The Joint United Nations Programme on HIV and AIDS (UNAIDS) and the World Health Organization (WHO) have produced country-specific estimates of HIV/AIDS biannually since 1997 [1–3]. These estimates are a primary source of information about the extent and spread of the HIV/AIDS epidemic and its impact. They are also used as inputs in preparing estimates of children orphaned by AIDS, economic and development impacts of the epidemic, and an estimation of the resource needs for prevention, care and treatment. The importance of having comparable country-specific estimates of HIV/AIDS is growing as organizations use estimates to determine how to allocate resources and as one basis of a judgement of need when countries seek funding from international mechanisms.
The procedures used for the end of 1997 estimates have been described in detail elsewhere . Those procedures were based on the work of WHO's Global Programme on AIDS and utilized the AIDS epidemic software (epimodel) that had been developed in the early 1990s when few surveillance data were available . In 1999, UNAIDS created a Reference Group on Estimates, Modelling and Projections that provides guidance on the procedures and assumptions used in preparing estimates of HIV/AIDS and its impact. This group is composed of researchers from various disciplines, and meets yearly to review recent research that can help improve the estimates.
On the basis of the recommendations of the Reference Group and feedback from national programmes, UNAIDS and WHO developed and implemented new procedures for making the end of 2001 estimates of HIV/AIDS . This paper describes the procedures and assumptions used to make the end of 2001 estimates of HIV/AIDS and its impact for 140 low and middle-income countries.
Methods for end of 2001 estimates
Two broad steps are required to make estimates of HIV/AIDS. First, point prevalence estimates and epidemic curves are developed using HIV prevalence data. Next, a set of assumptions about the survival time after HIV infection, sex ratio and other factors, along with prevalence curves are used to derive estimates of HIV incidence and AIDS mortality for adults, and incidence, prevalence and mortality for children. The processes and assumptions used vary with the data available and with the level of epidemic. For all countries, the United Nations Population Division's estimates of population, by age and sex and proportion of population living in urban areas, were used .
The HIV prevalence data come primarily from the national programmes’ surveillance systems with additional data from ad-hoc research. The data are available in the HIV Surveillance Database maintained by the US Census Bureau .
Estimating prevalence over time in countries with generalized epidemics
UNAIDS/WHO have defined generalized epidemics as those with HIV in pregnant women above 1% on a national basis . In countries with generalized epidemics, the procedures used HIV prevalence in pregnant women to approximate the prevalence in all adults, male and female, between the ages of 15 and 49 years. Estimates of HIV prevalence among pregnant women and in community surveys among all men and women aged 15–49 years are available from a number of sites. As shown in Fig. 1, the prevalence in pregnant women is a good proxy measure of adult prevalence [10–19]. Prevalence data from pregnant women are sorted into two geographical categories: major urban areas and outside major urban areas. Epidemic curves are then fit to these data sets using the UNAIDS/WHO Estimation and Projection Package (EPP).
The EPP uses four parameters to derive a best fit to data points by minimizing the least squares difference between the fitted curve and the full set of data points . In this model, the rate of spread of HIV is determined by the rate of spread of the epidemic or reproductive potential, r, the peak prevalence primarily by the fraction of the population considered to be at-risk of infection at the start of the epidemic f(0), and the final endemic prevalence by a parameter φ, which specifies the behavioural response of the population to the epidemic. The three parameters along with the start date of the epidemic, t(0), are calculated by fitting the curve to the prevalence data. This curve fitting approach has been described in detail elsewhere  and the software is available (http://www.unaids.org).
The fitted curves give yearly HIV point prevalence estimates for urban and non-urban areas. The actual prevalence used for the non-urban areas was adjusted as many countries’ surveillance systems do not cover rural areas well. It was assumed that HIV prevalence is lower in rural areas, and therefore if a country's system did not reflect the population in those areas the non-urban prevalence produced by the EPP was adjusted downwards by 20% to reflect this bias.
