Thesis (Complete)

UvA-DARE (Digital Academic Repository)

Engaging the private sector in public health challenges in Namibia de Beer, I.H.

Publication date 2017 Document Version Final published version License Other Link to publication

Citation for published version (APA): de Beer, I. H. (2017). Engaging the private sector in public health challenges in Namibia.

General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

Download date:06 Oct 2021

INGRID DE BEER

NAMIBIA CHALLENGES IN PUBLIC HEALTH IN PUBLIC HEALTH PRIVATE SECTOR PRIVATE ENGAGING THE

Engaging the Private Sector in Public Health Challenges in Namibia | Ingrid de Beer ENGAGING THE PRIVATE SECTOR IN PUBLIC HEALTH CHALLENGES IN NAMIBIA

INGRID DE BEER The studies included in this thesis were a result of operational and original research conducted on PharmAccess Foundation Namibia programs over a period of 12 years. The studies were financially supported by USAID, the Postcode Loterij, through STOP AIDS NOW and the AIDSFONDS and the Ministry of Foreign Affairs Netherlands.

PharmAccess Foundation is a Netherlands based organization, with offices in Nigeria, Ghana, Kenya, Tanzania and until 2016 in Namibia. The organization is dedicated to improving access to healthcare in Africa with innovative financing mechanisms. PharmAccess mobilizes public and private resources for the benefit of doctors and patients through health insurance, loans to doctors, clinical standards and quality improvements, private investment and operational research (www.pharmaccess.org).

2 ENGAGING THE PRIVATE SECTOR IN PUBLIC HEALTH CHALLENGES IN NAMIBIA

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Universiteit van Amsterdam op gezag van de Rector Magnificus

prof.dr.ir.K.I.J Maex ten overstaan van een door het College voor Promoties ingestelde commissie, in het openbaar te verdedigen in de Agnietenkapel op

dinsdag 19 december 2017, te 12.00 uur

door Ingrid Helene de Beer geboren te Bruck an der Mur, Oostenrijk

3 Promotiecommissie: Promotor: prof. dr. T.F. Rinke de Wit AMC - UvA Copromotor: prof. dr. J. van der Gaag AMC - UvA Overige leden: prof. dr. F.G.J. Cobelens AMC - UvA prof. dr. A.P. Hardon Universiteit van Amsterdam prof. dr. M.P. Prahdan Universiteit van Amsterdam dr. F.G. Feeley Boston University School of Public Health prof. dr. S.R. Moyo Namibia University of Science and Technology

Faculteit der Geneeskunde

4 LIST OF ACRONYMS 7

CHAPTER 1 INTRODUCTION 1 9

1.1 Background to the PharmAccess approach to improving healthcare 9 1.1.1 The healthcare landscape in Namibia, 2004 11 1.1.2 The PharmAccess interventions in Namibia 2004-2016 13 1.2 Engagement of the private sector 14 1.3 Strategies to stimulate demand by improving health awareness of patient/client 14 1.4 Strategies to stimulate supply by developing innovative new healthcare service provision 16 1.5 Strategy to innovate new payer mechanisms using temporary subsidization 16 1.6 Strategies to collect and present evidence to support policy-making 17 1.7 Hypotheses 17 References 18

CHAPTER 2 DIAGNOSTIC ACCURACY OF 2 ORAL FLUID-BASED TESTS FOR HIV 21 SURVEILLANCE IN NAMIBIA

CHAPTER 3 ANONYMOUS HIV WORKPLACE SURVEYS AS AN ADVOCACY TOOL FOR 25 AFFORDABLE PRIVATE HEALTH INSURANCE IN NAMIBIA

CHAPTER 4 UNIVERSITY STUDENTS AND HIV IN NAMIBIA: AN HIV PREVALENCE SURVEY 33 AND A KNOWLEDGE AND ATTITUDE SURVEY

CHAPTER 5 HEALTH CARE OPTIONS FOR COMMERCIAL FARM WORKERS IN NAMIBIA 43

CHAPTER 6 ASSESSING THE COSTS OF MOBILE VOLUNTARY COUNSELING AND TESTING 57 AT THE WORK PLACE VERSUS FACILITY BASED VOLUNTARY COUNSELING AND TESTING IN NAMIBIA

CHAPTER 7 MOBILE PRIMARY HEALTHCARE SERVICES AND HEALTH OUTCOMES OF 69 CHILDREN IN RURAL NAMIBIA

CHAPTER 8 INNOVATION IN NAMIBIA: PRESERVING PRIVATE HEALTH INSURANCE AND 83 HIV/AIDS TREATMENT

5 CHAPTER 9A INCIDENCE OF HIV IN WINDHOEK, NAMIBIA: DEMOGRAPHIC AND 93 SOCIO-ECONOMIC ASSOCIATIONS

CHAPTER 9B A CAUTIOUS NOTE ON HOUSEHOLD SURVEYS IN POOR SETTINGS 103

CHAPTER 10 PREVALENCE AND KNOWLEDGE ASSESSMENT OF HIV AND 105 NON-COMMUNICABLE DISEASE RISK FACTORS AMONG FORMAL SECTOR EMPLOYEES IN NAMIBIA

CHAPTER 11 DIABETES, HIV AND OTHER HEALTH DETERMINANTS ASSOCIATED WITH 119 ABSENTEEISM AMONG FORMAL SECTOR WORKERS IN NAMIBIA

CHAPTER 12 DISCUSSION & RECOMMENDATIONS 133 Strategy 1: To stimulate demand by improving health awareness of the patient/client 138 Strategy 2: To stimulate supply by developing innovative new healthcare service provision 141 Strategy 3: To innovate new payer mechanisms using temporary subsidization 143 Strategy 4: To collect and present evidence to support policy-making 145 Reflection and general recommendations 146 Conclusion 151 References 152

ADDENDUM 157 Summary (in English) 157 Samenvatting (Dutch summary) 161 List of Publications 166 PhD Portfolio 167 Acknowledgements 171 About the author 172

6 LIST OF ACRONYMS

AIDS Acquired Immuno-Deficiency Syndrome ART Anti-Retroviral Therapy CoW City of Windhoek CBO Community-based Organization CSO Civil Society Organization FBO Faith Based Organization GFATM Global Fund for HIV/Aids, TB and Malaria HBC Healthworks Business Coalition HIF Health Insurance Fund HIV Human Immunodeficiency Virus HIVREF Health Is Vital Risk Equalization Fund KYE Know Your Epidemic KYR Know Your Response MoHSS Ministry of Health and Social Services NABCOA Namibia Business Coalition on AIDS NAMAF Namibia Medical Aids Federation NCD Non-Communicable Diseases NGO Non-Governmental Organization NIMART Nurse Initiated Management of Anti-Retroviral Therapy NIP Namibia Institute of Pathology NMBF Namibia Medical Benefit Fund OOP Out of Pocket OVC Orphans and Vulnerable Children PEPFAR The United States President’s Emergency Plan for AIDS Relief PPP Public Private Partnership PSEMAS Public Service Medical Aids Scheme REF Risk Equalization Fund SSC Social Security Commission TB Tuberculosis THE Total health expenditure UHCAN Universal Health Coverage Advisory Committee UNDP United Nations Development Program WHO World Health Organization

7 8 1 INTRODUCTION

ENGAGING THE PRIVATE SECTOR IN PUBLIC HEALTH CHALLENGES IN NAMIBIA

1.1 Background to the PharmAccess approach to improving healthcare

Engaging the private sector in healthcare is a central theme in the work of PharmAccess Foundation, a Dutch not-for-profit organization, based in Amsterdam, with offices in several African countries, including Namibia. A number of studies have shown how private sector can be engaged and leveraged for healthcare in low- and middle income countries e.g. (Rosenberg, Hartwig, & Merson, 2008) (Igumbor, et al., 2014) (Mc Pake & Hanson, 2016) (Whyle, 2016). This thesis will describe interventions developed and applied in Namibia to engage its private sector as an agent to develop and leverage private resources for public health. Interventions and subsequent evaluations are presented according to the PharmAccess model of transitioning the vicious circle of poor healthcare in Africa into a virtuous cycle of sustainable healthcare stimulating both demand and supply (Schellekens, et al., 2007). The PharmAccess model to change the healthcare environment from the vicious circle of mediocre quality and poorly accessible healthcare into a virtuous cycle of inclusive good quality healthcare recognizes the key stakeholders of healthcare systems as: the patient, the provider and the payer, all three of them functioning in a policy environment that sets the rules of interaction and exchange.

The vicious circle of poor demand and supply of healthcare, which is so common in many sub-Saharan countries (Kirigia & Barry, 2008), is, according to the PharmAccess theory of change (www.pharmaccess. org) caused by a combination of underutilizing the private healthcare sector, low quality of care due to lack of standards, insufficient investments especially in providers at the base of the pyramid and high out of pocket costs for patients due to low levels of pre-payment and risk pooling. This reflects on a healthcare system where trust between stakeholders is essentially lacking: patients/clients not trusting the quality and availability of healthcare services, healthcare providers not trusting patients/clients and payer/insurers to pay, payers/insurers not trusting providers to provide quality care, policy makers not being able to implement policy or regulations resulting in a lack of trust from providers, payers and patients (Schellekens, et al., 2007). The unavailability of prepayment and risk pooling mechanisms for patients results in high out of pocket expenses when confronted with catastrophic health events, with severe economic consequences (Preker A.,

9 Lindner, Chernichovsky, & Schellekens, 2103). This situation is aggravated on the supply side by low quality health care due to lack of investment, high risk and low efficiency of healthcare providers. All of the afore- mentioned can result in healthcare markets with unbalanced supply and demand sides, which is to the detriment of all stakeholders involved: patients/clients, providers, payers and policy makers Africa (Schelle- kens, et al., 2007).

To balance healthcare supply and demand, and establish trust in the healthcare system, the vicious circle would need to be reversed so that healthcare markets in sub-Saharan Africa can attract the resources they need to function well and deliver more inclusive quality healthcare. Trust in the system from all stakeholders is required and this is at the center of the PharmAccess interventions in Africa (Schellekens, et al., 2007). For example, this approach includes enhancing trust in quality of healthcare facilities by establishing and implementing internationally recognized standards for quality improvement, through SafeCare (www.safecare.org). Such standards help grading a healthcare facility with a ‘quality mark’ that allows for benchmarking and targeted and actionable quality improvement plans (Johnson, et al., 2016). Applying such institutional standards make the quality improvement process more transparent, creating more trust with patients as well as with payers and thus easing access to finance. Another example involves PharmAccess’ interventions in inclusive healthcare financing, all aimed at reducing out of pocket payments through stimulating prepayment, risk and income solidarity mechanisms and cross subsidization to further stimulate the demand side and avoid catastrophic health expenses amongst patients. Various interventions supported by the Health Insurance Fund (www.pharmaccess.org/activity/health-insurance) are exemplary for this approach. Trust between investors and providers is enhanced through interventions of the Medical Credit Fund (www.medi- calcreditfund.org) that facilitates the provision of loans to healthcare providers in Africa, under conditions of quality improvement as defined by SafeCare. Lately, PharmAccess emphasis is particularly on digitalization of healthcare exchanges (for patients, providers and payers), spearheaded by the establishment of platforms such as M-Tiba for mobile health wallets in Kenya (www.m-tiba.co.ke). M-Tiba further increases transparency and thus creates additional efficiencies in healthcare provision (e.g. avoidance of leakage), identifies low-hanging fruit with respect to healthcare quality interventions, such as over-prescription of medicines, etc. In this process, the role of the government and policy makers is fully recognized, especially to enact transparent and efficient policy and regulatory frameworks to stimulate both supply and demand of healthcare.

All in all, to create a virtuous cycle of healthcare PharmAccess engages a strategy which aims to promote inclusive health markets by stimulating and aligning demand and supply to reduced out of pocket payments, through pre-payments, risk sharing and risk reduction models (Preker, et al., 2103) that strengthen primarily, but not exclusively, the private sector. The ultimate goal is to make health markets work more efficiently for low and middle income groups in Africa by raising more money within the healthcare system. The preferred mode of operation is through the establishment of strategic partnerships (Lange, et al., 2008) to build capacity of local stakeholders who provide health and health financing services to low and middle income groups.

The principle that strengthening health systems requires a holistic approach that includes patients, providers, payers and policy makers in both the public and private sector is at the center of the PharmAccess approach. The idea is to perform interventions that stimulate improvements inside the health system simultaneously for all pertinent stakeholders, to avoid fragmentation. Private sector activities should always compliment the public sector specifically when filling gaps in the healthcare system and are not intended to replace them. To avoid this ‘crowding out’ (Kronick & Gilmer, 2002) of either sector (public or private), through the influx of

10 donor funding, public private partnerships (Sulzbach, et al., 2011) using innovative mixed funding models are encouraged by PharmAccess, leading to local ownership and aiming at increasing the total amount of money in the healthcare system (Van der Gaag & Stimac, 2008). Local ownership provides a foundation for domestic sustainability (De Savigny & Adam, 2009). In this way donor money can function as a catalyst for innovations and partnerships whilst leveraging additional public and private funds leading to the crowding in (Ejughemre & Oyibo, 2014) of investments into the healthcare system within the policy framework of a country. It is further recognized that the health system does operate in isolation, it is only a part of a bigger system where it interacts with political, socio-economic, cultural and other factors (WHO, 2010).

1.1.1 The healthcare landscape in Namibia, 2004 Upon commencement of the PharmAccess interventions in 2004, Namibia was classified by the World Bank as a lower middle income country (World Bank, 2004), with a GDP per capita of US$3,298 (World Bank, 2004), amongst a population of just under 2 million people (Namibian Central Bureau of Statistics, 2006) in a country of over 825,000 square kilometers (Namibia Statistics Agency, 2013). Namibia has been and remains burdened with one of the world’s most inequitable income distribution (UNDP, 2007) with a Gini index of 63.3 (World Bank, 2003). This inequality in income distribution was also reflected in the healthcare system of the country.

Prior to and since the independence of Namibia, from South Africa, in 1990, the Ministry of Health and Social Services (MoHSS) had been the predominant and largest healthcare provider in Namibia. In 2004 the MoHSS was providing free healthcare to all Namibians, who had access to public health facilities in the country. Since most public health facilities were concentrated in urban areas, access to healthcare in rural areas was specifically challenging.

In parallel to the public healthcare providers managed by the MoHSS, Namibia’s private healthcare sector was (and remains) very prominent, of good quality and well organized on a formal level. This private healthcare sector was comprised of both for-profit and not-for-profit entities. The private for-profit sector in Namibia consists of private health care providers, medical professional associations, health insurers, third party administrators and medical aid funds. Private not-for profit facilities in the period prior to the influx of large amounts of vertical donor funding for Human Immunodeficiency Virus (HIV), tuberculosis (TB) and malaria (~2004/5), were mostly limited to faith based organizations (FBO’s) managing hospitals and clinics in partnership with the government. With the financial support from international donor funds, non-governmental organizations (NGO’s) and community based organizations (CBO’s) that delivered prevention, care and treatment supporting vertical programs started emerging. Although the public sector in Namibia had over three times the number of healthcare facilities (hospitals and clinics) than the private sector, the private sector employed 72% of doctors and 89% of nurses in the country (O’Hanlon, et al., 2010), providing good quality services to less than 20% of the population.

The financing of healthcare in the country was also divided along the lines of public and private health expenditure with little cross subsidization. Total health expenditure (THE) per capita in 2004 was US$205 per year, with health expenditure being 12.4% of total government expenditure. The payers for healthcare in Namibia were government (47.9%), public insurance (20.6%), private insurance (18.3%), out of pocket (OOP) (3.7%), companies (1%) and donors/non-governmental (8.4%). Despite the low OOP expenditure, 22% of total health expenditure (private insurance and OOP) came from households (MoHSS, 2008).

11 Prepayment for healthcare in Namibia was mostly organized by medical aid funds – not for profit mutual funds – providing both inpatient and outpatient healthcare benefits sought from private health providers. These benefits were based on maximum annual benefit scales, which varied in accordance with the price of the medical aid plan purchased. Short term health insurance coverage for catastrophic health care events was offered by various private insurance companies in Namibia.

Approximately 16% of the population (NAMAF, 2004) was enrolled on some form of prepayment for healthcare either through private medical aid and/or health insurance in 2004. More than half of these enrollees belong to the Public Services Medical Aids Scheme (PSEMAS). This scheme is a medical aid fund for government employees and their dependents only, through which government employees contribute to a highly subsidized pre-paid scheme enabling them to access health services in the private sector. Due to the high cost of private health insurance, access to affordable private healthcare remained out of reach of the far majority of low and middle income employed Namibians (Hohmann & Skolnic, 2004). Instead, they made use of free public health facilities, managed by the MOHSS, resulting in huge demand and overburdening of the public health system (Feeley, de Beer, de Wit, & Van der Gaag, 2006). At the start of this millennium over 60% of formal sector employees were uninsured (NABCOA, 2004). A very small number of employers provided prevention and/or treatment services to their employees, and only in some cases to the dependents of employees, either directly or by contracting with private sector service providers (NABCOA, 2004).

Formal dialogue and cooperation between the public and private for-profit healthcare sector, especially amongst policy makers, was extremely limited in 2004. Some of this sentiment stemmed from the historic inequalities with the private health care sector having served the higher income populations who could afford to pay for these services, while the majority of the previously disadvantaged populations and those of low and middle income were excluded. A mistrust of the public sector of the profit motive of the private sector was mirrored by a lack of trust of the private sector in the inability of the public sector to serve the needs of the Namibian population. Although the MoHSS had since Independence provided the legislative and regulatory framework for the healthcare sector, the private sector remained largely self-regulated in terms of quality and tariff management. Medical aid tariffs were (and are) negotiated between associations of providers and funders (NAMAF, 2004).

At the time when PharmAccess commenced programs in Namibia in 2004, the HIV prevalence rate had peaked with 22% of pregnant women testing HIV positive (MoHSS, 2003). The Namibian government had commenced piloting an anti-retroviral (ART) treatment program at a few sites but the need for antiretroviral treatment far exceeded the government’s health facilities ability to provide this service much needed treatment. Private health insurance in 2004 excluded treatment for HIV, with the exception of the most expensive health insurance plans (Feeley, et al., 2006). This was primarily because of the unknown and potentially unbearable risk associated with HIV, as perceived by Namibian medical aid funds. Both the public and private sector at the time continued to face increasing numbers of people requiring ART, with both sectors incurring direct and indirect costs related to absenteeism, illness and loss of morale.

By 2004, partnerships were being facilitated between the public sector, NGOs and CBOs specifically supported by international vertical donor funding as provided by GFATM and PEPFAR, resulting in the private not-for-profit sector providing a key role in the national HIV response (O’Hanlon, et al., 2010). Despite some attempts by the for-profit private healthcare sector over time to propose different public private partnerships, especially for infrastructure development and outsourcing of public healthcare services to private facilities,

12 there was limited appetite for this in the absence of government policy and regulations on public-private partnerships (Feeley, et al., 2006).

1.1.2 The PharmAccess interventions in Namibia 2004-2016 In 2004, PharmAccess embarked on the Okambilimbili program, funded by the Dutch Postcode Loterij, in Namibia to strategically address the problem of HIV/AIDS and the lack of access to HIV/AIDS treatment services at the time. The PharmAccess goal was to establish a prepayment mechanisms including HIV/AIDS risk coverage for lower and middle income people through existing medical aid funds. This mechanism would be combined with a simultaneous effort to determine and improve the quality of care delivered through an independent case management entity. These PharmAccess interventions were based on the principles of establishing sustainable solutions through structurally sound financing mechanisms; while harnessing the potential and under-utilized capacity of the private sector by integrating HIV care into the general healthcare sector and setting up partnerships between donors, the private sector and the public sector (Gustafsson- Wright, et al., 2010). The Okambilibili program was implemented for a period of five years. To evaluate the program a three-year panel household study was conducted in the City of Windhoek (CoW), the Capital of Namibia. These studies included anonymous oral fluid HIV screening of participating Windhoek City household members, providing the first household HIV prevalence and incidence data for the CoW for policy makers and response planning.

Building on the foundations of the Okambilimbili program a number of further gaps in the Namibian healthcare system were identified, which in 2007 led to the implementation of the HIV prevalence surveillance services provided in an informal public private partnership by PharmAccess Foundation, with the Namibia Business Coalition on AIDS (NABCOA), the Namibia Institute of Pathology (NIP) and MoHSS. This was a first step towards public private partnership (PPP) between the MoHSS and the not-for-profit private sector through PharmAccess and NABCOA, the membership of the latter being the for-profit-private sector. Dialogue between the public sector and the not-for-profit private health sector was encouraged through the Okambil- imbili program, paving the way for the expansion of the partnership. The provision of evidence of the prevalence of HIV in the private sector and a better understanding of the role that the private sector could play in providing HIV workplace programs - including access to treatment - opened the door for dialogue with the then prevalent policy makers.

In 2008, based on a demand for on-site testing in workplaces, the Bophelo! program was initiated to provide mobile wellness screening services, including HIV testing to workplaces. This initiative to de-stigmatize HIV testing by including screening for non-communicable diseases, shifted the focus beyond HIV and engaged the private sector in responding to the management and financing thereof. The mobile wellness screening at workplaces highlighted a significant gap in access to healthcare services, especially to rural, remote and underserved urban communities, which neither the public nor private sector were in a position to fill.

In an attempt to reduce the gap in healthcare service delivery, a pilot mobile primary health care clinic program was piloted by PharmAccess in partnership with the MoHSS in 2010. Upon the successful implementation of the pilot program, the Mister Sister mobile primary health care clinics were established in 2011. These mobile clinics were operated and managed by PharmAccess Foundation to provide pre-paid primary health care services to rural and remote communities, financed by employer contributions for employees, donor contributions for vulnerable groups and the provision of free medicines to the Mister Sister program by the MoHSS.

13 In 2011 the Bophelo! mobile wellness screening clinics and program was incorporated into the portfolio of NABCOA to provide these services on an ongoing basis to its membership. In 2014, NABCOA expanded its mandate beyond HIV/AIDS and became known as Healthworks Business Coalition (HBC), focusing on the integration of HIV in workplaces into general healthcare. In 2016, the Mister Sister mobile clinics were tran- sitioned to the HBC portfolio and continue to operate under the umbrella of the business coalition through the public private partnership agreement with the MoHSS.

The studies presented in this thesis are a direct result of operational research conducted on the Okambilim- bili, Bophelo! and Mister Sister programs implemented by PharmAccess Foundation and its partners to engage the private health sector in responding to public health challenges in Namibia, by stimulating demand through patient awareness, encouraging innovations in supply, experimenting with innovative payer subsidies and providing evidence to influence policy makers.

1.2 Engagement of the private sector

This thesis contains studies evaluating the PharmAccess strategies and activities of engaging the private sector in the public healthcare challenges in Namibia. These strategies involved interventions geared towards all key stakeholders of the healthcare system: the patient, the provider, the payer and the policy maker: Strategy 1: To stimulate demand by improving health awareness of the patient/client; Strategy 2: To stimulate supply by developing innovative new healthcare service provision; Strategy 3: To innovate new payer mechanisms using (temporary) subsidization; Strategy 4: To collect and present evidence to support policy-making.

The above strategies employed in Namibia by PharmAccess were supported by the PharmAccess paradigm that, in principle, healthcare provision is an economic exchange with demand and supply components. Complexity is generated by the fact that in a more mature healthcare system the decider (patient), provider (clinic) and payer (insurance) are separate entities, creating market distortions. The policy maker provides the context within which the healthcare exchange between the patient, the provider and the payer can take place. Therefore, the PharmAccess paradigm recommends interventions at all levels of the healthcare system, as indicated above. All in all the goal is to provide the right diagnosis, at the right time, for the right patient at the right cost within the right legal and regulatory framework, with both public and private sector contributing a complementary role. The above stakeholder groups are addressed in this thesis in no specific order of importance and it is recognized that no one can function in isolation, as (and this will be demonstrate in this thesis) interventions aimed at stimulating one stakeholder have an effect either directly or indirectly on the others.

1.3 Strategies to stimulate demand by improving health awareness of patient/client

Everything that is done in healthcare should begin and end with the patient in mind. In order to do so it is important to understand the needs of the patient, encourage the patient to understand their own needs and

14 in so doing stimulate development of healthcare services that are relevant to the demand, so there is a basis for (pre-) payment. With its goal to simulate healthcare prepayment mechanisms for (lower) middle-income people, PharmAccess naturally addressed workers in the formal sector in Namibia and started engaging with private sector employers. An entry point was to make employers and employees aware of their HIV status and use this to stimulate demand for health insurance/pre-payment. In order to reach out to as many people as possible, it was considered beneficial to use a non-invasive HIV test that would be acceptable, also for those who generally avoid blood-based laboratory tests (described in Chapter 2).

With the availability of a validated non-invasive rapid HIV screening device, PharmAccess Foundation proceeded under project Okambilimbili, in collaboration with the NABCOA and the NIP, to market anonymous HIV prevalence surveillance services (using the OraQuick® testing devices) to private sector companies. In Chapter 3 the findings of anonymous HIV workplace surveys, using the OraQuick® rapid HIV test are presented. It was assumed that providing the management of companies with HIV prevalence estimates of their workforce would create awareness amongst management and motivate these decision makers to enrol their employees on health insurance.

As the evidence around the prevalence of HIV became more prominent through these surveys, PharmAc- cess’ operational research, during the same period, aimed at exploring different market segments for the health insurance products which were being developed through the Okambilimbili program to leverage more private sector funding for the HIV response.

At the time of the implementation of the Health is Vital Risk Equalization Fund (HIVREF) presented in Chapter 8 and the engagement of the private sector in enrolling their employees for health insurance, a key question was raised by NABCOA regarding the prevalence of HIV infection in the future workforce, namely tertiary education students in Namibia and the potential need for a student HIV insurance linked to the risk equalization fund. To explore whether tertiary education students were a potential target market for HIV insurance through the HIVREF, HIV prevalence as well as HIV/AIDS knowledge and attitudes and general access to healthcare among students at the Polytechnic of Namibia and the University of Namibia was assessed (Chapter 4).

In 2007, PharmAccess Foundation was approached by the Agricultural Employers Association, agriculture being one of the core industries of the Namibian economy, to assist in the development of an HIV policy for the commercial farming sector in Namibia. Since access to care and treatment is a key component of a comprehensive workplace HIV policy options of providing access to HIV testing and access to treatment services needed to be investigated. One of the key challenges anticipated in the investigation of healthcare options for this sector was the vast geographical spread of the target population. In order to find a solution to enable rural commercial employers to implement a policy which facilitates access to care and treatment PharmAccess investigated healthcare options for rural employers and employees, whilst assessing this sector as a potential market for health insurance (Chapter 5).

15 1.4 Strategies to stimulate supply by developing innovative new healthcare service provision

As indicated by the demand studies, there was a need for better access to healthcare in remote areas in Namibia. However, given the vastness of the country, the small population size, the establishment of healthcare facilities in every remote location was not considered cost-effective. Therefore, PharmAccess Namibia responded to the gaps in the supply side of the healthcare system by developing two innovative mobile service provision systems: the Bophelo! and Mister Sister mobile clinics (Chapters 6 and 7), resulting in the first public private partnership agreements for mobile services in Namibia.

In Chapter 6, an overview is presented of the program and the effectiveness of mobile services is evaluated compared to fixed site voluntary counselling and testing for HIV (VCT) services both in terms of accessing high risk populations and cost effectiveness. This data, collected in 2009, provided a primary source of evidence in Namibia of health conditions within the general workforce across different industries to create greater awareness amongst management and policy makers of healthcare conditions in the working population and encourage enrolment in health insurance. All in all, the mobile testing services appeared to be able to access geographically hard-to-reach populations with high numbers of participants testing for HIV for the first time.

In response to the needs of the rural employers and their workforce (presented in Chapter 5), and the gap in referral services identified in the Bophelo! program, PharmAccess in 2010 commenced with piloting of the Mister Sister mobile clinics. Through this initiative, primary health care services were offered to rural and remote populations in Namibia in partnership with the MoHSS, who provided licensing, medication, consumables at no cost. In return PharmAccess provided the services to rural and remote populations, charging employers a prepaid subscription fee per employee per month. To sustain this program, PharmAccess leveraged additional private sector resources through corporate social investment funding and international donor funding to provide primary healthcare services to vulnerable groups en-route to the rural workplaces for whom services were contracted (Chapter 7).

1.5 Strategy to innovate new payer mechanisms using temporary subsidization

While developing insights in how to reach out to lower middle-income people to stimulate prepayment of healthcare, it transpired to PharmAccess that Namibian medical aids were not willing to develop pertinent insurance products including HIV coverage. This was mostly related to the perception of unknown and potentially unbearable costs that would result from such packages. Moreover, medical aids were hesitant because such products could potentially cannibalize their existing medical aid schemes. Therefore, PharmAccess stimulated the idea of a separate risk equalisation fund for HIV, removing the unknown HIV risk from the medical aid packages and allowing them to compete on a level playing field for lower-income health insurances. This HIVREF was accompanied by an independent quality control mechanism: a HIV patient case management system (Chapter 8).

16 1.6 Strategies to collect and present evidence to support policy-making

To transform any healthcare system evidence needs to be collected to motivate policy changes that affect healthcare exchanges. This was particularly true for the work conducted by PharmAccess in Namibia around mobilizing resources in the response to HIV/AIDS. Policy makers for the purpose of this thesis are not only defined as politicians, academics or groups of people who make laws and regulations to govern the healthcare system, but also management and decision makers of entities like business coalitions, universities, professional associations and private sector companies, who make decisions about whether or not to provide and pay for health care for their employees and dependents. Chapters 9a and Chapter 9b describe results from household surveys that were meant to evaluate the impact of affordable healthcare insurances (2006-2009) in the City of Windhoek. Using the OraSure® oral fluid testing devices for HIV surveillance, these surveys generated the first general population HIV prevalence data in Namibia. Information was shared with policy makers, which proved valuable with respect to resource allocation of antiretroviral treatment programs.

It was learned over the years that addressing HIV in a ‘vertical’ manner was not the optimal approach. Inclu- sion of other diseases, and in particular chronic non-communicable diseases appeared a way to de-stigmatize HIV and simultaneously address very important health problems in a more efficient manner. InChapters 10 and Chapter 11 it is illustrated how the Bophelo! mobile screening services were utilized to improve health awareness of HIV and non-communicable diseases (NCD), like diabetes and hypertension in the private sector amongst both employers and employees.

From continuous dialogue with the private sector through the Bophelo! program, it was evident that a key motivation for employers to provide wellness programs and health insurance for their employees was to have a healthier workforce with less absenteeism. PharmAccess wished to further substantiate this by specifically addressing work absenteeism as a consequence of diabetes, hypertension, HIV and other health determinants. The findings are presented inChapter 11.

1.7 Hypotheses

This thesis is addressing the hypothesis that private healthcare sector contributions can accelerate and complement ongoing public healthcare sector responses to HIV/AIDS and other health challenges in Namibia, by focussed interventions aimed at patients, providers, payers and policy makers. The crux of combining private and public healthcare responses in resource-constrained settings is to avoid crowding out: the contributions of one sector mitigating or replacing those of the other. All in all, the goal is to increase the total amount of money in the healthcare system. This thesis evaluates various interventions and their potential to mobilize (parts of) the Namibian private sector to develop services complementary to simultaneously developing public sector healthcare provision. Mixed methods are used, including qualitative behavioural studies and quantitative biomedical surveys. Topics covered in this thesis are pertaining to a wide array of research questions, linked to the PharmAccess strategies, such as:

17 1.7.1 Strategy 1 - to stimulate demand by improving health awareness of the patient/client 1.7.1.1 Does an understanding of the HIV prevalence, collected through various surveys support private sector organizations to enrol their internal stakeholders in pre-paid low-income health insurances including HIV in Namibia? 1.7.1.2 Are students of academic institutions in Namibia a target market for private low-income health insurance including HIV? 1.7.1.3 Is it feasible to strengthen health service delivery to the commercial farming sector in Namibia by introducing private affordable health insurance that includes HIV coverage? 1.7.2 Strategy 2 - to stimulate supply by developing innovative new healthcare service providers 1.7.2.1 Is mobile HIV testing in Namibia through private initiatives an affordable complement to serve populations that are beyond public sector reach and enrol those into care? 1.7.3 Strategy 3 - to innovate new payer mechanisms using temporary subsidization 1.7.3.1 Can low income health insurance/medical aid funding be developed sustainably in Namibia with short-term donor subsidy, to avoid crowding out of the private sector and to leverage private sector resources for HIV care and treatment? 1.7.4 Strategy 4 - to collect and present evidence to support policy-making 1.7.4.1 What is the prevalence, incidence and socio economic distribution of HIV in Namibia’s capital city, Windhoek, and how can this be used for policy-making? 1.7.4.2 What is the prevalence of non-communicable diseases, and absenteeism, at private workplaces in Namibia, and how could those be addressed through private initiatives?

The papers presented in this thesis describe the analysis of various innovative programs managed by Phar- mAccess in Namibia to leverage private sector resources over a time period of more than a decade (2004 -2016). The chapters are organized along the lines of the above strategies. A discussion of the findings and recommendations for the future for further engagement of the private sector in addressing public health challenges in Namibia and strengthening public private partnerships is presented in Chapter 12.

••• References

De Savigny, D., & Adam, T. (2009). Systems thinking for health systems strengthening. World Health Organization. Ejughemre, U., & Oyibo, P. (2014). Healthcare financing in Nigeria: A systematic review assessing the evidence of the impact of health insurance on primary health care delivery. Journal of Hospital Administration., Issue 1. Vol 4. Feeley, F., de Beer, I., de Wit, T., & Van der Gaag, J. (2006). The health insurance industry in Namibia: Baseline Report. Center for International Health. Gustafsson-Wright, E., de Beer, I., Gaeb, E., van Rooy, G., van der Gaag, J., & Rinke de Wit, T. (2010). The Okambilimbili Health Insurance Project in Namibia: lessons learnt. The Netherlands: PharmAccess Foundation, Amsterdam Institute for Global Health and Development. Hohmann, J., & Skolnic, R. (2004). Options and scenarios for HIV/AIDS risk cover for low-income employees within NABCOA member companies in Namibia. Windhoek: NABCOA. Igumbor, J., Pacoe, S., Rajap, S., Townsend, W., Sargent, J., & Darkoh, E. (2014, October). A South African public-private partnership HIV treatment model: Viability and success factors. PLoS One, 9(10). Johnson, M., Schellekens, O., Stewart, J., van Ostenberg, P., de Wit, T., & Spieker, N. (2016). An Innovative Approach for

18 Improving Quality Through Strandards, Benchmarking and Improvement in Low-and Middle-Income Countries. The Joint Commission Journal on Quality and Patient Safety, 42(8), pp.350-AP11. Kirigia, M., & Barry, S. (2008). Health challenges in Africa and the way forward. International Archives of Medicine, 1:27. Kronick, R., & Gilmer, T. (2002). Insuring low-income adults: does public coverage crowd out private? Health Affairs, 21(1):225-239. Lange, J., Schellekens, O., Lindner, M., & van der Gaag, J. (2008). Public–private partnerships and new models of healthcare access. Current Opinion in HIV and AIDS., 3(4), pp.509-513. Mc Pake, B., & Hanson, K. (2016). Managing the public-private mix to achieve universal health coverage. The Lancet, 388(10044), 622-630. MoHSS. (2008). Namibia National Health Accounts 2001/2-2006/7. Republic of Namibia Ministry of Health and Social Services. MoHSS, R. o. (2003). Report of the 2002 National HIV Sentinel Survey. Republic of Namibia Ministry of Health and Social Services. NABCOA. (2004). Healthcare survey amongst members of the NABCOA. Namibia Business Coalition on AIDS. NAMAF. (2004). NAMAF Annual Conference . Namibia Statistics Agency. (2013). Profile of Namibia. http://cms.my.na/assets/documents/p19dpmrmdp1bqf19s2u8pis- c1l4b1.pdf. Namibian Central Bureau of Statistics. (2006). 2001 Population Projections. http://cms.my.na/assets/documents/ p19dmrhrpm1bpl1jre1p3quqm1tjc1.pdf. NSA. (2011). Namibia Population and Housing Census. Namibia Statistics Agency, Government of the Republic of Namibia. O’Hanlon, B., Feeley, F., de Beer, I., Sulzbach, S., & Vincent, H. (2010). Namibia Private Sector Assessment. Strengthe- ning Health Outcomes through the Private Sector, Abt Associates Inc. Preker, A., Lindner, M., Chernichovsky, D., & Schellekens, O. (2103). Scaling Up Affordable Health Insurance : Staying the Course. RoN. (n.d.). www.gov.na/about-namibia. Government of the Republic of Namibia. Rosenberg, A., Hartwig, K., & Merson, M. (2008). Government-NGO collaboration and sustainability of orphans and vulnerable children projects in southern Africa. Journal of Evaluation and Program Planning, 51-60. SafeCare. (2016). www.safecare.org. Schaeffer C. (2014). Medical Aids Overview Namibia. http://www.namaf.org.na/index.php?page=resources. Schellekens, O., Lindner, M., Lange, J., & van der Gaag, J. (2007, October 17). A new paradigm for increased access to healthcare in Africa. Annual IFC / Financial Times Essay Competition. SHOPS Project. (2012). Namibia Private Health Sector Assessment. Brief. Bethesda: Strengthening Health Outcomes through the Private Health Sector Project, Abt Associates. Sulzbach, S., De, S., & Wang, W. (2011). The private sector role in HIV/AIDS in the context of an expanded global response: expenditure trends in five sub-Saharan African countries.Health Policy and Planning., Vol. 1: 26 pages i72-i84. UNAIDS. (2004). 2004 Epidemiological Fact Sheet. http://data.unaids.org/Publications/Fact-Sheets01/namibia_en.pdf. UNDP. (2007). Human Development Indicators. In: Human development report 2007/8. UNDP http://hdr.undp.org/en/ media/HDR_20072008_EN_Indicator_tables.pdf. UNDP. (2007). Human Development Report 2007/8, fighting climate change: human solidarity in a divided world. http:// hdr.undp.org/en/media/HDR_20072008_EN_Indicator_tables.pdf. Van der Gaag, J., & Stimac, V. (2008). Towards a new paradigm for health sector development. Amsterdam Institute for International Development. WHO. (2010). Improving health systems efficiency as a means of moving towards universal health coverage. Background

19 paper, 28. Geneva: WHO. Whyle, E. (2016). Models of public private engagement for health service delivery and financing in Southern Africa: a systematic review. Health Policy and Planning, pii: czw 075. World Bank. (2003). http://data.worldbank.org/indicator/SI.POV.GINI. World Bank. (2004). https:// http://data.worldbank.org/indicator/NY.GDP.PCAP.CD?locations=NA. World Bank. (2004). How does the World Bank classify countries? https://datahelpdesk.worldbank.org/knowledgebase/ articles/378834-how-does-the-world-bank-classify-countries. World Bank. (2009). data.worldbank.org/country/namibia. World Bank. (2011). http://data.worldbank.org/indicator/SI.POV.GINI

20 2 DIAGNOSTIC ACCURACY OF 2 ORAL FLUID-BASED TESTS FOR HIV SURVEILLANCE IN NAMIBIA

21 Letters to the Editor J Acquir Immune Deﬁc Syndr Volume 48, Number 1, May 1, 2008

6. Lebrecht D, Vargas-Infante YA, Setzer B, et al. Africa, where the HIV-1 epidemic im- by venipuncture, which is routinely per- Uridine supplementation antagonizes zalcita- poses an ever-increasing burden.2 In formed in pregnant women for manda- bine-induced microvesicular steatohepatitis in mice. Hepatology. 2007;45:72–79. developing countries with a high HIV tory syphilis (rapid plasma reagin [RPR]) 7. Venhoff N, Zilly M, Lebrecht D, et al. Uridine prevalence, most HIV testing for sur- screening. pharmacokinetics of Mitocnol, a sugar cane veillance purposes is conducted as part The reference standard used was extract. AIDS. 2005;19:739–740. of seroprevalence surveys among spe- the Namibian national algorithm for HIV 8. Weinberg ME, Roman MC, Jacob P, et al. cific population groups, such as women testing, which consists of serial dual Single-dose and cumulative pharmacokinetics of the food supplement NucleomaxXÒ and attending antenatal clinics (ANCs) and enzyme-linked immunoassay (ELISA) mechanism for enhanced bioavailability of persons attending voluntary counseling testing on blood plasma to detect anti- uridine. Antivir Ther. 2007;12(Suppl 2):L24. and testing (VCT) sites. In addition to bodies to HIV-1 and HIV-2. The baseline 9. Ashour OM, Naguib FN, el Kouni MH.5- invasive HIV diagnostic tests using test was the AxSYM HIV 1/2 gO assay (m-Benzyloxybenzyl)barbituric acid acyclonu- cleoside, a uridine phosphorylase inhibitor, and whole blood, plasma, or serum, non- (Abbott Laboratories, Abbott Park, IL). 2#,3#,5#-tri-O-acetyluridine, a prodrug of uri- invasive tests are available that are based A sample rate by a cutoff rate (S/CO) dine, as modulators of plasma uridine concen- on detecting antibodies to HIV present in value of >10.00 was considered clearly tration. Implications for chemotherapy. oral mucosal transudate, or oral fluid. reactive. Specimens with an S/CO value Biochem Pharmacol. 1996;51:1601–1611. These are more user-friendly and may be of >0.90 to 10.00 were considered in- 10. Koch EC, Schneider J, Weis R, et al. Uridine excess does not interfere with the antiretroviral more acceptable, especially to hard-to- determinate and were retested in dupli- efficacy of nucleoside analogue reverse transcrip- reach populations, such as commercial cate after recentrifugation. Specimens tase inhibitors. Antivir Ther. 2003;8:485–487. sex workers.3,4 Therefore, oral fluid– found to be reactive or repeatedly in- 11. Moore KH, Shaw S, Laurent AL, et al. based tests have good potential for use in determinate were confirmed with the Lamivudine/zidovudine as a combined formu- lation tablet: bioequivalence compared with screening and surveillance activities in Access HIV 1/2 New test (Bio-Rad Lab- lamivudine and zidovudine administered con- high-prevalence developing countries. oratories, Marnes-la-Coquette, France). currently and the effect of food on absorption. Reports from Europe and North The OraSure oral fluid specimens J Clin Pharmacol. 1999;39:593–605. America have demonstrated good diag- were tested with the Oral Fluid Vironos- 12. Combivir (lamivudine/zidovudine) prescribing nostic accuracy for 2 oral fluid–based tika HIV Uni-Form II microELISA information. GlaxoSmithKline. Available at: http://us.gsk.com/products/assets/us_combivir. tests that have been approved by the US system in accordance with the manu- pdf. Accessed December 2007. Food and Drug Administration (FDA): facturer’s instructions. Specimens that 13. McComsey GA, O’Riordan M, Setzer B, et al. OraQuick Rapid HIV-1/2 Antibody Test yielded results discrepant with the refer- Uridine supplementation in HIV lipoatrophy: (OraSure Technologies, Inc., Bethlehem, ence standard were checked for technical pilot trial on safety and effect on mitochondrial 5–7 indices. Eur J Clin Nutr. 2007 May 30 [Epub PA [‘‘OraQuick’’]) and Oral Fluid errors and were retested. All ELISA ahead of print]. Vironostika HIV Uni-Form II micro- testing for this study was performed at ELISA (bioMe´rieux Inc., Durham, NC), the Namibia Institute of Pathology in which is used in combination with an Windhoek, which is the national medical Diagnostic Accuracy of appropriate oral specimen collection de- reference laboratory in Namibia. vice.3,8,9 Test performance data from field OraQuick test results were read 2 Oral Fluid–Based settings in developing countries, partic- after 20 to 40 minutes and classified in Tests for HIV ularly sub-Saharan Africa, where HIV-1 accordance with the manufacturer’s infections are caused by a high variety of instructions. OraQuick external quality Surveillance in (‘‘non-B’’) subtypes,10 are limited, how- control procedures (control kits) were ever.4,11,12 This comparative study aimed performed in accordance with the man- Namibia to evaluate the diagnostic accuracy of the ufacturer’s recommendations. 2 oral fluid–based HIV tests for surveil- Only available specimen sets, con- To the Editor: lance purposes in a high-prevalence sisting of at least 1 oral fluid specimen Surveillance data on HIV preva- population in Namibia. (OraQuick and/or OraSure) and the lence among various at-risk populations A cross-sectional study was con- paired plasma specimen, were included are essential for the development and ducted in pregnant women 18 years of in the analysis. The reference standard evaluation of HIV control initiatives.1 age or older with unknown HIV status results were compared with the paired This is particularly true in sub-Saharan who attended the 2 ANCs of Katutura test results of the OraQuick and/or State Hospital, Windhoek, Namibia, and OraSure devices. The main outcomes for Oshakati State Hospital, Oshakati, Na- diagnostic accuracy were sensitivity, This study was supported by a research grant from mibia, for VCT and/or antenatal care in specificity, and predictive values, with The Netherlands Ministry of Foreign Affairs June and July 2006. Participants pro- exact 95% confidence intervals (CIs). awarded to the Amsterdam Institute for In- vided 2 oral fluid specimens using the Concordance between the OraQuick and ternational Development (DSI/SB-531/06). OraQuick and OraSure (OraSure Tech- OraSure devices was evaluated using the The authors acknowledge the support of Orasure Technologies, Inc. for the donation nologies, Inc.) oral devices in accordance k statistic. of kits. Orasure Technologies, Inc. did not with the manufacturer’s instructions, The study was approved by the provide funding for this study, nor did it have which were labeled and processed anon- Research Ethics Committee at the Na- any role in the design and conduct of the study ymously. For reference standard testing, mibian Ministry of Health and Social or in the preparation of the manuscript. paired blood specimens were obtained Services. Written informed consent was

116 q 2008 Lippincott Williams & Wilkins

22 J Acquir Immune Deﬁc Syndr Volume 48, Number 1, May 1, 2008 Letters to the Editor

obtained from all participants before OraQuick test performance data our knowledge, this study is the first to enrollment. Participants did not receive from resource-limited countries are lim- report on ELISA test performance with study test results (anonymous, unlinked), ited; however, overall, they show high oral fluid in a population predominantly but they accessed routine VCT as pro- specificity (100%) and varying sensitiv- infected with subtype C. In view of the vided at the 2 study sites. ity (75% to 100%).4,11,12 Notably, a recent dominance of subtype C in high-preva- A specimen set was available for field study on the OraQuick test device lence areas, such as southern and East 273 participants (175 at Katutura and 98 in a high-prevalence population in rural Africa and India, the validated oral fluid at Oshakati). Testing of the blood speci- India showed 100% (95% CI: 98 to 100) tests can be of great value to enhance mens by reference standard identified 70 sensitivity and 100% (95% CI: 99 to surveillance efforts where these are HIV-1 antibody-positive and 203 HIV-1 100) specificity,4 which is consistent with needed most. antibody-negative specimens, yielding our findings. The HIV infection rates we found a prevalence of 25.6% (95% CI: 20.4 The 3 discrepant OraSure-Vironos- in this study among participants from to 30.8). One participant had an in- tika test results (ie, 2 false-negative Katutura (25.1%) and Oshakati (26.5%) determinate test result at reference stan- results [Katutura and Oshakati] and 1 are consistent with results of the 2006 dard baseline testing, which turned out to false-positive result [Katutura]) were national sentinel serosurvey among preg- be negative at confirmatory testing. No checked, and the tests were repeated; nant women, which reported estimated OraQuick and OraSure test results were however, no additional evaluations were HIV-1 prevalences of 21.1% and 27.1%, reported as invalid or indeterminate. performed to resolve the true serostatus respectively, for these sites.17 Although OraQuick results were available for all of the specimens. This is a limitation of this study was not designed to measure 273 women, showing an accuracy of the study. A possible explanation for the prevalence and 95% CIs are wide be- 100% (Table 1). OraSure results were 2 false-negative results is the collection cause of the limited sample size, this excluded from analysis for 6 (2.2%) of insufficient oral specimen volume by finding adds to the assumption that the women because of labeling errors. The the collection device; this limitation has study population was representative of OraSure test recorded 3 (1.1%) discrep- been reported previously and warrants the general ANC population. ant results among 267 tests, namely, 2 the development of safeguards against Whereas the OraQuick (rapid) test false-negative results and 1 false-positive inadequate sampling.13 Alternatively, the would be most suitable for areas with result, yielding a sensitivity of 97.1% false-negative results may be explained limited laboratory resources, the OraSure- (95% CI: 89.9 to 99.6) and a specificity by the presence of low HIV antibody Vironostika tests would be preferred of 99.5% (95% CI: 97.2 to 100) (see levels during the early phase of infection. when complete confidentiality is re- Table 1). Concordance between Ora- Furthermore, it cannot be ruled out quired, when real-time diagnosis is not Quick and OraSure test results was high completely that the reference standard necessary, and in high-volume batch (k = 0.97; 95% CI: 0.95 to 0.99). used produced false-positive test results testing, rendering the OraSure-Vironos- This study, which is the first formal because it did not include a true confir- tika test particularly suitable for anony- field evaluation of the OraQuick and matory test, such as a Western blot or p24 mous surveillance testing. Although the OraSure devices in a resource-limited antigen test. Finally, it can be speculated manufacturer already offers differential setting in southern Africa, demonstrates that the 1 false-positive result may be pricing for the OraQuick and OraSure that the OraQuick test is 100% accurate associated with cross-reactivity of non- test kits in sub-Saharan Africa, oral and that the OraSure test has high specific antibodies in the patient speci- fluid–based tests are still more expensive specificity (99.5%) but slightly lower men as a result of chronic immune compared with the conventional HIV sensitivity (97.1%) in a high-prevalence system activation in Africans attributable tests. If their use expands over time, population in Namibia. The high nega- to an increased load of environmental further reduction of prices might be tive predictive value of both tests pathogens.14,15 negotiable. Inclusion in the World Health ($99%) renders them particularly suit- Immunoassay test performance on Organization’s HIV Test Kit Bulk Pro- able for surveillance purposes. Our serum or plasma may be affected by HIV curement Scheme might be helpful to findings contribute to the limited existing viral variation, including subtypes.16 In achieve this.18 data on diagnostic accuracy of oral fluid– Namibia, although no data have been In conclusion, the OraQuick and based HIV tests among populations in published to date, it is assumed that, like OraSure oral fluid–based HIV tests resource-limited settings that are infected elsewhere in southern Africa, subtype C showed high diagnostic accuracy in a with non-B subtypes.4,11,12 is the predominant HIV-1 subtype.10 To high-prevalence population in Namibia,

TABLE 1. Diagnostic Accuracy of the OraQuick HIV-1/2 Rapid Antibody test (OraQuick) and the Oral Fluid Vironostika HIV Uni-Form II microELISA System, in Combination With the OraSure Oral Specimen Collection Device (OraSure), on Oral Fluid Specimens Compared With the Namibian National Algorithm for HIV Testing (Reference Standard) in Namibia Positive Negative No. True- True- False- False- Sensitivity Speciﬁcity Predictive Predictive Test Specimens Positive Negative Negative Positive (95% CI) (95% CI) Value Value OraQuick 273 70 203 0 0 100% (94.9 to 100) 100% (98.2 to 100) 100% 100% OraSure 267 67 197 2 1 97.1% (89.9 to 99.6) 99.5% (97.2 to 100) 98.5% 99.0%

q 2008 Lippincott Williams & Wilkins 117

23 Letters to the Editor J Acquir Immune Defic Syndr Volume 48, Number 1, May 1, 2008 supporting their public health utility in REFERENCES 10. Hemelaar J, Gouws E, Ghys PD, et al. Global surveillance initiatives in resource-lim- 1. Mishra V, Vaessen M, Boerma JT, et al. HIV and regional distribution of HIV-1 genetic testing in national population-based surveys: subtypes and recombinants in 2004. AIDS. ited settings in sub-Saharan Africa. The 2006;20:W13–W23. advantages of convenience and user- experience from the Demographic and Health Surveys. Bull World Health Organ. 2006;84: 11. Bhore AV, Sastry J, Patke D, et al. Sensitivity friendliness may improve patient partic- 537–545. and specificity of rapid HIV testing of pregnant ipation, thus facilitating the collection of 2. WHO-UNAIDS. AIDS Epidemic Update. De- women in India. Int J STD AIDS. 2003;14: reliable prevalence data among various cember 2007. Geneva, Switzerland: WHO; 37–41. 12. Reynolds SJ, Ndongala LM, Luo CC, et al. at-risk populations. Additional appli- 2007. 3. Gallo D, George JR, Fitchen JH, et al. Evalua- Evaluation of a rapid test for the detection of cations, such as in VCT, should be in- tion of a system using oral mucosal transudate antibodies to human immunodeficiency virus vestigated. Ongoing evaluation of oral for HIV-1 antibody screening and confirmatory type 1 and 2 in the setting of multiple trans- fluid testing against the national refer- testing. OraSure HIV Clinical Trials Group. mitted viral subtypes. Int J STD AIDS. 2002; ence standard as part of a formal quality JAMA. 1997;277:254–258. 13:171–173. 13. Mortimer P, Parry J. Non-invasive virological assurance program is warranted. The 4. Pant PN, Joshi R, Dogra S, et al. Evaluation of diagnostic accuracy, feasibility and client diagnosis: are saliva and urine specimens development of safeguards against in- preference for rapid oral fluid-based diagnosis adequate substitutes for blood? Rev Med Virol. adequate sampling volume is indicated. of HIV infection in rural India. PLoS ONE. 1991;1:73–78. 2007;2:e367. 14. Biggar RJ, Gigase PL, Melbye M, et al. 5. Delaney KP, Branson BM, Uniyal A, et al. ELISA HTLV retrovirus antibody reactivity Raph L. Hamers, MD* Performance of an oral fluid rapid HIV-1/2 test: associated with malaria and immune com- experience from four CDC studies. AIDS. plexes in healthy Africans. Lancet. 1985;2: Ingrid H. de Beer, MA† 2006;20:1655–1660. 520–523. Harold Kaura, MSc‡ 6. Jafa K, Patel P, MacKellar DA, et al. In- 15. Messele T, Abdulkadir M, Fontanet al, et al. Miche`le van Vugt, MD, PhD*§ vestigation of false positive results with an oral Reduced naive and increased activated CD4 Lucille Caparos, MSc‡ fluid rapid HIV-1/2 antibody test. PLoS ONE. and CD8 cells in healthy adult Ethiopians Tobias F. Rinke de Wit, PhD* 2007;2:e185. compared with their Dutch counterparts. Clin 7. Wesolowski LG, MacKellar DA, Facente SN, Exp Immunol. 1999;115:443–450. *PharmAccess Foundation et al. Post-marketing surveillance of OraQuick 16. Koch WH, Sullivan PS, Roberts C, Center for Poverty-Related Communicable whole blood and oral fluid rapid HIV testing. et al. Evaluation of United States-licensed Diseases, and Academic Medical Center AIDS. 2006;20:1661–1666. human immunodeficiency virus immu- of the University of Amsterdam 8. Granade TC, Phillips SK, Parekh B, et al. noassays for detection of group M viral Amsterdam, The Netherlands Detection of antibodies to human immunode- variants. J Clin Microbiol. 2001;39:1017– †PharmAccess Foundation Namibia ficiency virus type 1 in oral fluids: a large-scale 1020. Windhoek, Namibia evaluation of immunoassay performance. Clin 17. Ministry of Health and Social Services. Results ‡Namibia Institute of Pathology Diagn Lab Immunol. 1998;5:171–175. of the 2006 national sentinel survey among Windhoek, Namibia 9. Soto-Ramirez LE, Hernandez-Gomez L, pregnant women. Press release, April 17, 2007. Sifuentes-Osornio J, et al. Detection of specific Windhoek, Namibia: Ministry of Health and §Division of Infectious Diseases antibodies in gingival crevicular transudate Social Services; 2007. Tropical Medicine, and AIDS by enzyme-linked immunosorbent assay for 18. World Health Organization. WHO HIV Test Academic Medical Center of the diagnosis of human immunodeficiency virus Kit Bulk Procurement Scheme. Available at: University of Amsterdam type 1 infection. J Clin Microbiol. 1992;30: http://www.who.int/diagnostics_laboratory/ Amsterdam, The Netherlands 2780–2783. procurement/. Accessed June 10, 2007.

