HIV/AIDS M&E - Getting Results 43024 These reports describe activities, challenges and lessons learned during the World Bank Global Global AIDS Monitoring and Evaluation Team (GAMET)'s work with countries and other partners. HIV/AIDS Program How GAMET helps countries to improve their HIV response through Epidemic, Response and Policy Syntheses The Global AIDS Monitoring and Evaluation Team (GAMET) analysis can help governments make more effective A prerequisite for an effective HIV response is to HIV/AIDS program and policy decisions. "know your epidemic", and ensure that the response addresses the specifics of each country's epidemic. A synthesis may be rapid, or more extensive ("full"). Recognizing this, an increasing number of countries Both rapid and full syntheses can address the same are requesting Epidemic, Response and Policy questions and focus either on one country or on a Syntheses ­ one of the products/services offered by geographically and culturally coherent group of GAMET. These Syntheses analyze all existing countries. A rapid synthesis looks at all main existing epidemiological and behavioral data to gain new data sources, reviews the findings with a small core insights into the epidemic in a country or region, and group of key stakeholders, and reaches well-founded looking also at data on the response, draw out policy conclusions within a few weeks. A "full" synthesis may and program implications. This provides a strong take up to a year, includes all available data as well as basis for tailoring national responses carefully to the new modeling and analysis and perhaps additional data country's epidemic -- the starting point for effective collection, and builds a broad, inclusive consensus HIV prevention, treatment, care and mitigation. through extensive review and discussion before publishing its definitive findings. The ways in which rapid This note explains what Epidemic, Response and and full syntheses vary are shown in Table 1, which Policy Syntheses are, their goals and approach and describes the ends of a continuum. how they differ from other types of HIV analyses. It suggests criteria for deciding when a synthesis is Table 1: Rapid / Full Synthesis -Typical characteristics worth doing ­ when it is likely to yield insights that will enable decision-makers to improve the country's Rapid synthesis Full synthesis HIV response. It describes the data needed, questions that syntheses investigate, the Time to 1 ­ 2 months 4 to 8 months methodological steps, and a checklist for a good complete synthesis report. Illustrative examples are cited of Popula- General population, General population syntheses recently completed by the Global AIDS tions of or specific sub- and all relevant sub- Monitoring and Evaluation Team (GAMET). focus populations populations Typical To provide an To analyze and What is an Epidemic, Response and Policy purpose overview of the HIV understand the Synthesis? epidemic and implications of newly response in a available data on the An Epidemic, Response and Policy Synthesis begins country/ region, for epidemic; to provide with an analysis of HIV prevalence, incidence, trends, example, as part of information to geographic and population distribution, epidemic phase the preparation of a fundamentally re- and potential trajectory, and the major risk factors and World Bank project, think or change the transmission dynamics in a country or region. It then or as part of a HIV response in a assesses the relevance, focus, coverage, quality and midterm review of a country or region, cost of major HIV responses in relation to the national HIV strategic usually before the epidemiological analysis and policy environment, to plan, or to answer development of new identify opportunities to improve intervention relevance, specific questions National HIV focus, coverage, quality and cost-effectiveness. The about the epidemic Strategic Plans. and response. findings and implications of this empirical, objective Table continues on next page much; when the HIV epidemic is changing (i.e. rapid Rapid synthesis Full synthesis changes in prevalence; or when sub-national HIV prevalence studies reveal trends or prevalence Data All major primary All primary data levels or patterns of infection that are very different sources data sources (i.e. sources, plus new to national trends, patterns and prevalence. consulted sentinel data, analysis of existing behavioral data, data (e.g. new review published of routine monitoring Regional syntheses are most applicable for a group research, and data, meta-analysis of of countries that constitute a geographically and existing reviews of existing data, culturally coherent entity, such as Central Asia, South HIV responses) mathematical Asia, West Africa, Central America or the Middle East - modeling of what we can call a "logical geographic area of affinity". transmission modes, Many regional groupings used by organizations are not etc.) meaningful for HIV. For example, the World Bank East Review Discussion with a Build consensus Asia and Pacific region includes China and Papua New and small core group amongst all partners, Guinea, which do not make for meaningful comparisons discussion of government and extensive peer review or aggregation. Similarly, the Latin America and other key partners before publication Caribbean region (LAC) combines two major and (several minor) sub-regions. There is nothing one can Nature of Conclusions are Is a definitive study ­ say that is equally meaningful to both poles of the EAP findings well-founded (i.e. i.e. a reference for or LAC regions. In contrast, South Asia and Central Asia fully substantiated other studies in the are geographically, linguistically, historically and by the data), but future not definitive (i.e. economically, logical geographic areas of affinity. not the full and In short, unless a region is coherent and meaningful final reference on from an HIV perspective, individual country syntheses the drivers in the are preferable to a regional one. But special country or region) circumstances may apply; for example, the six countries GAMET - Great Lakes area - Ethiopia of the Great Lakes Initiative on AIDS commissioned a examples (6 countries) - China joint synthesis that focused specifically on sub- - Vietnam - South Asia (4 main population groups in common or that move across their borders, whose mobility or experience of violence makes - Kenya countries) - West Africa (16 