Adjusted HIV prevalence in pregnant women in urban and rural areas was applied to the population (15–49 years) in urban and non-urban areas to produce estimates of the number of adults living with HIV/AIDS in the two areas. When combined, this gave the estimate of adults living with HIV/AIDS in the country.
Estimating prevalence over time in countries with low-level and concentrated epidemics
For countries that have low-level (prevalence below 5% in populations at higher risk) or concentrated epidemics (prevalence above 5% in at least one population at higher risk, but below 1% in pregnant women) one of two procedures was used depending on the data available.
Countries with sentinel surveillance systems
The basic approach for low-level and concentrated epidemics is to divide the adult (15–49 years) population into sub-groups based on exposure to HIV. For each sub-group, prevalence data and estimates of population size are used to estimate the number of individuals living with HIV/AIDS. By summing the number of infected individuals in these groups, one obtains an estimate of the total number of adults living with HIV in the country.
In countries with low-level or concentrated epidemics, surveillance systems are designed to track the epidemic among individuals most exposed. The actual groups included varied with the nature of the epidemic in a country and the availability of data. For all countries, estimates of sex workers and men who have sex with men were used. In countries with substantial populations who inject drugs, this population was included. Depending on the available behavioural and prevalence data, either clients of sex workers or men with sexually transmitted infections were also included as groups at higher risk.
The default estimates of the size of the groups at higher risk came from a review of available literature on estimates of population sizes. This review was part of a process of developing estimates of needs for a future AIDS vaccine conducted by WHO and UNAIDS . In this process, regional workshops were held where provisional estimates were reviewed and refined by country participants. Regional averages were used (e.g. south Asia) for countries where no data were available.
HIV prevalence data were not always available for all groups. When unavailable, we used prevalence data from another population at higher risk. For example, few countries have HIV prevalence data on men who have sex with men or clients of sex workers, so we used prevalence from male sexually transmitted infection patients.
Adjustments must be made to estimate the number of individuals in the population at lower risk who have HIV/AIDS. First, the size of the higher-risk sub-groups was subtracted from the population aged 15–49 years to produce the estimated size of the lower-risk population. Next, the estimate of prevalence for this group was based on data collected from pregnant women or other general population prevalence data. However, the measured prevalence in pregnant women may overestimate prevalence in the population at lower risk for several reasons. First, few countries have prevalence data for rural areas, where prevalence is often lower. Second, women in the populations at higher risk also become pregnant, increasing the antenatal clinic prevalence. In countries where prevalence is very low, women from groups at higher risk can alter the prevalence among pregnant women substantially.
These biases were handled in two ways. In the first, the adjusted prevalence rate was calculated by: (i) calculating the number of HIV-positive women from the populations at higher risk; (ii) applying unadjusted prevalence from pregnant women to the population of women aged 15–49 years to calculate the total number of HIV-positive women; (iii) subtracting the number of HIV-positive women from high-risk populations from the number of HIV-positive women; and (iv) dividing this number by the lower-risk population to yield an adjusted prevalence rate for the lower-risk population.
In the second approach, information on rural women was used to make the adjustment. For countries where prevalence data for pregnant women in rural areas were available, this prevalence was applied to the low-risk rural population. If no rural data were available, we assumed that prevalence in these areas was lower than in urban areas, and a lower bound on the prevalence from pregnant women in urban areas was used.
Another approach to estimating the number of individuals from low-risk groups who are infected with HIV/AIDS was used for some countries with more recent epidemics. For these countries, estimates of low-risk populations infected were made by estimating the number of regular sexual partners of infected individuals and then applying a yearly transmission probability.
By summing the number of individuals living with HIV/AIDS in each group, point prevalence estimates were made for multiple years, and a start date of the epidemic was set. Using these point estimates, a prevalence curve was fit for the overall epidemic in the country. An example of how estimates were made for countries with low-level and concentrated epidemics is shown in the Appendix.