118 q 2008 Lippincott Williams & Wilkins

24 3 ANONYMOUS HIV WORKPLACE SURVEYS AS AN ADVOCACY TOOL FOR AFFORDABLE PRIVATE HEALTH INSURANCE IN NAMIBIA

25 Journal of the International AIDS Society BioMed Central

Research Open Access Anonymous HIV workplace surveys as an advocacy tool for affordable private health insurance in Namibia Ingrid de Beer†1, Hannah M Coutinho*†2, Peter J van Wyk3, Esegiel Gaeb4, Tobias Rinke de Wit2,5 and Michèle van Vugt2,6

Address: 1PharmAccess Foundation Namibia, Windhoek, Namibia, 2PharmAccess Foundation, Center for Poverty-related Communicable Disease, Academic Medical Center, Amsterdam, The Netherlands, 3Namibia Business Coalition for AIDS, Windhoek, Namibia, 4Namibia Institute of Pathology, Windhoek, Namibia, 5Center for Poverty-related Communicable Diseases, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands and 6Division of Infectious Diseases, Tropical Medicine and AIDS, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands Email: Ingrid de Beer - [email protected]; Hannah M Coutinho* - [email protected]; Peter J van Wyk - [email protected]; Esegiel Gaeb - [email protected]; Tobias Rinke de Wit - [email protected]; Michèle van Vugt - [email protected] * Corresponding author †Equal contributors

Published: 11 November 2009 Received: 9 February 2009 Accepted: 11 November 2009 Journal of the International AIDS Society 2009, 12:32 doi:10.1186/1758-2652-12-32 This article is available from: http://www.jiasociety.org/content/12/1/32 © 2009 de Beer et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract Background: With an estimated adult HIV prevalence of 15%, Namibia is in need of innovative health financing strategies that can alleviate the burden on the public sector. Affordable and private health insurances were recently developed in Namibia, and they include coverage for HIV/AIDS. This article reports on the efficacy of HIV workplace surveys as a tool to increase uptake of these insurances by employees in the Namibian formal business sector. In addition, the burden of HIV among this population was examined by sector. Methods: Cross-sectional anonymous HIV prevalence surveys were conducted in 24 private companies in Namibia between November 2006 and December 2007. Non-invasive oral fluid-based HIV antibody rapid tests were used. Anonymous test results were provided to the companies in a confidential report and through presentations to their management, during which the advantages of affordable private health insurance and the available insurance products were discussed. Impact assessment was conducted in October 2008, when new health insurance uptake by these companies was evaluated. Results: Of 8500 targeted employees, 6521 were screened for HIV; mean participation rate was 78.6%. Overall 15.0% (95% CI 14.2-15.9%) of employees tested HIV positive (range 3.0-23.9% across companies). The mining sector had the highest percentage of HIV-positive employees (21.0%); the information technology (IT) sector had the lowest percentage (4.0%). Out of 6205 previously uninsured employees, 61% had enrolled in private health insurance by October 2008. The majority of these new insurances (78%) covered HIV/AIDS only. Conclusion: The proportion of HIV-positive formal sector employees in Namibia is in line with national prevalence estimates and varies widely by employment sector. Following the surveys, there was a considerable increase in private health insurance uptake. This suggests that anonymous HIV workplace surveys can serve as a tool to motivate private companies to provide health insurance to their workforce. Health insurance taken up by those who are able to pay the fees will alleviate the burden on the public sector.

Page 1 of 7 (page number not for citation purposes)

26 Journal of the International AIDS Society 2009, 12:32 http://www.jiasociety.org/content/12/1/32

Background Sustainability of public HIV/AIDS prevention and treat- HIV predominantly affects adults of working age. On a ment programmes in the long run is questionable given global scale, the majority of these adults live in sub-Saha- their heavy reliance on donor funds. In addition, the ran Africa [1], where Namibia is among the countries necessity to integrate these programmes into existing pri- hardest hit by the epidemic. According to most recent esti- mary health care systems and improve the efficacy of these mates, adult HIV prevalence in Namibia is 15.3%, with a systems will greatly increase the costs, logistical challenges plausibility range of 12.4-18.1% [2]. and required human resources [3]. Additional, complementary approaches, such as health insurance, are there- Large-scale implementation of highly active antiretroviral fore required to enable the long-term success of global treatment (HAART) in sub-Saharan Africa is currently tak- efforts to improve health care in developing countries. ing place. An estimated 2.1 million people in this region are now receiving antiretroviral treatment under World Major benefits of health insurance include protection of Health Organization (WHO) guidelines, which comes individuals against catastrophic health expenditures, down to approximately one out of every three HIV- increased solidarity through financial risk pooling, and infected people in need of treatment [3]. As a conse- the possibility to channel "vertical" funds, such as for quence, analogous to developments in the western world HIV/AIDS, into general health financing [12,13]. Cur- after the introduction of HAART, a shift towards HIV/ rently, the majority of those with access to health insur- AIDS as a chronic disease is taking place in the region, ance in sub-Saharan Africa are the urban elite, in with opportunistic infections and co-morbidity becoming particular higher income formal sector workers, who can increasingly important [3]. obtain coverage (partly) subsidized through their employers [9,13]. The life-long quality care and treatment that is required for the masses of HIV-infected patients will further In Namibia, approximately 12.5% of the population was increase the demands placed on the already overburdened covered by health insurance in 2004 [14]. PharmAccess and understaffed public health care systems in sub-Saha- Foundation, a not-for-profit organization that aims to ran Africa. Notwithstanding the extraordinary global improve access to affordable and sustainable quality surge in funding, the financial costs of the HIV/AIDS epi- health care provision in sub-Saharan Africa, supported the demic are expected to rise more than four-fold if preven- launch of several Namibian health insurance packages tion and treatment scale up continues at the same pace as aimed at low- and middle-income workers. Crucial in this today [4]. was the development of a risk equalisation fund for HIV/ AIDS (HIVREF) in 2006, which enabled individual health Moreover, the region is facing a general transition in insurance providers to share the risks for this disease. health challenges, with chronic non-communicable dis- Thus, otherwise competing health insurers can collabo- eases, such as diabetes and cardiovascular diseases, taking rate in this unique solidarity fund [9]. over from infectious diseases as the most important cause of morbidity and mortality [5]. Because chronic diseases As a special option, employer and/or employee groups are more expensive to treat and cause long-term disability, that cannot afford the primary health insurance can pur- the demands on health care infrastructure and capacity are chase an "HIV/AIDS only" package, covered by the expected to further increase [5]. HIVREF. This HIV/AIDS health insurance is compulsory for all employees of a company that decides to enrol, The current HIV-1 prevention and treatment strategies in while enrolment in the majority of primary health insur- sub-Saharan Africa are largely being implemented ances is voluntary. through civil society and the public sector. Although the private business sector is affected by the epidemic [6] and PharmAccess Foundation recently conducted several workplace programmes were the first to pioneer HIV-1 anonymous HIV workplace surveys in the formal business treatment in the region [7,8], public HIV/AIDS treatment sector in Namibia with the aim of stimulating employers programmes have largely taken over, supported by large to provide the affordable health insurance products that international funds [4]. Today, only a limited number of we have described to their employees. It was hypothesized multinationals and an even smaller number of small and that providing companies with HIV prevalence estimates medium enterprises (SMEs) offer an HIV/AIDS proof their workforces would create awareness among the gramme to their employees [9-11]. Approximately 26% of management and thereby lead to health insurance uptake. the sub-Saharan African companies that have HIV policies This article reports the results of these surveys and is the provide antiretroviral treatment to their workers [7]. first quantitative documentation of the burden of HIV among employees in the Namibian formal sector.

Page 2 of 7 (page number not for citation purposes)

27 Journal of the International AIDS Society 2009, 12:32 http://www.jiasociety.org/content/12/1/32

Methods only sex and age. Data on age were collected either in exact Survey design and implementation years or in age categories, depending on the size of the Between November 2006 and December 2007, cross-sec- company, to ensure confidentiality and encourage maxi- tional anonymous HIV surveys were conducted among mum participation. employees of 24 private companies throughout Namibia. The surveys were conducted by PharmAccess Foundation Impact assessment on health insurance uptake Namibia, in partnership with the Namibia Business Coa- It was hypothesized that providing companies with HIV lition on AIDS (NABCOA) and the Namibia Institute of prevalence estimates of their workforce would create Pathology. awareness among the management and thereby lead to health insurance uptake. To test this hypothesis, impact NABCOA was launched in 2003 to mobilize the private assessment was conducted as follows. In October 2008, business sector in the national HIV/AIDS response [15]; it when all cross-sectional HIV workplace surveys had been did so through its "Healthy Workforce, Healthy Business" conducted, the number of new insurance policies taken programme. Companies that expressed interest in HIV up by employees after the survey had been conducted was prevalence surveillance following this programme were reviewed. Data were obtained from several databases that referred to PharmAccess for implementation of HIV work- record data on insurance policies of the main providers in place surveys. The major incentive for companies to par- Namibia. PharmAccess has access to these databases as ticipate in these surveys was to obtain information to part of its external quality control responsibilities. Infor- develop or improve HIV/AIDS workplace programmes. mation on uptake of insurances that were not recorded in this database was obtained directly from the companies. In each company, surveys were prepared and conducted as follows. First, awareness-raising presentations were pro- Statistical analyses vided to the management, which stressed the value of HIV prevalence estimates for internal HIV/AIDS policy. In Statistical analyses were performed with SPSS version 15.0 addition, indirect effects of the surveys, such as increased for Windows, Chicago: SPSS Inc. For significance testing, awareness about HIV/AIDS among employees, were dis- Chi square and Student's T-test were used for dichoto- cussed. Second, education and sensitization sessions were mous and continuous variables, respectively. P-values < held for both management and employees on the process 0.05 were considered statistically significant. of surveillance and the importance of participation. Dur- ing these sessions, the importance of access to treatment Results and the need to mitigate the impact of HIV on the busi- HIV test results ness was highlighted. The availability of affordable health Table 1 shows overall HIV results of the surveys, stratified insurance packages was introduced as a risk-mitigation by industry and company, as well as by new insurance intervention. Third, anonymous and voluntary HIV prev- uptake. Overall, 6521 of 8500 targeted employees partici- alence surveys were conducted. Finally, anonymous sur- pated in the HIV surveys in 24 companies located vey results were presented to the management and throughout Namibia. Participation rates within compa- advocacy meetings were held to stimulate company nies varied from 61.3% to 97.3%, with a mean (95% CI) uptake of affordable private health insurance, including participation rate of 78.6% (78.3-78.8%). In total 980 out HIV/AIDS coverage, for employees. of 6521 employees tested HIV positive, suggesting an HIV prevalence of 15.0% (95% CI 14.2-15.9%). This propor- HIV testing and confidentiality tion varied from 3.0-23.9% between companies (Table 1). For HIV testing, OraQuick Rapid HIV-1/2 Antibody Tests (OraSure Technologies, Inc, Bethlehem, PA Figure 1 shows the proportion of employees who tested ["OraQuick"]) were used. This non-invasive HIV rapid HIV positive, stratified by employment industry. Trans- test was validated in Namibian high-risk populations in port, manufacturing, agriculture, fishing and mining 2005, showing 100% sensitivity and 100% specificity appear to be "high-risk industries", defined as those with [16]. a proportion of HIV-positive employees greater than the overall survey mean of 15.0%. The mining sector had the Because HIV results of the survey were not disclosed on an highest proportion of HIV-positive employees (21.0%), individual level, all participating employees were encour- whereas this was lowest in the information technology aged to visit a voluntary counselling and testing facility to sector (4.0%). However, in the latter sector, only a small obtain their HIV status in accordance with national HIV number of employees were tested. testing requirements. To guarantee confidentiality and ensure willingness to participate among employees, col- In order to guarantee confidentiality, data on sex were not lection of demographic data was limited and included collected in 11 of the 24 companies, resulting in registered

Page 3 of 7 (page number not for citation purposes)

28 Journal of the International AIDS Society 2009, 12:32 http://www.jiasociety.org/content/12/1/32

Table 1: HIV results by company and new insurances taken up by October 2008

Industry Company Participation rate1 Participation by sex (M/F)2 HIV positive New insurances3 Insurance type4

No. % No. No. % No. % Transport 1 308/447 68.9 132/176 49 15.9 113 25.3 Traditional Tourism 2 165/239 69.0 - 26 15.8 178 74.5 HIV only 3 127/149 85.2 - 6 4.7 118 79.2 HIV only Retail 4 714/863 82.7 - 77 10.8 578 70 HIV only Manufacturing 5 349/425 82.1 - 53 15.2 297 69.9 HIV only 6 511/525 97.3 - 54 10.6 359 68.4 HIV only 7 88/105 83.8 - 21 23.9 87 82.9 HIV only 8 202/215 94.0 149/53 29 14.4 98 45.6 HIV only 9 248/296 83.8 205/43 52 21.0 289 95.7 HIV only 10 400/653 61.3 332/68 88 22.0 924 145.5 7 HIV only Wholesale 11 115/137 83.9 54/61 9 7.8 18 9 13.1 Traditional 12 54/61 88.5 35/19 3 5.6 4 9 6.6 Traditional IT 13 25/31 80.6 - 1 4.0 12 8 12.1 HIV only Services 14 74/92 80.4 - 4 5.4 15 77/82 93.9 - 18 23.4 68 82.9 HIV only 16 383/625 61.3 311/72 65 17.0 324 51.8 Affordable 17 235/319 73.7 112/123 7 3.0 7 2.2 Traditional 18 102/131 77.9 86/16 8 8.8 0 0 - 19 155/161 5 96.3 44/111 5 3.2 3 9 1.9 Traditional Financial services 20 279/374 74.6 128/151 39 14.0 0 9 0- Fishing 21 664/1049 63.3 - 121 18.2 774 73.8 HIV only 22 287/435 6 66.0 86/201 49 17.1 328 75.4 HIV only Agriculture 23 154/177 87.0 - 26 16.9 153 86.4 HIV only Mining 24 805/909 88.7 724/82 169 21.0 47 9 5.2 Traditional

Total or mean N = 24 6521/8500 78.6 2398/1175 980 15.0 4779 56.2 -

1 Participation rate is defined as number of participating employees relative to target population. Target population is defined as total number of employees within company at time of workplace survey 2 Sex was recorded for 3573 of 6521 (54.8%) participating employees 3 Percentage of new insurances is defined as number of new insurances relative to the total number of employees per company at the time of the survey 4 Traditional insurance, which existed prior to introducing affordable insurance products, entails income dependent individual monthly premiums of N$800-2300; for affordable insurance, the age dependent monthly premium is N$250-350; for HIV coverage only, the monthly premium for all is N$30 5 N = 155/161 (96.3%) employees were on site on the day the survey was performed; use of this number would result in participation rate of 100% 6 N = 292/435 (67.1%) employees were on site on the day the survey was performed; use of this number would result in participation rate of 98.3% 7 Temporary employees, who were not part of the survey, were included in the new insurances taken up by this company 8 Insurance data of companies 13 and 14 could not be evaluated separately and were thus combined 9 All employees were insured at the time of the workplace survey sex for 3572 (54.8%) employees. In the 13 companies age is shown in Figure 2. Individuals in their 40 s had the where sex was registered, the majority of participants were highest risk to test HIV positive, whereas those younger male (67.1%; between company range of 28.4-89.9%; than 30 years old had the lowest risk. Sex-stratified analy- Table 1). In all, 431 of 2394 men (18.0%; 95% CI 16.5- sis showed an equal HIV distribution for women across 19.6%) and 142 of 1175 women (12.1%; 95% CI 10.3- age categories; the range of women testing HIV positive 14.0%) tested HIV positive (p < 0.0001). was 13.5-14.6% across age categories. Results were quite different for men. In the 31-50 years age group, 23% of Age was registered for 6514 (99.9%) employees. Exact age men tested HIV positive, versus 15% in men younger than was registered for 2718 (41.7%) employees in nine of 24 31 or older than 50 years. Of note, these estimates are companies; in this subgroup, mean age (95% CI) was 35.1 based on a relatively small proportion of the cohort (34.8-35.5) years, with a range of 18 to 69 years. Mean age (27.7%) because of the large amount of missing data on (95% CI) among employees who tested HIV positive and sex. negative was 36.1 (35.2-36.9) and 35.0 (34.5-35.4) years, respectively (p = 0.03). For the remaining 3796 employees, age was registered in categories. HIV distribution by

Page 4 of 7 (page number not for citation purposes)

29 Journal of the International AIDS Society 2009, 12:32 http://www.jiasociety.org/content/12/1/32

had been conducted in the 24 companies, 4779 new insurances were registered (Table 1). This comes down to coverage of 56% of the employees working at one of these companies at the time of the survey, assuming a constant workforce. The broad range of new insurances, varying from 0-146% between companies, can be explained in part by the fact that five of the 24 companies already provided health insurance to all their employees at the time of the survey. In addition, one company provided insurance to both permanent and temporary employees; the latter group did not participate in any of the HIV surveys (Table 1). Exclusion of these six companies resulted in 3783 new insurances in the remaining 18 companies, which employed 6205 individuals at the time of the survey. This suggests that 61% of the previously uninsured workforce was insured in October 2008, assuming a stable ProportionFigure 1 of HIV-positive employees stratified by industry Proportion of HIV-positive employees stratified by workforce. The majority (78%) of the new insurance industry. Numbers at bottom of bars represent mean par- products offered by these companies covered HIV/AIDS ticipation rate per industry category. Error bars represent only. Subsidization by employers ranged from 50% to 95% confidence intervals. The horizontal line represents 100% of the monthly premium for the newly purchased mean percentage of HIV-positive employees in the entire private health insurances. cohort. Discussion This study describes results of anonymous HIV workplace Impact assessment on health insurance uptake surveys among employees of 24 private companies in It was hypothesized that HIV workplace surveys would Namibia. The primary aims were to: (1) estimate HIV result in increased uptake of affordable private health prevalence among formal sector employees; and (2) use insurance by formal sector employees. In October 2008, these prevalence estimates as a tool to advocate imple- which was between 10 and 21 months after the surveys mentation of affordable health insurance for employees, including HIV/AIDS coverage.

Our finding that 15% of employees tested HIV positive is in line with national prevalence estimates [2], despite the fact that formal sector employees are not a representative sample of the general population. Interestingly, among workplace survey participants whose sex was registered, men were 1.5 times more likely than women to test HIV positive. This finding contrasts with national and international HIV prevalence data in general populations, where women are generally infected at higher rates [17]. Perhaps formal sector employment, and thus increased financial independence, is a protective factor against HIV/AIDS for women. However, this finding may be biased by the large amount of missing data on sex.

Impact assessment showed that new health insurance uptake was considerable, which suggests that anonymous HIV workplace surveys can trigger implementation of pri- ProportionFigure 2 of HIV-positive employees stratified by age Proportion of HIV-positive employees stratified by vate health insurance in the Namibian formal sector. After age. Data shown represent 86.8% of the cohort. Numbers at presentation of survey results to the company manage- the bottom of the bars represent total number of tested indi- ment, 18 of the 19 companies that did not yet provide viduals per age category. Error bars represent 95% confi- health insurance for employees expressed a willingness to dence intervals. The horizontal line represents mean do so. percentage of HIV-positive employees in the entire cohort.

Page 5 of 7 (page number not for citation purposes)

30 Journal of the International AIDS Society 2009, 12:32 http://www.jiasociety.org/content/12/1/32

Upon evaluation in October 2008, which was 10 to 21 Limitations of this study need to be discussed. First, we months after the surveys were conducted, 61% of previ- were unable to directly measure an impact of our surveys ously uninsured employees were enrolled in private on health insurance status of employees. Data on the health insurance. Because we had no access to registration number of insured employees prior to conducting the sur- data of some Namibian health insurance providers, this veys, or insurance premium subsidization by employers figure is likely to be an underestimation of the number of following the surveys, could not be collected due to the newly purchased insurances. The current data demon- operational nature of our research. Instead, we used an strate that even SMEs can be persuaded to invest in health- overview of newly registered insurances of the main insur- care solutions for their workforces, despite their limited ance companies as a proxy for employee insurance status resources compared to large companies [11]. several months after conducting the surveys. This indirect impact assessment assumed that the workforce of the Implementation of affordable health insurance in the pri- companies remained constant, since we were unable to vate business sector, including HIV/AIDS coverage, is rel- obtain data on employee turnover. evant for several reasons. First, approximately 5% of health services in Namibia are currently delivered through Nevertheless, the considerable increase in new insurance the private health sector [15]. Through implementation uptake by employees does suggest our surveys may have and expansion of affordable private health insurance triggered this. Prior to the surveys, none of the companies linked to output-based contracts with the private health offered insurance covering HIV/AIDS only, which was sector, this underutilization can be improved. This will taken up by the majority of employees. In addition, to the alleviate the burden currently placed on public health best of our knowledge, there were no targeted marketing programmes [9,13]. These public programmes can subse- campaigns by insurance companies following our surveys. quently focus their resources on the poorest segments of the population, which are unable to pay for health insur- Second, because self-selection following the NABCOA ance. campaign was the reason for companies to participate in our surveys, participation bias cannot be excluded. Second, raising HIV awareness and knowledge remains Finally, health care requirements of HIV/AIDS-related important considering the large amount of stigma that morbidity in sub-Saharan Africa have become more com- remains a major issue in many sub-Saharan African socie- plex and demanding since large-scale treatment has ties [18]. By offering health insurance that covers HIV, become available. Insurance products that focus on HIV/ companies may be able to promote more openness about AIDS only are therefore outdated. In Namibia, we are cur- this disease among employees. rently piloting "wellness workplace surveys" that focus on both HIV/AIDS and chronic diseases, such as diabetes and Third, HIV infection appears more concentrated among cardiovascular diseases. These surveys may motivate pri- the employed and more mobile members of society vate companies to provide health insurance products with [17,19]. A household survey performed in Windhoek, more extensive coverage to their workforces, in particular Namibia's capital city, in 2006 to evaluate the effect of with respect to chronic diseases that require lifelong treat- affordable health insurance on the population level found ment. that the relative risk to test HIV positive for employed versus unemployed adults aged 15 to 49 years was 1.5 [20]. Conclusion Moreover, HIV can be regarded as an occupational health In conclusion, this study describes results from the largest hazard in certain employment sectors, for example, in the workplace-based HIV survey performed in Namibia to mining sector, where this increased risk is related to the date. The proportion of HIV-positive formal sector large number of migrant workers [21]. employees is in line with national prevalence estimates and varies widely by employment sector. The considera- Targeting such high-risk populations will not only serve ble increase in health insurance uptake suggests that public health needs, but also result in a healthier work- anonymous HIV workplace surveys can serve as a tool to force and subsequently lead to greater productivity, a implement private health insurance in the formal busi- reduced need for worker replacement [6,19,22,23] and ness sector. direct financial gains for the private business sector. To overcome the notion among SME managers that HIV/ To sustain current HIV/AIDS prevention and treatment AIDS is not a relevant problem among their workforces strategies in developing countries, cooperation of private [11], anonymous HIV workplace surveys can aid in creat- and public efforts is required. Private health insurance, ing awareness and making informed decisions. paid by those who can afford the premiums, can alleviate the burden on the public health system [9] and thereby

Page 6 of 7 (page number not for citation purposes)

31 Journal of the International AIDS Society 2009, 12:32 http://www.jiasociety.org/content/12/1/32

make an important contribution to sustainable health 12. Palmer N, Mueller DH, Gilson L, Mills A, Haines A: Health financing to promote access in low income settings-how much do we care systems in the developing world. know? Lancet 2004, 364:1365-1370. 13. Sekhri N, Savedoff W: Private health insurance: implications for Competing interests developing countries. Bull World Health Organ 2005, 83:127-134. 14. Hohmann , Skolnic : Options and scenarios for HIV/AIDS risk The authors declare that they have no competing interests. cover for low-income employees within NABCOA member companies in Namibia. In Report of the Feasibility Study Windhoek; NABCOA Survey; 2004. Authors' contributions 15. (MoHSS) MoHaSS: Follow-up to the Declaration of Commit- IDB conceived the project, collected data and edited the ment on HIV/AIDS. In Namibia Country Report 2005 Windhoek, manuscript. HMC analyzed the data and wrote the manu- Namibia: Ministry of Health and Social Services; 2005. 16. Hamers RL, de Beer IH, Kaura H, van Vugt M, Caparos L, Rinke de script. PJW conceived the project and edited the manu- Wit TF: Diagnostic accuracy of 2 oral fluid-based tests for HIV script. EG was responsible for the HIV test results and surveillance in Namibia. J Acquir Immune Defic Syndr 2008, edited the manuscript. TRW conceived the project and 48:116-118. 17. Mishra V, Vaessen M, Boerma JT, Arnold F, Way A, Barrere B, et al.: supervised and edited the manuscript. MVV conceived the HIV testing in national population-based surveys: experience project and supervised and edited the manuscript. All from the Demographic and Health Surveys. Bull World Health Organ 2006, 84:537-545. authors gave final approval of the version to be published. 18. Holzemer WL, Uys LR: Managing AIDS stigma. Sahara J 2004, 1:165-174. Acknowledgements 19. Piot P, Greener R, Russell S: Squaring the circle: AIDS, poverty, and human development. PLoS Med 2007, 4:1571-1575. The authors acknowledge the contributions of various stakeholders in the 20. Janssens W, Gaag J van der, Rinke de Wit T: Some pitfalls in the Namibian medical aid funding (Health is Vital and Namibia Health Plan) and estimation of HIV prevalence. AIID Research Series 08-03/1, medical administration industry (My Health Namibia, Methealth Namibia, Amsterdam, AIID 2008. Prosperity Health Group Namibia and Medscheme Namibia) that provided 21. Desmond N, Allen CF, Clift S, Justine B, Mzugu J, Plummer ML, et al.: data for several analyses. The authors acknowledge the support of Orasure A typology of groups at risk of HIV/STI in a gold mining town in north-western Tanzania. Soc Sci Med 2005, 60:1739-1749. Technologies Inc, PA, USA, for its donation of kits. Orasure Technologies 22. Larson BA, Fox MP, Rosen S, Bii M, Sigei C, Shaffer D, et al.: Early Inc did not provide funding for this study, nor did it have any role in the effects of antiretroviral therapy on work performance: pre- design and conduct of the study, nor in the preparation of the manuscript. liminary results from a cohort study of Kenyan agricultural The Namibia Institute of Pathology provided the external quality assurance workers. AIDS 2008, 22:421-425. 23. Fox MP, Rosen S, MacLeod WB, Wasunna M, Bii M, Foglia G, Simon on the surveillance conducted during these studies. The study was funded JL: The impact of HIV/AIDS on labour productivity in Kenya. by a grant from The Netherlands Postcodeloterij through Stop AIDS Now! Trop Med Int Health 2004, 9:318-324. and the Netherlands AIDS Fonds.

References 1. UNAIDS/WHO: Sub-Saharan Africa AIDS epidemic update regional summary. In 2007 AIDS epidemic update Geneva: UNAIDS and WHO; 2007. 2. UNAIDS/WHO: Epidemiological Fact Sheet on HIV and AIDS, Namibia. Surveillance UWWGoGHAaS. Geneva: UNAIDS/WHO; 2008. 3. WHO/UNAIDS/UNICEF: Towards Universal Access: scaling up priority HIV/AIDS interventions in the health sector. In Progress Report 2008 Geneva: WHO; 2008. 4. Cohen J: HIV/AIDS. The great funding surge. Science 2008, 321:512-519. 5. Abegunde DO, Mathers CD, Adam T, Ortegon M, Strong K: The burden and costs of chronic diseases in low-income and middle-income countries. Lancet 2007, 370:1929-1938. 6. Social Impact Assessment and Policy Analysis Corporation (Pty) Ltd (SIAPAC): Impact Assessment of HIV/AIDS on the Municipal- ities of Ongwediva, Oshakati, Swakopmund, Walvis Bay and Windhoek. In Summary Report Volume 1. Windhoek, Namibia: USAID Namibia; 2003. 7. Borght S Van der, Rinke de Wit TF, Janssens V, Schim van der Loeff Publish with BioMed Central and every MF, Rijckborst H, Lange JM: HAART for the HIV-infected employees of large companies in Africa. Lancet 2006, scientist can read your work free of charge 368:547-550. "BioMed Central will be the most significant development for 8. Borght SF Van der, Collier AC, Rinke de Wit T, Richards SC, Feeley disseminating the results of biomedical research in our lifetime." FG: A successful workplace program for voluntary counseling and testing and treatment of HIV/AIDS at Heineken, Sir Paul Nurse, Cancer Research UK Rwanda. Int J Occup Environ Health 2007, 13:99-106. Your research papers will be: 9. Feeley F, Connelly P, Rosen S: Private sector provision and financing of AIDS treatment in Africa: current develop- available free of charge to the entire biomedical community ments. Curr HIV/AIDS Rep 2007, 4:192-200. peer reviewed and published immediately upon acceptance 10. Ramachandran V, Shah MK, Turner GL: Does the private sector care about AIDS? Evidence from firm surveys in East Africa. cited in PubMed and archived on PubMed Central AIDS 2007, 21(Suppl 3):S61-72. yours — you keep the copyright 11. Rosen S, Feeley F, Connelly P, Simon J: The private sector and HIV/AIDS in Africa: taking stock of 6 years of applied Submit your manuscript here: BioMedcentral research. AIDS 2007, 21(Suppl 3):S41-51. http://www.biomedcentral.com/info/publishing_adv.asp

Page 7 of 7 (page number not for citation purposes)

32 4 UNIVERSITY STUDENTS AND HIV IN NAMIBIA: AN HIV PREVALENCE SURVEY AND A KNOWLEDGE AND ATTITUDE SURVEY

33 de Beer et al. Journal of the International AIDS Society 2012, 15:9 http://www.jiasociety.org/content/15/1/9

RESEARCH Open Access University students and HIV in Namibia: an HIV prevalence survey and a knowledge and attitude survey Ingrid H de Beer1*†, Huub C Gelderblom2,3,4†, Onno Schellekens5, Esegiel Gaeb6, Gert van Rooy7, Alta McNally8, Ferdinand W Wit9 and Rinke de Wit F Tobias5,9,10,11,12

Abstract Background: With an overall adult HIV prevalence of 15.3%, Namibia is facing one of the largest HIV epidemics in Africa. Young people aged 20 to 34 years constitute one of the groups at highest risk of HIV infection in Namibia. However, little is known about the impact of HIV on this group and its access to healthcare. The purpose of this study was to estimate HIV prevalence, to assess the knowledge of and attitudes towards HIV/AIDS, and to assess access to healthcare among university students in Namibia. Methods: We assessed HIV/AIDS knowledge and attitudes, HIV prevalence and access to healthcare among students at the Polytechnic of Namibia and the University of Namibia. HIV prevalence was tested through anonymous oral fluid-based tests. Results: Half (n = 2790/5568) of the university students and 45% (n = 2807/6302) of the Polytechnic students participated in the knowledge and attitudes surveys. HIV/AIDS knowledge was reasonable, except for misperceptions about transmission. Awareness of one’s own HIV status and risks was low. In all, 55% (n = 3055/ 5568) of university students and 58% (n = 3680/6302) of Polytechnic students participated in the HIV prevalence survey; 54 (1.8%) university students and 103 (2.8%) Polytechnic students tested HIV positive. Campus clinics were not the major providers of healthcare to the students. Conclusions: Meaningful strategies addressing the gap between knowledge, attitude and young people’s perception of risk of HIV acquisition should be implemented. HIV prevalence among Namibian university students appears relatively low. Voluntary counselling and testing should be stimulated. Efforts should be made to increase access to healthcare through the campus clinics.

Background The overall adult HIV prevalence in Namibia is esti- Namibia in southern Africa has approximately 2.2 mil- mated at 15.3%, which is among the highest in the lion inhabitants and is classified as a middle-income world [1]. The HIV prevalence among pregnant women country. The Namibian health system has both a public is 19.9% [2]. The estimated HIV prevalence is 10.3% health service through the Ministry of Health and Social among 15- to 24-year-old females, and 3.4% among 15- Services (MoHSS) and a relatively well-established pri- to 24-year-old males [1]. In 2008, it was estimated that vate health sector. However, as the country is experien- 204,000 Namibians live with HIV, with an estimated 39 cing a large HIV epidemic, HIV/AIDS places a new infections occurring every day, 44% of which are in significant burden on the Namibian health system [1]. young people between the ages of 15 and 24 years [3]. AIDS has been the most prominent cause of death in Namibia since 1996, and in 2007, was the cause of 25% of all deaths [4]. In recent years, comprehensive volun- * Correspondence: [email protected] tary counselling and testing (VCT) and HIV treatment † Contributed equally 1PharmAccess Foundation Namibia, Windhoek, Namibia programmes have been established in Namibia [5]. In Full list of author information is available at the end of the article

© 2012 De Beer et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

34 de Beer et al. Journal of the International AIDS Society 2012, 15:9 Page 2 of 8 http://www.jiasociety.org/content/15/1/9

2009, 66 public antiretroviral treatment (ART) sites were September 2007 at the polytechnic. HIV/AIDS knowl- operational throughout the country. ART coverage edge and attitudes (KA) were evaluated using a self- increased from 3% to 60% between 2003 and 2007. Pro- administered 16-question quantitative survey. This sur- jections for coverage until 2013 stand at 80%. By March vey also assessed healthcare access. 2008, 50,600 people, including approximately 8,000 Namibian nurses contracted by PharmAccess Founda- through the private sector, were receiving ART [3]. tion collected the HIV samples, which were analyzed by According to recent studies [2,5], young people aged laboratory technicians from the Namibia Institute of 20 to 34 years constitute one of the groups at highest Pathology. HIV antibody status was assessed using the risk of HIV infection in Namibia. This age group forms oral fluid-based OraQuick Rapid HIV-1/2 Antibody Test about 25% of the Namibian population. Overall, the (OraSure Technologies, Inc., Bethlehem, PA ["Ora- level of education in Namibia is high. According to the Quick"]) [14]. The OraQuick is an FDA-approved, non- last demographic and health survey, in 2006/2007, more invasive, rapid diagnostic test that is suitable for epide- than half of the 20- to 34-year-old group attained the miological purposes, and has been officially validated by secondary level at school and up to 10% reached a the Namibian Institute of Pathology [14] and was higher educational level [6]. University students form an approved by the MoHSS for surveillance purposes in important constituency in interventions against HIV and 2006. AIDS. They are also identified as an interesting target group, as they represent the future leaders and eco- Organization of the KA and HIV prevalence surveys nomic backbone of the country. Both institutions implemented broad-scale awareness The Polytechnic of Namibia and the University of campaigns using campus radio, posters, flyers and infor- Namibia, both located in Windhoek, are the two largest mation boards prior to the surveys to raise awareness tertiary education institutions in the country, educating and encourage participation. more than 95% of Namibian university students. These Participation in both surveys was voluntary and anon- institutions provide primary healthcare and curative ser- ymous. Students received an explanation about the pur- vices to the students on campus. They give family plan- pose and objectives of the study before being asked for ning and health education on sexually transmitted consent and to fill in the questionnaire. infections, such as HIV/AIDS. However, to our knowl- The surveys were conducted at strategically placed edge, no data exist on the HIV prevalence at Namibian sites throughout the Windhoek campuses of UNAM institutions of higher learning, or on the impact of HIV and the Polytechnic, over a period of five days from and access to healthcare of university students. Few stu- 07h00 to 20h00, in order to facilitate the participation dies have focused on students as a group in sub-Saharan of as many students as possible, irrespective of location Africa [7-13]. or class schedule. The purpose of this study was: (1) to assess students’ Test sites were separated into three separate areas to knowledge of and attitudes towards HIV/AIDS; (2) to ensure confidentiality: 1) an area where students would estimate HIV prevalence among university students; and fill out the KA survey; 2) an area where the oral fluid (3) to assess their access to healthcare. We report that swab for the HIV survey was taken by a nurse; and 3) a among university students in Namibia, we found a rea- laboratory. When students participated in both surveys, sonable overall HIV knowledge, but identified some the results of the KA survey were linked to the results gaps. We also found relatively low overall HIV preva- of the HIV prevalence survey using an anonymous bar- lence, although it was high in some sub-groups, and low code system. use of existing campus health facilities. Informed consent and ethical clearance Methods The surveys were performed at the request of the man- Study population agement committees of the university and the polytech- The target group population consisted of all students of nic and in coordination with respective student the University of Namibia (UNAM) and the Polytechnic representative platforms for operational purposes to of Namibia in Windhoek, comprising more than 11,800 inform the institutions’ HIV programmes. The study students. interventions were approved by the participating institutions ethical committees as a part of the HIV manage- Surveys ment programme. Individual students only participated Two surveys were conducted separately: a survey on voluntarily in HIV testing after oral informed consent. knowledge and attitudes (KA) towards HIV/AIDS and HIV testing was performed anonymously and no results an HIV prevalence survey. The surveys were conducted were returned to participating individuals. No individual from 6 to 10 August 2007 at UNAM and from 10 to 17 identifiers were collected. Those students who wished to

35 de Beer et al. Journal of the International AIDS Society 2012, 15:9 Page 3 of 8 http://www.jiasociety.org/content/15/1/9

know their HIV status were referred to VCT centres respondents had misperceptions about transmission of established in Windhoek. The analysis of data for this HIV by deep kissing (50% UNAM, 44% Polytechnic), paper was conducted after the operational surveys of the witchcraft (26% UNAM, 27% Polytechnic) and shaking individual institutions using existing anonymous data. hands (14% UNAM, 15% Polytechnic). Almost all respondents knew that HIV could not be transmitted by Statistical analysis sharing food with an HIV-positive person (98% UNAM, Graphical representation and statistical analysis were 97% Polytechnic). performed using GraphPad Prism version 4.0b for Knowledge of treatment possibilities of HIV/AIDS Macintosh (GraphPad Software, San Diego, CA, USA), appeared high: 90% of UNAM and 86% of polytechnic Microsoft Excel 2004 for Mac (Microsoft, Seattle, WA, respondents indicated that they knew that there was no USA), and SPSS version 16 for Macintosh (SPSS Inc., treatment that can eradicate HIV/AIDS from a human’s Chicago, IL, USA). Statistical analysis was done using body, that traditional healers could not remove HIV the Chi-square test for categorical variables. A two-sided from a human’s body (96% UNAM, 94% Polytechnic), p value < 0.05 was considered statistically significant. that a person with tuberculosis was not always HIV positive (95% UNAM, 93% Polytechnic) and that ART Results could enable HIV-infected people to live longer, heal- Participation in the KA survey thier lives (94% UNAM, 89% Polytechnic). Half (n = 2790/5568) of the UNAM students and 45% Television/radio (73% UNAM, 66% Polytechnic), print (n = 2,807/6,302) of the Polytechnic students partici- media (70% UNAM, 61% Polytechnic) and health insti- pated in the KA surveys (Table 1). The participation tutions (66% UNAM, 64% Polytechnic) were mentioned rate in the KA survey was significantly higher among by most students as major sources of information about female students at both institutions: 65% of female HIV/AIDS. Fewer students mentioned information on UNAM students compared with 40% of male UNAM campus (45% UNAM, 27% Polytechnic) and the campus students (p < 0.0001, Chi-square), and 51% of female nurse/counsellor (15% UNAM, 11% Polytechnic) as Polytechnic students compared with 39% of male Poly- major sources of information on HIV/AIDS. technic students (p < 0.0001, Chi-square) (Table 2). The general attitude towards people living with HIV/ Of the 5,597 students who participated in the KA sur- AIDS was positive for the majority of respondents, indi- vey, most were females (60%). The age of the partici- cating that they would be willing to attend classes with pants ranged from 14 to 53 years (mean = 21.9+/- 4.3); an HIV-positive fellow student (95% UNAM, 94% Poly- a large majority (93%) were 18 to 30 years old. Among technic), or attend lectures presented by an HIV-positive the respondents, 30% were first-year students, 27% were lecturer (96% UNAM, 95% Polytechnic). However, only second-year, 20% were third-year, and 23% were fourth- 75% of UNAM and 77% of Polytechnic respondents year or more. indicated that they would eat food that was prepared by an HIV-positive person. Although 50% of UNAM and Outcome of the KA survey 54% of Polytechnic respondents had indicated that they Basic knowledge of HIV/AIDS was good: 95% of UNAM knew HIV could not be transmitted through deep kis- and 93% of Polytechnic respondents knew what HIV/ sing, only 31% of UNAM and 37% of Polytechnic AIDS was; and 97% of UNAM and 96% of Polytechnic respondents indicated that they themselves would actu- correctly defined the difference between HIV and AIDS. ally kiss an HIV-positive person. Knowledge of HIV prevention appeared high as 92% of Awareness of the existence of the institutional HIV/ all respondents knew that using a condom could protect AIDS programmes was moderate (72% UNAM, 71% against HIV infection. However, a number of Polytechnic). Highest on the list of programme services that students desired were availability of counselling (63% UNAM, 59% Polytechnic), HIV/AIDS-related Table 1 Number of participants in KA survey and/or HIV information (51% UNAM, 56% Polytechnic), support for prevalence survey HIV-positive people (64% UNAM, 50% Polytechnic) and Institution HIV survey yes HIV survey no Total condom distribution (36% UNAM, 36% Polytechnic). UNAM KA yes 2,789 (50%) 1 (0%) 2,790 (50%) KA no 266 (5%) 2,512 (45%) 2,778 (50%) Participation in the HIV prevalence survey Total 3,055 (55%) 2,513 (45%) 5,568 (100%) In all, 55% (n = 3055/5568) of UNAM students and 58% (n = 3,680/6,302) of Polytechnic students participated in Polytechnic KA yes 2,706 (43%) 101 (1.6%) 2,807 (45%) the HIV prevalence survey (Table 1). The participation KA no 974 (15%) 2,521 (40%) 3,495 (55%) rate in the HIV surveillance was significantly higher Total 3,680 (58%) 2,622 (42%) 6,302 (100%) among female students (67% of female UNAM students

36 de Beer et al. Journal of the International AIDS Society 2012, 15:9 Page 4 of 8 http://www.jiasociety.org/content/15/1/9

Table 2 Participation of students in the HIV prevalence survey and KA survey according to gender and institution Total (n) Participated in HIV survey Participated in KA survey n %n% Female students UNAM 3,053 1,950 (67%) 1,792 (65%) Male students UNAM 2,515 1,105 (44%) 998 (40%) All students UNAM 5,568 3,055 (55%) 2,790 (50%) Female students polytechnic 3,042 2,141 (71%) 1,542 (51%) Male students polytechnic 3,260 1,539 (44%) 1,267 (39%) All students polytechnic 6,302 3,680 (58%) 2,807 (45%) Participation rates in the HIV survey and KA survey were significantly higher among female students at both UNAM and the Polytechnic, p < 0.0001, Chi-square compared with 44% of male UNAM students, p < (Figure 1). The HIV prevalence was significantly higher 0.0001, Chi-square, and 71% of female Polytechnic stu- in female Polytechnic students (3.5%, n = 75/2141) than dents compared to 47% of male Polytechnic students, p in male Polytechnic students (1.8%, n = 28/1539; p = < 0.0001, Chi-square) (Table 2). Participation was espe- 0.0031, Chi-square) (Figure 1). HIV prevalence was cially low (22%) among part-time male Polytechnic stu- three times as high in part-time Polytechnic students dents (data not shown). The participation rate was (5.4%, n = 55/1016) than in full-time Polytechnic stu- higher in full-time Polytechnic students than in part- dents (1.8%, n = 47/2653; p < 0.0001, Chi-square, not time Polytechnic students (p < 0.0001, Chi-square, data shown) and highest in part-time female Polytechnic stu- not shown). dents (6.0%, n = 42/704, data not shown). HIV prevalence increased according to age in both institutions Outcome of the HIV prevalence survey with a peak in the 35-39 age group (Figure 2). Of the 3,055 UNAM students and 3,680 Polytechnic students who participated in the anonymous HIV preva- Behaviour and practices relating to own HIV status lence surveillance, 54 UNAM students (1.8%) and 103 The majority of students are unaware of their HIV sta- Polytechnic students (2.8%) tested HIV positive (Figure tus and risks. Overall, 46% of UNAM and 42% of Poly- 1). The HIV percentage appeared higher in female technic respondents reported knowing their HIV status; UNAM students (2.1%, n = 40/1950) than in male 64% (n = 32/50) of the UNAM respondents (signifi- UNAM students (1.3%, n = 14/1105), but this difference cantly higher than HIV-negative students who com- was not statistically significant (p = 0.15, Chi-square) pleted the KA survey, p = 0.0015, Chi-square) and 53%

Figure 1 Numbers of screened students and HIV prevalence, according to gender and institution. The size of the bars on the right depict the prevalence. The absolute numbers of HIV-positive students are depicted in the bars as “n = ...”. #The HIV prevalence was significantly higher among female students, compared with male students, at the polytechnic, p = 0.031, Chi-square.

37 de Beer et al. Journal of the International AIDS Society 2012, 15:9 Page 5 of 8 http://www.jiasociety.org/content/15/1/9

Figure 2 Number of students screened, and distribution of HIV-positive students according to age and institution. The size of the bars on the right depict the prevalence. The absolute numbers of HIV-positive students are depicted in the bars as “n=...”. In total, 2,905 of 3,680 (79%) participating Polytechnic students and 2,774 of 3,055 (91%) participating UNAM students disclosed their age and were included in this analysis. Note that although the overall percentage of HIV-positive students was higher at the Polytechnic with 2.8%, compared with 1.8% at UNAM, the percentage of HIV-positive students at the Polytechnic who disclosed their age was lower than among the UNAM students who disclosed their age. See also Figure 3.

