activated and many such Cold War and hence considerably different in their emergencies will happen in situations and countries nature. The first major CHE after the end of the Cold where multiple clusters are already active due to the War was Rwanda and with those dates we made sure to underlying conditions with the complex emergency include research on Rwanda but exclude research on exacerbating these conditions beyond the scope of an CHEs during the Cold War. ongoing UN country programme. We initially included all languages, but if no one in the research team could be found who understood the Methods language an article was published in, we would have The description of methods follows the Preferred excluded that article for practical reasons. Because Reporting Items for Systematic Reviews and Meta- all articles found were either in English, French or Analyses statement as far as applicable to qualitative Spanish, no articles were excluded due to language systematic reviews.6 No review protocol was published barriers. beforehand. Search strategy and data sources Inclusion criteria Our search strategy was developed in discussion For this review, we had to define three terms on which we could formulate clear inclusion criteria: (1) risk between the authors and based on previous experience factors, (2) communicable diseases and (3) CHEs. and extensive background reading. The search was In order to capture all risk factors and risk factor composed of terms for communicable diseases, mechanisms that might not have been labelled risk including specific diseases that have very often occurred factors or been mentioned as a side note, we decided to in previous CHEs and terms for CHEs. We searched the not include terms for risk factors in our search strategy. following bibliographic databases: Scopus, Medline, However, they were applied as an inclusion criterion. Embase and International Bibliography of Social Risk factors for this purpose were anything mentioned Sciences (IBSS). The search strategy for Medline as increasing the risk of a communicable disease is presented in figure . 1 Search terms for Medline outbreak happening or as a reason for an outbreak and Embase included subject headings that were having happened or as a mechanism that promoted not available in Scopus and IBSS. The search was favourable conditions for communicable disease conducted in May 2017. Additionally, we searched spread in CHEs. Only those risk factors that apply at the relevant websites of Medecins Sans Frontièrs, the population or setting level were included, as this WHO and the United Nations High Commissioner for review does not focus on the individual. Risk factors Refugees, the United Nations Children and Education were eligible for inclusion if they could plausibly apply Fund and ReliefWeb (UNOCHA). The search strategy in CHEs. was adapted for the individual websites according to Communicable diseases were defined as infectious the technical and search engine capacities provided by diseases transmissible ‘by direct contact with an the websites. All terms were searched in abstracts and affected individual or the individual’s discharges or by titles, keywords and relevant subject where possible. indirect means (as by a vector)’.7 References of included publications were also checked. Definitions for CHEs, sometimes also simply called Reviews were included. complex emergencies, are plentiful; however, as most agencies involved in the management of this type of Study selection disaster agree on some key issues, we used the United Nations Office for the Coordination of Humanitarian Based on the inclusion criteria, CCH and JB screened Affairs (UNOCHA) definition: “(M)ultifacetedtitles and abstracts of all articles identified via humanitarian crisis in a country, region or society bibliographic databases independently. In case of where there is a total or considerable breakdown of disagreement, full text was obtained. An article authority resulting from internal or external conflict was included for full-text review if either screener and which requires a multi-sectoral, international did not reject it. CCH and JB next screened full texts response that goes beyond the mandate or capacity of independently and decision about final inclusion was any agency and/or the ongoing United Nations country reached discursively. We sought access via libraries programme”.1 As such, emergencies such as the 2013– and contacted authors of conference abstracts directly.

10 BMJ Global Health Initiatives RESEARCH

Fig 1 | Search strategy in Medline

Data analysis and synthesis they gave no indications of any risk factors. Twenty- Due to the qualitative and heterogeneous nature of the two articles were included directly from searches with evidence found, this is a qualitative systematic review. an additional four articles retrieved from the reference The data were analysed using thematic synthesis.9 lists of included articles. Articles were predominantly Primary coding was done by CCH, except for one article in English. One article was in Spanish and one in in Spanish, which was primary coded by JB. JB or CCH French. confirmed the primary codes and added secondary Twelve main clusters of risk factors were identified codes for all articles. Coding was done by hand and that all exhibit a high level of inter-relatedness, codes were transcribed into custom-made coding feedback loops and interaction on various levels. sheets, recording quotes, codes and subcodes. Based These risk factor clusters provide an analytical lens on the codes and subcodes, descriptive and analytical and many individual risk factors can be grouped into themes were developed. primary and secondary (and sometimes even tertiary) clusters. Table 1 gives an overview of the included articles, the setting they describe and the risk factor Results and discussion clusters identified in them. Our literature search retrieved 153 articles after de- duplication and eight grey-literature documents (as Main risk factor clusters shown in figure 2). Articles were mainly excluded if • WASH2 10–23: Water, sanitation and hygiene are they did not focus on CHEs or applied a significantly central elements to limit the risk of communicable different definition of CHEs than this review does, if diseases in populations experiencing an emergency. they did not focus on communicable diseases and if As such, they are also central to CHEs and often in

BMJ Global Health Initiatives 11 RESEARCH

Fig 2 | Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram. IBSS, International Bibliography of Social Sciences

a more precarious state than in other emergencies. potentially increased contact of naive populations WASH risk factors include issues such as lack with new disease vectors. Early camp structures 2 10 12 14–17 19–21 of safe drinking water, lack of (such as layout of tents and siting of toileting areas) 10 15 19 22 18 21 22 hygiene, hygiene behaviour, lack of can lead to further complications. Early layout often 2 19–21 24 25 20 soap, lack of bed nets (as vector control develops as an ad hoc response to mass population is usually seen as a part of WASH in humanitarian displacement but may prove completely unsuitable response) and general water scarcity,2 10 12 14–17 19–21 as the camp expands. as well as lack of adequate sanitation and latrines. • Nutrition2 10 12 13 15 17 19 20 22–24 34: While nutrition These factors considerably increase the risk for 2 10 13 15 17 19 20 22 24 34 diarrhoeal diseases and compound risks for other factors such as malnutrition, food 2 10–12 17 19 types of communicable diseases especially if they shortages and exposure to contaminated 19 20 are coupled with other risk factor categories such as food are mainly risk factors at the individual overcrowding and mass population displacement. level, they also pose increased risk to populations as • Overcrowding2 10 13 15 17–20 22–24: Overcrowding in a whole if a sufficient percentage of the population CHEs is usually a function of either mass population is exposed. Nutrition factors are related to increased displacement or entrapment. While overcrowding susceptibility to communicable diseases with can also be an issue in ad hoc shelters after the resulting greater shedding and transmission widespread destruction of homes and infrastructure, to others. At the population level, nutritional it is more prevalent if populations are forced to factors can exacerbate other risk factors and risk become refugees or internally displaced persons factor clusters, for example by increasing the risk and are forced into camps. Overcrowding affects of violence and social unrest. Root causes for both hygiene-related diseases, such as diarrhoeal nutrition risk factors lie mainly in other risk factor diseases, but also increases the transmission rate of clusters such as insecurity and armed conflict or diseases such as measles and other infections that mass displacement and inadequate humanitarian spread from person to person. response. 2 12 19 20 23 • Mass population displacement2 10 12 14 15 17–20 23 24 26–34: • Living conditions : Poor living conditions Mass population displacement is a trigger for most are a combination of inadequate shelter, risk factor categories and as such possibly the main overcrowding and other individual factors in the immediate surroundings of an individual or group risk factor in CHEs. Mass population displacement of individuals. A key risk for people uprooted from is usually associated with large numbers of people their normal lives in CHEs and subject to inadequate moving into camp settings, often associated resources and shelter is indoor air pollution.2 19 20 with overcrowding, inadequate shelter and poor This is due to indoor fires, both for cooking purposes 2 10 15 17–20 29 WASH conditions. Additionally, and for heating.2 19 20 populations are displaced into regions and areas • Insecurity2 10 14 19 23 28 30 31 33 35: Insecurity is a with insufficient resources and services and with multifaceted bundle of risk factors that is one of

12 BMJ Global Health Initiatives RESEARCH

Table 1 | List of articles included in the analysis Article Setting Risk factor clusters Abubakar et al22 South Sudan; Internally Displaced Persons (IDPs) camps Infrastructure, economy, mass population displacement, nutrition, overcrowding, water, sanitation and hygiene (WASH) Bompangue et al26 Democratic Republic of Congo; mainly refugee camps Humanitarian response, mass population displacement Brennan and Nandy10 Complex emergencies Health and public health services, HIV-specific risk factors, humanitarian re- sponse, insecurity, mass population displacement, nutrition, overcrowding, WASH Burkle18 Complex emergencies Infrastructure, mass population displacement, overcrowding, living conditions, WASH Burkle24 Complex emergencies; paediatric populations Economy, health and public health services, mass population displacement, nutrition, overcrowding, WASH Chaignat and Monti12 Complex emergencies Environment, health and public health services, humanitarian response, living conditions, mass displacement, nutrition, WASH Close et al13 Complex emergencies Nutrition, overcrowding, mass population displacement, health and public health services, WASH Connolly et al2 Complex emergencies Economy, environment, health and public health services, HIV-specific risk fac- tors, infrastructure, insecurity, mass displacement, living conditions, overcrowd- ing, nutrition, WASH Coulombier et al14 Complex emergencies Health and public health services, insecurity, mass population displacement, WASH Cuadrado and Gonzalez23 Complex emergencies Environment, WASH, insecurity, mass population displacement, nutrition, overcrowding, health and public health services, living conditions, economy, infrastructure Fisher et al15 Complex emergencies Environment, health and public health services, HIV-specific risk factors, mass population displacement, overcrowding, living conditions, nutrition, WASH Goma Epidemiology Group (1995) Rwanda; refugee camps Environment, WASH Guthmann et al16 Sudan; IDPs WASH Howard et al27 Afghanistan Economy, mass population displacement, health and public health services Howard et al25 Afghanistan Economy, infrastructure Khaw et al28 Complex emergencies Health and public health services, HIV-specific risk factors, insecurity, mass population displacement Kolaczinski (2005) Afghanistan Health and public health services Kolaczinski et al (2005) Afghanistan Insecurity, health and public health services Kolaczinski and Webster (2003) East Timor Health and public health services, mass population displacement, overcrowding, living conditions Leyenaar30 Complex emergencies Economy, HIV-specific risk factors, insecurity, mass displacement Liddle et al31 Somalia Economy, infrastructure, health and public health services, insecurity, mass displacement MMWR (2011) Horn of Africa Mass population displacement, health and public health services Salama and Dondero33 Complex emergencies HIV-specific risk factors, insecurity, mass population displacement, health and public health services Toole and Waldman17 Complex emergencies and displacement crises Health and public health services, mass population displacement, overcrowding, living conditions, nutrition, WASH WHO34 Complex emergencies Environment, health and public health services, humanitarian response, mass population displacement, nutrition WHO20 Afghanistan and neighbours Environment, health and public health services, living conditions, mass displace- ment, overcrowding, nutrition, WASH WHO19 Liberia Economy, environment, health and public health services, HIV-specific risk factors, infrastructure, WASH, insecurity, living conditions, mass population displacement, overcrowding, nutrition

the main root causes for increased mortality (all funding, infrastructure in CHEs is often inadequate, causes) in complex humanitarian emergencies. especially in response to mass influx of people either Insecurity is composed of factors such as armed in camps or in the community. Lack of infrastructure conflict,10 social disruption10 19 30 33 and political also often comes with a lack of domestic instability.2 The specific nature of insecurity differs coordination,2 19 31 which additionally inhibits from complex emergency to complex emergency. efficient coordination with international response. However, by our (UNOCHA) definition, most, if not A lack of resources,2 31 water,2 10 12 14–17 19–21 all, complex emergencies experience a high level of electricity,19 funding22 and staff22 makes the affected severe violence either from inter-state or from intra- population more dependent on an international state conflict. Insecurity triggers other factors such response. as a lack of an adequate humanitarian response as • Humanitarian response10 12 26 34: By our (UNOCHA) it poses risks to aid workers and inhibits access to definition, a complex emergency demands beneficiaries. Additionally, it also inhibits access a multifaceted, multiagency international for the population to health services and has a high humanitarian response. However, poor response can potential to disrupt all other services. itself become a risk for the spread of communicable • Infrastructure2 18 19 22 23 31: Due to insecurity and diseases. Problems can lie with the response itself, also in some cases long-term neglect and lack of due to a lack of international commitment or a

BMJ Global Health Initiatives 13 RESEARCH

lack of professionalism of the responding agencies there are some very specific additional risk factors and organisations.12 Problems can also arise that are associated with an increase in the incidence domestically due to restrictions by governments of HIV in complex emergencies. Key risk factors for or warring parties, unsafe conditions in which aid an increased transmission of HIV include sexual workers cannot properly work without unacceptable and gender-based violence,2 10 15 19 28 30 33 increased levels of risk for themselves or lack of access for rates of sex work,2 10 19 28 30 33 use of unsafe blood various reasons.10 34 This also includes lack of products and conflict-related increased demand organisational motivation22 and poor institutional for (potentially unsafe) blood products,2 19 28 lack support10 and complex international issues such of infection control in healthcare facilities,2 19 28 as the lack of a binding legal framework for the lack of condoms2 28 and an increased use of illicit 24 protection of internally displaced populations. drugs.19 28 33 A high sexually transmitted infection 2 12 15 19–21 23 34 • Environment : Environmental factors prevalence can be linked to an increased risk of can increase the likelihood of communicable 15 contracting HIV. Lack of healthcare access and diseases outbreaks, and this is true beyond the lack of antiretroviral therapy increase the likelihood context of CHEs. However, many environmental of vertical transmission,30 and mass population factors, which would not have mattered otherwise, can be triggered by mass population displacement, displacement can lead to increased contact (sexual especially if populations are displaced into areas and otherwise) with populations with a higher 10 28 33 with a higher prevalence of environmental risk prevalence. factors. Environmental risk factors include weather and climate factors, such as cold and dust storms,2 20 Risk factor cascades but also vector habitats,19 20 34 increased contact with The risk factor clusters as well as individual risk factors animals19 20 and endemic diseases.2 12 19 Mass often interact and exacerbate one another. Some risk population displacement potentially puts people factors and risk factor cluster are particularly likely at risk from these factors and also exacerbates the to start risk cascades, especially mass population factors themselves due to the additional stress displacement (as illustrated in figure 3) and insecurity placed on the local environment by camps and (as illustrated in figure 4). by an influx of large numbers of people, often One of the key mechanisms for driving risk factors accompanied with significant land use changes.19 for communicable diseases in complex emergencies is • Economy2 19 23 25 27 30 31: While economic factors such mass displacement (as shown in figure 3), especially as poverty and lack of resource are certainly issues mass displacement into camp settings.18–20 Camp that are important in humanitarian emergencies, settings enforce a high dependence on outside they are not of the highest importance in CHEs. support for the residents. This makes residents more Poverty and economic degradation have the at risk for other risk factors. Mass displacement can ability to further exacerbate the root causes of the reduce access to healthcare and even if access to underlying conflict but only indirectly increase the healthcare is maintained the level and quality might likelihood of communicable disease outbreaks. be poor.2 10 15 17 18 Mass displacement thus tends to • Health and public health trigger all risk factors associated with lack of access to services2 10 12–15 17 19 20 23 24 27–29 31–36: Breakdown of healthcare and increases the risks for communicable health and public health services is probably one of diseases both at individual and community levels. the main risk factors for communicable diseases in This is often coupled with living conditions that are CHEs both for individuals and for populations. Lack conducive to increased transmission of communicable of access to health and medical care is a key risk diseases and put the individual more at risk.2 12 19 20 factor for severe progressions of most communicable This includes the lack of adequate shelter, which is diseases for the individual.2 10 12 15 17 19 20 28 29 31 33 34 It especially prone to increase vector-borne diseases and also facilitates the further spread of communicable respiratory diseases, especially in areas with cold diseases such as tuberculosis and makes detection temperatures.2 10 15 17 19 20 24 29 Overcrowding—often of cases and outbreaks harder. Additionally, in together with inadequate shelter and lack of sufficient complex emergencies, public health services WASH—increases the likelihood of triggering including vaccination, communicable disease hygiene risk factors and also the transmission rate of prevention and control measures, and surveillance respiratory infections and diseases such as measles. are no longer available making disease outbreaks For respiratory infections, this is further exacerbated more likely, harder to detect and harder to by conditions that lead to the use of indoor fires and control.2 10 12 13 15 17 19 20 24 27 31–33 35 This breakdown of subsequent indoor air pollution.2 19 20 services can be seen as a function of the underlying Additionally, as human populations become more conflict but is further compounded if there is not overcrowded, transmission of infections becomes

enough political will to provide adequate health more efficient, that is, the reproductive ratio (R0) 2 37 protection. of the infection increases. As R0 increases, the • HIV-specific risk factors2 10 15 19 28 30 33: HIV is a unique threshold immunisation coverage needed to achieve and often overlooked concern in CHEs. While many herd immunity also increases.38 Consequently, of the aforementioned risk factors also apply to HIV, immunisation coverage that was previously sufficient

14 BMJ Global Health Initiatives RESEARCH

Fig 3 | Mass population displacement cascade. WASH, water, sanitation and hygiene

is inadequate to prevent outbreaks. One of the main poses risks to aid workers’ security both for domestic/ problems, especially in overcrowded camps, is the national and international/expatriate staff.10 34 Aid provision of safe water and adequate hygiene. If organisations are—understandably—increasingly WASH conditions deteriorate, especially diarrhoeal reluctant to accept very high risks to their personnel, disease risk increases considerably. Any insufficiency leading to gaps in provision of services, which in WASH is more pronounced when coupled with high would otherwise have been filled by a humanitarian population density, as experienced in camp situations. response. Insecurity also increases the risk of the loss However, mass displacement, even when not coupled of domestic experts in disease prevention due to injury, with displacement into camps, also triggers additional death and flight.42 risk factors. Displacement can be into areas with These are only some aspects of two of the endemic diseases to which the displaced population many mechanisms by which CHEs drive risks for has no immunity.12 Additionally, mass displacement communicable diseases. We identified further makes populations vulnerable to environmental factors cascades triggered by economics and infrastructure as well as reinforcing these.12 21 Mass displacement can and risk factor cluster interaction for WASH and health exacerbate insecurity and therefore reignite a vicious systems risk factors. However, the level of complexity circle leading to further displacement and breakdown in these types of emergencies makes it impossible to of healthcare, services and infrastructure. capture all levels of interaction adequately. It is not so Insecurity itself, whether exacerbated by mass much that complex emergencies create different risk displacement or not, is an important triggering factors than other humanitarian crises but that they mechanism for communicable disease risk factors exacerbate any individual risk factors and compound in CHEs (as shown in figure 4). Insecurity, including interaction effects. Levels of risk factors will invariably political instability, armed conflict and social be higher in a complex emergency and the amount disruption, destroys services that previously prevented of interacting risk factors creates a ‘perfect storm’44 the spread of communicable diseases or disallows where a multifaceted, well-funded and logistically and access to these services by making accessing them politically highly integrated humanitarian response unsafe.2 10 14 19 28 30 31 33 36 39 40 This is particularly is not possible due to political, financial or security important for healthcare services that in the last reasons. These conditions make the danger of one or few years have increasingly become a target of more outbreaks of communicable diseases extremely armed conflict and attacks, decreasing the safety of high. both staff and patients.41–43 Additionally, disease While complex humanitarian emergencies do not prevention programmes are likely to be disrupted trigger risk factors that are unknown in other types of and infrastructure to be destroyed.15 17 20 36 With emergencies and disasters, they produce much higher regard to humanitarian response, which can under levels of risk and often tend to trigger more of the certain circumstance step into the place of previously known risk factors as well as risk factor cascades. Risk government-provided services, insecurity makes an factors related to poor sanitation and hygiene,45–52 adequate humanitarian response difficult.10 34 Not nutrition,46 53–55 mass population displacement and only will access to affected populations be difficult, overcrowding47 53 56–60 have been discussed extensively especially in situations when insecurity and active in the academic literature as being important in most fighting lead to entrapment or even to siege situation, types of emergencies, while risk factors resulting from as recently seen in Syria and Iraq, but insecurity also an inadequate humanitarian response, armed conflict

BMJ Global Health Initiatives 15 RESEARCH

Fig 4 | Insecurity cascade

and a breakdown in government services are generally Conclusion more associated with complex emergencies and other CHEs pose a significant threat to public health. The situations linked to failing statehood, such as civil war. described cascades, interactions and feedback loops The question remains of how to make useful this are only some of the most striking examples. The information on risk factors and their interactions. increased exposure to very many interacting risk While many of the risk factors and even starting points factors and the resulting risk factor cascades created of risk factor cascades are addressable, the context by a complex emergency encourages a perfect storm of of a complex emergency often prevents any such communicable diseases risk. interventions. A key first step in any attempt to address However, despite these extremely increased these issues in a given complex emergency is a rapid risks and the exceptional situation that CHEs 3 61–63 but thorough initial needs assessment, including pose, we did not find a correspondingly high level an assessment of the most critical risk factors present of academic engagement with the issue. Most in that specific complex emergency in order to develop of the included articles discussed situations of an evidence-based intervention strategy. However, mass displacement into camps, which is arguably it is unclear how to best undertake such a needs the best studied situation concerning complex assessment. Moreover, beyond the development of emergencies. However, conflicts like Syria and evidence-based risk assessment and management Yemen demonstrate that this might not be the most methods, there is a need for more rigorous research important situation in the 21st century. Syria and into the operational and structural barriers that make Yemen feature high levels of entrapment,64–67 as it difficult to address risk factors in CHEs. they are characterised by limited or no displacement due to a lack of safe humanitarian corridors. This Limitations situation coincides with a high level of most other This systematic review included subjective risk factors, especially lack of access to healthcare, interpretation as risk factors were rarely the main focus lack of humanitarian response, lack of WASH and of the included articles. Authors do not always clearly other services, food insecurity and high levels of describe the risk factors and their mechanisms. This insecurity. We conclude that more rigorous research introduced an interpretative and subjective element on the risk of communicable disease outbreaks in within the included articles, which became more complex humanitarian emergencies could elucidate subjective due to the level of interpretation required to opportunities to either prevent or better manage complete the thematic synthesis. However, the authors such events. Such research should be undertaken in maintained constant feedback to one another and collaboration between practitioners and academics. discussed challenges, interpretations and limitations More CHE research on entrapment situations is to ensure reliability and validity of the findings to especially desirable, in response to the nature of the degree that a qualitative analysis allows. We are recent conflicts. therefore confident that our interpretation properly Contributors: All authors (CCH, JB, PRH) contributed to the reflects the data, although agreeing that other conceptualisation of the research. Primary coding was done by interpretations are possible and may be equally valid. CCH, except for one article in Spanish, which was primary coded by This review was necessarily a qualitative synthesis as JB. JB and CCH confirmed the primary codes and added secondary codes for all articles. CCH wrote the draft manuscript and JB and PRH the evidence base (heterogeneous and qualitative in contributed feedback to and revisions of the manuscript. All authors nature) did not support quantitative analysis. revised and approved the final version of the manuscript.