Countries with reported HIV cases
For many countries in eastern Europe and central Asia, the primary data available are HIV case reports. For these countries, adjustments for under-detection and reporting were made by multiplying the yearly officially reported cases by three (for the low estimate) and five for the high estimate for all countries. These adjustment values were determined in consultation with national programmes in the regions and by comparing the reported to estimated ratio in the Ukraine, which has implemented a sentinel surveillance system. The yearly estimates of new infections and assumptions about survival were then used to produce national yearly estimates of prevalence.
Using Spectrum and epidemic patterns to generate estimates of incidence and mortality
The software package Spectrum  was used to produce estimates of adult incidence and mortality, as well as estimates for children infected via mother-to-child transmission. Spectrum uses the prevalence curves and applies a set of assumptions to produce age and sex-specific estimates of incidence, prevalence, and mortality for adults and children. These assumptions are: (i) female-to-male prevalence ratio; (ii) effects of HIV on fertility; (iii) transmission of HIV from mother to child; (iv) survival time from infection to death for adults and children; (v) age patterns of prevalence; and (vi) effects and coverage levels for highly active antiretroviral therapy (HAART).
Female-to-male prevalence ratio over time
For generalized epidemics, the assumption was that early in the epidemic there would be more men infected than women, whereas later in the epidemic there would be more women. The ratio of women to men increased from 0.23 at the beginning of the epidemic, stabilizing at 1.2 nine years later. For countries with concentrated epidemics, where injecting drug users, male clients of sex workers, and men who have sex with men are often major contributors to the epidemic, the female-to-male ratio starts much lower at 0.07, and grows to approximately 0.40 after 30 years. Finally, for countries with concentrated epidemics where injecting drug use plays little role and heterosexual transmission is higher, an intermediate pattern of the female-to-male ratio was used in which the ratio stabilizes at 1 after 10 years.
Effects of HIV on fertility
For each country, age-specific patterns of fertility from the United Nations Population Division were used as the basis for estimating the number of children born to HIV-positive mothers. Adjustments were made to these national fertility patterns to address the representativeness of HIV-positive young women and to correct for the effects of HIV/AIDS on fertility. First, for women aged 15–19 years who are HIV positive, fertility was assumed to be 50% higher than the national rate for that age because all HIV-positive women are sexually active and are exposed to the risk of pregnancy. For all other age groups, the age-specific fertility rate was reduced by 20% to account for the reduction of fertility as a result of HIV/AIDS [11,19,22,23]. The fertility of HIV-negative women was adjusted downwards (15–19 years) or upwards (20–49 years) to compensate for the increase or reduction among HIV-positive women in order to leave the overall fertility rate in each age group unchanged.
Transmission of HIV from mother to child
For all countries, we assumed that 32% of children born to HIV-positive mothers would be infected with HIV [6,24]. For countries where there was sufficient coverage of programmes for the prevention of mother-to-child transmission, the transmission rate was adjusted on the basis of the level of coverage and the type/effectiveness of the therapy offered.
Survival time from infection to death
For all countries, the survival time for adults who are not receiving antiretroviral therapy was described by a Weibull function with a median survival time of 9 years (see Fig. 2). Survival time was set at 9.4 years for women and 8.6 years for men, to reflect the earlier age of infection among women and the consequent longer survival. For children, a double Weibull distribution was used to describe survival time, with median survival of less than 3 years for children not receiving antiretroviral therapy .
Age patterns of prevalence
Age-specific estimates were made by distributing the total number of adult infections across each 5-year age group from 15–19 to 45–49 years for both men and women. These patterns were based on data from population-based prevalence studies and AIDS case reports that reflect age and sex variations in prevalence [7,25–27].