(n = 24/45, p = not significant compared with HIV- HIV/AIDS prevention, transmission and treatment was negative students who completed the KA survey) of the high, although there were some important mispercep- Polytechnic respondents who tested HIV positive during tions; that HIV prevalence among respondents was the HIV prevalence survey (and who also completed the lower than expected (1.8% at UNAM and 2.8% at the KA survey) reported that they knew their status. Polytechnic); and that campus health facilities were Only 40% of UNAM and 39% of Polytechnic respon- underused. dents indicated that they thought that they were at risk The level of overall knowledge about HIV/AIDS pre- of becoming infected with HIV. Of the respondents who vention, transmission and treatment is encouraging. Stu- tested positive for HIV during the prevalence surveil- dies from other higher institutions in Africa show lance, 52% of UNAM and 54% of Polytechnic respon- similar results, keeping in mind that the respondents dents indicated that they believed that they were at risk had at least secondary level of education and that infor- of contracting HIV, suggesting that HIV-positive stu- mation campaigns about HIV/AIDS had been conducted dents are better aware of their status and may be more in Namibia [8,15]. However, as in other studies per- aware of present or past risk behaviour. formed in sub-Saharan African countries [9,16], we observed some important misperceptions, such as the Healthcare access belief that HIV can be transmitted by deep kissing, With respect to healthcare delivery, students mentioned witchcraft and shaking hands. Overall attitudes towards state hospitals (38% UNAM, 41% Polytechnic), private people living with HIV/AIDS were positive, except for doctors (40% UNAM, 38% Polytechnic) and state clinics eating food prepared by an HIV-positive person or kis- (26% UNAM, 36% Polytechnic) as their primary access sing an HIV-positive person. As Tebourski et al demon- points. The campus clinics were the primary source of strated [16], there was a discrepancy between the healthcare for 39% of UNAM respondents (third-ranked knowledge that HIV cannot be transmitted in certain provider), and 18% of Polytechnic respondents (fourth- ways and willingness to engage in that behaviour. These ranked provider). knowledge and attitude issues should be addressed in HIV/AIDS educational campaigns, and meaningful stra- Discussion tegies that address the gap between knowledge, risky In this study among more than 5,000 university students sexual behaviour and young people’s perception of their in Namibia, we found: that overall knowledge about vulnerability to AIDS must be implemented [8].

38 de Beer et al. Journal of the International AIDS Society 2012, 15:9 Page 6 of 8 http://www.jiasociety.org/content/15/1/9

An alarming majority of HIV-positive and HIV-nega- Access to healthcare was not optimal. The campus tive students appeared to be unaware of their HIV sta- clinics were underused, particularly at the Polytechnic tus and their risk of acquiring or transmitting HIV. (fourth-ranked provider). Several explanations can be Many students (28% at UNAM and 29% at the Polytech- stated. First, the campus clinics are now financed as nic) did not know of the institutional HIV/AIDS aware- part of the overall running expenses of the institution; ness programmes, and a large majority of students no guarantee of quality of medical services can be obtained information about HIV/AIDS primarily from offered. Second, even if seeking care at overburdened media rather than from the university. This suggests public facilities away from campus is time consuming that awareness efforts should be improved, that educa- and expensive, students may do not want to change tional efforts at the university should be increased [9], their habits. Efforts should be made to increase access and that VCT should be further encouraged. to healthcare through the campus clinics.

Figure 3 HIV prevalence in relation to the amount of information provided by 3,680 polytechnic students who participated in the HIV prevalence survey. The overall HIV prevalence among the 3,680 Polytechnic students was 2.8%.

39 de Beer et al. Journal of the International AIDS Society 2012, 15:9 Page 7 of 8 http://www.jiasociety.org/content/15/1/9

The overall HIV prevalence among students was additional information they provided. Students who par- higher at the Polytechnic (2.8%) than at the university ticipated in the KA survey and also provided their age (1.8%), but appeared to be lower than expected, based had a lower HIV prevalence, and students who did not on Namibian population HIV prevalence data (3.7% and provide any additional information had a significantly 6.5% in persons aged 20-24 years and 25-29 years, higher HIV prevalence (Figure 3, Chi-square). respectively) [17]. The difference in overall HIV preva- The KA study may have been biased as it was based lence at the Polytechnic and UNAM may be related to on self-reported behaviours. Responses may have been the different demographics of the two student affected by memory bias or social desirability bias. populations. Although social desirability bias tends to be less when Two biases may be present. The first is the tertiary using self-administrated questionnaires compared with education level of the respondents in this study versus face-to-face questionnaires [20], it may still result in the general population that is assessed in the national underestimating sexual risk behaviour and attitudes surveys [1]. A recent household survey in Windhoek regarding HIV [21,22]. found that the HIV prevalence was lower among those who had finished secondary or higher education [17]. Conclusions Second, only 57% of all students participated in the In conclusion, at two universities in Namibia, we found HIV prevalence survey. Refusal to participate can gener- that students had moderate to good HIV/AIDS knowl- ateabiasknownas“volunteer bias”, which limits the edge, yet also had some important misperceptions about ability to generalize research findings and jeopardizes HIV/AIDS treatment and transmission, and low knowl- the validity of research outcomes [18]. In our study, for edge of their own HIV status. HIV prevalence was rela- example, there was 68% participation in full-time female tively low, but may be underestimated. There was Polytechnic students, of whom 2.3% were HIV positive, considerable interest in institutional HIV awareness pro- versus 22% in part-time male Polytechnic students, of grammes. The campus clinics were underused. These whom 4.8% were HIV positive. Hence, biases may have findings motivate continued and intensified prevention been introduced that most probably lead to underesti- and education initiatives through institutional HIV/ mations of actual HIV prevalence. AIDS awareness programmes. In a recent household survey in Windhoek, it was estimated that the HIV prevalence amongst non-survey par- Acknowledgements ticipants was four times higher than among participants We are grateful to the students who participated in the surveys and [19]. If we assume that the HIV prevalence among the acknowledge the PharmAccess staff for assistance with conducting the non-participating population of students is four times surveys, the management and staff of UNAM and the Polytechnic for their support during the surveys. higher than among those who participated, the HIV pre- The survey was financially supported through project Okambilimbili by the valence rates would be 4.2% at UNAM and 6.3% at the Nationale Postcode Loterij, The Dutch Aidsfonds, HIVOS and Stop Aids Now! Polytechnic, which is comparable to the national preva- Author details lence data [5]. This also reflected in a sub-analysis of 1PharmAccess Foundation Namibia, Windhoek, Namibia. 2Master’s Program the Polytechnic data where we compare HIV prevalence in Global Public Health, New York University, New York, NY, USA. 3Current according to the amount of information that students address: Hasso Plattner Research Laboratory, University of KwaZulu Natal, Durban, South Africa. 4Current address: International Trachoma Initiative, Task provided: many Polytechnic students who participated Force for Global Health, Emory University, Atlanta, USA. 5PharmAccess in the HIV prevalence survey did not provide their age Foundation, Amsterdam, The Netherlands. 6Namibia Institute of Pathology, 7 (2,905 of the 3,680 students in the HIV survey, 79%, but Windhoek, Namibia. Multidisciplinary Research Center, University of Namibia, Windhoek, Namibia. 8Polytechnic of Namibia, Windhoek, Namibia. 9Center this sub-set included only 55 of the 103 HIV-positive for Poverty-related Communicable Diseases, Academic Medical Center of the students, 54%), or participate in the KA survey (2,706 of University of Amsterdam, Amsterdam, The Netherlands. 10Center for Internal the 3,680 students in the HIV survey, 74%, also partici- Medicine, Center for Infection and Immunity (CINIMA), Amsterdam, The Netherlands. 11Amsterdam Institute for Global Health and Development pated in the KA survey, including 45 of the 103 HIV- (AIGHD), Amsterdam, The Netherlands. 12Academic Medical Center of the positive students,44%). University of Amsterdam, Amsterdam, The Netherlands. Most of the 55 HIV-positive students who disclosed Authors’ contributions their age were in the 20-24-year age group. Given the IHdB designed and coordinated the study, participated in the data analysis, fact that relatively fewer HIV-positive students (percen- and drafted the manuscript. HCG participated in the data analysis, tage wise) provided their age or participated in the KA performed the statistical analysis, and drafted the manuscript. OS participated in the design and coordination of the study, and helped draft survey, it is difficult to determine a relationship between the manuscript. EG, GvR, AMcN and FWW participated in the design and age and HIV status from the present data. This is even coordination of the study, and participated in the data analysis. TFRdW more evident when the HIV prevalence is calculated in conceived of the study, participated in its design and coordination, and helped draft the manuscript. All authors read and approved the final sub-groups of respondents based on the amount of manuscript.

40 de Beer et al. Journal of the International AIDS Society 2012, 15:9 Page 8 of 8 http://www.jiasociety.org/content/15/1/9

Competing interests doi:10.1186/1758-2652-15-9 The authors declare that they have no competing interests. Cite this article as: de Beer et al.: University students and HIV in Namibia: an HIV prevalence survey and a knowledge and attitude Received: 5 August 2011 Accepted: 22 February 2012 survey. Journal of the International AIDS Society 2012 15:9. Published: 22 February 2012

References 1. UNAIDS/WHO: Epidemiological Fact Sheet on HIV and AIDS 2008, Update. 2. Republic of Namibia, Ministry of Health and Social Services: Report of the 2006 National HIV Sentinel Survey 2007. 3. Republic of Namibia, Ministry of Health and Social Services, Directorate Special Programmes: Estimates and Projections of the Impact of HIV/AIDS in Namibia 2008. 4. Republic of Namibia, Ministry of Health and Social Services: Ministry of Health and Social Services System Review 2008. 5. Republic of Namibia, Ministry of Health and Social Services, United Nations General Assembly Special Session (UNGASS): Country Report. Reporting Period April 2006-March 2007 2008. 6. Republic of Namibia, Ministry of Health and Social Services: Namibia Demographic and Health Survey 2006-07 2008, 428. 7. Terry PE, Mhloyi M, Masvaure T, Adlis S: An examination of knowledge, attitudes and practices related to HIV/AIDS prevention in Zimbabwean university students: comparing intervention program participants and non participants. Int J Infect Dis 2006, 10:38-46. 8. Odu OO, Asekun-Olarinmoye EO, Bamidele JO, Egbewale BE, Amusan OA, Olowu AO: Knowledge, attitudes to HIV/AIDS and sexual behaviour of students in a tertiary institution in south-western Nigeria. Eur J Contracept Reprod Health Care 2008, 13:90-96. 9. Harding AK, Anadu EC, Gray LA, Champeau DA: Nigerian university students’ knowledge, perceptions, and behaviours about HIV/AIDS: are these students at risk? J R Soc Promot Health 1999, 119:23-31. 10. Adewole DA, Lawoyin TO: Characteristics of volunteers and non- volunteers for voluntary counseling and HIV testing among unmarried male undergraduates. Afr J Med Med Sci 2004, 33(2):165-170. 11. Baganizi E, Saah A, Bulterys M, Hoover DR, Celentano D, Alary M: Prevalence and incidence rates of HIV-1 infection in Rwandan students. AIDS 1997, 11(5):686-687. 12. Bateman C: HIV/AIDS on tertiary education campuses-data at last. S Afr Med J 2009, 99(1):18-20. 13. Keller P, McCarthy K, Mosendane T, Tellie M, Venter F, Noble L, Scott L, Stevens W, van Rie A: HIV prevalence among medical students in Johannesburg, South Africa. S Afr Med J 2009, 99(2):72. 14. Hamers RL, de Beer IH, Kaura H, van Vugt M, Caparos L, de Rinke Wit TF: Diagnostic accuracy of 2 oral fluid-based tests for HIV surveillance in Namibia. J Acquired Immune Defic Synd 2008, 48(1):116-118. 15. Svenson L, Carmel S, Varnhagen C: A review of the knowledge attitudes and behaviour of university students concerning HIV/AIDS. Health Promot Int 1997, 12:61-68. 16. Tebourski F, Ben Alaya D: Knowledge and attitudes of high school students regarding HIV/AIDS in Tunisia: does more knowledge lead to more positive attitudes? J Adolesc Health 2004, 34:161-162. 17. Gustafsson-Wright E, Janssens W, van der Gaag J: The inequitable impact of health shocks on the uninsured in Namibia. Health Policy Plan 2011, 26(2):142-156. 18. Callahan CA, Hojat M, Gonnella JS: Volunteer bias in medical education research: an empirical study of over three decades of longitudinal data. Med Educ 2007, 41:746-753. 19. Janssens W, van der Gaag J, de Rinke Wit T: Non-response bias in the measurement of HIV/AIDS prevalence in Namibia Amsterdam: Amsterdam Submit your next manuscript to BioMed Central Institute for International Development; 2008. and take full advantage of: 20. Catania JA, McDermott LJ, Pollack LM: Questionnaire response bias and face-to-face interview sample bias in sexuality research. J Sex Research • Convenient online submission 1986, 22:52-72. 21. Zaba B, Pisani E, Slaymaker E, Ties Boerma J: Age at first sex: • Thorough peer review understanding recent trends in African demographic surveys. Sex Transm • No space constraints or color ﬁgure charges Infect 2004, 80:28-35. • Immediate publication on acceptance 22. Braunstein SL, van de Wijgert JH, Nash D: HIV incidence in sub-Saharan Africa: a review of available data with implications for surveillance and • Inclusion in PubMed, CAS, Scopus and Google Scholar prevention planning. AIDS Rev 2009, 11:140-156. • Research which is freely available for redistribution 23. Republic of Namibia, Ministry of Health and Social Services: Report on National Testing Day 2008. Submit your manuscript at www.biomedcentral.com/submit

41 42 5 HEALTH CARE OPTIONS FOR COMMERCIAL FARM WORKERS IN NAMIBIA

 Health care options for commercial farm workers in Namibia

I De Beer1, HM Coutinho2, L Guariguata1, HT Fortsch4, R Hough1, TF Rinke de Wit3 1PharmAccess Foundation, Windhoek, Namibia 2PharmAccess Foundation, Amsterdam, the Netherlands 3Advocacy, Technology and Research, PharmAccess Foundation, Amsterdam Institute for Global Health and Development, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands 4Agricultural Employers' Association, Windhoek, Namibia

Submitted: 20 November 2009; Revised: 20 September 2010; Published: 22 February 2011 De Beer I, Coutinho HM, Guariguata L, Fortsch HT, Hough R, Rinke de Wit TF

Health care options for commercial farm workers in Namibia Rural and Remote Health 11: 1384. (Online), 2011

Available from: http://www.rrh.org.au

A B S T R A C T

Introduction: Limited access to health care in rural areas is a challenge in Namibia. In 2007 a survey was conducted among employers of commercial farms to assess the feasibility of introducing private, affordable health insurance that including HIV/AIDS coverage for commercial farm workers in Namibia. Healthcare access and utilization by people living and working on commercial farms were evaluated to gain insight into the possibility to strengthen health service delivery in this sector. Method: A cross-sectional survey of all members of the Agricultural Employers’ Association was conducted by telephone interview in a one-year period from 2006. The population sampled included 1708 employers in farms throughout 8 regions of Namibia. Results: In total, 1402 farm employers (82%) agreed to participate, representing 1414 farms and an average of 10.2 employees (range 0–342; 95% CI: 9.50, 10.94) per farm. Employers surveyed reported 95% of farms (95%CI: 93.6, 95.9) had access to at least one medical facility. Employers on the majority of farms (94.7%; 95%CI: 93.6, 95.9) reported that employees had visited at least one medical facility, most frequently using clinics (79.2%, 95%CI: 77.09, 81.32), doctors (50.1%; 95%CI: 47.53, 52.74) and mobile clinics (45.7%; 95%CI: 43.10, 48.28). Employers were significantly more likely to use private doctors (2=616.2, df=1, p<0.0003), travel longer distances (t=-11.34, df=1,470.5, p<0.0003) to reach them, and more likely to have health insurance coverage than employees (2=1,098, df=1, p<0.0003). Employers on several farms reported covering costs for health-related

transport (92.1%; 95%CI: 90.7, 93.5), medical consultations (62.2%; 95%CI: 59.7, 64.8), and providing free medications (88%; 95%CI: 86.2, 89.7). Only 0.85% of farms (95%CI: 0.37, 1.33) reported having employees enrolled in health insurance, but 77.1% (95%CI: 74.6, 79.4) of employers not providing insurance said they were willing to share costs for health insurance for employees. In addition, 148 farms had persons with formal medical training on site. Conclusions: Employees on commercial farms in Namibia and their dependants do not have adequate access to formal health care and in particular to HIV/AIDS-related services. Access could be improved by strengthening and expanding the on-site health services provided by some farmers, including basic care and support for on-site health professionals. While few employees were covered by health insurance, employers expressed great willingness to co-pay for a basic plan that would include outpatient care.

Key words: Namibia, access to care, co-payment, commercial farming, insurance.

Introduction government expenditure on health (7.6% of GDP12), and an extensive coverage of people on anti-retroviral therapy

(ART) for HIV/AIDS13, there is a lack of trained health Access to quality health care is a significant challenge in professionals serving in the public sector. In 2008, 72% of many parts of sub-Saharan Africa where the density of doctors (n=774) and 46% of registered nurses (n=2,989) in trained healthcare providers is the lowest in the Namibia were in private practice, serving less than 20% of world1. Despite trends towards urbanization, much of Sub- the population14. On average, the private sector employs Saharan Africa is heavily rural with limited access to 8.8 health workers per 1000 people, compared with 2 health hospitals and clinics in urban centres2. Many studies workers per 1000 people in the public sector14. document barriers to care for people living in rural settings worldwide, including the poor availability of trained staff, A potential partner in the provision of better health services barriers to information, basic equipment and materials for is the commercial agricultural sector, the largest private care, and physical distance to care3-6 . A number of studies employer in Namibia. As a significant contributor to the have evaluated strategies for improving healthcare access in economic development of Namibia, this sector has already rural Sub-Saharan Africa, including the use of community been engaged in addressing the HIV/AIDS epidemic, which health workers and participation in health insurance7-9 . still represents a significant health burden in the country11,15- However, few have looked at employers' willingness to pay 17 . For example, in 2005 the Agricultural Employers’ for health insurance as a strategy to improving healthcare Association (AEA), a membership-based organization access. representing employers in the commercial agricultural

sector, and the Namibia Farm Workers Union (NAFWU) Namibia, in particular, has the second lowest population signed an HIV/AIDS policy in 2005 to manage and mitigate density in the world, with many of its people living in rural the impact on the commercial agricultural sector11. However, settings and working in the agricultural sector10. There are there is no current system in place to provide basic primary 13 political regions in Namibia and commercial farming care to those employed in agriculture in Namibia14. takes place in 8 of these, with the remaining regions having communal land11. Providing access to health care for the Besides an increase in health professionals working in the largely dispersed population presents a serious challenge. public sector, another potential opportunity for improving While Namibia is significantly improving its healthcare access to care for those in rural settings is the introduction of financing and delivery system due to exceptionally high health insurance and cost-sharing approaches7. Health

insurance can be an important mechanism to avoid survey was piloted with 10 AEA-member commercial catastrophic individual and household healthcare farming employers and, following this, the questions were expenditure18. While a number of innovated insurance amended to better assess the primary goals of the survey. In schemes targeted at Namibians who could not afford private addition, a list of frequently asked questions was health insurance have been developed, at the last report only compiled. Once finalized, the survey was carried out using a 12.5% of the population was covered by health insurance, structured questionnaire with telephone interviews. Any and this represented people with middle to high incomes19-22 AEA-member commercial farm was eligible for inclusion . Given the small number of Namibians covered by health and all were contacted regarding participation. Of the insurance, engaging employers could provide a way to 1708 members invited to participate, 1402 employers expand that coverage and reduce high healthcare costs to (82.1%) agreed to the telephone survey and were included in individuals. Understanding the patterns of utilization among this study. The employers represented 1414 farms those employed in the agriculture sector, barriers of access to throughout 8 regions of Namibia. care, and the willingness of employers to pay for insurance schemes could help improve access to basic care for many The survey gathered data on: Namibians. • general characteristics of the farm and number of In 2007, at the request of the AEA, the PharmAccess employees and dependants Foundation in Namibia conducted a survey of AEA member • use of medical facilities and health expenditure by commercial farm employers to describe utilization of health all people working or living on farms services by employers and employees and associated costs; • local available medical expertise and services on describe employer knowledge, attitudes, and practices site regarding the provision of health services and support to • availability of first aid and provision of medicines employees; assess the status of accessible medical service on farms delivery; and assess employer willingness to pay for • general knowledge of HIV/AIDS as specified in health insurance schemes, including cost-sharing and AEA policy, including prevention reducing out-of-pocket expenditures. The survey was • willingness to pay for health insurance. designed to inform the implementation plan of the AEA  workplace policy. Although AEA members comprise only Personal identifiable information was not collected in order one-third of the estimated number of commercial farms, their to respect the privacy of respondents and eliminate any risk farms are geographically representative of the greater of breach of confidentiality. The telephone interviews were farming sector in Namibia. conducted by 20 peer enumerators selected from AEA members in each region and lasted approximately 30 min Methods each. Prior to the survey, enumerators received a one-day training covering the content of the questionnaire,

Survey design and target population interviewing skills, and responses to frequently asked questions. Employers were asked to provide information

about their farms regarding employee and employee- In 2007, the PharmAccess Foundation in Namibia conducted dependants’ healthcare-seeking behaviour. After the survey, a cross-sectional survey of members of the AEA to assess the peer enumerator answered questions for respondents utilization and provision of health services on farms (in from among the frequently asked questions collected from particular HIV/AIDS services) and possibilities for the the pilot of the survey, or relayed questions which the development of employer-provided insurance schemes. The

enumerator could not answer to the AEA leadership for by doctors (50.1%; 95%CI: 47.5, 52.7) and mobile clinics follow up. (45.7%; 95%CI: 43.1, 48.3). The vast majority of health utilization by employees was from public sources. From all Statistical analyses types of health facilities, employees used health services significantly more than employers and used significantly Data collected were entered into a database using MS Access more public facilities than private (Table 1). 2007. Statistical analysis was performed using R v2.10.0 (http://cran.r-project.org/bin/windows/base/) statistical The great majority of farm employers (90.7%; 95%CI: 89.2, software. Bivariate analyses of categorical variables were 92.2) reported they had visited at least one medical facility assessed using Pearson’s X2, continuous variables were over the previous 12 months. However, employers mostly assessed using Welch’s t-test, and an alpha of 0.05 was visited doctors (65.0%; 95%CI: 62.5, 67.5), dentists (49.5%; considered statistically significant. 95%CI: 46.9, 52.1) and optometrists (31.7%; 95%CI: 29.3, 34.1), and used primarily private facilities. Except for the use Results of mobile clinics, traditional healers, and nurses, employers reported travelling significantly longer distances to access

General characteristics health facilities than their employees (Table 1). The location of all commercial farms in Namibia and AEA members who

participated in the survey in relation to health facilities is Of the 1708 members invited, 1402 employers (82.1%) shown (Fig1). In general, health facilities are located at representing 1414 farms agreed to participate in the considerable distances from the widely dispersed farms. telephone survey; the farms were located throughout

8 geographical regions of Namibia (Fig1). Surveys were Costs of care administered on a ‘per-farm’ basis, so that employers who managed more than one farm responded more than once to The health-related costs assessed in this survey included the the survey. Therefore, in the results when referring to costs of a consultation, medications and transportation to and employers it is implied that they are representing a unique from the health facility. Employers were significantly more farm. In total, 87.1% of employers (95%CI: 85.4, 88.9) likely to pay for employees’ transportation costs (mean=2.8; resided permanently on a commercial farm. They reported a 95%CI: 2.57, 3.04; 2=303.5, df=1, p<0.0003) than the costs total of 8721 permanent employees, 5728 temporary  associated with medical consultations (mean=1.1; 95%CI: employees, and 14 754 employee dependants on all farms 0.96, 1.32; p<0.0003) or medications (mean=0.61; 95%CI: over the 12 months leading up to the survey date. The mean 0.47, 0.75; 2=538.1, df=1, p<0.0003) (Table 2). The number of employees per farm was 10.2 (range 0–342; 95%  situation was similar for employee dependants. Although the CI: 9.5, 10.9). average costs of transportation were not collected, the

Utilization, access and expenditures average cost of a medical consultation (NA$43.43; 95%CI: 36.67, 50.19) was almost double that of medications

(NA$26.84; 95%CI: 20.47, 33.20). The average one-way Employers (n=1414) were asked to report on the healthcare- distance travelled to a consultation was 63.36 km (95%CI: seeking behaviour of themselves and their employees in the 59.63, 67.08), which could result in several hours of 12 months prior to the survey (Table 1). A great majority of employees’ work lost. employers (94.7%; 95%CI: 93.6, 95.9) reported that employees had visited at least one medical facility and most frequently used clinics (79.2%, 95%CI: 77.1, 81.3), followed

Figure 1: Location of employers in relation to other commercial farms and public health facilities in Namibia.

Provision of health care and medications certification (16.8%; 95%CI: 14.8, 18.7). The details of why these professionals were on site or of their primary role was Because of the long travel distances often needed to reach not collected in this survey. However, their presence basic care, a great majority of farms reported providing care represents an opportunity for farm worker to avoid the costs for minor illnesses for employees (95.9%; 95%CI: 94.9, of and travel time for health care otherwise necessary. In 97.0) and their dependants (87.9%; 95%CI: 86.2, 89.6) on addition, employers reported providing different types of site. The details of how this care was provided were not basic medications to employees and their dependants on site captured by this survey; however, employers were asked (Table 4). The most provided medication was for minor whether any person living on the farm had some form of injuries, but medications for colds and influenza, diarrhoea, professional medical training. The study found that more and analgesics were also commonly available on farms at no than 10% of farms (10.47%; 95%CI: 8.9, 12.2) had at least cost to employees. Some employers reported selling one person in residence who had received professional medications to employees but free medications were most medical training (Table 3); however, this does not include common. the most common form of training: basic first aid

TableTable 1: Employer-reported 1: Employer-reported utilization utilization of and acces access sto to health health facilities facilities on farmson farms

FacilityFacility VariableVariable UtilizationUtilization PublicPublic Sector Sector DistanceDistance km km Type Total % (95%CI) P-value % (95%CI) P-value mean 95%CI P-value Type Total( n) % (95%CI) P-valuedf=1 % (95%CI) df=1P-value mean 95%CI df=1 P-value Hospital (n) df=1 df=1 df=1 Hospital Employee 1414 41.4 (38.9, 44.0) 93.0 (90.9, 95.1) 107.2 (100.8, 113.5) EmployeeEmployer 1414 1414 41.4 25.8 (38.9, (23.5, 44.0) 28.1) <0.003 93.0 9.6 (6.6, (90.9, 12.6) 95.1 ) <0.003 236.0 107.2 (218.8, (100.8, 253.3) 113.5) <0.003 Clinic Employer 1414 25.8 (23.5, 28.1) <0.003 9.6 (6.6, 12.6) <0.003 236.0 (218.8, 253.3) <0.003 Clinic Employee 1414 79.2 (77.1, 81.3) 97.7 (96.8, 98.6) 63.8 (61.3, 66.4) EmployeeEmployer 1414 1413 79.2 10.2 (77.1, (8.6, 81.3) 11.8) <0.003 68.197.7 (60.4, (96.8, 7 5.7)98.6) <0.003 88.8 63.8 (75.8, (61.3,101.8) 66.4) <0.003 Mobile Employerclinic 1413 10.2 (8.6, 11.8) <0.003 68.1 (60.4, 75.7) <0.003 88.8 (75.8, 101.8) <0.003

Mobile clinic Employee 1414 45.7 (43.1, 48.3) 98.5 (97.5, 99.4) 15.5 (12.7, 18.2) Employer 1414 15.1 (13.3, 17.0) <0.003 97.7 (95.6, 99.7) <0.003 12.6 (7.7, 17.5) 0.317 Employee 1414 45.7 (43.1, 48.3) 98.5 (97.5, 99.4) 15.5 (12.7, 18.2) Doctor <0.003 0.317 EmployerEmployee 1414 1410 15.1 50.3 (13.3, (47.7, 17.0) 52.9) 53.397.7 (49.6, (95.6, 57.0) 99.7) <0.003 98.1 12.6 (93.0, (7.7,103.3) 17.5) Doctor Employer 1414 65.0 (62.5, 67.5) <0.003 2.1 (1.1, 3.0) <0.003 155.1 (146.8, 163.4) <0.003 Dentist Employee 1410 50.3 (47.7, 52.9) 53.3 (49.6, 57.0) 98.1 (93.0, 103.3) EmployerEmployee 1414 1410 65.0 23.0(62.5, (20.9, 67.5) 25.2) <0.003 66.22.1 (61.0, (1.1, 71.3)3.0) <0.003 106.6 155.1 (88.6, (146.8, 124.5) 163.4) <0.003 Dentist Employer 1414 49.5 (46.9, 52.1) <0.003 1.3 (0.4, 2.1) <0.003 226.5 (214.8, 238.3) <0.003 OptometristEmployee 1410 23.0 (20.9, 25.2) 66.2 (61.0, 71.3) 106.6 (88.6, 124.5) EmployerEmployee 1414 1399 49.5 7.2(46.9, (5.9, 52.1) 8.7) <0.003 64.41.3 (55.0, (0.4, 73.7)2.1) <0.003 120.7 226.5 (102.0, (214.8, 139.4) 238.3) <0.003 Optometrist Employer 1414 31.7 (29.3, 34.1) <0.003 0.7 (-0.1, 1.4) <0.003 278.9 (262.2, 295.6) <0.003 TraditionalEmployee healer 1399 7.2 (5.9, 8.7) 64.4 (55.0, 73.7) 120.7 (102.0, 139.4) Employee 1383 3.4 (2.4, 4.4) NA 196.8 (132.9, 260.8) Employer 1414 31.7 (29.3, 34.1) <0.003 0.7 (-0.1, 1.4) <0.003 278.9 (262.2, 295.6) <0.003 Employer 1414 0.1 (-0.05, 0.3) <0.003 NA NA 100 (21.6, 178.4) 0.167 TraditionalNurse healer EmployeeEmployee 1383 1358 3.4 1.5(2.4, (0.8, 4.4) 2.0) 15.0 (-0.6,NA 30.6) 37.5 196.8 (2.4, (132.9, 75.6) 260.8) EmployerEmployer 1414 1413 0.1 (-0.05, 0.6 (0.2, 0.3) 1.0) <0.0030.04 11.1 (-9.4,NA 31.6) 0.77NA 38 100 (15.9, (21.6, 104.8) 178.4) 0.422 0.167 Nurse

Employee 1358 1.5 (0.8, 2.0) 15.0 (-0.6, 30.6) 37.5 (2.4, 75.6) Employer 1413 0.6 (0.2, 1.0) 0.04 11.1 (-9.4, 31.6) 0.77 38 (15.9, 104.8) 0.422 Table 2: Employer-reported coverage of costs for utilization by employees and their dependants

Farms Units TableCost type 2: Employer-reported Responding n coverage Covering costof costs for utilization byMean, employees range (95%CI) and their dependants % (95%CI) Transportation to medical facility Trips – n Distance travelled – km Farms Employee 1414 92.1 (90.7, 93.5) 2.80, 0–60 (2.57, 3.04) Units 63.36, 0–618 (59.63, 67.08) Cost typeDependant Responding 1414 n Covering 83.9 (82.0, cost 85.9) Mean, range (95%CI) Medical consultation fee % (95%CI) Number of incidents Cost of incidents (NAD) TransportationEmployee to medical facility 1414 62.2 (59.7, 64.8)* 1.14, 0–25Trips (0.96, – n 1.32) 43.43,Distance 0–999 (36.67, travelled 50.19) – km EmployeeDependant 1414 1414 92.1 53.6 (90.7, (51.0, 93.5) 56.2)* 2.80, 0–60 (2.57, 3.04) 63.36, 0–618 (59.63, 67.08) DependantMedications 1414 83.9 (82.0, 85.9) Number of incidents Cost of incidents (NAD) MedicalEmployee consultation fee 1414 48.2 (45.6, 50.8)* 0.61,Number 0–20 of(0.47, incidents 0.75) 26.84, Cost 0–800 of (20.47,incidents 33.20) (NAD) Dependant 1414 41.8 (39.2, 44.4)* EmployeeNAD, Namibian $. 1414 62.2 (59.7, 64.8)* 1.14, 0–25 (0.96, 1.32) 43.43, 0–999 (36.67, 50.19) Dependant*Compared with covering 1414 transportation 53.6 costs, (51.0, signi 56.2)*ficantly different p<0.0003 Medications Number of incidents Cost of incidents (NAD) Employee 1414 48.2 (45.6, 50.8)* 0.61, 0–20 (0.47, 0.75) 26.84, 0–800 (20.47, 33.20) Dependant 1414 41.8 (39.2, 44.4)* NAD, Namibian $. *Compared with covering transportation costs, significantly different p<0.0003

Table 3: Employer-reported on-site presence of trained health professional

Professional training† No. farms reporting Farms healthcare professional % (95%CI) General practitioner 15 1.06 (0.52, 1.59) Registered nurse 72 5.09 (3.95, 6.24) Nursing assistant 23 1.63 (0.97, 2.29) Pharmacist 17 1.20 (0.63, 1.77) Optometrist 3 0.21 (-0.03, 0.45) Paramedic 11 0.78 (0.32, 1.24) Midwife 26 1.84 (1.14, 2.54) Homeopath 2 0.14 (-0.05, 0.34) Other training 15 1.06 (0.53, 1.59) First Aid Certification 237 16.76 (14.81, 18.71) Any trained professional 148 10.47 (8.87, 12.06) †1414 Farms responded.

Table 4: Employer-reported provision of medications and care on-site in farms

Medication provided Dispensed at no cost Dispensed at cost Farms Providing Farms Providing responding medication responding medication (n) % (95%CI) (n) % (95%CI) Employees Cold & influenza 1366 88.9 (87.7, 90.6) 1338 31.5 (29.0, 33.9) Diarrhoea 1365 86.0 (84.2, 87.8) 1336 22.9 (20.7, 25.1) Analgesics 1365 88.9 (87.7, 90.6) 1335 36.5 (34.0, 39.0) Contraception (other 1361 4.7 (3.4, 5.8) 1337 7.8 (6.4, 9.2) than condoms) Minor injury 1364 91.9 (90.5, 93.3) 1337 17.1 (15.4, 19.4) Other 1279 1.9 (1.2, 2.7) 1280 0.4 (0.06, 0.7) Employee dependants Cold & influenza 1253 86.5 (84.7, 89.7) 1232 34.5 (32.0, 37.0) Diarrhoea 1253 83.5 (81.5, 85.4) 1229 25.4 (23.1, 27.6) Analgesics 1252 85.4 (83.6, 87.3) 1232 38.6 (36.1, 41.2) Contraception (other 1252 5.7 (4.5, 7.0) 1231 8.8 (7.3, 10.2) than condoms) Minor injury 1253 89.2 (87.6, 90.8) 1232 19.9 (17.8, 22.0) Other 1175 1.0 (0.9, 2.2) 1175 0.2 (- 0.01, 0.5)

HIV/AIDS 19 farms (43.1%; 95%CI: 29.3, 58.7) the employees paid for these costs themselves. The publicly funded ART program As requested by the AEA, questions regarding an employer’s costs NA$8 per consultation; medications are covered by the knowledge, attitudes and practices for HIV/AIDS were also state. Access to HIV prevention measures on farms was also collected. Employers on 44 farms (3.1%; 95%CI: 2.2, 4.0) limited. Employers on 203 farms (14.3%; 95%CI: 12.2, reported they knew of employees receiving ART for 15.8) reported that they were aware of 1130 employees or HIV/AIDS. Employers on only 8 farms (18.1%; 95%CI: 6.7, their dependants who had gone for HIV testing in the 29.4) covered the costs associated with the publicly 12 months prior to the survey. However, this may not reflect funded ART program for employees with HIV/AIDS; on the actual number of employees or their dependants being

tested, especially regarding the associated sensitivity and willing to pay the full monthly premium for their employees stigma. In addition, the following HIV-related services were (16.0%; 95%CI: 14.1, 17.9) or employee dependants (3.7%; being provided to employees on some farms: condoms 95%CI: 2.7, 4.7), among those who would not pay the full (14.6%; 95%CI: 12.7, 16.4), HIV-specific education (11.3%; premium, willingness to share the cost (co-payment) was 95%CI: 9.7, 13.0), training on condom use (13.1%; 95%CI: much higher (employees: 77.1%; 95%CI: 74.6, 79.4; 11.25, 14.75), and peer education (1.8%; 95%CI: 1.1, 2.5). employee dependants: 38.7%; 95%CI: 36.1, 41.3). Employers were willing to co-pay on average NA$44 While relatively few services were provided to employees on (approximately US$6.30) per employee and NA$39 farms, knowledge regarding HIV by employers was (approximately US$5.60) per dependant per month. comparatively strong. Employers on 89.8% farms (95%CI: Employers also indicated they were prepared to co-pay for 88.2, 91.4) were able to correctly identify two possible an average of 2.8 dependants per employee. Employers on routes of HIV infection, and reject at least three common farms that provided free medicines to employees and their misconceptions about HIV transmission. The most dependants, and those who reported paying for transportation widespread misconception was that HIV can be transmitted to medical care were significantly more likely to be willing by mosquito bites (12.8%; 95%CI: 11.1, 14.5). Also, 20% of to co-pay insurance (2=5.70, df=1, p=0.01; 2=8.84, df=1, employers (95%CI: 17.9, 22.1) did not know that mother-to- p=0.002, respectively). child HIV transmission can be prevented. Discussion Insurance availability and prospects for future investment Access to health care in rural settings is notoriously limited

in Sub-Saharan Africa, where populations are often highly The survey assessed the health insurance coverage reported dispersed5. The low population density means that many by employers for themselves, employees, and their rural Namibians suffer from this lack of access to care. dependants. Just over half the employers reported their farms Commercial farms are one of the largest employers in provided health insurance, covering both in- and outpatient Namibia representing 65 000 employees and employee services, for themselves (57.5%; 95%CI: 54.9, 60.1) and dependants who often live permanently on the farms. their own dependants (53.7%; 95%CI: 51.1, 56.3). Partnerships with the agricultural sector organisations, such Conversely, almost no employers reported health insurance as the AEA, provide an opportunity to improve access to coverage on farms for employees (0.85%; 95%CI: 0.37, care for a large number of people. 1.33) and their dependants (0.50%; 95%CI: 0.13, 0.86).

Farms provided significantly more insurance to employers Currently, the provision of health care, including health and their dependants than for employees and employee insurance, for people living on AEA farms is at the dependants ( 2=1098, df=1, p<0.0003).  discretion of the employer and not a standard policy. From

the survey results, employees are making use of a variety of Employers were asked about their willingness to pay for health services, mostly in the less expensive public sector, private health insurance for employees that would cover only and travelling long distances to do so, bearing most of the out-patient care through a mobile clinic. The hypothetical cost themselves. Employers, by contrast, are consistently monthly cost of this scheme was calculated to be a NA$100 using more private services and travelling even longer premium (approximately US$14.30) per person. This distances. It is possible that employers use private services premium is significantly cheaper than any existing health due to a perceived increase in quality of more expensive insurance product on the Namibian market because it does care. More employers benefit from health insurance not include in-patient care. While few employers were

coverage than employees. The cost of full private health Engaging in public-private partnerships, as in other insurance is prohibitive for the average farm worker who developing countries, could also be utilized in often must also provide for his or her family who also live on Namibia26. With the adoption of health insurance, the use of the farm. While this study found a very small number of the private health sector could increase and significantly employers who provide health insurance for employees, alleviate the burden on the public health sector19,27. Health there is a strong willingness to share insurance costs with insurance could also lead to gains from a healthier workforce employees, which offers an opportunity to improve access to with greater productivity, a reduced need for worker care. replacement, and direct financial gains for the commercial farming sector28-30 . Currently uninsured employers and their Another opportunity identified in this study was that some families (43% of the participants in this survey) could also farm employers supported costs for employees and their benefit from such an insurance scheme. dependants, providing free basic medications and often covering the cost of travel or health consultations. These Other novel approaches to improving access to care, such as employers were more likely to be willing to co-pay for the strengthening of mobile clinics and the introduction of health insurance for employees than employers who did not remote-care communications such as telehealth have been provide this type of support. It may be that these employers shown to work in other rural settings and could be applied in saw value in investing in the health of their employees to Namibia31-33 . A recent meta-analysis found that home-based avoid costs related to absenteeism. Some farms also housed a telehealth has a positive effect on clinical outcomes33. While number of trained health professionals, presumably to telemedicine has not yet been implemented in Namibia, the provide care for farm residents. Although the employment geography of the country suggests a potentially important status of these health professionals was not clear from the role for this technology32. In addition, affordable fingerprint study, their presence represents a resource that could be technology, which has been used in other developing engaged and strengthened to provide basic care on remote countries with success, could be used for identification and farms. medical monitoring33.

In Sub-Saharan Africa there is an increasing lack of This study also identified a significant gap in the provision healthcare workers16, a persistent burden of HIV/AIDS and of basic services for HIV/AIDS prevention, education and an emerging epidemic of chronic disease23. Task shifting, or management. Very few farms supported ART for their delegating tasks from more- to less-specialized health employees and there was no evidence of an established workers, has been proposed as a potential solution to this structure for voluntary counselling and testing for HIV. The problem24,25. The resident, trained health professionals burden of HIV/AIDS is significant in Namibia13 and all identified in this study could potentially be motivated or re- sectors, especially where employees live with their families trained to provide healthcare to employees on a cache of in the workplace, must be engaged to curb the nearby farms. Building the capacity of these professionals to epidemic. Although the study showed that employers had provide care by giving training on relevant public health good knowledge of the transmission of HIV/AIDS, some topics and providing basic diagnostic and treatment tools misconceptions were identified and systematic support or would greatly improve access to health care in these remote education was not being provided for those living and regions, as has been shown in previous studies24. working on their farms. The introduction of health insurance for employees could lower the threshold for seeking HIV- Given the cost of current private Namibian health insurance testing; however, this would need to be supported by the packages, novel approaches must be taken to make cost- employers or the insurance scheme. sharing by employers an affordable and attractive option.

Limitations and biases this study demonstrates, the commercial farming sector in Namibia is a potential partner for improving access through All information collected was from employers only who cost-sharing insurance schemes and leveraging the existing were asked to respond on behalf of employees and employee health services on many farms. The introduction of dependants. This is the greatest limitation to the findings as affordable private health insurance, including HIV/AIDS it presents a risk for information bias, recall bias, and coverage, for commercial farm workers is feasible. This reporting bias. Ideally, employees and their dependants study did not assess the costs associated with health would have been questioned directly in order to limit this insurance; however, such information would be valuable in bias. However, that was outside the scope of this strengthening the case for greater coverage, and future study. Questions were designed to limit speculation by studies would be useful to identify the most cost-effective employers on employee behaviour and instead provide solution. Innovative interventions such as mobile clinics, evidence or counts of events. To avoid recall bias, employers telehealth and the involvement of medical professionals in were asked to report on only the 12 months prior to the task shifting to competent lay people residing on farms administration of the survey. would also increase access and greatly benefit resident workers. In addition, there was a risk of reporting bias if employers gave inaccurate information, either regarding the provision Acknowledgements of health services or the health status of their employees, leading to a possible inflation of the numbers. Because the The authors acknowledge the help and support of the AEA has no specific policy requiring employers to provide members of the AEA and commercial farmers in Namibia health services to employees – and thus there are no without which this work would not have been possible. In repercussions for not providing services – this risk was addition, they thank visiting scientist Huub Gelderblom for considered to be low. his valued assistance and input, by contributing to the

concepts and critically reading the manuscript. While conducting interviews by telephone was deemed to be the most feasible way of reaching the largest number of participants, this may have had an impact on willingness to participate, leading to a selection bias. Ideally, a random sample of employers would References have been used to avoid a systematic selection of any particular kind of respondent. In addition, having AEA-member moderators 1. Dal Poz MR, Kinfu Y, Drager S, Kunjumen T. Counting health conduct the survey may have had an impact on employer workers: definitions, data, methods and global results. Geneva. responses. Again, this approach was taken because t i was the WHO: Department of Human Resources for Health Evidence and most feasible and could provide the largest number of Information for Policy, 2006. responses. It was also potentially beneficial to have employers speaking with colleagues rather than with an ‘outsider’. 2. United Nations. World Population Prospects: the 2008 Revision. Geneva: United Nations Population Division, 2008. Conclusion 3. Mueller KJ, Ortega ST, Parker K, Patil K, Askenazi A. health

status and access to care among rural minorities. Journal of Health The study results underscore the limited access to health care Care of the Poor and Underserved; 10(2): 230-249. for people living in rural communities in Sub-Saharan

Africa, particularly in lower income groups. However, as

4. Raso G, Utzinger J, Silué KD, Ouattara M, Yapi A et al. 13. WHO. HIV/AIDS: Dispatch from Namibia. (Online) 2009. Disparities in parasitic infections, perceived ill health and access to Available: http://www.who.int/hiv/mediacentre/namibia/en/index. health care among poorer and less poor schoolchildren of rural Côte html (Accessed 23 May 2010). d'Ivoire. Tropical Medicine and International Health 2005; 10(1): 42-57. 14. Namibia Ministry of Health and Social Services. Health and Social Services System Review. Windhoek, Namibia: Namibia 5. Tanser F, Gijsbertsen B, Kobus H. Modelling and understanding Ministry of Health and Social Services, 2008. primary health care accessibility and utilization in rural South Africa: An exploration using a geographical information system. 15. UNAIDS and WHO. Sub-Saharan Africa: AIDS epidemic Social Science and Medicine 2006; 63(3): 691-705. update regional summary. Geneva: WHO, 2007.

6.Martínez A, Villarroel V, Seoane J, del Pozo F. Analysis of 16. Cohen J. The great funding surge. HIV/AIDS 2008; 321(5888): information and communication needs in rural primary health care 512-519. in developing countries. IEEE Trans Inf Technol Biomed 2005; 9(1): 66-72. 17. WHO/UNAIDS/UNICEF. Towards Universal Access: scaling up priority HIV/AIDS interventions in the health sector. Geneva: 7. Jutting JP. Do Community-based health insurance schemes WH0, 2008. improve poor people’s access to health care? Evidence from rural Senegal? World Development 2004; 32(2): 273-288. 18. Ranson MK. Reduction of catastrophic health care expenditures by community-based health insurance scheme in Gujarat, India. 8. Wilkinson D, Gouws E, Sach M, Karim SS. Effect of removing Bulletin of the World Health Organisation 2002; 80(8): 613-621. user fees on attendance for curative and preventive primary health care services in rural South Africa. Bulletin of the World Health 19. Sekhri N, Sayedoff W. Private health insurance: implications Organisation 2001; 79(7): 665-71. for developing countries. Bulletin of the World Health Organisation 2005; 83(2): 127-134. 9. Asenso-Okyere WK, Osei-Akoto I, Anum A, Appiah EN. Willingness to pay for health insurance in a developing economy. A 20. Feeley F, de Beer I, Rinke de Wit T, vander Gaag J. The health pilot study of the informal sector of Ghana using contingent insurance industry in Namibia: Baseline Report. Namibia: valuation. Health Policy 1997; 42(3): 223-237. Windhoek, 2006.

10. United Nations Population Division. Population Database. 21. Hohman S. Options and scenarios for HIV/AIDS risk cover for (Online) 2008. Available: http://esa.un.org/ (Accessed 5 May low income employees within NABCOA member companies in 2010). Namibia. Namibia, Windhoek: 2004.

11. The Agricultural Employer's Association. Policy on managing 22. Schellekens O, de Beer I, Lindner ME, van Vugt M, HIV/AIDS within the commercial agricultural sector. Windhoek, Schellekens P, Rinke de Wit T. Innovation in Namibia: preserving Namibia: The Agricultural Employer's Association, 2006. private health insurance and HIV/AIDS treatment. Health Affairs 2010; 28(6): 1799-1806. 12. The World Bank. Health expenditure, total (% of GDP). (Online) 2007. Available: http://data.worldbank.org/indicator/SH. 23. Abegunde D, Mathers C, Adam T, Ortegon M, Strong K. The XPD.TOTL.ZS (Accessed 23 May 2010). burden and costs of chronic diseases in low-income and middle- income countries. Lancet 2004; 370(9603): 1929-1938.

24. Morris M, Chapula B, Chi B, Mwango A, Chi H, Mwanza J et 29. Larson B, Fox M, Rosen S, Bii M, Sigei C, Shaff er D et al. al. Use of task-shifting to rapidly scale-up HIV treatment services: Early effects of antiretroviral therapy on work performance: experiences from Lusaka, Zambia. BMC Health Services Research preliminary results from a cohort study of Kenyan agricultural 2007; 9(5): no pp. workers. AIDS 2008; 22(3): 421-425.

25. Samb B, Celletti F, Holloway J, Van Damme W, De Cock K, 30. Fox M, Rosen S, MacLeod W, Bii M, Foglia G, Sim on J. The Dybul M. Rapid expansion of the health workforce in response to impact of HIV/AIDS on labour productivity in Kenya. Tropical the HIV epidemic. New England Journal of Medicine 2007; Medicine and International Health 2004; 9(3): 318-324. 357(24): 2510-2514. 31. Bagayoko C, Muller H, Geissbuhler A. Assessment of Internet- 26. Widdus R. Public-private partnerships for health: their main based telemedicine in Africa (the RAFT project). Computed and targets, their diversity, and their future directions. Bulletin of the Medical Imaging Graphics 2006; 30(6-7): 407-416. World Health Organisation 2001; 79(8): 713-720. 32. Wootton R. Telemedicine support for the developing world. 27. Feeley F, Connelly P, Rosen S. Private sector provision and Journal of Telemedicine and Telecare 2008; 14(3): 109-114. financing of AIDS treatment in Africa: current developments. Current HIV/AIDS reports 2007; 4(4): 192-200. 33. Ozuah P, Reznik M. The role of telemedicine in the care of children in under-served communities. Journal of Telemedicine and 28. Piot P, Greener R, Russell S. Squaring the circle: AIDS, Telecare 2004; 10(Suppl 1): 78-80. poverty, and human development. PLos Medicine 2007; 4(10): 1571-1575.

