Nicholas C. Grassly1; Geoffrey P. Garnett
Department of Infectious Disease Epidemiology, Imperial College London, St Mary's Hospital, Norfolk Place, London W2 1PG, England
The emerging HIV epidemics in countries of Asia and Eastern Europe will contribute significantly to the future of the HIV pandemic. Forecasts of the scale of these epidemics are subject to massive uncertainty, however, mainly because of the sensitivity of predictions to small alterations in parameters that are difficult to estimate. In most of these countries, HIV is currently concentrated among vulnerable populations such as injecting drug users, sex workers and their clients, or men who have sex with men. This distribution suggests an alternative to disease forecasting based on the techniques of risk assessment routinely used by environmental epidemiologists. Exposure mapping, dose–response curves and the concept of acceptable risk are some of the tools that may be useful for HIV risk management. This approach is illustrated by a description of exposure in Indonesia and an assessment of currently accepted risk of death for different causes including HIV in the Russian Federation. Although inappropriate for forecasts of heterosexual HIV transmission, mathematical models are shown to be useful for making qualitative predictions about the relative importance of different behaviours for the spread of HIV over time and for interpreting observed trends in HIV prevalence from sentinel surveillance sites.
Keywords: HIV infections/epidemiology; Disease outbreaks; Forecasting; Risk assessment/methods; Models, Theoretical (source: MeSH, NLM).
Source: Bull World Health Organ, vol.83 no.5, May 2005
The "next wave"?
Recent predictions of a catastrophic "next wave" of the HIV pandemic in the populous countries of Asia and Eastern Europe (1, 2) are a cause of controversy within international development agencies and have been condemned by some of the national governments concerned (3). The predictions suggest that prevalent HIV infections in China, India and the Russian Federation alone could reach between 35 million and 68 million in 2010, compared with the current global total of 38 million estimated by the Joint United Nations Programme on HIV/AIDS (UNAIDS) and WHO (4). Undoubtedly the drug-using and sexual behaviours that transmit HIV are increasing in these countries (5), and numbers of newly diagnosed HIV infections have surged in recent years (4). However, next-wave scenarios are contingent on a generalized heterosexual spread occurring on a scale similar to that seen in sub-Saharan Africa. Is this possible?
While many studies have shown the correlation between an individual's sexual behaviour and infection with HIV, there is little evidence for population-level measures that can explain differences in adult HIV prevalence (6–8). This lies at the heart of the difficulty in forecasting the scale of heterosexual HIV epidemics, because no measures of sexual behaviour in the population have been identified as accurate predictors of the size of an epidemic. Current forecasts therefore have to rely on scenarios proposed by experts using Delphi or similar techniques or on mathematical models. Models that attempt to forecast the size of a heterosexual HIV epidemic tend to be based either on fitting epidemic curves to available prevalence data and extrapolating (9) or on mathematical descriptions of sexual behaviour and the associated transmission of HIV (10). There is a massive level of uncertainty in such forecasts (5). Thus, while next-wave scenarios can be deemed unlikely or derived through applying a worst-case scenario systematically (11), they can only ever be disproved after the event. It is the role of UNAIDS, WHO and the scientific community to make clear what can and cannot be said about the future of the HIV pandemic. If accurate forecasts are so problematic, is there an alternative approach that can be adopted?
In most countries of Asia and Eastern Europe, in contrast to sub-Saharan Africa, HIV currently remains concentrated within identifiable vulnerable populations such as injecting drug users and their sex partners, men who have sex with men, and sex workers and their clients. In many of these countries it has been possible to estimate the size of these populations, using data from surveys and censuses or indirect estimates based on multiplier or capture–recapture techniques (12). HIV prevalence estimates from sentinel surveillance sites among these populations can then be multiplied by population size to estimate the number of people infected with HIV. UNAIDS and WHO have developed software and held training workshops worldwide to assist national AIDS programme staff and others in the implementation of this approach (13) (see Table 1 for an example from Indonesia).
Once HIV reaches vulnerable populations where condom use is inconsistent or needle sharing is common, epidemics are usually inevitable (14). Ignoring the possibility of heterosexual HIV transmission in the wider adult population, it is possible to explore the effect of different assumptions about the peak prevalence and timing of HIV epidemics among vulnerable populations on future epidemic size. Assumed peak prevalence levels can be informed by comparison with long exposed populations from the same or neighbouring countries. Using this approach, it has been estimated that the epidemics in Eastern Europe and Asia are likely to add about 21 million infections to the HIV pandemic by 2010 if there is not a rapid and expanded response (15). Thus, even without a generalized heterosexual epidemic, the need for action is clear.