them vulnerable to HIV, or whose behaviors put them at countries) particular risk. Deciding on the appropriate scope of a Great Lakes Initiative on AIDS Synthesis synthesis Six countries in Africa ­ Burundi, Democratic Republic of Congo, Kenya, Rwanda, Tanzania, and Uganda ­ The decision on how extensive a synthesis should be formed "The Great Lakes Initiative on AIDS" (GLIA) to rests primarily on its purpose and the questions it seeks complement their national AIDS program efforts to to answer, and on the availability of data, time and reduce HIV infections and mitigate the socio-economic budget. A variant of the 80/20 rule might well apply: a impact of HIV in the Great Lakes Region. In order to rapid synthesis may provide 80% of the benefit for 20% identify sub-populations and interventions for GLIA to of the cost of a full synthesis, but there are some focus on, so as to add value to the efforts of each circumstances when a full synthesis is valuable. individual country, and use the comparative Rapid syntheses are most applicable when there advantages of a regional program, the GLIA council of are not enough data for a full synthesis but some ministers asked for an analytic study that would provide good quality data are available and existing data an evidence base for deciding on GLIA's focus. So a have not been well triangulated; when there is not rapid synthesis was undertaken, that proceeded as time or funding for a full synthesis; when a full follows: synthesis (to determine the drivers of the epidemic) (1) First, data were extracted from all existing has been undertaken within the last few years, and documents and other data sources, on the size and recent data do not suggest significant changes in the HIV prevalence and risk factors of all populations epidemic. identified as being at risk or vulnerable to HIV infection Full syntheses are more applicable when or transmission. Eight sub-populations were identified epidemics are not well characterized; when a lot of for detailed analysis ­ fishermen and fisherwomen, new data is available that has not been analyzed 2 military, prisoners, long-distance truck drivers, female sex workers, refugees and displaced people, and women affected by sexual violence. The lives of these populations are marked by mobility, conflict and/or violence, vulnerability to HIV and risk factors that the epidemiological evidence indicates result in significant contributions to HIV transmission. (2) The number of people living with HIV (PLHIV) was estimated for each sub-population, by multiplying the number of people in the group by the median HIV prevalence estimate based on the best data available. (No estimation of size or prevalence was possible for female sex workers because of difficulties in defining and identifying them as a distinct group.) (3) The national HIV strategic plans of the GLIA countries were all carefully reviewed to assess the Fishermen at Kaloka, Lake Victoria, Kenya extent of gaps in coverage of interventions serving these populations. Comparison with other types of analysis (4) Then, comparing the data on HIV prevalence, risk factors and size of the identified populations and Other types of analysis may serve as inputs into an contributions to HIV transmission, with the national HIV Epidemic, Response and Policy Synthesis: strategic plans, the study defined which vulnerable populations the GLIA should focus on, and the type of HIV modeling studies that apply epidemic models interventions that existing evidence suggests as most that draw on data from other countries to extrapolate likely to prevent HIV infections and mitigate impact for or "fill in gaps" in incomplete country data may be an each population. important source for a synthesis, but a synthesis will also include numerous other data sources, including The small study team was led by an analytical surveillance, research and evaluation evidence. epidemiologist, working with a research assistant, and an M&E specialist, who conceptualized the study, Program reviews and evaluations use a range of coordinated the work, provided technical oversight, and methods to answer a specific research question, helped draw out the policy implications of the findings, narrower in scope than a synthesis. Provided they synthesizing information from 285 different studies. The are based on rigorous, objective evidence, program work was funded jointly by the Global AIDS Monitoring reviews and evaluations may also be an input into a and Evaluation Team (GAMET) and the GLIA synthesis. However, subjective, impressionistic Secretariat. Reviews, data, and other support for the evaluations are unlikely to be useful input for a work was provided by a number of partners: the synthesis. National AIDS Commission staff who represent the Reviews of proven HIV prevention approaches ­ countries in GLIA, the UNAIDS M&E advisors in each "what works, where" ­ for example, reviews of country, World Bank AIDS Team Leaders and effective HIV prevention approaches for sex ACTAfrica, the World Health Organisation, the German workers, can help understand the factor and Technical AID agency GTZ, the International circumstances associated with program Organization for Migration (IOM), United Nations High effectiveness. Commissioner for Refugees (UNHCR) and the United National UNECA Economic Commission for Africa "Think pieces" explore potentially new approaches, especially in contexts where existing efforts appear (UNECA). to be faltering, such as Southern Africa. The work was completed quickly ­ the idea was first discussed in April 2007, terms of reference were written What data sources are needed for a synthesis? in July, and work began in September. An initial draft report was reviewed in November, then presented and discussed in a technical workshop in December with Six core data sources are needed for all Epidemic, approximately 70 delegates from GLIA countries and Response and Policy Syntheses: technical agencies. The findings and recommendations National population-based household surveys, which feed into GLIA's strategic planning in January 2008. provide the best estimate of overall epidemic size. 3 Antenatal surveillance data, which provide the best the synthesis must add discernable value, either insight into trends and the most comprehensive site- through new information or new integration of existing specific data. knowledge. Bio-behavioral surveillance of most-at-risk- populations (MARPS), i.e. data on behaviors as well 2.There are adequate national data to synthesize. as on individual HIV status of the study sample Without sufficient data of reasonable quality, a good population. synthesis cannot be done ­ instead, further original data collection is the priority. Liberia and the Maldives Research studies examining HIV risk factors and are two examples of countries which require improved intervention impact. primary data collection, because existing data are Credible, preferably peer reviewed models of the inadequate to support a synthesis. sources of new infections in a country or region. 