55 56 6 ASSESSING THE COSTS OF MOBILE VOLUNTARY COUNSELING AND TESTING AT THE WORK PLACE VERSUS FACILITY BASED VOLUNTARY COUNSELING AND TESTING IN NAMIBIA

57       

        

  

                                                

   

58                                                                                                                                                                                                                                                                                                             

59                                                                                                                                                                                                                                                                                                                                                                                                                        

60                                                                                                                

61                                                                                                                                                           •           •                                                                                                                                                                                                                     

62                                                                                                                                                                               

63                                                                                                                                                                                                                                                                                                                                   

64                                                                                                                                                                                                                                                                                    

65                                                                                                                                                                                                                                                                                                                                                         

66                                                                                                                                                                                                                                                                                                                

67                                                                           

68 7 MOBILE PRIMARY HEALTHCARE SERVICES AND HEALTH OUTCOMES OF CHILDREN IN RURAL NAMIBIA

69      

     

 



                                                                    

                              

70                                                                                                                                                                                                                                                                                                                                   

71                                                                                                                                                                                                                                                                                                                                                                               

72                                                                                                                                                                                                                                                                                                                                    

73     

                                                                                                                         

74      

                                                                                                                                                                                                      

75                                                                                                                                                                                                                                                                                                                

76                                                                                                                                                                                                                                                

77                                                                                                                                                                                                                                                                          

78                                                                                                                                                                                                                                                                                                                                                                                                            

79                                                                                                                                                                                                                                                                                                                                               

80                                                                                                                                                                                                                                                                                                                     

81                                                                             

82 8 INNOVATION IN NAMIBIA: PRESERVING PRIVATE HEALTH INSURANCE AND HIV/AIDS TREATMENT

83 MarketWatch

MarketWatch Innovation In Namibia: Preserving Private Health Insurance And HIV/AIDS Treatment A novel mechanism supported by donors helped this middle-income country subsidize private health insurance premiums and maintain private HIV/AIDS services. by Onno P. Schellekens, Ingrid de Beer, Marianne E. Lindner, Michele van Vugt, Peter Schellekens, and Tobias Floris Rinke de Wit

ABSTRACT: Namibia, a lower-middle-income country in sub-Saharan Africa, suffers from a huge HIV/AIDS burden. An influx of donor funding in 2004–2007 increased support for publicly provided HIV care and treatment. This raised concern that private funding would be “crowded out,” thereby leading to a reduction in the overall resources used to treat patients. In 2006 the Namibian medical aid industry, with donor support, created a special fund to subsidize private health insurance, including HIV/AIDS services. The program allowed both low- and higher-income people to be covered. Crowding out valuable private resources was avoided and the quality of HIV/AIDS services improved. [Health Aff (Millwood). 2009;28(6):1799–806]

amibia is a lower-middle-income health spending is relatively high by African country in Southwest Africa with a standards, as is spending for HIV/AIDS Nyearly per capita income of around (US$65.25).2, 3 US$3,000 (Exhibit 1)—well above the Afri- n HIV/AIDS burden. Namibia has a huge can average.1 However, Namibia has one of the HIV/AIDS burden. AIDS has been the most im- most unequal distributions of income of any portant cause of death in Namibia since 1996.4, 5 country in the world.1 What’s more, health Between 2004 and 2007, disease-specific (ver- outcomes for the poorest one-fifth of the pop- tical) funding—in particular, from the U.S. ulation are considerably worse than those for President’s Emergency Plan for AIDS Relief the richest one-fifth.1 The composition of (PEPFAR)6 and the Global Fund to Fight Namibia’s health spending is atypical in Af- AIDS, Tuberculosis, and Malaria7—increased rica, with a relatively small proportion of out- by approximately US$350 million, of which of-pocket payments and a large contribution most was channeled through the public sys- of government financing and private health tem. HIV/AIDS funding especially increased. insurance.2 At US$167 per year, per capita n Donor funding and crowding out. In

The authors are all affiliated with PharmAccess Foundation in Amsterdam, the Netherlands—the donor supporting the Health Is Vital Risk Equalization Fund, which is the subject of this paper. Onno Schellekens ([email protected]) is managing director. Ingrid de Beer is project director,Namibia. Marianne Lindner is senior policy adviser. Michele van Vugt is senior medical officer. Peter Schellekens is consultant physician. Tobias Rinke de Wit is director of advocacy, technology, and research.

HEALTH AFFAIRS ~ Volume 28, Number 6 1799

EXHIBIT 1 Basic Socioeconomic Indicators In Namibia

Indicator Amount

Populationa 2,031,000 million Gross domestic product (GDP) per capita (US$ 2005)a US$3,016 Gini coefficienta 74.3 Under-5 mortality rate (per 1,000 live births)a 55 (poorest)/31 (richest) HIV prevalence (adults ages 15–49) (2004)b 15% Estimated number of HIV-positive people (2004)b 175,000 Antiretroviral therapy coverageb 22% (2004)/68% (2006) Health spending per capitac US$167 Percent of GDP spent on health carec 4.9% Percent of total health spending (2004) financed throughc Government 69% Private prepaid 24% Out-of-pocket payments 7% Percent of population covered by health insurance (2004)d 12.5%

SOURCES: See below. a United Nations Development Program. Human development indicators. In: Human development report 2007/2008, fighting climate change: human solidarity in a divided world [Internet]. New York (NY): UNDP; 2007 [cited 2009 Aug 26]. Available from: http://hdr.undp.org/en/media/HDR_20072008_EN_Indicator_tables.pdf. b World Health Organization/UNAIDS. Namibia: Epidemiological fact sheet on HIV and AIDS. Geneva (Switzerland): WHO/ UNAIDS; 2008 Sep. c World Health Organization. World health statistics [Internet]. Geneva (Switzerland): WHO; [cited 2009 Apr 16]. Available from: http://www.who.int/whosis d Namibia Medical Aids Federation. Annual conference documentation. Unpublished documentation. Windhoek (Namibia): NAMAF; 2005.

2004 it was anticipated that because of the ex- health and large increases in (disease-specific) pected increases in donor funding for HIV/ donor funding, including Namibia, need to AIDS, the private health sector would experi- take this crowding-out effect into account. ence increased competition from the public Such countries are simply too poor to forgo sector, resulting in crowding-out effects. private contributions for health. Crowding out is the phenomenon whereby new To address the expected crowding-out ef- or expanded public programs meant to cover fects, a dedicated risk pool for HIV/AIDS—the the uninsured have the unintentional effect of Health Is Vital Risk Equalization Fund prompting the privately insured to switch to (HIVREF)—was established in Namibia in the new program. This means that costs in- 2006. Donor funding for HIV/AIDS was chan- curred by the government could be much neled through this fund and used to subsidize higher than expected, and health care im- private health insurance premiums for lower- provements as a result of policy change might income, uninsured people. The PharmAccess not be as robust as they should be. Foundation selected Namibia for this inter- Some strategies to combat the effect of vention because it is a small country with a crowding out are, for example, to limit eligibil- positive government attitude toward innova- ity to the uninsured, subsidize employer- tions in health care financing, a relatively well- based insurance, or apply a premium to fami- established private health insurance sector, lies at higher levels of income eligibility.8–11 and substantial donor funding for HIV/AIDS. Countries with low per capita budgets for The “risk equalization” fund aimed to pre-

1800 November/December 2009

85 MarketWatch

vent those who could pay for private insurance important developments took place in the from seeking HIV/AIDS treatment and care in Namibian medical aid market. Medical aid the expanded public sector by (1) inducing al- schemes’ existing members (higher-income ready insured higher-income groups to ear- populations) were interested in improved mark (part of) their premium specifically for HIV/AIDS coverage, as public-sector care fell privately delivered HIV/AIDS services, and (2) short of their needs. Uninsured lower/middle- using HIV/AIDS donor funds to facilitate the income groups showed interest in basic health introduction of a new lower-cost health insur- insurance packages, including HIV/AIDS ance product (including HIV/AIDS treatment) treatment, at an affordable price. Simulta- for lower-income uninsured groups. This pa- neously, the schemes came under pressure per describes the effects of the donor- from employers and the government to intro- supported risk equalization fund for HIV/ duce lower-cost packages, including HIV/ AIDS in the Namibian medical aid market. AIDS coverage for middle- and lower-income formal-sector workers. The Namibian busi- Before The Fund ness sector joined its forces on HIV/AIDS, add- In 2004, before the risk equalization fund ing to the pressure to find sustainable solu- for HIV/AIDS was established, Namibia’s in- tions for HIV/AIDS in the workplace. surance (”medical aid”) schemes, acting as pri- Namibian trade unions increased their pres- vate not-for-profit entities with voluntary sure on companies to seek sustainable health membership, offered mostly “high-option” care solutions. At the same time, donor fund- products with extensive coverage for both in- ing to the Namibian government through dis- patient and outpatient services. In 2004 al- ease-specific programs started to rise, which most 250,000 Namibians (12.5 percent of the wasexpectedtoleadtoamajorincreaseinac- population) had some form of voluntary pri- cess to HIV/AIDS treatment and care in the vate health insurance.12, 13 Everyone else paid public sector. for health care out of pocket or sought care n How the fund works. Notwithstanding from the public sector, where basic services these dynamics in the medical aid market, the were delivered largely free of charge. schemes initially appeared hesitant to enter Mostly higher- and middle-income formal- the low-cost market because of two main fac- sector workers were insured.14 Annual premi- tors. First, given their relatively small individ- ums ranged between US$684 and US$3,420 ual risk pools and the lack of actuarial data on per person—many times higher than the coun- health risk in the uninsured population, they try’s annual per capita health spending of considered the unknown risk of a new, poten- US$167. Employers paid a large portion of tially large, previously uninsured population workers’ premiums. HIV/AIDS treatment ben- as too high. Second, there was concern that in- efits were generally included in these high-end troducing low-cost products might lead to insurance plans, but spending limits meant “cannibalization” as higher-income, already in- that antiretroviral drugs often had to be pur- sured people might replace their traditional chased out of pocket in the second half of the high-end products with these lower-cost insurance year, or people had to resort to the products, thus jeopardizing the sustainability (donor-funded) public sector. of participating schemes. Most lower-income formal-sector workers The schemes’ reluctance was addressed did not have the resources to prepay for the when, with donor support, the Health Is Vital high-end coverage, even with 50 percent pre- Risk Equalization Fund was established in mium contributions from their employers.15 2006.16 Under the fund, individual medical aid Therefore, this group had to pay out of pocket schemes’ HIV/AIDS risk is shared in a larger or rely on the public sector for general health joint risk pool. This induces economies of care, including HIV/AIDS treatment and care. scale, prevents the exclusion of people with n Paving the way. In 2004 a number of specific conditions (such as HIV/AIDS), and

HEALTH AFFAIRS ~ Volume 28, Number 6 1801

86 MarketWatch

pools resources to improve (perceived) quality the fund combined with a low-cost insurance of care. package (Product 2) covering general health The fund operated as follows. (1) It targeted care (including HIV/AIDS treatment) and the both higher- and lower-income formal-sector provision of a subsidy for the HIV/AIDS part workers. These workers all had monthly in- through the fund would generate demand for comes above the minimum wage (US$135 per Product 2 from lower-income groups. This month). It did not include the poorest of the would result in additional prepaid resources poor, who could not afford to contribute to the for health care that formal-sector workers pay premium. (2) For the already insured higher- for. (3) The larger HIV/AIDS risk pool would income groups, a new product covered HIV/ enable better disease management and data AIDS only, providing enhanced coverage for systems, improving actuarial data and man- HIV/AIDS and improved quality (Product 1). agement practices. The annual premium for this new product was US$35.52 per person in 2006—funded out of What The Fund Achieved existing premiums and not subsidized. People With a total subsidy payment of who bought it generally had a monthly income US$175,000 from PharmAccess in 2007, the above US$650. (3) For uninsured lower- risk equalization fund achieved the following. income workers, a new low-cost general n Increased enrollment. By the end of health insurance product comprising full pri- 2007, only twenty months after the fund’s es- mary and secondary care (including HIV/AIDS tablishment, 34,040 formal-sector workers coverage) was made available (Product 2).17 had enrolled in its two products (Exhibit 2). The average annual premium was US$284 per This included both higher- and lower-income person in 2006. For this product the donor workers. Enrollment constituted 30 percent of provided a yearly subsidy for the HIV/AIDS the part of the medical aid market that was ac- component of US$53.28 per person, for three tually reached by the fund (approximately years. Only workers with monthly income be- 90,000 members), 12 percent of the Namibian low US$650 could buy Product 2. health insurance market (out of approximately To solve the potential problem of cannibal- 296,000 enrollees by the end of 2007), and ization, a one-year waiting period for enroll- about 2 percent of the entire Namibian popu- ment in Product 2 by insured people was im- lation (of two million) (Exhibit 3). posed, which alleviated insurers’ concerns that n Additional prepaid resources. Of the people would switch immediately to the 34,040 people in the fund, nearly 27,000 who lower-cost product. In addition, setting a were already insured paid a total of maximum income eligibility limit of US$650 US$950,000 through Product 1 as earmarked also alleviated concerns. contributions to the fund out of their existing Products 1 and 2 offered the same HIV/ premiums to obtain HIV/AIDS coverage. The AIDS benefits to all, regardless of income or subsidy of US$175,000 thus resulted in US$1.1 risk profiles. These benefits included hospital- million in additional voluntarily prepaid re- ization, medication, laboratory monitoring sources that otherwise would not have been and radiology tests, and counseling and con- raised, paid by 7,283 previously uninsured sultations; the annual limit of the package was lower-income Namibians who bought new approximately US$15,000. lower-cost insurance (Product 2) or an HIV/ n Anticipated effects. The following ef- AIDS-only package (Product 1) (Exhibit 4). fects of the risk equalization fund were antici- These additional resources were 6.5 times pated: (1) Higher-income populations would higher than the original total premium sub- be willing to pay an earmarked portion of their sidy. HIV/AIDS premium into the pool in exchange n Improved quality of care. The quality for enhanced HIV/AIDS coverage and better of HIV/AIDS services measurably improved (perceived) quality. (2) The establishment of because of the risk equalization fund.18 This

1802 November/December 2009

87 MarketWatch

EXHIBIT 2 Enrollment In Health Is Vital Risk Equalization Fund Products In Namibia, 2007

Product 1: Product 2: full product HIV/AIDS only including HIV/AIDS Total

Number of already insured people upgraded to risk equalization fund products 26,757 0 26,757 Number of previously uninsured people enrolleda 4,005 3,278 7,283 Total enrolled in products 30,762 3,278 34,040 Average premium per person per year US$35.52a US$284 –b Average subsidy per person per year None US$53.28 –b

SOURCE: My Health Administrators. Unpublished reports provided to PharmAccess Foundation, 2007. NOTE: Dollar figures are U.S. dollars. a Members of employer groups who could not afford Product 2 could buy Product 1. In that case, the employer paid the full premium, which for this group was US$53.28. b Not applicable. was supported mainly by the establishment of with individual patients (medication adher- a dedicated administration and disease man- ence, adverse effects, toxicities, and determi- agement entity in conjunction with the fund, nation and interpretation of immunological known as MyHealth. Core elements of this ar- and virological progression markers). Key in- rangement were the appointment of HIV case dicators monitored included patients’ CD4+ managers and implementation of a compre- T-cell counts, plasma HIV-1 RNA loads, hemo- hensive database for patient monitoring.19 globin, white blood cell counts, and liver en- The case managers closely followed the im- zymes (Exhibit 5). plementation of a treatment plan that was es- The disease management function thus led tablished by the treating clinicians, who had to to the introduction of standardized treatment adhere to guidelines to be able to claim pay- protocols and data requirements in the private ment. The arrangement allowed for follow-up sector, all in line with the Namibian and

EXHIBIT 3 Uptake Of Health Is Vital Risk Equalization Fund Products In Namibia, 2006–2007 Number taking up products (thousands) 30 Total 25

20 Previously insured

Previously uninsured 5 On HIV therapy 0 1/2006 7/2006 1/2008 7/2008 SOURCE: MyHealth Administrators. Unpublished reports provided to PharmAccess Foundation, 2007.

HEALTH AFFAIRS ~ Volume 28, Number 6 1803

88 MarketWatch

EXHIBIT 4 Additional Resources Prepaid As A Result Of The Risk Equalization Fund In Namibia

Annual premium Total annual premium Number enrolled per person (US$) payments (US$)

Already insured people who bought Product 1 26,757 35.52 950,000 Previously uninsured who bought Product 1 4,005 53.28 213,000 Previously uninsured who bought Product 2 (full product) 3,278 284.00 931,000 Total previously uninsured (Products 1 + 2) 7,283 53.28/284 1,144,000 Premium subsidya (Product 2) 3,278 53.28 175,000 Net additional premium contributions 7,283 53.28/284 970,000

SOURCE: PharmAccess Foundation. Calculations based on My Health enrollment and premium data, 2007. NOTE: Product 1 covers HIV/AIDS only. a Available only to the 3,278 previously uninsured people who bought Product 2.

World Health Organization guidelines for favorably with the international literature on quality of care. antiretroviral therapy in Africa.20 At the individual patient level, the quality of Furthermore, more people undertook vol- care improved greatly. At entry in the insurance untary counseling and testing: 7,634 had been scheme, about 30 percent of patients (n = 516) tested by the end of 2007.21 This led to in- suffered from a severe loss of resistance (CD4 creased detection and uptake of HIV-positive counts below 200 per microliter of blood). Af- patients (1,736 by the end of 2007). Neverthe- ter a year and a half, the resistance of more less, the percentage of HIV-positive people than 75 percent of them had recovered; they within the fund-insured population was still were out of the danger zone, albeit still not about three times lower than reported figures completely healthy. These patients, although of for Namibia. Hence, further efforts are needed course still needing lifelong treatment, thus re- to detect more HIV-positive people within the gained a “normal” life. These results compare population reached by the fund.

EXHIBIT 5 Health Effects Of The Health Is Vital Risk Equalization Fund In Namibia, 2002–2007 Number receiving laboratory services and visits (thousands) 2.5 Hemoglobin 2.0 ALAT (indicator of liver disease) VL (plasma HIV-1 RNA load) 1.5 CD4+ cells Number of visits to clinics 1.0

0.5

0.0 2002 2003 2004 2005 2006 2007 SOURCE: MyHealth Disease Management Program. Unpublished data provided to PharmAccess Foundation, 2007. NOTES: These results demonstrate improved highly active antiretroviral therapy (HAART) laboratory monitoring, with increased frequencies of biochemical tests for HAART toxicities, while at nearly every visit a CD4 count and plasma HIV-1 RNA load test was performed, on average twice per year, in line with current regulations. The risk equalization fund began operations in 2006.

1804 November/December 2009

89 MarketWatch

Finally, the administration and disease prevent crowding out, likely relieving the bur- management arrangement generated actuarial den on the public sector. The intervention also data on service use, treatment costs, and en- demonstrates that through such a mechanism, rollment and retention of insured people. The quality of care can be improved. average annual cost of treating a patient was Despite the program’s early success in mo- approximately US$850. The fund’s adminis- bilizing additional voluntarily prepaid private trative costs were approximately US$10 per funds through a donor-supported risk equal- person per year.22 ization fund, the willingness of the target n Success factors, scalability, and groups to enroll has been increasingly affected replicability. Critical success factors of the by the expected improvements in the availabil- Namibian intervention include government ity of HIV/AIDS services in the public sector, policy that supports a role for the private sec- leading to crowding out and reversing the tor in health and a willingness to accept subsi- gains made. dies for those in higher income brackets. Ade- Donors should take these points to heart. quate private delivery structures are also Donors need to realize that large donor fund- needed to ensure that the quality of private ing for health can have detrimental effects care is at least as good as that of public care. such as these. In poor countries, where a This is relevant because private voluntary means-tested mandatory contribution for schemes function only if a product offers more health care cannot be enforced, a strategy benefits than the cost of membership, a prod- should be applied that avoids crowding out uct compares favorably with alternatives, and and thus includes voluntary private insurance the perceived quality of medical services deliv- and risk equalization. The Namibian experi- ered is competitive. In particular, willingness ence shows that this is possible. to prepay depends to a great extent on the (perceived) quality of the care received.23–25 The current research was funded by a grant from the Further scaling up of this innovation in Netherlands Postcode Loterij, through STOP AIDS Namibia would require higher premium subsi- NOW! and its subsidiary, the Dutch AIDS Fonds. The dies for a longer period of time. Experience study sponsors had no involvement in the study design, demonstrated that the prevailing insurance data collection, analysis, or interpretation of the data fees for low-cost products were still too high or the decision to submit the paper for publication. for many lower-income workers.26 Investment Actuarial data for the current paper were kindly in insurers and providers focusing on low-cost provided by MyHealth, the disease management primary care services would also be required. organization of the Health Is Vital Risk Equalization A potential risk factor in the scaling up could Fund (HIVREF) in Namibia. The grant supporting the be the lack of a sufficiently strong regulatory current study was also used to (partly) support the framework, as widely described in the litera- premiums of affordable medical aid (insurance) ture.27 Finally, replication of the Namibian ex- products for individuals in Namibia. All novel perience to other, even poorer countries would Namibian affordable medical aid products that include require higher subsidies and greater invest- HIV/AIDS coverage received subsidies. No conflict of ment in private health structures. interest has occurred. The Possible And The Reality NOTES This intervention demonstrates that a risk 1. United Nations Development Program. Human equalization mechanism can raise additional development indicators. In: Human development prepaid resources for health that are several report 2007/2008. Fighting climate change: hu- times higher than the total subsidy provided, man solidarity in a divided world [Internet]. financed by higher-income populations. It New York: UNDP; 2007 [cited 2009 Aug 26]. shows that in a country with sizable donor Available from: http://hdr.undp.org/en/media/ HDR_20072008_EN_Indicator_tables.pdf funding for HIV/AIDS, such a mechanism can

HEALTH AFFAIRS ~ Volume 28, Number 6 1805

90 MarketWatch

2. World Health Organization. World health statis- for HIV/AIDS risk cover for low-income employ- tics [Internet]. Geneva: WHO; [cited 2008 Oct ees within NABCOA member companies in 26]. Available from: http://www.who.int/whosis/ Namibia. Report on the feasibility study, October data 18–November 25, 2004. Windhoek: Namibia 3. Republic of Namibia, Ministry of Health and So- Business Coalition on AIDS; 2004. cial Services. United Nations General Assembly 16. The donor is PharmAccess Foundation, a Dutch Special Session (UNGASS) country report. Re- not-for-profit organization engaged in strength- porting period April 2006–March 2007 ening health systems in sub-Saharan Africa. The [Internet]. Windhoek: Ministry of Health and PharmAccess project supporting the Namibian Social Services; [cited 2009 Aug 26]. Available innovation is called the Okambilimbili project from: http://data.unaids.org/pub/Report/2008/ (which, in the Oshivambo language, means “but- namibia_2008_country_progress_report_en.pdf terfly,” as a sign of positive change). 4. World Health Organization/UNAIDS. Namibia: 17. Members of employer groups who could not af- epidemiological fact sheet on HIV and AIDS. ford Product 2 could buy Product 1. In that case, Geneva: WHO/UNAIDS; 2008 Sep. the employer paid the full premium. 5. Feeley F, DeBeer I, Rinke de Wit T, vander Gaag 18. This concerns quality improvement within the J. The health insurance industry in Namibia: private system. A comparison of the quality and baseline report [Internet]. Boston: Boston Uni- price of public care is not available. versity, Center for International Development; 19. The donor, PharmAccess, provided external Amsterdam Institute for International Develop- quality assurance through the database. ment; and PharmAccess Foundation; 2006 Jun 20. Keiser O, Anastos K, Schechter M, Balestre E, [cited 2009 Aug 26]. Available from: http://sph Myer L, Boulle A, et al. Antiretroviral therapy in .bu.edu/cihd/images/stories/scfiles/cih/final_ resource-limited settings 1996 to 2006: patient namibia_insurance_situation__june_2006.pdf characteristics, treatment regimens, and moni- 6. United States President’s Emergency Plan for toring in sub-Saharan Africa, Asia, and Latin AIDS Relief [Internet]. Washington: PEPFAR; America. Trop Med Int Health. 2008;13(7):870–9. [cited 2009 Feb 16]. Available from: http://www 21. Unpublished data from the MyHealth disease .pepfar.org management program, provided to PharmAccess 7. Global Fund to Fight AIDS, Tuberculosis, and Foundation, 2007. Malaria [Internet]. Geneva: Global Fund; c2009 22. Unpublished data from the Health Is Vital Risk [cited 2009 Feb 16]. Available from: http://www Equalization Fund, provided to PharmAccess, .theglobalfund.org 2009. 8. Cutler DM, Gruber J. The effect of Medicaid ex- 23. Carrin G. Community-based health insurance pansions on public insurance, private insurance, schemes in developing countries: facts, problems, and redistribution. Am Econ Rev. 1996;86(2): and perspective. Geneva: WHO; 2003. 378–83. 24. Preker AS, Carrin G, Dror D, Jakab M, Hsiao W, 9. Kronick R, Gilmer T. Insuring low-income Arhin-Tenkorang D. Effectiveness of community adults: does public coverage crowd out private? health financing in meeting the cost of illness. Health Aff (Millwood). 2002;21(1):225–39. Bull World Health Organ. 2002;80(2):143–50. 10. van der Gaag J, Stimac V. Towards a new para- 25. Litvack J, Bodart C. User fees plus quality equals digm for health sector development. Amsterdam: improved access to health care: results of a field Amsterdam Institute for International Develop- experiment in Cameroon. Soc Sci Med. 1993;37 ment; 2008 Sep. (3):369–83. 11. World Bank. Better health in Africa, experience 26. Asfaw A, Gustafsson-Wright E, van der Gaag J. and lessons learned. Washington: World Bank; Willingness to pay for health insurance: an anal- 1994. ysis of the potential market for new low-cost 12. MyHealth Administrators. Unpublished reports health insurance products in Namibia. Amster- provided to PharmAccess Foundation. Wind- dam: Amsterdam Institute for International De- hoek: MyHealth Administrators; 2007. velopment; forthcoming. 13. Namibia Medical Aids Federation. Conference 27. See, for example, Jack W. Health insurance re- management report. Windhoek: NAMAF; 2005. form in four Latin-American countries: theory 14. This paper discusses only private formal-sector and practice. Washington: World Bank; 2000. workers. It does not discuss civil servants, who are insured under a separate arrangement. 15. Hohmann J, Skolnic RA. Options and scenarios

1806 November/December 2009

91 92 9 a INCIDENCE OF HIV IN WINDHOEK, NAMIBIA: DEMOGRAPHIC AND SOCIO-ECONOMIC ASSOCIATIONS

93 Incidence of HIV in Windhoek, Namibia: Demographic and Socio-Economic Associations

Marielle Aulagnier1*, Wendy Janssens2,6, Ingrid De Beer3, Gert van Rooy4, Esegiel Gaeb5, Cees Hesp1, Jacques van der Gaag2,6,7, Tobias F. Rinke de Wit1,8 1 PharmAccess Foundation, Amsterdam, The Netherlands, 2 Department of Development Economics, VU University Amsterdam, Amsterdam, The Netherlands, 3 PharmAccess Foundation Namibia, Windhoek, Namibia, 4 Multi-disciplinary Research Center, University of Namibia, Windhoek, Namibia, 5 Namibia Institute of Pathology, Windhoek, Namibia, 6 Amsterdam Institute for International Development, Amsterdam, The Netherlands, 7 Brookings Institution, Washington, D.C., United States of America, 8 Department of Global Health, Academic Medical Center of the University of Amsterdam, Amsterdam Institute for Global Health and Development, Amsterdam, The Netherlands

Abstract

Objective: To estimate HIV incidence and prevalence in Windhoek, Namibia and to analyze socio-economic factors related to HIV infection.

Method: In 2006/7, baseline surveys were performed with 1,753 private households living in the greater Windhoek area; follow-up visits took place in 2008 and 2009. Face-to-face socio-economic questionnaires were administrated by trained interviewers; biomedical markers were collected by nurses; GPS codes of household residences were recorded.

Results: The HIV prevalence in the population (aged.12 years) was 11.8% in 2006/7 and 14.6% in 2009. HIV incidence between 2007 and 2009 was 2.4 per 100 person year (95%CI = 1.9–2.9). HIV incidence and prevalence were higher in female populations. HIV incidence appeared non-associated with any socioeconomic factor, indicating universal risk for the population. For women a positive trend was found between low per-capita consumption and HIV acquisition. A HIV knowledge score was strongly associated with HIV incidence for both men and women. High HIV prevalence and incidence was concentrated in the north-western part of the city, an area with lower HIV knowledge, higher HIV risk perception and lower per-capita consumption.

Discussion: The HIV incidence and prevalence figures do not suggest a declining epidemic in Windhoek. Higher vulnerability of women is recorded, most likely related to economic dependency and increasing transactional sex in Namibia. The lack of relation between HIV incidence and socio-economic factors confirms HIV risks for the overall urban community. Appropriate knowledge is strongly associated to lower HIV incidence and prevalence, underscoring the importance of continuous information and education activities for prevention of infection. Geographical areas were identified that would require prioritized HIV campaigning.

Citation: Aulagnier M, Janssens W, De Beer I, van Rooy G, Gaeb E, et al. (2011) Incidence of HIV in Windhoek, Namibia: Demographic and Socio-Economic Associations. PLoS ONE 6(10): e25860. doi:10.1371/journal.pone.0025860 Editor: Abdisalan Mohamed Noor, Kenya Medical Research Institute - Wellcome Trust Research Programme, Kenya Received May 3, 2011; Accepted September 13, 2011; Published October 4, 2011 Copyright: ß 2011 Aulagnier et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This work was supported by a grant from the Netherlands Ministry of Foreign Affairs (DSI/SB, project number 13298). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected]

Introduction This was greatly facilitated by the rapid scale up of vertical donor funding (PEPFAR and GFATM) which resulted in 28.5% of the Namibia, with a population of approximately 2.2 million is 2008/2009 Total Health Expenditure of N$4,945/US$706 classified as a middle income African country; although income million being spent on HIV/AIDS. 51% of this amount is donor disparity is one of the largest in the world [1]. In 2004 more than funded [5]. Recent evidence suggested that national HIV 38% of the population was living below the poverty line [2]. prevalence in Namibia started to decline or stabilize since 2002 With a reported national prevalence of 15.3% in the adult [6]. However, since prevalence is still high, fighting the HIV population of 15–49 years [3], Namibia is experiencing one of the epidemic requires continuous vigilance. Namibia is facing rapid largest HIV epidemics in Africa. Since 1996, AIDS has been the urbanization, especially towards the capital city of Windhoek [7], leading cause of death and contributed to a drop in life expectancy which might exacerbate the HIV epidemic [1,8–10] as economic from 65 to 61 years, between 1990 and 20081. Over the past two inequality is heightening in city areas and has been found to be decades, the Government of Namibia has prioritized HIV and highly associated with sexually transmitted diseases and HIV [11]. AIDS in its development undertakings. This resulted, amongst Ensuring population’s good health presents many challenges others, in impressive increase in access to antiretroviral treatment within the complex urban context of developing countries [9,12– with a .75% coverage according to the new WHO guidelines [4]. 14]. This paper presents HIV incidence and prevalence estimates

PLoS ONE | www.plosone.org 1 October 2011 | Volume 6 | Issue 10 | e25860

94 Incidence of HIV in Windhoek-Namibia from household surveys conducted in Windhoek-Namibia between Biomedical markers. Biomedical markers were collected by 2006/7 and 2009. Associations with HIV incidence are explored qualified nurses. Anonymous HIV tests were performed on with demographic and socio-economic factors, which may participants aged 12 years or older who(se parents) provided contribute to guiding future prevention and control efforts in the informed consent. HIV screening was performed on non-invasive capital city of Namibia. oral fluid samples collected with OraSureH HIV-1 Oral Specimen Collection Device (OraSure Technologies, Inc., Bethlehem, PA). Methods Samples were labelled with a barcode that corresponded to the barcode on the survey questionnaires and shipped to the Namibia Ethics Statement Institute of Pathology (NIP). At NIP a HIV test was performed The study was approved by the Research Ethics Committee at using the Oral Fluid Vironostika HIV Uni-Form II Micro-ELISA the Namibian Ministry of Health and Social Services (MOHSS). (bioMe´rieux Inc., Durham, NC). This procedure had previously Anonymity of HIV test results was safeguarded through specific been validated and approved by the MOHSS for use in bio-medical protocols that linked the results to the household anonymous epidemiological surveys [16]. HIV test results from survey dataset using anonymous identification numbers. Partici- NIP could thus be linked to the household survey forms but not to pants’ names and addresses were kept separate from the data. particular individuals, since the list of their names and Individuals who wished to know their test results were referred to corresponding barcodes was kept separate at MRCC and existing VCT centers in Windhoek, since individual HIV results combination of information with NIP was not possible. HIV test cannot be based on a single test as performed during the current results were not shared with survey participants; those who wished surveys. to know their HIV status were referred to the appropriate facilities for HIV testing according to the national protocol. Study design and target population Household socio-economic status and welfare. Household In 2004, a project was started in Namibia named ‘‘Okambi- consumption was measured to investigate household socio- limbili’’ (‘‘Butterfly’’) which stimulated the development and economic status and welfare. Annual household consumption was implementation of low-cost voluntary basic health insurances for calculated by summing weekly food consumption multiplied by 52, low income workers, by temporarily subsidizing medical aid monthly consumption of regular non-food items multiplied by 12 (health insurance) premiums for low income workers and creating and annual consumption of remaining non-food items in each a risk-equalization fund for HIV and AIDS [15], thus improving household. Subsequently, per capita consumption was calculated by access to affordable health care and reduction of risk of dividing annual household consumption by the number of members catastrophic healthcare expenditures. In order to evaluate the in each household [17]. Individuals were assigned to a consumption project’s impact, large-scale household surveys were implemented category based on their per capita consumption level in comparison in a random sample of the greater Windhoek area in 2006/7 and to the poverty line (264 N$/38US$ per month in the 2003 Namibia re-conducted in 2008 and 2009. These surveys measured health Household Income and Expenditure Surveys [2]), corrected for indicators, biomedical markers and socio-economic characteristics inflation. The poverty line was calculated at 302 N$/43 US$ per in Namibian households. month in 2006 and 373 N$/53US$ per month in 2009. The The sample design for the original survey was a representative consumption categories were defined as follows: 1) below poverty stratified two stage probability sample. The first stage consisted of line, 2) between poverty line and twice the poverty line, 3) between geographical areas (primary sampling units (PSU), as defined for twice and three times poverty line and 4) more than three times national census purposes) from Windhoek and the second stage poverty line. consisted of randomly selected households in these areas. Target HIV knowledge and risk perception score. An HIV population consisted of private households in the greater Wind- knowledge score was constructed by summing all questions on hoek area, excluding people in hospitals, hostels and prisons. HIV knowledge in the questionnaire (see Appendix S1 for more Surveys were conducted with the same households in 2006/7, information). Correct responses were scored 1 and incorrect 2008 and 2009. Households that had relocated their residence responses 0. The score ranged from 0 to 11. The personal HIV between surveys were traced wherever possible; field workers risk perception score was constructed using the following question: revisited the households a minimum of 3 times to contact ‘‘what do you think your chances are of getting HIV/AIDS?’’. absentees. The response score were No risk at all (0), Small (1), Moderate (2), High (3), Don’t know (4). The individuals who refused to answer Data collection were eliminated from the analyses. Socio-economic questionnaire. Face-to-face questionnaires were administrated by trained interviewers from the Multi- Data analysis disciplinary Research and Consultancy Centre (MRCC) of the HIV incidence. Incidence rate was defined as number of new University of Namibia, after household consent. The demographic HIV infections per 100 person-years (PY). Person-years were and socio-economic questionnaire collected information on based on all participants aged 12 years or older in 2006/7 gender, age, education level, employment status, income, uninfected at baseline and followed up until round 3 of the survey household composition, housing characteristics, household (Figure 1). Individuals who acquired HIV infection were assumed consumption, etc. In addition the survey collected specific to have been infected halfway through the observation period. The information on health (preventive and reproductive health, calculation does not account for the window period of infection. chronic diseases, illness and injury, hospitalizations), healthcare This will not affect the estimates assuming a constant HIV spending and health behaviour (alcohol consumption, smoking, incidence rate over the years 2007–2009. contraception use, age at first sexual intercourse, HIV testing, etc.). Factors associated with HIV status and geographical Information regarding children under 12 years old was obtained analysis. Simple logistic regressions were used to study from a household adult; information from adolescents 12–18 years demographic and socio-economic factors associated with HIV old was collected by direct interviews after consent from their incidence separately in males and females, older than 18 years. parents. Variables with p,0.15 were introduced into backward multiple

PLoS ONE | www.plosone.org 2 October 2011 | Volume 6 | Issue 10 | e25860

95 Incidence of HIV in Windhoek-Namibia

Figure 1. Follow up of individuals aged 12 years or older, who participated to the baseline survey and were seronegative at baseline (2007). doi:10.1371/journal.pone.0025860.g001 logistic regression (exit threshold p.0.15). Statistical analyses were years or older did work during the past 12 months. Nearly one fifth performed with SPSS, version 13.0 (SPSS Inc., Chicago, 2002). (19.6%) of the individuals declared a consumption level below the We considered that variables were significant at p, = 0.05 and national poverty line (302 N$/month) and 22.8% subscribed for a that 0.05,p,0.10 indicated a trend at the limit of significance. health insurance policy in 2006. For each PSU three sets of latitude/longitude coordinates were recorded using hand-held GPS devices. The geographical centre HIV knowledge, attitudes, practices and risk perception of each PSU was calculated by averaging its latitudes/longitudes to Among individuals who declared to ever have had sexual obtain the arithmetic mean. Subsequently, data was fed into intercourse, the mean age for first sexual intercourse was 18.8 QlikView 9.0 (Qliktech, Radnor, Pennsylvania) and superimposed years (sd+/23.5). The mean age for sexual debut was 19.2 years on Google Maps (Google Inc, Mountain View). Each bubble (sd +/23.4) in females and 18.6 years (sd +/23.7) in males represents a PSU with the size of the bubble proportional of the (p = 0.02). In the population aged 15 years and older in 2006, scores. Because of the small incident HIV sample in each PSU, the 32.4% were living in couple (married or consensual union) and incidence map is not shown in the current manuscript. 64.6% were never married. Nearly half of the individuals (47.1%) perceived themselves to Results be not at risk for HIV infection in 2008, 14.3% at low risk, 7.8% at moderate risk and 4.5% at high risk. Younger people (,25 years) Socio-economic characteristics were more likely than older ones to view themselves as being not at The original baseline survey sample aimed at 2,000 households risk (50.0% vs 45.6%, p,0.001). The mean score for HIV and realized a voluntary participation of 1,753 (88%) in the socio- knowledge was 5.19 (sd+/24.7). This score was significantly economic part of the questionnaire in 2006. The households were higher in population aged 25 years or older than in the younger again visited in 2008 (1,154 households i.e. 66% of baseline population (5.6 vs 4.5, p,0.001). household participants) and 2009 (861 households i.e. 75% of 2008 household participants). Three years after baseline, 56.5% of the males and 57.7% of the females still participated. Attrition was HIV incidence mostly due to mobility of the target population and refusal to again The incidence analysis includes all individuals aged 12 years or participate. Eight hundred and sixty one (861) households older who tested HIV negative at baseline in 2006/2007 and for representing 3,168 individuals aged 12 years or older in 2006 whom HIV status was known in 2009. Where relevant, HIV participated in all three rounds of survey. The mean household positive status was prospectively extrapolated for missing test size was 4.8 persons in 2006 (Table 1). Households lost to follow- results in future survey rounds. HIV negative status was up were slightly younger, more affluent and with higher education retrospectively extrapolated for missing test results in prior survey level (data not shown). rounds. Twenty-two individuals tested positive in 2006/7 with Among the sample, 51.8% were females and the mean age in borderline OD-values , 2 times the cut-off point and tested 2006 was 30.3 years (standard deviation (sd) +/213.3). Nearly negative in the second independent HIV test. The results of the 10% of individuals of school-age (.5 years) never attended school, second test were used in this analysis. This yields a baseline sample 12.0% had a pre-school or primary level, 75.4% a secondary level of 2,659 HIV negative individuals. For 1,543 of the initially HIV and 3.1% a higher level. About half (47.4%) of the people aged 15 negative people, HIV status is known in 2009 (Figure 1).

PLoS ONE | www.plosone.org 3 October 2011 | Volume 6 | Issue 10 | e25860

96 Incidence of HIV in Windhoek-Namibia

Table 1. Socio-economic characteristics of individuals ($12 years) who participated in the 3 rounds of surveys (n = 3,168) and comparison with the Namibian population [44].

Category Subcategory Population (. = 12 years) Namibian population (DHS 2006/07)

Household size (mean) 4.8 4.5 Gender Female 51.8% 53.0% Age (mean) 30.3 (+/213.3) NA Females (.15 y) Males (.15 y) Females (15–49 y) Males (15–49 y) Marital status Never married 65.1% 64.2% 57.9% 65.0% Married/Union 30.2% 34.7% 35.2% 30.8% Divorced/separated 1.8% 0.9% 4.4% 3.9% Widowed 2.9% 0.1% 2.6% 0.3% Females* Males* Females (20–49 y) Males (20–49 y) Age at first sexual intercourse (median) 19.0 18.0 18.9 18.0 Females (.5 y) Males (.5 y) Females (.5 y) Males (.5 y) School grade None 7.9% 11.1% 14.6% 15.6% Primary 9.5% 14.7% 41.6% 44.7% Secondary 79.8% 70.7% 37.5% 33.3% Higher 2.8% 3.5% 4.7% 4.9% Employed during the past year (.15 years) 47.4% NA Per capita consumption level ,302 N$/month 19.6% NA 302–603 N$/month 26.3% 604–905 N$/month 15.9% $906 N$/month 38.2% Covered by health insurance policy 22.8% NA Personal HIV perception risk, 2008 No risk at all 47.1% NA Small 14.3% Moderate 7.8% High 4.5% Don’t know/refused 26.4% HIV knowledge score, 2009 (mean) 5.2 (+/24.7) NA

*Among individuals aged .12 years who had sexual intercourse; NA: Not available. doi:10.1371/journal.pone.0025860.t001

Incidence estimates were based on a total of 3,858 person-years 15.6]) and concerned more frequently people aged 25 years or older of follow up on people aged 12 years or older in 2006. During the (Table 2). For females, HIV prevalence peaks in the 25–39 age- follow-up period, 93 individuals had acquired HIV infection by group but for men it continues to be high after 40 years old. 2009, resulting in an overall incidence rate of HIV infection of 2.4 per 100 PY (95%CI = [1.9–2.9]). The incidence rate was lower in Factors associated with HIV incidence male populations (1.6; 95%CI = [1.0–2.1] per 100 PY) than in For adult women (.18 years old), very few socio-economics female population (2.6; 95%CI [2.0–3.3] per 100 PY) (Table 2). factors were significantly associated with HIV incidence in HIV incidence was lower in the youngest age group (12–24 years) univariate analyses (Table 3): younger age at 1st sexual intercourse among both sexes: 0.8 per 100 PY in males and 1.9 per 100 PY in (p = 0.07; borderline significant) and lower consumption level females. The mean age of women who became HIV positive by (p = 0.01). There was no evidence for difference in HIV incidence 2009 was 31.8 versus 36.6 in men population (p = 0.06). by marital status, school grade, employment status, health insurance subscription and number of sexual partners declared HIV prevalence in the past year. HIV acquisition was significantly less frequent In 2006/7, the overall rate of HIV-infection was 11.8% among among women with higher HIV knowledge score (p,0.001). the participant aged 12 years or older and 13.5% among Multiple logistic regression indicated that only HIV knowledge participants aged 15–49 years. In 2009, of the 3,787 individuals score remained associated with incident HIV (p,0.001). Trends aged 12 years or older who participated in the survey 2,119 with lower consumption level were observed (p = 0.12) (Table 3). accepted to perform an HIV test. The overall HIV prevalence was Among adult men, there was little evidence that HIV acquisition 14.6% (95%CI = [13.1–16.1]) in this population and 15.8% was associated with any socioeconomic factor. Only personal HIV (95%CI = [14.1–17.6]) in the population aged 15–49 years risk perception (p = 0.08; borderline significant) and HIV knowl- (Table 2). edge score (p,0.001) were associated with HIV incidence. In The HIV prevalence was higher among females (15.6%; multiple logistic regression, HIV knowledge score remained the only 95%CI = [13.5–17.7]) than among males (13.4%; 95%CI = [11.2– significantly association with HIV acquisition (Table 3).

PLoS ONE | www.plosone.org 4 October 2011 | Volume 6 | Issue 10 | e25860

97 Incidence of HIV in Windhoek-Namibia

Table 2. HIV incidence and HIV prevalence estimates by age and sex.