The concentrated nature of the HIV epidemics in most Asian and Eastern European countries suggests the utility of a novel approach to the management of HIV in these countries. The techniques of risk assessment, routinely used by environmental epidemiologists, can be used to map exposure to HIV. Behavioural surveillance can identify the high risk behaviours that transmit HIV and can be used to map the distribution of vulnerable populations within a country (16). The need for mapping emerges from the heterogeneous distribution of risk behaviours and hence HIV within countries. For example, in India, injecting drug use is concentrated in the north-east of the country and in urban centres such as Chennai, Delhi and Mumbai, where HIV transmission associated with needle sharing has led to significant epidemics of HIV (17, 18). These maps may be supplemented by surveillance of HIV-related knowledge and of biological markers of risk, such as the sexually transmitted infections that facilitate HIV transmission (e.g. herpes simplex virus type-2), or bloodborne infections such as hepatitis C. Procedures for the estimation of the size of the vulnerable populations can then define the magnitude of exposure to HIV by the different transmission routes in different parts of the country (12).
A key step in risk assessment is to identify what is an acceptable risk. In the United Kingdom, for example, the Health and Safety Executive considers a one-in-a-million excess risk of death per year from environmental pollutants as the level of acceptable risk at which no further improvement in safety needs to be made (19). It is often illuminating to compare policy on acceptable risk with currently accepted risk revealed in morbidity and mortality statistics. For example, using estimates of cause-specific death rates (20) and HIV case reports (21), it is possible to estimate the average time it takes a 15–24-year-old living in the Russian Federation to acquire a one-in-a-million chance of death from specific causes (Table 2). The accepted risk of death from AIDS revealed by these statistics should highlight the need for HIV prevention strategies to bring the accepted risk in line with what is considered acceptable. In reality, this average masks a great deal of heterogeneity in risk. Moral judgements tend to be made about the culpability of those becoming infected with HIV, and the concept of acceptable risk for these people implicitly revised.
Effective surveillance of vulnerable populations identifies prevalent HIV infections. However, the incidence of new HIV infections can show a markedly different pattern. For example, in 2002 in Indonesia, although prevalent infections were equally distributed between injecting drug users, and sex workers and their clients (Table 1), the majority of incident infections were among injecting drug users (22). Patterns of incidence for a given year can be crudely estimated from simple measures of risk behaviour without the need for a complex dynamic model. For example, the average number of visits to different sex workers reported by male clients (e.g. x per year), together with the probability of HIV transmission per sex worker visited (P), and HIV prevalence among sex workers (y), can give an estimate of the expected number of new infections among clients that year » N-N(1-Py)x, where N is the number of clients. Simple estimates of expected incidence based on similar calculations for different vulnerable groups can act as a useful guide for HIV prevention (22). Interventions that focus on HIV prevention among vulnerable populations have been found in the past to be both effective and cost-effective, even where HIV transmission has become more generalized (23, 24).
The relationship between simple measures of risk behaviour and HIV incidence is analogous to the dose–response relationship central to the assessment of health risks in environmental epidemiology. Dose–response relationships need to account for the timing of health outcomes, and this is especially important for infectious diseases with long incubation and infectious periods. The relationship is simplified by focusing on HIV status rather than on health outcomes, and it is possible that measures of risk at the population level can be linked to HIV prevalence some years later. For vulnerable populations the relationship may be reasonably well defined, since high prevalence of unsafe behaviour such as needle sharing typically results in rapid HIV transmission and an epidemic that saturates at high prevalence. Of course, policy-makers are also (and often solely) concerned with risk of HIV to the wider adult population. In sub-Saharan Africa there is evidence, albeit limited, that living in a community with a certain risk profile — economically active, mobile population, many female bar workers, close to a large town — increases an individual's risk of infection over and above his or her own sexual behaviour (8). In Asia and Eastern Europe, although analogous research has not been carried out, models suggest that adults living where injecting drug use and sex work are common are at a higher risk of infection irrespective of their own sexual behaviour (5). Thus, risk in the general population is likely to be correlated with risk in other groups as well as with patterns of sexual behaviour. Epidemiological theory suggests a threshold relationship, such that below a certain level of sexual activity in the population the risk may be minimal (25). This contrasts with linear relationships seen, for example, between risks of ischaemic heart disease and blood pressure, cholesterol or obesity (26). As noted earlier, however, this relationship is not well understood, and therefore dose–response curves for the general adult population will be poorly defined. Furthermore, as the time frame expands, the problems inherent in forecasts of HIV prevalence will return. What role, then, can be played by transmission dynamic models?