16 BMJ Global Health Initiatives RESEARCH

Funding: The research was funded by the National Institute for Health 23 Cuadrado P, González P. Cólera epidémico en situaciones de Research Health Protection Research Unit (NIHR HPRU) in Emergency emergencia compleja. Rev Esp Salud Pública 2014;88:191–201. Preparedness and Response at King’s College London in partnership 24 Burkle FM. Pediatric issues in complex emergencies. Ambulatory with Public Health England (PHE), in collaboration with the University Child Health 2001;7:119–26. of East Anglia, Norwich Medical School. 25 Howard N, Shafi A, Jones C, et al. Malaria control under the Taliban regime: insecticide-treated net purchasing, coverage, Disclaimer: The views expressed are those of the authors and not and usage among men and women in eastern Afghanistan. necessarily those of the NHS, the NIHR, the Department of Health or Malar J 2010;9:7. Public Health England. 26 Bompangue D, Giraudoux P, Piarroux M, et al. Cholera epidemics, war Competing interests: None declared. and disasters around Goma and Lake Kivu: an eight-year survey. PLoS Negl Trop Dis 2009;3:e436. Patient consent: Not required. 27 Howard N, Chandramohan D, Freeman T, et al. Socio-economic Provenance and peer review: Not commissioned; externally peer factors associated with the purchasing of insecticide-treated nets in reviewed. Afghanistan and their implications for social marketing. Trop Med Int Health 2003;8:1043–50. Data sharing: No additional data are available. 28 Khaw AJ, Salama P, Burkholder B, et al. HIV risk and prevention This is an open access article distributed in accordance with the in emergency-affected populations: a review. Disasters Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, 2000;24:181–97. which permits others to copy, redistribute, remix, transform and 29 Kolaczinski J, Webster J. Malaria control in complex emergencies: the build upon this work for any purpose, provided the original work example of East Timor. Trop Med Int Health 2003;8:48–55. is properly cited, a link to the licence is given, and indication of 30 Leyenaar J. Human immuno-deficiency virus and infant feeding whether changes were made. See: https://​creativecommons.​org/​ in complex humanitarian emergencies: priorities and policy licenses/​by/​4.​0/. considerations. Disasters 2004;28:1–15. 31 Liddle KF, Elema R, Thi SS, et al. TB treatment in a chronic complex © Author(s) (or their employer(s)) 2018. Re-use permitted under emergency: treatment outcomes and experiences in Somalia. Trans R CC BY. Published by BMJ. Soc Trop Med Hyg 2013;107:690–8. 32 Centers for Disease Control and Prevention (CDC). Measles— 1 Macias L. Complex Emergencies. Austin: Center for Climate Change Horn of Africa, 2010–2011. MMWR Morb Mortal Wkly Rep and African Political Stability;. 2013. 2012;61:678–84. 2 Connolly MA, Gayer M, Ryan MJ, et al. Communicable diseases 33 Salama P, Dondero TJ. HIV surveillance in complex emergencies. AIDS in complex emergencies: impact and challenges. Lancet 2001;15(Suppl 3):S4–12. 2004;364:1974–83. 34 WHO. Outline strategy for malaria control in complex emergencies. 3 Brown V, Guerin PJ, Legros D, et al. Research in complex humanitarian Geneva: WHO, 2000. emergencies: the Médecins Sans Frontières/Epicentre experience. 35 Kolaczinski J, Graham K, Fahim A, et al. Malaria control in Afghanistan: PLoS Med 2008;5:e89–6. progress and challenges. Lancet 2005;365:1506–12. 4 Paquet C, Hanquet G. Control of infectious diseases in refugee and 36 Kolaczinski J. Roll Back Malaria in the aftermath of complex displaced populations in developing countries. Bull Inst Pasteur emergencies: the example of Afghanistan. Trop Med Int Health 1998;96:3–14. 2005;10:888–93. 5 Duffield A, Reid G, Shoham J, et al. Evidence base for interventions in 37 Swift L, Hunter PR, Lees AC, et al. Wildlife trade and the emergence of complex emergencies. Lancet 2005;365:842–3. infectious diseases. Ecohealth 2007;4:25–30. 6 Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for 38 Guerra F, Bolotin S, Lim G, et al. The basic reproduction number (R₀) systematic reviews and meta-analyses: the PRISMA statement. Ann of measles: a systematic review. Lancet. in print. Intern Med 2009;151:264–9. 39 Sparrow A, Almilaji K, Tajaldin B, et al. Cholera in the time of 7 Anon. Merriam Webster's Medical Dictionary. Springfield: Merriam war: implications of weak surveillance in Syria for the WHO's Webster, 2007. preparedness—a comparison of two monitoring systems. BMJ Glob 8 Nicoll A, Editorial team C. Preventing and controlling disease Health 2016;1:e000029. outbreaks in a complex emergency situation: discussion of the 40 Devi S. Syria's health crisis: 5 years on. Lancet 2016;387:1042–3. tsunami aftermath. Weekly releases 2005;10. 41 Dewachi O, Skelton M, Nguyen VK, et al. Changing therapeutic 9 Thomas J, Harden A. Methods for the thematic synthesis of geographies of the Iraqi and Syrian wars. Lancet 2014;383:449–57. qualitative research in systematic reviews. BMC Med Res Methodol 42 Fouad FM, Sparrow A, Tarakji A, et al. Health workers and the 2008;8:45. weaponisation of health care in Syria: a preliminary inquiry for The 10 Brennan RJ, Nandy R. Complex humanitarian emergencies: a major Lancet–American University of Beirut Commission on Syria. Lancet global health challenge. Emerg Med 2001;13:147–56. 2017;390:2516–26. 11 Burkle FM. Globalization and disasters: issues of public health, 43 Terry F. Violence against health care: insights from Afghanistan, state capacity and political action. Journal of International Affairs Somalia, and the Democratic Republic of the Congo. International 2006;59:241–XIII. Review of the Red Cross 2013;95:23–39. 12 Chaignat CL, Monti V. Use of oral cholera vaccine in complex 44 Burkle FM, Greenough PG. Impact of public health emergencies on emergencies: what next? Summary report of an expert meeting and modern disaster taxonomy, planning, and response. Disaster Med recommendations of WHO. J Health Popul Nutr 2007;25:244–61. Public Health Prep 2008;2:192–9. 13 Close RM, Pearson C, Cohn J. Vaccine-preventable disease and 45 Ezzati M, Lopez AD, Rodgers A, et al. Selected major risk factors the under-utilization of immunizations in complex humanitarian and global and regional burden of disease. Lancet emergencies. Vaccine 2016;34:4649–55. 2002;360:1347–60. 14 Coulombier D, Pinto A, Valenciano M. [Epidemiological surveillance 46 Kouadio IK, Aljunid S, Kamigaki T, et al. Infectious diseases following during humanitarian emergencies]. Med Trop 2002;62:391–5. natural disasters: prevention and control measures. Expert Rev Anti 15 Fisher ML, Brahmbhatt D, Powers R, et al. Health issues in Infect Ther 2012;10:95–104. humanitarian emergencies. International Disaster Nursing 47 Ivers LC, Ryan ET. Infectious diseases of severe weather- 2010:427–48. related and flood-related natural disasters. Curr Opin Infect Dis 16 Guthmann JP, Klovstad H, Boccia D, et al. A large outbreak of hepatitis 2006;19:408–14. E among a displaced population in Darfur, Sudan, 2004: the role of 48 Cronin AA, Shrestha D, Cornier N, et al. A review of water and water treatment methods. Clin Infect Dis 2006;42:1685–91. sanitation provision in refugee camps in association with selected 17 Toole MJ, Waldman RJ. The public health aspects of complex health and nutrition indicators—the need for integrated service emergencies and refugee situations. Annu Rev Public Health provision. J Water Health 2008;6:1. 1997;18:283–312. 49 Artelt T, Kaase M, Scheithauer S. Challenges regarding infectious 18 Burkle FM. Lessons learnt and future expectations of complex diseases in migrants: special aspects in the care of female refugees. emergencies. BMJ 1999;319:422–6. [German]. Gynakologe 2017;50:134–8. 19 WHO. Communicable disease toolkit Liberia. Geneva: WHO, 2004. 50 Cutler SJ, Sj C. Refugee crisis and re-emergence of forgotten infections 20 WHO. Communicable disease profile Afghanistan and neighbouring in Europe. Clin Microbiol Infect 2016;22:8–9. countries. Geneva: WHO, 2001. 51 Jones G, Haeghebaert S, Merlin B, et al. Measles outbreak in a refugee 21 Group GE. Public health impact of Rwandan refugee crisis: what settlement in Calais, France: January to February 2016. Euro Surveill happened in Goma, Zaire, in July, 1994? Goma Epidemiology Group. 2016;21:30167. Lancet 1995;345:339–44. 52 Stich A. [Frequent infectious diseases in migrants]. Internist 22 Abubakar A, Azman AS, Rumunu J, et al. The first use of the global 2016;57:409–15. oral cholera vaccine emergency stockpile: lessons from South Sudan. 53 Watson JT, Gayer M, Connolly MA. Epidemics after natural disasters. PLoS Med 2015;12:e1001901. Emerg Infect Dis 2007;13:1–5.

BMJ Global Health Initiatives 17 RESEARCH

54 Waring SC, Brown BJ. The threat of communicable diseases following 62 Salama P, Spiegel P, Talley L, et al. Lessons learned from natural disasters: a public health response. Disaster Manag Response complex emergencies over past decade. Lancet 2005;3:41–7. 2004;364:1801–13. 55 Young H, Jaspars S, Nutrition JS. Nutrition, disease and death in times 63 Bradt DA, Drummond CM. Rapid epidemiological assessment of famine. Disasters 1995;19:94–109. of health status in displaced populations—an evolution toward 56 Wilder-Smith A. Tsunami in South Asia: what is the risk of post- standardized minimum essential data sets. Prehosp Disaster Med disaster infectious disease outbreaks? Ann Acad Med Singapore 2002;17:178–85. 2005;34:625–31. 64 United Nations. Life in War Zones remains grim, with cities turned into 57 Gayer M, Legros D, Formenty P, et al. Conflict and emerging infectious death traps, civilian suffering ‘pushed to the limits’, Secretary-General diseases. Emerg Infect Dis 2007;13:1625–31. tells Security Council. New York: United Nations, 2017. 58 Sharara SL, Kanj SS, War KS. War and infectious diseases: challenges 65 Brophy Z. Aid agencies struggle to support Yemen's trapped civilians: of the Syrian civil war. PLoS Pathog 2014;10:e1004438. the New Arab. 2015. https://www.alaraby.​ co.​ uk/​ english/​ news/​ ​ 59 Toole M. A global public health challenge. Infectious Disease Clinics 2015/4/​ 2/​ aid-​ agencies-​ struggle-​ to-​ support-​ yemens-​ trapped-​ ​ of North America 1995;9:353–66. civilians 60 Thomas SL, Thomas SD. Displacement and health. Br Med Bull 66 Gladstone R. Why so many children are being killed in Aleppo. 2004;69:115–27. New York Times 2016. 61 Moss WJ, Ramakrishnan M, Storms D, et al. Child health in complex 67 Global Centre for the Responsibility to Protect. Illegal blockade of emergencies. Bull World Health Organ 2006;84:58–64. starving civilians in Yemen atrocity alert. 2017.

18 BMJ Global Health Initiatives RESEARCH

Evaluation of a programme for ‘Rapid Assessment of Febrile Travelers’ (RAFT): a clinic-based quality improvement initiative Farah Jazuli,1 Terence Lynd,1 Jordan Mah,1 Michael Klowak,2 Dale Jechel,1 Stefanie Klowak,3 Howard Ovens,4,5 Sam Sabbah,6,7 Andrea K Boggild3,7,8

For numbered affiliations see ABSTRACT (n=10, 6.5%), ­laboratory-confirmed influenza (n=8, end of article. Background 5%) and lobar pneumonia (n=8, 5%). Correspondence to: Fever in the returned traveller is a potential medical Dr Andrea K Boggild Conclusions andrea.​ boggild@​ utoronto.​ ca​ emergency warranting prompt attention to exclude In addition to provision of more timely care to Additional material is published life-threatening illnesses. However, prolonged ambulatory febrile returned travellers, we reduced ED online only. To view please visit evaluation in the emergency department (ED) bed-usage by providing an alternate setting for follow- the journal online. may not be required for all patients. As a quality up malaria screening, and treatment of infectious Cite this as: Jazuli F, Lynd T, improvement initiative, we implemented an algorithm diseases manageable in an outpatient setting, but Mah J, et al. Evaluation of a programme for for rapid assessment of febrile travelers (RAFT) in an requiring specific therapy. ‘Rapid Assessment of Febrile ambulatory setting. Travelers’ (RAFT): a clinicbased quality improvement Methods Introduction initiative. BMJ Open Criteria for RAFT referral include: presentation to Fever in the returned traveller is a common syndrome, 2016;6:e010302. doi:10.1136/ the ED, reported fever and travel to the tropics or bmjopen-2015-010302 occurring in 17% of ill returned Canadian travellers subtropics within the past year. Exclusion criteria and new immigrants presenting for care after travel.1 Prepublication history and include Plasmodium falciparum malaria, and additional material is available. Although often due to self-limited infections, such as To view please visit the journal fulfilment of admission criteria such as unstable vital travellers’ diarrhoea, fever after travel may indicate (http://dx.doi.org/10.1136/ signs or significant laboratory derangements. We serious and potentially life-threatening causes, such bmjopen-2015-010302). performed a time series analysis preimplementation as malaria, dengue or typhoid fever, as was the case Received: 19 October 2015 and postimplementation, with primary outcome of in 28% of febrile returned Canadian travellers or new Revised: 12 April 2016 wait time to tropical medicine consultation. Secondary immigrants studied recently.1 Fever in the returned Accepted: 4 July 2016 outcomes included number of ED visits averted for traveller is necessarily encountered by front-line repeat malaria testing, and algorithm adherence. Republished with Canadian practitioners such as family physicians, permission from BMJ Open Results 2016;6:e010302. walk-in physicians and emergency department (ED) From February 2014 to December 2015, 154 patients physicians who do not specialise in infectious diseases. were seen in the RAFT clinic: 68 men and 86 women. Thus, standardised management protocols, algorithms Median age was 36 years (range 16–78 years). Mean and guidelines are needed to assist in the management time to RAFT clinic assessment was 1.2±0.07 days of this commonly imported syndrome. (range 0–4 days) postimplementation, compared to A major gap in the care of febrile returned travellers 5.4±1.8 days (range 0–26 days) prior to implementation exists in Canada. Fever in this population constitutes (p<0.0001). The RAFT clinic averted 132 repeat malaria a potential medical emergency warranting immediate screens in the ED over the study period (average 6 per exclusion of life-threatening travel-acquired infections. month). Common diagnoses were: traveller's diarrhoea One to two Canadians are reported to die each year (n=27, 17.5%), dengue (n=12, 8%), viral upper due to delayed diagnosis or treatment of malaria,2 respiratory tract infection (n=11, 7%), chikungunya and many more become critically unwell and require admission to intensive care.3 However, most febrile returned travellers will have more benign aetiologies, Strengths and limitations of this study such as traveller’s diarrhoea (TD) or respiratory tract Fever in the returned traveller is a medical emergency, potentially heralding infections;14 yet there is no standardised ’system’ life-threatening infections such as malaria, or more benign aetiologies, some of for close follow-up and monitoring of such patients which may self-resolve. in the critical first few days of their illness, when As a clinic-based quality improvement initiative, we designed and implemented deterioration may occur or a serious diagnosis may an algorithm for ‘Rapid Assessment of Febrile Travelers’ based on Canadian declare itself. Thus, many patients are either admitted national fever assessment guidelines, and demonstrated a significant reduction to hospital for observation, or are discharged from in wait-time to Tropical Medicine assessment and emergency department (ED) the ED with ambulatory infectious diseases follow- bed-usage. up, days to weeks later. This gap in care translates We also demonstrated a reduction in repeat ED visits for follow-up malaria into over-utilisation of acute care, such as the ED and screening after an initial negative. general medicine inpatient service, and also leads to under-provision of care for those who present early We did not have access to the full range of hospital administrative ED data that in their potentially serious illness with more benign would permit quantification of economic savings; thus, we did not perform an appearing clinical parameters.5 economic analysis. National Canadian guidelines on the assessment We identified an aetiological diagnostic gap in 14% of patients, a rate that is of febrile returned travellers have been published,6 similar to other studies of fever in the returned traveller. and we have adapted these guidelines into an ED

BMJ Global Health Initiatives 19 RESEARCH

decision-algorithm to standardise the evaluation General Hospital from three EDs of Mount Sinai and disposition of such patients, through creation of Hospital (MSH) and the University Health Network the ’Rapid Assessment of Febrile Travelers’ (RAFT) (UHN) between October and December 2013. This Programme. The RAFT algorithm triages patients to enabled establishment of a baseline ’referral wait time’ either hospital admission, in the case of unstable vital to definitive consultation for febrile returned travellers signs, significant laboratory derangements, volume at our institution. depletion or Plasmodium falciparum malaria, or to same-day referral to the RAFT Clinic in the Tropical Algorithm development Disease Unit (TDU) of Toronto General Hospital, if In consultation with the three EDs of MSH and UHN, patients fail to fulfil admission criteria (figure ). 1 In we adapted the national fever assessment guidelines6 addition, the RAFT algorithm provides management into a simple decision-algorithm for use in the ER advice and diagnostic stewardship to the participating (figure 1), as well as supporting materials for clinicians EDs. The RAFT Programme aims to fill the identified and patients (see online supplementary files 1 and 2). care gap and, in the process, improves patient flow, Patients are eligible for RAFT Clinic referral if they: utilisation and delivery of service and clinical outcomes. report subjective fever or are objectively febrile in the We herein report our primary and secondary outcome ED; have travelled outside North America to a tropical measures of quality and performance at 10 months or subtropical destination in the past year; are being postimplementation of the RAFT programme. assessed in the ED between 08:00 on Sunday to 08:00 on Friday. Exclusion criteria for RAFT Clinic referral Methods include: unstable vital signs; significant laboratory Preimplementation derangements; volume depletion; fulfilment of other We collected data on the turnaround time of referrals standard hospital admission criteria; and initial for fever in the returned traveller to the TDU at Toronto malaria screening that is positive for P. falciparum

Fig 1 | Algorithm for assessment of fever in the returned traveller. †The Rapid Tropical Assessment Clinic is designed to ensure definitive disposition of a febrile returned traveller within 24 hours of their initial emergency room presentation. Between Friday after 8:00 andS unday before 8:00, as well as statutory holidays and the obligatory ambulatory closure for 2 weeks over the Christmas/New Year’s block, the Rapid Tropical Assessment Clinic is unavailable. During these times, if the patient does not have Plasmodium falciparum or otherwise fulfil admission criteria, the patient should still be referred to General Internal Medicine or Infectious Diseases for disposition (as per standard historical procedure). §Additional investigations should be based on clinical judgement. For example, a febrile returned traveller with diarrhoea should also have stool investigations; a febrile returned traveller with dysuria should have urine Culture&Sensitivity±STI screening, etc.*If the malaria screen is positive for Plasmodium vivax, P. ovale or P. malariae (ie, non-P. falciparum), please initiate chloroquine therapy: 4 tablet loading dose (600 mg base), followed by 2 tablets 6 hours later. Completion of therapy will be organised by the Rapid Tropical Assessment Clinic. If the malaria screen is positive for P. vivax and the patient travelled to Papua New Guinea or Indonesia, please initiate Malarone therapy: 4 tablets orally×1 with food. Completion of therapy will be organised by the Rapid Tropical Assessment Clinic. Chloroquine tablet: 150 mg base (in a 250 mg tablet). Treatment course: Loading dose of 600 mg base, followed by 300 mg base 6 hours later. This is followed by 300 mg base at 24 and 48 hours for a total of 1.5 g base. Malarone tablet: fixed combination of 400-mg atovaquone+100-mg proguanil. Treatment course: 4 tablets orally once daily with food ×3 days. CBC, complete blood count; ICU, intensive care unit; ID, infectious diseases; LFT, liver function test; NP, nasopharyngeal; P. falciparum, Plasmodium falciparum; STI, sexually transmitted infections

20 BMJ Global Health Initiatives RESEARCH

malaria. For patients referred to the TDU via the RAFT Results algorithm, we offer a same-day or next-day assessment Over the first 22 months of implementation, 198 clinic (RAFT Clinic), which runs concurrently with patients were referred to the RAFT clinic and 154 regularly scheduled ambulatory tropical medicine (78%) were assessed, while 44 patients referred to the clinics. The RAFT Clinic is operational on weekdays, RAFT clinic failed to present to the clinic as instructed. and is staffed by 1–2 administrative assistants, 1 of 2 Of those who did not come to the clinic on referral, staff physicians on each day and up to three rotating 19 (43%) were contacted and rescheduled, 9 (20%) resident trainees. A roster of eight staff physicians felt completely well and did not want an appointment in tropical medicine offer coverage of TDU staff and 13 (30%) were lost to follow-up, though not seen physician absences. No additional hospital resources again in the ED for their presenting illness. Of the 154 were committed to development or implementation patients referred and assessed in the RAFT clinic, 68 of the programme. Since the RAFT clinic operates (44%) were men and 86 (56%) were women. Median concurrently with regularly scheduled tropical age was 36 years (range 16–78 years). English was the medicine clinics, there is no net increase in resources first language of 81% (n=124). Median time between allocated. presentation to the ED and evaluation by an ED physician was 76 min (range 9–359 min; 90th centile Implementation 168 min), compared to a provincial median of 60 min The RAFT programme (ED algorithm and RAFT Clinic and 90th centile of 180 min (Ovens, unpublished in the TDU) were implemented simultaneously at the data). Median time between initial presentation to the three EDs of MSH/UHN at the end of February 2014. In- ED and to the first malaria screen was 105 min (range services were provided to the staff of participating EDs, 0–911 min). Median time between initial presentation and wall posters were mounted in each ED. Binders to the ED and to final disposition was 290.5 min (range containing the national fever assessment guidelines, 32–936 min; 90th centile 537.8 min), compared to a and copies of the RAFT algorithm (figure 1), patient provincial median of 180 min and 90th centile of handout (see online supplementary file 1) and RAFT 420 min (Ovens, unpublished data). referral form (see online supplementary file 2) were Mean time to RAFT Clinic assessment following strategically placed in each ED, and online. Pocket ED discharge was 1.2±0.07 days (range 0–4 days) cards of the algorithm were provided to ED staff as postimplementation, compared to 5.4±1.8 days (range well as Infectious Diseases staff, residents and fellows 0–26 days) prior to implementation (p<0.001). Time who typically field calls regarding febrile returned to RAFT Clinic assessment did not differ by age, sex, travellers. All Infectious Diseases staff were reminded first language or family physician status; however, we of the RAFT programme at monthly business meetings noted an increased time to assessment for referrals during the initial implementation period. made on Friday and Saturday (p<0.0001; table 1). No patient was admitted to hospital during or following care of their travel-acquired illness in the RAFT clinic. Postimplementation evaluation Twenty-two patients (14%) had a repeat visit to the ED Following institutional review board approval at MSH prior to assessment in the RAFT clinic, and for those and UHN, we extracted demographic, clinical and patients, their second malaria screen was performed. health systems data on all RAFT patients evaluated The RAFT clinic averted an average of six repeat between 28 February 2014 and 31 December 2015, malaria screens in the ED per month, as these were and entered them into a password protected MS now being performed in the RAFT clinic, rather than Access database. The primary outcomes of interest using ED resources. This translates to a 24 hour per were turnaround time of referrals for febrile returned month reduction in ED bed-usage, assuming a 4 hour travellers postimplementation. Secondary outcomes stay in the ED for triage, blood work and the malaria of interest included: number of repeat ED visits prior result to be returned. to the RAFT clinic visit (ideally 0–1 day later), bed- Top regions of exposure were: the Caribbean (n=43, usage averted for repeat 24 hour malaria screening 28%), sub-Saharan Africa (n=31, 20%), South Asia and adherence to the algorithm regarding laboratory (n=24, 16%), Southeast Asia (n=17, 11%), Central investigations. America/Mexico (n=17, 11%) and South America (n=14, 9%). Among 79 different countries visited by Analysis RAFT patients, the most common countries of exposure Descriptive statistics (mean, SD, median, range) were were: India (n=18, 12%), the Dominican Republic calculated for continuous variables, and differences (n=15, 10%), Cuba (n=9, 6%), Tanzania (n=9, 6%), were compared using Student’s t-test or, in the case Brazil (n=8, 5%), Thailand (n=8, 5%), Mexico (n=7, of non-normal distribution, the Mann-Whitney Rank 4.5%), Jamaica (n=6, 4%) and South Africa (n=5, Sum test. Categorical variables were quantified by 3%). The median trip duration was 14 days (range proportions, and differences were compared using 3–1095 days). Yates’ corrected χ2 analysis. All computations were The median temperature of RAFT Clinic patients performed using the GraphPad Prism software at presentation in the ED was 37.1°C (range 35.2– (GraphPad, USA). Level of significance was set at 40.7°C). Adherence to the recommended initial blood p<0.05. work algorithm was variable. Ninety-nine per cent of

BMJ Global Health Initiatives 21 RESEARCH

patients (n=152) had a complete blood count (CBC) laboratory-confirmed influenza (n=8, 5%), lobar drawn, while 98% (n=151) had electrolytes and pneumonia (n=8, 5%), acute urinary tract infection creatinine drawn. Ninety-four per cent of patients (n=7, 4.5%) and rickettsioses (n=6, 4%) (table 2). (n=144) received malaria screening. Among those Acute HIV was diagnosed in two febrile returned who did not receive malaria screening, 6 of 10 travellers, and Plasmodium vivax malaria in another travelled to areas without appreciable malaria risk two (table 2). Two cases of P. falciparum malaria were such as the USA (n=2), Cuba (n=1), Trinidad (n=2) and diagnosed among RAFT Clinic patients; however, both Mexico (n=1), while 4 (40%) travelled to areas where of these patients had been appropriately referred to malaria screening would have been indicated such as and assessed by the inpatient Infectious Diseases Nicaragua (n=2), the Dominican Republic (n=1) and consultation service, as per the algorithm, and then the Philippines (n=1). Adherence to the remainder of referred to RAFT by the Infectious Diseases team, the suggested blood work, in decreasing order, was as rather than coming to RAFT off-protocol directly from follows: liver function tests (hepatic transaminases, the ED. All cases of chikungunya fever were acquired bilirubin, alkaline phosphatase) 90% (n=138), blood in the Caribbean or Central America: three in Jamaica, cultures 88% (n=135) and urinalysis 65% (n=100). three in the Dominican Republic and one each in the Diagnoses were classified into major common British Virgin Islands, Costa Rica, the Dominican presenting febrile syndromes, such as gastrointestinal Republic, St. Lucia and St. Vincent. (n=44, 29%), respiratory (n=39, 25%), vector-borne (n=32, 21%), sexually transmitted infection (STI)/ genitourinary (n=13, 8%), lymphadenopathy (n=2, Discussion 1%), skin and soft-tissue infections (n=2, 1%), Implementation of the RAFT clinic led to a 78% musculoskeletal (n=1, 0.6%), non-specific viral reduction in the time to assessment by ambulatory syndrome (n=19, 12%) and no final aetiological tropical medicine, and this enabled febrile returned diagnosis (n=3, 2%). Non-infectious causes were travellers to be followed closely during the critical first found in three travellers (2%). Common aetiological few days of illness during which clinical deterioration diagnoses were: TD (n=27, 17.5%), dengue fever can occur. That we did not have any patients requiring (n=12, 8%), viral upper respiratory tract infection admission following assessment in the RAFT clinic (URTI) (n=11, 7%), chikungunya fever (n=10, 6.5%), supports that such an ED algorithm and programme can be implemented safely. For the two patients with P. falciparum malaria, the RAFT algorithm was followed Table 1 comparison of RAFT clinic referral time by and those patients were appropriately referred to the patient demographics Infectious Diseases consultation service for evaluation, Referral to prior to being sent to RAFT by the Infectious Diseases TDU wait time (days) Mean SD Median Range team. Conversely, the benign, self-limited nature of Sex* many travel-acquired illnesses was reiterated by the Male 1.2 0.8 1 0–3 number of patients seen in the EDs and referred to RAFT, Female 1.2 0.8 1 0–4 but who felt better by the following day and declined Age (years)† the appointment. Since no additional resources <19 1 0 1 1 are committed to running the RAFT programme as 19–50 1.2 0.8 1 0–4 >50 1.3 1.0 1 0–3 patients are accommodated into the regular schedule, First language‡ the impact of ’no shows’ on clinic operations is English 1.2 0.8 1 0–4 negligible. Averaged over a year, our results suggest Non-English 1.3 0.8 1 0–3 that implementation of a RAFT programme can avert Family Doctor§ ∼72 repeat ED visits and 288 hours of ED bed-usage for Yes 1.2 0.8 1 0–4 a second malaria screening, thereby enhancing patient Unknown 1.3 0.8 1 0–3 Day of the week¶ care and reducing workload in the ED. Monday 1.4 1.0 1 0–3 Our RAFT algorithm was derived from national Tuesday 0.8 0.5 1 0–2 guidelines on the approach to febrile returned Wednesday 1 0.4 1 0–2 travellers.6 Yet adherence to the recommended Thursday 0.8 0.4 1 0–1 minimum blood work was variable, with excellent Friday 1.9 1.4 3 0–3 adherence to malaria screening, CBC, electrolytes Saturday 2.0 0.6 2 1–3 Sunday 1.2 0.7 1 0–4 and creatinine, and lesser adherence to tests such as *No difference by the Mann-Whitney Rank Sum test, p=0.558. hepatic transaminases and urinalysis. Liver function †No difference by One-Way ANOVA on Ranks with Dunn’s post hoc test, tests are often perturbed in febrile returned travellers p=0.543. ‡No difference by the Mann-Whitney Rank Sum test, p=0.493; non- with diagnoses such as dengue, enteric fever, Epstein- English first languages included Spanish (n=5), Mandarin (n=4), French Barr virus (EBV), cytomegalovirus (CMV), leptospirosis (n=4), Tagalog (n=3), Hindi (n=2), Portuguese (n=2), Greek (n=2) and 1 and the viral hepatitides, even in the absence of overt each of Bosnian, Bulgarian, Guyanese, Korean, Russian, Tamil, Thai and Ukrainian. jaundice. Furthermore, hyperbilirubinaemia is one §No difference by the Mann-Whitney Rank Sum test, p=0.19. of the diagnostic criteria for severe malaria.7 The ¶p<0.0001 by One-Way ANOVA on Ranks with Dunn’s post hoc test. ANOVA, analysis of variance; RAFT, Rapid Assessment of Febrile pattern of abnormality of liver tests can be useful Travelers; TDU, Tropical Diseases Unit. in refining the differential diagnosis. For instance,