INCIDENCE1 2007–2009 PREVALENCE2 2009

N HIV incidence (per 100 PY) 95%CI N1 HIV prevalence (%) 95%CI

Population . = 12 years 1543 2.4 [1.9–2.9] 2119 14.6 [13.1–16.1] Males 760 1.6 [1.0–2.1] 947 13.4 [11.2–15.6] 12–24 years 319 0.8 [0.2–1.4] 377 6.6 [4.1–9.2] $25 Years 441 2.2 [1.3–3.1] 570 17.9 [14.8–21.1] Females 958 2.6 [2.0–3.3] 1172 15.6 [13.5–17.7] 12–24 years 368 1.9 [1.0–2.7] 401 6.2 [3.9–8.6] $25 Years 590 3.1 [2.2–4.0] 771 20.5 [17.7–23.4] Population 15–49 years 1290 2.4 [1.9–3.0] 1623 15.8 [14.1–17.6]

PY: Person-years; CI: Confidence interval; 1Sample population: individuals aged 12 years or older in 2007, who participated to the baseline survey and were seronegative at baseline. 2Sample population: persons who participated to the survey in 2009 and accepted to perform a test. doi:10.1371/journal.pone.0025860.t002

Geographic distribution Recent evidence suggests that HIV prevalence is declining since The geographic distribution per PSU of HIV prevalence, HIV 2002 in Namibia [6]. However, the current survey in Windhoek does personal risk perception, HIV knowledge score and mean per- not substantiate this finding. In fact, the Windhoek HIV prevalence capita consumption in the Greater Windhoek Area in the year appears to remain stable: in the order of 15%. It is possible that with a 2009 are represented in Figures 2a–d. It is noticeable that high .75% coverage of HIV patients by antiretroviral therapy in HIV prevalence is concentrated in the north-western part of the Windhoek, more HIV positive patients remain alive and thus city (Figure 2a). These areas are also characterized by smaller HIV contribute to increased overall HIV prevalence estimations. knowledge scores (Figure 2b), higher personal HIV risk perception The data presented in this manuscript are collected through (Figure 2c), and lower per-capita consumption (Figure 2d). household surveys, a methodology that was accepted by the WHO to be more representative of a population [30]. Careful analysis of Discussion our survey data led to the observation that individual surveyors can contribute to data aberrations, but these cases were excluded This is the first paper reporting on HIV incidence in Windhoek, from the current analyses [31]. Therefore, it can be stipulated that Namibia as determined by three consecutive household surveys the current Windhoek HIV incidence and prevalence data are the (2006–2009). A high HIV incidence was observed in Windhoek in most recent and most representative to date. both sexes: 2.6 per 100 PY in females and 1.6 per 100 PY in males. With respect to the representativeness of the Windhoek data, The female HIV incidence in Windhoek appeared comparable to several factors should be taken into consideration that could data reported from urban adult population in Zambia and introduce biases. Most importantly, data may be influenced by the Zimbabwe but lower than data reported in Malawi or South attrition rate of 43% over two years. However, very similar attrition Africa [18–20]; for men the HIV incidence was lower than the rates are observed in other three year follow-up surveys in African data observed in the only one incidence survey performed in adult settings [32–35] and therefore our data should be comparable to the urban men in Kenya [21]. literature. In line with the literature [33], non-response rates for The overall HIV prevalence in this study for the age group HIV testing in the sample tends to be higher among the wealthier, 15–49 years was 15.8% (95%CI = [14.1–17.6]). This is similar better educated and younger households. The wealthier and higher to the most recent statistics published by WHO for Namibia educated population in Windhoek has lower HIV prevalence [36]. (15.3% in 2007) in the same age group [3] but higher than the Therefore HIV incidence and prevalence might be overestimated. latest estimation [22] published recently for 2008/2009 Previous research has indicated that correction for non-response (13.3%). Moreover, the HIV prevalence data in Windhoek based on such observed characteristics tends to have small and confirm the general observation that women acquire HIV insignificant effects on HIV prevalence estimates [37–38]. In infection on average 5 years earlier than men and that contrast, recent evidence in Namibia suggests that the bias due to prevalence in younger women (#25 years) is higher than in unobserved characteristics may be significant and that non-respon- the corresponding male age group [23]. It is underscored that dents may be two to three times more likely to be infected than age and sex differentials in distribution of HIV infection remain respondents [39]. key drivers of generalized epidemics, highlighting the vulner- Thus, on the one hand the attrition of the relatively more ability of young women. One other reason of HIV proliferation affluent survey participants may lead to upward estimations of in women population might be the increasingly acceptable form overall population HIV incidence and prevalence. On the other of transactional sex in Namibia [24]. Research demonstrated hand, attrition may have been predominated by participants with that ‘‘transactional sex among women was associated with a a higher risk of HIV infection. Based on the latter, HIV prevalence 54% increase in odds of being HIV seropositive’’ [25] and would be underestimated. If these factors would balance each concern more often the poorest communities who want to other out and with the currently estimated HIV incidence of 2.4%, achieve higher socioeconomic status [25–28]. This result we can estimate HIV prevalence including those individuals who confirms the trend found between women wealth status and were HIV negative in previous surveys but who did not participate HIV acquisition [1,29]. in the last (2009) survey. This calculation yields an estimated

PLoS ONE | www.plosone.org 5 October 2011 | Volume 6 | Issue 10 | e25860

98 Incidence of HIV in Windhoek-Namibia 0.001 , – 0.74 0.65–0.85 –– – – Univariate and multiple 1 0.001 , 0.79 0.08 0.73 Univariate analysis Multiple analysis 2009 and individuals tested positive in 2008 but did not perform a test in 115 3.8 1.1–14.1 580 0.73 0.65–0.83 7561 1.66 0.55–5.03 1 MEN 85 1 n (580) OR 95%CI p OR 95%CI p 8056 3.2 3.9 0.9–10.9 1.3–11.4 205 1 130 1.05 0.33–3.20 0.82 370 1 158 1 427 0.92 0.89–1.12 0.92 72 0.94 0.24–3.64 224 0.61162 0.16–2.40 0.87 0.38–1.98 0.73 580243 1.02 0.99–1.05 1 0.12 – – 19811 1.37 0.57–3.29 2.33 0.27–20.0 5433327 0.82 0.42 0.17–3.91 0.19–1.97 0.88 0.19–6.63 369 1.94 0.72–5.21 0.19 0.001 , – 0.12 – – 1 0.74 0.64–0.78 –– 0.80 0.34–1.89 –– Multiple analysis 0.52 0.17–1.62 0.36 0.15–0.89 0.001 0.08 0.01 0.63 0.07 , 0.60 Univariate analysis 144 2.11 1.0–4.4 613 0.91 0.82–1.01 749 0.73 0.67–0.80 n (749) OR 95%CI P OR 95%CI P 83 1.25 0.4–3.5 329 1 484196 1 1.18 0.75–2.83 0.27 749328 0.98 1 0.97–1.01 0.15 – – 216 0.6955 0.36–1.33 1 258 1 4946525 0.74 1.24 0.12–4.61 0.36–4.19 2.36 0.44–12.6 WOMEN 418 0.86 0.49–1.52 0.60 /month 99 0.36 0.13–1.02 $ $ /month 151 1.1 0.50–2.10 /month 116 1 /month 275 0.30 0.14–0.85 $ $ 302* N 906 N Moderate to highDon’t know 85 2.60 1.1–6. Small , No risk at all 604–905 N 302–603 N $ No Yes Never married Married Consensual unionDivorc./sep./widowNone 82 52 0.65 1.01 0.24–1.72 0.39–2.86 Primary Secondary Tertiary Employed Unemployed Socio-demographic characteristics associated with HIV incidence between 2007 and 2009 among women and men aged 18 years or older sexual intercourse st Sample population consist of all2009) individuals - with see HIV Figure negative 1. status at – baseline TB: and tuberculosis a – known * HIV poverty status in line 2009 in (ie 2006. individuals who performed a test in HIV knowledge score, 2009 Health Insurance policy doi:10.1371/journal.pone.0025860.t003 Table 3. 1 Age Marital status Age at 1 Personal HIV risk perception School-grade analysis. Employment status, past year Characteristics in 2006 Per capita consumption level

PLoS ONE | www.plosone.org 6 October 2011 | Volume 6 | Issue 10 | e25860

99 Incidence of HIV in Windhoek-Namibia

Figure 2. Geographic representation of the city of Windhoek with ‘‘bubble-graphs’’ representing values of various indicators. Figure 2a, HIV Prevalence per PSU, 2009/The HIV prevalence ranged from 0 to 47%. Figure 2b, Mean HIV Personal Risk Perception per PSU, 2009: The HIV risk perception was measured on a 1 to 4 scale. The mean score per PSU ranged from 0 to 1.7. Figure 2c, Mean HIV knowledge score per PSU, 2009: HIV knowledge was measured according to a 1–11 scale. The mean score per PSU ranged from 2.3 to 11.0. Figure 2d, Mean per-capita consumption level in household per PSU, 2009: Mean Per capita consumption level per PSU ranged from 485$ to 125,800$ per month. One outlier of 530,800$ per month has not been included in this graph. doi:10.1371/journal.pone.0025860.g002 overall cumulative HIV prevalence of 16% over the full three among urban residents compared to rural populations [1,8] and years of surveying in Windhoek. strongly associated with migrations [41]. Therefore, the HIV Lastly, while prospective cohort studies are considered the ‘‘gold prevalence and incidence is likely to be lower in other parts of standard’’ approach to establish determinants of HIV incidence, Namibia, probably with the exception of the Caprivi strip. they remain subject to biases [20] since these determinants are Continued high urban HIV incidence in Namibia might also be based on self-reported behaviours. Individuals’ reports may not partly explained by the high income disparity in Windhoek. reflect actual behaviour because of memory bias or social Economic inequality was found to be highly associated with sexually desirability bias, especially when considering attitudes and behav- transmitted diseases and HIV [11] and has been demonstrated to iours regarding sexual activity or HIV [40]. Our surveys are based worsen health outcomes across all economic strata in society [42]. on actual HIV test results and are more reliable in this respect. The geographical analyses in this paper highlight the concen- It has to be emphasized that this survey was performed in an tration of HIV in the North-West (NW) part of the city. This urban area, characterized by persistent migrations and strong salient finding is not easily picked up by statistical analyses alone. economic inequalities. It is known that HIV prevalence is higher These are known to be the lower income areas occupied largely by

PLoS ONE | www.plosone.org 7 October 2011 | Volume 6 | Issue 10 | e25860

100 Incidence of HIV in Windhoek-Namibia previously disadvantaged Namibians, often migrants from within obvious that individuals with better HIV knowledge can engage and outside Namibia. People in these areas reside more closely in themselves in safer behaviours [43]. For this reason, the current informal housing, without the same access to amenities as the paper is a call for increased HIV information and education high/middle income areas in the South-East (SE). Thus, in prevention campaigns in Windhoek, with a particular focus on the Windhoek, HIV prevalence appears at least geographically to NW part of the city. The campaign should cut across all socio- coincide with areas with poorer living standards. However, economic strata of society, focus on women and impact should be bivariate and multivariate analyses of HIV incidence did not measured by continued representative HIV incidence surveys. reveal significant relations. This may be because power to detect any association was somehow lower for incidence analysis or Supporting Information because the factors driving the geographical correlation between poverty and HIV infection have changed over time, e.g. from Appendix S1 Construction of the HIV Knowledge score. economic to knowledge-related mechanisms. (DOC) The lack of a relationship between HIV incidence and socio- demographic factors such as marital status, education level, Acknowledgments employment, etc. confirmed once again that HIV risks concern all economic strata in African urban societies [20]. Continued efforts The authors wish to acknowledge Raph Hamers for critical suggestions. are required to combat the epidemic at the general population level. In this respect, it is clearly demonstrated by the current survey that Author Contributions the only socio-demographic factors that remain significantly Conceived and designed the experiments: WJ IDB JVDG TRW. associated with HIV incidence as well as prevalence are related to Performed the experiments: IDB GVR EG. Analyzed the data: MA WJ proper knowledge of the disease (knowledge scores) and an adequate CH. Contributed reagents/materials/analysis tools: EG. Wrote the paper: perception of the risk of HIV infection. The geographical analyses MA TRW. Edited the manuscript: WJ IDB GvR EG CH JVDG TRW. confirmed these observations by indicating correlations between Contributed to the acquisition of funding: JVDG TRW. HIV knowledge, HIV risk perception and HIV prevalence. It is

References 1. Fox AM (2010) The social determinants of HIV serostatus in sub-Saharan 18. Kumwenda N, Hoffman I, Chirenje M, Kelly C, Coletti A, et al. (2006) HIV Africa: an inverse relationship between poverty and HIV? Public Health Rep incidence among women of reproductive age in Malawi and Zimbabwe. Sex 125 Suppl 4: 16–24. Transm Dis 33: 646–651. 2. Schmidt M (2009) Poverty and Inequality in Namibia: An Overview. Windhoek: 19. Padian NS, van der Straten A, Ramjee G, Chipato T, de Bruyn G, et al. (2007) Institute for public policy research. 8 p. Diaphragm and lubricant gel for prevention of HIV acquisition in southern 3. WHO (2010) World health Statistics. Geneva: WHO. 177 p. WHO website. African women: a randomised controlled trial. Lancet 370: 251–261. Available: http://www.who.int/whosis/whostat/EN_WHS10_Full.pdf. Ac- 20. Braunstein SL, van de Wijgert JH, Nash D (2009) HIV incidence in sub-Saharan cessed 2011 September 14. Africa: a review of available data with implications for surveillance and 4. WHO, UNAIDS, UNICEF (2010) Towards universal access: Scaling up priority prevention planning. AIDS Rev 11: 140–156. HIV/AIDS interventions in the health sector. 150 p. WHO website. Available: 21. Bailey RC, Moses S, Parker CB, Agot K, Maclean I, et al. (2007) Male http://www.who.int/hiv/pub/2010progressreport/report/en/index.html. Ac- circumcision for HIV prevention in young men in Kisumu, Kenya: a cessed 2011 September 14. randomised controlled trial. Lancet 369: 643–656. 5. Government of Namibia, Health Systems 20/20 Project, WHO, UNAIDS 22. MoHSS (2009) 2008/9 estimates and projections of the impact of HIV/AIDS in (2010) Namibia Health Resource Tracking: 2007/2008 & 2008/2009. Bethesda, Namibia – Report, December 2009. MD: Health Systems 20/20 project. Abt Associates Inc. 23. Abdool Karim Q, Sibeko S, Baxter C (2010) Preventing HIV infection in 6. UNAIDS (2010) Report on the Global AIDS Epidemic.364 p. UNAIDS website. women: a global health imperative. Clin Infect Dis 50 Suppl 3: S122–129. Available: http://www.unaids.org/globalreport/global_report.htm. Accessed 24. USAID, MOHSS (2009) HIV/AIDs in Namibia: Behavioral and Contextual 2011 September 14. Factors Driving the Epidemic. Windhoek, USAID. 98 p. 7. Frayne B (2004) Migration and urban survival strategies in Windhoek, Namibia. 25. Dunkle KL, Jewkes RK, Brown HC, Gray GE, Mc Intryre JA, Harlow S (2004) Geoforum 35: 489–505. Transactional sex among women in Soweto, South Africa: prevalence, risk 8. Garcia-Calleja JM, Gouws E, Ghys PD (2006) National population based HIV factors and association with HIV infection. Soc Sci Med 59: 1581–92. prevalence surveys in sub-Saharan Africa: results and implications for HIV and 26. MacPhail C, Campbell C (2001) ‘I think condoms are good but, I hate those AIDS estimates. Sex Transm Infect 82 Suppl 3: iii64–70. things’: condom use among adolescents and young people in a Southern African 9. Kyobutungi C, Ziraba AK, Ezeh A, Ye Y (2008) The burden of disease profile of township. Soc Sci Med 52: 1613–27. residents of Nairobi’s slums: Results from a Demographic Surveillance System. 27. Hunter M (2002) The materiality of every day sex: thinking beyond Popul Health Metr 6: 1. ‘prostitution’. African studies 61: 99–120. 10. Vearey J, Palmary I, Thomas L, Nunez L, Drimie S (2010) Urban health in 28. Jewkes RK, Levin JB, et al. (2003) Gender inequalities, intimate partner violence Johannesburg: the importance of place in understanding intra-urban inequalities and HIV preventive practices: findings of a South African cross-sectional study. in a context of migration and HIV. Health Place 16: 694–702. Soc Sci Med 56: 125–34. 11. Holtgrave DR, Crosby RA (2003) Social capital, poverty, and income inequality 29. Shelton JD, Cassell MM, et al. (2005) Is poverty or wealth at the root of HIV? as predictors of gonorrhoea, syphilis, chlamydia and AIDS case rates in the Lancet 366: 1057–8. United States. Sex Transm Infect 79: 62–4. 30. WHO (2007) WHO website. Available: http://www.who.int/mediacentre/ 12. Galea S, Vlahov D (2005) Urban health: evidence, challenges, and directions. news/releases/2007/pr61/en/index.html. Accessed 2011 September 14. Annu Rev Public Health 26: 341–365. 31. Janssens W, de Beer I, Coutinho H, van Rooy G, van der Gaag J, Rinke de 13. Harpham T (2009) Urban health in developing countries: what do we know and Wit TF (2010) Estimating HIV prevalence: A cautious note on household where do we go? Health Place 15: 107–116. surveys in poor settings. BMJ 341: c6323. 14. Vlahov D, Galea S, Gibble E, Freudenberg N (2005) Perspectives on urban 32. Mwaluko G, Urassa M, Isingo R, Zaba B, Boerma JT (2003) Trends in HIV and conditions and population health. Cad Saude Publica 21: 949–957. sexual behaviour in a longitudinal study in a rural population in Tanzania, 15. Schellekens O, de Beer I, Lindner M, van Vugt M, Schellekens P, Rinke de 1994–2000. AIDS 17: 2645–2651. Wit T (2009) Innovation in Namibia: Preserving Private Health Insurance and 33. Gregson S, Garnett GP, Nyamukapa CA, Hallett TB, Lewis JJ, et al. (2006) HIV HIV/AIDS treatment. Health Affairs 28: 6. decline associated with behavior change in eastern Zimbabwe. Science 311: 664–6. 16. Hamers RL, de Beer IH, Kaura H, van Vugt M, Caparos L, et al. (2008) 34. Lopman B, Lewis J, Nyamukapa C, Mushati P, Chandiwana S, et al. (2007) HIV Diagnostic accuracy of 2 oral fluid-based tests for HIV surveillance in Namibia. incidence and poverty in Manicaland, Zimbabwe: is HIV becoming a disease of J Acquir Immune Defic Syndr 48: 116–118. the poor? AIDS 21 Suppl 7: S57–66. 17. Grosh M, Glewwe P (2000) Designing household survey questionnaires for 35. Wambura M, Urassa M, Isingo R, Ndege M, Marston M, et al. (2007) HIV developing countries: lessons from fifteen years of the Living Standards prevalence and incidence in rural Tanzania: results from 10 years of follow-up in Measurement Study. Washington, DC: The World Bank. an open-cohort study. J Acquir Immune Defic Syndr 46: 616–623.

PLoS ONE | www.plosone.org 8 October 2011 | Volume 6 | Issue 10 | e25860

101 Incidence of HIV in Windhoek-Namibia

36. Gustafsson-Wright E, Janssens W, van der Gaag J (2011) The Inequitable 41. Voeten HA, Vissers DC, Gregson S, Zaba B, White RG, et al. (2009) Strong Impacts of Health Shocks on the Uninsured in Namibia. Health Policy and Association Between In-Migration and HIV Prevalence in Urban Sub-Saharan Planning 26: 142–156. Africa. Sex Transm Dis 2. National Center for Biotechnology Information 37. Marston M, Harriss K, Slaymaker E (2008) Non-response bias in estimates of (NCBI) Website. Available: http://www.ncbi.nlm.nih.gov/pubmed/19959971. HIV prevalence due to the mobility of absentees in national population-based Accessed 2011 September 14. surveys: a study of nine national surveys. Sex Transm Infect 84 Suppl 1: 42. Daniels N, Kennedy BP, Kawachi I (1999) Why justice is good for our health: i71–7. the social determinants of health inequalities. Daedalus 128: 215–51. 38. Mishra V, Barrere B, Hong R, Khan S (2008) Evaluation of bias in HIV 43. Stringer EM, Sinkala M, Kumwenda R, Chapman V, Mwale A, et al. (2004) seroprevalence estimates from national household surveys. Sex Transm Infect 84 Personal risk perception, HIV knowledge and risk avoidance behavior, and their Suppl 1: i63–70. relationships to actual HIV serostatus in an urban African obstetric population. 39. Janssens W, Van der Gaag J, Rinke de Wit TF (2008) Non-response bias in the J Acquir Immune Defic Syndr 35: 60–66. measurement of HIV/AIDS prevalence in Namibia. Amsterdam: Amsterdam 44. MOHSS (2008) Namibia. Demographic and Health Survey, 2006–2007. Institute for International Development. Windhoek. 428 p. 40. Zaba B, Pisani E, Slaymaker E, Ties Boerma J (2004) Age at first sex: understanding recent trends in African demographic surveys. Sex Transm Infect 80: 28–35.

PLoS ONE | www.plosone.org 9 October 2011 | Volume 6 | Issue 10 | e25860

102 9b A CAUTIOUS NOTE ON HOUSEHOLD SURVEYS IN POOR SETTINGS

103 These letters are selected from rapid responses posted bmj.com To submit a rapid on bmj.com. Selection is usually made 12 days after print publication of the article to which they respond. response go to any article LETTERS on bmj.com and click “respond to this article”

ESTIMATING HIV PREVALENCE of household survey data are recommended to accurately reflects the results of the pooled trials, avoid basing HIV global needs assessments on a range of ARRs would need to be reported. A cautious note on household flawed prevalence rates.2 3 Most journals have stringent word limits for Wendy Janssens research fellow , Amsterdam Institute abstracts. Fitting all the existing requirements surveys in poor settings for International Development, 1105 BM Amsterdam, for abstracts that conform to the CONSORT/ Netherlands [email protected] The World Health Organization’s HIV prevalence Ingrid de Beer general manager , PharmAccess Foundation PRISMA guidelines within these word limits estimates have recently been adjusted Namibia, First Floor Angola House, Windhoek, Namibia is difficult. Accurately conveying valid and downwards, mostly because of new data from Hannah M Coutinho medical officer , PharmAccess generalisable measures of absolute risk requires population based surveys.1 But such surveys Foundation, 1105 BM Amsterdam, Netherlands a large number of words. Unless word limits Gert van Rooy research fellow , Multidisciplinary Research are limited by surveyor bias—they are typically and Consultancy Centre (MRCC), University of Namibia, for abstracts are relaxed, or the format for performed on large numbers of respondents Windhoek, Namibia reporting clinical trials is restructured, what by small numbers of surveyors—and this could Jacques van der Gaag director, senior fellow , Amsterdam existing requirement should be removed so that disproportionately influence (worldwide) HIV Institute for International Development, 1105 BM absolute risks can be added? Amsterdam, Netherlands and Brookings Institution, Mark J Bolland senior research fellow prevalence estimates. Washington, DC 20036, USA In 2007 a population based household survey [email protected] Tobias F Rinke de Wit director, advocacy, technology and Andrew Grey associate professor of medicine , Department in a sub-Saharan country randomly assigned research , PharmAccess Foundation, 1105 BM Amsterdam, of Medicine, University of Auckland, Auckland 1142, New eight trained nurses to perform medical Netherlands Zealand Competing interests: None declared. interviews and collect oral fluid samples for Competing interests: None declared. A longer version of this note is posted on www.pharmaccess. anonymous HIV testing on 2452 people. The 1 Gigerenzer G, Wegwarth O, Feufel M. Misleading org and www.aiid.org . communication of risk. BMJ 2010 ; 341 : c4830 . estimated HIV prevalence was 12.7%. 1 WHO. Global HIV prevalence has levelled off. 2007. (12 October.) The figure shows weekly HIV positivity www.who.int/mediacentre/news/releases/2007/pr61/ Cite this as: BMJ 2010;340:c6331 estimates collected during the survey, stratified en/index.html . 2 Janssens W, de Beer I, Coutinho HM, van Rooy G, van der by nurse. HIV positivity in samples obtained Gaag J, Rinke de Wit TF. A cautious note on household by one nurse (“H”) increased to more than survey HIV prevalence estimates in resource-poor Absolute risk reduction settings. BMJ 2010 . www.bmj.com/content/341/bmj. 80% during the second phase of the survey. c3942/reply#bmj_el_243786 . Reported results from trials of screening for No significant differences were found in age, 3 Roehr B. International AIDS relief stagnated in 2009. abdominal aortic aneurysms in men over 65 BMJ 2010 ; 341 : c3942 . (20 July.) sex, education, income, marital status, or years provide another example of misleading Cite this as: BMJ 2010;341:c6323 household demographics of respondents visited communication of risk.1 All trials report a by nurse “H” compared with the other nurses. DATA OPENNESS 40-50% relative reduction in mortality for The areas assigned to nurse “H” were identical abdominal aortic aneurysms with screening. to those assessed by three other nurses, none Sounds fantastic, but the absolute risk of whom showed similar results. When all Reporting of absolute risk reduction is a paltry 0.4%. All publications 313 respondents sampled by nurse “H” were In arguing that abstracts should always contain included the absolute data but focused on excluded, HIV prevalence dropped from 12.7% absolute risks, Gigerenzer and colleagues the relative risk reduction. For example, in the (95% CI 11.4 to 14.0) to 9.6% (7.3 to 11.8). do not tackle the practical limitations of this report of the 10 year results of the UK MASS trial If one nurse in 10 produced 50% false recommendation. 1 They do not mention for the “What this study adds” box (which many positive results, estimated HIV prevalence whom the absolute risks should be reported. busy readers may limit themselves to) stated: would be 1.2, 1.5, or 2.6 times higher than it Absolute risk reduction (ARR) depends on “About half of all aneurysm related deaths should be in 15%, 6%, or 2% HIV prevalence baseline risk and the event rate, each of which should be prevented by a national screening settings, respectively.1 Retrospective analyses varies by age, sex, prevalent disease, and programme.”2 Although no deliberate attempt comorbidities, so ARRs should be reported to mislead was made, enthusiasm for screening Nurse separately for each variable. Furthermore, event was not tempered with reflection on the small A C E G B D F H rates are often lower in trial participants than in size of the absolute risk reduction. 100 unselected populations. To increase the external Paul E Norman vascular surgeon , University of Western Australia, WA 6009, Australia 80 validity of the ARR, investigators may use [email protected] existing published event rates. If this approach 60 Competing interests: None declared. is followed, how can the uncertainty in the 1 Gigerenzer G, Wegwarth O, Feufel M. Misleading 40 population event rates be combined with that in communication of risk. BMJ 2010 ; 341 : c4830 . the observed relative risk reduction to produce (12 October.) 20 2 Thompson SG, Ashton HA, Gao L, Scott RAP; on behalf accurate confidence intervals for the ARR? of the Multicentre Aneurysm Screening Study Group. 0 Meta-analyses often pool trials with different Screening men for abdominal aortic aneurysm: 10 HIV positive respondents (%) 1 2 3 4 5 6 7 8 9 10 11 year mortality and cost effectiveness results from the event rates and different ARRs. In this situation, randomised Multicentre Aneurysm Screening Study. BMJ Week how should the ARR be reported? Because 2009 ; 338 : b2307 . Prevalence of HIV positive results from eight nurses neither the average ARR nor that for any one trial Cite this as: BMJ 2010;341:c6333

BMJ | 20 NOVEMBER 2010 | VOLUME 341 1063

104 10 PREVALENCE AND KNOWLEDGE ASSESSMENT OF HIV AND NON-COMMUNICABLE DISEASE RISK FACTORS AMONG FORMAL SECTOR EMPLOYEES IN NAMIBIA

105 RESEARCH ARTICLE Prevalence and Knowledge Assessment of HIV and Non-Communicable Disease Risk Factors among Formal Sector Employees in Namibia

Leonor Guariguata1*, Ingrid de Beer2, Rina Hough2, Pancho Mulongeni2, Frank G. Feeley3, Tobias F. Rinke de Wit1,4

a11111 1 PharmAccess Foundation, Amsterdam, the Netherlands, 2 PharmAccess Foundation, Windhoek, Namibia, 3 Boston University School of Public Health, Boston, Massachusetts, United States of America, 4 Amsterdam Institute for Global Health and Development (AIGHD), Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands

* [email protected]

OPEN ACCESS Abstract Citation: Guariguata L, de Beer I, Hough R, Mulongeni P, Feeley FG, Rinke de Wit TF (2015) Prevalence and Knowledge Assessment of HIV and Introduction Non-Communicable Disease Risk Factors among The burden of non-communicable diseases (NCDs) is growing in sub-Saharan Africa com- Formal Sector Employees in Namibia. PLoS ONE 10(7): e0131737. doi:10.1371/journal.pone.0131737 bined with an already high prevalence of infectious disease, like HIV. Engaging the formal employment sector may present a viable strategy for addressing both HIV and NCDs in Editor: Philip Anglewicz, Tulane University School of Public Health, UNITED STATES people of working age. This study assesses the presence of three of the most significant threats to health in Namibia among employees in the formal sector: elevated blood pres- Received: November 7, 2014 sure, elevated blood glucose, and HIV and assesses the knowledge and self-perceived risk Accepted: June 4, 2015 of employees for these conditions. Published: July 13, 2015

Copyright: © 2015 Guariguata et al. This is an open Methods access article distributed under the terms of the A health and wellness screening survey of employees working in 13 industries in the formal Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any sector of Namibia was conducted including 11,192 participants in the Bophelo! Project in medium, provided the original author and source are Namibia, from January 2009 to October 2010. The survey combined a medical screening credited. for HIV, blood glucose and blood pressure with an employee-completed survey on knowl- Data Availability Statement: Data from all years of edge and risk behaviors for those conditions. We estimated the prevalence of the three con- the Bophelo! Project are available for download at ditions and compared to self-reported employee knowledge and risk behaviors and Figshare: DOI: http://dx.doi.org/10.6084/m9.figshare. possible determinants. 1342026. URL: http://figshare.com/articles/Bophelo_ Project_January_2009_October_2010/1342026. Results Funding: The Namibia Global Fund Program provided funding for the HIV screening of the 25.8% of participants had elevated blood pressure, 8.3% of participants had an elevated Bophelo! program. The funding for the Bophelo! is random blood glucose measurement, and 8.9% of participants tested positive for HIV. Most provided by The Dutch Foreign Ministry. The participants were not smokers (80%), reported not drinking alcohol regularly (81.2%), and Bophelo! Project is a partnership between the PharmAccess Foundation, the Namibian Business had regular condom use (66%). Most participants could not correctly identify risk factors for Coalition on AIDS, and the Namibian Institute of hypertension (57.2%), diabetes (57.3%), or high-risk behaviors for HIV infection (59.5%). In

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 1/12

106 Prevalence and Knowledge Assessment of HIV and NCDs in Namibia

Pathology, supported by the Ministry of Health and multivariate analysis, having insurance (OR:1.15, 95%CI: 1.03 – 1.28) and a managerial Social Services. The project was initially created position (OR: 1.29, 95%CI: 1.13 – 1.47) were associated with better odds of knowledge of using funds from the Dutch Postcode Loterij, through the Dutch Aidsfonds, Stop AIDS Now!, HIVOS and diabetes. the Global Fund to fight AIDS, TB and Malaria. The funders had no role in the study design, data Conclusion collection and analysis, decision to publish, or preparation of the manuscript. The prevalence of elevated blood pressure, elevated blood glucose, and HIV among employees of the Namibian formal sector is high, while risk awareness is low. Attention Competing Interests: The authors have declared that no competing interests exist. must be paid to improving the knowledge of health-related risk factors as well as providing care to those with chronic conditions in the formal sector through programs such as workplace wellness.

Background The burden of non-communicable disease (NCDs) is growing in sub-Saharan Africa[1,2] and Namibia in particular is heavily affected[1]. The World Health Organization indicates that Namibia has the 2nd highest rate of deaths attributable to cardiovascular disease and diabetes in the African region and is in the top 20 globally. The country also has the 4th highest mortality attributable to NCDs in the African region, a prevalence of 9.2% of raised blood glucose, and the 8th highest prevalence of high blood pressure in the world (49.1%)[1]. This growing problem is coupled with the already high burden of infectious diseases like HIV and tuberculosis. The prevalence of HIV in Namibia among people aged 15–49 is estimated at 13.1% or 160,000 people aged 15 and over[3]. For employers, maintaining the health of employees translates to reduced costs[4–6]. Stud- ies have shown that poor health among employees leads to decreases in productivity and increased absenteeism[7–10]. Health promotion programs and the adoption of employer-paid or contributed health insurance can lead to decreased absenteeism and reduced costs for employers[4,11–14]. Three conditions which have emerged as serious contributors to the health burden of Namibia (hypertension, diabetes, and HIV) have all been shown to be preventable through education on risk factors, lifestyle changes and behavior modification programs, and regular screening and education[15–21]. As a contribution to this, the formal sector employment in Namibia represents an opportunity to provide education and screening through company wellness programs, which can in turn reduce costs and improve the health of employees and complement public sector efforts in these areas. The purpose of this analysis was to conduct a secondary analysis of a large dataset derived from a workplace wellness survey to estimate the prevalence of three conditions in the study population: elevated blood pressure, elevated blood glucose, and HIV. In addition, this study puts the biomedical findings in the context of the knowledge and self-perceived risk of employees for these conditions and their related risk behaviors and determinants in order to identify areas for improvement in education and screening.

Methods The study is based on a secondary analysis of data obtained from a health and wellness screening survey conducted of 11,192 participants in the Bophelo! Project in Namibia, from January 2009 to October 2010. The survey is the largest ever performed in the formal sector in Namibia

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 2/12

107 Prevalence and Knowledge Assessment of HIV and NCDs in Namibia

and reached employees in 13 industries, including self-reported data as well as a biomedical assessment. Information was collected per sector of each company as well as whether the company had an HIV prevention and education workplace program in place including services such as regular information sessions and free voluntary screening. Participation in the screening and survey was entirely voluntary and strictly confidential. To protect confidentiality, no identifiers were stored with the information gathered and participants gave written informed consent before inclusion. The survey was performed by two mobile clinics from the Bophelo! Project, a partnership between PharmAccess, the Namibian Business Coalition on AIDS and the Namibia Institute of Pathology. As part of the project, wellness surveys and health screening services were offered to companies along with sensitization and information sessions provided before employee screenings. The mobile clinic visited a company site for as long as required to see all participating employees (up to 32 people per mobile clinic per day) and each screening visit took approximately a total of 45–60 minutes consisting of pre-test counseling (30–45 minutes), testing, and post-test counseling sessions. All services were provided by Ministry of Health and Social Services (MOHSS)-trained testers and counselors. While screening results were being obtained, the participant received health education and confidential counseling. Participants who tested positive for any condition or had screening results outside of a normal range were encouraged to seek follow-up consultation and were issued a referral letter to local medical services. Only de-identified information was collected from the screenings without any possibility of linking back to the participant. The study was approved by the ethical board of the MOHSS of Namibia. Participants who provided informed consent were asked to complete a questionnaire with demographic (age, sex, marital status, smoking), job status information (contract type, type of position), and knowledge, self-perceived risk and self-reported presence of the conditions: HIV/AIDS, diabetes, hypertension and cardiovascular disease. Knowledge of conditions was assessed by asking participants to respond to true-false statements about related symptoms and risk behaviors. For hypertension, participants were asked whether the following activities could lower their risk of hypertension: maintaining a healthy body weight, increasing physical activity, avoiding salt in the diet, stopping smoking, avoiding excessive alcohol intake, and whether or not it was possible for an individual to lower their own risk of hypertension. Similarly, participants were asked about diabetes-related risk behaviors including: maintaining a healthy weight, increasing physical activity, and lowering one’s own risk for diabetes. Questions for HIV/AIDS risk were: whether a mother could transmit HIV to her child, whether mother-to- child transmission could be prevented, if condom use can prevent HIV infection, identifying true and false statements regarding transmission routes for HIV including food sharing, shaking hands, kissing, and sexual behavior. In addition, participants were asked whether a healthy looking person could have HIV, whether a person with tuberculosis always has HIV, whether traditional healers can cure HIV/AIDS, and whether having sex with a virgin can cure AIDS. For the purposes of analysis, “good” knowledge of the conditions was considered a correct answer on each of the respective questions. Knowledge was therefore calculated as a binomial variable (All correct answers/Not all correct answer) for use in univariate and multivariate analysis. In addition, participants who agreed received a medical screening for HIV status, random blood glucose, systolic and diastolic blood pressure and anthropometric markers. A finger prick was conducted on all participants to collect a blood sample for HIV and blood glucose. The following assays and methods were used for collecting medical screening information.

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 3/12

108 Prevalence and Knowledge Assessment of HIV and NCDs in Namibia

HIV Testing The Determine HIV 1/2 Assay by Inverness rapid test strip and Trinity Biotech Uni-Gold HIV 1/2 test kit were used in conjunction for HIV testing, according to Namibian VCT regulations. If the result was discordant between the two tests, a third (tie-breaker) test was performed using the Inverness Clearview Complete HIV 1/2 test kit.

Blood glucose test Blood glucose was assessed through a non-fasting capillary blood draw. Values outside a normal range (3.0 mmol/L– 6.6 mmol/L) were repeated by a second draw and the second result taken for analysis. The Accutrend Plus GCT meters were used to conduct the test (Roche Molecular Diagnostics, USA). The meters were operated at 18 to 32 degrees Celsius. Diabetes was considered likely if the result was 11 mmol/L in accordance with the American Diabetes Association guidelines, and an elevated value was considered a result 6.6 mmol/L. Partici- pants with values 11 mmol/L were referred for follow-up to a medical facility. Blood pressure determination Systolic and diastolic blood pressure was measured using a sphygmomanometer in millimetres of mercury (mmHg) by using the MG150f Digital Blood Pressure monitor (Rossmax Interna- tional Ltd, Taiwan). Blood pressure was measured on the upper arm, in a sitting position. The measurement was taken three times on each participant and the average of the three readings was used. A blood pressure reading above 140/90 mmHg was considered elevated; above 153/ 103 mmHg was considered high and referral was made to a medical facility.

Statistical Analysis Data were collected in DOS and stored using SPSS. All statistical analyses were conducted using R Project for Statistical Computing version 2.10.0 (www.r-project.org). The chi-squared test for significance was used for comparisons in contingency tables with p 0.05 considered statistically significant. Participants were allowed to refuse any or all of these screenings. Not all subjects took every test because either they refused to have it, or the employer opted not to provide the particular screening to that company. In this case, the subjects were excluded from analysis for a particular screening and only respondents included. Multivariate analysis was carried out using logistic regression analysis. Where necessary, variables with several categories were modified to binary variables for ease of interpretation in multivariate analyses and where the additional information provided by the categories was not significant.

Results Demographics, self-reported health, and health-related behaviors The majority of survey participants were below the age of forty, male, and had not completed a secondary school level education (Table 1). In addition, most were working in manual labor and not in administration, management, or supervisory positions. The majority of respondents worked in food manufacturing and fishing. For self-reported health status, almost 80% reported being in “good” or “excellent” health (n = 8,730). Ten per cent of participants reported having hypertension. Furthermore, 25.8% found to have an elevated blood pressure on screening. Fifty-nine per cent of respondents who

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 4/12

109 Prevalence and Knowledge Assessment of HIV and NCDs in Namibia

Table 1. Demographic and screening results for hypertension, high blood glucose and HIV of formal sector employees in Namibia.

Demographics N % 95%CI Age 20–29 3,150 28.1 (27.3–29.0) 30–39 3,524 31.5 (30.6–32.4) 40–49 2,363 21.1 (20.4–21.9) 50–59 1,154 10.3 (9.8–10.9) 60 226 2 (1.8–2.3) Sex Male 7,356 65.7 (64.8–66.6) Female 3,834 34.3 (33.4–35.1) Education Below Secondary 5,078 45.4 (44.4–46.3) Secondary or Higher 5,073 45.3 (44.4–46.2) Job Type Administration 976 8.7 (8.2–9.3) Management 552 4.9 (4.5–5.3) Supervisor 648 5.8 (5.4–6.2) Labour 8,732 78 (77.2–78.8) Sector Administration and Business 408 3.6 (3.3–4.0) Hotel and Restaurants 331 3.0 (2.7–3.3) Food Manufacturing 2,504 22.4 (21.6–23.1) Retail and Trade 1,145 10.2 (9.7–10.8) Tourism 525 4.7 (4.3–5.1) Wholesale 245 2.2 (1.9–2.5) Agriculture 1,112 9.9 (9.4–10.5) Utilities 1,264 11.3 (10.7–11.9) Infrastructure 691 6.2 (5.7–6.6) Fishing 1,969 17.6 (16.9–18.3) Manufacturing 63 0.6 (0.4–0.7) Mining 318 2.8 (2.5–3.2) Storage and Transport 617 5.5 (5.1–5.9) Screening Results N % 95%CI Blood pressure 140/90 mmHg (Elevated/High) 2,888 25.8 (25.0–26.6) <140/90 mmHg (Normal) 8,223 73.5 (72.7–74.3) Declined or Not Tested 81 0.7 (0.6–0.9) Blood glucose 6.6 mmol/L (Elevated) 933 8.3 (7.8–8.8) <6.6 mmol/L (Normal) 9,583 85.6 (85.0–86.3) Declined or Not Tested 676 6.0 (5.6–6.5) HIV Positive 993 8.9 (8.4–9.4) Negative 8,848 79.1 (78.3–79.8) Declined or Not Tested 1,351 12.1 (11.5–12.7) doi:10.1371/journal.pone.0131737.t001

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 5/12

110 Prevalence and Knowledge Assessment of HIV and NCDs in Namibia

reported being diagnosed with hypertension were female. Less than 1.5% of participants reported having been diagnosed with diabetes. With regard to risk behaviors, the majority of participants reported eating fruits daily, 20% reported being smokers and 19.8% reported drinking alcohol once per week. The majority of participants reported they did not have sex outside their regular relationships (76.6%) and the majority also reported using a condom always if having extra-marital sex (66.0%). A total of 8.9% of respondents tested positive for HIV upon screening.

Self-perceived risk For all three conditions, excluding non-respondents and those who reported having the condition, the majority of respondents reported no knowledge of or felt they had no risk or a small risk of developing the condition (Fig 1). With regard to self-perceived risk of each condition, almost half of respondents considered they had a low to no risk of hypertension (48.1%), or did not know what the condition was (14.9%). Similarly, 44.5% of respondents thought they had low or no risk of developing diabetes, and 14.0% reported they did not know anything about diabetes. Conversely, only 5.7% of reported no knowledge of HIV/AIDS and 41.5% of respondents thought they had an at least moderate risk of contracting HIV. When factoring in screening results, 21% of those who reported they had no knowledge of hypertension were found to have elevated blood pressure. Similarly, 21% of those reporting no knowledge of HIV tested positive upon screening. Nine per cent of those reporting no knowledge of diabetes were found to have elevated blood glucose on screening.

Knowledge assessment Participants were asked a series of questions relating to each of the three conditions. For questions related to hypertension, 42.8% of participants correctly responded to all questions on risk factors and associated behaviors for hypertension (Fig 1). A significantly larger proportion of those previously diagnosed with hypertension were able to correctly identify the risk factors compared to the rest of respondents (55.9% vs. 41.3% respectively, χ2 = 86.2, p<0.05). Simi- larly, 42.7% of participants were able to correctly answer all questions relating to the risk of diabetes and those previously diagnosed were significantly more likely to answer diabetes-related questions correctly than the rest of respondents (65.6% vs.42.5%, χ2 = 37.4, p<0.05) (Fig 1). 3.2% of all respondents found on screening to have elevated blood pressure had no knowledge of the condition. Similarly, 1.3% of all respondents with elevated blood glucose on screening had no knowledge of diabetes, and 0.6% of those with positive HIV screening reported no knowledge of HIV. Less than half of participants were able to correctly answer all questions correctly related to risk factors and behaviors for HIV/AIDS (40.5%), although there were more questions (n = 11) assessed than for diabetes (n = 3) or hypertension (n = 6). In addition, 95% of respondents were able to answer at least two-thirds of HIV-related knowledge questions correctly (Fig 1), compared to 64% for hypertension and 48% for diabetes. Those who screened positive for HIV were slightly better at answering behavior questions correctly (42.5%) compared to those who screened negative (39.9%; χ2 = 6.82, p = 0.03).

Univariate and multivariate knowledge analysis On univariate analysis and including only industry as a covariate to account for any clustering, increasing age and having insurance were significantly associated with good knowledge of both hypertension and diabetes (Table 2). Having an administrative or managerial position was associated with better odds of good knowledge for all three conditions (Table 2). In addition,

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 6/12

111 Prevalence and Knowledge Assessment of HIV and NCDs in Namibia

Fig 1. Self-perceived risk and knowledge of hypertension, diabetes, and HIV among total respondents and those with abnormal screening results in Namibian formal sector employees. Light grey bars represent responses among the total population, while the dark grey bars indicate what proportion of those respondents had abnormally high values on screening for blood pressure and blood glucose, or positive HIV screening. doi:10.1371/journal.pone.0131737.g001

being female was associated with significantly better knowledge of HIV compared to men (Table 2). In multivariate analysis controlled additionally for age, sex, education, insurance, and job type (Table 2). After controlling for covariables, increasing age was associated with good knowledge of hypertension and diabetes; being female was associated with better knowledge of HIV; and having insurance or having administrative position were associated with diabetes

Table 2. Factors associated with knowledge* of hypertension, diabetes and HIV among formal sector employees in Namibia.

Crude Model: OR (95% CI) Adjusted Model: OR (95% CI)

Hypertension Diabetes HIV Hypertension Diabetes HIV Variables Age (continuous) 1.02 (1.02, 1.01 (1.01, 1.00 (0.99, 1.02 (1.02, 1.02 (1.01, 1.00 (0.99, 1.02) 1.02) 1.00) 1.03) 1.02) 1.00) Sex (Female vs. Male) 1.08 (0.99, 1.04 (0.96, 1.29 (1.19, 1.09 (0.99, 1.02 (0.92, 1.26 (1.14, 1.18) 1.14) 1.41) 1.21) 1.12) 1.39) Education ( Secondary vs. Below 0.91 (0.84, 0.99 (0.91, 1.02 (0.94, 0.90 (0.82, 0.93 (0.84, 0.92 (0.83, Secondary) 0.99) 1.07) 1.11) 1.00) 1.02) 1.01) Insrance (Yes vs. No) 1.15 (1.05, 1.28 (1.17, 1.02 (0.93, 1.06 (0.95, 1.15 (1.03, 0.97 (0.87, 1.26) 1.40) 1.12) 1.18) 1.28) 1.08) Job type (Adminstrative/Managerial vs. Manual 1.11 (1.00, 1.37 (1.24, 1.12 (1.01, 1.03 (0.90, 1.29 (1.13, 1.11 (0.97, labour) 1.23) 1.52) 1.24) 1.18) 1.47) 1.27)

*Knowledge is deﬁned as correctly answering all questions related to each of the three conditions on the given survey. The crude model included industry as a covariate. The adjusted model further included all the covariates listed (age, sex, education, insurance, job type). doi:10.1371/journal.pone.0131737.t002

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 7/12

112 Prevalence and Knowledge Assessment of HIV and NCDs in Namibia

knowledge (Table 2). None of the independent variables was strongly associated with better knowledge of a particular disease. No other significant associations or trends were for any of the other independent variables described for any of the three conditions.

Discussion The Bophelo! Project survey presented here covered the majority of key industries in Namibia and the sample size represents approximately 3% of the total Namibian formal sector workforce[22]. The results presented in this paper demonstrate overall low knowledge and self-perceived risk for three major conditions among a large Namibian population working in the formal sector. Those who were previously diagnosed showed overall better knowledge of their conditions than total respondents, suggesting that some education by their care providers is reaching them. Of the three conditions, respondents demonstrated better overall knowledge of HIV, which may reflect efforts by the MOHSS in population wide HIV-related education. Knowledge on blood pressure and blood glucose was overall low. An assessment of African migrants in Glasgow showed a similar pattern with low knowledge of chronic diseases and infectious disease perceived as the greatest threat[23]. Low knowledge levels could also be partly explained by the fact that the overall disease burden of NCDs in Namibia is lower and has a shorter history than that for HIV[24,25], but there is an expectation of a rapid rise in CVD and diabetes for low- and middle-income countries in the next generation[26]. Self-perceived risk has also been shown to affect health-seeking behavior and behavior change. A study of perceptions of risk of HIV/AIDS and sexual behavior in Kenya found a strong positive association between self-perceived risk and risky sexual behavior[27]. Where there is discordance between self-perceived risk and actual risk, the necessary treatment may be delayed or risk behaviors amplified; a relationship that has been seen in studies of HIV [19,28,29] as well as diabetes[30,31] and hypertension[32] in African populations. In this study, majority of participants were not able to accurately assess their own risk of any of the three conditions presented regardless of screening result which may influence not only health- seeking behavior, but also the self-monitoring and management necessary to prevent complications related to any of the three conditions. From multivariate analysis, it is difficult to determine the independent factors related to gaps in knowledge from demographic and work-related information. None of the independent variables had a strong relationship with knowledge of a particular condition. Studies have shown some benefit to workplace wellness education programs in improvements on cardiac risk factors[6], obesity[14], cardiovascular disease and diabetes[33], and physical activity[34]. In general, Rula and Hobgood reported that “workplace health promotion programs founded on objective health metrics can motivate employee health-risk reduction”[35]. However, targeted education may be more effective. Abubakari et al. found that diabetes knowledge and illness perception varied by ethnicity, socioeconomic status, and other demographic factors[30], and similar results have been seen with self-reported health status and NCD risk factors in the US[36]. Further research is needed to determine target groups and those who would be particularly vulnerable to gaps in knowledge among formal sector employees. This study demonstrates the gaps in knowledge related to three conditions: diabetes, hypertension, and HIV, which have been shown to have significant impacts on employees and employers[7, 12, 25–28]. Directed workplace programs may provide the best opportunity for closing such knowledge gaps. Work environments may provide a venue for screening and education regarding healthy lifestyles and behaviors. Research has shown that introducing wellness programs in the workplace can have benefits not only for the individual and their ability to maintain a healthy lifestyle and reduce risks for certain disease, but also as a way for employers

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 8/12

113 Prevalence and Knowledge Assessment of HIV and NCDs in Namibia

to reduce costs related to absenteeism and lost productivity from illness[6,13,37]. Evidence for all three conditions shows that early intervention after diagnosis can significantly help prevent complications and progression[15,17,20], especially in an asymptomatic phase. These data could also be used to encourage employers to expand existing workplace programs focused on one disease to include education on other conditions and risk behaviors.

Limitations Because this is a cross-sectional survey, it is impossible to determine causation and only associations between independent variables and the outcomes (knowledge) can be ascertained. In addition, the survey was meant as an awareness and management information exercise for companies and was not powered to detect particular associations. This increases the potential for type II statistical errors or the probability of finding a significant result when one does not actually exist. In addition, the screening procedures are not the best standard in studies for prevalence of diabetes, hypertension, or HIV. A random blood glucose test, in particular, is the least sensitive measure and may be subject to large variations depending on when the subject last ate, their level of physical activity, and other factors. As a result, the screening results for high blood glucose should be interpreted with caution and are only an indication of possible elevated blood glucose rather than providing a basis for studying prevalence. Similarly, diagnostic standards for hypertension require that blood pressure be measured on two separate occasions to confirm a diagnosis. The scope of the wellness survey, as well as the resources available for the study precluded more rigorous screening measures. These do, however, represent the first indication of measures of risk factors for NCDs in a large Namibian population as far as the authors are aware. Because participation was voluntary, there may be a selection bias, which would make results not applicable to the general population. No particular adjustments were made for missing data as they were not found to be systematically distributed throughout the sample. None of the variables had a greater than 5% proportion of missing data. While this survey collected information on those who refused to learn their results, it did not provide information on those who refused to have the test done at all. Given the sample size of this largest survey amongst Namibian formal sector employees, the results could be interpreted as the best available representative health data in the Namibian formal sector to date.

Conclusions Despite increasing awareness of chronic conditions in sub-Saharan Africa, knowledge among formal sector workers in Namibia is poor. Opportunities exist for improving the knowledge and risk perception of NCDs and HIV among formal sector employees in this country. How- ever, more research is needed to ascertain prevalence of these conditions and to determine the most effective strategies for management and prevention.

Acknowledgments The authors acknowledge the role of the Bophelo! partners contribution, as listed below, and to the Bophelo! programme for the collection of the data utilized for this study: Namibia Ministry of Health and Social Services—for approving the Bophelo! data collection protocol and providing permission to conduct and publish this study; Namibia Institute of Pathology—for providing the quality assurance of the medical screening process;

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 9/12

114 Prevalence and Knowledge Assessment of HIV and NCDs in Namibia

Namibia Business Coalition on AIDS—for facilitating the Bophelo! intervention in the participating workplaces; The private sector companies who contracted the Bophelo! services; The employees of the companies who participated in the Bophelo! survey; The Namibia Global Fund Program which provided funding for the HIV screening of the Bophelo! program The Dutch Foreign Ministry for providing funding through MFS to fund the technical assistance provided by PharmAccess Foundation in the management of the Bophelo! programme.

Author Contributions Conceived and designed the experiments: IDB TFRDW. Performed the experiments: PM RH. Analyzed the data: LG RH PM FF. Wrote the paper: LG IDB PM FF TFRDW.

References 1. WHO | World Health Organization—Global Health Observatory [Internet]. WHO. [cited 2015 Mar 3]. Available: http://www.who.int/gho/ncd/en/. 2. Kirigia JM, Sambo HB, Sambo LG, Barry SP. Economic burden of diabetes mellitus in the WHO African region. BMC Int Health Hum Rights. 2009; 9(1):6. 3. UNAIDS. Namibia Country Report [Internet]. [cited 2015 Mar 3]. Available: http://www.unaids.org/en/ regionscountries/countries/namibia/. 4. Pegus C, Bazzarre TL, Brown JS, Menzin J. Effect of the Heart At Work program on awareness of risk factors, self-efficacy, and health behaviors. J Occup Environ Med Am Coll Occup Environ Med. 2002 Mar; 44(3):228–36. 5. Loeppke R, Nicholson S, Taitel M, Sweeney M, Haufle V, Kessler RC. The impact of an integrated population health enhancement and disease management program on employee health risk, health conditions, and productivity. Popul Health Manag. 2008 Dec; 11(6):287–96. doi: 10.1089/pop.2008.0006 PMID: 19108644 6. Milani RV, Lavie CJ. Impact of worksite wellness intervention on cardiac risk factors and one-year health care costs. Am J Cardiol. 2009 Nov 15; 104(10):1389–92. doi: 10.1016/j.amjcard.2009.07.007 PMID: 19892055 7. Rosen S, Vincent JR, MacLeod W, Fox M, Thea DM, Simon JL. The cost of HIV/AIDS to businesses in southern Africa. AIDS Lond Engl. 2004 Jan 23; 18(2):317–24. 8. Sarafidis PA, Lasaridis A, Gousopoulos S, Zebekakis P, Nikolaidis P, Tziolas I, et al. Prevalence, awareness, treatment and control of hypertension in employees of factories of Northern Greece: the Naoussa study. J Hum Hypertens. 2004 Sep; 18(9):623–9. PMID: 15029221 9. Collins JJ, Baase CM, Sharda CE, Ozminkowski RJ, Nicholson S, Billotti GM, et al. The assessment of chronic health conditions on work performance, absence, and total economic impact for employers. J Occup Environ Med Am Coll Occup Environ Med. 2005 Jun; 47(6):547–57. 10. Prabhakaran D, Shah P, Chaturvedi V, Ramakrishnan L, Manhapra A, Reddy KS. Cardiovascular risk factor prevalence among men in a large industry of northern India. Natl Med J India. 2005 Apr; 18 (2):59–65. PMID: 15981439 11. Clark MJ, Curran C, Noji A. The effects of community health nurse monitoring on hypertension identification and control. Public Health Nurs Boston Mass. 2000 Dec; 17(6):452–9. 12. Marseille E, Saba J, Muyingo S, Kahn JG. The costs and benefits of private sector provision of treatment to HIV-infected employees in Kampala, Uganda. AIDS Lond Engl. 2006 Apr 4; 20(6):907–14. 13. Perez AP, Phillips MM, Cornell CE, Mays G, Adams B. Promoting dietary change among state health employees in Arkansas through a worksite wellness program: the Healthy Employee Lifestyle Program (HELP). Prev Chronic Dis. 2009 Oct; 6(4):A123. PMID: 19754999 14. Archer WR, Batan MC, Buchanan LR, Soler RE, Ramsey DC, Kirchhofer A, et al. Promising practices for the prevention and control of obesity in the worksite. Am J Health Promot AJHP. 2011 Feb; 25(3): e12–26.