3.A synthesis is the right product to offer. In Program monitoring and evaluation data. Southern Africa's unparalleled epidemics, where HIV is exceptionally high in most geographic and The major data sources for a full synthesis are population groups and where existing approaches are summarized in Appendix 1. A rapid synthesis would use not working, a synthesis may not be useful. For some (but not all) of the primary data listed there. example, Swaziland and Botswana have very small populations and highly homogeneous epidemics, in When is a synthesis valuable? which HIV prevention is failing. The real challenge is to understand the socio-cultural context of entrenched There are five major preconditions for a synthesis to be risk behaviors and to identify effective HIV prevention worth doing: approaches. A think piece that seeks to build a new understanding of the unique HIV epidemic in each 1.There is something important and distinctive to country and new ways to respond may be more say about a national or regional epidemic. This would relevant than a synthesis. be the case if the epidemic is largely uncharacterized (as in Papua, Indonesia); poorly characterized (as in 4.There is broad national support. A synthesis must West Africa, when it had been assumed that the not be imposed by external parties. Narrow support epidemic was broadly similar to Southern Africa); the from a single national actor -- even the Ministry of subject of debate (as in India); if there has been limited Health or the National AIDS Council -- will not suffice if integrative analysis using multiple data sources (as in other key national partners resist the idea or do not Central America and the Caribbean); or the think that a synthesis is worth doing. epidemiological patterns and response priorities are poorly matched and need rethinking (such as Ghana, 5.The synthesis is done by world class national or after new data became available). It is hard to have international individuals or institutions that are something important and distinctive to say about an experienced in epidemic syntheses. A synthesis obvious or well delineated HIV epidemic (such as should never be attempted unless there is expertise to Estonia's, where HIV among ethnically Russian IDU is do it that is unquestionably world class. Adequacy will the obvious driver). not suffice ­ world class expertise is necessary. This To put this another way, the business case for a poses particular challenges where HIV is a peripheral synthesis must be unmistakable. There must be an issue and HIV expertise is consequently limited ­ even initial anticipation of the major unanswered policy including major countries such as China or Egypt. question/s the synthesis will address, what the storyline might be and the major insights it is likely to Synthesis methodology/approach afford. For example, in South Asia and West Africa, syntheses aimed to characterize the epidemics across large, culturally coherent regions and in particular to There is no simple road map or recipe for doing a highlight the role, and policy and programming synthesis, any more than there are clear guidelines for challenges of sex work (and to a lesser extent MSM). writing a research paper. There are, however, some The key issue in Indonesia was to characterize the principles and key questions to ask, which define a heterogeneity of the epidemic and differentiate the logical approach in undertaking a synthesis: epidemic in Papua from the rest of the country. 1. First frame a possible storyline ­ an hypothesis In short, the work must anticipate in advance a clear, about the epidemic. Before embarking on a synthesis, important storyline, with key policy messages. "So it is important to frame an initial storyline, which is then what?" syntheses where the probable storyline is constantly tested against the evidence, refined and tediously obvious and unoriginal are not worth doing ­ abandoned if the evidence refutes it. 4 For example, the Vietnam and China syntheses began data. This information then helps to answer questions with the storyline that HIV was highly concentrated, with about the character of the epidemic. injecting drug use as the major driver of HIV transmission, igniting sexual transmission and amplifying 3. Investigate the epidemic's trends and phase. epidemic potential - by implication, effective injecting What are the trends (i.e. changes in HIV prevalence in drug use programs could radically curb epidemic the general population and in sub-populations over time) potential. This initially controversial storyline was and phase (i.e. where, on the likely epidemic trend line, decisively supported by the evidence synthesized. is the country)? What major behavioral factors appear to Sadly, in the absence of effective injecting drug use be associated with a rising, stable or falling epidemic? programs, the predicted ignition of HIV in sex work has occurred and the epidemic character is slowly shifting. Epidemic trends are critically important. A soaring epidemic implies that effective responses have not been identified and that significant transmission is occuring in the general population. A falling epidemic implies that HIV is in decline and is coming under control. This implies a transition in the pattern of infections from transmission among the general population to transmission from already infected people to their stable partners. This distinction has significant implications for interventions and priorities ­ whether to focus more on uninfected people (what used to be called primary prevention) or focus more on people living with HIV (secondary prevention). Sharing drug injecting equipment is a major driver of HIV 4. What is the character and potential of the epidemic? Based on the evidence on the epidemic's The Indonesia synthesis began with the hypothesis that