22 BMJ Global Health Initiatives RESEARCH

Table 2 | Final diagnoses issued to 154 febrile returned be collected on all febrile returned travellers in travellers evaluated in the RAFT Clinic between 28 order to inform the differential diagnosis, even in February 2014 and 31 December 2015 the absence of right upper quadrant pain and overt Syndrome/aetiology Number Per cent jaundice. Urinalysis, independent of urine culture, Gastrointestinal syndromes 44 29 is also helpful in refining the differential diagnosis Traveller’s diarrhoea, no confirmed 27 17.5 and may be abnormal in those without frank urinary aetiology symptoms, but with common travel-related diagnoses Campylobacter 3 2 Salmonella, non-typhoidal 3 2 such as pyelonephritis (often occurring in the setting Salmonella typhi/enteric fever 3 2 of traveller’s diarrhoea), STIs including chlamydia Strongyloidiasis 2 1 and gonorrhoea, and leptospirosis, which leads to Giardiasis 1 0.6 significant proteinuria. Conversely, the temptation to Dientamoeba fragilis 1 0.6 perform urine culture on febrile returned travellers Viral enteritis 1 0.6 without signs or symptoms of bacterial cystitis Gastritis 1 0.6 Clostridium difficile colitis 1 0.6 or pyelonephritis should be resisted in order to Postinfectious irritable bowel 1 0.6 avoid inappropriate antimicrobial treatment of syndrome asymptomatic bacteriuria. Even in the absence of a Respiratory syndromes 39 25 dedicated RAFT programme, we advise adherence Viral upper respiratory tract infection 11 7 to the national guidelines for assessment of febrile Lobar pneumonia 8 5 Influenza-like illness 4 3 travellers, though, as demonstrated in this analysis, Influenza A 4 3 adherence overall was quite good. Influenza B 4 3 Respiratory tract infections are the third most Mononucleosis and mono-like 3 2 common cause of fever in the returned traveller in single- syndrome due to EBV or CMV centre and multicentre analyses.68–13 We noted that Lower respiratory tract infection, 1 0.6 non-lobar pneumonia respiratory syndromes were the second most common Haemophilus influenzae 1 0.6 presentation among this group of febrile returned Group A streptococcus pharyngitis 1 0.6 travellers, including eight cases of laboratory-confirmed Acute sinusitis 1 0.6 influenza. We noted at least 1 ’off season’ transmission Coxsackie virus 1 0.6 of travel-acquired influenza A imported back to Canada Vector-borne, non-localising 32 21 in the month of July, which reinforces the point that Dengue fever 12 8 Chikungunya fever 10 6.5 influenza circulates with reciprocal seasonality in the Rickettsioses 6 4 temperate southern hemisphere (ie, during their winter 14 Malaria, Plasmodium vivax 2 1 months, which are our summer months). Influenza Malaria, Plasmodium falciparum 2 1 can circulate year-round in the tropics, so clinicians STI/genitourinary 13 8 must have it on their differential diagnosis and perform Acute urinary tract infection, including 7 4.5 nasopharyngeal (NP) swabs on returned travellers urosepsis Acute HSV-1 2 1 with influenza-like illness, regardless of the month or Acute HIV 2 1 season here. Testing laboratories should be alerted to Syphilis, secondary 1 0.6 the potential for off-season transmission of influenza in Chlamydia trachomatis 1 0.6 travellers, so that they may adjust their NP screening Fever with lymphadenopathy 2 1 algorithms accordingly. Lymphadenitis, bacterial 1 0.6 In single-centre and multicentre analyses, TD Toxoplasmosis 1 0.6 Skin and soft-tissue infections 2 1 is the most common cause of non-malarial fever 68–13 Cellulitis 1 0.6 in the returned traveller. Invasive bacterial Shingles 1 0.6 gastroenteritides (eg, Campylobacter, Salmonella) Musculoskeletal 1 0.6 are common specific causes of fever in the returned Septic arthritis 1 0.6 traveller, but a specific aetiological confirmation is Non-specific viral syndrome 19 12 unlikely due to the insensitivity of stool culture. Of 33 Non-infectious 3 2 Temporal arteritis 1 0.6 febrile returned travellers with presumed bacterial TD Toxidrome, cocaine 1 0.6 in this study, only six had stool culture positivity for a Syncope 1 0.6 typical bacterial enteropathogen, despite >80% of TD No diagnosis 3 2 being bacterial.15 This lack of aetiological confirmation CMV, cytomegalovirus; EBV, Epstein-Barr virus; HSV-1, herpes simplex among individuals with TD has led to widespread virus type 1; RAFT, Rapid Assessment of Febrile Travelers; STI, sexually transmitted infection. implementation of empiric treatment strategies (eg, 3 days of ciprofloxacin15); however, this approach may foster increased fluoroquinolone resistance predominant elevation of hepatic transaminases among endogenous flora, and is counter to the occurs in arboviral infection, enteric fever, EBV, CMV tenets of antimicrobial stewardship. High-sensitivity, and viral hepatitis. Conversely, a more cholestatic multiplex stool pathogen detection assays, now even picture is suggestive of leptospirosis, biliary commercially available,1617 have the potential to better obstruction and even viral alcalulous cholecystitis. direct antimicrobial treatment decisions in returned Thus, hepatic transaminases and bilirubin should travellers with TD.

BMJ Global Health Initiatives 23 RESEARCH

While fatal malaria has been continually imported febrile returned travellers by frequent in-services and by febrile returned travellers, it was the 2013–2015 reminders to ED staff, as well as posted RAFT signage Ebola virus disease (EVD) crisis, during which time and binders in the ED. All fellows on-call overnight for fatal and non-fatal cases of EVD were exported from the Infectious Diseases service had RAFT pocket cards West Africa, that really brought the need for a travel and signage in their reviewing room. Similarly, we do history to the forefront. While we did not have any not have a system that forces a general travel history, patients with EVD during our enrolment period, though with triage protocols mandating the collection ∼17% of diagnoses (n=26) in this population of of travel history to specific geographic regions such febrile returned travellers were notifiable at both the as the Middle East (due to Middle East Respiratory provincial and federal level, including influenza, HIV, syndrome coronavirus) and West Africa (due to Ebola salmonellosis, campylobacteriosis, typhoid fever, virus), we believe that travel history is most likely giardiasis and malaria, indicating public health import requested from all febrile patients entering the ED. At and/or potential communicability.18 Our diagnosis of present, nurses are automatically prompted at triage four febrile returned travellers with acute HIV, syphilis to document a travel history within 21 days should and genital chlamydia infection reinforces the need for a patient present with fever, cough, dyspnoea or a thorough sexual and behavioural history, especially diarrhoea; thus, we feel that the likelihood of missing in the context of known disinhibition on the part of travel-acquired illness in our EDs is low. travellers.19 In their cross-sectional study of >112 000 ill returned international travellers, Matteelli et al20 Conclusions documented STIs in 0.9%, many of which were acute Through implementation of a RAFT programme, HIV. Common STIs including secondary syphilis, acute we have been able to provide more timely care to HIV, acute HSV1 or HSV2, and gonococcemia can all ambulatory febrile returned travellers and, in doing lead to fever in the returned traveller. As such, these so, fill a gap in care faced by such travellers prior to diagnoses should remain on the differential diagnosis implementation of the programme. We have also and be excluded in the sexually active febrile returned reduced ED bed-usage by providing an alternate setting traveller with a compatible history and clinical picture. for follow-up malaria screening. In addition, we have The several limitations of this analysis should be offloaded the responsibility for treatment of infectious acknowledged. First, owing to limitations in the scope diseases that can be managed in an outpatient setting, and funding of the study, we do not have the full range but require specific therapy, such as acute urinary tract of hospital administrative ED data that would permit infections, from the ED. Our programme underscores quantification of economic savings; thus, an economic the range of febrile illnesses that are imported to analysis was neither planned nor performed. Although Canada by travellers on a daily basis, and reinforces we cannot apply a dollar value to the economic savings the need to combine history, physical examination and of an ED visit averted for fever after travel, owing to the a minimum set of laboratory investigations to exclude variability of this metric, we believe that our programme potentially life-threatening imported illnesses such as offers a systems-level improvement in care as most malaria and bacteraemia. patients appreciate timely definitive management and avoidance of ED visits if possible. Second, illnesses AUTHOR AFFILIATIONS with very short incubation periods, such as influenza 1 Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada and URTIs, may be over-represented and erroneously 2McMaster University, Hamilton, Ontario, Canada attributed to travel. We cannot definitively exclude 3Tropical Disease Unit, Division of Infectious Diseases, UHN-Toronto the possibility that some cosmopolitan causes of General Hospital, Toronto, Ontario, Canada 4Department of Family and Community Medicine, University of fever in this group of returned travellers were locally Toronto, Toronto, Ontario, Canada acquired. Third, our ability to comment on the full 5Schwartz/Reisman Emergency Medicine Institute, Sinai Health spectrum of aetiological illness in this population is System, Toronto, Ontario, Canada 6Department of Emergency Medicine, University Health Network, limited by the application of specific diagnostic tests Toronto, Ontario, Canada deemed to be clinically relevant to the patient. Our 7Department of Medicine, University of Toronto, Toronto, Ontario, goal was not to more precisely define the spectrum of Canada 8Public Health Ontario Laboratories, Public Health Ontario, Toronto, illness encountered, but to reduce time to diagnosis, Ontario, Canada as we assume this leads to better outcomes and to a Contributors: FJ, TL and JM contributed to the study design, data more efficient use of hospital resources. In 19 patients, collection and analysis, and to the critical appraisal of the manuscript. ’non-specific viral illness’ was the final diagnosis, and DJ, MK and SK contributed to data collection and analysis, and to in three patients the diagnosis remained unknown, the critical appraisal of the manuscript. HO and SS contributed to programme implementation, data collection and to the critical although symptoms resolved uneventfully and without appraisal of the manuscript. AKB conceived the programme and specific therapy. Understanding the full spectrum study, contributed to the study design, data collection, analysis of aetiological illness in such a population would and interpretation, and was primarily responsible for writing the manuscript. require additional sophisticated and investigational Funding: This research received no specific grant from any funding diagnostics. Finally, we did not have a system by agency in the public, commercial or not-for-profit sectors. which to capture febrile returned travellers who Competing interests: None declared. may have fulfilled algorithm criteria but not sent to Ethics approval: IRB of University Health Network and REB of Mount RAFT. We mitigated the risk of failure to capture all Sinai Hospital.

24 BMJ Global Health Initiatives RESEARCH

Provenance and peer review: Not commissioned; externally peer 10 Wilson ME, Weld LH, Boggild A, et al., GeoSentinel Surveillance reviewed. Network. Fever in returned travelers: results from the GeoSentinel Surveillance Network. Clin Infect Dis 2007;44:1560–8. Data sharing statement: No additional data are available. doi:10.1086/518173 This is an Open Access article distributed in accordance with the 11 Antinori S, Galimberti L, Gianelli E, et al. Prospective observational Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, study of fever in hospitalized returning travelers and migrants which permits others to distribute, remix, adapt, build upon this work from tropical areas, 1997–2001. J Travel Med 2004;11:135–42. non-commercially, and license their derivative works on different doi:10.2310/7060.2004.18557 terms, provided the original work is properly cited and the use is non- 12 Parola P, Soula G, Gazin P, et al. Fever in travelers returning from commercial. See: http://creativecommons.​ org/​ licenses/​ by-​ nc/​ 4.​ 0/​ tropical areas: prospective observational study of 613 cases hospitalised in Marseilles, France, 1999–2003. Travel Med Infect Dis 1 Boggild AK, Geduld J, Libman M, et al. Travel acquired infections 2006;4:61–70. and illnesses in Canadians: surveillance report from CanTravNet 13 West NS, Riordan FA. Fever in returned travelers: a prospective surveillance data, 2009–2011. Open Med 2014;8:e20–32. review of hospital admissions for a 2(1/2) year period. Arch Dis Child 2 Kain KC, MacPherson DW, Kelton T, et al. Malaria deaths in 2003;88:432–4. doi:10.1136/adc.88.5.432 visitors to Canada and in Canadian travellers: a case series. 14 Mutsch M, Tavernini M, Marx A, et al. Influenza virus infection CMAJ 2001;164:654–9. in travelers to tropical and subtropical countries. Clin Infect Dis 3 McCarthy AE, Morgan C, Prematunge C, et al. Severe malaria in 2005;40:1282–7. doi:10.1086/429243 Canada, 2001–2013. Malar J 2015;14:151. doi:10.1186/s12936- 15 Steffen R, Hill DR, DuPont HL. Traveler's diarrhea: a clinical review. 015-0638-y JAMA 2015;313:71–80. doi:10.1001/jama.2014.17006 4 Boggild AK, Geduld J, Libman M, et al. Travel acquired infections 16 Lalani T, Tisdale MD, Maguire JD, et al. Detection of enteropathogens in Canada: CanTravNet 2011—2012. Can Commun Dis Rep associated with travelers’ diarrhea using a multiplex Luminex-based 2014;40:313–25. assay performed on stool samples smeared on Whatman FTA Elute 5 Boggild AK, Page AV, Keystone JS, et al. Delay in diagnosis: malaria cards. Diagn Microbiol Infect Dis 2015;83:18–20. doi:10.1016/j. in a returning traveller. CMAJ 2009;180:1129–31. doi:10.1503/ diagmicrobio.2015.05.011 cmaj.090171 17 Spina A, Kerr KG, Cormican M, et al. Spectrum of enteropathogens 6 Boggild A, Ghesquiere W, McCarthy A , For the Committee to Advise detected by the FilmArray GI Panel in a multicentre study of on Tropical Medicine and Travel (CATMAT). Fever in the returning community-acquired gastroenteritis. Clin Microbiol Infect international traveller: initial assessment guidelines. Can Commun 2015;21:719–28. doi:10.1016/j.cmi.2015.04.007 Dis Rep 2011; 37:1–15. 18 Public Health Agency of Canada. List of nationally notifiable 7 Committee to Advise on Tropical Medicine and Travel (CATMAT). diseases. Ottawa, ON: Public Health Agency of Canada. Canadian recommendations for the prevention and treatment of http://dsol-​ smed.​ phac-​ aspc.​ gc.​ ca/​ dsol-​ smed/​ ndis/​ list-​ eng.​ php​ malaria: an advisory committee statement. Public Health Agency of (accessed 14 Jul 2015). Canada , 2014. publications.​ gc.​ ca/​ collections/​ collection_​ 2014/​ ​ 19 SalitI E, Sano M, Boggild AK, et al. Travel patterns and risk aspc-phac/​ HP40-​ 102-​ 2014-​ eng.​ pdf​ behaviour of HIV-positive people travelling internationally. CMAJ 8 Bottieau E, Clerinx J, Schrooten W, et al. Etiology and outcome of 2005;172:884–8. doi:10.1503/cmaj.1040877 fever after a stay in the tropics. Arch Intern Med 2006;166:1642–8. 20 Matteelli A, Schlagenhauf P, Carvalho AC, et al., GeoSentinel doi:10.1001/archinte.166.15.1642 Surveillance Network. Travel-associated sexually transmitted 9 O'Brien D, Tobin S, Brown GV, et al. Fever in returned travelers: infections: an observational cross-sectional study of the GeoSentinel review of hospital admissions for a 3-year period. Clin Infect Dis surveillance database. Lancet Infect Dis 2013;13:205–13. 2001;33:603. doi:10.1086/322602 doi:10.1016/S1473-3099(12)70291-8

BMJ Global Health Initiatives 25 Short research report

Prevalence and factors associated with the use of ­antibiotics in non-bloody diarrhoea in children under 5 years of age in sub-Saharan Africa Asa Auta,1 Brian O Ogbonna,2 Emmanuel O Adewuyi,3 Davies Adeloye,4,5 Barry Strickland-Hodge6

1 For numbered affiliations see ABSTRACT every year. Most of these episodes and deaths occur end of article. Objectives among children in Africa with about 440 million cases Correspondence to: Dr Asa To estimate the prevalence and determine the and 350 000 deaths annually.2 Auta, School of Pharmacy and factors associated with the use of antibiotics in the Biomedical Sciences, University The appropriate treatment of diarrhoea is simple, yet of Central Lancashire, Preston management of non-bloody diarrhoea in children it remains a problem in many low-income and middle- PR1 2HE, UK; [email protected] under 5 years of age in sub-Saharan Africa (SSA). income countries (LMICs). Recent reviews of diarrhoea Additional material is published Methods management in children in LMICs have revealed a high online only. To view please visit the journal online. We conducted a meta-analysis of demographic and degree of inappropriate practice including excessive 3–5 Cite this as: Auta A, Ogbonna health survey data sets from 30 countries in SSA. fluid curtailment and antibiotic prescribing. The BO, Adewuyi EO, et al. Arch Pooled prevalence estimates were calculated using existing WHO guideline recommends the use of oral Dis Child Epub ahead of print: random effects model. 2Χ tests were employed to rehydration solution (or an intravenous electrolyte 19 March 2018. doi:10.1136/ archdischild-2017-314228 determine the factors associated with the antibiotic solution in cases of severe dehydration) as well as use. zinc supplementation and continued feeding for the Received: 4 October 2017 6 Revised: 2 March 2018 Results treatment and management of diarrhoea in children. Accepted: 7 March 2018 The pooled prevalence of antibiotic use among cases The guideline only recommends the use of antibiotics Republished with permission of non-bloody diarrhoea in children under 5 years in cases of bloody diarrhoea, suspected cholera or from Archives of Disease of age was 23.1% (95% CI 19.5 to 26.7). The use of associated sepsis. in Childhood Published 19 March 2018; doi:10.1136/ antibiotics in children with non-bloody diarrhoea in The inappropriate use of antibiotics in children with archdischild-2017-314228 SSA was associated with (p<0.05) the source of care, diarrhoea can result in the development of antibiotic place of residence, wealth index, maternal education resistance. In addition, the majority of antibiotics can and breastfeeding status. increase the risk of diarrhoea because of their effect 7 Conclusion on gut microflora. Very few studies have focused on We found an unacceptably high use of antibiotics to understanding the extent of antibiotic use in episodes treat episodes of non-bloody diarrhoea in children of non-bloody diarrhoea in children in sub-Saharan 8 under the age of 5 in SSA. Africa (SSA). Rogawski et al reported that 48.2% and 21.5% of cases of non-bloody diarrhoea in Haydom in Tanzania and Venda in South Africa, respectively, were Introduction treated with antibiotics. Also, Opondo et al9 found that Diarrhoea is one of the leading causes of morbidity and 64.5% of cases of non-bloody diarrhoea in children mortality in children under 5 years of age worldwide were treated inappropriately with antibiotics in eight with about 1.7 billion episodes and 578 000 deaths district hospitals in Kenya. However, these studies were not nationally representative and did not explore What is already known on this topic the factors associated with the use of antibiotics in children with non-bloody diarrhoea. We conducted The appropriate treatment of diarrhoea is simple, yet it remains a problem in a meta-analysis of demographic and health survey manyv low-income and middle income countries. (DHS) data sets from 30 countries in SSA to determine Most cases of non-bloody diarrhoea in children are self-limiting and are caused the prevalence and factors associated with the use of by microorganisms which are not susceptible to antibiotic therapy. antibiotics in children with non-bloody diarrhoea. Use of antibiotics to treat non-bloody diarrhoea in children increases the risk of adverse effects and the development of resistant bacteria. Methods What this study adds Data source We utilised nationally representative data to estimate the prevalence of We conducted a meta-analysis of DHS data on the antibiotic use in episodes of non-bloody diarrhoea in children under the age of 5 treatment of non-bloody diarrhoea with antibiotics in sub-Saharan Africa (SSA). in children under the age of 5 in 30 countries in SSA. DHS are nationally representative household surveys We found that over one in five cases of non-bloody diarrhoea were treated with conducted by ICF Macro/MEASURE DHS on behalf of antibiotics in SSA. national ministries of health with financial support We highlight the need to educate prescribers and parents in SSA on appropriate from many international partners including the United management of diarrhoea and the consequences of inappropriate use of States Agency for International Development.10 The antibiotics in children. standard DHS uses identical methodology including

26 BMJ Global Health Initiatives Short research report

Fig 1 | Meta-analysis for the prevalence of antibiotic use among cases of non-bloody diarrhoea

the probability sampling strategy and survey households, type and geographical location of child’s instrument to collect data that are comparable across residence. countries. We employed a random effects meta-analysis to Our study only included country data sets that calculate pooled prevalence estimates of the use of were collected from 2000 to 2016 and contained antibiotics in children with diarrhoea. A random disaggregated data on the type of diarrhoea—bloody effects meta-analysis was used because it allows for and non-bloody diarrhoea in children under the age of heterogeneity across studies. Despite the similarity 5. The data sets of 38 countries in SSA were available of the DHS study design across countries, we from DHS programme website. Of these 38 data sets, expected heterogeneity due to differing population 30 met the inclusion criteria. Details of the included parameters including geographical distribution and countries are contained in figure 1. socioeconomic conditions in different countries and regions of SSA. A test of heterogeneity of the DHS Data analysis data obtained for the different countries showed a All DHS data sets were downloaded with permission high level of inconsistency (I2 >50%) thereby agreeing from the DHS programme website and the data were with our decision to use the random effects model in analysed using Stata V.14 and Microsoft Excel 2016. our analysis. Furthermore, we performed sensitivity The variables from the DHS data sets extracted and analysis by excluding from our analysis one country included in our analysis were prevalence of children data at a time and the impact of excluding the data with diarrhoea; type of diarrhoea, whether bloody or was evaluated on the summary results. This was done non-bloody; proportion of children treated/untreated; to examine the effect of outliers and test the robustness proportion of children who were treated for diarrhoea of our findings. and type and sources of treatment. Other variables Subgroup analyses were performed to determine included were the sociodemographic characteristics of whether factors such as sex, age, type of residence, children and their households including age and sex wealth index, education and sources of care were of child, mother’s educational level, wealth index of associated with the use of antibiotics in children under

BMJ Global Health Initiatives 27 Short research report

Table 1 | Prevalence of antibiotic use in non-bloody diarrhoea by demographic category CI 55.9% to 61.3%) and Sierra Leone 47.1% (95% Number of ­children CI 43.9% to 50.3%) (figure 1). The regional estimate with non-bloody Prevalence estimate Test for associa- was lowest in east Africa, 18.7% (95% CI 13.9% to Category diarrhoea % (95% CI) tion or trend 23.6%) and highest in central Africa, 27.6% (95% CI Sex of child 16.9% to 38.3%). The higher estimate in central Africa Male 18 268 23.5 (19.9 to 27.1) X2=1.778, Female 16 809 22.9 (16.2 to 39.8) p=0.182 was due to the contribution of Congo-Brazzaville. The Age of child sensitivity analysis conducted by excluding the Congo- 0 9307 21.1 (17.6 to 24.5) X2=0.018, Brazzaville data yielded a pooled estimate of 21.8% in 1 11 395 25.0 (21.2 to 28.8) p=0.893 the central region, which is comparable to southern 2 7040 22.9 (18.9 to 26.9) African region. The SSA pooled estimate obtained 3 4413 22.1 (18.0 to 26.2) following the sensitivity analysis (21.9%; 95% CI 4 2922 22.0 (17.8 to 26.2) 18.8% to 24.9%) was comparable to our previous Currently breastfeeding No 13 160 24.1 (20.0 to 28.2) X2=13.485, estimate. Yes 21 917 22.4 (18.9 to 26.0) p<0.001 Table 1 summarises the data based on the subgroup Type of residence analyses performed. The subgroup analyses suggested 2 Urban 10 771 26.7 (22.5 to 31.0) X =109.137, that the use of antibiotics in children with non-bloody p 0.001 Rural 24 306 21.6 (18.2 to 25.1) < diarrhoea in SSA was significantly associated with Wealth index* 2 (p<0.05) the source of care, type of residence, wealth Poor 16 596 20.7 (16.9 to 24.5) X =176.047, Middle 6789 22.0 (18.6 to 25.5) p<0.001 index, maternal education and breastfeeding status. Rich 11 668 27.6 (23.6 to 31.6) The results revealed that antibiotics were commonly Mother’s highest level of education* used in children who sought advice or treatment from No education 14 740 20.5 (17.2 to 23.8) X2=184.584, private hospitals/clinic and community pharmacies Primary 12 360 24.0 (20.4 to 27.7) p<0.001 with pooled estimates of 41.1% (95% CI 34.1% Secondary 7352 27.9 (23.3 to 32.4) to 47.4%) and 41.8% (95% CI 34.8% to 48.9%), Higher 611 33.0 (26.1 to 39.8) Sources of care* respectively. Government hospitals 3131 38.5 (32.3 to 44.8) X2=3447.205, Government health centres 6184 35.2 (29.9 to 40.4) p< 0.001 Discussion Government health posts/ 2603 30.9 (24.4 to 37.5) The main contributions of our study lie in the use of dispensaries nationally representative data, the comprehensiveness Private hospitals/clinics 1347 41.1 (34.1 to 47.4) Community pharmacies 1391 41.8 (34.8 to 48.9) of the factors explored and the application of meta- Shops 2321 33.6 (25.5 to 41.7) analysis to provide pooled estimates on the prevalence Markets 1236 21.6 (6.9 to 36.3) of antibiotic use in episodes of non-bloody diarrhoea Traditional practitioners 1011 5.0 (3.6 to 6.5) in children under the age of 5 in SSA. We found a Others 1626 16.8 (11.8 to 21.9) high use of antibiotics to treat episodes of non-bloody Did not seek treatment from 14 189 9.3 (7.3 to 11.2) diarrhoea in children under the age of 5 in SSA: any provider n=35 077. over one in five cases of non-bloody diarrhoea were *Category with some missing data. treated with antibiotics. Most cases of non-bloody diarrhoea in children are self-limiting and caused by microorganisms which are not susceptible to antibiotic therapy such as rotavirus or for which the 5 years of age in SSA. The Χ2 tests for association or efficacy of antibiotic therapy is somewhat uncertain where appropriate for trend were calculated and results such as in campylobacter infections.9 11 Up to 13% were considered statistically significant at

28 BMJ Global Health Initiatives Short research report

diarrhoea. Despite these limitations, this study 1 Liu L, Oza S, Hogan D, et al. Global, regional, and national causes of child mortality in 2000–13, with projections to inform provides additional insight into the management of post-2015 priorities: an updated systematic analysis. Lancet diarrhoea in SSA and could prompt appropriate health 2015;385:430–40. system response. 2 Walker CL, Rudan I, Liu L, et al. Global burden of childhood pneumonia and diarrhoea. Lancet 2013;381:1405–16. AUTHOR AFFILIATIONS 3 Carvajal-Vélez L, Amouzou A, Perin J, et al. Diarrhea 1 School of Pharmacy and Biomedical Sciences, University of Central management in children under five in sub-Saharan Africa: Lancashire, Preston, UK does the source of care matter? A countdown analysis. 2Department of Clinical Pharmacy and Pharmacy Management, BMC Public Health 2016;16:830. Faculty of Pharmaceutical Sciences, Nnamdi Azikiwe University, 4 Perin J, Carvajal-Velez L, Carter E, et al. Fluid curtailment during Awka, Nigeria 3Statistical and Genomic Epidemiology Laboratory, Institute of Health childhood diarrhea: a countdown analysis. BMC Public Health and Biomedical Innovation, Queensland University of Technology, 2015;15:588. Brisbane, Australia 5 Diallo AF, Cong X, Henderson WA, et al. Management of childhood 4Nigerian Urban Reproductive Health Initiative, Abuja, Nigeria diarrhea by healthcare professionals in low income countries: an integrative review. Int J Nurs Stud 2017;66:82–92. 5Johns Hopkins Centre for Communication Programs, Baltimore, 6 World Health Organization. The treatment of diarrhoea: a manual for Maryland, USA 6 physicians and other senior health workers. 2005 http://​apps.​who.​ Faculty of Medicine and Health, School of Healthcare, University of int/iris/​ bitstream/​ 10665/​ 43209/​ 1/​ 9241593180.​ pdf​ (accessed 02 Leeds, Leeds, UK Oct 2017). The authors gratefully appreciate ICF International, USA, for providing 7 Diniz-Santos DR, Silva LR, Silva N. Antibiotics for the empirical the NDHS data sets for this study. treatment of acute infectious diarrhea in children. Braz J Infect Dis 2006;10:217–27. Contributors: AA conceived and designed the study. All authors 8 Rogawski ET, Platts-Mills JA, Seidman JC, et al. Use of antibiotics in oversaw its implementation. AA and DA planned and carried out children younger than two years in eight countries: a prospective the analyses. AA, BO and EO wrote the initial draft and all authors cohort study. Bull World Health Organ 2017;95:49–61. contributed writing to subsequent versions of the manuscript. All 9 Opondo C, Ayieko P, Ntoburi S, et al. Effect of a multi-faceted quality authors reviewed the study findings, read and approved the final improvement intervention on inappropriate antibiotic use in children version of the manuscript before submission. with non-bloody diarrhoea admitted to district hospitals in Kenya. Funding: The authors have not declared a specific grant for this BMC Pediatr 2011;11:109. research from any funding agency in the public, commercial or not-­for- 10 Measure DHS. Methodology: survey process. http://​dhsprogram.​com/​ profit sectors. What-We-​ Do/​ Survey-​ Process.​ cfm​ (accessed 16 May 2017). 11 O’Ryan G M, Ashkenazi-Hoffnung L, O’Ryan-Soriano MA, et Competing interests: None declared. al. Management of acute infectious diarrhea for children Provenance and peer review: Not commissioned; internally peer living in resource-limited settings. Expert Rev Anti Infect Ther 2014;12:621–32. reviewed. 12 Liu J, Platts-Mills JA, Juma J, et al. Use of quantitative molecular © Article author(s) (or their employer(s) unless otherwise stated in diagnostic methods to identify causes of diarrhoea in the text of the article) 2018. All rights reserved. No commercial use is children: a reanalysis of the GEMS case-control study. Lancet permitted unless otherwise expressly granted. 2016;388:1291–301.