Mathematical models of HIV transmission are useful for exploring the relationship between risk behaviour — including that of vulnerable populations such as sex workers — and risk of infection in the wider population. Many models have been published illustrating the importance, for instance, of numbers of sex partners, rates of needle sharing, transmission probabilities, heterogeneities in their values and the mixing between different groups (5, 27, 28). The difficulty of accurately measuring these key variables means that such models are better suited to generating qualitative rather than quantitative predictions. Such qualitative predictions can be compared with observed surveillance data indicating whether gross changes in the epidemiology of HIV are under way. For example, following an epidemic of HIV among injecting drug users, a model based on behavioural data from urban areas of the Russian Federation predicts significant sexual transmission among drug users and also to others through both commercial and noncommercial sex (Fig. 1). Although models are unable to predict the final extent of this sexual transmission, they highlight the importance of preventing sexual transmission among injecting drug users as well as among their non-injecting partners.
Comparison of qualitative model predictions with sentinel surveillance HIV prevalence data and, where available, behavioural data makes clear where changes in patterns of exposure — and therefore prevention — must be occurring. This can also highlight significant gaps in surveillance systems. For example, simple models of HIV transmission among sex workers or injecting drug users estimate significant HIV prevalence among individuals who have stopped sex work or injecting drug use and are therefore no longer considered vulnerable. In the case of female sex workers, their infections may be detected at antenatal clinic surveillance sites if they become pregnant, but men who cease injecting drug use will not be picked up by most surveillance systems and may represent an overlooked source of prevalent and incident HIV infections.
Mathematical models are also essential to estimating the likely impact of behavioural change resulting from interventions on the incidence of HIV. This may be particularly important when transferring a successful intervention from one location to another where the epidemiological context may differ (29). In the past, models of HIV transmission have been used to highlight the importance of good intervention coverage and the efficiency gained by targeting vulnerable populations (30). Ongoing behavioural surveillance as part of a risk assessment framework can help track trends in behaviour and identify effective and ineffective programmes. The integrated analysis of behavioural and biological surveillance data can provide a compelling argument for the success of national programmes (31).
The risk assessment approach described in this paper offers an alternative to scaremongering based on uncertain forecasts. The world health report 2002 has advocated bringing risk assessment into the heart of decisions about the allocation of resources in reducing the global burden of disease (32). A focus on risk may also help an effective response to the ongoing HIV pandemic.
Competing interests: none declared.
1. Eberstadt N. The future of AIDS. Foreign Affairs Nov/Dec 2002 Vol. 81.
2. The next wave of HIV/AIDS: Nigeria, Ethiopia, Russia, India, and China. National Intelligence Council USA; 2002. ICA 2002-04 D, available from: http://www.odci.gov/nic
3. Sharma DC. US charity donation for HIV/AIDS in India sparks government row. Lancet 2002;360:1574.
4. 2004 Report on the global HIV/AIDS epidemic: 4th global report. Geneva: Joint United Nations Programme on HIV/AIDS; 2004.
5. Grassly NC, Lowndes CM, Rhodes T, Judd A, Renton A, Garnett GP. Modelling emerging HIV epidemics: the role of injecting drug use and sexual transmission in the Russian Federation, China and India. International Journal of Drug Policy 2003;14:25-43.
6. Auvert B, Buve A, Ferry B, Carael M, Morison L, Lagarde E, et al. Ecological and individual level analysis of risk factors for HIV infection in four urban populations in sub-Saharan Africa with different levels of HIV infection. AIDS 2001;15:S15-30.
7. Drain PK, Smith JS, Hughes JP, Halperin DT, Holmes KK. Correlates of national HIV seroprevalence: an ecologic analysis of 122 developing countries. Journal of Acquired Immune Deficiency Syndromes 2004;35:407-20.
8. Bloom SS, Urassa M, Isingo R, Ng'weshemi JZL, Boerma JT. Community effects on the risk of HIV infection in rural Tanzania. Sexually Transmitted Infections 2002;78:261-6.
9. UNAIDS Reference Group on Estimates, Modelling and Projections. Improved methods and assumptions for estimation of the HIV/AIDS epidemic and its impact: recommendations of the UNAIDS Reference Group on Estimates, Modelling and Projections. AIDS 2002;16:W1-16.