size, trends and phase, the character and potential of HIV outside the Papua provinces was fueled by injecting the epidemic may be delineated. Epidemics may be drug use and linked sex work transmission, and that HIV characterized in many ways. "Hyperendemic epidemics" in Papua was circulating as a result of casual refer to the unparalleled generalized epidemics of transmission in the general population. However, the Southern Africa, in which prevalence is much higher recent population-based survey suggests that sex work than in any other region or sub-region. "Generalized remains a major driver even in Papua, evidence that epidemics" refer to epidemics elsewhere that are compelled a revision of the storyline. sustained by sexual transmission in the general population. "Concentrated and low-prevalence The West Africa synthesis was prompted by the epidemics" are sustained primarily by transmission conviction that sex work plays a larger role than among specific most-at-risk populations. acknowledged in West African epidemics and that programs for sex workers receive insufficient emphasis. It is important to emphasize two points in characterizing epidemics. First, epidemic definitions should be In Kenya and Ethiopia, syntheses were spurred by the based primarily on transmission dynamics, not availabililty of new data that had not been assimilated arbitrary thresholds. We used to say that HIV adequately. The initial storyline or hypothesis was that epidemics were generalized if HIV was over 1% in the there was greater diversity than recognized in each general population, and concentrated if HIV was below national epidemic, with male circumcision playing a 1% in the general population, and over 5% in sub- major role in the diversity. It must be emphasized that an populations. A transmission-based definition asserts that initial storyline is not a preconception, prejudice or HIV is concentrated if HIV transmission is mainly among opinion, but a sophisticated initial reading of the most-at-risk populations and if protecting vulnerable evidence by practitioners familiar with the evidence and groups would protect HIV spreading into the wider context. community; and generalized if protecting most-at-risk- populations would not protect the wider community 2. Determine the size of the epidemic. Once a broad, (because there is already significant transmission refutable storyline (hypothesis about the epidemic) is occurring outside most-at-risk populations). This sketched based on an overarching understanding of the definition is consistent with and encourages closer epidemic and context, the next step is to determine the analysis of transmission patterns, and has clear estimated number of people living with HIV in the programmatic implications. country. This will be based on national estimates, antenatal, population-based and most-at-risk-population 5 The Kenya rapid synthesis In developing the "Total War on AIDS" Project (TOWA) funded by the World Bank to support the Kenya National HIV and AIDS Strategic Plan 2005/6-2009/10 (KNASP), it became clear that better information on the epidemic could help make programs more strategic, prioritized, evidence-based and results-orientated. So the National AIDS Control Council (NACC) of Kenya agreed that GAMET should: (a) examine all existing studies of HIV transmission and drivers of the epidemic in Kenya, and identify opportunities to use this analysis to improve intervention priorities, design and implementation, and (b) review the first Call for Proposals for TOWA grants and recommend ways to strengthen the results-orientation of future grant rounds. It took the team of two ­ an epidemiologist and an HIV Prevention specialist ­ about two weeks to synthesize and summarize the existing information on the epidemic in Kenya. They were able to complete the work this quickly because the NACC and National AIDS Program of the Ministry of Health had compiled an excellent data base of all existing studies and data which was immediately accessible. Supported over several years by the Canadian International Development Agency (CIDA) and the US Centers for Disease Control, Kenya has been a leader in Africa in surveillance, program evaluation, and research on most-at-risk-groups. But despite the considerable collection of data on the trends, current patterns and drivers of the epidemic in Kenya, the NACC did not have a clear updated and integrated picture of the HIV epidemic. The GAMET team began with an epidemic "story line" or hypothesis that was tested rigorously against the data, refined, discussed with experts and specialists, and refined more. Overall, HIV prevalence has fallen decisively in Kenya ­ among pregnant women attending antenatal clinics HIV fell from a peak of 13.4% in 2000 to 4.6% in 2006. There is clear evidence of behavior change -- more condom use, fewer sexual partners and later age of initiating sex. However, HIV in Kenya remains tenacious and entrenched, with still an estimated 55,000 new infections each year. This places an insupportable burden upon AIDS treatment programmes, and makes it essential to revitalize and strengthen HIV prevention programs to intensify declines in new infections, with particular focus on areas of highest HIV. Much higher prevalence in Nyanza and parts of Nairobi provinces is driven largely by low levels of male circumcision, as well as high transmission among lakeshore fishing communities. Epidemiological modeling by UNAIDS (Gouws et al 2006) indicates that 58% of new infections are from casual and low- risk heterosexual sex and another 28% are transmission within discordant couples; sex workers and their clients account for 12%, and sex between men and injecting drug use each account for close to 5% of new infections. The team outlined an agenda for intensified prevention driven by the synthesis analysis of Kenya's surveillance, research and modeled sources of new infections that prioritizes interventions to tackle transmission clusters in the general population, between couples and among most-at- risk-populations and their sexual partners. A working draft report on these conclusions on the epidemic and their implications for HIV policies and programs was shared with core experts, and fed into strategic discussions, planning and programming by the National AIDS program in Kenya. A polished draft was circulated more widely for review and discussion, and then a final draft provided input for the UNAIDS Kenya