BMJ Global Health Initiatives 29 Short report

Infectious disease outbreaks: how online clinical decision ­support could help Kieran Walsh

BMJ Publishing Group, London ABSTRACT However, it is not clear how doctors might use WC1H 9JR, UK This paper describes an evaluation of how doctors clinical decision support in the context of an epidemic. Correspondence to: might use an online clinical decision support tool They will likely be busy during an epidemic—this Dr Kieran Walsh BMJ Publishing Group, London to improve the care that they would provide to might mean that they will want even faster answers. WC1H 9JR, UK; kmwalsh@ bmj. patients with infectious disease and what features Medical advice can change continually throughout com they would expect in such a clinical decision support an epidemic—this might mean that they would place Additional material is published tool. Semistructured interviews were conducted by even more value on updated content. In this paper, online only. To view please visit telephone with doctors to evaluate the utility of a the journal online. we describe an evaluation of how doctors might use a clinical decision support tool in helping them to Cite this as: Walsh K. BMJ clinical decision support tool—BMJ Best Practice—to Stel Epub ahead of print: improve the care that they would provide to patients improve the care that they would provide to patients Jul 21, 2018 . doi:10.1136/ with infectious disease and to assess the features with a rare infectious disease and what features they bmjstel-2018-000368 that they would value in such a tool. The doctors were would expect in this tool. Received: 14 June 2018 primarily interested in how they could use the tool to Revised: 25 June 2018 improve care. They were short of time and so needed Accepted: 30 June 2018 to be able to access the content that they needed Methodology Republished with permission really quickly. They expected content that was both BMJ Best Practice is the online clinical decision support from BMJ Simulation and tool of the BMJ. Doctors who are part of a user group at Technology Enhanced evidence based and current, and they used a range of Learning Published 21 devices to access the content. They used desktops, BMJ were asked by email to take part in an evaluation July 2018; doi:10.1136/ laptops, mobiles and sometimes mobile apps. as to the utility of BMJ Best Practice in helping them to bmjstel-2018-000368 Doctors view the utility of clinical decision support improve the care that they would provide to patients in the management of rare infectious diseases from with a rare infectious disease and as to what features a number of perspectives. However, they primarily they would expect in BMJ Best Practice. Those who see utility in the tools as a result of their capacity to responded to the email were asked to use BMJ Best improve clinical practice in infectious diseases. Practice to help them solve a clinical problem involving a patient who had been working in Liberia and who had Introduction clinical features and risk factors that were suggestive of Ebola or its differentials (ie, other serious infectious Infectious diseases are a threat to human health diseases). Semistructured interviews were then around the world. In recent years, there have been conducted by telephone with the doctors according to a growing number of outbreaks of rare infectious a schedule which was founded on existing literature in diseases. These include outbreaks of Ebola, Zika and this field.3 4 Interviews lasted approximately 20 min. influenza.1 2 These infections pose special challenges Detailed field notes were taken during the interviews. to learners and educators alike in medical education. Notes were analysed using thematic analysis.5 This Under normal conditions they are rare and so it is was used to allow new concepts and themes to difficult to dedicate too much time or resources to emerge from the data. Interviews were continued them in undergraduate or postgraduate curricula. until data saturation was achieved—this was defined However, the situation can quickly change to that of an as two successive interviews in which no new themes epidemic. In these circumstances, doctors and other emerged.6 healthcare professionals need instant education and support. It is in these circumstances that online clinical decision support could play a major role in controlling Results outbreaks of infectious diseases. Eighteen learners took part in the semistructured Online clinical decision support provided must be interviews. Four key themes emerged from the aligned to the needs of the healthcare professional interviews. learners.3 In this regard, it is clear that, under normal circumstances, healthcare professionals need certain Theme one: improving clinical practice as a result of features in clinical decision support. They need clinical decision support decision support tools that provide evidence-based The learners were primarily interested in how they knowledge and that will also work at the point of could use the tool to improve care. They felt that the care. The clinical decision support also needs to be tool would change their practice or at the very least continually updated. Clinical decision support must be confirm that their practice was correct. The learners intuitive to use and must fit with the clinical workflow were most interested in improving their practice with of the healthcare professional. They must also work on regard to diagnosis and differential diagnosis. Some whatever device the healthcare professional is using— users felt that it would also help them improve their this might be on a desktop, mobile device or an app. clinical management—for example, in isolating the

30 BMJ Global Health Initiatives Short report

Box 1: Themes from the interviews Theme 3: content needs to be evidence based and continually updated Theme 1: improving clinical practice as a result of clinical decision support The learners expected content that was both evidence Learner comments based and current. This was especially true of the ‘I found it quite directive and it would have helped me in clinical care. I would have sections that guide treatment. Learners were concerned known what to do in management—or at least the first steps. Having the steps laid out about the currency of the content as they realised that in algorithms is helpful’. this can be a rapidly changing field of healthcare— ‘It would definitely change my practice—if there was a need to change but often it is particularly in the midst of an outbreak. The learners more about confirming that my practice is correct which is reassuring and worthwhile realised that their patients would be looking things up also’. and so wanted to be as least as updated as their patients ‘I found the resource fairly easy to use. I quickly made the diagnosis of Ebola using it’. were. Different learners had different views on exactly ‘I put fever headache and the other symptoms into the search box and got sensible answers—like Ebola and influenza’. what currency of knowledge meant. Some thought that ‘I entered the data into the search and came up with Lassa, Ebola and Marburg. The it meant updated within the last 2 years; some thought search box worked’. that it meant the content should be updated more frequently; and some thought that it depended on Theme 2: Time is of the essence the precise context. Learners realised that sometimes Learner comments there was a weak underlying evidence base and that ‘I am always quite short of time so I am not sure that I would have much time to read in these circumstances expert consensus opinion was the entire content or even a section of it. It needs to be easy to read and search and the next best alternative. Learners appreciated lots of find’. references and links to other resources—they felt that ‘Speed of using the resource is vital—seconds do count when you are short of time’. this would add credibility. The learners did not have Theme 3: Content needs to be evidence based and continually updated time to look at all the evidence themselves and so Learner comments needed to be able to trust the provider to do this for ‘I would expect the content to be continually updated—especially the treatment them. Comments related to this theme are shown in sections as this changes all the time. So, the latest research should be taken into box 1. consideration in the content—that would be highly valued by me’. Some learners pointed out the fact that there can ‘The evidence base of the resources is very important to me. The content must be be a conflict between the currency and the evidence- validated. Lots of references add credibility. Some guidelines are too long so it is based credentials of a resource—in that the most recent great to have short summaries but these summaries must be evidence based’. updates to content might not be completely evidence Theme 4: content needs to be accessible based. Learner comments ‘I used my laptop to look it up in this case—but I usually use my desktop computer. I Theme 4: content needs to be accessible didn’t notice that there was an app—but an app is great as I am often offline’. Learners used a range of devices to access the content. ‘Being able to use the content on a mobile phone is great. A downloadable app is also They used desktops, laptops, mobiles and sometimes valuable as there are lots of environments where you can’t get any type of signal’. mobile apps. Some of the users did not always have access to a desktop computer or to a wifi network and so especially appreciated the mobile app. They patient. However, there was concern that the advice had high expectations of the interfaces and thought from a clinical decision support tool in management that they should have an equally good experience might not be compatible with that of a local hospital regardless of the device that they were using. Some guideline. After using the tool to solve the clinical learners used different devices to access the content— scenario, 17 of the 18 learners correctly thought that depending on the context in which they were using Ebola or another viral haemorrhagic fever was the most the site. Many users commented that in the future they likely diagnosis or a priority differential that needed to expected to use their mobile more frequently and that be ruled out. younger generations of learners were more likely to Comments related to this theme are shown in box 1. want to use the mobile site. Comments related to this theme are shown in box 1. Theme 2: time is of the essence The learners were short of time and so needed to be able to access the content that they needed really Conclusions quickly. They needed content that was concise and Learners view the utility of BMJ Best Practice in the easy to search and navigate. Learners needed to have management of rare infectious diseases from a number answers to their questions ideally within seconds of perspectives. They see utility in the tool as a result and at most within minutes. The need for immediate of their ability to improve clinical practice in infectious access to clinical knowledge partly depended on the diseases—however, they feel that they will only be exact context that they were acting in. Sometimes this able to exploit clinical decision support if the content was when they were with a patient—in which case is evidence based and continually updated and they needed an answer very quickly. But sometimes accessible in a timely manner in a range of different they were looking things up after they had seen the clinical devices. patient—in this circumstance they had more time. There are limitations to this evaluation. All the Comments related to this theme are shown in box 1. learners were users of online resources—they may

BMJ Global Health Initiatives 31 Short report

not represent users who do not use online resources. Contributors: KW conceived and carried out this work; wrote up the However, the growing ubiquity of online resources manuscript and approves this final version. He is accountable for all aspects of the work. means that the number of learners who do not use Funding: The authors have not declared a specific grant for this these resources grows smaller year by year. All were research from any funding agency in the public, commercial or not-for- users of a single clinical decision support tool: once profit sectors. again they may not be representative of users of other Competing interests: KW works for BMJ, which produces BMJ Best Practice. This provides clinical decision support in a range of tools. This was a small evaluation—however, it used a infectious and non-infectious diseases. protocol that allowed the operationalisation of data Provenance and peer review: Not commissioned; internally peer saturation to achieve an adequate sample size. reviewed. In recent years, infectious disease outbreaks © Author(s) (or their employer(s)) 2018. No commercial re-use. See have increased in frequency and severity. In the rights and permissions. Published by BMJ. future it will be impossible to educate all healthcare 1 Beeching NJ, Fenech M, Houlihan CF, et alBMJ 2014;349:g7348. professionals to manage infectious disease outbreaks 2 Petersen LR, Jamieson DJ, Powers AM, et alN Engl J Med using conventional face to face means. New methods 2016;374:1552–63. 3 Kwag KH, González-Lorenzo M, Banzi R, et al. Providing doctors such as clinical decision support tools will be with high-quality information: an updated evaluation of web- needed. However, for these tools to be used, they will based point-of-care information summaries. J Med Internet Res need to be current, evidence based and accessible 2016;18:e15. 4 Beeler PE, Bates DW, Hug BL. Clinical decision support systems. and will need to be designed to help doctors achieve Swiss Med Wkly 2014;144:w14073. their primary outcome—improving the care that 5 Joffe H, Yardley L. Content and thematic analysis. In: Marks they deliver to patients. They will also need to be DF, Yardley L, eds. Research methods for clinical and health quick to access—time is of the essence when caring psychology. London: Sage Publications, 2004:56–68. 6 Francis JJ, Johnston M, Robertson C, et al. What is an adequate for patients during a serious outbreak of infectious sample size? Operationalising data saturation for theory-based disease. interview studies. Psychol Health 2010;25:1229–45.

32 BMJ Global Health Initiatives Analysis

Emerging and re-emerging infectious disease threats in South Asia: status, vulnerability, preparedness, and outlook Without investment in surveillance and early detection the region remains vulnerable to infectious disease threats, say Buddha Basnyat and colleagues

outh Asia, despite decreas- Although there are frequent reports virus, while 41% have been infected with ing rates of infectious disease, of sporadic cases of suspected emerging chikungunya.2 7 Although the majority accounts for a significant pro- infectious disease syndromes and limited of dengue infections are inapparent, the portion of their global burden. outbreaks of emerging infections such majority of chikungunya infections appear The sub-continent is also in the as Nipah virus, Chandipura virus, and to be symptomatic.7 In many countries, Smidst of rapid economic growth; large Crimean-Congo Haemorrhagic Fever dengue, which was once a childhood scale changes in land use, access to water (CCHF), South Asia has not in recent history infection, is increasingly seen in adults.8 and sanitation, and agricultural produc- experienced a large outbreak of an emerging The incidence of chikungunya in South tion; environmental degradation; and infection. However, factors associated with Asian countries is lower than dengue and technological transformation, all against vulnerability (Table 1) to the emergence of since there is only one serotype, people do a background of uneven health system infectious diseases—such as population not experience repeated infections.7 Almost capacity. South Asia, defined by the World density, national and international travel, all chikungunya infections, however, are Bank as Afghanistan, Bangladesh, Bhutan, bio-diversity, land use change, zoonotic symptomatic and some people develop India, Maldives, Nepal, Pakistan, and Sri reservoirs, weak healthcare and public disabling polyarthritis which can last for Lanka, is home to a quarter of the world’s health systems, and deficiencies in water several months.9 Co-infection with these population. Existing infectious disease and sanitation—indicate that South Asia is two viruses appears to occur often, possibly challenges—including tuberculosis, HIV, at high risk. Preparedness and the ability as both viruses are transmitted by the same and malaria—have been augmented by to detect and respond to a disease outbreak vector Aedes aegypti.9 It is concerning that emerging and growing threats such as den- are critical for national, regional, and global the Zika virus is also transmitted by the gue, chikungunya, healthcare associated health security. same vector, which is abundant in all South infections, and antimicrobial resistance. Asian countries. If Zika is introduced to the These emerging and re-emerging infec- Current status of emerging and epidemic region it is likely to spread quickly, cause tious disease challenges threaten to create infections complications in pregnancies, and add economic disruption and potentially large We have focused on the diseases below as to the burden of neurological infections morbidity and mortality burdens. Here we they include the most important emerging caused by other flavi-viruses such as review the status, vulnerability, and pre- and re-emerging illnesses in South Asia. the West Nile virus and the Japanese paredness for emerging and re-emerging Enteric fever is often diagnosed in patients Encephalitis virus. infectious diseases and describe the state of with fever but is increasingly difficult to preparedness and surveillance for threats treat with fluoroquinolones. Zoonotic infections such as Zika, Middle East respiratory syn- South Asia has been identified as a hot spot drome coronavirus (MERS-CoV), and avian Vector borne viral infections for the emergence of zoonotic infectious influenza. The main burden of vector borne viral infec- diseases.10 11 The endemic zoonoses have tions in the region is attributable to dengue re-emerged or emerged in newer areas or and chikungunya, while Zika virus is also with newer clinico-epidemiological presen- Key messages likely to emerge. Of the 390 million dengue tations, often with more serious manifesta- • South Asia accounts for a significant infections that are estimated to occur annu- tions. proportion of the global burden of ally worldwide, over 70% occur in South Livestock may act as intermediate 1 infection diseases, although in recent Asia. amplifying hosts, facilitating the transfer history it has not experienced a large Although sporadic cases of dengue of pathogens from their normal ecological outbreak of an emerging infection. infection were seen in many South Asian niche into humans. Examples are countries in the 1960s, regular epidemics • The region remains seriously vul - Japanese encephalitis (JE), Nipah virus, only occurred in the early 1990s in India and Crimean-Congo haemorrhagic fever nerable to existing and new threats 2 including Zika, Ebola, MERS-CoV, and and Sri Lanka. Dengue emerged in (CCHF). JE has spread to newer areas in avian influenza. epidemic proportions in Bangladesh in the subcontinent against a backdrop of 2000,3 in Pakistan in 2006,4 in Nepal a large proportion of undiagnosed cases Surveillance and preparedness for • in 2010,5 and more recently in Bhutan of acute encephalitis syndrome.12 JE has early detection of outbreaks is crucial in 2013.6 In India and Sri Lanka, by become endemic to the Kathmandu valley in this region inhabited by one fourth around 40 years of age 90% to 95% of region after its introduction in early 2000.13 of the world’s population. adults have been infected with the dengue Nipah virus emerged in Malaysia in the

BMJ Global Health Initiatives 33 Analysis

Table 1 | Key areas of vulnerability to emerging infectious diseases in South Asia Factor Situation in South Asia Association with vulnerability to emerging and epidemic infections Population size and South Asia is home to one quarter of world’s population, with Absolute population size and the intensity of contact between people are density Bangladesh and India being amongst the most densely populated key determinants—along with transmissibility of the infectious agent and countries in the world the susceptibility of the population to infection—that determine the scale of an outbreak of an infection that is transmitted from person to person Land use The rate of land use change in most of South Asia is now slow. Changes in land use may alter ecosystems and the interaction of animal Where land can be agriculturalised it has been, and much of the hosts with humans, giving rise to new opportunities for amplification and / land is already extensively cropped. The rate of forest growth is or spill over to humans56 positive in India, there being net reforestation. Biodiversity57 South Asia is not especially bio-diverse but India may be a hot The impact of declining biodiversity is variable, and may act to either spot of bat to human virus sharing 58 increase or decrease the emergence of infectious diseases57 Insect and tick vectors59 Mosquito and tick vectors are widely present for some serious As Zika virus has shown, the presence of competent vectors can lead to infections60-62 dramatic introductions and transmission of pathogens Livestock density Cattle and goats are raised in large numbers across South Asia. Livestock may act as intermediate hosts for a range of zoonotic infections As incomes increase the demand for meat will increase and including CCHF, fascioliasis, bovine TB, brucellosis, and leptospirosis livestock farming will intensify from cattle; and fascioliasis, brucellosis, Orf virus, and Q fever from sheep. Rift Valley fever could be introduced successfully to South Asia63 Poverty and human South Asia has had a period of sustained economic growth, Poverty is a risk factor for almost every infection, but is a particular development index declining poverty rates, and improved human development. vulnerability for epidemic infections. Poverty is associated with crowding, Nevertheless the region is home to a very large number of poor poor sanitation, poor nutrition, and poor access to preventive, diagnostic, people with poor infrastructure and therapeutic healthcare. It is no coincidence that Ebola raged out of control in three of the poorest countries in Africa Healthcare systems The public healthcare systems in South Asia are poor, with patchy Healthcare systems can contribute to the emergence and transmission of coverage, limited resources, overcrowding, and inadequate infectious diseases in several ways: infrastructure • Resistance to antibacterial, antiviral, and anti-malarial drugs can be promoted by poor use of these drugs • Healthcare settings can amplify infections. The introduction and transmission of MERS-CoV in Korea is a good example

late 1990’s, initially being misdiagnosed deaths, while the rhetoric on elimination outbreaks of scrub typhus in Nepal were as JE, and there have since been frequent continues to grow.27 28 Brucellosis, bovine mistakenly diagnosed as enteric fever and outbreaks in Bangladesh and, to a lesser tuberculosis, and a range of food borne suboptimal treatment with ceftriaxone extent, India. CCHF has recently been diseases contribute to the morbidity and administered.35 36 Finally, even where blood recognised in humans in South Asia for the mortality attributable to zoonotic infections culture facilities are available, because first time.14 15 but are struggling to gain the attention of of the paucity of the typhoid organism in The highly pathogenic avian influenza policy makers in the subcontinent,29 30 the blood, blood culture growth is usually (HPAI) virus A/H5N1, which was despite resulting in 150 million illnesses, restricted to about 50%. So without proper introduced to the subcontinent in 2005 175 000 deaths, and 12 million disability diagnostics, it is hard not only to prescribe through wild birds, has since become adjusted life years.31 appropriate treatment but also to estimate endemic across large parts of north east the true burden of typhoid fever in South India and Bangladesh, across porous Enteric fever and antimicrobial resistance Asia. international borders. It has resulted in Enteric (typhoid) fever is caused by bac- In recent years, the appearance in South losses of around US$500 million (£398m; teria Salmonella typhi and Salmonella Asia of fluoroquinolone resistant H 58 16-18 32 €460m). Rates of neuroleptospirosis paratyphi. S typhi is the most common typhoid organisms has made the treatment and leptospirosis have been rising in bacterial organism grown in blood cultures of this disease even more challenging.37 north India and Sri Lanka and have been in South Asia. These organisms cause indis- associated with disseminated intravascular tinguishable clinical features that generally Polio coagulation.19 20 The growing proportion comprise of high fever for at least 3 days The elimination of polio from many coun- 33 of severe cases of leptospirosis has created with no localising signs. tries in the region, most recently in India, massive pressures on the healthcare Typhoid fever is one of the most common is a landmark achievement. However, two delivery systems in affected countries.21 diagnoses in cases of fever in the region. out of the three remaining countries in the Scrub typhus, which has been grossly Blood cultures (the gold standard for the world with endemic polio are in South Asia. under reported in South East Asian diagnosis of typhoid fever) are not readily There were only 74 wild type polio cases in countries,22 is now increasingly being seen available in most, especially rural, regions Pakistan and Afghanistan in 2015, and 33 in newer ecological niches such as urban of South Asia. Instead the widely available in 2016. However, unrest in these countries landscapes.23-25 Anthrax is endemic in large Widal test, a slide agglutination test threatens progress in eradication and is a parts of South Asia. In the border areas of developed in 1896, is often used to make high priority for regional cooperation. India, Bangladesh, and Myanmar, poor a definitive diagnosis of enteric fever.34 vaccination and surveillance have been Unfortunately, it is difficult to establish a Vulnerabilities of South Asia accompanied by increasing anthrax cases, “cut off” for this test. The test also cross South Asia already has some of the most which prompted Bangladesh to announce a reacts with malaria parasites, rickettsial densely populated areas in the world and “red alert” in 2010.26 organisms, and dengue virus, all very is projected to see rapid shifts in urbanisa- Rabies remains endemic in eight common causes of undifferentiated febrile tion, population density, dietary patterns, countries in South East Asia, with illness in South Asia. and subsequent demand for animal source 1.4 billion people at risk. The region The use of the Widal test has sometimes foods and agriculture intensification—all contributes about 45% of global rabies had dire consequences, such as when of which are considered important drivers