Effects and coverage levels of highly active antiretroviral therapy
Survival time was adjusted for individuals receiving HAART. HAART was assumed to increase survival by an average of 3 years for low and most middle-income countries. This increase in survival for those on HAART was based on the assumption that, at least initially in these countries, the provision of HAART would be without the level of case management found in many high-income countries and with fewer drug regimens available. These factors plus the patient adherence problems and inherent therapy failures would limit the initial impact of therapy on survival.
For the 2001 estimates, we assumed no appreciable HAART coverage in any country with a generalized epidemic. HAART coverage was sufficiently high in several countries of south America, eastern Europe and Asia to be included in the estimates of mortality. For some middle-income countries, it was assumed that HAART could extend average survival by up to 5 years, depending on the level of case management.
2001 country-specific estimates
The procedures described above were used to produce the 2001 estimates of prevalence and mortality for adults and children in 140 low and middle-income countries. For high-income countries, existing national estimates of prevalence and mortality were reported. These estimates are presented in Table 1. In addition, estimates of the number of children under the age of 15 years who had lost one or both parents as a result of AIDS were made. The procedures for this estimate have been described in detail elsewhere [6,28].
There have been many improvements in the methods since those that were used for the 1997 and 1999 estimates. We discuss below the improvements over the previous methods and the remaining limitations.
The methods for the 2001 estimates are improved over those used in the previous rounds of estimates [1,2,4]. First, the new curve-fitting approach incorporated into the EPP is a major improvement on epimodel, which used only a start date and a single year prevalence estimate to define the epidemic curve. Also, by deriving separate epidemic curves for major urban and non-urban areas from fitting all data, the epidemic in a country is described more accurately. Second, reviews of recent research led by the Reference Group has refined assumptions about survival, the effects of HIV/AIDS on fertility, the representativeness of women in antenatal care and other factors that improve the estimates of prevalence and mortality . Third, estimates of the size of exposure groups have improved, yielding better estimates in countries with low-level and concentrated epidemics . Finally, the end of 2001 estimates also benefit from two more years of prevalence data.
A major determinant of the quality of estimates is the availability of data. Whereas almost all countries have some information about the prevalence of HIV/AIDS, the quality and quantity of data varies tremendously . One of the major weaknesses in estimating HIV prevalence in countries is the representativeness of the data produced by the surveillance systems. In countries with generalized epidemics, the number of assumptions required to translate HIV prevalence in pregnant women to adult prevalence are fewer and are supported by more research. There are biases, however, even in these countries. One major bias is that HIV prevalence data from pregnant women are primarily collected in urban or peri-urban sites and information on urban/rural variations remains sparse. Although adjustments for this have been made in the estimates, data from truly rural areas are lacking for many countries.
Studies in several African countries comparing prevalence among pregnant women with prevalence in the 15–49-year-old population (men and women) suggest that prevalence in pregnant women attending antenatal clinics is a good proxy measure of prevalence in the 15–49-year-old population [10–19]. In other regions, this relationship is less clear. A recent study in Cambodia found good agreement between the prevalence in pregnant women and the general prevalence in women in urban areas. In rural areas, the prevalence among pregnant women was higher than in the community-based sample . In some cases, data in pregnant women overestimate prevalence in others it underestimates, meaning that there is no substitute for local data and local analysis of the situation. In addition, changes in the contraceptive rate can also alter the relationship between prevalence in pregnant women and the general population .
In low-level and concentrated epidemics, the coverage in key subgroups including sex workers, injecting drug users, and men who have sex with men is often quite limited, and the range of possible sizes of these populations in each country is quite large. This is perhaps the greatest source of error in estimates of HIV prevalence. In addition, prevalence data for injecting drug use are often collected in detention settings or treatment clinics. Data collection among men having sex with men in most developing countries has been rare, leaving huge gaps in our understanding of this important component of the global pandemic. Furthermore, these populations often change over time, e.g. injecting drug use becomes more common in some places as prices go down or may become less common if other oral or inhaled drugs gain popularity . Also, in larger countries such as Indonesia, Russia, India and China models should be made at a regional or provincial level. Whereas for some countries an attempt was made to build regional models, few had sufficient data to make these estimates. Finally, country estimates made primarily based on HIV case reports have a large degree of uncertainty as few data exists on the completeness of reporting.