Country synthesis. This work on the drivers of the disease is continuing, and will be used in reviewing and revising the national HIV/AIDS strategy, as well as in the second Call for Proposals under the TOWA in mid- 2008. "Kenya today" by Joseph Bertiers/AfricanColours.net 6 There is also the important and under-recognized fishing communities. Numerous ethnographic studies in category of mixed epidemics, in which new infections are Kenya and elsewhere in Africa have found steeply driven by most-at-risk populations and the general elevated HIV prevalence in fishing communities, population. Many epidemics in Africa, the Caribbean and associated with cash income, separation from family and Papua New Guinea/Papua today are probably mixed. community, sparse living conditions, hazardous Mixed epidemics require a careful balance between lifestyles, juxtaposed with a dynamic trading milieu and targeted and general population interventions. Mixed considerable mobility and unique sexual micro-cultures epidemics take different forms. Some mixed epidemics (which, in Suba, include jaboya where female fish buyers reflect a (very) roughly equal balance among different exchange sex with boat owners for primary purchasing transmission sources and require roughly equal rights and abila where women servicing boat houses emphasis on targeted and general population provide sex and housekeeping services). interventions. This may be the case in parts of East Africa. For example, Kenya's fęted decline in HIV 6. What are the major transmission dynamics; i.e. infection may be due to two things: effective targeted the modes and sources of transmission? In addition interventions for vulnerable groups in large cities and to analyzing geographic and community patterns in HIV along major highways, and large-scale behavior change epidemics, a vital aim of a synthesis is to understand the and partner reduction in the wider population. main transmission modes and sources of new infections and to ensure that programs are based on this analysis Countries may exhibit different epidemic characteristics so they can effectively prevent new infections. This can in different areas of the country. For example, Indonesia be posed as a simple question: How did the last 1,000 appears to have a concentrated epidemic outside Papua new infections in a given country occur? There are four and a more generalized epidemic in the Papua main approaches for trying to better understand provinces. Similarly, Kenya may have relatively transmission patterns and sources of infection: concentrated transmission in the Eastern and Western Provinces and more generalized patterns in Nyanza. a) Simple comparisons of HIV prevalence: Simplest Tanzania may have more concentrated epidemic is to compare HIV prevalence across different patterns in Zanzibar and more generalized patterns in communities. A steep gradient (large differences in inland Kagera. Mozambique's epidemic may be more HIV prevalence) implies that high prevalence concentrated in most-at-risk populations in its Northern communities are contributing significantly to Region Provinces of Cabo Delgard, Nampula and transmission. For example, HIV prevalence among Niassa and driven by more generalized transmission in sex workers in Ghana was 78% in Accra, the capital the central provinces of Sofala, Manica and Zambezia. and 81% in Kumasi, the second city, compared to 2% prevalence in the general population. Intuitively, Second, epidemic patterns may evolve over time, these data suggest that sex workers are contributing sometimes rapidly, and the gender ratio of HIV significantly to HIV transmission ­ a hypothesis prevalence is a key indicator of the extent to which supported by subsequent studies. an epidemic is concentrated or generalized. A higher male HIV prevalence ratio implies a concentrated b) A more systematic approach is to ask HIV-infected epidemic, driven by men having sex with each other or people how they became infected and to use their sex workers or sharing injecting equipment. An even sex answers to infer transmission modes. ratio or higher female ratio implies a generalized epidemic, in which significant numbers of women are c) Risk factor studies are a more sophisticated infected through sexual transmission. approach, and examine the concomitants of HIV infection in a research sample. 5. Consider major contrasts and patterns in the epidemic. There can be important variances in HIV However, there is a crucial limitation in these types of across communities, sub-populations, and parts of a analyses ­ they tend to reflect HIV prevalence (which country. How heterogeneous is the HIV epidemic, with includes all infections, recent and historical), and not HIV respect to geography and sub-communities? The incidence (new infections in the previous year). HIV answer helps to understand diverse transmission prevention programs must effectively target the largest patterns and then assess how adequately HIV sources of incident HIV infection, that is, new infections. prevention resources are directed towards priority Many countries have HIV prevalence data, but very few geographic areas and communities. It also helps to have good HIV incidence data. Some countries have characterize HIV epidemics and to understand risk limited HIV incidence data from longitudinal studies, factors and transmission patterns. For example, the such as vaccine trials or HIV prevention trials, but these highest prevalance district in Kenya, Suba in Nyanza samples are usually small and not representative of the province, is inhabited by the Luo who do not practice wider population ­ for example, prevention trials among male circumcision, and also has over 120 lakeshore sex workers in Nairobi, Kenya. Test techniques (such as 7 Ethiopia ­ full synthesis the detuned elisa test, the BED assay or the avidity/affinity test) to identify recent infections using The Government of Ethiopia requested a synthesis to cross-sectional survey blood samples are being take account of new data from the first population- developed, but remain experimental. based survey of HIV prevalence and related behaviors. A GAMET team of three epidemiologists d) Mathematical modeling: In the absence of worked a total of about 200 days, well over three full measured HIV incidence data, countries may be months of work each. able to model their sources of incident infection, using various available computer models or A study Steering