34 BMJ Global Health Initiatives Analysis

of zoonoses.38 While these factors are also Preparedness establishing effective intersectoral indicators of economic growth, human and There is limited regional capacity to iden- coordination measures without adversely animal healthcare systems have failed to tify, respond to, and mitigate emerging reducing the core competencies of keep pace with the corresponding need for infectious disease threats in South Asia. participating agencies; devising acceptable, sectoral and intersectoral surveillance and The main reason seems to be a lack of polit- effective, and sustainable policies, coordination.39 ical will. While mechanisms for regional including trade laws, that tackle risks Institutional capacity for collective conversations do exist, they have without endangering livelihoods; and epidemiological and laboratory response, not translated into the kind of operational investment in human and animal healthcare especially at sub-national levels, remains capabilities that the European Centers capacity at individual (caregivers) and limited, more so in veterinary sectors. for Disease Prevention and Control or the systems (infrastructure) levels. ­Surveillance systems often are incomplete European Commission are able to facilitate At the global scale, South Asia must in their reach, fail to identify early warning within the European Union. Moreover, the become more engaged in the health security signals, and communicate poorly across composition of the World Health Organiza- agenda. The government of India is a founder sectors. Despite the visibility following tion’s South East Asia Regional Office leaves contributor to the Coalition for Epidemic emergence of HPAI, limited capacity out key countries (Afghanistan and Paki- Preparedness Innovation (CEPI), an initiative and infrastructure has kept wildlife stan) in the region, hindering cooperation that aims to accelerate the development of departments in most countries from being at WHO level. new vaccines for high threat pathogens. active participants in preparedness and Clearly civil societies and research This leadership is commendable, but such response efforts. communities in the region need to research and development pipelines needs The outbreak of MERS-CoV in South work together to tackle these risks and to be linked to strengthened surveillance, Korea in 2015 served as a warning about vulnerabilities and to lobby for more political response, and research platforms within the vulnerability to infectious disease commitment. Multiple approaches need South Asia to ensure that it can be evaluated outbreaks of overstretched, crowded, to be considered to overcome the limited and implemented locally. Finally, global and unprepared healthcare systems.40 preparedness of the region.48 science communities can help enhance Healthcare systems are variable, but Regional cooperative surveillance regional conversation to encourage local in South Asia there are the substantial programmes, such as those promoted collaboration. challenges of inadequate infrastructure, by the Connecting Organizations for Tackling these systemic shortcomings poor quality services, fragmented health Regional Surveillance (CORDS) initiative, needs a concerted approach that takes a information systems, and weak controls may strengthen regional surveillance and medium to long term view of outcomes. over the private healthcare sector.41 This preparedness. The South Asian Association Focus should be on building strong, poses a risk of the late detection, and for Regional Cooperation (SAARC) should intersectorally connected systems, with potentially explosive amplification, of be activated and work alongside inter- aligned policies driven by One Health, epidemic prone infections within the governmental agencies such as WHO, the as institutions come together to review healthcare system. Food and Agriculture Organization of the programmes and policies. Fundamentally, the current state of United Nations, and the World Organization affairs in the subcontinent flows from for Animal Health. Outlook severe policy neglect. The One Health The European Commission and Early detection of outbreaks is crucial for initiative is a worldwide, cross-sector the European Developing Countries their early control. Mathematical models approach to addressing vector borne and Clinical Trials Partnership (EDCTP) have have suggested that it might be possi- zoonotic diseases. While the world has supported the establishment of regional ble to contain an emerging pandemic of moved towards incorporating One Health clinical research networks for emerging avian influenza if detection and reporting in their policy discourse, South Asia has and epidemic prone infectious diseases of cases that suggest human to human been slow to adopt this approach, except in Europe, Latin America, and Africa. transmission happen within around for initial signs in Bangladesh.42 There A major regional funding initiative is three weeks of the first case.50 51 A report have been limited efforts to build One needed to establish a similar, sister in 2010 of 398 WHO notified outbreaks Health capacity in the subcontinent, clinical research network in South Asia that happened between 1996 and 2009 initiated mostly by external agencies,43 that can tackle the combined threats of found that only 7% occurred in the WHO but policymakers remain disengaged in the antimicrobial resistance and epidemic South East Asia region and that, over the absence of a convincing case made by the prone infections. However, especially in period ­studied, the timeliness of detec- research community.44 45 This disconnect the context of South Asia, public health tion had improved, although significant is further exemplified by the fact that research, including systems and policy delays in public notification remained.52 veterinary research has focused more on research, should not take a back seat to A 2016 update has not changed the over- increasing animal productivity, while clinical research. all picture.53 This implies that outbreaks neglecting zoonotic potential.46 The fragmentation of prevention in South Asia are less common than else- The trepidations experienced during measures and the need to break the where, since the WHO South East Asia the outbreaks of Ebola in West Africa and traditional governance silos of human region has around 27% of the world subsequently during Zika cases in South and animal health systems is a significant population yet only 7% of reported out- America is justified given the fact that, if challenge. Facets of the response may breaks. This is surprising given that the a rapidly spreading infection finds its way include establishing harmonised characteristics of South Asia show its vul- into South Asia, it could wreak havoc before international commitments to enforce nerability to emerging and epidemic infec- being brought under control.47 minimal assurances to confront zoonoses49; tions. We must not ignore this ­warning

BMJ Global Health Initiatives 35 Analysis

that outbreaks­ are less well detected and 3 Sharmin S, Viennet E, Glass K, Harley D. 22 Aung AKSD, Spelman DW, Murray RJ, Graves S. reported in this region. The emergence of dengue in Bangladesh: Rickettsial infections in Southeast Asia: implications epidemiology, challenges and future disease risk. for local populace and febrile returned travelers. Am The revised International Health Trans R Soc Trop Med Hyg 2015;109:619-27. J Trop Med Hyg 2014;91:451-60. doi:10.4269/ Regulations (IHR) whose aim is to help doi:10.1093/trstmh/trv067. ajtmh.14-0191. the international community prevent and 4 Pandey BD, Morita K, Khanal SR, et al. Dengue 23 Park S-WHN-Y, Ha NY, Ryu B, et al. Urbanization of virus, Nepal. Emerg Infect Dis 2008;14:514-5. scrub typhus disease in South Korea. PLoS Negl respond to acute public health risks that doi:10.3201/eid1403.070473. Trop Dis 2015;9:e0003814. doi:10.1371/journal. can cross borders was implemented in 5 Subedi D, Taylor-Robinson AW. Epidemiology of pntd.0003814. 2007. They appear to have had a positive dengue in Nepal: History of incidence, current 24 Wei Y, Huang Y, Luo L, Xiao X, Liu L, Yang Z. Rapid effect on detection and reporting, and prevalence and strategies for future control. J Vector increase of scrub typhus: an epidemiology and Borne Dis 2016;53:1-7. spatial-temporal cluster analysis in Guangzhou the latest data on IHR core capacities 6 Zangmo S, Klungthong C, Chinnawirotpisan P, et al. City, Southern China, 2006-2012. PLoS One (2013) shows good levels in South East Epidemiological and molecular characterization of 2014;9:e101976. doi:10.1371/journal. Asia.52 54 This indicated that the region is dengue virus circulating in Bhutan, 2013-2014. PLoS pone.0101976. Negl Trop Dis 2015;9:e0004010. doi:10.1371/ 25 Sethi S, Prasad A, Biswal M, et al. Outbreak well prepared. However the core capacity journal.pntd.0004010. of scrub typhus in North India: a re-emerging 55 indicators have been criticised. 7 Rodríguez-Barraquer I, Solomon SS, Kuganantham epidemic. Trop Doct 2014;44:156-9. Clearly, there has been insufficient work P, et al. The hidden burden of dengue and doi:10.1177/0049475514523761. on epidemic preparedness in South Asia. Chikungunya in Chennai, India. PLoS Negl Trop 26 Mondal SP, Yamage M. A retrospective study on Dis 2015;9:e0003906. doi:10.1371/journal. the epidemiology of anthrax, foot and mouth Unless immediate attention is given to pntd.0003906. disease, haemorrhagic septicaemia, peste des petits preparedness and ability to respond, the 8 Pang J, Leo YS, Lye DC. Critical care for dengue in ruminants and rabies in Bangladesh, 2010-2012. region remains vulnerable to existing and adult patients: an overview of current knowledge and PLoS One 2014;9:e104435. doi:10.1371/journal. future challenges. Curr Opin Crit Care 2016;22:485- pone.0104435. new threats including Zika, Nipah, MERS- 90. doi:10.1097/MCC.0000000000000339. 27 Gongal G, Wright AE. Human rabies in the CoV, and avian influenza. 9 Saswat T, Kumar A, Kumar S, et al. High rates of WHO Southeast Asia Region: forward steps for Contributors and sources: All the authors co-infection of Dengue and Chikungunya virus elimination. Adv Prev Med 2011; 1:383870. in Odisha and Maharashtra, India during 2013. doi:10.4061/2011/383870. contributed to the initial draft. RL put the segments Infect Genet Evol 2015;35:134-41. doi:10.1016/j. 28 Chowdhury F-RBA, Basher A, Amin MR, Hassan of the contribution together. PH outlined the table. meegid.2015.08.006. N, Patwary MI. Rabies in South Asia: fighting All the authors worked on and approved the final 10 Jones KE, Patel NG, Levy MA, et al. Global for elimination. Recent Pat Antiinfect Drug version. trends in emerging infectious diseases. Nature Discov 2015;10:30-4. doi:10.2174/157489 2008;451:990-3. doi:10.1038/nature06536.­ 1X10666150410130024. Competing interests: We have read and understood 11 Grace DMF, Ochungo P, Kruska R, et al. Mapping of 29 Falenski A, Mayer-Scholl A, Filter M, Göllner C, BMJ’s policy on declaration of interests and declare poverty and likely zoonoses hotspots. 2012. https:// Appel B, Nöckler K. Survival of Brucella spp. no conflict of interest. cgspace.cgiar.org/handle/10568/21161. in mineral water, milk and yogurt. Int J Food Provenance and peer review: Commissioned; 12 National Vector Borne Disease Control Program. Microbiol 2011;145:326-30. doi:10.1016/j. Japanese encephalitis. 2015. http://nvbdcp.gov.in/ ijfoodmicro.2010.11.033. externally peer reviewed je-new.html. 30 Thakur ASM, Katoch VC, Dhar P, Katoch RC. A study on Republished with permission from BMJ 13 Partridge J, Ghimire P, Sedai T, Bista MB, the prevalence of bovine tuberculosis in farmed dairy 2017;357:j1447 Banerjee M. Endemic Japanese encephalitis in cattle in Himachal Pradesh.Veterinary World, 2010. the Kathmandu valley, Nepal. Am J Trop Med Hyg 31 Coker RJHB, Hunter BM, Rudge JW, Liverani M, 1,2 3 Ramanan Laxminarayan, Manish Kakkar, 2007;77:1146-9. Hanvoravongchai P. Emerging infectious diseases Peter Horby,4 Gathsaurie Neelika Malavige,4,5 14 Yadav PDRC, Patil DY. D Majumdar T, Mourya DT. in southeast Asia: regional challenges to control. Buddha Basnyat4,6,7 Crimean-Congo hemorrhagic fever: current scenario Lancet 2011;377:599-609. doi:10.1016/S0140- in India. Proc Natl Acad Sci, India, Sect B Biol Sci 6736(10)62004-1. 1 Center for Disease Dynamics, Economics, and 2014;84:9-18doi:10.1007/s40011-013-0197-3. 32 Parry C, Basnyat B. Typhoid and paratyphoid Policy, Washington, DC, USA 15 Hussain B, Iqbal A, Abudakar M. Crimean-Congo fevers. In: Warrell D, Cox TM, Firth JD, eds. 2 hemorrhagic fever (CCHF): an emerging disease in Oxford Textbook of Medicine.5th ed. Oxford Princeton Environmental Institute, Princeton Pakistan. Vetenary Sciences: Research and Reviews University Press, 2010doi:10.1093/ University, Princeton, New Jersey, USA 2016;2:11-22. med/9780199204854.003.070608_update_002. 3Public Health Foundation of India, New Delhi, 16 World Health Organization. H5N1 avian influenza: 33 Maskey AP, Day JN, Phung QT, et al. Salmonella timeline of major events. 2014. www.who.int/ enterica serovar Paratyphi A and S. enterica serovar India influenza/human_animal_interface/avian_influenza/ Typhi cause indistinguishable clinical syndromes in 4Centre for Tropical Medicine and Global Health, H5N1_avian_influenza_update.pdf. Kathmandu, Nepal. Clin Infect Dis 2006;42:1247-53. Nuffield Department of Clinical Medicine, 17 Chakrabarti AKPS, Pawar SD, Cherian SS, et al. doi:10.1086/503033. Characterization of the influenza A H5N1 viruses 34 Olopoenia LA, King AL. Widal agglutination test - 100 University of Oxford, UK of the 2008-09 outbreaks in India reveals a years later: still plagued by controversy. Postgrad 5Faculty of Medical Sciences, University of Sri third introduction and possible endemicity. Med J 2000;76:80-4. doi:10.1136/pmj.76.892.80. Jayawardanapura, Sri Lanka PLoS One 2009;4:e7846. doi:10.1371/journal. 35 Basnyat B. Typhoid versus typhus fever in post- pone.0007846. earthquake Nepal. Lancet Glob Health 2016;4:e516- 6Oxford University Clinical Research Unit, Patan 18 Marinova-Petkova A, Feeroz MM, Rabiul Alam SM, 7. doi:10.1016/S2214-109X(16)30094-8. Academy of Health Sciences, Lalitpur, Nepal et al. Multiple introductions of highly pathogenic 36 Arjyal A, Basnyat B, Nhan HT, et al. Gatifloxacin 7 avian influenza H5N1 viruses into Bangladesh. versus ceftriaxone for uncomplicated enteric fever Global Antibiotic Resistance Partnership, Emerg Microbes Infect 2014;3:e11. doi:10.1038/ in Nepal: an open-label, two-centre, randomised Nepal emi.2014.11. controlled trial. Lancet Infect Dis 2016;16:535-45. 19 Sethi S, Sharma N, Kakkar N, et al. Increasing doi:10.1016/S1473-3099(15)00530-7. Correspondence to: buddha.basnyat@ndm. trends of leptospirosis in northern India: a 37 Ryan ET. Troubling news from Asia about treating ox.ac.uk clinico-epidemiological study. PLoS Negl Trop Dis enteric fever: a coming storm. Lancet Infect 1 Bhatt S, Gething PW, Brady OJ, et al. 2010;4:e579. doi:10.1371/journal.pntd.0000579. Dis 2016;16:508-9. doi:10.1016/S1473- The global distribution and burden of 20 Bandara M, Ananda M, Wickramage K, Berger E, 3099(15)00542-3. dengue. Nature 2013;496:504-7. Agampodi S. Globalization of leptospirosis through 38 Regmi ADJ. Effects of urbanization on global food doi:10.1038/nature12060. travel and migration. Global Health 2014;10:61. demand. 2013. https://naldc.nal.usda.gov/naldc/ 2 Jeewandara C, Gomes L, Paranavitane SA, et al. doi:10.1186/s12992-014-0061-0. download.xhtml?id=34108&content=PDF. Change in dengue and Japanese encephalitis 21 Schneider MCJM, Jancloes M, Buss DF, et al. 39 Chatterjee P, Kakkar M, Chaturvedi S. Integrating seroprevalence rates in Sri Lanka. PLoS One Leptospirosis: a silent epidemic disease. Int J Environ one health in national health policies of developing 2015;10:e0144799. doi:10.1371/journal. Res Public Health 2013;10:7229-34. doi:10.3390/ countries: India’s lost opportunities. Infect Dis Poverty pone.0144799. ijerph10127229. 2016;5:87. doi:10.1186/s40249-016-0181-2.

36 BMJ Global Health Initiatives Analysis

40 Cowling BJ, Park M, Fang VJ, Wu P, Leung GM, Wu JT. global surveillance and response to emerging B Biol Sci 2004;359:1049-58. doi:10.1098/ Preliminary epidemiological assessment of MERS- zoonotic diseases.National Academies Press, 2009: rstb.2004.1480. CoV outbreak in South Korea, May to June 2015. Euro 205-33. 57 Keesing F, Belden LK, Daszak P, et al. Impacts of Surveill 2015;20(25):7-13. 49 Sekar N, Shah NK, Abbas SS, Kakkar M. Roadmap to biodiversity on the emergence and transmission 41 Patel V, Parikh R, Nandraj S, et al. Assuring health Combat Zoonoses in India Initiative. Research options of infectious diseases. Nature 2010;468:647-52. coverage for all in India. Lancet 2015;386:2422-35. for controlling zoonotic disease in India, 2010-2015. doi:10.1038/nature09575. doi:10.1016/S0140-6736(15)00955-1. PLoS One 2011;6:e17120. doi:10.1371/journal. 58 Brierley L, Vonhof MJ, Olival KJ, Daszak P, 42 Institute of Epidemiology Disease Control and pone.0017120. Jones KE. Quantifying Global Drivers of Zoonotic Research. Strategic framework for One Health 50 Longini IM Jr, , Nizam A, Xu S, et al. Containing Bat Viruses: A Process-Based Perspective. Am Nat approach to infectious diseases in Bangladesh. pandemic influenza at the source. Science 2016;187:E53-64. doi:10.1086/684391. 2012. 2005;309:1083-7. doi:10.1126/science.1115717. 59 Kilpatrick AM, Randolph SE. Drivers, dynamics, 43 Nguyen-Viet H, Doria S, Tung DX, Mallee H, Wilcox 51 Ferguson NM, Cummings DAT, Cauchemez S, et al. and control of emerging vector-borne zoonotic BA, Grace D. Ecohealth research in Southeast Asia: Strategies for containing an emerging influenza diseases. Lancet 2012;380:1946-55. past, present and the way forward. Infect Dis Poverty pandemic in Southeast Asia. Nature 2005;437:209- doi:10.1016/S0140-6736(12)61151-9. 2015;4:5. doi:10.1186/2049-9957-4-5. 14. doi:10.1038/nature04017. 60 Ghosh S, Nagar G. Problem of ticks and tick- 44 Abbas SSKM, Kakkar M. Rabies control in India: 52 Chan EH, Brewer TF, Madoff LC, et al. Global capacity borne diseases in India with special emphasis on a need to close the gap between research and for emerging infectious disease detection. Proc Natl progress in tick control research: a review. J Vector policy. Bull World Health Organ 2015;93:131-2. Acad Sci U S A 2010;107:21701-6. doi:10.1073/ Borne Dis 2014;51:259-70. doi:10.2471/BLT.14.140723. pnas.1006219107. 61 Dhiman RC. Emerging vector-borne zoonoses: 45 Kakkar M, Venkataramanan V, Krishnan S, Chauhan 53 Kluberg SA, Mekaru SR, McIver DJ, et al. Global eco-epidemiology and public health implications RS, Abbas SS. Roadmap to Combat Zoonoses in Capacity for Emerging Infectious Disease Detection, in India. Front Public Health 2014;2:168. India (RCZI) initiative. Moving from rabies research to 1996-2014. Emerg Infect Dis 2016;22:E1-6. doi:10.3389/fpubh.2014.00168. rabies control: lessons from India. PLoS Negl Trop Dis doi:10.3201/eid2210.151956. 62 Dash AP, Bhatia R, Sunyoto T, Mourya DT. 2012;6:e1748. doi:10.1371/journal.pntd.0001748. 54 Hardiman M. The revised International Health Emerging and re-emerging arboviral 46 Thomas D, Zerbini E, Parthasarathy R, Vaidyanathan Regulations: a framework for global health diseases in Southeast Asia. J Vector Borne A. Increasing animal productivity on small mixed security. Int J Antimicrob Agents 2003;21:207-11. Dis 2013;50:77-84. farms in South Asia: a systems perspective. doi:10.1016/S0924-8579(02)00294-7. 63 Liverani M, Waage J, Barnett T, et al. Understanding Agric Syst 2002;71:41-57doi:10.1016/S0308- 55 Ijaz K, Kasowski E, Arthur RR, Angulo FJ, Dowell and managing zoonotic risk in the new 521X(01)00035-X. SF. International Health Regulations--what livestock industries. Environ Health Perspect 47 Eisenstein M. Disease: Poverty and pathogens. gets measured gets done. Emerg Infect Dis 2013;121:873-7. doi:10.1289/ehp.1206001. Nature 2016;531:S61-3. doi:10.1038/531S61a. 2012;18:1054-7. doi:10.3201/eid1807.120487. 48 Governance challenges for zoonotic disease 56 McMichael AJ. Environmental and social surveillance, reporting and response. InKeusch influences on emerging infectious diseases: Cite this as: BMJ 2017;357:j1447 GTPM, Gonzalez MC, Scott KA, Tsai P, eds. Sustaining past, present and future. Philos Trans R Soc Lond http://dx.doi.org/10.1136/bmj.j1447­

BMJ Global Health Initiatives 37 Editorial

Health workers are vital to sustainable development goals and universal health coverage Interventions to support a bigger and better trained work force must be prioritised

Lara Fairall, associate professor generates much debate, the current focus of medications and vaccines could make in Eric Bateman, emeritus professor research and innovation falls short of what underserved communities.5 Furthermore, Knowledge Translation Unit, University of Cape is needed to support health providers in the hope that high tech gadgets and mobile Town Lung Institute, Cape Town, South Africa their daily struggle to meet patients’ needs. apps might substitute for scarce health staff in Proposed innovations tend to be narrow in severely resource constrained settings is not Correspondence to: L Fairall scope, appealing in their gadgetry, but myopic supported by evidence.6 Few of the roughly [email protected] and potentially distracting on the long slow 200 000 health apps currently available have On international woman’s day earlier this journey towards cost effective, accessible had their long term effectiveness or safety month many major news outfits carried health systems for all. evaluated in target populations. Technical the story of Salome Karwah, who in 2014 The search for a blockbuster, game solutions often amount to what Nicholas featured on the cover of Time magazine for changing drug, while enticing, overlooks Muraguri, principal secretary in the Kenyan her work during the Ebola outbreak in west the gains that existing effective, cheap Ministry of Health, refers to as a “junkyard Africa.1 Karwah has died at the age of 28 from complications arising during the birth of her fourth child. Tragic for many reasons, her death highlights persistently high maternal mortality rates, especially in Africa, which contributes 20% of the world’s births but 40% of the world’s maternal deaths.2 It also highlights that stigma can be lethal; her husband reported that hospital workers avoided her because she had worked with and survived Ebola. And it highlights that providers can be patients too. But perhaps most importantly, Karwah’s premature death is a reminder that the health system improvements promised during crises such as the Ebola outbreak are slow to materialise. Responses such as fast tracking the development of an Ebola vaccine tend to eclipse the scaling up of proved, cost effective interventions to meet the health related sustainable development goals. These identify maternal health as a priority, along with child survival and reductions in malaria, HIV, tuberculosis, and non-communicable diseases, under the compelling call to action that healthcare reform must ensure universal coverage for all major causes of morbidity, “leaving no-one behind.” 3 The interagency and expert group on sustainable development goal indicators recognised the pivotal role of health workers in the quest for universal coverage, calling for their “recruitment, development and training in developing countries.” 4 In many of these countries, where doctors are in short supply, this means depending on non-physicians, whether nurses, clinical officers, or healthcare extension workers. We question whether this vital aspect of healthcare delivery is receiving the attention it deserves. Six of the 394 Nigerian Community Health Extension Workers who received PACK training display the PACK Although financing universal health Nigeria guide outside their Primary Healthcare Centre. After training, 93% of trainees in the Nigeria pilot coverage in a fiscally challenged world states report that they now use the PACK Nigeria guide in the majority of their consultations

38 BMJ Global Health Initiatives Editorial of equipment that is dumped when it goes with growing lists of health problems without 3 The sustainable development goals report 2016: leaving 7 no one behind. https://unstats.un.org/sdgs/report/2016/ wrong.” the tools, medications, and equipment they leaving-no-one-behind We must also acknowledge the limitations need? PACK, or the Practical Approach to Care 4 Inter-Agency and Expert Group on SDG Indicators. of task sharing with non-professional or Kit, is an example of one such tool, comprising Sustainable development goals 3: Ensure healthy lives and promote well-being for all at all ages. Targets and indicators. community health workers. In many countries simplified algorithms and checklists covering 2016. https://sustainabledevelopment.un.org/sdg3 these are relatively untrained, low paid, or all of adult primary care with onsite training 5 Wirtz VJ, Hogerzeil HV, Gray AL, et al. Essential medicines voluntary workers—often women—drawn to support frontline providers to provide for universal health coverage. Lancet2017;356:403-76. 9 doi:10.1016/S0140-6736(16)31599- from their own communities. Although evidence informed care. Tested in a series 9 pmid:27832874 community health workers offer an intuitive of trials,10-13 it is currently in use throughout 6 Agarwal S, Perry HB, Long L-A, Labrique AB. Evidence and valuable way of augmenting access to South Africa and being piloted in Brazil, on feasibility and effective use of mHealth strategies by frontline health workers in developing countries: care, they should not divert attention from Nigeria, and Ethiopia. It is time to curb our systematic review. Trop Med Int Health2015;356:1003-14. the need to attract and empower professional obsession with quick fixes and get alongside doi:10.1111/tmi.12525 pmid:25881735 7 Jack A. Caring for a growing world. Financial Times 2016 Nov primary care providers, especially now that frontline health providers, working with 24. https://www.ft.com/reports/maternal-child-health most low and middle income countries face them to codesign the interventions they need 8 Ventevogel P. Integration of mental health into an increasing burden from multimorbidity and to deliver the sustainable development goals primary healthcare in low-income countries: avoiding medicalization. Int Rev Psychiatry2014;356:669-79. lifelong disease. Task shifting to community and letting them lead the way so that none of doi:10.3109/09540261.2014.966067 pmid:25553784 health workers, without adequate support us is left behind. 9 Fairall L, Bateman E, Cornick R, et al. Innovating to improve primary care in less developed countries: towards a global from trained professionals, can easily become We thank Ruth Cornick and Audry Dube for comments on 8 model. BMJ Innov2015;356:196-203. doi:10.1136/ task dumping, reliant on the goodwill of the earlier drafts. bmjinnov-2015-000045 pmid:26692199 world’s poorest women to fill a gap that health Competing interests: We have read and understood BMJ 10 Fairall LR, Zwarenstein M, Bateman ED, et al. systems are failing to close. policy on declaration of interests and declare LF and EDB Effect of educational outreach to nurses on tuberculosis case detection and primary care of A final area of concern is data driven quality are employees of the KTU. The KTU has partnered with BMJ, which provides continuous evidence updates for PACK respiratory illness: pragmatic cluster randomised improvement of health services. Effective in and will be marketing PACK related support services and controlled trial. BMJ2005;356:750-4. doi:10.1136/ bmj.331.7519.750 pmid:16195293 settings with good information systems and licences to PACK content in future. BMJ cofunds the PACK 11 Zwarenstein M, Fairall LR, Lombard C, et al. internet connectivity, data driven quality global development director. PACK receives no funding Outreach education for integration of HIV/AIDS improvement can be burdensome when from industry. EDB sits on the boards and speakers’ bureau care, antiretroviral treatment, and tuberculosis of multiple drug companies with interests in asthma and care in primary care clinics in South Africa: applied in poorly resourced countries because chronic obstructive pulmonary disease. He is funded by drug PALSA PLUS pragmatic cluster randomised it requires struggling health systems to lay companies to give lectures and conduct research on the trial. BMJ2011;356:d2022. doi:10.1136/bmj. cables, install devices, and establish complex same conditions. d2022 pmid:21511783 12 Fairall L, Bachmann MO, Lombard C, et al. Task shifting high tech systems. The data generated may Provenance and peer review: Commissioned; not externally peer reviewed. of antiretroviral treatment from doctors to primary-care serve only to highlight deficiencies in services nurses in South Africa (STRETCH): a pragmatic, parallel, and contribute further to the alienation Republished with permission from BMJ 2017;356:j1357 cluster-randomised trial. Lancet2012;356:889-98. doi:10.1016/S0140-6736(12)60730- of providers from managers, and health 1 Barker A, Von drehle D. Person of the year: the Ebola fighters. 2 pmid:22901955 planners from ministries and donors. Time 2014 Dec 10. http://time.com/time-person-of-the- 13 Fairall LR, Folb N, Timmerman V, et al. Educational year-ebola-fighters/ It is time to refocus attention on the outreach with an integrated clinical tool for 2 Alkema L, Chou D, Hogan D, et al. United Nations nurse-led non-communicable chronic disease recruitment, empowerment, and retention of Maternal Mortality Estimation Inter-Agency Group management in primary care in South Africa: a healthcare providers, who in many countries collaborators and technical advisory group. Global, pragmatic cluster randomised controlled trial. PLoS regional, and national levels and trends in maternal Med2016;356:e1002178. doi:10.1371/journal. are non-physician clinicians like Karwah. mortality between 1990 and 2015, with scenario- pmed.1002178 pmid:27875542 Where are the interventions to recognise and based projections to 2030: a systematic analysis by the UN Maternal Mortality Estimation Inter-Agency support these providers who arrive at work Group. Lancet2016;356:462-74. doi:10.1016/S0140- Cite this as: BMJ 2017;362:j1357 day after day to face growing queues of people 6736(15)00838-7 pmid:26584737 http://dx.doi.org/10.1136/bmj.j1357

