Calculating HIV/AIDS Estimates

By Joint United Nations Programme on HIV/AIDS (UNAIDS)

One question that is raised often all around the continent of Africa is: How do the international agencies like The Joint United Nations Programme on HIV and AIDS (UNAIDS) and the World Health Organization (WHO) arrive at their estimates for epidemics such as HIV/AIDS? This is an important question because estimates are used not only to determine how international resources to fight HIV/AIDS will be allocated to countries but also remain the primary source of information about the extent of the epidemic and its impact for both researchers and lay people.

Introduction
This paper describes the procedures and process used to make the 2001 round of UNAIDS/WHO estimates of HIV/AIDS. The paper focuses on the different approaches used to make estimates of prevalence in countries with generalized and low-level and concentrated epidemics as well as on new curve-fitting software that was developed to produce epidemic curves for each country. In addition, it presents the assumptions used (e.g. survival from infection to death, the rate of mother-to-child transmission) that are required to derive estimates of incidence and mortality in adults, as well as prevalence, incidence and mortality in children. The paper describes the general process by which the estimation and modelling procedures have been refined and improved over time. The paper also discusses the limitations and weaknesses of the procedures and the data used to make the estimates, and suggests areas where further improvements need to be made.

Calculation of Estimates
UNAIDS/WHO estimates are based on all available data, including surveys of pregnant women, population-based surveys such as household surveys conducted by individual countries such as Kenya, Mali, Zambia and Zimbabwe, as well as other surveillance information. UNAIDS views such information as complementary and useful in helping to estimate the number of people living with HIV in a country.

There have been steady improvements in the modelling methodology used by UNAIDS/WHO and partners, along with better data from country surveillance. These have led to lower global HIV/AIDS estimates, not just for the current year but also for past years, despite the continued expansion of the global epidemic. Current estimates therefore cannot be compared directly with estimates from previous years, nor with those that may be published subsequently.

UNAIDS and WHO continue to work with countries, partner organizations and experts to improve data collection. 

UNAIDS/WHO publish updated country estimates biannually. The most recent estimates for end 2003are contained in the 2004 Report on the global AIDS epidemic, released at the International AIDS Conference in Bangkok (11-16 July 2004). 

For more information, see the Questions & Answers sections on UNAIDS methodology and epidemic in Africa:
Q&A II: Basic facts about the HIV/AIDS epidemic and its impact
Section II: The status of the global epidemic and modes of transmission in different regions
Section III: Other epidemiological and related issues

Methodology
UNAIDS/WHO, in close consultation with countries, employs a six-step method to obtain national estimates of HIV prevalence. The following is a brief description of the methodology, software and assumptions produced by the UNAIDS reference group on estimates, modelling and projections to provide the relevant technical basis for the UNAIDS/WHO global estimates and projections of HIV prevalence. The software packages and their manuals may be downloaded from Epidemiological software and tools.

In countries with a generalized epidemic, national estimates of HIV prevalence are based on data generated by surveillance systems that focus on pregnant women who attend a selected number of sentinel antenatal clinics. This data is entered into the Estimation and Projection Package (EPP) software which fits a simple epidemiological model to find the best fitting curve that describes the evolution of adult HIV prevalence over time. This adult prevalence curve along with national population estimates and epidemiological assumptions are then entered into the Spectrum software program to calculate the number of people infected, new infections and deaths.

Assumptions 
This method assumes that in countries with a generalized epidemic HIV prevalence among pregnant women is a good approximation of prevalence among the adult population (aged 15-49).Studies conducted at subnational level in a number of African countries have provided the evidence for this assumption (by directly comparing HIV prevalence among pregnant women at antenatal clinics to that detected among the adult population in the same community).

In countries with a low level or concentrated epidemic, national estimates of HIV prevalence are primarily based on surveillance data collected from populations at high risk (commercial sex workers, men who have sex with men, injecting drug users) and estimates of the size of populations at high and low risk. This information is entered into point prevalence and projection spreadsheet models, the Workbook Method, to find the best fitting curve that describes the evolution of adult HIV prevalence over time. This adult prevalence curve along with the national population estimates and epidemiological assumptions are then entered into the Spectrum software program to calculate the number of people infected, new infections and deaths.

More detailed explanation of methods and assumptions may be found on the UNAIDS reference group on estimates, modelling and projections website http://www.epidem.org/Default.htm.  

* UNAIDS, "Epidemiology: How Do UNAIDS/WHO arrive at Estimates?"

Download PDF version [59Kb] Back to main page