Committee, chaired by HAPCO (the epidemiological analytic methods. These include National AIDS Council), included all the key partners, estimating the "population attributable fraction" as and helped ensure wide and intensive consultation was done in Accra, Ghana; the Asian Epidemic with the government and other stakeholders. Initially, Model developed by the East-West Center with an inception report outlining the study scope and support from USAID and others; and the recently methodology, and a tentative table of contents and developed UNAIDS incidence model being used in study outline were circulated for comment, and Africa. Studies and models can improve our guided the work. A rough initial draft of the findings understanding of the major sources of new infection and recommendations underwent intensive internal and transmission dynamics. However, they are only review, and then a revised draft was circulated widely some of the inputs that go into a synthesis, and must for review. At a four-day retreat of key stakeholders, be carefully substantiated and triangulated using the draft was read and discussed very carefully, data from other sources. which led to an extensively revised and more policy- and action-oriented third draft with a much clearer 7. Assess the extent to which the major responses, storyline and clearer, more explicit policy and including national strategic priorities, investments program implications. and interventions, match the major drivers of HIV transmission. With the epidemic size, trends, character The study synthesized findings from an exhaustive and heterogeneity delineated, as well as sources of literature review, extensive additional analysis of the recent infection and transmission patterns/modes, there primary data, and several commissioned background is sufficient understanding of the character of an papers. The final draft report was edited to improve epidemic to assess the extent to which major national the presentation and refine the analysis. The year- HIV prevention responses match and tackle the major long process culminates in a dissemination process sources of transmission. For example, in Ghana, where planned for January 2008. a careful study found that 76% of new inections among sexually active adults were attributable to sex work, the The synthesis concluded that Ethiopia's epidemic is World Bank estimated that only 0.8% of the World Bank- much more concentrated than previously thought, funded project (GARFUND) resources were devoted to with much higher prevalence along highways, in programs to make sex work safer. Across West Africa, market towns and development nodes, among sex important and relevant programs for sex workers have workers, and in Gambella district near the border with been discontinued as funding sources changed. Sudan, where much lower rates of male circumcision Conversely, a regional highways project focusing on are an important contributing factor to higher HIV commercial sex along Southern Africa's highways and rates than the rest of Ethiopia. Transmission within borders enumerated few sex workers and subsequently discordant couples also contributes significantly to found that most adult men reported casual, rather than new infections, needing new prevention approaches. commercial sex. In both Papua New Guinea and Papua Indonesia, the majority of HIV resources are spent in the It is too early to gauge the impact of the synthesis capital city, yet evidence suggests that HIV and risk study ­ that will be seen in the extent to which the behavior are higher in the largely unreached rural areas. programmatic implications change the direction and Over time, in mature, declining HIV epidemics, emphasis of the national response. transmission occurs increasingly from people with HIV to their spouses, yet there is limited emphasis on couple- based HIV prevention. In Uganda and elsewhere in East Africa, there is a strong emphasis on HIV prevention among young people, yet the data indicate that the peak age of HIV prevalence and incidence is among older men. Globally, there is immense opportunity to better align HIV prevention priorities and investments with the major sources of HIV transmission ­ this is perhaps the single most important role a synthesis can play. Kedija learns about family planning and HIV/AIDS 8 8. To what extent do the major interventions reflect making their significance and implications clear. For proven approaches and global best practice example, a recent Kenya rapid synthesis arguing that evidence? It is not sufficient to ensure that HIV male circumcision in Nyanza is critically important prevention investments address the major sources of includes an analysis of how many HIV infections could HIV transmission ­ it is also important to ensure they be prevented each year by male circumcision and how use proven, evidence-based approaches. For example, much smaller Nyanza's epidemic would be in 10 years if in China, harm reduction approaches emphasize specified levels of male circumcision were to be methadone maintenance therapy. However, recent achieved. It is important for the conclusions to be World Bank cost-effectiveness studies and syntheses relevant, important and contain new insights. The danger suggest that needle-syringe programs may be far more that syntheses will yield "blinding glimpses of the cost-effective. In Vietnam, and elsewhere in Asia, drug obvious" can be anticipated and avoided at the start, addicts are placed in detention centres, despite clear when the storyline is sketched, ensuring that the evidence that the majority relapse into continued drug synthesis is likely to have something important to say. use upon release. Worse still, HIV infection increases steeply during detention, suggesting that the detention Hallmarks of an excellent synthesis centres are incubators of HIV transmission and may significantly accelerate national epidemics. Identifiying and describing effective approaches is particularly a) At the start, there is a clear initial hypothesis or relevant for countries that acknowledge the character of storyline that is then tested against the data, and their HIV epidemic and want more differerentiated refuted, confirmed or refined. assistance ­ as one Asian program manager said in frustration: "Don't just tell us to do harm reduction, we b) A compelling, clear, short Executive Summary sets out know that, tell us how to do it really well." The key point the key findings and policy and program here is to ensure that syntheses do not simply go over implications, clearly explaining the importance and well-trodden ground, but tackle current challenges ­ potential impact of the recommendations. which may increasingly focus on the question: "Given our improved understanding of our specific situation, c) The synthesis answers the following questions: what is most likely to work?" What is the magnitude of the epidemic? 