Surveillance systems themselves are not static – they change and evolve. Normally, surveillance begins in sites believed to have higher prevalence, but as the system expands, lower prevalence sites are added. Moreover, coverage provided by sites can change over time with changes in the healthcare structure . This means that analysis must look at how trends have changed in those sites that have been sampled consistently.
Estimating the uncertainties
One important consideration is the uncertainty in the estimates of prevalence. Both for the end of 1999 and 2001 estimates ranges were produced for country-specific estimates of prevalence and mortality. The ranges varied from plus or minus 20% for countries with the best data, to plus or minus 35% for those countries with less or poorer data. Estimates of incidence and mortality by definition are even less robust as they are based on the estimates of prevalence, and additional assumptions about survival time are applied to produce the estimates of mortality and incidence.
The estimates by UNAIDS/WHO are produced in consultation with national AIDS programmes. Many countries have their own processes for producing estimates and projections of HIV/AIDS. Countries such as Thailand, Brazil, South Africa and Uganda (along with many others) have well established processes for collecting data and producing updated estimates [32–35]. UNAIDS/WHO follow a consultative process in which provisional estimates are produced and sent to country programmes for review and comment, and based on the feedback, the provisional estimates are improved.
In this review process, countries are asked to comment on the provisional estimates. If there is broad agreement (within plus or minus 20%) on the point prevalence estimates of prevalence and mortality for adults and children, UNAIDS/WHO use the country's estimate and report the range from the original estimate. If there is disagreement on the estimates, data and assumptions are compared to try to reach consensus on the best estimates. Finally, if agreement on the best estimate is not reached a point estimate is not published.
The authors would like to acknowledge the work of the national AIDS programmes. In addition, the time and work of the scientists and researchers who have participated in the meetings of the UNAIDS Reference Group on Estimates, Modelling and Projections has been instrumental in improving the methods and procedures presented here.
1. The United Nations Joint Programme of HIV/AIDS (UNAIDS). Report on the global HIV/AIDS epidemic, June 1998
. Geneva: UNAIDS; 1998.
2. The United Nations Joint Programme of HIV/AIDS (UNAIDS). Report on the global HIV/AIDS epidemic, June 2000
. Geneva: UNAIDS; 2000.
3. The United Nations Joint Programme of HIV/AIDS (UNAIDS). Report on the global HIV/AIDS epidemic, 2002
. Geneva: UNAIDS; 2002.
4. Schwartländer B, Stanecki KA, Brown T, Way PO, Monasch R, Chin J, et al
. Country-specific estimates and models of HIV and AIDS: methods and limitations. AIDS
5. Chin J, Lwanga, SK. Estimation and projection of adult AIDS cases: a simple epidemiological model. Bull WHO
6. The UNAIDS Reference Group on Estimates, Modelling and Projections. Improved methods and assumptions for estimation of HIV/AIDS epidemic and its impact: recommendations of the UNAIDS Reference Group on Estimates, Modelling and Projection. AIDS
7. Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat. World population prospects. The 2000 revision
. New York: United Nations; 2001.
8. United States Bureau of the Census, Population Division, International Programs Center. HIV/AIDS surveillance data base
. Washington, DC: US Bureau of the Census; June 2002.
9. World Health Organization and the Joint United Nations Programme on HIV and AIDS (UNAIDS). Guidelines for second generation HIV surveillance
. Geneva: UNAIDS; 2000.