BMJ Global Health Initiatives 39 Analysis

Implementing One Health as an integrated approach to health in Rwanda

t is increasingly clear that resolu- challenges. This approach drives inno- the majority of human pathogens are tion of complex global health prob- vations that are important to solve both zoonotic (60%) and three-quarters of new lems requires interdisciplinary, acute and chronic health problems and and emerging pathogens are zoonotic from intersectoral expertise and coop- offers synergy across systems, resulting in wildlife species.45 However, One Health, eration from governmental, non- improved communication, evidence-based which is larger than simply zoonosis (other Igovernmental and educational agencies. solutions, development of a new genera- examples include land use, water toxins, ‘One Health’ refers to the collaboration of tion of systems-thinkers, improved sur- forest degradation and climate change multiple disciplines and sectors working veillance, decreased lag time in response, (see online supplementary appendix A)), locally, nationally and globally to attain and improved health and economic sav- can have a great impact on people and the optimal health for people, animals and ings. Several factors have enabled the One quality of their lives as well as local and the environment. One Health offers the Health movement in Rwanda including an national economies. opportunity to acknowledge shared inter- elaborate network of community health Although perhaps the most discussed, ests, set common goals, and drive toward workers, existing rapid response teams, infectious diseases are not the only team work to benefit the overall health of a international academic partnerships will- relevant One Health concerns affecting nation. As in most countries, the health of ing to look more broadly than at a single the globe. Waste dumped in or near Rwanda’s people and economy are highly disease or population, and relative equity water flows through streams, rivers and dependent on the health of the environ- between female and male health profes- lakes, affecting entire communities— ment. Recently, Rwanda has developed sionals. Barriers to implementing this strat- crops, animals and people. Meanwhile, a One Health strategic plan to meet its egy include competition over budget, poor as the human population on the planet human, animal and environmental health communication, and the need for improved increases, humans are forced to live in technology. Given the interconnectedness closer proximity to both wild and domestic of our global community, it may be time for animals, which increases exposure to Key messages countries and their neighbours to follow new pathogens, and forces the sharing What is already known about this Rwanda’s lead and consider incorporating of limited supplies of water (see online topic? One Health principles into their national supplementary appendix B). Add to strategic health plans. • One Health is the term that refers to this the growing pressures to increase the collaboration of multiple disci- agricultural production, global warming plines, sectors and multiple groups Why One Health? with the resultant decrease in water working locally, nationally and glob- One Health refers to the collaboration of supplies, changing microbial patterns, ally to attain optimal health for peo- multiple disciplines, sectors and groups and deforestation, and the result is a ple, animals and the environment. working locally, nationally and globally to deterioration of natural resources and a What are the new findings? attain optimal health for people, animals reduction in many of the protections and 1 checks and balances that have previously • There is little information on how to and the environment. Recent examples institutionalise and operationalise of new and emerging diseases in animals been afforded to human populations. One Health. Rwanda has set out to and humans (Ebola, Middle East respira- Despite the complex nature of these achieve, in policy and practice, what tory syndrome, avian flu (H5N1), swine challenges, most governments have offices has yet to be implemented across any flu (H1NI), severe acute respiratory syn- or units that focus vertically on specific nation: an evidence-based, intercon- drome) show how quickly balance changes diseases, and these offices often compete nected system to address ‘One Health’ and how vulnerable humans, animals and with each other for limited resources. Non- 23 problems. In this paper, we pre - crops are to disease outbreaks. Infectious government organisations (NGOs) that sent Rwanda’s ‘One Health’-oriented diseases are transmitted between humans align themselves with one issue or disease response to global grand challenges and animals by a variety of routes includ- are often disconnected from interventions as a call to action. ing direct contact (rabies), the environ- for others. It is this sort of Brownian approach to improving health that Recommendations for policy ment (anthrax), via food (campylobacter/ salmonella/brucella/bovine tuberculo- results in poor communication between If successful, Rwanda’s One Health • sis), or through bites by arthropod vectors disciplines, duplication of services with approach will result in speedier (malaria/leishmaniasis/Rift Valley fever). resulting higher costs, ignoring of common achievement of meaningful health As we have recently seen with the Ebola antecedent causes of poor health, and outcomes with more innovative solu- and Zika outbreaks, in our interconnected inefficient utilisation of available resources tions to pressing health problems, world, an animal pathogen can catch a ride such as specialised reference laboratory and will serve as a model for other on the sole of a shoe, beneath a finger nail, facilities. Academic communities, local countries that may benefit from incor- or in respiratory passages, and travel from and national governments, and scientists porating One Health principles into one remote corner of the globe to another worldwide are now recognising that the their national strategic environmen- in less than a day. Furthermore, zoonotic ill- next logical step in problem solving is to tal, livestock and health plans. ness is not a small or insignificant­ problem; ­ connect interdisciplinary and government

40 BMJ Global Health Initiatives Analysis

agency experts so that they can focus on health indicators (eg, improved maternal • Address issues that relate to land use the root causes of illness and the need health, reduction in HIV, reducing planning, reducing contact between for prevention and detection rather than malaria and other vector borne illnesses), humans, domestic and wildlife with responding separately and acutely to each and environmental sustainability all minimal changes to critical habitat; and disease. The activities and conditions of depend on interdependent systems, • Address nutritional access by developing each individual, each region, and each shared responsibility, involvement of the safer practices related to bush meat and country affect others on a variety of levels community, and collaboration across animal consumption. including economic, cultural, physical, government agencies, content specialists This multipronged strategic plan is social and more. While very few sub- and policies—all ideas embodied by One problem focused rather than discipline Saharan African countries are taking steps Health, a burgeoning global approach focused, and seeks to bring together the to put these ideas into action, Rwanda is to integrated health. The government of newly realigned University of Rwanda, unique in that it is well on its way. Rwanda has therefore framed policies and the Ministries of Health, Agriculture and priorities to drive toward an integrated, Animal Resources and Education, The holistic-system approach to promoting One Health in Rwanda Wildlife Unit of the Rwanda Development health. Moreover, it has led to the adoption Known as ’the land of 1000 hills’, Rwanda Board, and other ministries and civil of the One Health approach by the East has a north–south mountain range, various society. The strategic plan reflects Rwanda’s African Community, and Rwanda is water sheds, rain forests and grazing lands. belief that complex health problems can also working with its neighbours to The nation confronts various challenges: be addressed through integrated policy address regional issues that recognise energy sustainability, natural gas extrac- and interventions that simultaneously the inextricable connection between the tion from beneath Lake Kivu, a growing and holistically address multiple causes of health of the country’s people, animals population, land degradation, crop raiding, poor health (eg, poverty, limited education, and environment and the importance wildlife poaching6; a loss of biodiversity, unsafe and scarce water, lack of sanitation, of this interconnection in development. conversion of forests to farm land and the food insecurity, gender inequality, and The concept and approach of One Health 14 risk of soil overexploitation; and climate close proximity of humans and animals). provides an opportunity for the Rwandan change resulting in an increasingly vari- To further reinforce One Health principles government to expand its reforms to able rainfall. In addition, Rwanda is one of and uphold accountability to the strategic address important interdisciplinary, the most densely populated (415 people/ plan, Rwanda has set out to meet three intersectoral health problems and work to square mile) countries in the world,7 where core One Health objectives over the next meet the Sustainable Development Goals. One Health disasters can quickly affect few years. Rwanda has therefore set out to achieve, large populations. Further, areas with high in policy and practice, what has yet to Objective 1: Rwanda’s government response population density are more prone to food be implemented across any nation—an Rwanda’s One Health response (table 1) insecurity, soil erosion, decreased graz- evidence-based, interconnected system to goes beyond the traditional approach of ing lands, and forest degradation, which 13 address One Health problems. disease surveillance, outbreak investiga- in turn leads to increased food insecurity tions and response. It also includes new and other measures of poor health.89 The The Rwanda One Health strategic plan competencies around leadership/govern- eastern part of Rwanda is home to pasto- In 2015, the Government of Rwanda devel- ance, efficiencies in resource utilisation, ral communities, which move from place oped and approved a One Health strategic disaster management, delivery of health- to place in search of water and pastures to plan to streamline cross-sectoral and insti- care, systems-related approaches, and feed their animals. Movement is not limited tutional interventions, minimise duplica- vigorous attention to training for life-long to the national borders, thus pastoralists tion of efforts, and maximise the use of learning. In the past, the Ministry of Health, are at risk of picking up animal pathogens public resources. The goals are to:Promote the Ministry of Agriculture and Animal that can be disastrous to the livestock popu- integrated disease surveillance, prevention Resources, other government organisa- lation in Rwanda such as foot and mouth and response (animals, humans and agri- tions, academic institutions and NGOs had disease and contagious bovine pleural culture); separate roles with little overlap. Despite pneumonia, both of which have become limited resources, Rwanda’s One Health endemic.1011 These diseases have high Improve education and communication • approach is intended to develop collabora- mortality and thus affect food security and among animal, human and environmen- tive leaders committed to improving health the economic well-being of these nomads. tal professionals; equity and social justice by addressing Contagion between animals (wild and Expose and integrate students engaged • health disparities that impact on efficiency domestic) and humans does not happen in professional education at university by promoting shared resources and col- in only one direction. In 2011, one of the level to concepts related to One Health; laboration among those working at the Promote interprofessional collabora- mountain gorillas, which provide large eco- • animal (wildlife, livestock and companion tion around innovation, research and tourism revenue for Rwanda, succumbed to animals), ecosystem and human health a human virus (human meta-pneumovirus) ­discovery; interface. passed on by a tourist.12 • Develop educational tools for pre-univer- Through these experiences, Rwanda sity education that introduces concepts Objective 2: Rwanda’s One Health community has learned that the eradication of hunger of One Health; response through initiatives such as Girinka (one • Develop policy focused on upstream Another important goal of Rwanda’s One cow per family (see online supplementary drivers of disease emergence including Health Strategy is to empower and mobi- appendix B)), improvements in public land use, water access and deforestation; lise various experts and lay workers and

BMJ Global Health Initiatives 41 Analysis

­establish a One Health workforce to pre- great opportunity to articulate One Health narrow sub-population. In the new model, pare, coordinate and manage epidemio- as a cross-disciplinary approach. driven by the One Health strategic plan, logical outbreaks of infectious, toxic or As a start, Rwanda has created a One interventions are highly coordinated. environmental health concern or health Health curriculum embedded in its Master Finally, within Rwanda there is relative events. For example, the Rwandan strategic of Global Health Delivery programme equity between female and male health plan requires the inclusion of veterinarians, which integrates collaborative problem- professionals, making it easier to address wildlife experts and environmental experts solving approaches with elements of important gender and cultural issues who work on emergency management com- infectious disease, epidemiology, ecology, relevant to improving One Health. mittees. Similarly, disease surveillance of environment, finances, food safety and Resolution of One Health problems both zoonotic and potential zoonotic dis- leadership. Plans are also underway to often pits one discipline or sector against ease is monitored by a multidisciplinary integrate One Health modules into the another with resultant perceptions of team. This is a bottom-to-top approach Master of Public Health and Epidemiology ’winners’ and ’losers’, at least in terms of that involves community health workers courses, developing a 1-month community- resources. Other barriers that need to be (CHWs), community-based animal health focused field boot camp in 2017 to further overcome include a lack of experts trained workers, NGOs, health clinics, hospitals, train animal–human–ecosystem providers in a One Health approach, competition park rangers, farmers and domestic animal in integrated problem solving, leadership for government resources, battles over owners. These experts are prepared and and communication skills related to One curricular time in training programmes, trained to act rapidly and collaboratively Health. Finally, a vibrant One Health issues related to licensing and certification, given evolving information. students’ club for undergraduates was and interdisciplinary turf wars. There has One idea moving forward is to create a established in 2012, the first of its kind in long been a need to develop infrastructure hub-and-spoke network using the nearly the region, that links virtually with other such interdisciplinary laboratories and 45 000 CHWs (spokes) linked to hubs health sciences schools around the world structures that promote interdisciplinary, (centres of expertise) through mobile who share a commitment to learning interministerial collaboration focused on phone technology. Perhaps one day the related to One Health. It consists of students problem solving (eg, childhood diarrhoea CHWs will be rebranded ’One Health from veterinary medicine, environmental linked to bovine mastitis). In Rwanda, the CHWs’ (OHCHWs) given that they are well health, nursing, medicine and agriculture. key ministries related to One Health have situated to quickly identify unusual events The goal of the club is to bring One Health already coalesced to form a ministerial or problems affecting humans, animals or skills, approaches and attitudes to a new ’Social Cluster’ which meets monthly, ecology/agriculture. Hub centres would be generation of scientists and problem with the goal of ensuring that there is connected via the internet to district centres solvers who will embrace the importance little competition for resources between and eventually to a central repository and of working together to serve the community ministries and that shared issues are command centre. OHCHWs would routinely rather than working in silos. addressed collectively. However, additional collect local information on the health of efforts to create a robust infrastructure humans, animals and crops and notify hub Enablers of and barriers to implementing One that would support collaboration and centres when there are sudden changes or Health in Rwanda interdisciplinary training would further concerns. While Rwanda has been forward think- enable Rwanda’s One Health response. ing in developing a One Health-focused The One Health approach is in evolution and will still require a cultural shift in Objective 3: Rwanda’s One Health educational national strategic plan, several impor- (academic) response tant factors have enabled this innovative Rwanda as power and organisational Interprofessional team work and collabora- change. Following the 1994 genocide, structures become realigned to provide tion such as that mentioned above has the Rwanda has benefited greatly from two dec- new reporting structures, new offices, new best chance of becoming routine if educa- ades of social and political stability from education and new lines of communication. tion and training starts early and focuses transparent governance with local, regional Moving forward, Rwanda’s government on core competencies that stress problem and national representation. The elaborate needs to fund the implementation and solving, team work, leadership, creativ- network of CHWs has been a key element embrace the concept of ’oneness’ such ity, conflict management, communica- in primary healthcare delivery. Rwanda’s that the separate ministries can develop tion, project management, transparency rapid response teams previously developed common policies, approaches and and outcomes. However, despite extensive in response to outbreaks of Ebola and yel- evaluations that can feed into action plans capacity-building efforts in Rwanda, there low fever in neighbouring Uganda and the and improved health infrastructure such remains an undersupply of physicians, vet- Democratic Republic of Congo are now as providing better equipped laboratories erinarians and environmental scientists. being used to address other One Health and data tracking. Academics need to think To combat this problem, several colleges problems by coordinating surveillance, beyond the traditional silos (medicine, and universities have recently coalesced information sharing, and planning of risk public health, veterinary medicine, into the University of Rwanda to improve reduction and communication. engineering, etc) in ways that will stimulate opportunities for interprofessional train- The country also benefits from many innovation and encourage problem solving. ing, interdisciplinary scholarship and international academic partnerships research innovation, and work force expan- involving medicine, public health, Conclusion sion. Harmonisation of the environmental veterinary medicine, agriculture and the As in most countries, the health of Rwan- health programmes offered by the old vet- environment. However, until recently, da’s people and its economy are highly erinary college and the old ’human health most of these partnerships were solitary dependent on the health of the environ- sciences’ college has now provided another and often focused on one disease or a ment. One Health offers the opportunity

42 BMJ Global Health Initiatives Analysis

Table 1 | One Health strategic objectives Objective Strategies 1 (government level): promote and strengthen interdisciplinary collaboration to Improve communication and interactions between ministries responsible for animal, promote a One Health approach human and environmental issues and regional agencies Engage in One Health strategic planning focused on systems-thinking that considers the diverse range of complex and inter-related One Health issues impacting on animals, human health and the environment at the local, national and international level Collaborate with the East African Community to expand the One Health concept across the region, given that toxins, infections and environmental degradation do not respect political borders Provide financial incentive and support for One Health initiatives to incentivise col- laborative problem solving 2 (community/NGO level): strengthen surveillance, prevention, early detection, rapid Improve the capacity within Rwanda to conduct community surveillance, treatment response, and control of zoonosis in both animals and humans and monitoring of outcomes of One Health problems including emerging and re- emerging zoonotic diseases, neglected diseases, and other public health events of international concern that pose a threat to human health Introduce technologies including computers, mobile phone data collection applica- tions, and tele-conferencing to improve detection, monitoring and intervention related to One Health problems at the community level Promote timely and goal-directed communication between local communities, minis- tries, NGOs and neighbouring nations 3 (academic level): build capacity and promote applied research at the human, Improve training capacity of both professionals and mid-level providers to develop animal and ecosystem interface skills necessary to identify, monitor and respond to One Health problems that may cross outside of their area of expertise Modify health science and environmental training programmes/curriculum to promote graduation competencies related to collaboration and cross-disciplinary problem solving Develop training programmes for existing professionals to promote the sharing of knowledge, skills and resources to address current and future One Health needs Train, keep current and incentivise One Health problem solvers to stay in Rwanda. Despite the huge investment of national resources, it is not unusual for trained health experts to leave the country for economic gain or even to be pulled away for other national service Protect national resources include the gorilla population and other wild animals that could be damaged by exposure to life-threatening human infectious diseases

to recognise shared interests, set common SN and HM drafted and revised the paper based 1Ministry of Health, Kigali, Rwanda goals and drive toward team work to benefit on comments from all members of the One Health 2University of Minnesota, Minneapolis, the health of a nation. Rwanda’s One Health strategic planning task force. All members of the Minnesota, USA approach provides innovations that are strategic planning committee developed the plan 3University of California Davis, Davis, California, important to both acute (disaster or emerg- for the paper. PC provided review and comments on USA drafts of the paper. PF and AB reviewed the final draft ing zoonotic disease) and chronic (animal, 4University of Rwanda, Butare, Rwanda human and ecosystem) health problems of the paper and provided comments. 5Harvard Medical School, Boston, Competing interests: None declared. and offers synergy across systems, resulting Massachusetts, USA in improved communication, development Provenance and peer review: Not commissioned; 6Partners in Health, Rwanda of a new generation of systems-thinkers, externally peer reviewed. 7 improved surveillance, decreased lag time Data sharing statement: No additional data are USAID Preparedness and Response Project 8 of response, and improved health and eco- available. Rwanda Agricultural Board, Kigali, Rwanda nomic savings. Given the interconnected- Transparency declaration: TN, the manuscript’s 9Rwanda Development Board, Kigali, Rwanda ness of our global community in which guarantor, affirms that the manuscript is an honest, 10Centers for Disease Control and Prevention humans, animals and the environment accurate and transparent account of the study being (CDC), Kigali, Rwanda impact on each other and do not respect reported, that no important aspects of the study have 11Rwanda Biomedical Center, Kigali, Rwanda geopolitical boundaries, it may be time for been omitted, and that any discrepancies from the 12 University of Global Health Equity all countries and their neighbours to follow study as planned (and, if relevant, registered) have Rwanda’s lead and consider incorporating been explained. Dr Michael Wilkes; mswilkes@​ ucdavis.​ edu​ One Health principles into their national Thierry Nyatanyi,1,2 Michael Wilkes,3,4,5 Additional material is published online strategic health plans. Haley McDermott,3,6 Serge Nzietchueng,2,7 only. To view please visit the journal Handling editor: Seye Abimbola. Isidore Gafarasi,8 Antoine Mudakikwa,9 online (http://dx.doi.org/10.1136/ 9 10 bmjgh-2016-000121). Twitter: Follow Jean Felix Kinani @ohac_gorilla, Jean Felix Kinani, Joseph Rukelibuga, jfkinani, gorillahealth Jared Omolo,10 Denise Mupfasoni,10 Republished with permission from BMJ Global Contributors: The following are members of the Adeline Kabeja,11 Jose Nyamusore,11 Health 2017;2:e000121. One Health Strategic Planning Task Force, Ministry Julius Nziza,4 Jean Leonard Hakizimana,11 of Health, Rwanda: TN, SN, IG, AM, JFK, JR, JO, DM, Julius Kamugisha,11 Richard Nkunda,11 AK, JNy, JNz, JLH, JK, RN, RK, ER, PF and AB provided Robert Kibuuka,4 Etienne Rugigana,4 5,6,12 4 guidance on this project. TN and MW were the primary Paul Farmer, Philip Cotton, This is an Open Access article distributed in authors. TN is the guarantor of the article. TN, MW, Agnes Binagwaho5,12 accordance with the Creative Commons Attribution

BMJ Global Health Initiatives 43 Analysis

Non Commercial (CC BY-NC 4.0) license, which permits Lond B Biol Sci 2001;356:983–9. doi:10.1098/ 12 Palacios G, Lowenstine LJ, Cranfield MR, et al. others to distribute, remix, adapt, build upon this rstb.2001.0888 Human metapneumovirus infection in wild work non-commercially, and license their derivative 6 Wronski T, Bariyanga J, Apio A, et al. Interactions mountain gorillas, Rwanda. Emerging works on different terms, provided the original work between wildlife, humans and cattle: activity Infect Dis 2011;17:711. doi:10.3201/ is properly cited and the use is non-commercial. See: patterns of a remnant population of impala on the eid1704.100883 http://creativecommons.​ org/​ licenses/​ by-​ nc/​ 4.​ 0/​ degraded Mutara Rangelands, Rwanda. Rangeland J 13 Lee K, Brumme ZL. Operationalizing the 2015;37:357–65. doi:10.1071/RJ15025 One Health approach: the global governance 1 Association AVM. One Health: a new professional 7 Commission RNC. Fourth population and housing challenges. Health Policy Plan 2013;28:778–85. imperative, One Health Initiative Task Force: final census. Kigali, Rwanda, 2012. doi:10.1093/heapol/czs127 report. Schaumburg, IL, USA: American Veterinary 8 Blarel B, Hazell P, Place F, et al. The economics of 14 Collaboration F-O-W. Tripartite Concept Note: Medical Association, 2008. http://www avma org/ farm fragmentation: evidence from Ghana and sharing responsibilities and coordinating onehealth/default.​ asp​ Rwanda. World Bank Econ Rev 1992;6:233–54. global activities to address health risks at the 2 Gibbs E. Emerging zoonotic epidemics in the doi:10.1093/wber/6.2.233 animal-human-ecosystems interfaces. World interconnected global community. Vet Rec 9 Drechsel P, Gyiele L, Kunze D, et al. Population Health Organization, 2010. http://www oie int/ 2005;157:673. doi:10.1136/vr.157.22.673 density, soil nutrient depletion, and economic growth fileadmin/Home/eng/Current_Scientific_ Issues/ 3 Mwacalimba KK, Green J. ‘One health’ and in sub-Saharan Africa. Ecol Econ 2001;38:251–8. docs/pdf/FINAL_CONCEPT_NOTE_Hanoi pdf development priorities in resource-constrained doi:10.1016/S0921-8009(01)00167-7 (accessed 2013). countries: policy lessons from avian and pandemic 10 Ekou J. Dairy production and marketing in Uganda: influenza preparedness in Zambia. Health Policy Plan current status, constraints and way forward. Afr J 2015;30:215–22. Agric Res 2014;9:881–8. 4 Jones KE, Patel NG, Levy MA, et al. Global 11 Tekleghiorghis T, Moormann RJ, Weerdmeester K, Cite this as: Nyatanyi T, Wilkes M, McDermott H, trends in emerging infectious diseases. Nature et al. Foot-and-mouth disease transmission et al. Implementing One Health as an 2008;451:990–3. doi:10.1038/nature06536 in Africa: implications for control, a review. integrated approach to health in Rwanda. 5 Taylor LH, Latham SM, Mark E. Risk factors for Transbound Emerg Dis 2016;63:136–51. BMJ Global Health 2017;2:e000121. human disease emergence. Philos Trans R Soc doi:10.1111/tbed.12248 doi:10.1136/bmjgh-2016-000121

44 BMJ Global Health Initiatives Analysis

Building resilient health systems: a proposal for a resilience index Health system resilience begins with measurement of critical capacities ahead of crisis say Margaret E Kruk and colleagues

he 2014 west African Ebola epi- resilient health systems—systems that can Exploring the concept of resilient health demic shone a harsh light on withstand health shocks while maintaining systems the health systems of Guinea, routine functions.10 The issue of how While the construct of resilience has been Liberia, and Sierra Leone. While global bodies can support countries in widely used in diverse fields, including decades of domestic and inter- withstanding future health shocks is ecology, engineering, and psychology, it is Tnational investment had contributed to playing out now in the election of the new relatively new to health.13-17 With a plethora substantial progress on the Millennium director general of WHO, with several of frameworks and catchphrases crowding Development Goals,1 2 national health candidates making health system resilience the global health lexicon, there are legiti- systems remained weak and were unable part of their election planks. mate questions about the value added by to cope with the epidemic. Routine care of Based on recent literature, this paper the concept. We identify three contribu- the population also deteriorated during defines health system resilience as “the tions of the concept of resilience to the the outbreak.1-4 Surveillance systems did capacity of health actors, institutions, health systems field. not function effectively, allowing Ebola to and populations to prepare for and First, resilience emphasises the functions spread within and between the countries. effectively respond to crises; maintain health systems need (figure 1) to respond Global institutions were slow to respond to core functions when a crisis hits; and, and adapt to health shocks, introducing a the crisis, squandering an opportunity to informed by lessons learnt during the dynamic dimension into more static health stem its course.5-7 crisis, reorganise if conditions require system models which can help the system Since then, diverse panels of experts it.”12 Health system resilience is relevant cope with surges in demand and adapt to have pointed to political and technical in all countries facing health shocks— changing epidemiology and population deficiencies in multilateral organisations whether sudden (Ebola, earthquakes, expectations of care.18 A rigidity of mission in tackling health crises.8-11 These reports terror attacks, refugees), slower moving characterises the operations of many have noted that the first line of defence (new pathogens such as Zika becoming countries’ health systems, whose ethos and against future pandemics is an effective endemic or epidemiologic transition), or organisation is better suited to yesterday’s national health system. They have also the more chronic stresses that characterise disease burden than tomorrow’s, focusing called for better measurement of public even times that seem calm (drug shortages, mostly on basic, episodic care, unequipped health capacity, and investment to build loss of key health personnel, smaller to provide advanced care for infections, outbreaks of endemic diseases). Yet, while longitudinal care for a broad spectrum of health system resilience has been defined chronic diseases, or emergency care needed Key messages and widely discussed, there is debate about to respond to the rising tide of injuries. • National health systems are the first whether the concept has anything new Second, the concept contributes useful line of defence against health crises to add to discussions on health system new ideas to health systems from other • Health systems today are rigid and strengthening, and how resilience can best sectors. Solutions for supply chains and slow to adapt; they must become be built and measured. logistics to respond to surges in demand 19 20 more resilient to effectively respond As shown in figure 1, resilient health from other fields may be relevant. to crises and maintain core services systems are aware, integrated, diverse, Building trust and promoting meaningful self regulating, and adaptive. These community engagement have been studied • Resilience requires planning and investment in slow variables (for features do not arise in a vacuum: they in other fields, such as environmental instance, health workers, managers, require a foundation of strong local and sustainability and political science, but national leadership, a committed health have not been well operationalised in information systems) and fast vari- 21-23 ables (such as isolation wards, pro- workforce, sufficient infrastructure, health systems science. Resilience tective equipment, surveillance). It and global support. The last point is draws on complex systems notions requires methodical building of col- especially worth emphasising: resilience identified as important in health systems laboration and trust with communi- is not self sufficiency. Crises do not respect but rarely acted upon, such as the ties ahead of crisis. geopolitical boundaries and so resilience interconnectedness of health and non- requires thoughtful interconnectedness or health actors and the importance of The resilience index proposed here • “smart dependency.” feedback loops.24 is designed to help countries assess In this paper, we argue that the concept Finally, the concept of resilience helps whether their health systems can of resilience adds substantial value to bridge disparate health and development withstand future shocks; it should the health systems discourse, and we agendas—such as universal health be tested in countries at high risk of propose measures of health system coverage, the Global Health Security health shocks. resilience. Agenda, and the Sustainable Development

BMJ Global Health Initiatives 45 Analysis

engagement is needed to ensure responsive health services that people trust and want to use.12 Imposed technocratic solutions will not bring about needed change, and the particular arrangements needed to promote resilience have to emerge from the country’s context. Value judgments about what constitutes resilience for whom should be made explicit. Ordinary people may lack the power to shape the health system response or hold it to account; the process of building resilience should foster that power. Building resilience should be integrated with existing efforts to strengthen health systems and its success should be judged on equitable health gains rather than the security of wealthy nations.