9. Synthesis reflection ­ in what respects and to What are the trends and phase of the epidemic, its what extent, does the synthesis challenge character and potential? conventional wisdom and thinking about the HIV What are the major contrasts/patterns in the response? How sure is the evidence base for such epidemic? challenges? What are the major caveats? It is important for a synthesis to summarize how it challenges What transmission dynamics and sources give rise conventional wisdom ­ for example, the South Asia to most new infections? synthesis argued that South Asia's epidemics were To what extent do the major responses, including smaller and more concentrated than generally national strategic priorities, investments and perceived, and that HIV responses should be narrowly interventions, match the major drivers of HIV focused on most-at-risk-populations. This was somewhat transmission? controversial at the time ­ it is important to evaluate how sound the evidence base is for any challenge to the Do major interventions reflect proven approaches generally accepted wisdom. It is also important to state and global best practice? the major caveats clearly. In South Asia, for example, In what respects and to what extent does the there were limited HIV prevalence data for North India synthesis challenge established wisdom and until the National Family Health Survey (NFHS) was thinking about the HIV response? How sure is the completed ­ when it was, it subsequently validated the evidence base for such challenges? What are the existing sparse data, but the caveat was essential. major caveats? 10. Conclude the synthesis by setting out the major How can the conclusions be aligned with the policy implications for decision makers. A synthesis realities of the current policy environment? is a policy document (not just an analytical piece), and must have important policy and program implications for d) The conclusions are clearly stated and flow logically decision makers. These implications must be (i) spelled from the data analysis, offering relevant, important, out unmistakably in the report, (ii) clearly summarized in new insights on the epidemic and how the response a separate conclusion, and (iii) form a major part of the might be more effective in preventing HIV infections. Executive Summary. They must be written clearly, simply and in ways that "speak to" decision makers, 9 e) The report clearly describes the methodology used, including review mechanisms and approvals. It cites all data sources in full, documents all methods of data collection, and notes all sample sizes so that readers can appropriately interpret the findings. The reasons more countries want syntheses ... Syntheses can be the foundation for a new or revised national strategic plan, guiding government and development partner investments in HIV in a country. They can be used to review the validity of areas being funded. They can identify "hot spot" areas and key sub- populations where HIV prevention activities need to focus. And they can help assess the existing and needed coverage and scale of the highest priority and most cost-effective prevention interventions. The synthesis study in China, for example, found that fewer than 1% of IDUs were being reached with needle/syringe programs, which are about one third of the cost of methadone maintenance programs per infection averted. Syntheses support a renewed focus on effective HIV prevention, and reflect the strong realization that expanding and sustaining access to treatment depends on effective prevention. They can enable policy makers and planners to rethink the country's HIV response, to choose HIV prevention interventions that are targeted, realistic, appropriate and more likely to be effective. References, further information Photograph copyrights: World Bank. AIDS in South Asia: Understanding and "Men fishing in Kenya" (p3) is reproduced with the kind Responding to a Heterogenous Epidemic. World permission of the California Academy of Sciences. More Bank, Washington DC, 2006. pictures from the Manzanita Image Project are at: www.worldbank.org/AIDS > publications > scroll http://www.calacademy.org/research/library/manzanita/html/ down or click here for this report Injecting drug use photo (p5) accessed on line at: More details on the work in Papua Indonesia are http://satudunia.oneworld.net/ezimagecatalogue/catalogue/vari provided in the Getting Results note, November ations/30881-300x300.jpg 2007 and the full report Risk Behavior and HIV Prevalence in Tanah Papua 2006 ­ Results of the Picture "Kenya today" (p7) accessed on line at kenya.africancolours.net/content/9679 and reproduced by IBBS 2006 in Tanah Papua, Statistics Indonesia kind permission of the artist, Joseph Bertiers and Ministry of Health, is available on line at: http://siteresources.worldbank.org/INTINDONESIA/R "Kedija learns about family planning and HIV/AIDS" near her esources/Publication/PapuaHIV_en.pdf village in Siraro Woreda, Ethiopia. (p8) She has three children, Gouws E et al. 2006. Short term estimates of adult not counting the ones that have passed away. She is the HIV incidence by mode of transmission: Kenya and second wife of her husband, who has recently turned his attention to his new third wife. Photograph by kind permission Thailand as examples. Sex Transm Infect, 82(suppl of Photoshare, © 2005 Netsanet Assaye, photo I.D. 2006-371, III):iii51-55. www.photoshare.org Links to synthesis reports will be added at www.worldbank.org/GAMET as they become T available 10 Appendix 1: Illustrative Data Guide for Syntheses 17. Identify, quantify and rank the major correlates of A: Epidemic Review Data sources HIV status AIDS Case Reporting Ethnographic and Qualitative Studies 1. Obtain all AIDS case reporting at national level 18. Obtain all relevant ethnographic and qualitative studies, particularly those dealing with HIV, drug 2. Critically review the quality and limitations of the use, sex work, sexuality (including homosexuality