10. Fontanet AL, Messele T, Dejene A, Enquselassie F, Abebe A, Cutts F, et al
. Age- and sex-specific HIV-1 prevalence in the urban community setting of Addis Ababa, Ethiopia. AIDS
11. Gregson S, Terceira N, Kakowa M, Mason PR, Anderson RM, Chandiwana SK, et al
. Study of bias in antenatal clinic HIV-1 surveillance data in high contraceptive prevalence populations in sub-Saharan Africa. AIDS
12. Fylkesnes K, Ndhlovu Z, Kasumba K, Mubanga Musonda R, Sichone M. Studying dynamics of the HIV epidemic: population-based data compared with sentinel surveillance in Zambia. AIDS
13. Kilian AHD, Gregson S, Ndyanabangi B, Walusaga K, Kipp W, Sahlmüller G, et al
. Reductions in risk behaviour provide the most consistent explanation for declining HIV-1 prevalence in Uganda. AIDS
14. Kwesigabo G, Killewo JZJ, Urassa W, Mbena E, Mhalu F, Lugalla JLP, et al
. Monitoring of HIV-1 infection prevalence and trends in the general population using pregnant women as a sentinel population: 9 years experience from the Kagera region of Tanzania. J Acquir Immune Defic Syndr
15. Ministry of Health, Republic of Rwanda. 1997 population-based serosurvey
. Kigali: Ministry of Health; January 1998.
16. Kigadye RM, Klokke A, Nicoll A, Nyamuryekung'e KM, Borgdorff M, Barongo L, et al
. Sentinel surveillance for HIV-1 among pregnant women in a developing country: 3 years’ experience and comparison with a population serosurvey. AIDS
17. Wawer MJ, Serwadda D, Gray RH, Sewankambo NK, Li C, Nalugoda F, et al
. Trends in HIV-1 prevalence may not reflect trends in incidence in mature epidemics:data from the Rakai population-based cohort, Uganda. AIDS
18. Kahindo M, Nyang J, Chege J. Multicentre study on factors determining the differential spread of HIV in Africa – preliminary results of the Kisumu study site.
In: 2nd National Conference on HIV/AIDS
. Kenya, 28–30 October 1998; and sentinel data from the National AIDS and STDs Control Programme, Nairobi.
19. Glynn JR, Buvé A, Caraël M, Kahindo M, Macauley IB, Musonda RM, et al
. Decreased fertility among HIV-1 infected women attending antenatal clinics in three African cities. J Acquir Immune Defic Syndr Human Retrovirol
20. Esparza J, Chang ML, Widdus R, Madrid Y, Walker N, Ghys PD. Estimation of ‘‘needs’’ and ‘‘probable uptake’’ for HIV/AIDS prevention vaccines based on possible policies and likely acceptance. Vaccines
21. Stover J. aim: a computer program for making HIV/AIDS projections and examining the social and economic impacts of AIDS.
Washington, DC: The Futures Group International/Policy Project; May 2002 (http://www.futuresgroup.com
22. Gray RH, Wawer MJ, Serwadda D, Sewankambo N, Li C, Wabwire-Mangen F, et al
. Population-based study of fertility in women with HIV-1 infection in Uganda. Lancet
23. Desgrées du Loû A, Msellati P, Yao A, Noba V, Viho I, Ramon R, et al
. Impaired fertility in HIV-1 infected pregnant women: a clinic-based survey in Abidjan, Côte d'Ivoire. AIDS
24. DeCock KM, Fowler MG, Mercier E, de Vincenzi I, Saba J, Hoff E, et al
. Prevention of mother-to-child HIV transmission in resource-poor settings: translating research into policy and practice. JAMA
25. Fylkesnes K, Musonda RM, Sichone M, Ndhlovu Z, Tembo F, Monze M. Declining HIV prevalence and risk behaviours in Zambia: evidence from surveillance and population-based surveys. AIDS
26. Buvé A, Caraël M, Hayes RJ, Auvert B, Ferry B, Robinson NJ, et al
. Multicentre study on factors determining differences in rate of spread of HIV in sub-Saharan Africa: methods and prevalence of HIV infection. AIDS
2001, 15 (Suppl. 4)
27. Carpenter LM, Nakiyingi A, Ruberantwari S, Malamba A, Kamali J, Whitworth JAG. Estimates of the impact of HIV-1 infection on fertility in a rural Ugandan cohort. Health Transition Rev
1997, 7 (Suppl. 2)
28. USAID, UNICEF, and UNAIDS. Children on the brink 2002: a joint report on orphans estimates and program strategies
. Washington, DC: USAID; 2002.