Resilience in action Fig 1 | Resilient health system framework12 We present three case studies, in which sev- eral of the authors were involved, where a Goals—lending fresh impetus to the need to shocks. These factors may include range of large health shocks contributed to invest in health systems.25-28 It identifies unfavourable trade terms, weak citizen to improved health system resilience: the immediate and longer term payoffs engagement, and chronic health system chronic system dysfunction aggravated by of well functioning, responsive, and deficiencies.30-33 a population influx in Lebanon; sudden adaptable health systems and highlights There are also worries about short term and severe infectious disease outbreak in the unacceptable costs of inaction. timeframes for action when problems are Liberia; and repeated, anticipated disaster By containing outbreaks, returning to multifactorial, and a paradoxical push for shocks in Indonesia. baseline function faster, and mitigating national self reliance when threats readily Awareness is the capacity to detect and other shocks, resilient health systems cross borders.31 interpret local warning signs and quickly can contribute to economic stability.29 While these concerns highlight the call for support. Liberia’s initial paralysis The recognition that health systems are potential for resilience to be used as during the Ebola epidemic was partly the front line for dealing with the next big shorthand for a narrow preparedness caused by poor understanding, at all levels, threat to global health security amplifies agenda, they do not accurately represent of the disease severity. Self regulation is the urgency of strengthening them and the meaning of health system resilience the ability to isolate threats and maintain draws in new actors and ideas.10 as intended here. Building resilience is core functions under stress. While Ebola The increasing attention to resilience much more than preparedness; it involves treatment units are a classic example of self in global health has, however, prompted investment in institutions, preconditions regulation (in Liberia’s case, these came criticisms of the concept. One is that it is (like an effective health workforce) and too late to mitigate spread), Lebanon’s an imposed, technocratic solution that other “slow variables.” Communities emergency vaccination and surveillance obscures the socioeconomic and political should not have to shoulder crises efforts, and Indonesia’s regional crisis factors that lead to inadequate responses alone; instead meaningful government mitigation centres can also be seen as

Box 1: Integrated approaches to care for diverse needs: working with non-state actors during the Syrian refugee influx in Lebanon Since the beginning of the Syrian civil war in 2011, Lebanon has had an unprecedented influx of refugees, increasing its population by 1.5 million, or 30%.34 The Syrian crisis persists today, placing continuing strain on Lebanon’s health system. Lebanon’s health system has demonstrated resilience by rapidly mobilising and expanding its diverse primary care capacity in the public and private sectors. Initial refugee health relief focused on short term assistance delivered by multiple organisations.37 38 The fragmentation of early relief efforts motivated the ministry of public health to establish a steering committee to streamline relief funding and encourage transparency and accountability across international and national health actors.39 Primary healthcare grew to be the central platform for the response. In 2015, the government and its multi-sectoral partners (including UNHCR, UNDP, World Bank, and NGOs) established 20 new public health centres and directly supported 100 private health centres, increasing primary care capacity by 40%.40 Covered services include non-communicable disease screening, nutrition services, and mental health support.40 41 Additions to the epidemiological surveillance system improved the ability to detect emerging diseases, contributing to the country’s quick response to polio threats.42 Despite early successes in primary care, access to Lebanon’s secondary and tertiary healthcare systems continues to be a challenge for refugees.40 41 Recent estimates suggest that approximately 26% of the refugee population needs secondary healthcare, however 23% of those are unable to access it, primarily because of high fees (71%).43 Financial assistance is limited to specific conditions and requires co-payment, which contributes to substantial financial burdens for refugees.40-44

46 BMJ Global Health Initiatives Analysis

Box 2: Learning from failure: communicating with communities during Liberia’s Ebola crisis At the peak of the 2014 Ebola epidemic, Liberia reported 300 to 400 new cases each week and had the highest incidence of Ebola deaths of the affected west African nations.45 Meanwhile, non-Ebola patients were neglected—health facilities lacked testing and isolation capacity and thus turned down patients who appeared sick.46 Some facilities simply stopped providing services altogether. Some urban and rural communities resisted surveillance and disease control efforts, believing Ebola was purposely introduced by the government and foreign institutions to gain profits from emergency response activities.46-48 Trust was further eroded by inadequate response from Ebola task forces and help hotlines when neighbours fell ill.47 Gradually, Ebola treatment units opened and health facilities resumed services. At the same time, the ministry of health and partner NGOs launched a series of public health messages beginning with “Ebola kills,” intended to emphasise the gravity of the epidemic.49 This approach backfired. Communities reasoned that if Ebola was fatal then affected people should avoid treatment units and instead wait to die at home, supported by family.49 Public messages gradually evolved to “the earlier you report Ebola, the more likely you are to survive.” Traditional leaders were enlisted to support community training in all 88 counties and spread messages in local dialects.50 To improve the effectiveness of the epidemic response, communities were directly engaged in surveillance. In West Point, Monrovia’s largest slum, community and traditional leaders were assembled to discuss concerns and propose a locally driven solution for Ebola surveillance in the densely populated area.50 51 A system for active case finding developed. Leaders recruited community volunteers to complete ministry led surveillance training, which eventually led to the deployment of 152 active case finders and 15 psychosocial support workers.51 Active case finders and psychosocial support workers helped identify potential Ebola cases, reduce caregiver transmission, and promote burials by trained “safe and dignified” burial teams.

homoeostatic innovations for containing ministry of health has expanded primary partnerships allow government officials health threats. care to tackle the multiple health needs of to weave the experience, expectations and Indonesia’s case also shows the value of both refugees and citizens. This has been capabilities of affected people into the learning and adaptation: in anticipation of done in part through consultation and containment strategy for a more powerful future catastrophic weather events crisis contracting with private sector providers, and empathetic response. Identifying ways mitigation centres were instituted after the including faith based providers; an to work effectively with local leaders was country experienced poor coordination example of integration among diverse a critical lesson from Liberia during the after tsunamis. In each of these case health actors who in the past may not have recent Ebola epidemic. Community leaders studies, most elements of resilience worked together. were critical in case finding, community emerged after a crisis rather than ahead of Integration also draws attention to the mobilisation, and other epidemic control it. As we note below, future research should key mediating role that broader state- measures. consider how the elements of resilience society relations play during crises, perform when adopted before the event. including the recognition of people as Measuring resilience capacity: the resilience index The value of having diverse healthcare producers of their health and thus as Recent international panels reviewing the providers that can coordinate with each co-architects of an effective crisis response. Ebola response have called for measure- other is seen in the case of Lebanon, which Involving people and communities in ment of health system resilience capacity is now hosting 1.8 million refugees from crafting a response depends on—and ahead of crises.10 35 36 Building on the con- Syria, increasing its population by over is a potential means of—strengthening ceptual framework described in figure 1, 30%.34 To meet the challenge of much government accountability to its citizens. we have outlined a set of preliminary larger numbers of people seeking care, the Stronger mechanisms for state-society measures of national health system resil-

Box 3: Improving self regulation: coordinating multiple actors during natural disasters in Indonesia Spread across three major geologic fault lines, Indonesia experiences periodic earthquakes and tsunamis. Each recent disaster has tested the country’s health system and led to progressive adaptation. The 2004 Indian Ocean tsunami devastated the province of Aceh.52 Overnight, 106 health facilities in Aceh were damaged or destroyed, and more than half of the health workforce was displaced or killed.53 54 The government struggled to organise a response and assistance was further delayed by security concerns: Aceh had been the site of recent battles between the government and the Free Aceh Movement, a guerrilla separatist group.55 56 When aid arrived, provision was chaotic with duplication of efforts in some areas and gaps in services in others. It took two weeks to establish a disaster coordination centre, and nearly a month for the Aceh health system to resume function. Two years later during the 2006 Yogyakarta earthquake, the national response was remarkably different. Hours after the earthquake the president of Indonesia temporarily relocated his office to Yogyakarta to support the National Disaster Management Agency emergency efforts.57 While 67 of 115 health centres in Yogyakarta were damaged or severely destroyed, domestic health teams were quickly mobilised to provide emergency relief.58 59 The response to this earthquake—both more efficient and more locally driven—was informed by lessons learnt from Aceh and the absence of conflict in the area. Learning from these experiences, Indonesia established nine regional crisis mitigation centres in 2009.60 Strategically located in disaster prone areas, these centres are proactively equipped with staff, vehicles, and emergency supplies, and perform community outreach with local health facilities in between natural disasters, teaching basic first aid and natural disaster response.61 62

BMJ Global Health Initiatives 47 Analysis

Table 1 | Resilience index Characteristics* Aims Measures Rationale Aware Know health system capacity 1 Distribution of health system assets Real time geo-registry of HWs, supplies, and facilities (including NGOs and private operations) and weaknessesa can realistically gauge available national capacities 2 Health service utilisation trends Routine health monitoring helps system detect service fluctuations and accurate assessments of crisis impact, and rate of return to baseline after a shock Know risks and population 3 Presence of active epidemiologic Routine surveillance is necessary to detect disease threats and trigger mitigation mechanisms surveillance systema,b 4 Functioning civil registration and Basic knowledge of population demographics is important for estimating health threats and vital statistics system trends, and understand crisis impact Communicate 5 List of decision makers in key Point persons across sectors must be immediately accessible for communication, decision sectorsa making, and sounding alarms 6 Breadth of functioning Communities must be able to notify and sound alarms—this requires an environment of free communication channelsa speech and freedom of press, and functioning, open platforms for timely communication (hotlines, community committees, social media) Diverse Effectively respond to range of 7 Scope of health services available Including services that respond to population health needs and expectations in basic primary health needs in primary carec care package will promote routine health system utilisation and confidence in the health system 8 Quality of care for sentinel Health outcomes, healthcare utilisation during crisis, and trust in health authorities require conditions in basic packagec competent and respectful care Adequately finance health 9 Financing of healthcare: adequacy Total health system funding must be sufficient to support functioning services; financing systems; prevent financial harm of government health expenditure systems should aim to reduce catastrophic and impoverishing health spending63-66 and financial protectionc Self regulating Isolate threat and maintain core 10 Memorandums of understanding Establishing agreement about roles for private providers—not for profit and for profit—in crisis function with non-state providers expands service provision in emergencies and may promote collaboration in times of calm 11 Database of service delivery A routinely updated global, open access library of service delivery models tested and deemed alternatives for affected and effective in past crises promotes inter-country learning and lowers redundant reinvention and unaffected populationsa perpetuation of failed ideas Leverage outside capacity 12 Collaboration agreements with Agreements on nature of collaboration (timing, type of support, roles or responsibilities) regional and global actors during emergencies is a form of smart dependency and contributes to a faster, more effective response29 Integrated Coordinate with non-health actors 13 Existence of a national emergency Ready coordination systems encourages fast decision making and implementation, curbing (education, transport, police, coordination system and leadersa potential effects of emergencies media, private enterprise) 14 Frequency of joint planning Rehearsal of preparedness plans and regular collaboration establishes norms of intersectoral sessions and drillsa teamwork 15 Process for development of a One Acknowledging human ties to the environment and other species encourages an inclusive Health strategyb understanding of public health vulnerabilities Engage citizens and communities 16 Index of Ministry of Health and Quick action in responding to community needs can foster trust and promote containment of to build trust government responsiveness to health shock community need 17 Population trust in health system Trust in government and the health system is essential to effective service delivery and for acceptance of government messages in crises—this is true in government run and mixed provider health systems6768 18 Platforms for dialogue with Regular input about health system functioning from citizens will improve emergency planning community leaders and establish communication channels for routine and emergency needs 19 In-country social scientists with Tapping experts in sociology, anthropology, and related disciplines strengthens experience working with health understanding of key social structures in crisis response, local health determinants and the departments local appropriateness and acceptability of interventions Link healthcare provision to 20 Availability of district health staff Public health staff serve to promote public health practices and act as sentinels for potential public health with public health trainingb outbreaks connecting local clinics to surveillance and monitoring system Coordinate primary and referral 21 Agreement on roles and referral Defined agreements on the role of primary and referral facilities reduces confusion and service care protocols for facilities delay, and streamlines service delivery for patients Adaptive Shift resources to meet need 22 Formal provisions to reallocate Flexible spending of funds—national and international—speeds up and better targets funds in emergency emergency response in fast changing situations Promote rapid local decision 23 Management capacity of district For decentralised responses, local health teams must be able to interpret local data and local making or local health teamsc leaders must be able to make quick and sound operational decisions 24 Agreements on delegation of Pre-crisis agreements permitting local decision making in crisis with sufficient support hasten authority and funding in crises response time to evolving challenges Evaluate to improve 25 Mechanisms for, and capacity to, Rigorous monitoring during crisis and independent evaluation post-crisis permits course track progress and evaluate health correction and points to needed reforms. National capacity for data use and, more broadly, system performance in crisis and in a culture of open inquiry and evaluation needs to be built in times of calm to deliver during a times of calmb crisis. *Characteristics are interrelated and interdependent. Decision making and coordination should occur across these characteristics. a, b, cindicate concepts similar to proposed International Health Regulation, Global Health Security Agenda, and sustainable development goals, respectively.

ience (table 1). They include existing health measures from non-health fields, and new ing, for example) and fast (such as provi- system and preparedness metrics (from proposed measures that need further devel- sions to reallocate money in emergencies) the International Health Regulations, the opment and testing. This proposed resil- drivers of resilience. In contrast to tradi- Global Health Security Agenda, and the ience index balances slow (availability of tional health security frameworks, many Sustainable Development Goals), relevant district health staff with public health train- of our indicators reflect ­characteristics

48 BMJ Global Health Initiatives Analysis

of “everyday” resilience; they not only 2Harvard Medical School, Boston, USA 7 The silver bullet of resilience. Lancet2014;384:930. encourage daily function but also proac- 3 doi:10.1016/S0140-6736(14)61613- Harvard University, Cambridge, USA 5pmid:25220958 tively reduce the likelihood of rising system 4Office of Health Systems, US Agency for 8 United Nations High-level Panel on the Global threats. The index can thus inform develop- International Development, Washington, DC, Response to Health Crises. Protecting humanity from ment of national health plans. It can also future health crises. 2016. www.un.org/News/dh/ USA infocus/HLP/2016-02-05_Final_Report_Global_ expose gaps in function and measurement 5 World Bank, Washington, DC, USA Response_to_Health_Crises.pdf capacity where regional and global coop- 6 9 Moon S, Sridhar D, Pate MA, et al. Will Ebola eration can contribute. American University of Beirut, Beirut, change the game? Ten essential reforms before the Lebanon next pandemic. The report of the Harvard-LSHTM The index does not prescribe national 7 Independent Panel on the Global Response to Ebola. benchmarks. Given the heterogeneity of John Snow Inc., Monrovia, Liberia Lancet2015;386:2204-21. doi:10.1016/S0140- health systems and national contexts, 8Mission of Japan to the European Union, 6736(15)00946-0pmid:26615326 benchmarks for resilience indicators should Brussels, Belgium 10 Commission on a Global Health Risk Framework for the Future. The neglected dimension of global 9 be set within countries to accommodate World Health Organization, Freetown, Sierra security: a framework to counter infectious disease the local context. The next step would be to Leone crises. 2016. www.nap.edu/read/21891/chapter/1 11 World Economic Forum. Managing the risk and review and extend this list as needed, and 10 London School of Hygiene and Tropical impact of future epidemics: options for public-private to develop indicators for the new measure Medicine, London, UK cooperation. 2015. www.weforum.org/reports/ constructs, with input from community 11The Rockerfeller Foundation, New York, USA managing-risk-and-impact-future-epidemics-options- leaders and non-health sector actors. While public-private-cooperation 12UNICEF, New York, USA 12 Kruk ME, Myers M, Varpilah ST, Dahn BT. What is the index is meant to be prospective (used 13 a resilient health system? Lessons from Ebola. in advance of a crisis), some proposed Institute for Healthcare Improvement, Lancet2015;385:1910-2. doi:10.1016/S0140- measures include routinely collected service Cambridge, USA 6736(15)60755-3pmid:25987159 delivery and quality indicators that over 14World Health Organization, Geneva, 13 Zhang X, Miller-Hooks E, Denny K. Assessing the role of network topology in transportation time can indicate the “slope” of resilience Switzerland network resilience. J Transp Geogr2015;46:35- (the extent and speed with which a system 15Ministry of Health, Jakarta, Indonesia 45doi:10.1016/j.jtrangeo.2015.05.006 14 Dalziell EP, McManus ST. Resilience, vulnerability, returns to baseline or better after a shock). 16 Ministry of Economic Cooperation and and adaptive capacity: implications for The validity of the resilience index should Development, Bonn, Germany system performance. 1st International Forum be tested against actual performance during for Engineering Decision Making (IFED). Stoos, Acknowledgments: This paper stemmed from recent health shocks in several settings. Switzerland; 2004. discussions at The Rockefeller Foundation’s Bellagio 15 Anderies J, Ryan P, Walker B. Loss of resilience, crisis, Conclusion Center. We are grateful to The Rockefeller Foundation and institutional change: lessons from an intensive for supporting the meeting and related research agricultural system in southeastern Australia. Before the failure of health systems during on resilience. The views expressed here are the Ecosystems (N Y)2006;9:865-78doi:10.1007/ the Ebola outbreak is forgotten, we need to s10021-006-0017-1 responsibility of the authors and may not reflect the consider how to make them more resistant 16 Allenby B, Fink J. Toward inherently secure and views of the foundation. resilient societies. Science2005;309:1034-6. to crises and more flexible in their response. Author contributions: MEK conceived the idea doi:10.1126/science.1111534pmid:16099973 The concept of resilience adds dynamism 17 Holling C. Resilience and stability of ecological and structure of this paper and wrote the first draft and urgency to the longstanding work of systems. Annu Rev Ecol Syst1973;4:1doi:10.1146/ with assistance from EJL. All authors contributed annurev.es.04.110173.000245 health system strengthening and gives an intellectual content, edited the manuscript, and 18 WHO. Everybody’s business: strengthening health opportunity to learn from other sectors. approved the final version for submission. systems to improve health outcomes.WHO, 2007. 19 Niyato D, Wang P, Hossain E. Reliability Country experiences as varied as Lebanon, Not commissioned; externally peer reviewed Liberia, and Indonesia demonstrate how analysis and redundancy design of smart Competing interests: Fred Martineau has had a grid wireless communications system resilience can be built after health crises. grant from the Wellcome Trust for demand side management. IEEE - Wirel Commun2012;19doi:10.1109/ Proposed measures of health system resil Republished with permission from BMJ 2017;357:j2323 ience can improve our assessment of coun- MWC.2012.6231158 1 Government of the Republic of Sierra Leone. 20 Walker B, Carpenter S, Rockstrom J, Crépin tries’ progress in building resilience and Millennium development goals progress report A-S, Peterson G. Drivers, “slow” variables, indicate areas for action. We hope imple- 2010. 2010. www.undp.org/content/dam/undp/ “fast” variables, shocks, and resilience. Ecol mentation of these ideas can energise policy- library/MDG/english/MDG%20Country%20Reports/ Soc2012;17:art30doi:10.5751/ES-05063-170330 Sierra%20Leone/sierraleone_september2010.pdf 21 Dhillon RS, Kelly JD. Community trust and the makers and ultimately benefit families and 2 Republic of Liberia. Progress, prospects, and Ebola endgame. N Engl J Med2015;373:787-9. communities in times of crisis and beyond. challenges towards achieving the MDGs. 2010. doi:10.1056/NEJMp1508413pmid:26222382 Margaret E Kruk,1 Emilia J Ling,1 www.lr.undp.org/content/dam/liberia/docs/docs/ 22 Gilson L. Trust and the development of health care 2 3 MDG%20Report%20Liberia%202010.pdf as a social institution. Soc Sci Med2003;56:1453- Asaf Bitton, Melani Cammett, 3 Petherick A. Ebola in west Africa: learning the 68. doi:10.1016/S0277-9536(02)00142- 4 5 Karen Cavanaugh, Mickey Chopra, lessons. Lancet2015;385:591-2. doi:10.1016/ 9pmid:12614697 Fadi el-Jardali,6 Rose Jallah Macauley,7 S0140-6736(15)60075-7pmid:25682526 23 Kutalek R, Wang S, Fallah M, Wesseh CS, Gilbert J. Mwihaki Kimura Muraguri, Shiro Konuma,8 4 Mlambo MK, Kamara AB, Nyende M. Ebola interventions: listen to communities. Lancet 9 10 Financing post-conflict recovery in Africa: Glob Health2015;3:e131. doi:10.1016/S2214- Robert Marten, Frederick Martineau, the role of international development 109X(15)70010-0pmid:25618243 11 12 Michael Myers, Kumanan Rasanathan, assistance. J Afr Econ2009;18(Suppl 1):i53-76 24 De Savigny D, Adam T, eds. Systems thinking for Enrique Ruelas,13 Agnès Soucat,14 doi:10.1093/jae/ejp010. health systems strengthening. Alliance for Health Anung Sugihantono,15 Heiko Warnken16 5 Park M. WHO announces changes after Policy and Systems Research.WHO, 2009. widespread Ebola criticism. 2015. 25 Nicholson D, Yates R, Warburton W, Fontana G. Correspondence to: M E Kruk mkruk@hsph. http://edition.cnn.com/2015/05/18/health/ Delivering universal health coverage: a guide for harvard.edu who-ebola-reform policymakers. Report of the WISH Universal Health 1 6 Gostin LO. Reforming the World Health Organization Coverage Forum 2015. 2015. http://wish-qatar. Harvard TH Chan School of Public Health, after Ebola. JAMA2015;313:1407-8. doi:10.1001/ org/2015-summit/reportsen/universal-health- Boston, US jama.2015.2334pmid:25871658 coverage-en