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 10 / 12

115 Prevalence and Knowledge Assessment of HIV and NCDs in Namibia

15. Lindström J, Eriksson JG, Valle TT, Aunola S, Cepaitis Z, Hakumäki M, et al. Prevention of Diabetes Mellitus in Subjects with Impaired Glucose Tolerance in the Finnish Diabetes Prevention Study: Results From a Randomized Clinical Trial. J Am Soc Nephrol. 2003 Jul 1; 14(suppl 2):S108–S113. 16. Lee L-L, Watson MC, Mulvaney CA, Tsai C-C, Lo S-F. The effect of walking intervention on blood pressure control: a systematic review. Int J Nurs Stud. 2010 Dec; 47(12):1545–61. doi: 10.1016/j.ijnurstu. 2010.08.008 PMID: 20863494 17. Huang S, Hu X, Chen H, Xie D, Gan X, Wu Y, et al. The positive effect of an intervention program on the hypertension knowledge and lifestyles of rural residents over the age of 35 years in an area of China. Hypertens Res Off J Jpn Soc Hypertens. 2011 Apr; 34(4):503–8. 18. Totsikas C, Röhm J, Kantartzis K, Thamer C, Rittig K, Machann J, et al. Cardiorespiratory fitness deter- mines the reduction in blood pressure and insulin resistance during lifestyle intervention. J Hypertens. 2011 Jun; 29(6):1220–7. doi: 10.1097/HJH.0b013e3283469910 PMID: 21505353 19. Albarracín D, Gillette JC, Earl AN, Glasman LR, Durantini MR, Ho M-H. A test of major assumptions about behavior change: a comprehensive look at the effects of passive and active HIV-prevention interventions since the beginning of the epidemic. Psychol Bull. 2005 Nov; 131(6):856–97. PMID: 16351327 20. Green EC, Halperin DT, Nantulya V, Hogle JA. Uganda’s HIV prevention success: the role of sexual behavior change and the national response. AIDS Behav. 2006 Jul; 10(4):335–346; discussion 347– 350. PMID: 16688475 21. Fisher HH, Patel-Larson A, Green K, Shapatava E, Uhl G, Kalayil EJ, et al. Evaluation of an HIV prevention intervention for African Americans and Hispanics: findings from the VOICES/VOCES Community- based Organization Behavioral Outcomes Project. AIDS Behav. 2011 Nov; 15(8):1691–706. doi: 10. 1007/s10461-011-9961-7 PMID: 21573724 22. Namibia Statistics Agency. The Namibia Labour Workforce Survey 2012 Report. Windhoek; 2013. 23. Cooper M, Harding S, Mullen K, O’Donnell C. “A chronic disease is a disease which keeps coming back ...it is like the flu”: chronic disease risk perception and explanatory models among French- and Swahili-speaking African migrants. Ethn Health. 2012 Dec 1; 17(6):597–613. doi: 10.1080/13557858. 2012.740003 PMID: 23153320 24. Hall V, Thomsen RW, Henriksen O, Lohse N. Diabetes in Sub Saharan Africa 1999–2011: Epidemiol- ogy and public health implications. a systematic review. BMC Public Health. 2011 Jul 14; 11(1):564. 25. Dzudie A, Kengne AP, Muna WFT, Ba H, Menanga A, Kouam Kouam C, et al. Prevalence, awareness, treatment and control of hypertension in a self-selected sub-Saharan African urban population: a cross- sectional study. BMJ Open. 2012; 2(4). 26. Guariguata L, Whiting DR, Beagley J, Linnenkamp U, Hambleton I, Cho NH, et al. Global estimates of diabetes prevalence in adults for 2013 and projections for 2035 for the IDF Diabetes Atlas. Diabetes Res Clin Pract. 2013 Dec; 27. Akwara PA, Madise NJ, Hinde A. PERCEPTION OF RISK OF HIV/AIDS AND SEXUAL BEHAVIOUR IN KENYA. J Biosoc Sci. 2003 Jul; 35(03):385–411. 28. Fenton KA, Chinouya M, Davidson O, Copas A. HIV testing and high risk sexual behaviour among Lon- don’s migrant African communities: a participatory research study. Sex Transm Infect. 2002 Aug 1; 78 (4):241–5. PMID: 12181459 29. Prata N, Morris L, Mazive E, Vahidnia F, Stehr M. Relationship between HIV Risk Perception and Con- dom Use: Evidence from a Population-Based Survey in Mozambique. Int Fam Plan Perspect. 2006 Dec 1; 32(4):192–200. PMID: 17237016 30. Abubakari A-R, Jones MC, Lauder W, Kirk A, Anderson J, Devendra D, et al. Ethnic differences and socio-demographic predictors of illness perceptions, self-management, and metabolic control of type 2 diabetes. Int J Gen Med. 2013 Jul 29; 6:617–28. doi: 10.2147/IJGM.S46649 PMID: 23935384 31. Mayega RW, Etajak S, Rutebemberwa E, Tomson G, Kiguli J. “Change means sacrificing a good life”: perceptions about severity of type 2 diabetes and preventive lifestyles among people afflicted or at high risk of type 2 diabetes in Iganga Uganda. BMC Public Health. 2014 Aug 21; 14(1):864. 32. Commodore-Mensah Y, Samuel LJ, Dennison-Himmelfarb CR, Agyemang C. Hypertension and overweight/obesity in Ghanaians and Nigerians living in West Africa and industrialized countries: a systematic review. J Hypertens. 2014 Mar; 32(3):464–72. doi: 10.1097/HJH.0000000000000061 PMID: 24445390 33. Freak-Poli R, Wolfe R, Peeters A. Risk of cardiovascular disease and diabetes in a working population with sedentary occupations. J Occup Environ Med Am Coll Occup Environ Med. 2010 Nov; 52 (11):1132–7. 34. Kolbe-Alexander T, Greyling M, da Silva R, Milner K, Patel D, Wyper L, et al. The relationship between workplace environment and employee health behaviors in a South African workforce. J Occup Environ Med Am Coll Occup Environ Med. 2014 Oct; 56(10):1094–9.

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 11 / 12

116 Prevalence and Knowledge Assessment of HIV and NCDs in Namibia

35. Rula EY, Hobgood A. The impact of health risk awareness on employee risk levels. Am J Health Behav. 2010 Oct; 34(5):532–43. PMID: 20524883 36. Loprinzi PD. Factors influencing the disconnect between self-perceived health status and actual health profile: implications for improving self-awareness of health status. Prev Med. 2015 Apr; 73:37–9. doi: 10.1016/j.ypmed.2015.01.002 PMID: 25584985 37. Guariguata L, de Beer I, Hough R, Bindels E, Weimers-Maasdorp D, Feeley FG, et al. Diabetes, HIV and other health determinants associated with absenteeism among formal sector workers in Namibia. BMC Public Health. 2012; 12:44. doi: 10.1186/1471-2458-12-44 PMID: 22257589

PLOS ONE | DOI:10.1371/journal.pone.0131737 July 13, 2015 12 / 12

117 118 11 DIABETES, HIV AND OTHER HEALTH DETERMINANTS ASSOCIATED WITH ABSENTEEISM AMONG FORMAL SECTOR WORKERS IN NAMIBIA

119 Guariguata et al. BMC Public Health 2012, 12:44 http://www.biomedcentral.com/1471-2458/12/44

RESEARCH ARTICLE Open Access Diabetes, HIV and other health determinants associated with absenteeism among formal sector workers in Namibia Leonor Guariguata1*, Ingrid de Beer2, Rina Hough2, Els Bindels2, Delia Weimers-Maasdorp2, Frank G Feeley III3 and Tobias F Rinke de Wit1,4

Abstract Background: As countries in sub-Saharan Africa develop their economies, it is important to understand the health of employees and its impact on productivity and absenteeism. While previous studies have assessed the impact of single conditions on absenteeism, the current study evaluates multiple health factors associated with absenteeism in a large worker population across several sectors in Namibia. Methods: From March 2009 to June 2010, PharmAccess Namibia conducted a series of cross-sectional surveys of 7,666 employees in 7 sectors of industry in Namibia. These included a self-reported health questionnaire and biomedical screenings for certain infectious diseases and non-communicable disease (NCD) risk factors. Data were collected on demographics, absenteeism over a 90-day period, smoking behavior, alcohol use, hemoglobin, blood pressure, blood glucose, cholesterol, waist circumference, body mass index (BMI), HIV status, and presence of hepatitis B antigens and syphilis antibodies. The associations of these factors to absenteeism were ascertained using negative binomial regression. Results: Controlling for demographic and job-related factors, high blood glucose and diabetes had the largest effect on absenteeism (IRR: 3.67, 95%CI: 2.06-6.55). This was followed by anemia (IRR: 1.59, 95%CI: 1.17-2.18) and being HIV positive (IRR: 1.47; 95%CI: 1.12-1.95). In addition, working in the fishing or services sectors was associated with an increased incidence of sick days (IRR: 1.53, 95%CI: 1.23-1.90; and IRR: 1.70, 95%CI: 1.32-2.20 respectively). The highest prevalence of diabetes was in the services sector (3.6%, 95%CI:-2.5-4.7). The highest prevalence of HIV was found in the fishing sector (14.3%, 95%CI: 10.1-18.5). Conclusion: Both NCD risk factors and infectious diseases are associated with increased rates of short-term absenteeism of formal sector employees in Namibia. Programs to manage these conditions could help employers avoid costs associated with absenteeism. These programs could include basic health care insurance including regular wellness screenings. Keywords: Absenteeism, Namibia, Work force, Sub-Saharan Africa, Non-communicable disease, HIV

Background case for many low- and middle-income countries where Absenteeism due to health factors is mainly studied in employees must pay for care directly out-of-pocket. As high-income countries. In many of these countries, a result, absenteeism in the workforce in low- and mid- social mechanisms such as health insurance and national dle-income countries may place a greater burden on health systems are in place to support and care for employees who do not have access to health care, as employees who are faced with illness. This is not the well as to employers who must bear the burden of lost productivity.

* Correspondence: [email protected] As low and middle-income countries work to compete 1PharmAccess Foundation, Trinity Building C, Pietersbergweg 17, 1105, BM in the global market, the health of their workforce is Amsterdam, Zuidoost, the Netherlands increasingly vital to enhance productivity. For many Full list of author information is available at the end of the article

© 2012 Guariguata et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

120 Guariguata et al. BMC Public Health 2012, 12:44 Page 2 of 12 http://www.biomedcentral.com/1471-2458/12/44

sub-Saharan African countries, infectious diseases like from this paper will provide evidence that may help fill HIV/AIDS are still having a serious impact despite the that gap and drive further research in this area. large-scale campaigns for prevention, care and anti-ret- Namibia represents a country in economic and epide- roviral therapy (ART) [1]. Non-communicable diseases miologic transition in sub-Saharan Africa. It has one of (NCDs) are playing an important role, as changing diets the region’s highest rates of gross-national income per and sedentary habits are adopted, particularly in urban capita at USD 4,210 in 2008, compared to an average of settings [2]. Conditions like cardiovascular disease and just USD 1,082 for all of sub-Saharan Africa [19]. How- diabetes are increasingly affecting countries in economic ever, the country is also characterized by one of the transition and are expected to increase dramatically over highest income disparities in the world [19]. While the next 20 years [3,4]. Despite these trends, funding to Namibia posts better than average performance on a prevent and manage NCDs is significantly behind that number of economic measures when compared to the of communicable disease [5]. This is remarkable in light rest of sub-Saharan Africa [19], it has one of the highest of the fact that cost-effective and cost-saving interven- national rates of HIV/AIDS at 13.3% [1]. The govern- tions have been shown to prevent NCDs in developing ment has engaged in a substantial program to provide countries [6] and are adaptable to the workplace. anti-retroviral therapy at a low cost and improve access For many sub-Saharan African countries, unlike high- to care [1]. This has resulted in almost universal cover- income countries which have many more social support age of anti-retroviral therapy. The majority of health mechanisms in place, individuals pay out-of-pocket for services in Namibia are concentrated in a few cities in their own care [7-9] which creates a significant barrier the north and center of the country [20]. This leaves a to access. As a result, individuals in the workforce faced portion of the population without easy access to basic with the high cost of managing disease have few care. While the Ministry of Health and Social Services options. There is some evidence of employers filling this (MOHSS) has developed and implemented a program to gap by providing basic health care in the workplace improve access to care for more remote regions [20] it [10]. Employers must balance the provision of health can still be a long distance to travel. As a consequence, services and health insurance to their workers while large employers often provide basic healthcare services maintaining profitability, taking into account the double on site. In addition, some employers offer (co)payment burden of infectious and non-infectious chronic dis- of healthcare insurances for their employees (and depen- eases. If the worker is not paid for some or all of the dents) through one of the existing Namibian medical aid sick time, household income falls at the same time that schemes [10]. However, provision of these services is medical care costs may increase. Furthermore, sickness based on the perceived needs of the employee and related absenteeism is an indicator of the prevalence employer rather than on actual evidence of disease risks and severity of health conditions in the work force, an and determinants. indicator that is not collected through the health system. This study is based on the results of wellness surveys Loss of productivity due to health-related absenteeism that were performed at the request of employers from 7 remains a barrier to productivity and thus to economic different Namibian industry sectors. These wellness sur- development. Gathering information regarding employee veys are the most extensive yet in that country. The health can help employers make informed decisions to industry sectors assessed were retail, agriculture, fishing, keep their workforce healthy and productive and reduce services, wholesale, tourism and transport. In addition, absenteeism as well as improving access to care for the survey provides information on the prevalence of individuals. certain risk factors and conditions among different Current research into absenteeism and its determi- industries in the Namibian private sector. The primary nants, concentrated in high-income countries, has found goals of this paper are: that with regards to health, NCDs are contributing significantly to increased sick leave [11-14]. However, simi- 1. Describe the demographic and work profile of lar research among employees in sub-Saharan Africa has participants including industry, contract type, job focused primarily on the contribution of infectious dis- type, age, gender, and the prevalence of risk factors ease and in particular HIV/AIDS [15-17]. In addition, and conditions such as hypertension, diabetes, HIV, since HIV/AIDS has largely affected those of working hepatitis B, anemia and syphilis in the Namibia for- age, it is important to understand the impact of being mal sector. infected on employees and the workplace. In a previous 2. Describe the associations of the above determi- study we reported on the HIV prevalence amongst the nants to self-reported sickness absence. workforce in Namibia [18]. There is a gap in research looking more generally at the determinants of absentee- Through seeking to answer these questions, this paper ism among workers in sub-Saharan Africa. The findings aims to understand the determinants of absenteeism in

121 Guariguata et al. BMC Public Health 2012, 12:44 Page 3 of 12 http://www.biomedcentral.com/1471-2458/12/44

the Namibian formal sector in order to improve the information (contract type, type of position), and self- provision of healthcare to Namibian workers. reported presence of the following conditions: HIV/ AIDS, diabetes, and hypertension. In addition, partici- Methods pants who agreed received a medical screening including From 2009 to 2010, a voluntary cross-sectional wellness risk factors (haemoglobin, systolic and diastolic blood survey and health screening of workers was conducted pressure and anthropometric markers: height, weight by PharmAccess Foundation at the request of 42 com- and waist circumference) and prevalence of diseases panies across seven industry sectors in Namibia. Partici- (HIV status, hepatitis B antigens, syphilis, random blood pation in the screening and survey was entirely glucose for diabetes). A finger prick was conducted on voluntary and strictly confidential. To protect confiden- all participants to collect a blood sample for the follow- tiality, no identifiers were stored with the information ing rapid tests: HIV, syphilis antigen, hepatitis B antigen, gathered and participants gave written informed consent haemoglobin and blood glucose. Participants were before inclusion. allowed to refuse any or all of these screenings. The survey was performed by two mobile clinics from The following assays and methods were used for col- the Bophelo! Project, a partnership between PharmAc- lecting medical screening information according to man- cess, the Namibian Business Coalition on AIDS and the ufacturer instructions. Any test with an inconclusive or Namibia Institute of Pathology. As part of the project, discordant result was repeated. For HIV testing, the wellness surveys and health screening services are Determine HIV 1/2 Assay by Inverness rapid test strip offered to companies based in Namibia along with sensi- and Trinity Biotech Uni -Gold HIV 1/2 test kit were tization and information sessions provided before used in conjunction for HIV testing, according to Nami- employee screenings. The mobile clinic visits a company bian VCT regulations. If the result was discordant site for as long as required to see all participating between the two tests, a third (tie-breaker) test was per- employees (up to 32 people per van per day) and each formed using the Inverness Clearview Complete HIV 1/ screening visit takes an average of 30 min for pre- test 2 test kit. Hepatitis B testing was done using the Deter- counseling, testing, and post- test counseling sessions. mine HBsAg Whole Blood Assay rapid test strip (Inver- All services are provided by MOHSS-trained testers and ness Medical, USA) was used. The Determine Syphilis counselors, as part of a national task shifting effort. TP Whole Blood Assay (Inverness Medical, USA) was While screening results are being obtained, the partici- used for presence of syphilis antibodies. The Accutrend pant receives health education and confidential counsel- Plus GCT meters were used to test blood glucose ing. Participants who tested positive for any condition (Roche Molecular Diagnostics, USA). HemoCue Hb 201 or have screening results outside of a normal range are + Analyser and HemoCue Hb 201+ microcuvettes encouraged to seek follow-up consultation and are (HemoCue, USA) were used to conduct haemoglobin issued a referral letter to local medical services. Only tests. Systolic and diastolic blood pressure was measured de-identified information was collected from the screen- in millimetres of mercury (mmHg) by using the ings without any possibility of linking back to the parti- MG150f Digital Blood Pressure monitor (Rossmax Inter- cipant. The study was approved by the Ministry of national Ltd, Taiwan). The measurement was taken Health and Social Services of Namibia (MOHSS). three times on each participant and the average of the A total of 10,701 employees were offered the possibi- three readings was used. A blood pressure reading lity to participate in the survey across all companies; above 140/90 mmHg was considered elevated; above 8,355 provided written informed consent, of which 100 153/103 mmHg was considered high and referral was were excluded from analysis because of pregnancy, as made to a medical facility. thiswasconsideredtobearisk for potential bias for Bodyweight was measured with light clothes on, with- absenteeism. 589 were excluded because of being out shoes on an analogue scale and the recorded value younger than 20 years of age which could also bias find- was rounded to the nearest 0.5 kg. Height was measured ings on absenteeism. In summary, 7,666 participants in centimeters using a non-stretch retractable measure- were included in this analysis. Participants were asked ment tape to the nearest 0.5 cm. These were used to to recall how many days they had missed work due to calculate body mass index according to the formula sickness in the 90 days prior to administration of the below. survey. To preserve confidentiality, this measure was weight kg based entirely on participant recall and not compared to BMI = 2 2 employer records. height m Participants who provided informed consent were asked to complete a questionnaire with demographic A BMI of 25-30 was considered overweight; above 30: (age, sex, marital status, smoking), job status obese. Waist circumference was measured at the level of

122 Guariguata et al. BMC Public Health 2012, 12:44 Page 4 of 12 http://www.biomedcentral.com/1471-2458/12/44

the umbilicus in standing position with the hands by the Outcome of interest side and the feet together at the end of a normal expira- The outcome of interest used for all univariate and tion. A no-stretch retractable tape was used. Measure- expanded analyses was a self-reported measure collected ments were rounded to the nearest 0.1 cm. A waist from the survey. Participants were asked to name the circumference above 102 cm in men and 88 cm in number of days they had been absent from work in the women was considered above normal and indicative of previous 90 day period due to health reasons. central obesity. Only aggregate results were reported back to the Modeling employer. Individual screening results were given All univariate analyses were done using non-parametric directly to the participant. Based on individual partici- tests to allow for the non-normal distribution of the pant results, the person was advised to make an outcome and controlled for by industry. Initially, Pois- appointment with a clinician for confirmation and son’s regression models were fit to the data and tested further medical follow-up. In addition, participants with using likelihood ratio tests for over dispersion against a screening results outside the above indicated normal negative binomial models. This test consistently found values or who tested positive for any condition were that the Poisson’s model was significantly over-dispersed given a letter of referral for a follow-up visit with a clini- with respect to the negative binomial model. Thus, cian including a description of the findings of the negative binomial regression models were used for both screening. the univariate models and expanded models in agree- Two variables were created based on a combination of ment with published standards for the analysis of absen- findings from clinical screening and participant teeism data [21]. In univariate and expanded analyses, responses. Diabetes was classified as known diabetes or incidence rate ratios (IRR) for sick days were calculated those whose random blood glucose was 11.1 mmol/L or to measure the strength of association for independent higher. Similarly, hypertension was classified as known variables to the outcome. For all statistical tests, an hypertension, as well as those who were found at alpha of 0.05 or less was considered statistically signifi- screening to have systolic blood pressure above 154 cant. Univariate models used number of sick days in the mmHg or diastolic blood pressure above 104 mmHg. last 90 days as the outcome and only a single predictor Data were collected and stored using SPSS. All partici- (each risk factor, demographic variable, and condition). pants received counseling and information on their Expanded models used number of sick days in the last results and were invited to speak with a trained health 90 days as the outcome and assessed the impact of risk professional or to schedule a follow-up with a physician. factors controlling for industry, gender, age, job status, and contract type. Statistical analysis We constructed dummy variables for the industry All statistical analyses were conducted using R Project variable in order to compare the rates of sickness for Statistical Computing version 2.10.0 (http://www.r- absence for that variable in regression analyses. The project.org). 890 participants had missing values for the retail sector was chosen as a control group simply as a outcome variable and were thus excluded from univari- matter of convenience. The survey was not designed ate and expanded analyses. In many cases, missing data and implemented with a particular pre-selected control was mostly a result of companies refusing to provide a group. The retail industry was chosen as the control particular screening to employees. These companies group post-hoc because it was the largest and because it were then dropped from the corresponding part of the did not have extreme values for the prevalence of any of analysis. There were no significant differences in demo- the screening-related conditions. The control group was graphic and working sector variables for those with used only for analysis regarding industry sector only. missing data compared with those included in the analysis. Descriptive statistics were calculated based on an Results intent-to-treat model where all those included with suf- Demographic and job-related information ficient outcome data were analyzed. However, in most A description of demographic variables is presented in cases, results on screenings were not available for some Table 1 including all 7,666 participants. For all sectors participants as a result of the choice of companies not combined, the majority of participants were men (65.3%, to offer that screening, rather than the choice of the 95%CI: 64.2-66.4), and the mean age of participants was employee not to undergo the test. Differences between 36.3 (95%CI: 36.0-36.5) years. The majority of employees industries for continuous variables were tested using were working in non-administrative labor positions one-way analysis of variance. Chi-squared analysis was (78.6%, 95%CI: 77.2-79.0) and had permanent contracts used to test differences between industries for categori- (70.5%, 95%CI: 69.4-71.5). Across all companies, the cal variables. mean number of sick days was 0.92 (95%CI: 0.83-1.01)

123 Guariguata et al. BMC Public Health 2012, 12:44 Page 5 of 12 http://www.biomedcentral.com/1471-2458/12/44

Table 1 Distribution of demographic and job-related hotel and restaurants industry with a mean age of 31.2 information study participants and the oldest population worked in the electricity, gas Categorical variables n % 95% CI and water sector (mean age 40.0). There were also sig- Sex nificant differences in the proportion of males and 2 Male 5,005 65.3% 64.2-66.4 females by industry (c = 411.9, d.f. = 6, p < 0.0001), job 2 Female 2,661 34.7% 33.6-35.8 type (c = 736.5, d.f. = 6, p < 0.0001) and contract type (c2 = 891.6, d.f. = 6, p < 0.0001). The fishing industry, Job Type one of the largest employers in Namibia, had signifi- Labour 5,986 78.1% 77.2-79.0 cantly fewer employees on permanent contracts and the Administration 1,626 21.2% 20.3-22.1 vast majority working in labour jobs. The electricity, gas Contract Type and water industry had the highest rate of sick days Permanent 5,403 70.5% 69.4-71.5 (1.2) followed by the fishing sector (1.1) while the agri- Part-time/Short-term 2,263 29.5% 28.5-30.5 culture sector reported the lowest number of sick days Industry (0.5). The agriculture, fishing, electricity, gas and water, Retail and Trade 2,086 27.2% 26.2-28.2 and wholesale and retail trade industries all had signifi- Agriculture 465 6.1% 5.5-6.6 cantly higher sick day rates than the retail sector which was used as a reference group. Fishing 1,869 24.4% 23.4-25.3 Electricity, gas & water 1,140 14.9% 14.1-15.7 Risk factors Wholesale trade and repair of motor 1,031 13.4% 12.7-14.2 vehicles The majority reported they had never smoked (82.1%, 95%CI: 79.8-81.6), almost one third of participants Hotels and Restaurants 300 3.9% 3.5-4.3 (27.0%) presented with an elevated or high blood pres- Transport, Storage and 775 10.1% 9.4-10.8 Communication sure, many (39.2%) had a BMI of 25 or greater and 18.7% (95%CI: 17.8-19.6) had central obesity for their Sick absence gender (Table 3). Any absence 1,431 18.6% 17.6-19.7 Analysis by industry shows that in terms of risk fac- 5 or more days absent 334 4.3% 3.9-4.8 tors, the fishing, agriculture and electricity, gas and Continuous Variables n 95% CI water industries all had a prevalence of employees with Age elevated or high blood pressure above 28%. The preva- Mean 36.3 36.0-36.5 lence of elevated or high blood pressure was signifi- 2 Median 35 cantly different among industries (c = 100.0, d.f. = 6, p Range 20- < 0.0001) The lowest prevalence of abnormal blood 79 pressure was found in the wholesale and retail trade and Missing data 0 repair of motor vehicles industry which was still 17.1% Sick (Table 2). In addition, at least a quarter of employees in Days all industries were found to be overweight or obese. The Mean 0.92 0.83-1.01 majority of employees across all industries did report Median 0 smoking, although the agriculture industry did report Range 0-90 the highest prevalence of smokers at 34.8%. Missing data 890 Conditions and diseases The overall prevalence of HIV in participants was 9.1% in the 90 days previous to administration of the survey (95%CI: 8.4-9.7). Prevalence of antibodies against hepati- with a skewing towards few to no sick days. A total of tis B surface antigen (HbsAg) was 6.8% (95%CI: 6.3-7.4) 18.6% of all employees surveyed reported having at least while prevalence of syphilis antibodies was 1.1% (95%CI: one sick day with 4.3% reporting having missed five or 0.9-1.4). In addition, 1.7% (95%CI: 1.0-2.4) had diabetes more days. The greatest percentage of absences due to according to accepted guidelines [22]. sickness was in the fishing industry (5.6%) (Table 2) fol- There was more variation in conditions and disease lowed by the retail and trade industry (5.2%) with the prevalence between industries than those found for risk lowest percentage of absences in the agriculture and factors. The prevalence of hypertension ranged from hotels and restaurants industries (0.6%). 10.7% in the hotels and restaurants industry to 21.7% in There was a significant difference (F statistic = 72.531, the fishing industry. Similarly, the hotels and restaurants D.F. = 6, p < 0.0001) in ages by industry, presented in industry had the lowest prevalence of diabetes at just Table 2, where the youngest population worked in the 0.7% compared to 3.6% at the electricity, gas and water

124 http://www.biomedcentral.com/1471-2458/12/44 Table 2 Demographic, risk factor, and disease information by industry Guariguata Demographic Information Age Sex Job type Contract type Sick days Sickness Industry n Mean Male Female Labour Administration Permanent Part-time/ Mean Standard Any More than 5 al et

Short-term deviation absence days absent . M ulcHealth Public BMC Retail and Trade 2,086 33.5 63.0% 37.0% 75.2% 23.9% 70.5% 29.5% 0.7 3.2 5.2% 1.0% Agriculture 465 37.1 13.7% 38.7% 91.4% 8.2% 90.5% 9.5% 0.5 2.7 0.6% 0.0% Fishing 1,869 37.6 46.5% 48.0% 93.2% 6.4% 45.9% 54.1% 1.1 4.7 5.6% 1.3% Electricity, gas & water 1,140 40.0 43.1% 21.1% 61.8% 37.7% 82.8% 17.2% 1.2 4.6 3.6% 1.0%

Wholesale trade and 1,031 36.4 37.8% 23.6% 81.4% 17.8% 81.9% 18.1% 1.0 3.8 3.1% 0.8% 2012, repair of motor vehicles

Hotels and Restaurants 300 31.2 6.7% 53.7% 81.0% 18.3% 63.3% 36.7% 0.7 4.0 0.6% 0.2% 12 Transport, Storage and 775 36.2 29.1% 21.5% 60.0% 38.8% 87.2% 12.8% 0.9 2.7 2.5% 0.6% :44 Communication Risk Factors Blood Pressure BMI Waist Smoking behaviour Circumference Industry n Normal (< Elevated or High (140/ High (≥ 154/ Underweight Normal (18.5- Overweight Obese (≥ Normal Above Never Smokes 140/90 mm/ 90-153/103 mm/Hg) 104 mm/Hg) (< 18.5) 24.9) (25-29.9) 30) limit smoked Hg) Retail and Trade 2,086 72.3% 17.4% 10.1% 7.0% 57.9% 21.4% 12.4% 86.2% 13.3% 81.8% 16.1% Agriculture 465 71.0% 19.8% 8.8% 17.2% 58.7% 13.5% 10.3% 88.0% 11.8% 64.7% 34.8% Fishing 1,869 67.3% 18.4% 13.9% 3.7% 50.5% 26.6% 18.9% 77.8% 21.7% 85.0% 13.1% Electricity, gas & water 1,140 71.8% 18.9% 9.1% 4.4% 46.9% 28.9% 18.9% 80.1% 18.9% 82.5% 15.5% Wholesale trade and 1,031 82.3% 12.0% 5.7% 6.7% 64.6% 18.3% 9.9% 88.2% 11.5% 81.0% 17.4% repair of motor vehicles Hotels and Restaurants 300 79.3% 13.7% 7.0% 6.7% 48.7% 28.3% 16.3% 79.3% 20.7% 78.76% 19.0% Transport, Storage and 775 74.8% 15.7% 9.2% 5.4% 50.3% 28.5% 15.4% 82.3% 17.4% 74.6% 24.6% Communication Conditions and Diseases Industry n Hypertension Diabetes Anemia HIV Hepatitis B Syphilis Retail and Trade 2,086 14.0% 0.9% 6.5% 8.3% 7.3% 0.8% Agriculture 465 12.9% 0.9% 3.9% 5.6% 8.0% 3.9% Fishing 1,869 21.7% 1.8% 10.2% 14.3% 4.2% 0.2% Electricity, gas & water 1,140 21.1% 3.6% 4.7% 8.2% 7.6% 1.6% Wholesale trade and 1,031 12.7% 1.6% 2.9% 7.9% 9.2% 1.7% repair of motor vehicles 12 of 6 Page Hotels and Restaurants 300 10.7% 0.7% 4.7% 5.7% 5.3% 0.0% Transport, Storage and 775 15.0% 1.9% 4.6% 4.6% 7.6% 1.9% Communication

125 Guariguata et al. BMC Public Health 2012, 12:44 Page 7 of 12 http://www.biomedcentral.com/1471-2458/12/44

Table 3 Distribution of findings from biomedical Univariate analysis screening Given the high variability of in demographic informa- Variable Levels n % 95% CI tion, risk factors, and prevalence of conditions and dis- Risk Factors eases between industries, univariate analysis of Blood Pressure independent variables to rate of sick days was conducted Normal (< 140/90 mm/Hg) 5,580 72.8% 71.8-73.8 controlling for industry sector and are presented as Crude IRR (incidence rate ratio) in Table 4. For demo- Elevated (140/90-153/103 mm/Hg) 1,301 17.0% 16.1-17.8 graphic information, having a part-time or fixed term ≥ High ( 154/104 mm/Hg) 767 10.0% 9.3-10.7 contract was significantly associated with a reduction in BMI sick days compared to those with permanent contracts. Underweight (< 18.5) 476 6.2% 5.7-6.7 In addition working in the fishing sector (IRR = 1.53, Normal (18.5-24.9) 4,160 54.3% 53.1-55.4 95%CI: 1.23-1.90), the services sector (IRR = 1.70, 95% Overweight (25-29.9) 1,851 24.1% 23.2-25.1 CI: 1.32-2.20) or the wholesale sector (IRR = 1.39, 95% Obese (≥ 30) 1,146 14.9% 14.1-15.7 CI: 1.07-1.79) were significantly associated with a higher Waist Circumference rate of sick days, while those in the agriculture sector Normal 6,195 80.8% 79.9-81.7 had significantly less sick days (IRR = 0.66, 95%CI: 0.46- 0.94) compared to retail, the reference group. Above limit 1,435 18.7% 17.8-19.6 No risk factor had a significant association with an Smoking behaviour increase in sick days. However, there were associations Smokes 1,346 17.6% 16.7-18.4 with respect to conditions and disease. Diabetes had the Never smoked 6,186 80.7% 79.8-81.6 highest effect size on the rate of sick days (IRR = 3.40, Conditions and diseases 95%CI: 1.91-6.04). HIV infection was also associated Hypertension with a significant increase in sick days (IRR = 1.55, 95% No 6,171 80.5% 79.6-81.4 CI: 1.19-1.74) as was anaemia (IRR:1.82, 95%CI: 1.21- Yes 1,277 16.7% 15.8-17.5 2.23). Diabetes No 6,788 88.5% 87.8-89.3 Expanded models Analysis of all health screenings were remodeled con- Yes 119 1.6% 1.3-1.8 trolling for industry, sex, age, job status, and contract Haemoglobin type. Results of these expanded models are also pre- Normal 5,654 73.8% 72.8-74.7 sented in Table 2 as Adjusted IRR (incidence rate ratio). Anaemia 479 6.2% 5.7-6.8 None of the covariates were found to be significant HIV effect modifiers, as most of the adjusted incidence rate Positive 694 9.1% 8.4-9.7 ratios remained close to the values found in univariate Negative 6,057 79.0% 78.1-80.0 analysis. Diabetes, elevated blood pressure, anaemia and Hepatitis B being HIV positive were the only conditions significantly Positive 524 6.8% 6.3-7.4 associated with absenteeism. Negative 5,621 73.3% 72.3-74.3 Discussion Syphilis This paper describes the association of multiple health Positive 88 1.1% 0.9-1.4 determinants with an important labor productivity vari- Negative 6,074 79.2% 78.3-80.1 able (absenteeism) in an African context. The industries represented are similar to those of the formal private industry (c2 = 30.1, d.f. = 6, p < 0.0001). The fishing sector of Namibia, although the sample is not represen- industry also had a high prevalence of anemia at 10.2% tative of the entire formal sector of the country. Agricul- and the highest prevalence of HIV at 14.3% compared ture is the largest industry in Namibia (15.9%), followed to just 4.6% for the transport, storage and communica- by employment in private households (10.9%) (not cap- tion industry. Prevalence rates of HIV varied signifi- tured in this study) and retail (9.1%) which was the lar- cantly between industries (c2 = 125.1, d.f. = 6, p < gest group represented by this study [23]. Other sectors 0.001). The prevalence of hepatitis B antigen was also not represented in this study include public administra- above 7% for all sectors except for fishing and the hotels tion, education, mining and real estate. However, other and restaurants industries. The prevalence of syphilis important industries such as services, retail, tourism, antigen was relatively low, although the agriculture sec- fishing, and agriculture were included. The distribution tor did have a prevalence of 3.9%. of participants among different sectors does not reflect

126 Guariguata et al. BMC Public Health 2012, 12:44 Page 8 of 12 http://www.biomedcentral.com/1471-2458/12/44

Table 4 Results of univariate and expanded models where the outcome is self-reported health-related absence from work in the previous 90 days Demographic Crude Rate Crude IRR p-value Adjusted Rate Adjusted IRR p-value Information Ratio (IRR) 95% CI Ratio (IRR) 95% CI Industry < 0.0003 Retail and Trade Agriculture 0.66 (0.46-0.94) 0.02 Fishing 1.53 (1.23-1.90) < 0.01 Electricity, gas & water 1.70 (1.32-2.20) < 0.01 Wholesale trade and repair of 1.39 (1.07-1.79) 0.01 motor vehicles Hotels and Restaurants 1.01 (0.66-1.55) 0.96 Transport, Storage and 1.20 (0.90-1.59) 0.21 Communication Contract Type Permanent Part-time/Short-term 0.74 (0.63-0.89) < 0.001 Job Type Labour Administration Work 1.03 (0.85-1.24) 0.80 Age Age 1.00 (0.99-1.01) 0.97 Sex Male Female 1.07 (0.90-1.26) 0.41 Crude Rate Crude IRR Adjusted Rate Adjusted IRR Risk Factors Ratio (IRR) 95% CI p-value Ratio (IRR) 95% CI p-value BMI BMI z-score 1.04 (0.96-1.13) 0.32 0.98 (0.90-1.07) 0.72 Waist Circumference Normal Above limit 1.21 (0.99-1.48) 0.06 1.10 (0.90-1.34) 0.42 Blood Pressure Normal (< 140/90 mm/Hg) Elevated (140/90-153/103 mm/ 0.75 (0.60-0.92) 0.01 0.74 (0.59-0.91) 0.01 Hg) High (≥ 154/104 mm/Hg) 1.10 (0.84-1.43) 0.50 1.03 (0.78-1.35) 0.85 Smoking Status Never smoked Smoker 1.09 (0.89-1.34) 0.40 1.15 (0.94-1.42) 0.18 Hypertension No Yes 1.14 (0.92-1.41) 0.22 1.07 (0.86-1.34) 0.55 Diabetes No Yes 3.40 (1.91-6.04) < 0.01 3.67 (2.06-6.55) < 0.01 Haemoglobin Normal Anaemia 1.82 (1.35-2.46) < 0.01 1.59 (1.17-2.18) < 0.01

127 Guariguata et al. BMC Public Health 2012, 12:44 Page 9 of 12 http://www.biomedcentral.com/1471-2458/12/44

Table 4 Results of univariate and expanded models where the outcome is self-reported health-related absence from work in the previous 90 days (Continued) HIV Negative Positive 1.55 (1.18-2.04) < 0.01 1.47 (1.12-1.95) 0.01 Hepatitis B Negative Positive 1.01 (0.75-1.37) 0.92 1.14 (0.85-1.55) 0.38 Syphilis Negative Positive 0.64 (0.31-1.37) 0.22 0.83 (0.40-1.69) 0.60 *adjusted IRR were calculated after controlling for industry, age, sex, job status, and contract type. All IRR were calculated by exponentiating the beta coefficients obtained from negative binomial regression the proportions present in the Namibian formal sector, strengthening the health system, access to care for Nami- but does cover a large range of activities. bians remains an area of concern. Public health facilities, Namibia enacted the Labour Act in 1992, which was which are the most accessible, are understaffed and revised in 2007 (Act 11, 2007). Both the old and new patients may have to pay out-of-pocket [27]. These bar- Labour Act gives all employees (permanent and short- riers to access may also be reducing absenteeism as term contracts) a right to sick leave. The Labour Act employees decide to continue working rather than trying states that during the first 12 months of employment, to seek care. employees who work 5 days per week accrue 1 day of Compared with current research on absenteeism in sick leave for every 5 weeks of employment. Employees the workplace, this study looks at a variety of health- that work 5 days per week are entitled to 30 working related determinants without focusing on any one con- days of sick leave and those working 6 days per week dition. It provides some insight into the prevalence of can take up to 36 working days of sick leave, in a 3 year different risk factors and conditions among workers in cycle. While the law may not be applied consistently various industries as well as the relative impact of those everywhere, the majority of companies comply in pro- factors on absenteeism. The majority of absenteeism viding paid sick leave [24]. A medical certificate is also research is focused on high-income countries [11-14], required for sick leave, which may act as a barrier to and highlights the effect of particular chronic conditions some seeking care given that they may have to pay out- on absenteeism. The findings of these studies consis- of-pocket. In addition, employers are required to pay tently find a strong association between NCDs, their employees full salary benefits for the allowed sick leave. risk factors, and increased absenteeism. For most absen- However, there are no studies published looking at teeism-related studies in high-income countries, the employer compliance with the law and the effect on recall period is longer and can be validated against employee absenteeism in Namibia. It is possible that employee records which help minimize recall bias. This employees may not use sick leave if they know they will level of information is not available for the current not be compensated or that there may be some threat study as with many studies looking at sickness work to their position from several absences. absence in a sub-Saharan context [9,15-17]. As a result, There are a number of reports examining absenteeism the true level of absenteeism in this population may be rates in Europe and the United States. These reports show underestimated as a result of recall bias or may be sub- average sick leave rates ranging from 5.1 days per year for ject to whether employees have access to basic care ser- employees in Europe to just over 8 days per year for vices provided by some of the employers, as with the employees in the United States and up to 12 days per year agriculture sector [10]. for public sector employees in those countries [25,26]. It is apparent from findings in Table 3 that NCDs and Assuming the same rates of absenteeism in this study hold their associated risk factors are affecting the working for the full year, we could expect there to be an average population in Namibia. This is consistent with findings rate of sick leaves taken by the study participants close to from studies in high-income countries [11,12]. This sur- 4 days per year. This calculation would put Namibian vey found a relatively high proportion of central obesity absenteeism rates in the formal sector somewhat below (18.7%) and elevated or high blood pressure (27.0%). In the estimates from the United States and Europe. Similar addition, diabetes had a significant impact on the rate of statistics for neighboring sub-Saharan African countries sick days although affecting a relatively low proportion were not readily accessible. Despite much progress toward of the population (1.7%). Moreover, anaemia which may

128 Guariguata et al. BMC Public Health 2012, 12:44 Page 10 of 12 http://www.biomedcentral.com/1471-2458/12/44

be caused by a large number of conditions, including impact of HIV/AIDS compared to diabetes may be a HIV, was also found to affect absenteeism. The mix of result of successful intervention programs already in NCD and communicable disease is in line with the epi- place among employers, including the exemplary high demiological transition that many low- and middle- ART coverage achieved in Namibia [30]. income countries face today [28]. The impact of conditions like hepatitis B and syphilis The most significant finding of this study is the asso- were negligible and did not contribute to increased rates ciation of diabetes with absenteeism. Random blood glu- of sick days in the short-term. Prevalence estimates for cose is not a rigorous test for conditions like impaired hepatitis B surface antigen were below those reported glucose tolerance or diabetes and is sensitive to whether for sub-Saharan Africa (> 8%) [3]. The same was true a person has eaten recently. However, the American for syphilis, which is well below average rates reported Diabetes Association has established that a random in general population surveys in Africa [3]. Given the blood glucose measurement above 11.1 mmol/L is very natural progression of these diseases in the absence of likely a sign of diabetes [22]. Therefore, although the treatment, it is possible that infection could lead to actual diagnosis of diabetes could not be made, the cur- long-term disability and contribute to future loss in rent glucose results suggest that diabetes has an impor- overall productivity for companies with a large propor- tant impact on absenteeism and employee health. tion of affected employees [24]. Almost half of those who were overweight or obese The fishing, services, and wholesale sectors had signifi- also had high blood pressure readings and elevated cantly higher rates of sick days when compared to the blood glucose. This finding is compatible with other stu- largest sector in this study, retail, even after controlling dies that have shown that the majority of people diag- for other factors. The prevalence of HIV and diabetes nosed with one non-communicable disease have co- were some of the highest in these sectors as well, which morbidity with another [28,29]. NCDs share many com- may indicate causal relationships. Conversely, the agri- mon risk factors and several of those are modifiable, cultural sector had significantly less absenteeism, per- especially through lifestyle changes. These shared factors haps because a number of farms in Namibia provide provide an opportunity for prevention strategies cen- health-related services on site to their employees [20]. tered on improving diet, physical activity, and smoking All of the above results are possibly subject to a bias reduction. Education strategies to reduce these risks toward workers who are fit enough to attend work on could be included in workplace wellness programs and the day of the screening. If this is the case, the effects could have an effect on both the health of employees reported here are an underestimate of the true impact and reducing absenteeism. of these and other health factors on sick days for work- Despite showing a less significant impact on absentee- ers in Namibia. ism in this study, HIV is still a significant determinant of increased sick days. In the current study it is not pos- Limitations sible to determine reasons for sick leave or to directly Because this is a cross-sectional survey, it is impossible relate absenteeism to HIV. It is possible that some to determine causation and only associations between employees missed work in order to access ART care in independent variables and the outcome (absenteeism) a location away from their workplace. Conversely, it is can be ascertained. In addition, the survey was meant as possible that a number of the HIV positive workers are an awareness and management information exercise for still in the early phase of infection and therefore are not companies and was not powered to detect particular yet on ART. These people could have more frequent associations. This increases the potential for type II sta- (short) episodes of sickness, as compared to HIV nega- tistical errors or the probability of finding a significant tive people or HIV positive people on ART. HIV posi- result when one does not actually exist. Because partici- tive workers could also have other HIV positive family pation was voluntary, there may be a selection bias members and take more frequent sick leave days to care which would make results not applicable to the general for them. population of employees. For the outcome variable Discovering the reasons behind diabetes- and HIV- (absenteeism), 890 people were missing data or refused related absenteeism may help the private sector provide to answer the question, which could bias the results as services that minimize costs lost to employee sickness. well. It could also be that HIV positive people preferen- For the fishing sector, prevalence of HIV appeared tially declined to participate in the HIV test. No particu- much higher than the average for the whole population lar adjustments were made for this missing data as this (14.3% compared to 9.0%, respectively). Where HIV/ is a cross-sectional survey and missing data were not AIDS is more prevalent in certain sectors, special atten- found to be systematically distributed throughout the tion should be given to providing adequate services such sample of industries. None of the variables had a greater as prevention education and ART. The lower relative than 5% proportion of missing data. While this survey

129 Guariguata et al. BMC Public Health 2012, 12:44 Page 11 of 12 http://www.biomedcentral.com/1471-2458/12/44

collected information on those who refused to learn interventions may be the most effective. If companies their results, it did not provide information on those were to purchase insurance for their employees, or pro- who refused to have the test done at all. vide wellness services, these should include both HIV- Finally, it should be stated that the primary outcome related prevention and treatment as well as information variable (sick days) was not verified against employer on the risks and management of NCDs. records and is subject to recall bias as with any self- reported information. This bias was minimized by using a Funding sources relatively short recall period of 90 days. Therefore, results The funding for the Bophelo! Project originated from cannot be extrapolated to longer time periods of observa- both the public sector (60% of all costs are funded tion. In addition, an important factor that could influence through international donors, more information at results, are the companies’ policies regarding sick leave. http://www.pharmaccess.org) and the domestic private These were not reviewed in the current study, but more sector (employers paid part of the costs at each screen- liberal leave policies might result in higher reported absen- ing site). The Bophelo! Project is a partnership between teeism. There is a risk of underreporting of absenteeism the PharmAccess Foundation, the Namibian Business due to cultural or workplace pressure which is not Coalition on AIDS, and the Namibian Institute of reflected in the findings. We did not review the workplace Pathology, supported by the Minsitry of Health and policies of each of the companies for this study, which Social Services. The project was initially created using would help to expand on this point, as it was beyond the funds from the Dutch Postcode Loterij, through the capacity of this analysis to do so. However, this would be Dutch Aidsfonds, Stop AIDS Now!, HIVOS and the an important aspect to consider in the future. Global Fund to fight AIDS, TB and Malaria. There is no information available on expected absentee rates for employees in the Namibian formal sector Disclosures which makes comparison of these results to a baseline The authors have no conflicts of interest to disclose. impossible. However, the findings of this study may help to prioritize areas of intervention for health of employ- Author details ees including risk factors related to NCDs and HIV. 1PharmAccess Foundation, Trinity Building C, Pietersbergweg 17, 1105, BM Amsterdam, Zuidoost, the Netherlands. 2PharmAccess Foundation, P.O. Box Conclusion 9895, Windhoek, Namibia. 3Boston University School of Public Health, 715 Albany Street, Talbot Building, Boston, MA 02118, USA. 4Amsterdam Institute While significant gains have been made in the provision for Global Health and Development (AIGHD), Academic Medical Center, of treatment for HIV/AIDS in Namibia, it is still a sig- University of Amsterdam, Amsterdam, the Netherlands. nificant factor contributing to absenteeism among work- Authors’ contributions ers. In addition, this study shows non-communicable LG - conducted the statistical analysis and wrote the manuscript. disease factors that are significantly contributing to IDB - contributed to the design and conduct of the study as well as increases in sick days, especially high blood glucose and providing information on the Namibian formal sector. RH - conducted the data collection, cleaning, and primary descriptive high blood pressure. Wellness and healthy living educa- analyses. DWM - designed the survey instrument and conducted the survey. tion programs have been shown to have significant FF - contributed to the analysis with methods on occupational health impacts on the incidence of chronic conditions in high- research and revised the manuscript. TFRDW - project lead and contributed to the writing, review, and income countries [25,31]. Similar approaches could be interpretation of findings of the analysis. All authors read and approved the developed and tested for the unique needs of this low- final manuscript. and middle-income country setting. Competing interests As other studies have shown that factors related to Leonor Guariguata is a consultant to the PharmAccess Foundation and a full NCDs have an impact on employee productivity and time employee of the International Diabetes Federation. absenteeism, companies should consider providing edu- Ingrid de Beer is a full time employee of the PharmAccess Foundation in Namibia. cation and basic screening services related to healthy liv- Rina Hough is a full time employee of the PharmAccess Foundation in ing and risk-reduction for NCDs. The impact of chronic Namibia. infection with other conditions such as hepatitis B and Els Bindels is a full time employee of the PharmAccess Foundation in Namibia. syphilis proved less pronounced in Namibia. Company- Delia Weimers-Maasdorp is a full time employee of the PharmAccess provided services to reduce absenteeism should focus on Foundation. the full spectrum of HIV prevention, treatment and Frank Feeley is professor of international health at the Boston University School of Public Health. care, as well as wellness education for risk factors Tobias Rinke de Wit is an employee of the PharmAccess Foundation in the related to NCDs including healthy eating and exercise. Netherlands. These recommended interventions should be implemen- Received: 31 October 2011 Accepted: 18 January 2012 ted with simultaneous cost-effectiveness analyses, Published: 18 January 2012 including the costing of absenteeism, to assess which

130 Guariguata et al. BMC Public Health 2012, 12:44 Page 12 of 12 http://www.biomedcentral.com/1471-2458/12/44

References 31. Namibian Ministry of Labour and Social Welfare: Namibia Labour Force 1. UNAIDS: Report on the Global HIV/AIDS Epidemic. Geneva 2008. Survey. Windhoek 2010. 2. The World Health Organization: Preventing Chronic Disease: A Vital Investment. Geneva 2005. Pre-publication history 3. The World Health Organization: Global Burden of Disease Report. Geneva The pre-publication history for this paper can be accessed here: 2004. http://www.biomedcentral.com/1471-2458/12/44/prepub 4. The International Diabetes Federation: The Diabetes Atlas. Brussels , 4 2008. 5. Negin J, Robinson H: Funding for HIV and non-communicable diseases doi:10.1186/1471-2458-12-44 Australia: Nossal Institute for Global Health at the University of Melbourne; Cite this article as: Guariguata et al.: Diabetes, HIV and other health 2010. determinants associated with absenteeism among formal sector 6. Narayan K, Zhang P, Kanaya A, et al: 30: The diabetes pandemic and workers in Namibia. BMC Public Health 2012 12:44. potential solutions. Disease Control Priorities in Developing Countries World Health Organization; 2006. 7. Zikusooka CM, Kyomuhang R, Orem JN, et al: Is health care financing in Uganda equitable? Afr Health Sci 2009, 9(Suppl 2):S52-S58. 8. Akazili J, Gyapong J, McIntyre D: Who pays for health care in Ghana? Int J Equity Health 2011, 10:26. 9. Huffman MD, Rao KD, Pichon-Riviere A, et al: A cross-sectional study of the microeconomic impact of cardiovascular disease hospitalization in four low- and middle-income countries. PLoS One 2011, 6:e20821. 10. de Beer I, Coutinho HM, Guariguata L, et al: Health care options for commercial farm workers in Namibia. Rural Remote Heal 2011, 11:1384. 11. Bertera RL: The effects of behavioral risks on absenteeism and healthcare costs in the workplace. J Occup Med 1991, 33:1119-1124. 12. Collins JJ, Baase CM, Sharda CE, et al: The assessment of chronic health conditions on work performance, absence, and total economic impact for employers. J Occup Environ Med 2005, 47:547-557. 13. Leynen F, Backer GD, Pelfrene E, et al: Increased absenteeism from work among aware and treated hypertensive and hypercholesterolaemic patients. Eur J Cardiovasc Prev Rehabil 2006, 13:261-267. 14. Jans MP, van den Heuvel SG, Hildebrandt VH, et al: Overweight and obesity as predictors of absenteeism in the working population of the Netherlands. J Occup Environ Med 2007, 49:975-980. 15. Tawfik L, Kinoti S: The impact of HIV/AIDS on health systems and the health workforce in sub-Saharan Africa. USAID, Bureau for Africa, Office of Sustainable Development; 2003. 16. Fox MP, Rosen S, MacLeod WB, et al: The impact of HIV/AIDS on labour productivity in Kenya. Trop Med Int Health 2004, 9:318-324. 17. Rosen S, Vincent JR, MacLeod W, et al: The cost of HIV/AIDS to businesses in southern Africa. AIDS 2004, 18:317-324. 18. de Beer I, Coutinho HM, van Wyk PJ, et al: Anonymous HIV workplace surveys as an advocacy tool for affordable private health insurance in Namibia. J Int AIDS Soc 2009, 2:7. 19. The World Bank: Namibia at-a-glance. 2008 [http://www.worldbank.org]. 20. In Poverty and Inequality In Namibia: an overview. Edited by: Schmidt M, Stiftung FE. Institute for Public Policy and Research; 2009:. 21. Sturman M: Multiple Approaches to Absenteeism Analysis Cornell University: Center for Advanced Human Resources; 1996. 22. American Diabetes Association: Clinical Practice Recommendations. Diabetes Care 2008, 31(Suppl 1):S1-108. 23. Feeley FG, Rosen S, Connelly P: The private sector and HIV/AIDS in Africa: Recent developments and implications for policy. In The changing HIV/ AIDS landscape: selected papers for the World Bank’s agenda in Africa. Edited by: Lule EDA. World Bank; 2009:. 24. Namibia. National Laws on labour, social security, and related human rights. International Labour Organization. [http://www.ilo.org/dyn/natlex/ docs/WEBTEXT/29328/64850/E92NAM01.htm]. 25. Mercer: Pan-European Employer Health Benefits Issues: 2010 Survey Report. London 2011. Submit your next manuscript to BioMed Central 26. United States Bureau of Labour Statistics: Program Perspectives: On Paid Sick and take full advantage of: Leave. Washington D.C 2010. 27. The World Health Organization: Namibia Country Cooperation Strategy 2010 - 2015. Geneva 2010. • Convenient online submission 28. Amuna P, Zotor FB: Epidemiological and nutrition transition in • Thorough peer review developing countries: impact on human health and development. Proc • No space constraints or color ﬁgure charges Nutr Soc 2008, 67:82-90. 29. Wolff J, Starfield B, Anderson G: Prevalence, expenditures, and • Immediate publication on acceptance complications of multiple chronic conditions in the elderly. Arch Intern • Inclusion in PubMed, CAS, Scopus and Google Scholar Med 2002, 162:2269-2276. • Research which is freely available for redistribution 30. World Health Organization: Towards universal access United Nations, World Health Organization; 2010. Submit your manuscript at www.biomedcentral.com/submit

131 132 12 DISCUSSION & RECOMMENDATIONS

The studies presented in this thesis largely focused on evaluating how the private sector can be engaged in public health challenges in Namibia. The findings indicate that there are multiple opportunities for doing so, such as understanding and subsequently stimulating demand from patients/clients (Chapters 2-5), establishment of innovative service provision for HIV and non-communicable diseases (Chapters 6-7), stimulating the market for prepaid services through risk equalisation (Chapters 7-8), performing surveys and advocating pertinent research results to policy makers within the public and private sector (Chapters 9-11).