AIDS case reporting and bi-sexuality) and gender 3. Analyze AIDS case reporting by gender, age, 19. Critically review studies and use the data to situate economic status, transmission mode, geographic the biological and behavioral data described origin and risk factors, carefully noting all major above in an integrated, contextualized format trends Models HIV Prevalence 20. Obtain all major HIV models and projections 4. Obtain all HIV prevalence surveys from blood donor, hospital patient, VCT, PMTCT, ANC and 21. Critically review the models and summarize their high risk group surveys characterization of HIV prevalence, incidence, trends and changes in infection patterns 5. Critically review the quality of the data, with particular reference to sampling and testing issues 22. Identify the major gaps and limitations in the models and estimate their effect upon the 6. Analyze HIV prevalence data by age, gender, accuracy of the models economic status transmission mode, geographic origin and risk factors, carefully noting all major 23. Make recommendations to strengthen further trends. Use statistical tests wherever there are modeling adequate sample sizes STI Data HIV Research Studies 24. Obtain all major STI data 7. Obtain all major HIV research studies, particularly 25. Critically examine the quality and patterns of the incidence and risk factor studies. data 8. Critically review the quality of the studies and 26. Examine the relationship between STI data and identify their methodological strengths and HIV trends weaknesses 27. Do STI data predict HIV infection levels? 9. Analyze HIV incidence data and trends, as well as 28. Summarize STI patterns and trends and their correlates of HIV infection implications Bio-behavioral Surveys Other studies 10. Obtain all major bio-behavioral surveys 29. Rapid or full epidemiological synthesis done in the 11. Critically review the quality of the studies, past 10 years including their internal consistency and 30. Qualitative study reports and anecdotal concordance with other data sources information (can include supervision reports, case 12. Analyze HIV incidence data and behavioral trends, studies, field reports, interviews, narratives, with particular emphasis on correlates of HIV testimonies, etc.) infection 31. UNAIDS country-specific epidemiological data and 13. Identify, quantify and rank the major correlates of estimates provide latest UNAIDS/WHO epidemic HIV status update. As of Jan 2008, latest was published in Dec 2006: www.unaids.org/en/HIV_data/epi2006/. Behavioral Surveys The UNAIDS/WHO Global HIV/AIDS Online 14. Obtain all major behavioral surveys (that do not Database collates the most recent country-specific include bio-markers data on HIV spread and impact, with information on risk behaviors (e.g. casual sex and condom 15. Critically review the quality of the studies, use), as well as health sector response including interviewer recruitment and training, tool information (e.g. data on people receiving translation and back-translation, sampling, the treatment, coverage of HIV counseling and interview process, including rapport building. Also testing) Available on line at: examine their internal consistency and www.unaids.org/en/HIV_data/Epidemiology/epidat concordance with other data sources abases.asp. Epidemiological fact sheets by 16. Analyze behavioral trends by risk group, age country can be downloaded at gender, economic level and other relevant factors www.who.int/globalatlas/default.asp. 11 B: HIV response review data 32. Obtain all major program reports describing 39. Qualitative study reports and anecdotal national AIDS responses, particularly reports from information: These can include supervision the national HIV M&E system and from major reports, case studies, field reports, interviews, development partners. Most countries have at narratives, testimonies, etc. least started to develop national HIV monitoring and evaluation systems. These systems typically C: Financing the HIV response generate periodic information products (reports), including a report to UNAIDS on the 25 UNGASS 40. National AIDS Spending Assessments and Public indicators. The UNAIDS website contains good HIV Expenditure Reports provide information information (www.unaids.org, search for `Global about spending on the HIV response in the AIDS Epidemic Update'), as do websites of country, and help the Bank team and client to development partners (www.theglobalfund.org, determine together, the most strategic use of the www.pepfar.gov, and the World Bank Africa Bank's investment in HIV in the country. Results Management System). Ministries of Finance, the NAC or the in-country 33. Review National HIV/AIDS policy and National M&E person are the most likely sources for these HIV/AIDS strategic plans data. 34. Analyze the quality and coverage of the services, in relation to projected need 35. Evaluate the distribution of the services, in relation to urban-rural divisions and socio-economic status Sources used to compile this list: of participants. Pay particular attention to SW, clients, MSM and prisoners Gorgens-Albino and Victor-Ahuchogu, 36. Summarize the major providers of services, by J., 2007. Planning for, Measuring and size and quality, including the following categories: Achieving HIV Results: A Handbook for public providers, the private sector and NGOs. Task Team Leaders of World Bank Assess the resources and capacity in each sector lending operations with HIV and recommend capacity improvements where components, Version 2.2, The World required Bank: Washington DC. 37. Critically evaluate national responses, relative to GAMET. 2007. How GAMET helps the need and in comparison to other countries countries to improve their HIV response 38. Rapid epidemiological and/or HIV response through Epidemic, Response and Policy analyses or detailed Epidemic, Response and Syntheses (ERPS). The World Bank: Policy Synthesis reviews done in the past 10 years Washington DC. About the authors: David Wilson is Lead Specialist at the World Bank, in the Global HIV/AIDS Program. dwilson@worldbank.org Marelize Görgens-Albino is an M&E expert with the Global AIDS Monitoring and Evaluation Team (GAMET) at the World Bank. mgorgens@worldbank.org "HIV/AIDS - Getting Results" series editor: Joy de Beyer, Global HIV/AIDS Program, jdebeyer@worldbank.org For more topics in the "Getting Results" series, please go to www.worldbank.org/aids > Getting Results December 2007 12