29. Walker N, Garcia-Calleja JM, Heaton L, Asamoah-Odei E, Poumerol G, Lazzari S, et al
. Epidemiological analysis of the quality of HIV sero-surveillance in the world: how well do we track the epidemic? AIDS
30. Saphonn V, Hor SP, Ly SP, Chuon S, Saidel T, Detels R. How well do antenatal clinic (ANC) attendees represent the general population? A comparison of HIV prevalence from ANC surveillance sites with a population-based survey of women aged 15–49 in Cambodia. Int J Epidemiol
31. Hanenberg R, Rojanapithayakorn W. Changes in prostitution and the AIDS epidemic in Thailand. AIDS Care
32. Thai Working Group on HIV/AIDS Projection. Projections for HIV/AIDS in Thailand: 2000–2020
. Bangkok: AIDS Division, Ministry of Public Health; 2001.
33. STD/AIDS Control Programme. HIV/AIDS surveillance report, June 2001
. Kampala, Uganda: Uganda Ministry of Health; June 2001.
34. Brazil Ministério da Saúde. Boletim Epidemiológico – AIDS XIII.
Brasília: Ministério da Saúde; 2000.
35. Department of Health, South Africa. National HIV and syphilis sero-prevalence survey of women attending public antenatal clinics in South Africa, 2000
. Pretoria, South Africa: Department of Health; 2001.
Table 2 shows an example of estimating adults living with HIV/AIDS in a country with a low-level or concentrated epidemic. This example is drawn from one country, but the adult population has been set at 50 000 000 as an example and the name of the country has been removed.
For this country, five groups at higher risk of infection were used as the basis of the estimate. For each of these populations, various sources of information were used to make the population size estimates. For injecting drug users, an estimate from the Harm Reduction Network was the basis of the low estimate, whereas the high estimate was from the national government. For the estimate of the number of men who have sex with men, the low estimate was from the national government, whereas the high estimate was based on a regional average. Likewise, the two estimates of the number of sex workers were based on an estimate from the national government and a regional average. For the number of men and women with sexually transmitted infections, the low and high estimates were based on prevalence surveys that had been run in several provinces. The low and high estimates bracketed the mean value found in those studies.
The estimated HIV prevalence was based on sentinel surveillance data from each of these groups. For no groups except injecting drug users was there wide geographical coverage, therefore the estimated range in prevalence was quite large to reflect the uncertainty in a national estimate of prevalence in these groups.
The estimated number of individuals living with HIV/AIDS in each of the groups at higher risk is simply the product of the estimated population size and prevalence. For each group, the range in the estimated number of individuals with HIV/AIDS is large, reflecting the uncertainty in both the estimate of the population size and prevalence. In addition to infections among individuals at higher risk, there were also data in this country suggesting that infections had occurred as a result of both unsafe medical injections and unsafe transfusion and blood donor practices. The national government had made estimates of individuals infected from these two modes of transmission, and these estimates were added to those that occurred among the groups at higher risk.
Finally, this country has a fairly recent epidemic and few data from populations at lower risk. The estimate of number of infected adults from those at lower risk was based on estimating the number of regular sexual partners of groups at higher risk that would be infected. This was done by using national behavioural data to estimate the percentage of the infected individuals who have regular sexual partners who are not part of the populations at higher risk. Then it was assumed that approximately 4% of partners would be infected per year. This approach yields a national estimate of 46 300 adults living with HIV/AIDS, with a range of 22 300 to 93 200. Cited Here...