BMJ Global Health Initiatives 49 Analysis

26 United Nations General Assembly. Implementation 42 Lebanon Ministry of Public Health. Maintaining health 55 Aceh redux: The tsunami that helped stop a war. of the International Strategy for Disaster Reduction. security, preserving population health, and saving 2014. www.irinnews.org/report/100960/aceh- United Nations, 2015. children and womens lives. 2016. www.moph.gov.lb/ redux-tsunami-helped-stop-war 27 United Nations General Assembly. DynamicPages/download_file/1669 56 Indonesia insists on restricting relief workers in Transforming our world: the 2030 agenda 43 United Nations Children’s Fund, United Nations High tsunami hit Aceh. http://reliefweb.int/report/ for sustainable development. 2015. https:// Commissioner for Refugees, United Nations World indonesia/indonesia-insists-restricting-relief- sustainabledevelopment.un.org/post2015/ Food Programme. Vulnerability assessment of Syrian workers-tsunami-hit-aceh. 2005. transformingourworld refugees in Lebanon, 2016. 2016. http://documents. 57 Gadjah Mada University, International Recovery 28 Kutzin J, Sparkes SP. Health systems strengthening, wfp.org/stellent/groups/public/documents/ena/ Platform. The Recovery Status Report: The Yogyakarta universal health coverage, health security and wfp289533.pdf and Central Java Earthquake 2006.International resilience. Bull World Health Organ2016;94:2. 44 United Nations High Commissioner for Refugees. Recovery Platform, 2009. doi:10.2471/BLT.15.165050pmid:26769987 Refugees from Syria.2015. 58 Leitmann J. Cities and calamities: learning from 29 Rodin J. The resilience dividend: being strong in a 45 World Health Organization. The Ebola outbreak in post-disaster response in Indonesia. J Urban world where things go wrong.1st ed. Public Affairs, Liberia is over. 2015. www.who.int/mediacentre/ Health2007;84(Suppl):144-53. doi:10.1007/ 2014. news/statements/2015/liberia-ends-ebola/en s11524-007-9182-6pmid:17356900 30 Grove KJ. Security beyond resilience. Environ Plann D 46 Matanock A, Arwady MA, Ayscue P, et al. Centers for 59 Elnashai AS, Amr S, Kim SJ, Yun GJ, Sidarta D. The Soc Space2017;35:184-94doi: Disease Control and Prevention. Ebola virus disease Yogyakarta earthquake of May 27, 2006. Mid- 10.1177/0263775816686583 cases among health care workers not working in America Earthquake Center. 2007. www.ideals. 31 Stephanie M. Topp WF, Veena Sriram, Kerry Scott. Ebola treatment units: Liberia, June-August, 2014. illinois.edu/handle/2142/8934 Critiquing the concept of resilience in health systems. MMWR Morb Mortal Wkly Rep2014;63:1077-81. 60 World Health Organization. Indonesia: improving News & Commentary. Health Systems Global, 2016. pmid:25412067 health services during emergencies. 2014. www.who. 32 Van de Pas R. Beyond resilience. 2015. www. 47 Peters MM. Community perceptions of Ebola int/features/2014/emergencies-indonesia/en internationalhealthpolicies.org/beyond-resilience response efforts in Liberia, Montserrado and 61 Health Minister Regulation on Regionalization of 33 Béné C, Wood RG, Newsham A, Davies M. Resilience: Nimba counties. 2014. www.ebola-anthropology. Centre for Post-Disaster Health Crisis Mitigation. No new utopia or new tyranny? reflection about the net/case_studies/community-perceptions-of-ebola- 783.Government of Indonesia, 2006. potentials and limits of the concept of resilience in response-efforts-in-liberia-montserrado-and-nimba- 62 Ministry of Health.Technical Guidance for Post- relation to vulnerability reduction programmes. IDS counties Disaster Health Crisis Mitigation, 2007. Working Papers 2012(405):1-61. 48 Kobayashi M, Beer KD, Bjork A, et al. Community 63 Hsu J, Price M, Huang J, et al. Unintended 34 United Nations Office for the Coordination of Knowledge, Attitudes, and Practices Regarding consequences of caps on Medicare drug benefits. Humanitarian Affairs. Lebanon crisis response plan Ebola Virus Disease - Five Counties, Liberia, N Engl J Med2006;354:2349-59. doi:10.1056/ 2015: annual report. 2016. http://reliefweb.int/ September-October, 2014. MMWR Morb Mortal Wkly NEJMsa054436pmid:16738271 report/lebanon/lebanon-crisis-response-plan-2015- Rep2015;64:714-8.pmid:26158352 64 Joyce GF, Escarce JJ, Solomon MD, Goldman DP. annual-report. 49 Pellecchia U, Crestani R, Decroo T, Van den Employer drug benefit plans and spending on 35 Moon S, Sridhar D, Pate MA, et al. Will Ebola Bergh R, Al-Kourdi Y. Social Consequences of prescription drugs. JAMA2002;288:1733-9. change the game? Ten essential reforms before the Ebola Containment Measures in Liberia. PLoS doi:10.1001/jama.288.14.1733pmid:12365957 next pandemic. The report of the Harvard-LSHTM One2015;10:e0143036. doi:10.1371/journal. 65 Arsenijevic J, Pavlova M, Groot W. Measuring Independent Panel on the Global Response to Ebola. pone.0143036pmid:26650630 the catastrophic and impoverishing effect of Lancet2015;386:2204-21. doi:10.1016/S0140- 50 Nyenswah TG, Kateh F, Bawo L, et al. Ebola household health care spending in Serbia. Soc Sci 6736(15)00946-0pmid:26615326 and its control in Liberia, 2014-2015. Emerg Med2013;78:17-25. doi:10.1016/j.socscimed.201 36 United Nations. Protecting Humanity from Future Infect Dis2016;22:169-77. doi:10.3201/ 2.11.014pmid:23267776 Health Crises of the High-level Panel on the Global eid2202.151456pmid:26811980 66 Dalal K, Aremu O. Fairness of utilizing Response to Health Crises.United Nations, 2016. 51 Fallah M, Dahn B, Nyenswah TG, et al. Interrupting health care facilities and out-of-pocket 37 Republic of Lebanon, Ministry of Public Health. Ebola Transmission in Liberia Through Community- payment burden: evidence from Cambodia. Health response strategy: a new approach in 2016 & Based Initiatives. Ann Intern Med2016;164:367-9. J Biosoc Sci2013;45:345-57. doi:10.1017/ beyond.Ministry of Public Health, 2015. doi:10.7326/M15-1464pmid:26746879 S0021932012000521pmid:22958391 38 United Nations High Commissioner for Refugees. 52 Borrero JC. Field survey: northern Sumatra 67 Ackatia-Armah NM, Addy NA, Ghosh S, Syria regional response plan: January to June and Banda Aceh, Indonesia and after the Dubé L. Fostering reflective trust between 2013. 2012. www.unhcr.org/uk/partners/ earthquake and tsunami of 26 December 2004. mothers and community health nurses to donors/50d192fd9/syria-regional-response-plan- 2005. www.eeri org/lfe/clearinghouse/sumatra_ improve the effectiveness of health and january-june-2013.html tsunami/observ1_php. nutrition efforts: An ethnographic study in Ghana, 39 Ammar W, Kdouh O, Hammoud R, et al. Health 53 Ministry of Health Indonesia. Tsunami aftermath in West Africa. Soc Sci Med2016;158:96-104. system resilience: Lebanon and the Syrian refugee Nanggroe Aceh Darussalam (NAD) province and its doi:10.1016/j.socscimed.2016.03.038pm crisis. J Glob Health2016;6:020704. doi:10.7189/ effects on health services: a rapid assessment for id:27131046 jogh.06.020704pmid:28154758 policy formulation, 2005. 68 Giordano GN, Lindström M. Trust and health: 40 Government of Lebanon and the United Nations. 54 United Nations Information Management testing the reverse causality hypothesis. Lebanon crisis response plan 2015-2016. 2014. Service, Rehabilitation and Reconstruction J Epidemiol Community Health2016;70:10-6. www.alnap.org/resource/20702 Agency. Tsunami recovery status report, doi:10.1136/jech-2015-205822pmid:26546287 41 United Nations High Commissioner for Refugees. December 2005. reliefweb.int/sites/reliefweb. Syria regional response plan. 2014. www.unhcr.org/ int/files/resources/84D4D7659153EE7949257 Cite this as: BMJ 2017;357:j2323 syriarrp6 0D800203196-unorc-idn-6dec.pdf http://dx.doi.org/10.1136/bmj.j2323

50 BMJ Global Health Initiatives COMMENTARY

Productive disruption: opportunities and challenges for innovation in infectious disease surveillance

Background to collect and analyse data to improve allocation decisions. Figure 1 illustrates the Infectious diseases place an unacceptable spatial estimates of disease burden using flow of data and potential hurdles faced by and disproportionate social and economic new Big Data sources, mobile-Health or national control programmes and the ways burden on low-income countries. National m-Health approaches or mechanistic and in which new approaches may be used in disease control programmes have the dif- parallel to traditional systems. statistical modelling techniques. While ficult task of allocating limited budgets for Data quality is often perceived as a major these advances leap ahead, however, interventions across regions of their coun- barrier to using passive surveillance data tries, based on often disparate datasets of many remain most useful for estimating to guide resource prioritisation. Data may 1 varying quality from a range of sources global disease distribution, rather than for be delayed or simply missing from core including clinics, hospitals, village health national control programme prioritisation. settings and, even where available, is workers, the private sector and non-govern- Translating these new techniques to frequently under-reported, potentially in 2 mental organisations (NGOs). Every stage inform policy in endemic settings remains spatially and temporally variable ways. For of the data collection and analysis pipeline challenging. The pronounced disconnect example, a problematic reporting pattern for surveillance systems may be affected by that emerges repeatedly is an apparent between health systems and academia a lack of capacity as well as by biases and increase in disease incidence that is may limit the utility of new approaches. The misaligned incentives for reporting and actually caused by increased surveillance managing data. Addressing these issues high burden of work placed on healthcare efforts and/or diagnostic capacity. Over- will be essential for effective reduction in workers in low-income settings further reporting is also a potential hazard when the burden of endemic infectious diseases limits their scope and time available local regions are financially incentivised globally as well as to preparing for emerg- for engagement with methodological to exaggerate their needs. Further, since ing epidemic threats. developments. many local health centres lack diagnostic Meanwhile, academic researchers—often capabilities, much of the large-scale data Despite ongoing challenges to in high-income settings—are developing rely on syndromic surveillance (influenza- implementation, however, there are increasingly sophisticated methods like-illness, diarrhoea, fever) with low promising analytical approaches that can specificity. Additionally, estimates of the leverage even patchy and low-quality data catchment population of health facility Summary box and diverse new data streams that can or district may be flawed, since data may be based on a single census from up to a • New innovations that could trans- be productively harnessed to strengthen form infectious disease surveillance decade ago, thus potentially both out of strategies for resource allocation when 3 and control, including the use of Big date and a poor reflection of seasonal integrated with existing surveillance 2 Data, mobile health approaches and fluctuations in population numbers. cutting edge quantitative methods, systems. We detail the data and analysis Collection of more detailed, highly curated offer hope for disrupting traditional challenges faced by national disease data at sentinel sites seems promising, but health systems and improving health control programmes, outline possible may add little to national decision-making because such high-quality sites are likely worldwide. solutions offered by analytical approaches to be limited in their spatial scale. All these • Much has been made of their poten- and new data-streams and conclude by tial, but very few have been trans - issues mean that where case numbers exist, outlining barriers to implementation. lated successfully into policy or scaled policy makers may have little confidence in up to a population level. them. Challenges associated with infectious disease We argue that there is currently a • surveillance systems Leveraging fragile data using statistical and lack of integration of new approaches, Generally, epidemiological data about mathematical principles making them unsustainable or unre- patients are reported by healthcare practi- alistic for most national control pro- The first question is what can be done with tioners via passive surveillance systems to grammes and that the gulf between existing infrastructure and data. With the a central database, which is used to deter- academia and policy makers remains a right statistical or theoretical tools, even major barrier to their implementation. mine trends over time in and map the geo- low-quality data can potentially be lever- graphic distribution of burden of disease aged to help inform strategic deployment We propose that these innovations • in different regions as well as the extent of control efforts. Thoughtful deployment must be designed with direct input and efficacy of interventions (active sur- of interpolation or geostatistical tools can from national control programmes veillance via sentinel sites may also inform be used to create smooth maps of burden and embedded within already exist- these efforts). These regional data in turn or intervention efforts across space, also ing health systems. serve as an important basis for resource allowing extrapolation to unmeasured

BMJ Global Health Initiatives 51 COMMENTARY

1 4 5 contexts.­ Autocorrelation models are or R0, which captures the degree to which parameters of particular pathogens can

also powerful tools, building on surveil- an outbreak is expected to grow (R0 >1) establish the degree to which the data

lance data to guide predictions about out- or shrink (R0 <1) (eg, deployed during the are reliable and forecast outbreaks and breaks of dengue, for example.6 Moving recent Ebola outbreak despite variation in emergence events and/or the impact of from statistical to mechanistic approaches, reporting rates).9 Allocation of resources interventions like vaccination (eg, roll- even if incidence reporting is erratic, towards ‘source’ populations, where out of cholera vaccination to contain an 13 dynamical signatures of the infectious pro- R0>1 then becomes possible—although epidemic). cess might still be detectable if additional maps of the locations and densities of rural data on features of cases are available, such populations (ie, denominator challenges) Complementing existing epidemiological as age, geographic location and gender. are also necessary. Infectious disease information with new data sources Age is a powerful covariate for infectious models can allow characteristics of the While these analytical strategies can com- disease dynamics, as age of infection is surveillance system, such as the magnitude pensate for the limitations of different data linked to the magnitude of transmission. of under-reporting, to be estimated where quality issues, a range of promising new High-transmission pathogens often have a the susceptible population can be inferred data streams are also available. Rapid tech- low average age of infection, as they move (eg, via susceptible reconstruction).10 nological advances make these increasingly quickly through immunologically naïve Where only syndromic surveillance is affordable, offering additional or new data populations.7 Conversely, implementation available, it may be possible to correct layers to include in epidemiological analy- of control efforts is likely to increase the for background rates of focal syndromes ses.1 14 15 We focus on three new data types average age of infection,8 so intervention to pull out the dynamics associated with that are tractable in the context of control efficacy may be measured using shifting a particular infection. This strategy has programme capacity: geospatial data, pas- age structure of cases. been used for influenza11 (influenza- sively collected mobile phone records and Where data are consistent through like-illness data are frequently available pathogen genomic data. Note that we do time, but not space, basic principles from but data on influenza are rare), enabling not discuss the many mHealth approaches infectious disease dynamics open the way investigation of signatures of climate effects to actively engaging with populations to estimating characteristics of pathogens. on the burden of infection,12 for example, directly, for surveillance or for interven- For example, the growth rate of an epidemic which has potential to contribute to tions like health education,16 but these also can be extracted from incidence, allowing planning efforts. Alternatively, simulation provide data that do not rely on traditional estimates of the net reproduction number, tools based on known epidemiological surveillance systems. In general, different

Fig 1 | Data flows through health systems (blue) and major challenges faced by control programmes (red). A subset of clinical cases, which often represent only a subset of total infections both asymptomatic and clinical, are first detected by local health workers, most typically in health facilities and hospitals. Local health workers are also responsible for following up individuals with chronic infections requiring multiple treatments over months or years. Some fraction of clinical cases are lab confirmed, depending on capacity, and reported to regional or district centres, which in turn report to national control programmes. Data are often aggregated before being reported centrally. NGOs and the private sector may also produce a significant amount of epidemiological data. National control programmes aggregate and analyse data to map the distribution of disease burden, intervention efficacy and so on. New direct mHealth approaches (eg, participatory surveillance) and passively collected data (eg, from mobile phones via Call Data Records (CDRs); and satellites) may be used directly by control programmes to map underlying risks and population distributions. At every level, capacity remains an enormous issue for routine surveillance, and training for new approaches will be challenging for most control programmes. At different levels of the health system, incentives for reporting accurately may be misaligned, and timeliness of reporting may be particularly problematic for emerging threats. NGOs, non-governmental organisations

52 BMJ Global Health Initiatives COMMENTARY

needs dominate at different phases across subscriber privacy.18 Indeed, the Genome sequencing costs have declined an epidemiological spectrum from emer- development of appropriate anonymisation strikingly in the last decades, making gence to elimination (figure 2), which will and aggregation protocols remains an it increasingly feasible for control determine which data are useful and how important priority for academics and programmes to integrate pathogen they should be analysed. public health practitioners and will require sequencing and molecular epidemiology Geospatial data are increasingly of high carefully balancing the ethical risks of into their data collection and analysis resolution and encompasses settlements reidentifying individuals with the benefits strategies. Unlike mobile phone data, and transport networks, indices of of predicting disease spread and identifying which offers an external view of mobility vegetation coverage, land use, land surface targets for intervention. Integration of this that can be used to model disease spread, temperature and wind speed. These data information into risk mapping by control analysis of pathogen genetic data provides can be combined with other geospatially programmes offers particular promise complementary insights into transmission referenced data such as meteorological where disease incidence is heterogeneous chains and pathogen gene flow between data from weather stations, population in time and space and mobility drives both locations. During outbreaks, sequencing densities or road networks to generate changing burden of infection and the can provide insights into the place and comprehensive estimates of environmental type of intervention needed. For example, time of the outbreak’s origin and the pace variables or indicators like remoteness during an emerging epidemic (eg, Ebola), of its spatial spread.20 For elimination or urbanicity relevant to communicable spatial containment of the disease and planning, pathogen sequencing can disease transmission. This can then be proactive surveillance (in the correct identify the regional or national origin of combined with point pattern data on vector locations) are essential, requiring specific presence via machine learning algorithms spatial targets and estimates of how people a particular isolate and estimate the rate to determine features such as the likely will move the infection to new regions.19 of migration between populations where range and local transmission intensity of At the other end of the spectrum, control this information is difficult to measure by vector borne infections like Zika virus.17 programmes aiming for elimination (eg, other means. For endemic pathogens like Mobile phone data—routinely collected malaria, measles) require accurate maps TB, where low incidence and relatively by operators and providing information of remaining foci of transmission and an stable prevalence in many places makes about the location and movements of understanding of the relative importance analytical inferences and spatial data from subscribers in real time—offer tremendous of local versus imported cases of disease.18 mobile phones or satellites less tractable, promise for control programmes to These data can also be leveraged to address genomics can provide key insights into measure disease spread, if appropriately issues in estimation of the denominator or the spread of drug resistance and the anonymised and aggregated to protect population at risk.3 connectivity between different populations.

Fig 2 | Optimal use of new approaches depends on epidemiological context. Different phases of epidemiological containment and control lend themselves to different analytical approaches and data sources. Here, we have highlighted the spatial dimensions of this issue, with emergence and elimination phases exhibiting high spatial heterogeneity. In these cases, pronounced heterogeneity produces signals in data that can be leveraged to model the spread of infection between populations. For endemic infections where prevalence is distributed throughout the country and controlling disease burden is the primary purpose of interventions, the use of age profiles of exposure and other analytical approaches may be used to enhance or make use of patchy or poor quality data

BMJ Global Health Initiatives 53 COMMENTARY

Engaging community health workers and Ministries of Health, donor agencies and Provenance and peer review: Not commissioned; researchers who collect and analyse data academics. externally peer reviewed. is key Caroline O. Buckee,1 Maria I E Cardenas,2 Data sharing statement: No additional data are 3 4 5 While our focus is on the potential of June Corpuz, Arpita Ghosh, Farhana Haque, available. Jahirul Karim,6,7 Ayesha S. Mahmud,1 existing data despite perceived inadequa- Richard J Maude,1,7,8 Keitly Mensah,9 © Article author(s) (or their employer(s) unless cies and in use of novel data streams, it is Nkengafac Villyen Motaze,10 otherwise stated in the text of the article) 2018. All important to note that country experience Maria Nabaggala,11 Charlotte Jessica Eland Metcalf,9 12 11 rights reserved. No commercial use is permitted points to many low-cost adjustments to Sedera Aurélien Mioramalala, Frank Mubiru, Corey M. Peak,1 Santanu Pramanik,4 unless otherwise expressly granted. 13 14 practice that could improve data availabil- Jean Marius Rakotondramanga, Eric Remera, Handling editor: Alberto L Garcia-Basteiro ity. In multiple settings, there are important Ipsita Sinha,7,8 Siv Sovannaroth,15 16 7 Republished with permission from BMJ Glob Health opportunities to share data across diseases, Andrew J Tatem, Win Zaw 2018;3:e000538. breaking down silos between disease sur- Correspondence to: Dr Caroline O. Buckee veillance and/or control programmes. This cbuckee@ hsph. harvard.edu effort would be enhanced by electronic 1Department of Epidemiology, Center for rather than paper-based reporting systems, Communicable Disease Dynamics, Harvard School of Public Health, Boston, Massachusetts, USA This is an open access article distributed in allowing standardised data collection, 2Centro de Atención y Diagnóstico de Enfermedades accordance with the terms of the Creative Commons aggregation and sharing. Perhaps most Infecciosas, Universidad Industrial de Santander, Attribution (CC BY 4.0) license, which permits others importantly, however, effectively engaging Bucaramanga, Colombia to distribute, remix, adapt and build upon this work, community health workers and others who 3Epidemiology Bureau, Department of Health of the for commercial use, provided the original work is Philippines, Manila, Philippines properly cited. See: http://creativecommons.​ org/​ ​ collect and report patient data will be key licenses/by/​ 4.​ 0/​ to strengthening surveillance. For example, 4Public Health Foundation of India, Vasant Kunj, New in Cambodia, malaria case reporting was Delhi, India 1 Hay SI, George DB, Moyes CL, et al. Big data 5 opportunities for global infectious disease improved when healthcare workers received Programme for Emerging Infections (PEI), Infectious Diseases Division (IDD), ICDDR, B, Dhaka, Bangladesh surveillance. PLoS Med 2013;10:e1001413. feedback about the incidence of malaria in 2 Buckee CO, Tatem AJ, Metcalf CJ. Seasonal 6Disease Control Department, Directorate General of population movements and the surveillance and their area using colour-coded stickers; the Health Services, Dhaka, Bangladesh control of infectious diseases. Trends Parasitol 7 effective use of mobile phones and tablets Mahidol-Oxford Tropical Medicine Research Unit, 2017;33:10–20. for reporting cases improved when health Faculty of Tropical Medicine, Mahidol University, 3 Tatem AJ. Mapping the denominator: spatial workers could use them for personal use Bangkok, Thailand demography in the measurement of progress. Int 8 and where internet coverage was restricted, Centre for Tropical Medicine and Global Health, Health 2014;6:153–5. Nuffield Department of Medicine, University of Oxford, 4 Kraemer MU, Hay SI, Pigott DM, et al. Progress and the flexible reporting using mobile phones Oxford, UK challenges in infectious disease cartography. Trends Parasitol 2016;32:19–29. via SMS and later smartphones greatly 9Dept of Ecology and Evolutionary Biology, Princeton 5 Takahashi S, Metcalf CJE, Ferrari MJ, et al. The improved timeliness of data collection. University, Princeton, NJ, USA geography of measles vaccination in the African 10 It will be essential to sustain resources Division of Epidemiology and Biostatistics, Great Lakes region. Nat Commun 2017;8:15585. for engaging with health workers involved Stellenbosch University, Stellenbosch, South Africa 6 Johansson MA, Reich NG, Hota A, et al. in generating and reporting surveillance 11Infectious Disease Institute, College of Health Evaluating the performance of infectious disease data and not to impose technological Sciences, Makarere University, Uganda forecasts: A comparison of climate-driven and 12 seasonal dengue forecasts for Mexico. Sci Rep M&E Service, Data & Malaria Survey Division, advances at the expense of motivating 2016;6:33707. National Malaria Control Program, Antananarivo, 7 Anderson RM, May RM. Infectious diseases of and retaining qualified, reliable workers Madagascar humans. Oxford: Oxford University Press, 1991. who form the foundation of an effective 13 Epidemiology Unit, Institut Pasteur, Madagascar 8 Ferrari MJ, Grenfell BT, Strebel PM. Think globally, health system. Barriers to implementing 14Rwanda Biomedical Center, Kigali, Rwanda act locally: the role of local demographics and approaches that use novel types of data 15Technical Bureau, National Malaria Control Program, vaccination coverage in the dynamic response of measles infection to control. Philos Trans R Soc Lond or new analytical tools are rooted in Pnom Penh, Cambodia B Biol Sci 2013;368:20120141. 16 human capacity (figure 1) at every level WorldPop, Department of Geography and 9 Shaman J, Yang W, Kandula S. Inference and within control programmes, from local Environment, University of Southampton, forecast of the current west african ebola healthcare centres up to the ministry of Southampton, UK outbreak in Guinea, sierra leone and liberia. PLoS health. While a considerable strength of Contributors: All authors conceived, planned and Curr 2014;6. 10 Bjørnstad ON, Finkenstadt B, Grenfell BT. novel data-streams such as satellite or contributed intellectually to this work. COB and CJEM Endemic and epidemic dynamics of measles: mobile phone data are that they bypass wrote the manuscript, and all authors edited the estimating epidemiological scaling with a many of the levels of reporting where data manuscript. time series SIR model. Ecological Monographs 2002;72:169–84. may be lost and incentives are misaligned, Funding: This work was the result of the 11 Charu V, Zeger S, Gog J, et al. Human mobility the insights they provide will be limited for Harvard-Wellcome Trust Workshop on Infectious and the spatial transmission of influenza setting public health agendas if they are Disease Modeling for Control Programs in in the United States. PLoS Comput Biol 2017. It was supported by Wellcome Trust Fund 2017;13:e1005382. not integrated with traditional reporting 12 Fisman DN, Tuite AR, Brown KA. Impact of el systems. For academics, one of the most 106866-Sustaining Health Award and by Award niño southern oscillation on infectious disease important contributions in this regard Number U54GM088558 from the National Institute hospitalization risk in the United States. Proc Natl will be providing training, so that public of General Medical Sciences. Acad Sci U S A 2016;113:14589–94. 13 Azman AS, Luquero FJ, Rodrigues A, et al. health officers have the technical capacity Disclaimer: The content is solely the responsibility Urban cholera transmission hotspots and their to understand the benefits of, and oversee of the authors and does not necessarily represent implications for reactive vaccination: evidence the implementation of, new methods. the official views of the National Institute of General from Bissau city, Guinea bissau. PLoS Negl Trop Dis Medical Sciences or the National Institutes of Health. 2012;6:e1901. Sustainable integrating of these approaches 14 Tatem AJ. Mapping population and pathogen will require sustained commitment from Competing interests: None declared. movements. Int Health 2014;6:5–11.

54 BMJ Global Health Initiatives COMMENTARY

15 Haque F, Ball RL, Khatun S, et al. 17 Messina JP, Kraemer MU, Brady OJ, et al. Mapping 20 Arias A, Watson SJ, Asogun D, et al. Rapid outbreak Evaluation of a smartphone global environmental suitability for Zika virus. Elife sequencing of Ebola virus in Sierra Leone identifies decision-support tool for diarrheal 2016;5:e15272. transmission chains linked to sporadic cases. Virus disease management in a resource-limited 18 Wesolowski A, Eagle N, Tatem AJ, et al. Quantifying Evol 2016;2:vew016. setting. PLoS Negl Trop Dis the impact of human mobility on malaria. Science 2017;11:e0005290. 2012;338:267–70. Cite this as: Buckee CO.,Cardenas MIE, Corpuz J, et al. 16 Tomlinson M, Rotheram-Borus MJ, 19 Salje H, Lessler J, Endy TP, et al. Revealing the Productive disruption: opportunities and challenges Swartz L, et al. Scaling up mHealth: microscale spatial signature of dengue transmission for innovation in infectious disease surveillance. where is the evidence? PLoS Med and immunity in an urban population. Proc Natl Acad BMJ Global Health 2018;3:e000538. 2013;10:e1001382. Sci U S A 2012;109:9535–8. doi:10.1136/bmjgh-2017-000538

References - Foreword, page 1: 1 World Health Organization. Declaration of alma-ata. 2016 http://apps.who.int/gb/ebwha/pdf_files/ environment interface in low-resource settings. PLoS International conference on primary health care. WHA69/A69_39-en.pdf?ua=1; (accessed 17 Negl Trop Dis. 2014;8(11):e3257 Alma-ata, USSR Sept 6–12:1978. September 2018). 8 World Economic Forum. Global Disease Outbreaks 2 Cueto M. The ORIGINS of Primary Health Care and 6 Bloom G, Merrett GB, Wilkinson A, et al. Antimicrobial [Internet]. 2018. Available from: http://reports.weforum. SELECTIVE Primary Health Care. American Journal of resistance and universal health coverage. BMJ Glob org/global-risks-2016/global-disease-outbreaks/ Public Health. 2004;94(11):1864-74. PubMed PMID: Health 2017;2:e000518. doi:10.1136/ bmjgh-2017- 9 Gronvall G, Boddie C, Knutsson R, Colby M. One 15514221. 000518 health security: an important component of the 3 Walsh J, Warren K. Selective Primary Health Care. New 7 Gebreyes WA, Dupouy-Camet J, Newport MJ, Oliveira global health security agenda. Biosecur England Journal of Medicine. 1979;301(18):967-974. CJ, Schlesinger LS, Saif YM, Kariuki S, Saif LJ, Saville W, Bioterror. 2014 Sep-Oct;12(5):221-4. 4 http://www.undp.org/content/undp/en/home/ Wittum T, Hoet A, Quessy S, Kazwala R, Tekola B, Shryock 10 Pandemic preparedness and strengthening health sustainable-development-goals.html T, Bisesi M, Patchanee P, Boonmar S, King LJ. The global systems [Internet]. World Bank. 2017 [cited 18 5 World Health Organization. Framework on integrated, one health paradigm: challenges and opportunities for September 2018]. Available from: http://www. people centred health services: report by the secretariat. tackling infectious diseases at the human, animal, and worldbank.org/en/topic/pandemics#1

BMJ Global Health Initiatives 55 Global Health Initiatives Better health outcomes for all

BMJ Global Health Initiatives Working with international partners, our Global empower health professionals in Health Initiatives focus on low and middle income countries 1. Building health research capacity. to improve patient care with the 2. Connecting thought leaders. best available knowledge, decision 3. Strengthening health systems. support and educational tools. 4. Sharing and disseminating knowledge.

Partner with us to make a diff erence Find out more: bit.ly/BMJGH

Contact Mitali Wroczynski Head of Strategic Partnerships, BMJ [email protected]