10. Anderson RM, May RM, Boily MC, Garnett GP. Rowley JT. The spread of HIV-1 in Africa – sexual contact patterns and the predicted demographic impact of AIDS. Nature 1991;352:581-9.
11. Walker N, UNAIDS, quoted in: Washington Post 1 October 2002.
12. Pisani E. Estimating the size of populations at risk for HIV: issues and methods. Arlington (VA): UNAIDS/IMPACT/Family Health International; 2002.
13. Walker N, Stover J, Stanecki KA, Zaniewski E, Grassly NC, Calleja-Garcia J-M, et al. The workbook approach to making estimates and projecting future scenarios of HIV/AIDS in countries with low level and concentrated epidemics. Sexually Transmitted Infections 2004; 80 Suppl 1:i10-i13.
14. Mills S. Back to behaviour: prevention priorities in countries with low HIV prevalence. AIDS 2000;14 Suppl 3:S267-73.
15. Stover J, Walker N, Garnett GP, Salomon JA, Stanecki KA, Ghys PD, et al. Can we reverse the HIV/AIDS pandemic with an expanded response? Lancet 2002;360:73-7.
16. Amon J, Brown T, Hogle J, MacNeil J, Magnani R, Mills S, et al. Surveillance surveys: BSS guidelines for repeated behavioral surveys in populations at risk of HIV. Arlington (VA): Family Health International; 2000.
17. Dorabjee J, Samson L. A multi-centre rapid assessment of injecting drug use in India. International Journal of Drug Policy 2000;11:99-112.
18. HIV estimates in India. Delhi: National AIDS Control Organisation (NACO); 2004 (available from: http://naco.nic.in).
19. Hunter PR, Fewtrell L. Acceptable risk. In: Fewtrell L, Bartram J, editors. Water quality – guidelines, standards and health: assessment of risk and risk management for water-related infectious disease. Geneva: World Health Organization; 2000.
20. Shkolnikov V, McKee M, Leon DA. Changes in life expectancy in Russia in the mid-1990s. Lancet 2001;257:917-21.
21. European Centre for the Epidemiological Monitoring of AIDS. HIV/AIDS surveillance in Europe. Mid-year report 2002. Saint-Maurice: Institut de Veille Sanitaire; 2002.
22. Pisani EP, Garnett GP, Brown T, Stover J, Grassly NC, Hankins C, et al. Back to basics in HIV prevention: a focus on exposure. BMJ 2003;326:1384-7.
23. Creese A, Floyd K, Alban A, Guinness L. Cost-effectiveness of HIV/AIDS interventions in Africa: a systematic review of the evidence. Lancet 2002;359:1635-42.
24. Merson MH, Dayton JM, O'Reilly K. Effectiveness of HIV prevention interventions in developing countries. AIDS 2000;14 Suppl 2:S68-84.
25. Anderson RM, May RM. Infectious diseases of humans: dyamics and control. Oxford: Oxford University Press; 1991.
26. Law MR, Wald NJ. Risk factor thresholds: their existence under scrutiny. BMJ 2002;324:1570-6.
27. Anderson RM, Medley GF, May RM, Johnson AM. A preliminary study of the transmission dynamics of the human immunodeficiency virus (HIV) the causative agent of AIDS. IMA Journal of Mathematics Applied in Medicine and Biology 1986;3:229-63.
28. Blower SM, Hartel D, Dowlatabadi H, Anderson RM, May RM. Drugs, sex and HIV: a mathematical model for New York City. Philosophical Transactions of the Royal Society of London, Series B 1991;321:171-87.
29. Grassly NC, Garnett GP, Schwartlander B, Gregson S, Anderson RM. The effectiveness of HIV prevention and the epidemiological context. Bulletin of the World Health Organization 2001;79:1121-32.
30. Garnett GP, Anderson RM. Strategies for limiting the spread of HIV in developing countries: conclusions based on studies of the transmission dynamics of the virus. Journal of Acquired Immune Deficiency Syndromes 1995;9:500-13.
31. Stoneburner RL, Low-Beer D. Population-level HIV decline and behavioral risk avoidance in Uganda. Science 2004;304:714-8.
32. The world health report 2002 – Reducing risks, promoting healthy life. Geneva: World Health Organization; 2002.
33. National estimates of adult HIV infection, Indonesia 2002. Jakarta: Ministry of Health of the Republic of Indonesia, Directorate General of Communicable Disease Control and Environmental Health; 2003. p. 53.