The research questions were formulated around the strategies of PharmAccess to stimulate healthcare exchange by key players in the healthcare system: the patient, the provider, and the payer; always in the context of the pertinent healthcare policy. Whilst the patient, provider and payer directly participate in the healthcare exchange, the policy is the context within which this exchange takes place and the policy makers determine the ‘rules of the game’. The findings and experience in testing PharmAccess hypothesis shows that coordinated exchange between these players is a key to stimulate change.

A summary of the interventions including the direct and indirect effect of the intervention on each stakeholder group is presented in Table 1.

133 Evidence of HIV prevalence in workforce for targeted interven - tions e.g. sectoral strategies and workplace policies Approved use of non-invasive oral fluid testing for HIV preva - lence surveillance based on validation data Policy environment “Rules of engagement”” Increase uptake of insurance amongst employer Prevalence information to facilitate insurance risk and premium calculations Non-invasive screening HIV test becomes available Payer HIV prevalence screening services available for the first time in Namibia Evidence supported establishment of Bophelo! mobile screening services Validated tests for use in Validated HIV prevalence surveil - lance services Supply (provider) Improve knowledge of HIV prevalence to generate demand for health Insurance Identified demand for patient HIV Counsel - (HCT) ling and Testing on site at companies Validation study Validation conducted to acquire tool to stimulate demand for HIV testing Demand (patient/ client) Intervention Effect Intervention Effect Intervention directed at stakeholder/ Effect on stakeholder 2. Anonymous HIV workplace surveys Strategy 1: to stimulate demand by improving health awareness of patients/clients 1. of OraQuick and Validation OraSure oral fluid HIV tests Table 1. Table PharmAccess strategy

134 Evidence supported the first PPP agreement for mobile primary health - care Evidence of successful PPP paving way for Mister Sister PPP Anonymous and comparative data increased employers awareness to provide access to healthcare for employees Evidence motivating need for HIV prevention and treatment programs for tertiary students Mixed funding model of payers – private employers, donors and MoHSS Mixed funding model made comparative cost cheaper due to private sector contributions Underutilized campus health facilities / Potential for student health insurance PPP for wellness screening using mobile clinics established Resulted in the establish - ment of the Mister Sister mobile primary health care clinics (public private partnership PPP) Identified need to include HCT and treatment at campus health facilities Awareness raising of Awareness health conditions of individual patients stimulate demand for health services Demand for mobile PHC services identified with referral Established demand for healthcare service demand and willing - ness to pay of client Increase awareness of HIV Identified demand for HCT on site Effect Intervention Intervention Effect Intervention Effect - Strategy 2: to stimulate supply by developing innovative new public private partnership healthcare providers 5. Comparative cost of mobile vs fixed site screening 4. Healthcare options for commer cial farm workers 3. University students HIV prevalence surveys

135 Evidence for PPP and proof of principle mixed funding model of Mister Sister clinics raised of gap Awareness in insurance market for low income earners raised of risk Awareness of crowding out private sector First household prevalence data in the mapped by country, socio economic and geographic status. First HIV household incidence data Data for health insurers of HIV and NCD risk in house - holds in Windhoek Mixed funding model donor and – employer, MoHSS contributions Creation of low income health insurance and temporarily avoiding crowding out First PPP to provide mobile PHC to employer groups and vulnerable children Creation of low income health insurance and the risk equalization fund Data for providers on HIV prevalence & NCD risk in household –awareness raising for screening, prevention and early detection Improved health outcomes in children raised Awareness amongst employers of need for health insurance Individual awareness raised of NCD Effect Intervention Intervention Effect Intervention Effect 8. HIV in Windhoek Prevalence and incidence of Strategy 3: to innovate new payer mechanisms using temporary subsidization 7. Innovation in Namibia: Preserving Private Health Insurance Strategy 4: to collect and present evidence support policy making 6. Mobile PHC and Health outcomes of children

136 Motivate investment in health insurance and move away from AIDS exceptionalism Quantify loss of labour to motivate investment of employers into health insurance and access to healthcare Knowledge sharing for limitations to HIV household surveys HIV and NCD preva - lence data for health insurers of risk in formal sector Evidence to stimulate employer to become payer for health insurance None Motivate focus on NCD screening and early detection for prevention and treatment None None Stimulate demand for health insurance for formal sector Stimulate demand for health insurance uptake in formal sector None Intervention Effect Effect Intervention Intervention Effect 11. Absenteeism related to prevalence of HIV and NCD in workplaces 10. Prevalence of HIV & NCD workplaces 9. Cautious note on estimating HIV prevalence

137 The following pages discuss per strategy the most important conclusions and put those into perspective of the general literature.

Strategy 1: To stimulate demand by improving health awareness of the patient/client Various interventions were performed by PharmAccess that were geared towards increasing demand for healthcare in the Namibian context. One postulation was that performing surveys for pertinent health problems would have multiple stimulatory effects: making people more aware of their health status (and hidden health problems) and thus more willing to prepay for healthcare, making employers more aware of their healthcare responsibilities towards employees and at the same time get more information on the extent of various health problems (in particular HIV), thus creating data to help health policy makers to make better-informed decisions. One approach was to assess whether an understanding of the HIV prevalence, collected through bio-medical and behavioural surveys support organizations to enrol their internal stakeholders in pre-paid low income health insurance for HIV in Namibia?

We begin in Chapter 2 by presenting the validation of the diagnostic accuracy of the OraQuick® HIV-1/2 Rapid Antibody test and the Oral Fluid Vironostika® HIV Uniform II microELISA system with the Namibian National Algorithm for HIV Testing. Although oral fluid based rapid HIV tests were available on the international market at the time, these tests had not been validated nor approved for use in Namibia. A validation study was thus undertaken by PharmAccess in collaboration with the Namibia Institute of Pathology (NIP).

This study was the first formal field evaluation of the OraQuick® and OraSure® devices, in a resource limited setting in Southern Africa. The study found that the OraQuick® test is 100% accurate (100% sensitivity and specificity) and that the OraSure® has a high specificity (99.5%) and slightly lower sensitivity (97.1%) The high negative predictive value of both tests rendered them particularly suitable for HIV surveillance purposes. Based on this validation the Namibian MoHSS approved the OraQuick® and OraSure® devices for surveillance purposes in Namibia. HIV testing for individual diagnostic purposes remained based on rapid blood testing utilizing the National testing algorithm. This study paved the way for the implementation of non-invasive HIV screening in Namibia, both in workplace and household studies.

Following the validation of non-invasive oral fluid HIV testing devices in Namibia, prevalence surveillances were conducted by PharmAccess in 24 Namibian private companies, from 11 different industries, between November 2006 and December 2007 (Chapter 3). Participation rates amongst the 8,500 employees ranged from 61.3-97.3% with a mean participation rate of 78.6%. The prevalence rate of 15.0% revealed that the overall HIV prevalence rate in these companies was in line with the national prevalence estimates in Namibia at the time (UNAIDS/WHO, 2008). However, it was also found that HIV prevalence varied widely between industries, ranging from 4% in the information technology sector to 21% in the mining sector. These prevalence rates were similar to findings of high HIV prevalence rates in workforces in Southern Africa (Evian, et al., 2004) and especially in miners (Corbett, et al., 2004) in South Africa.

The studies not only provided the first prevalence data from a portion of the working population across various industry sectors, but also demonstrated how knowledge of the HIV prevalence amongst the workforce stimulated enrolment in private health insurance for HIV in Namibia. Following the HIV prevalence surveillance, 61% of the previously uninsured employees had been enrolled in health insurance, the majority of these insurances (78%) covering HIV only. This engagement of the private sector to purchase health insurance including HIV coverage for their employees was deemed important for several reasons, but most impor-

138 tantly to reduce the burden on the public sector (Feeley, et al., 2007) (Sekhri & Savedoff, 2005), secondly raising awareness of HIV to reduce traditionally high stigma (Holzemer WL, 2004) and thirdly to manage apparent higher concentrations of HIV infection amongst employed and mobile members of society (Mishra, et al., 2006) (Piot, et al., 2007).

This study was important in dispelling the myth that HIV was a general population and not a workforce problem as maintained by many organizations especially in the SME sector in Namibia ( (NABCOA, 2004) (Hohmann & Skolnic, 2004) and can be convinced to invest in the health of their employees, which was supported by later studies (Rosen, et al., 2007).

Thus, these surveys helped identifying formal sector worker populations at higher HIV risks which helps targeting pertinent HIV interventions, both with respect to prevention and to treatment. Sharing the surveyed HIV prevalence data with higher management of companies coincided with considerable increase in private health insurance uptake, suggesting that anonymous HIV workplace surveys can indeed serve as a tool to motivate companies to invest in health insurance for their workforce and thus alleviating burden on the public sector. Regular screening of working populations and the consolidation of data, as was highlighted in the study, not only provided information for employers to plan workplace health responses but also gave valuable evidence for policy makers to plan for interventions and health insurers to plan for health risks. Both large and small companies purchased insurance for their employees following the availability of the evidence of the anonymous HIV prevalence surveillance. The consolidated anonymous data by industry made it possible for sector responses to be planned, the data from these surveys fed into to motivation for the establishment and strengthening of the Walvis Bay Corridor Group (supporting the transport, fishing and manufacturing sector) and the establishment of the HIV desk at the Chamber of Mines. The evidence of the HIV prevalence rates in the working population was availed to and utilized by the actuaries of medical aid funds in Namibia to calculate the expected risk, and contributed to the development of the HIV benefit structure within the low-income health insurance products developed as part of the HIVREF (Chapter 8).

The evidence of the HIV prevalence rates was also utilized by the disease management companies of the medical aid funds to motivate the establishment of case management capacity, and related investments. Especially as more countries look towards inclusive forms of community and national health insurance, the demonstration and quantification of the benefit of access to prepaid healthcare can be a key motivator for support and enrolment in the systems.

Expanding the findings on formal sector employees, additional studies were performed amongst those who are likely to become member of such sector: tertiary students. The question was considered whether students of academic institutions in Namibia were a target market for low-income health insurance for HIV? To this end the HIV prevalence, knowledge of HIV and health care utilization trends of students at the two largest academic institutions in Namibia were determined as presented in Chapter 4.

The study found, although it may have been underestimated, that students had relatively low HIV prevalence rate (1.8% at UNAM and 2.8% at Polytechnic) compared to the estimated national adult average of 15.3% (UNAIDS/WHO, 2008) at the time in Namibia. Studies performed in Ethiopia (Mulu, et al., 2014), South Africa (Dell, 2013) and Mali (White, et al., 2009) revealed 1.2%-3.4% HIV prevalence amongst university students. In South Africa and Namibia the prevalence rate of students were found to be well below the national adult average, while they were similar to the lower national adult average in Mali and Ethiopia. In all

139 cases, the need for intensified HIV prevention and treatment amongst University students is highly recommended to policy-makers.

Namibian campus health facilities were found to be underused, mostly due to privacy reasons. The findings motivated an intensification of prevention and education initiatives through the institutional HIV awareness programs and better marketing the use of the existing campus health facilities. Given the relatively low HIV prevalence found amongst students, the underutilization of campus health facilities and the relatively low reported willingness to pay towards an HIV health insurance, it was concluded that this group would not be the most viable target group for insurances covering HIV without significant and long-term subsidy. This study indeed could not motivate the policy makers at the tertiary institutions to find a mechanism to provide HIV insurance to students at the time. The low HIV prevalence amongst tertiary students brings hope and makes them an attractive group to include in any (future) general health insurance in the country, thus contributing to larger risk pools for sustainability and cross subsidization purposes.

Since agriculture is one of the largest industries in Namibia, and despite their dispersed geographic location, commercial farmers, their employees and dependents were considered by PharmAccess as an interesting target group for health insurance. A study was conducted to evaluate what are the healthcare challenges and the related needs for health services amongst rural and remote employers in the commercial farming sector in Namibia? It was assumed that commercial farm workers could be a good target group for private health insurance, which includes benefits for the treatment of HIV Chapter( 5).

It was found that this group indeed did not have adequate access to formal healthcare and particularly to HIV related services, similar to the experience in other African countries, where distance from healthcare facilities is a significant barrier to access (Kenny, et al., 2015) (Rees, et al., 2016). The willingness of farmers/ employers to contribute to the cost of health insurance in Namibia was high, however access to health services due to large distances posed a significant obstacle. Thus it was found that health insurance alone would not be a solution for rural and remote employers. Geographic distance barriers could only be overcome by bringing healthcare services closer to these work places. This study motivated the piloting and implementation of the Mister Sister mobile primary health care services public private partnership.

All in all it can be concluded that private interventions to create demand in the context of a public healthcare system that offers services for free or at marginal costs remains a serious challenge. All target populations addressed in the above studies could potentially pre-pay for healthcare. However, they also have access to free services. Apparently such services, even when overburdened, with long waiting times, medicine shortage and too short consultation times, are considered ‘good enough’ by most formal sector workers to not switch towards health insurance. The opportunity for national health insurance in Namibia would require the reformulation of the policy of free public healthcare for all, those who can pay should pay. Employer groups should be mandated to provide health insurance for their employees, this should not be a voluntary benefit. The Namibian Social Security Act (34 of 1994) makes provision for a National Medical Benefit Fund (NMBF), mandated to the Social Security Commission (SSC), which to date (end 2016) has not been established. In 2012, the SSC commissioned an actuarial study to recommend options for the NMBF. This study, which PharmAccess Foundation contributed to in terms of data provision and technical support, proposed a social health insurance model for Namibia funded through tax resources – either increased taxation or redistribution of existing taxation (Deloitte, 2012). In 2013 a ‘Universal Health Coverage Advisory Committee (UHCAN)’ was established headed by the MoHSS, the secretariat supported by the SSC, to develop a road map towards universal health coverage in

140 Namibia supported by a model of health insurance for all. PharmAccess was a member of this advisory committee. The fragmentation of the process of the NMBF development and the mandate for universal health coverage being seated with the MoHSS, has been a key determinant in the lack of political will and progress. Thus despite financial and technical support from various international and national partners, little progress has been made towards social or national health insurance in Namibia, unlike other African countries such as Ghana, Tanzania and South Africa.

Strategy 2: To stimulate supply by developing innovative new healthcare service provision It was evaluated whether mobile HIV testing in Namibia is an affordable alternative for hard-to-reach populations? The public-private partnership (PPP) model ‘Bophelo!’ providing mobile wellness screening at workplaces in Namibia, presented in Chapter 6, demonstrated the potential to mobilize private sector funding and be a cost-effective manner to reach individuals in workplaces and other hard-to-reach areas.

To reduce the stigma associated with HIV testing, the screening package included blood pressure and body mass index (BMI) readings; and blood screening for HIV, blood glucose, cholesterol, haemoglobin and Hepa- titis B. The inclusion of wellness screening in addition to stand-alone HIV testing illustrated a number of benefits at only a marginal additional cost and could thus assist to help address the growing issue of non-communicable diseases in Namibian workplaces. The Bophelo! PPP also provided evidence in the provision of mobile services to workplaces in Namibia to create the foundation for expansion to a public private partnership for populations that are hard-to-reach by conventional public or private services. Wellness screening, using the Bophelo! mobile clinics not only provided a better understanding of the costs associated with such an initiative, it also demonstrated that mobile services could be effectively run in Namibia across large distances. This motivated the Ministry of Health and Social Services to acquire clinics to provide outreach services for special programs in Namibia. The Bophelo! program demonstrated the effectiveness, and cost efficiency, of mobile testing with lower public cost due to significant contributions by the private sector. This program however highlighted a gap in access to healthcare facilities for referral and treatment after screening.

The Bophelo! program demonstrated, what was later to be confirmed in other literature (Bemelmans, et al., 2016), how the use of lay counsellors and task shifting could provide quality healthcare services at a lower cost. The findings of accessibility of testing services also confirm the findings of a later study conducted on mobile testing services in South Africa (Meehan, et al., 2015). This method of providing screening services has proven to be sustainable and continues in 2016, with both the Healthworks Business Coalition (previously NABCOA) offering mobile wellness screening partially funded by GFATM, and Mister Sister providing the services to companies contracted at a full cost recovery fee.

Upon the successful implementation of mobile primary healthcare services it was further aimed to assess whether and how leveraging private sector funded mobile clinics can benefit vulnerable groups? InChapter 7, the Mister Sister Mobile Clinics providing primary health care services in a PPP to rural and remote employers are presented. This PPP was formally established between PharmAccess and the MoHSS in 2011, after a successful pilot project amongst a small group of commercial farmers in late 2010 (PharmAc- cess, 2010). The services were implemented to reduce the geographic barriers to preventative and curative services as were identified in the earlier PharmAccess study presented inChapter 5.

141 The target population for the clinics were predominantly employees and dependents at rural and remote workplaces. Vulnerable groups such as women, children and the elderly residing at significant distance from healthcare facilities which the mobile clinic encountered en-route were also provided services (MSMHS, 2013). These services were funded through employer contributions for their employees and national and international donor contributions for vulnerable groups. The MoHSS provides all medication, vaccines and consumables for the primary health care services at no cost and provides services to patients referred from the mobile facilities. The mobile clinics provide primary health care according to the national treatment guidelines of the MoHSS and form an innovative private extension to public services. The patients served by the mobile clinics would have been patients needing to travel long distances to visit the public health facilities of the MoHSS. With the MoHSS providing the medication/ consumables at no cost, the employer contributing to the operating and transport costs for their employees and dependents and international/national donors contributing to the operating costs of service provision for vulnerable groups all parties benefit. It was concluded that the provision of regular preventive and therapeutic mobile clinic services improved health outcomes and reduced morbidity in vulnerable children who had access to these services. The greatest improvement in health indices was amongst Orphans and Vulnerable Children (OVC) whose disease burden was significantly reduced, indicating that mobile clinics may be an effective intervention in hard-to-reach and resource-limited settings. These findings were similar to The engagement of the private sector to contract for healthcare services for their employees was an important component to motivate this public private partnership, upon which subsidized care could subsequently be expanded to other vulnerable groups and poor communities, utilizing both domestic and foreign donor funding. Similar to the experience in Malawi (Geof- froy, et al., 2014) mobile clinics can improve access to healthcare and health outcomes for rural, remote and poor patients to compliment fixed site facilities, although the challenge remains how to move towards an integrated role for mobile clinics in the government healthcare system.

Using mechanisms such as this PPP could leverage private sector resources to not only improve the health of employees and dependents directly associated with the workplace, but vulnerable groups within the geographic area of service provision especially in hard to reach and underserved geographic areas in Africa. The expansion of the PPP has been possible, as a result, of increasingly attracting private sector funding, declining unit costs as a result of higher patient numbers, and increasing operational experience improving efficiencies. The provision of these mobile services are sustainable for service provision to employer groups for their employees and dependents, however continuing to serve vulnerable and poor populations will require partial subsidization from foreign or local donor funds unless the government outsources community mobile health services as part of an expanded PPP.

In late 2013 the MoHSS opened a fixed site clinic in the study location presented inChapter 7, which would provide daily services to the community and employer groups. This facilitated that the Mister Sister mobile clinics could move to new under-served areas. By 2014 the demand and willingness to pay for mobile primary care services amongst semi-urban and urban employers far outweighed the rural and remote employers. The lower cost of service provision in the semi-urban and rural areas due to lower travel and overnight costs for staff, in the absence of significant donor or government subsidy, led to a shift of provision of services to these areas. The Mister Sister mobile clinics in mid-2013 started providing service to both employer groups and communities in informal settlements in and around Windhoek. These services were funded predominantly by private sector contracting for the prepaid healthcare services to their employees and corporate social responsibility funding from a private medical aid scheme to poor communities living in informal settlements around the capital city. An evaluation of the mobile clinic program providing services

142 into informal settlements around Windhoek, conducted in 2013 (Sabin, et al., 2014) highlighted improvements in access to health services and reduced costs associated with travel and time, due to the availability of mobile clinics, with high levels of patient satisfaction with the quality of services provided.

Given the short period of time between the implementation of the services in the areas in Windhoek and the data collection for the evaluation, there was inconclusive evidence of the changes in the health status of patients. It could also not be concluded whether the availability of mobile clinics resulted in changes in health seeking behaviour of patients. Further research would have to be conducted into these topics to inform the impact of the mobile clinics. The various challenges identified during the evaluation included language barriers, high volumes of patient visits per day, which are very similar to the challenges faced by fixed health care sites in the City of Windhoek. The utilization of mobile clinics in informal settlements in and around the urban areas of Namibia however, represent an innovative manner to temporarily service communities who do not have access to such services until such a time as semi-mobile or fixed-site clinics can be established and permanently manned and sustainably financed. The Mister Sister clinics are the only primary health care clinics in Namibia to be SafeCare (www.safecare.org) certified. SafeCare is an innovative method of certification of healthcare facilities according to five SafeCare levels, determined by an assessment of the facility’s grading according to SafeCare standards and a quality improvement plan to facilitate measurable improvements between assessments (Johnson M. S., 2016). Mister Sister mobile clinics illustrate that good quality primary healthcare services can be provided in a public private partnership and that medicines provided free of charge by the MoHSS can be transparently accounted for and delivered to patients in under-served areas. The example of PPP contributes towards supporting the call for more PPP’s in Namibia and the SADC region as described in literature (Novignon & Olakojo, 2012). The expansion of the PPP to fixed and semi-mobile sites, contracting with private providers and encouraging more nurse led clinics, could contribute significantly towards mitigating the service-delivery gap currently experienced in Namibia.

The provision of primary healthcare services using mobile clinics could be expanded to include HIV treatment services as was successfully done in South Africa (Igumbor, et al., 2014) and other countries in sub-Sa- haran Africa (Roa, et al., 2011). At the end of 2016, the Mister Sister mobile primary healthcare programme (described in Chapter 7) had been identified by the government to become a Nurse Initiated Management of ART (NIMART) site. The staff of the mobile clinics underwent training on ART initiation and management in late 2016 and plans to commence roll out of NIMART services were in place for implementation in early 2017.

Strategy 3: To innovate new payer mechanisms using temporary subsidization The question was addressed whether low income health insurance/medical aid funding can be developed sustainably in Namibia with short-term donor subsidy to avoid the crowding out of the private sector and leverage private sector resources for HIV care and treatment. Chapter 8 demonstrates that through the innovative Health is Vital Risk Equalization Fund (HIVREF) payment mechanism additional prepaid resources for health could be raised that were several times higher than the short-term PharmAccess donor subsidy. This additional revenue was financed by higher income populations and demonstrated how cost sharing could be achieved while simultaneously the quality of care for HIV could be improved. The intervention confirmed that private voluntary schemes work when the benefits exceed the cost of membership. Moreover, the willingness to prepay appeared to depend to a great extent on the quality of the care received. The critical success factors of this intervention to leverage private sector resources for health included the government policy that supports the role of the private sector in healthcare; a willingness to accept subsidies for those other than the

143 poorest of the poor; and an adequate quality-controlled private sector healthcare service delivery infrastructure. This was accomplished despite an environment at the time where there was relatively little dialogue between the public and private healthcare sectors.

After the first 20 months (2006/2007), the HIVREF had enrolled over 34,000 people, which was equivalent to approximately 2% of the Namibian population. During the same period the Namibian public treatment program to fight HIV, TB and malaria, highly subsidized by PEPFAR and GFATM, was implemented and scaled-up nationwide. The successful implementation of the free access to HIV treatment in public health facilities had a direct effect on the new enrolment numbers of the HIVREF. The willingness of the target groups to enrol declined, enrolled groups resigned and by early 2009 enrolment rates on the HIVREF reduced to under 18,000, providing an example of crowding out of the private sector (Schellekens, et al., 2007). The HIVREF turned dormant in 2009, at which time the remaining members were transferred to a private insurance product of a participating health insurer. Despite the HIVREF’s dormancy low income health insurance products developed as part of the HIVREF in 2006-2009 were still in existence (at the time of writing this thesis at the end of 2016) being marketed and managed as part of smaller medical aid fund risk pools.

The evidence and experience of the HIVREF contributed towards a number of other international studies and publications (Preker, et al., 2010) (Preker, et al., 2013) (Schellekens, et al., 2007) (Gustafsson-Wright, et al., 2009). The experience of leveraging private sector resources through the HIVREF is supported by similar later findings in South Africa, demonstrating that private sector can play an important role in expanding health care coverage by mobilizing revenue to support the public healthcare system (van den Heever, 2012). Although South Africa is ahead of Namibia in the development of pre-payment for health care and health insurance (Erasmus, et al., 2016) (van den Heever A. M., 2016), the HIVREF mechanism remains a first in the region.

The gains made by the HIVREF demonstrated that both the financial and clinical management of HIV was effectively possible in the private sector. However, the availability of free public health services for HIV, predominantly financed by donors (MoHSS, 2015); (Abt Associates, 2008) resulted in crowding out of the results achieved. With free ART services available and a lack of a mandatory requirement for employed individuals to belong to a health insurance, there was less incentive for companies to insure their employees. Although this may have been a suitable response to the HIV epidemic during a period of ample donor funding, yet the consequences thereof are felt over a decade later when the donor funding to treat HIV is declining (Cairney & Kapilashrami, 2014) (Schiffman, 2008) and structures for health financing after donor exit have not been adequately put in place. Donors should take these points to heart realizing that large donor funding for health, and specifically for vertical programs, can have detrimental effects in crowding out the private sector. In countries where a means-tested mandatory contribution for health care cannot be enforced, a strategy should be applied that avoids crowding out and for example includes voluntary private insurance and risk equalization. The Namibian experience shows that this is, at least temporarily, possible.

Seven years after the 2009 reintegration of the HIVREF into the medical schemes, the UHCAN (SSC, 2016) is looking at mechanisms to pave the way for a national or social health insurance in Namibia. The principles of risk pooling of expensive diseases, the utilization of private sector administrators and disease management mechanisms could be re-established and form the basis of such a pre-paid medical benefit scheme. From the experience of the HIVREF and the related low-income health insurance products, the large-scale enrolment of those in formal employment on health insurance would have to be made mandatory or be incentiv-

144 ized. To undo the crowding out effect, the legal framework in Namibia would have to be revised, to restrict free public health services only for the poor, requiring all those who can afford to pay for health services to pay.

The findings of a 3-year panel survey conducted in the City of Windhoek, (Gustafsson-Wright, et al., 2010) estimated that for the middle three income quintiles surveyed, health insurance premiums could be set at between 4-6% of per capita income levels. This study also found that the potential for a health insurance market in Namibia was very large with a relatively high willingness to pay, even amongst the lowest income quintiles. This would reduce the need for tax-based public health funding and encourage growth in private sector health insurance generating volumes, which support cost reductions through economies of scale. The findings are valuable information for Namibia upon which to base information on national health insurance and universal health care coverage also using mixed models of financing (Uzochukwu, et al., 2015).

The role of the private sector, especially in the achievement of universal health coverage remains a subject of debate, although most broadly suggested is that public sector, private sector and funding agencies work together to mobilize domestic resources to both fund and deliver health services sustainably and especially fill gaps left by the public sector and international funders (Stallworthy, et al., 2014). Challenges such as the poor implementation of public private partnership agreements in the provision of universal health coverage services have hampered progress in countries such as Malawi (Abiiro, et al., 2014) thus suggesting that well-functioning public private partnerships that support access to community health insurance and health services should fully integrate both public and private sector.

Strategy 4: To collect and present evidence to support policy-making We presented in Chapter 9a the data providing the first ever HIV prevalence and incidence data from a 3-year panel household survey conducted in the City of Windhoek, to provide information to insurers and policy makers to plan the HIV response. The findings of HIV infection rates of 2-3 per 100PY provided an evidence base to advocate against complacency in the HIV response. This study underscored the importance of using incidence (new infection) to understand the HIV epidemic in addition to the more commonly used HIV prevalence rates, which represent both new and existing infections. This was especially important when looking at the findings of HIV amongst adolescents (12-24 years) in the city. HIV prevalence rates for male and female adolescents were similar in 2009, yet HIV incidence in adolescent females was 1-3 per 100PY and 0.2-1 per 100PY in male adolescents. This study found what has now become an accepted fact that young females are at heightened risk of HIV infection and specific strategies are needed to prevent infection in adolescent girls (Mavedzenge, et al., 2016). The methodology of the household surveys conducted in Namibia was utilised in other sub-Saharan African countries where PharmAccess had interventions and the lessons learnt could be shared. One of the key lessons resulting from the implementation of the study was the risk of surveyor bias as described in Chapter 9b, which highlights potential limitations of household surveys.

As a result of the data generated from these studies UNAIDS commissioned a Know Your Epidemic / Know Your Response (KYE/KYR) study (van Renterghem, et al., 2012); (Van Renterghem, et al., 2012) to encourage the CoW management to adopt an HIV strategic plan using the evidence. This strategic plan has been in development since 2012, yet has not been implemented, largely due to human and financial resource constraints. The findings of the HIV prevalence and incidence survey, especially the geographic representation of the burden of disease in the city, supported the negotiations of PharmAccess with the MoHSS, resulting in the provision of mobile primary health care and HIV screening services to the informal settlements in and

145 around Windhoek, by the Mister Sister mobile primary health care clinics since 2012 and the recent (2016) inclusion of the Mister Sister clinics in the MoHSS NIMART programming.

A further research question to improve health awareness amongst decision/policy makers in the private sector was to establish what the prevalence of non-communicable diseases, and the related absenteeism, is at workplaces in Namibia? Chapter 10 presents the findings of a larger Bophelo! wellness screening dataset on the prevalence, knowledge and self-perceived risk of employees of the three most significant health threats in Namibia, namely HIV, hypertension and diabetes. It was found that the prevalence of elevated blood pressure, elevated blood glucose, and HIV among employees of the Namibian formal sector is high, while risk awareness is low. This data suggested that attention must be paid to improving the knowledge of risk factors related to these conditions as well as providing care to those with chronic conditions in the formal sector through programs such as workplace wellness, motivating even greater investment by employers into the health and well-being of their employees.

Despite the high prevalence rates, risk awareness and knowledge of health related risk factors was low. These findings, together with the findings of the resultant absenteeism due to these conditions’ effect on productivity (Chapter 11) when presented to companies could be used as an advocacy tool. Such advocacy would encourage employers to expand their HIV workplace programs to wellness programs which include information, education and communication on NCD, as well as encourage access to health services for these chronic conditions. Evidence that both non-communicable disease risk factors and infectious diseases are associated with increased rates of short-term absenteeism of formal sector employees in Namibia is presented. This evidence supports the suggestions that programs to manage these conditions could help employers avoid costs associated with absenteeism. These workplace wellness programs were encouraged to include basic health care insurance, to provide effective access to care and include regular wellness screenings.

The results of this survey motivated the inclusion of treatment for NCD in the primary healthcare services provided by the Mister Sister mobile clinics (Chapter 7). The Bophelo! workplace (Chapter 10) and Windhoek household survey (Chapter 9a) data on NCD was included in larger international studies (Hendricks, et al., 2012) (Collaboration NCD Risk Factor, 2016). The strategy employed in Bophelo! to include screening for other conditions into a minimum package to reduce HIV testing stigma could be employed effectively in other developing countries to incorporate screening for NCD in HIV surveys.

Reflection and general recommendations

From the various PharmAccess studies it is evident that interventions aimed at specific stakeholders have effects on other stakeholders, illustrating the intricateness and daunting complexity of health systems in general. The various interventions have taken place over a more than a decade in different locations with many different stakeholders involved. The funding for the different components of the PharmAccess programs in Namibia came from various different sources at different moments with different priorities and reporting requirements. This reality contributed to a relatively fragmented approach to improving components of the healthcare system. To break out of the vicious circle of poor healthcare as per the PharmAccess theory of change, a more comprehensive and less fragmented approach may have had the potential for more intensive results. In other words: an important lesson learnt from the PharmAccess interventions in Namibia is that a more coordinated approach addressing all stakeholders in the healthcare exchange (patients, providers,

146 payers) simultaneously would probably have more impact. Over the years, the PharmAccess Namibia interventions were relatively dispersed and not always feeding into each other: the Windhoek household survey did not benefit the farmers or the formal sector employees; the HIVREF did not involve the mobile clinics; the HIV surveys amongst formal sector employees, students and farmers did not directly feed into health insurance. With more substantial and timely funding in place, an intervention could have been designed that involves all stakeholders of the healthcare exchange simultaneously. This lesson was learnt and put into practice with the various PharmAccess approaches in Nigeria, Ghana, Tanzania and Kenya. Whether this conclusion is indeed correct remains to be seen from the developments of PharmAccess interventions in these other countries.

Innovative financing models to assist governments to cover the costs of ARV treatment in the wake of declining donor funding can only be encouraged and expanded as more local funding models are required (Atun, et al., 2016). The responsibility for funding HIV/AIDS should be shared between public and private sector, especially in the light of declining donor resources in countries whose income is growing. The question is often what is the capacity of a country to finance their own HIV response in an environment of declining donor funding? As shown in some of the existing research, upper middle income countries such as Namibia could become financially self-reliant (Resch, Ryckman, & Hecht, 2015) to domestically finance healthcare and specifically if more cost effective provider services such as those of Bophelo! Chapter( 6) and Mister Sister mobile clinics (Chapter 7) can be expanded.

Further leveraging of domestic resources to fund the HIV epidemic could require the mobilization of private sector resources such as those of the HIVREF (Chapter 8) and similar models. In the absence of a social or national health insurance scheme, the HIVREF in fact represented a unique model, that could effectively combat the often-observed disadvantages of ‘HIV exceptionalism’ (Lima, et al., 2014) (Dionne, 2015) (Cailhol, et al., 2013) or ‘vertical funding’ for HIV (Mussa, et al., 2013) (Tan, et al. 2015) (Nattrass, et al., 2016). With stagnating PEPFAR and declining GFATM funds for antiretroviral treatment and in light of the current uncertainties with respect to US policy towards combating HIV/AIDS, the HIVREF could become attractive again. Particularly in Africa countries that are moving towards a national health insurance system (e.g. Ghana, Rwanda, South Africa, Kenya, Tanzania), PEPFAR funds could complement their general risk pools by covering the diagnosis and treatment of HIV/AIDS.

The development of the PharmAccess Namibia programs, as mentioned previously, was at times organic and opportunistic, based on the identification of gaps in one program, trying to find an innovative solution for such gap and then trying to secure funding to intervene. The need to align interventions to donor/funder priorities at times led to fragmentation of the interventions and the inability to break through the vicious circle. In hindsight, Namibia could be considered as a ‘phase zero’ country for PharmAccess for ideas and innovations that would later be further refined and appropriated for use in other countries. The PharmAccess experiences in Namibia provided: - the first healthcare crowding out experiences of HIVREF to contribute to PharmAccess Theory of Change - the first lessons in healthcare insurance, risk sharing and prepayment which formed the basis for the Health Insurance Fund (HIF) for Nigeria, Tanzania, Kenya - the first format for impact evaluation through household surveys as later used in Nigeria, Tanzania, Kenya - the first challenge to build financial models for diagnosis and treatment of a particular diseases, such as the HIV actuarial models later used elsewhere in Africa

147 - the first challenge to implement and improve quality of the management of the treatment of HIV (My Health Namibia) including the development of standard treatment protocols, setting up disease and quality management systems, highlighting the need for improvement of private healthcare facilities and patient safety - the first model of PPP to provide wellness screening services to specific target groups in a mixed-funding model - the first model for PPP mobile healthcare provision through ambulant clinics providing mixed funding model - the first opportunity to develop SafeCare standards for mobile health facilities

Besides the learning for the PharmAccess Group the programs and interventions have brought significant changes in the healthcare system of Namibia and especially broke through the ice of building trust between the public and private sector to jointly provide health services using a mixed funding model.

Managing the public-private mix will be one of the biggest challenges of the Namibian government going forward towards the achievement of the universal health coverage goals. Similar to the models and related challenges described by McPake and Hanson (McPake B, 2016) the Namibian government will have to ensure public-private partnerships that facilitate access to good quality and affordable healthcare. In the process of managing such public-private mix, the Namibian MoHSS may benefit from a more regulatory role, outsourcing service provision and health service management to both the for-profit and not-for-profit private sector.

The recent decline in donor funding for health in Namibia, as well as the mandate of the UHCAN warrants the consideration of the re-establishment of the HIVREF. The structure and mechanism of the HIVREF could be re-established as a transitional funding model during the PEPFAR phase out, and other donor transitions. This could be accomplished through the inclusion of contracted private healthcare providers, the expansion of existing public-private partnerships for service delivery (such as the Mister Sister clinics), and the utilization of private third-party administrators and disease/case managers monitored to a stringent quality monitoring and management system.

In hindsight, a more substantial investment of vertical donor funding in the HIVREF in 2008 might have leveraged the same proportionate amount of private sector funding and with the inclusion of other disease benefits could have created a platform for a national health insurance scheme in Namibia.

Overall, concluding from the results of studies described in this thesis, the following recommendations, are made which build on the recommendations made by Gustafsson-Wright at al. (Gustafsson-Wright, et al., 2010):

1 Continue stimulating and understanding the demand of the patient/ client through increased efforts to provide preventative health awareness both in formal sector and in communities: a Mandate or incentivize the provision of workplace programs, which provide linkages to care, for health in the formal sector; b Seek PPP’s to provide workplace type programs in the informal sector, linked to community health programs;

148 c Offer free or inexpensive testing for major health expenditure risks to both workplaces and communities; d Expand the existing wellness testing services offered in PPP through mobile facilities to remote and underserved areas; e Use the mobile facilities available to seek out key populations for testing and referral for HIV and other chronic diseases; f Establish a bi-directional referral system between the health care facility and the workplace/community and/or case manager to ensure that patients with chronic conditions remain within the continuum of care.

2 Stimulate supply by expanding innovative new healthcare service provision through the PPP on service delivery mechanism a Expand the PPP (example of Mister Sister) to existing fixed and semi-mobile sites and private health care providers; b Seek public private partnerships to services the minimum health insurance package and risk equalization mechanism supported products; c Establish PPP desk within MoHSS or inter-ministerial to seek out innovative and relevant PPP’s for health and contract for related services.

3 Innovate payer structures to channel health risks into risk equalization mechanisms a Restore the HIVREF for HIV and channel private premium contributions and donor transitional funding into this model; b Offload the public sector by outsourcing services to private providers in a public private partnership, similar to that of Mister Sister; c Expand the HIVREF, to include other expensive chronic conditions that are of importance to Namibia, such as cardio-vascular disease, TB, diabetes and possibly cancers. Donor contributions could be channeled into the fund to allow for price reductions in the private sector and to raise private sector funding from the higher income groups within the risk pool; d Vertical donor funding during donor exit/transition phases could be channeled through these funds and used to leverage private sector and other resources; e Investigate options to use the risk equalization fund mechanism as a foundation to establish the NMBF package for all employed but uninsured as contained in the Social Security Act 34 of 1994 (Republic of Namibia, 1994), engaging the private sector to avoid crowding out; f Explore the possibility of expanding the NMBF to a national health insurance scheme for Namibia, to include the poor.

4 Develop specific private health insurance for employed, including low income, groups a Provide a minimal insurance package supported by public, private and tax revenue through the REF; b Make health insurance mandatory for all formally employed; c Accelerate the NMBF for all uninsured employed to be enrolled in the minimum insurance package; d Utilize the combination of risk equalization mechanisms, the minimal insurance package and PPP structures for service delivery to establish this basis for a national health insurance; e Consider using the infrastructure of an existing medical scheme, such as the Public Service Employees Medical Aid Scheme (PSEMAS) with risk equalization mechanism(s) for expensive and chronic diseases to support such a minimum package.

149 5 Expand the role of MoHSS to facilitate policy making and regulation for public private partnerships a Include a focus on the contracting of the private sector to provide public health services within the MoHSS mandate; b Expand the focus and resource allocation of the MoHSS to include public private partnerships and outsourcing of services to both for-profit and not-for profit private sector; c Establish a PPP desk within the MoHSS to facilitate access to and from the private sector for the presentation of proposals or a coordinated call for proposals for the private sector to assist in public health challenges.

6 Establish an independent national health information hub to support evidence for policy making a Establish a national health sector management platform consisting of representatives of all stakeholders to develop a health sector development strategy that makes optimal use of both public and private resources including foreign donor funding and investments, while strengthening complemen- tarities and avoiding crowding out effects; b Link this national health information hub to the inter-ministerial PPP desk; c Establish a mandatory, possibly incentivized reporting requirement for private sector to overcome current reporting barriers; d Collect all pertinent information from both public and private sector and make this information available to all stakeholders on a regular basis; e Develop both the public and private sector capacity to on an ongoing basis collect, store, analyze, publish and disseminate pertinent health information for policy and planning purposes.

7 Establish national standards for quality of healthcare services and health information a Achieve international recognition of national standards through established certification e.g. SafeCare (www.safecare.org) and accreditation (Whittaker, 2011) e.g. Council for Health Service Accreditation of Southern Africa (www.cohsasa.co.za ) bodies; b Explore possibilities of establishing an independent national quality board to represent those standards and independently offer quality assessments of public, private and public/private providers; c Consider the temporary or partial subsidization of the certification and accreditation of the public and private health care facilities participating in the minimum package provision, to avoid cost barriers; d Develop pay for performance mechanisms for healthcare providers linked to the metrics of providers’ quality performance; e Assist existing HIV managed care programs to expand to other important diseases and ensure adher- ence to quality standards and national treatment protocols for these high risk diseases; f Develop a national curriculum for continuous medical education of healthcare staff that is linked to the country’s priorities in terms of high expense and high risk diseases; g Develop incentives for healthcare staff to increase the quality of output in both public and private sectors.

150 Conclusion

This thesis presents studies demonstrating how the private sector could be engaged successfully towards addressing public health challenges in Namibia. This engagement could be achieved not only in the response to HIV, but for other chronic and non-communicable diseases by stimulating demand through the patient/ client; directing demand through the establishment of innovative public-private partnership providers; innovating new payer mechanisms and providing evidence for policy making. The response should include leveraging private financing of healthcare, both through formal health insurance and pre-paid subscription for primary care services.

The re-establishment of the HIVREF and its related infrastructure, as well as the scaling up of the public private partnership’s Bophelo! and Mister Sister, could contribute significantly towards a national health insurance approach, addressing access-, financing- and quality- gaps in the Namibian healthcare system, especially in light of the declining donor resources for health. In a country such as Namibia, with a strong, well-established private sector, which has demonstrated a willingness to engage and support public health challenges, and a government which has demonstrated a willingness to engage with the private sector, the achievement of the goal of universal health coverage for a population of under 2.5 million people should be achievable.

151 ••• References

Abiiro, G., Mbera, G., & De Allegri, M. (2014, May). Gaps in universal health coverage in Malawi: a qualitative study in rural communities. BMC Health Servcies Research. Abt Associates. (2008). Namibia Health Accounts 2001/2-2006/7. Namibia: Republic of Namibia and the Health Systems 20/20 project. Atun, R., Silva, S., Ncube, M., & Vassall, A. (2016, June). Innovative financing for HIV response in sub-Saharan Africa. Journal of Global Health, 6(1). Bemelmans, M., Baert, S., Negussie, E., Bygrave, H., Biot, M., Jamet, C., . . . Ford, N. (2016, May). Sustaining the future of HIV counselling to reach 90-90-90: a regional country analysis. Journal of the International AIDS Society, 19(1). Cailhol, J., Craveiro, I., Madede, T., Makoa, E., Mathole, T., Parsons, A., . . . Lehmann, U. (2013). Analysis of human resources for health strategies and policies in 5 countries in Sub-Saharan Africa, in response to GFATM and PEPFAR-funded HIV-activities. Globalization and health, 9(1), p.1. Cairney, L., & Kapilashrami, A. (2014). Confronting ‘scale-down’: Assessing Namibia’s human resource strategies in the context of decreased HIV/AIDS funding. Global Public Health, 9(1-2). Collaboration NCD Risk Factor. (2016). Trends in adult body-mass index in 200 countries from 1975 to 2014: a pooled analysis of 1698 population based measurement studies with 19.2 million participants. Lancet, 387, 1377-1396. Corbett, E., Charalambous, S., Moloi, V., Fielding, K., Grant, A., Dye, C., . . . Churchyard, G. (2004). Human Immunodefi- ciency Virus and the Prevalence of Undiagnosed Tuberculosis in African Gold Miners. American Journal of Respiratory and Critical Care Medicine, Vol 170 No 6. De Savigny, D., & Adam, T. (2009). Systems thinking for health systems strengthening. World Health Organization. Dell, S. (2013). South Africa: Less HIV in universities than nationally. Global Window on Higher Education, 15:9. Deloitte. (2012). Social Security Commission, National Medical Benefit Fund. http://www.ssc.org.na/files/downloads/1c4_ NMBF%20Report%2031%20July%202012%20clean%20send.pdf. Dionne, K. (2015). HIV Exceptionalism: Development through Disease in Sierra Leone by Adia Benton (review). African Studies Review, 58(2), pp.257-258. Ejughemre, U., & Oyibo, P. (2014). Healthcare financing in Nigeria: A systematic review assessing the evidence of the impact of health insurance on primary health care delivery. Journal of Hospital Administration., Issue 1. Vol 4. Erasmus, D., Ranchod, S., Abraham, M., Carvounes, A., & Dreyer, K. (2016). Challenges and opportunities for health finance in South Africa: a supply and regulatory perspective. Insight Actuaries and Consultants. Evian, C., Fox, M., MacLeod, W., Slotow, S., & Rosen, S. (2004). Prevalence of HIV in workforces in southern Africa, 2000-2001. South Sudan Medical Journal, 94(2), pp.125-130. Feeley, F., Connelly, P., & Rosen, S. (2007). Private sector provision and financing of AIDS treatment in Africa: current developments. Curr HIV/AIDS Rep, 4:192-200. Feeley, F., de Beer, I., de Wit, T., & Van der Gaag, J. (2006). The health insurance industry in Namibia: Baseline Report. Center for International Health. Geoffroy, E., Harries, D., Bissell, K., Schell, E., Bvumbwe, A., Tayler-Smith, K., & Kizito, W. (2014). Bringing care to the community: expanding access to health. Public Health Action: International Union Against Tuberculosis and Lung Disease, Vol 4 No 4. Gustafsson- Wright, E., de Beer, I., Gaeb, E., van Rooy, G., van der Gaag, J., & Rinke de Wit, T. (2010). The Okambilimbili Health Insurance Project in Namibia: lessons learnt. The Netherlands: PharmAccess Foundation, Amsterdam Institute for Global Health and Development.

152 Gustafsson-Wright, E., Asfaw, A., & van der Gaag, J. (2009, November). Willingness to pay for health insurance: an analysis of the potential market for new low-cost health insurance products in Namibia. Social Science and Medicine, 69(9), 1351-9. Gustafsson-Wright, E., de Beer, I., Gaeb, E., van Rooy, G., van der Gaag, J., & Rinke de Wit, T. (2010). The Okambilimbili Health Insurance Project in Namibia: lessons learnt. The Netherlands: PharmAccess International, Amsterdam Institute for Global Health and Develpoment. Hendricks, M., Wit, F., Roos, M., Brewster, L., Akande, T., de Beer, I., . . . C., S. (2012, March). Hypertension in Sub-Sa- haran Africa: Cross Sectional Surveys in four Rural and Urban Communities. Plos ONE, 7(3). Hohmann, J., & Skolnic, R. (2004). Options and scenarios for HIV/AIDS risk cover for low-income employees within NABCOA member companies in Namibia. Windhoek: NABCOA. Hohmann, J., & Skolnic, R. (2004). Options and scenarios for HIV/AIDS risk cover for low-income employees within NABCOA member companies. Report on the feasibility study. Windhoek: NABCOA. Holzemer WL, U. L. (2004). Managing AIDS Stigma. Sahara J, I:165-174. Igumbor, J., Pacoe, S., Rajap, S., Townsend, W., Sargent, J., & Darkoh, E. (2014, October). A South African public-private partnership HIV treatment model: Viability and success factors. PLoS One, 9(10). Igumbor, J., Pascoe, S., Rajap, S., Townsend, W., Sargent, J., & Darkoh, E. (2014, October). A South African public private partnership HIV treatment model: viability and success factors. PLoS One, 9(10):e110635. Johnson, M. S. (2016). SafeCare: An Innovative Approach for Improving Quality Through Standards, Benchmarking, and Improvement in Low-and Middle-Income Countries. The Joint Commission Journal on Quality and Patient Safety, 42(8), pp.350-AP11. Johnson, M., Schellekens, O., Stewart, J., van Ostenberg, P., de Wit, T., & Spieker, N. (2016). An Innovative Approach for Improving Quality Through Strandards, Benchmarking and Improvement in Low-and Middle-Income Countries. The Joint Commission Journal on Quality and Patient Safety, 42(8), pp.350-AP11. Kenny, A., Basu, G., Ballard, M., Griffiths, T., Kentoffio, K., Niyonzima, J., . . . Kraemer, J. (2015). Remoteness and maternal and child healthservice utilization in rural Liberia: A population–based survey. Journal of Global Health, Vol. 5 No. 2. Kirigia, M., & Barry, S. (2008). Health challenges in Africa and the way forward. International Archives of Medicine, 1:27. Kronick, R., & Gilmer, T. (2002). Insuring low-income adults: does public coverage crowd out private? Health Affairs, 21(1):225-239. Lange, J., Schellekens, O., Lindner, M., & van der Gaag, J. (2008). Public–private partnerships and new models of healthcare access. Current Opinion in HIV and AIDS., 3(4), pp.509-513. Lima, V., Thirumurthy, H., Kahn, J., Saavedra, J., Cárceres, C., & Whiteside, A. (2014). Modeling scenarios for the end of AIDS. Clinical Infectious Diseases, 59(suppl 1), pp.S16-S20. Mavedzenge, S., Luecke, E., Lopez, A., Wagner, D., Hartmann, M., Lutnick, A., & Lambdin, B. (2016). HIV Testing among Key Populations, Adolescent Girls and Men in Eastern and Southern Africa: A Review of Research, Policy and Programming. Methods (2) 2016. Mc Pake, B., & Hanson, K. (2016). Managing the public-private mix to achieve universal health coverage. The Lancet, 388(10044), 622-630. McPake B, H. K. (2016, August). Managing the public–private mix to achieve universal health coverage. The Lancet, 388(10044), 622-630. Meehan, S., Leon, N., Naidoo, P., Jennings, K., Burger, R., & Beyers, N. (2015). Availability and acceptability of HIV counselling and testing services. A qualitative study comparing clients’ experiences of accessing HIV testing at public sector primary health care facilities or non-governmental mobile services in Cape Town, South Afr. BMC Public Health, 15:845.

153 Mishra, V., Vaessen, M., Boerma, J., Arnold, F., Way, A., & Barrere, B. (2006). HIV testing in national population-based surveys: experience from the Demographic and Health Surveys. Bulletin of the World Health Organization, 84:537- 545. MoHSS. (2008). Namibia National Health Accounts 2001/2-2006/7. Republic of Namibia Ministry of Health and Social Services. MoHSS. (2015). Namibia National Health Accounts 2012/3012. Namibia: Ministry of Health and Social Services. MoHSS, R. o. (2003). Report of the 2002 National HIV Sentinel Survey. Republic of Namibia Ministry of Health and Social Services. MSMHS, M. S. (2013). Background (online). http://www.mistersisterclinics.org/aboutClinic.php. Mulu, W., Abera, B., & Yimer, M. (2014). Prevalence of Human Immunodeficiency Virus infection and associated factors amongst students at Bahir Dar University. Ethiop. J. Health Dev., 28 (3): 170-177. Mussa, A., Pfeiffer, J., Gloyd, S., & Sherr, K. (2013). Vertical funding, non-governmental organizations, and health system strengthening: perspectives of public sector health workers in Mozambique. Human resources for health, 11(1), p.1. NABCOA. (2004). Healthcare survey amongst members of the NABCOA. Namibia Business Coalition on AIDS. NABCOA. (2004). HIV/AIDS amongst NABCOA member companies. NABCOA. NAMAF. (2004). NAMAF Annual Conference . Namibia Statistics Agency. (2013). Profile of Namibia. http://cms.my.na/assets/documents/p19dpmrmdp1bqf19s2u8pis- c1l4b1.pdf. Namibian Central Bureau of Statistics. (2006). 2001 Population Projections. http://cms.my.na/assets/documents/ p19dmrhrpm1bpl1jre1p3quqm1tjc1.pdf. Nattrass, N., Hodes, R., & Cluver, L. (2016). Changing Donor Funding and the Challenges of Integrated HIV Treatment. AMA Journal of Ethics, 18(7), p.681. Novignon, J., & Olakojo, S. (2012, December). The effects of public and private health care expenditure on health status in sub-Saharan Africa: new evidence from panel data analysis. Health Econ Rev, 2(1):22. O’Hanlon, B., Feeley, F., de Beer, I., Sulzbach, S., & Vincent, H. (2010). Namibia Private Sector Assessment. Strengthe- ning Health Outcomes through the Private Sector, Abt Associates Inc. PharmAccess. (2010). Report of the mobile clinic pilot project in Hochfeld. Windhoek: PharmAccess Foundation. Piot, P., Greener, R., & Russel, S. (2007). Squaring the Circle: AIDS, poverty, and human developemnt. PLoS Med, 4:1571-1575. Preker, A., Lindner, M., Chernichovsky, D., & Schellekens, O. (2013). Scaling Up Affordable Health Insurance. Washington DC: The International Bank for Reconstruction and Development / World Bank. Preker, A., Lindner, M., Chernichovsky, D., & Schellekens, O. (2103). Scaling Up Affordable Health Insurance : Staying the Course. Preker, A., Schellekens, O., & Lindner, M. (2010). Global Marketplace for Private Health Insurance. Washington DC: The International Bank for Reconstruction and Development / World Bank. Rees, C., Hawkesworth, S., Moore, S., Dondeh, B., & Unger, S. (2016). Factors Affecting Access to Healthcare: An Observational Study of Children under 5 Years of Age Presenting to a Rural Gambian Primary Healthcare Centre. PLoS ONE, 11(6): e0157790. doi:10.1371/journal.pone.0157790. Republic of Namibia, A. 3. (1994). http://www.ssc.org.na/files/downloads/1b2_SSC Act 34 of 1994.pdf. Resch, S., Ryckman, T., & Hecht, R. (2015, January). Funding AIDS programmes in the era of shared responsibility: an analysis of domestic spending in 12 low-income and middle-income countries. The Lancet. Global Health, 3(1), 52-61. Roa, P., Gabre-Kidan, T., Mubangizi, D., & Sulzbach, S. (2011, August). Leveraging the private health sector to enahance HIV service delivery in lower-income countries. Journal of Acquired Immune Deficiency Syndrome, 57 Supl 2:S111-9.

154 Rosen, S., Feeley, F., Connelly, P., & Simon, J. (2007). The private sector and HIV/AIDS in Africa: taking stock of 6 years of applied research. AIDS 2007. Rosenberg, A., Hartwig, K., & Merson, M. (2008). Government-NGO collaboration and sustainability of orphans and vulnerable children projects in southern Africa. Journal of Evaluation and Program Planning, 51-60. Sabin, L., Bachman DeSilva, M., Hollenbeck-Pringle, D., Ashigbie, P., Purcell, S., Robinson, A., . . . I, d. B. (2014). Improving access and quality of healthcare through mobile clinics in Windhoek. Center for Global Health and Development, Boston University. Schellekens, O., Lindner, M., Lange, J., & van der Gaag, J. (2007, October 17). A new paradigm for increased access to healthcare in Africa. Annual IFC / Financial Times Essay Competition. Schellekens, O., Lindner, M., Lange, J., & van der Gaag, J. (2007, October 15). A new paradigm for increased access to healthcare in Africa. Annual IFC / Financial Times Essay competition. Schiffman, J. (2008). Has donor prioritization of HIV/AIDS displaced aid for other health issues? Health Policy and Planning, 23(2), 95-100. Sekhri, N., & Savedoff, W. (2005). Private health insurance: implications for developing countries. Bulletin of the World Health Organization, 83:127-134. SSC. (2016). https://www.ssc.org.na/UHCAN/Universal-Health-Coverage-Advisory-Committee-of-Namibia/65/. Stallworthy, G., Boahene, K., Ohiri, K., Pamba, A., & Knezovich, J. (2014, June). Roundtable discussion: what is the future role of the private sector in health? Globalization and Health, 10:55. Sulzbach, S., De, S., & Wang, W. (2011). The private sector role in HIV/AIDS in the context of an expanded global response: expenditure trends in five sub-Saharan African countries.Health Policy and Planning., Vol. 1: 26 pages i72-i84. Tan, D., Upshur, R., & Ford, N. (2015). Global plagues and the Global Fund: challenges in the fight against HIV TB and Malaria. BMC International Health and Human Rights, 3(1), p.29. UNAIDS/WHO. (2008). Epidemiological Fact Sheet on HIV and AIDS. UNDP. (2007). Human Development Indicators. In: Human development report 2007/8. UNDP http://hdr.undp.org/en/ media/HDR_20072008_EN_Indicator_tables.pdf. Uzochukwu, B., Ughasoro, M., Etiaba, E., Okwuosa, C., Envuladu, E., & Onwujekwe, O. (2015, July-August). Health care financing in Nigeria: Implications for achieving universal health coverage.Nigerian Journal of Clinical Practice, 18(4), 437-44. van den Heever, A. M. (2012). The role of insurance in the achievement of universal coverage withn a developing country context: South Africa as a case study. BMC Public Health, 12 Suppl 1:S%(E pub 2012 Jun 22). van den Heever, A. M. (2016, June). South Africa’s universal health coverage reforms in the post-apartheid period. Health Policy, http://dx.doi.org/10.1016/j.healthpol.2016.05.012. Van der Gaag, J., & Stimac, V. (2008). Towards a new paradigm for health sector development. Amsterdam Institute for International Development. van Renterghem, H., Colvin, M., De Beer, I., Gunthorp, J., Odiit, M., Thomas, L., . . . Getcha, M. (2012). City epidemics matter. AIDS 2012. Van Renterghem, H., Colvin, M., de Beer, I., Gunthorp, J., Odiit, M., Thomas, L., . . . Getcha, M. (2012). The urban HIV epidemic in eastern and southern Africa: need for better KYE/KYR to inform adequate city repsonses. Journal of the International AIDS Society, 15. White, H., Kristensen, S., & Coulibaly, D. (2009). Prevalence and Predictors of HIV Infection amongst Malian Students. AIDS Care, 21:701-707. Whittaker, S. S. (2011). Quality standards for healthcare establishments in South Africa. repository.up.ac.za. WHO. (2010). Improving health systems efficiency as a means of moving towards universal health coverage. Background paper, 28. Geneva: WHO.

155 Whyle, E. (2016). Models of public private engagement for health service delivery and financing in Southern Africa: a systematic review. Health Policy and Planning, pii: czw 075. World Bank. (2003). http://data.worldbank.org/indicator/SI.POV.GINI. World Bank. (2004). https:// http://data.worldbank.org/indicator/NY.GDP.PCAP.CD?locations=NA. World Bank. (2004). How does the World Bank classify countries? https://datahelpdesk.worldbank.org/knowledgebase/ articles/378834-how-does-the-world-bank-classify-countries.

156 ADDENDUM

Summary (in English)

Engaging the private sector in African healthcare is a central theme in the work of PharmAccess Foun- dation, a Dutch not-for-profit organization, based in Amsterdam, with offices in several African countries, including Namibia. This thesis describes interventions developed and applied in Namibia to engage its private sector as an agent to develop and leverage private resources for public health. Interventions and subsequent evaluations are presented according to the PharmAccess model of transitioning the vicious circle of poor healthcare in Africa into a virtuous cycle of sustainable healthcare stimulating both demand and supply, while recognizing the key stakeholders of healthcare systems as: the patient, the provider and the payer, all three of them functioning in a policy environment that sets the rules of interaction and exchange.

The studies presented in this thesis describe and evaluate important elements of the PharmAccess strategy and activities of engaging the private sector in the public healthcare challenges in Namibia. Topics have been arranged according to strategies that: • stimulate demand by improving health awareness of patients/clients and enabling them to get access to better healthcare • stimulate supply by capacity building, introducing quality benchmarks and mobilizing innovative financing • stimulate prepayment and risk sharing for healthcare using innovative payer mechanisms • collect and present scientific evidence on the effectiveness of these interventions to eventually support policy-making.

The PharmAccess paradigm recommends interventions at all levels of the healthcare system. The ultimate goal is to provide the right diagnosis, at the right time, for the right patient at the right cost within the right legal and regulatory framework, with both public and private sector contributing a complementary role. The stakeholders in the healthcare system cannot function in isolation, and this thesis demonstrates that interventions aimed at stimulating one stakeholder have an effect either directly or indirectly on the others.

The first PharmAccess strategy to stimulate organized demand for healthcare is supported in this thesis by the publications presented in Chapters 2-5.

The validation of the diagnostic accuracy of the OraQuick® HIV-1/2 Rapid Antibody test and the Oral Fluid Vironostika® HIV Uniform II microELISA system with the Namibian National Algorithm for HIV Testing is presented in Chapter 2. This study was the first formal field evaluation of the OraQuick® and OraSure®

157 devices, in a resource limited setting in Southern Africa. The study found that the OraQuick® test is 100% accurate (100% sensitivity and specificity) and that the OraSure® has a high specificity (99.5%) and slightly lower sensitivity (97.1%) The high negative predictive value of both tests rendered them particularly suitable for HIV surveillance purposes. Based on this validation the Namibian MoHSS approved the OraQuick® and OraSure® devices for surveillance purposes in Namibia. This study paved the way for the implementation of non-invasive HIV screening in Namibia, both in workplace and household studies.

Chapter 3 presents the results of HIV prevalence surveillances that were conducted in 24 Namibian private companies, from 11 different industries. Participation rates amongst the 8,500 employees ranged from 61.3- 97.3% with a mean participation rate of 78.6%. An average HIV prevalence rate of 15% was found, although the HIV prevalence varied widely between industries, ranging from 4% in the information technology sector to 21% in the mining sector. The studies not only provided the first prevalence data from a portion of the working population across various industry sectors, but also demonstrated how knowledge of the HIV prevalence amongst the workforce stimulated enrolment in private health insurance for HIV in Namibia. Following the HIV prevalence surveillance, 61% of the previously uninsured employees enrolled in health insurance, the majority of these insurances (78%) covering HIV. This study was important in dispelling the myth that HIV was a general population and not a workforce problem.

In Chapter 4, the study of the HIV prevalence, knowledge of HIV and health care utilization trends of students at the two largest academic institutions in Namibia – the future intellectual workforce of Namibia – is presented. The study found, although it may have been underestimated, that students had relatively low HIV prevalence rate (1.8% at UNAM and 2.8% at Polytechnic) compared to the estimated national adult average of 15.3% at the time in Namibia. Need for intensified HIV prevention (to remain HIV negative) and treatment amongst those University students who were infected were recommended to policy-makers.

Chapter 5 presents a study conducted to evaluate the healthcare challenges and the related needs for health services amongst rural and remote employers in the commercial farming sector in Namibia. It was assumed that commercial farm workers could be a good target group for private health insurance, which includes benefits for the treatment of HIV. It was found that this group did not have adequate access to formal healthcare and particularly to HIV related services. The willingness of farmers/employers to contribute to the cost of health insurance in Namibia appeared high, however access to health services, particularly due to large distances posed a significant obstacle.

Research related to a second PharmAccess strategy to stimulate supply by developing innovative healthcare service provision is demonstrated in Chapters 6 and 7. Chapter 6 presents the public-private partnership (PPP) model ‘Bophelo!’ providing mobile wellness screening at workplaces in Namibia, demonstrated the potential to mobilize private sector funding and be a cost-effective manner to reach individuals in workplaces and other hard-to-reach areas. To reduce the stigma associated with HIV testing, the screening package included blood pressure and body mass index (BMI) readings; and blood screening for HIV, blood glucose, cholesterol, haemoglobin and hepatitis B. The inclusion of wellness screening in addition to stand-alone HIV testing illustrated a number of benefits at only a marginal additional cost and could thus assist to help address the growing issue of non-communicable diseases in Namibian workplaces.

158 In Chapter 7, the Mister Sister Mobile Clinics providing primary health care services in a PPP to rural and remote employers are presented. The services were implemented to reduce the geographic barriers to preventative and curative services as were identified in the earlier study presented inChapter 5. Using mechanisms such as this PPP could leverage private sector resources to not only improve the health of employees and dependents directly associated with the workplace, but vulnerable groups – orphans and vulnerable children- within the geographic area of service provision.

Another important PharmAccess strategy, to implement new healthcare prepayment and risk-sharing mechanisms, is presented in Chapter 8. This chapter addresses the question whether low income health insurance/medical aid funding can be developed sustainably in Namibia with short-term donor subsidy to avoid the crowding out of the private sector and leverage private sector resources for HIV care and treatment. It is shown that through the Health is Vital Risk Equalization Fund (HIVREF) additional prepaid resources for health could be raised that were several times higher than the short-term donor subsidy. This additional revenue was financed by higher income populations and demonstrated how cost sharing could be achieved while simultaneously quality of care for HIV was improved. After the first 20 months (2006/2007), the HIVREF had enrolled over 34,000 people, which was equivalent to approximately 15% of the Namibian formal sector workforce. However, during the same period the Namibian public treatment program to fight HIV, TB and malaria, highly subsidized by the US program PEPFAR and by the Global Fund, was implemented and scaled-up nationwide. Subsequent successful implementation of free access to HIV treatment in public health facilities had a direct effect on the new enrolment numbers of the HIVREF and resulted in crowding out of the temporary gains made.

An overarching PharmAccess strategy, to (make) collect and present scientific evidence pertinent to its interventions to support (local) policy-making is presented in Chapters 9-11.

Chapter 9a presents the data providing the first ever HIV prevalence and incidence data from a 3-year panel household survey conducted in the City of Windhoek. The findings of HIV infection rates of 2-3 per 100PY (2%-3% incidence) provided an evidence base to advocate against complacency in the HIV response. This study underscored the importance of using incidence to understand the HIV epidemic in addition to the more commonly used HIV prevalence rates. A remarkable lesson resulting from the implementation of the study was the risk of surveyor bias as described in Chapter 9b, which highlights potential limitations of household surveys. Chapter 10 presents the findings of a larger Bophelo! wellness screening dataset specifically for the prevalence, knowledge and self-perceived risk of employees, of HIV, hypertension and diabetes. It was found that the prevalence of elevated blood pressure, elevated blood glucose, and HIV among employees of the Namibian formal sector is high, while risk awareness is low. This data suggested that attention must be paid to improving the knowledge of risk factors related to these conditions as well as providing care to those with chronic conditions in the formal sector through workplace wellness programs.

In Chapter 11 evidence that both non-communicable disease risk factors and infectious diseases are associated with increased rates of short-term absenteeism of formal sector employees in Namibia is presented. This evidence suggests that workplace wellness programs, which include screening and basic health insurance benefits to manage these conditions could help employers avoid costs associated with absenteeism.

159 The work of PharmAccess in Namibia over the years illustrated the intricateness and daunting complexity of health systems in Africa in general. This thesis also demonstrates how Namibia could be considered as a ‘phase zero’ country for PharmAccess for ideas and innovations that would later be further refined and appropriated for use in other countries. For example, the HIFREV crowding out experience contributed to subsequent insurance models by the Health Insurance Fund (HIF) in Nigeria, Kenya, Tanzania and the household surveys framework for Namibia was later implemented in these countries too.

Besides the learning for the PharmAccess Group the programs and interventions have brought significant changes in the healthcare system of Namibia and built trust between the public and private sector to jointly provide health services using a mixed funding model. Recommendations are made in this thesis for the Namibian healthcare system in light of the declining donor funding and the country’s striving to achieve universal health coverage. The recommendations are to: • Continue stimulating and understanding the demand of the patient/client through increased efforts to provide awareness of wellness and individual healthcare needs both in formal sector and in communities; • Stimulate supply by expanding innovative quality healthcare service provision through the PPP on service delivery mechanism; • Innovate payer structures to channel health risks into prepayment and risk equalization mechanisms; • Develop specific private health insurances for formally employed, including low income, groups; • Expand the role of MoHSS to facilitate policy making and regulation for public private partnerships; • Establish an independent national health information hub to support evidence for policy making; • Establish national standards for quality of healthcare services and health information.

In conclusion the studies presented in this thesis motivate further engagement of the private sector towards addressing public health challenges in Namibia not only for HIV, but for other chronic and non-communicable diseases by stimulating demand through the patient/client, directing demand to quality providers through the establishment of innovative public-private partnerships, innovating payer mechanisms for healthcare and providing evidence for policy making.

160 Samenvatting (Dutch summary)

De private sector betrekken bij de Afrikaanse gezondheidszorg is een centraal thema in het werk van PharmAccess. PharmAccess is een Nederlandse non-gouvernementele organisatie met het hoofdkantoor in Amsterdam en diverse kantoren in Afrika, waaronder Namibië. Dit proefschrift beschrijft verschillende interventies die ontwikkeld en toegepast zijn in Namibië met als doel de private sector te betrekken bij de publieke gezondheidszorg om daarmee mogelijkheden te vergroten voor verbetering van kwaliteit en toegankelijkheid. De interventies en evaluaties van deze interventies worden in dit proefschrift gepresenteerd volgens het PharmAccess model. Dit model beoogt de vicieuze cirkel van benendenmaatse gezondheidszorg in Afrika te transformeren naar een cirkel van duurzame gezondheidzorg die de zorgvraag stimuleert en rekening houdt met alle betrokkenen in het zorgsysteem: de patiënt, de zorgaanbieder en de financier, die gezamenlijk functioneren onder het beleid van een zorgsysteem, dat de regels voor interactie en transactie definieert.

De studies in dit proefschrift beschrijven en evalueren belangrijke elementen van de strategie en activiteiten van PharmAccess om de private sector te betrekken bij de uitdagingen van de publieke gezondheidszorg in Namibië. Onderwerpen zijn gerangschikt naar gelang activiteiten die: • de zorgvraag stimuleren door het creeren van bewustwording rond gezondheid en het bevorderen van toegang tot betere zorg • het zorgaanbod stimuleren door capaciteitsopbouw, benchmarking op kwaliteit en het mobiliseren van innovatieve financiele stimuli • vooruitbetaling en risicospreiding voor zorg stimuleren door middel van innovatieve financieringsmod- ellen • wetenschappelijk bewijs verzamelen en presenteren aangaande de effectiviteit van deze interventies ter uiteindelijke ondersteuning van beleidsvorming.

Het PharmAccess paradigma pleit voor interventies op alle niveaus van het zorgsysteem. Het uiteindelijk doel is om de juiste diagnose te stellen, op het juiste moment, bij de juiste patiënt, tegen de juiste kosten, binnen het juiste juridische kader, met zowel een private- als een publieke sector die elkaar aanvullen. De verschillende stakeholders kunnen niet los van elkaar functioneren. Dit promotieonderzoek laat zien dat een inter- ventie bij een van de stakeholders direct of indirect effect heeft op de andere stakeholders.

De eerste strategie van PharmAccess, stimuleren van een georganiseerde zorgvraag, is onderbouwd door een aantal studies die gepresenteerd worden in de hoofdstukken 2-5.

De validatie van de diagnostische accuraatheid van OraQuick© HIV-1-2 Rapid Antibody test en Oral Fluid Vironostika© HIV Uniform II microELISA system met het Namibische nationale algoritme voor hiv-testen is gepresenteerd in hoofdstuk 2. Dit was de eerste officiële evaluatiestudie van OraQuick© test en OraSure© in een Zuidelijk Afrikaanse context. De studie wijst uit dat OraQuick© 100% betrouwbaar is (100% sensitiviteit en specificiteit). OraSure© heeft eveneens een hoge specificiteit (99,5%) en iets lagere sensitiviteit (97.1%). Door de hoge negatieve voorspellende waarden zijn beide testen bijzonder geschikt voor hiv-surveillance. Op basis van deze validatie heeft het Namibische Ministerie voor Volksgezondheid OraQuick© en OraSure© goedgekeurd voor surveillancedoeleinden. Deze studie heeft de weg vrij gemaakt in Namibië voor het gebruik van niet-invasieve hiv-testen bij verschillende studies.

161 In hoofdstuk 3 worden de resultaten gepresenteerd van het hiv-prevalentieonderzoek bij 24 Namibische bedrijven in 11 verschillende sectoren. De deelname onder 8500 werknemers varieerde van 61,3 tot 97,3%, met een gemiddelde deelname van 78,6%. De gemiddelde hiv-prevalentie was 15%. De spreiding was echter groot: van 4% in de informatie- en technologiesector tot 21% in de mijnbouwsector. Naast prevalentie- data voor de verschillende sectoren liet deze studie ook zien hoe kennis over hiv-prevalentie onder de beroepsbevolking de deelname aan private zorgverzekeringen in Namibië stimuleerde. Na het hiv-prevalentieonderzoek werd 61% van de onverzekerden alsnog verzekerd. De meerderheid daarvan (78%) betrof een zorgverzekering met dekking voor hiv. Deze studie ontzenuwde de mythe dat hiv een probleem was van alleen de algemene bevolking en niet van de beroepsbevolking.

Hoofdstuk 4 laat de resultaten zien van onderzoek naar hiv-prevalentie, kennis over hiv en het gebruik van gezondheidszorg bij studenten op de twee grootste universiteiten (UNAM en Polytechnic) – het toekomstige intellectuele potentieel van Namibië. De prevalentie onder studenten was relatief laag (1,8% op UNAM en 2,8% op Polytechnic) in vergelijking met de prevalentie onder volwassenen op nationaal niveau (15,3%). Er is mogelijk sprake van onderschatting. Meer aandacht voor preventie van hiv (om hiv-negatief te blijven) en behandeling van hiv onder studenten waren aanbevelingen aan beleidsmakers.

In hoofdstuk 5 worden de resultaten gepresenteerd van een studie naar de zorgverlening aan werknemers in de commerciële landbouw in de rurale en afgelegen gebieden van Namibië. De verwachting was dat werknemers in deze sector een goede doelgroep vormen voor private zorgverzekeringen, inclusief dekking voor hiv. Uit het onderzoek kwam echter naar voren dat deze groep geen goede toegang had tot de formele zorg en in het bijzonder tot de hiv-zorgverlening. De bereidheid van boeren/werkgevers om bij te dragen aan de kosten van een zorgverzekering bleek groot. Maar de toegang tot zorgverlening vormde een groot obstakel in het bijzonder als gevolg van grote afstanden.

Onderzoek naar de tweede strategie van PharmAccess, stimuleren van het zorgaanbod door het ontwikkelen van innovatieve zorgverlening, wordt gepresenteerd in de hoofdstukken 6 en 7.

In hoofdstuk 6 wordt het publiek-private samenwerkingsmodel ‘Bophelo!’ beschreven. Dit model biedt mobiele gezondheidsscreening op de werkvloer en laat de potentie zien om private financiering in te zetten om zo op een kosteneffectieve manier werknemers te bereiken in moeilijk bereikbare gebieden. Om het stigma op het testen op hiv te reduceren is de screening uitgebreid met het meten van de bloeddruk en BMI en bloedtesten voor bloedsuiker, cholesterol, hemoglobine en hepatitis B. Het toevoegen van deze ‘gezonde leefstijl’ parameters aan het hiv-testen leverde voordelen op, tegen beperkte additionele kosten en zou kunnen helpen bij de aanpak van leefstijl-gerelateerde aandoeningen bij Namibische werknemers.

In hoofdstuk 7 worden de Mister Sister Mobile Clinics gepresenteerd. Deze verlenen basiszorg in een publiek-private samenwerking in rurale en afgelegen gebieden. Deze zorgverlening wordt aangeboden om de geografische barrières te slechten voor preventieve en curatieve zorgverlening, zoals geconcludeerd in eerder onderzoek (zie hoofdstuk 5). Private-publieke samenwerking maakt het mogelijk om middelen uit de private sector aan te wenden voor het verbeteren van de gezondheid van werknemers en hun gezinsleden, maar ook van kwetsbare groepen zoals wezen en kwetsbare kinderen in hetzelfde geografische gebied.

Een andere belangrijke strategie van PharmAccess is het invoeren van nieuwe financieringsmechanismen voor zorg, die vooruitbetaling en risicospreiding met zich meebrengen. Onderzoek hiernaar wordt uiteengezet

162 in hoofdstuk 8. Dit onderzoek zoekt antwoord op de vraag of het meet korte-termijn donorgelden mogelijk is om een duurzame zorgverzekering te ontwikkelen voor mensen met een laag inkomen zodanig dat crowding out wordt voorkomen en extra private middelen voor hiv-behandeling en zorg worden gerecruteerd. Door de ‘Health is Vital Risk sector Equalization Fund’ (HIVREF) kwam extra financiering voor de zorg beschikbaar die vele malen hoger was dan de korte termijn donorgelden. Deze additionele bronnen werden beschikbaar gesteld door mensen met hogere inkomens, hiermee aantonend dat risicospreiding kan worden gerealiseerd, terwijl tegelijkertijd de kwaliteit van de hiv-zorgverlening verbetert.

Na de eerste 20 maanden (2006/2007) had HIVREF ruim 34.000 nieuwe verzekerden, ongeveer 15% van de in de formele sector werkende bevolking van Namibië. Echter, gedurende diezelfde periode werd het Namibische nationale programma voor hiv, tuberculose en malaria opgeschaald met substantiele subsidies van het Amerikaanse hulpprogramma PEPFAR en het Global Fund. De succesvolle lancering van een gratis behandelprogramma voor hiv had direct effect op het aantal nieuwe verzekerden bij HIVREF en leidde tot crowding out van de tijdelijk bereikte resultaten.

Een veelomvattende strategie van PharmAccess is het (laten) verzamelen en presenteren van wetenschappelijk bewijs rond verrichtte interventies ter ondersteuning van (locale) beleidsvorming. Deze komt aan de orde in de hoofdstukken 9-11.

In hoofdstuk 9a worden de resultaten gepresenteerd van een drie jaar lopende huishouden-enquête naar de incidentie en prevalentie van hiv in Windhoek. Het aantal nieuwe hiv-geïnfecteerden van 2-3 per 100 personen per jaar (2%-3% incidentie) was voldoende bewijs om beperkte waakzaamheid in de strijd tegen hiv aan te vechten. Deze studie toont het belang aan van onderzoek naar de incidentie van hiv om de epid- emie te begrijpen, in aanvulling op de vaker gebruikte prevalentie. Een opmerkelijke les die getrokken kon worden uit deze studie was het risico van onderzoekersbias zoals beschreven in hoofdstuk 9b, dat gaat over de mogelijke beperkingen van het uitvoeren van enquêtes onder huishoudens.

Hoofdstuk 10 geeft de resultaten van de Bophelo! gezondheidsscreening naar de prevalentie, kennis en risicoperceptie onder werknemers over hiv, hoge bloeddruk en diabetes. De prevalentie van hoge bloeddruk, verhoogde glucosewaarden en hiv was hoog, terwijl de risicoperceptie laag bleek. Dit indiceert dat er meer aandacht moet komen voor het verhogen van de kennis over risicofactoren die gerelateerd zijn aan deze aandoeningen. Ook moet de zorg voor mensen met deze aandoeningen verbeteren door introductie van gezondheidsprogramma’s op de werkvloer.

Hoofdstuk 11 laat zien dat zowel de risicofactoren voor niet-overdraagbare ziekten als voor infectieziekten zijn geassocieerd met verhoogd verzuim. Dit zou kunnen betekenen dat gezondheidsprogramma’s op de werkvloer, inclusief screening en een basale ziektekostenverzekering, werkgevers kunnen helpen om kosten te vermijden die samenhangen met verzuim.

Het werk van PharmAccess in Namibië van de afgelopen jaren illustreert de grote uitdadingen en complex- iteit van zorgsystemen in Afrika in het algemeen. Dit proefschrift laat ook zien dat Namibië een ‘fase nul’ land was voor PharmAccess waar ideeën en innovaties zijn onderzocht die na aanscherping kunnen worden hergebruikt in andere landen. De ervaring met crowding out van het HIVREF droeg bij aan de daarop volgende verzekeringsmodellen van het Health Insurance Fund (HIF) in Nigeria, Kenia, Tanzania en de huishoud-enquêtes in Namibie werden later in deze landen gebruikt.

163

Naast deze meerwaarde voor PharmAccess, hebben de programma’s en interventies het zorgsysteem van Namibië significant veranderd. Er is vertrouwen gecreëerd bij de publieke en private sector om gezamenlijk zorg te verlenen met een gemengd financieringsmodel. Op basis van de resultaten van de onderzoeken beschreven in dit proefschrift, kunnen aanbevelingen worden gedaan voor het Namibische zorgsysteem om financieel minder afhankelijk te worden van donoren en universele toegang tot gezondheidszorg te verwezen- lijken. Deze aanbevelingen luiden als volgt: • Doorgaan met het stimuleren van de zorgvraag van de patiënt/cliënt door meer inspanningen om bewustzijn te creëren rond welzijn en individuele zorgbehoefte, zowel in de formele als informele sector. • Stimuleren van het aanbod door middel van innovatieve en kwalitatieve zorgverlening met publiek-private samenwerking. • Innoveren van financieringsstructuren om gezondheidsrisico’s te kanaliseren richting mechanismen voor het vooruitbetalen en delen van risico’s. • Ontwikkelen van specifieke privé-zorgverzekeringen voor werknemers, inclusief de werknemers met lage inkomens. • Uitbreiden van de rol van het ministerie van volksgezondheid om het beleid en regulering ten aanzien van publiek-privaat partnerschap te faciliteren. • Oprichten van een onafhankelijke informatievoorziening rond gezondheidszorg, die wetenschappelijke onderbouwing voor beleid ondersteunt. • Vaststellen van nationale kwaliteitsstandaarden voor goede zorgverlening en gezondheidsinformatie.

Samenvattend pleit het gepresenteerde onderzoek in dit proefschrift voor intensievere betrokkenheid van de private sector bij het adresseren van de uitdagingen in de publieke gezondheidszorg in Namibië, niet alleen voor hiv maar ook voor andere aandoeningen. Dit is mogelijk door het stimuleren van de zorgvraag van de patiënt/cliënt, door het sturen van de vraag naar kwaliteitszorgaanbieders door middel van innovatieve publiek-private samenwerking, door de introductie van innovatieve financieringsmechanismen voor zorg en door het verzamelen van wetenschappelijk bewijs voor beleidsvorming.

164 List of Publications

Peer-reviewed publications: 1 Hamers RL, de Beer IH, Kaura H, van Vugt M, Caparos L, Rinke de Wit TF: Diagnostic accuracy of 2 oral fluid-based tests for HIV surveillance in Namibia. J Acquir Immune Defic Syndr 2008,48:116-118. 2 Hamers R.L., De Beer I.H., Kaura H., van Vugt M., Caparos L., Rinke de Wit T.F. Diagnostic accuracy of 2 oral fluid-based tests for HIV surveillance in Namibia. Epistola International. Edition 2008; 1:21-23. 3 Janssens W, Gustafsson-Wright E, de Beer I, van der Gaag J: A Unique Low-cost Private Health Insur- ance Program in Namibia: Protection from Health Shocks Including HIV/AIDS. Development Issues. Vol. 10/Number 2/ November 2008. 4 OP Schellekens, I de Beer, ME Lindner, M van Vugt, P Schellekens, TF Rinke de Wit. Innovation in Namibia: Preserving Private Health Insurance and HIV/AIDS treatment. Health Affairs. Volume 28, Number 6. November/December 2009. 5 De Beer I, Coutinho H, van Vugt M, Rinke de Wit TF. Anonymous HIV workplace surveys as an advocacy tool for affordable private health insurance in Namibia Journal of the International AIDS Society. 2009, 2:7. 6 Janssens W, de Beer I, Coutinho HM, van Rooy G, van der Gaag J, Rinke de Wit TF. A cautious note on household survey HIV prevalence estimates in resource-poor settings. BMJ 2010. 341: c6323. www.bmj.com/content/341/bmj. 7 De Beer I, Gelderblom H, McNally A, Van Rooy G, Schellekens O, Rinke de Wit T. HIV matters for Univer- sity Students. Journal of the International AIDS Society 2012, 15:9. 8 De Beer I, Coutinho HM, Guariguata L, Fortsch HHT, Hough R, Rinke de Wit TF. Health care options for commercial farm workers in Namibia. Rural and Remote Health 11 (online), 2011: 1384. http://www.rrh.org.au. 9 Leonor Guariguata, Ingrid de Beer, Rina Hough, Els Bindels, Delia Weimers-Maasdorp, Frank G Feeley and Tobias F Rinke de Wit; Diabetes, HIV and other health determinants associated with absenteeism among formal sector workers in Namibia. BMC Public Health 2012, 12:44 doi:10.1186/1471-2458-12- 44. 10 Marleen E. Hendriks, Ferdinand W. N. M. Wit, Marijke T. L. Roos, Lizzy M. Brewster, Tanimola M. Akande, Ingrid H. de Beer, Sayoki G. Mfinanga, Amos M. Kahwa, Peter Gatongi, Gert Van Rooy, Wendy Janssens , Judith Lammers , Berber Kramer , Igna Bonfrer , Esegiel Gaeb, Jacques van der Gaag , Tobias F. Rinke de Wit , Joep M. A. Lange, Constance Schultsz. Hypertension in Sub-Saharan Africa: Cross-Sectional Surveys in Four Rural and Urban Communities; PlosOne. March 2012, Volume 7, Issue 3, e3263f. 11 I de Beer, K Chani, P Mulongeni, TF Rinke De Wit, E Sweeney-Bindels, FG Feeley. Assessing the costs of mobile voluntary counseling and testing at the work place versus facility based voluntary counseling and testing in Namibia; Rural and Remote Health 15 (online), 2015: 3357. http://www.rrh.org.au. 12 Aulagnier M, Janssens W, de Beer IH, van Rooy G, Gaeb E, Hesp C, van der Gaag J and Rinke de Wit, T.F. Incidence of HIV in Windhoek, Namibia: demographic and socio-economic associations. PLOS One 6, e25860, 2011. 13 Leonor Guariguata, Ingrid de Beer, Rina Hough, Pancho Mulongeni, Frank G. Feeley, Tobias F. Rinke de Wit. Prevalence and knowledge assessment of HIV and non-communicable diseases among formal sector employees in Namibia. PlosOne, 13 July 2015 http://dx.doi.org/10.1371/journal.pone.0131737.

165 14 Aneni, EC, De Beer IH, Hanson L, Brennan A, Rijnen B, Feeley FG Mobile Primary Health Care Services and Health Outcomes of Children in Rural Namibia. Accepted for publication in the Rural and Remote Health Journal 2013. Rural and Remote Health (online), 2013: 2380. http://www.rrh.org.au.

Significant research reports 1 F Feeley, C Beukes, I de Beer. The Impact on Employer Operating Costs of Low Cost Health Insurance Including an HIV/Treatment Benefit; Results of a Study of Five Employers in Namibia Center for Global Health and Development, Boston University School of Public Health, Discussion paper 12, March 2010. 2 E Gustaffson Wright; I de Beer, E Gaeb, G van Rooy, J van der Gaag, T Rinke de Wit; The Okambilimbili Health Insurance Project in Namibia: Lessons Learnt. PharmAccess International, Amsterdam Institute of Global Health and Development, The Netherlands, October 2011. http://www.aiid.org/uploads/File/ publications/13_Evaluating%20the%20Okambilimbili%20Project%20-%20Final%20Report.pdf?PHP- SESSID=843e721de6b3e0880444023a29242f1b. 3 O Hanlon B; Feeley FF; De Beer I; Sulzbach S; Vincent H. Namibia Private Sector Assessment. Bethesda MD: Strengthening Health Outcomes through the Private Sector, Abt Associates, September 2010. http://www.shopsproject.org/sites/default/files/resources/Final%20Namibia%20PSA%204-19-11.pdf. 4 Keebler, D; De Beer I, Mulongeni P; Phatela N, Odiit M. Know Your Epidemic- study of the City of Wind- hoek, June 2012: UNAIDS. 5 Keebler, D; De Beer I, Mulongeni P; Phatela N, Odiit M. Know Your Response study of the City of Wind- hoek, June 2012: UNAIDS. 6 Keebler, D; De Beer I, Mulongeni P; Phatela N, Odiit M. Synthesis of the Know Your Epidemic-Know Your Response study of the City of Windhoek, June 2012: UNAIDS. 7 Sabin, L; De Beer I, Da Silva M, Feeley FF. Evaluation of Mister Sister mobile primary health care services in Namibia, February 2014. 8 Van Renterghem, H., Colvin, M., De Beer, I., Gunthorp, J., Odiit, M., Thomas, L., Getcha, M. (2012). City epidemics matter. AIDS 2012. 9 Van Renterghem, H., Colvin, M., de Beer, I., Gunthorp, J., Odiit, M., Thomas, L., Getcha, M. (2012). The urban HIV epidemic in eastern and southern Africa: need for better KYE/KYR to inform adequate city repsonses. Journal of the International AIDS Society, 15. 10 Feeley, F., de Beer, I., de Wit, T., & Van der Gaag, J. (2006). The health insurance industry in Namibia: Baseline Report. Center for International Health.

166 PHD PORTFOLIO

Name: Ingrid H de Beer PhD Period: UvA registration: 2016-2017; Research: 2005-2016 Name of PhD supervisors: Prof. Dr. T.F. Rinke de Wit and Prof. Dr. J. van der Gaag

Workload General courses Year Hours/ECTS University of Stellenboch Project management 2005 80/2.8

Specific courses Boston University School of Public Health - Fall Programme: Epidemiology 2012 80/2.8 Biostatistics 2012 80/2.8 Monitoring & Evaluation 2012 80/2.8 University of Kwazulu/Natal (HEARD) Know your Epidemic / Know your Response methodology 2012 40/1.4

Seminars, workshops and masterclasses University of Namibia (UNAM): Multi-Disciplinary Research Center: Seminar presenter (total 3) 2006-2009 15/0.5 Namibia University of Science and Technology Seminar attendance (Average 4 per year) 2014-2016 36/1.2 UNAM School of Medicine Seminar attendance (Average 2 per year) 2012-2016 30/1.1 Boston University School of Public Health Seminar presented (total 3) 2012 12/0.4 Seminars attended (total 6) 2012 18/0.6

167 Presentations at scientific conferences Innovative Health Insurance and Risk Equalization in Namibia [Poster 2006 7/0.25 presentation]: XVI International AIDS Conference, Toronto The Health is Vital Risk Equalization Fund and leveraging private sector 2006 14/0.5 resources [Oral]: NAMAF Annual Conference, Windhoek, Namibia. HIV Disease Management solutions through the Risk Equalization Fund 2007 14/0.5 [Oral]: Annual Conference of the HIV Clinicians Society of Namibia, Windhoek, Namibia Managed care, standard treatment guidelines and the need for an expert 2007 14/0.5 case review panel in Namibia [Oral]: NAMAF Annual Conference, Wind- hoek, Namibia. The Health in Vital Risk Equalization Fund for HIV in Namibia [Oral]. XVII 2008 14/0.5 International AIDS Conference, Mexico City, Mexico Integration of the HIVREF into medical aid schemes [Oral]: NAMAF Annual 2009 14/0.5 Conference, Swakopmund, Namibia The effects of mobile wellness screening of employees in the formal sector 2010 14/0.5 in Namibia. [Oral]: Annual Conference of the Medical Association of Namibia, Windhoek, Namibia

Other Member, World Health Organization (WHO) Technical Working Group to 2012-2013 28/1.0 review WHO Consolidated Guidelines for the use of ART for HIV Treatment & Prevention. Member, Universal Health Coverage Advisory Committee, Namibia 2013-2016 56/2.0 Member, Technical Working Group on Voluntary Male Circumcision, 2012-2015 28/1.0 Ministry of Health and Social Services, Directorate Special Programs, Namibia Member, Technical Working Group on HIV Counselling and Testing, 2011-2016 56/2.0 Ministry of Health and Social Services, Directorate Special Programs, Namibia

168 Mentoring Madelief Vosser – Undergraduate Social Work practicum, University of 2010 14/0.5 Amsterdam Steffie Hemelaar – Undergraduate Medicine practicum, University of 2011 28/1 Amsterdam Dr. Ehimen Aneni – MPH candidate Boston University School of Public 2012 14/0.5 Health Laura Hanson - MPH candidate Boston University School of Public Health 2012 14/0.5 Gloria Esiomeme - MPH candidate Boston University School of Public 2013 14/0.5 Health Devon Blaze Cain - MPH candidate Boston University School of Public 2014 14/0.5 Health Mikko Vaisanen – Nursing practicum, KAMK University of Applied 2015 28/1 Sciences Jere Ronkainen – Nursing Practicum, KAMK University of Applied 2015 28/1 Sciences

Parameters of esteem Visiting Fellow, Boston University School of Public Health (Social Entrepre- Aug-Dec neur in Residence) 2013 Board Member, Namibia Red Cross Society 2005-2006 Board Member, Orange Babies Namibia (HIV & PMTCT care & support) 2012-2016 Board Member, Positive Vibes Trust (HIV prevention and LGBTI rights) 2008-2016

169 170 ACKNOWLEDGEMENTS

Most sincere gratitude is expressed to the late Professor Joep Lange, the visionary founder and leader of PharmAccess without whom the work of PharmAccess in Namibia would not have been possible. The work contained in this booklet is a special tribute to him and his legacy through PharmAccess in Namibia.

I am extremely thankful to all the amazing hard working people who have joined me on this 12 year journey of working for - and researching the work of – PharmAccess Namibia. Too many people to all mention by name, the service providers, clients, partners and government officials, who made this journey of over a decade possible. Many have contributed directly and indirectly to the studies contained in this thesis and my PhD experience. A hearty thank you to all the study participants, whether individuals, groups or organizations, who made these studies possible.

I am immensely grateful to Prof. Tobias Rinke de Wit, who 13 years ago inspired me to leave my corporate career and join PharmAccess, who has motivated me to complete this PhD, undertook to be my promoter and who has been a good friend through 12 years of ups and downs. Professors Jacques van der Gaag and Rich Feeley, who have been inspirational on both a professional and personal level. A special thank you to Tobias, Rich and Jacques for supervising and co-authoring a number of the papers published over this period.

Many thanks to Professors Frank Cobelens, Anita Hardon, Menno Pradhan, Rich Feeley and Sylvester Moyo for their willingness to serve as members of my doctoral review committee. A sincere thank you to all my co-authors on the publications included in this thesis; RL Hamers, H Kaura, M van Vugt, L Caparos, OP Schellekens, ME Lindner, P Schellekens, H Coutinho, W Janssens, G van Rooy, L Guariguata, R Hough, E Sweeney-Bindels, D Weimers-Maasdorp, K Chani, P Mulongeni, M Aulagnier, E Gaeb, C Hesp, EC Aneni, L Hanson, A Brennan and B Rijnen.

I am particularly appreciative of the leadership, management and colleagues at PharmAccess in Amsterdam and other country offices, for their support of the daily operations in Namibia which made these studies possible.

A heartfelt thank you, last but not least, to my family and friends for their encouragement and support throughout the process, my mom Helene, late father Peter and siblings Doris, Artur and Gerald. A special thank you also to my partner Wolfgang Henckert for his continuous support and motivation to complete this thesis.

171 ABOUT THE AUTHOR

Ingrid de Beer was born Weissnar on 16.12.1970 in Bruck an der Mur, Austria. Her parents immigrated to South Africa in 1977, where she spent her childhood. After completing her studies in Political Science and Development Studies at the University of Johannesburg she moved to Namibia.

Ingrid worked in the private sector in Namibia for 12 years, in the field of human resource management, organizational development, corporate governance, internal audit and industrial relations. In 2003, whilst serving as the Director: Human Capital for the Ohlthaver & List Group of Companies in Namibia, Ingrid met Dr. Tobias Rinke de Wit of PharmAccess. At that time Namibia was experiencing the brunt of the HIV epidemic, and the effect was being felt in the group of companies Ingrid was working for. Trying to find a solution to the barriers of accessing HIV treatment motivated Ingrid to leave the corporate world and join PharmAccess in the fight against HIV/AIDS.

Over the 12 years working as Managing Director of PharmAccess Foundation Namibia, Ingrid developed an extreme passion for improving access to quality healthcare for Namibian, recognizing the important role the private sector has to play to compliment public sector initiatives.

In late 2016, the Namibian operations of PharmAccess were successfully merged with the Namibian Health- works Business Coalition and PharmAccess ceased operations in Namibia. In early 2017, Ingrid took up the position of Technical Director at PharmAccess in Dar es Salaam, Tanzania.

172 Engaging the Private Sector in Public Health Challenges in Namibia | Ingrid de Beer Copyright: University of Amsterdam, 2017 Ingrid de Beer | [email protected] Correspondence to: