AVERT - AVERTing HIV and AIDS

Understanding the HIV and AIDS epidemics

The term epidemic is used when HIV and AIDS are widespread in a community. In order to understand the many epidemics of HIV that are spreading around the world, and the AIDS epidemics which follow in their footsteps, it is necessary to look at certain figures. The figures we need include the number of people living with HIV (the HIV prevalence), the number of new infections (the HIV incidence), and the number of people who have died of AIDS.

HIV/AIDS is a complicated and hugely important world issue. There has been a great deal of study into the medical and social aspects of the world's epidemics, which has generated a great deal of data. This information is often presented as long numbers, graphs and tables, and sometimes the terminology used can be quite technical. In order to turn such information into an understanding of the epidemics, you need to know where the data come from and what they really mean.

There are two main types of national HIV and AIDS statistics:

  • Reports of actual cases tell us the minimum number of people who are affected, but are of limited use if many cases go unreported.
  • Estimates based on surveys give the proportion of people living with HIV, as well as other statistics, according to certain assumptions.

Reported HIV diagnoses

With reported diagnoses, each number indicates an actual positive result for a person's HIV test. This method of looking at an epidemic can give an extremely clear picture in terms of real people who have been affected by the virus, especially when looking at smaller areas. However, it is often not a reliable way of assessing wider trends because many people living with HIV have never taken an HIV test, and not all diagnoses are reported.

In general, national totals of reported HIV diagnoses are only really useful when they come from places with well-developed voluntary testing schemes, such as the USA, Western Europe, Canada or Australia. However, even in these countries, one in every three or four people living with HIV has never taken a test, and some test results go unreported. Some states of the USA do not report any HIV statistics through the name-based system used to compile national surveillance reports.

Another point to remember is that looking at the years in which people tested HIV positive does not tell you when they were infected - the test itself may come many years after infection occurred. And when looking at HIV reports, it's important to keep in mind that there might be more than one reason for trends in the data. An increase in diagnoses might not mean that more people are becoming infected with HIV than in previous years - it might mean, instead, that HIV testing has become more easily available than in recent years, or that stigmatisation of people living with HIV has declined, so more people are willing to be tested.

In terms of characteristics (such as age, gender, race, etc), HIV reports are not necessarily entirely representative of all HIV infections because some groups of people may be more likely to be tested than others. It is worth remembering this limitation when interpreting reported statistics by exposure category in particular.

Reports from the most recent years are usually affected by reporting delays (see Glossary, below).

Reported AIDS diagnoses and AIDS deaths

Most countries have a long history of reporting AIDS cases and in many places - including the USA, Canada and much of Africa - reporting of all diagnoses is compulsory.

A problem with AIDS case reporting is that different countries have different definitions of what actually constitutes AIDS. While a definition of AIDS is becoming standardised, there is still the problem that some resource-poor areas lack HIV testing facilities, and therefore have to diagnose AIDS on the basis of symptoms alone. Another problem is that people with AIDS do not commonly have this disease for long - it either kills them or, given access to antiretroviral medication, they frequently cease to have symptoms.

Antiretroviral treatment has made it difficult to interpret trends in AIDS diagnoses and AIDS deaths in countries where most people have access to the drugs. This is why, in recent years, reliable HIV reporting has become even more important in the richer parts of the world.

If someone is thought to have died from AIDS then this should be recorded on their death certificate. However, it is acknowledged that such records are less reliable in countries where AIDS is highly stigmatised, because doctors are sometimes inclined to spare shame to a family by mis-recording the cause of an AIDS death.

The World Health Organisation says of AIDS case surveillance that, "While giving a general idea of the increase of AIDS in a population, the figures do not reflect the actual prevalence of AIDS disease so much as the accuracy of detection, diagnosis and reporting of the disease syndrome". The proportion of AIDS cases reported varies from less than 10% in some countries to almost 90% in others.1

Estimated HIV prevalence

'HIV prevalence' is given as a percentage of a population. If a thousand truck drivers, for example, are tested for HIV and 30 of them are found to be positive, then the results of a study might say that HIV prevalence amongst truck drivers is 3%.

For the purpose of producing a national or international HIV prevalence figure, researchers include all people with HIV infection who are alive at a given point in time, whether or not they have progressed to AIDS.

In most cases, HIV prevalence cannot be accurately determined from reported cases because many infections are undiagnosed or unreported. The best estimates are mainly based on the results of surveys of large groups of people.

In a country with a generalised epidemic (a high level of infection in the whole population), the national estimate of HIV prevalence is usually mainly based on surveys of pregnant women attending antenatal clinics. Because antenatal clinics are well-attended in most such countries, these data provide a good basis for comparisons; they are also very reliable indicators of trends in prevalence. Surveys collect blood over a short period from all women attending a clinic for the first time (to avoid duplication). All identifying details are removed from the samples, except age group and location of clinic, before they are screened for HIV to determine prevalence.

Many studies have shown that HIV prevalence among pregnant women attending clinics is generally very similar to prevalence in the adult population as a whole. Often some small adjustments are made - for example to compensate for underreporting in the most rural areas, or for large differences between male and female rates of infection. Such refinements are made according to the findings of separate surveys of the general population.

Population based surveys are useful because they tell us how prevalence varies according to gender, race or other characteristics, but they are usually not the main source of national prevalence estimates. One reason for this is that population based surveys are much more complicated and expensive than antenatal surveys. Nevertheless, population based surveys are becoming more frequent, and their influence on national HIV prevalence estimates is increasing. Between 2001 and 2007, thirty countries in sub-Saharan Africa, Asia and the Caribbean conducted national population based surveys. The results of these surveys have led to significant revisions of HIV prevalence estimates for several countries - most notably India in 2007.

In a country with a low-level or concentrated epidemic (where high levels of infection are found only in specific groups), the national estimate of HIV prevalence is mainly based on data collected from populations most at risk - usually sex workers, injecting drug users or men who have sex with men - and on estimates of the sizes of the populations at high risk and at low risk. Reports of HIV diagnoses and AIDS deaths may also be taken into account.

Better understanding of the nature of an HIV epidemic allows better prevalence estimates to be produced. This is why, each year when a new set of estimates is brought out, the figures for previous years may change.

Estimated HIV incidence

'HIV incidence' is the number of new HIV infections in the population during a certain time period. People who were infected before that time period are not included in the total, even if they are still alive.

Unfortunately, directly measuring HIV incidence is a complex and expensive process, so incidence data for many resource-poor areas, and some rich ones, is difficult to find. However, one example is that (according to USAID) in the Masaka region of Uganda, HIV incidence fell from 7.6 per thousand per year in 1990 to 3.2 per thousand per year by 1998.

National estimates of HIV incidence are usually produced by computer models and are based on estimates of HIV prevalence. Such models apply a set of assumptions such as the survival time of those infected with HIV and the mother-to-child transmission rate. Trends in HIV prevalence among teenagers and young adults can give a rough idea of incidence because infections among this group are likely to have been recently acquired.

Estimated AIDS deaths

Computer models are also used to produce estimates of the number of people to have died of AIDS, based on HIV prevalence, according to the same set of assumptions.

The accuracy of the assumptions is monitored using surveys of the general population, and the estimates may be compared with reports of AIDS diagnoses, census records or death certificate data. Organisations like UNAIDS/WHO constantly review and improve their methods and assumptions, taking into account the latest research findings.

Margin of error

This is a phrase used to explain the precision of an estimate. Because the groups looked at by surveys can never be entirely representative of the wider population, and because no computer model is perfect, there will always be some degree of uncertainty attached to figures derived from them. For this reason, an estimate is often accompanied by a range or 'plausibility bound', and the width of the range is an indication of the uncertainty of the estimate. The wider the range, the greater the uncertainty.

The size of a plausibility bound is affected by the quality of the data and the number of steps and assumptions used to arrive at the estimate. Also, ranges tend to be larger when the numbers being estimated are smaller, because it is then likely that fewer people are included in surveys.

Another property of HIV-related plausibility bounds is that they tend to be wider in countries with low-level or concentrated epidemics. This is because, in low-level or concentrated epidemics, one needs to estimate both the numbers of people in the groups at higher risk of HIV infection and HIV prevalence rates in those groups.

Arguably the most important estimates in the world are the national ones produced by UNAIDS/WHO. These organisations say of their own estimates that they "are confident that the actual numbers of people living with HIV, people who have been newly infected or who have died of AIDS lie within the reported ranges".

Understanding HIV prevalence and incidence trends

Changes in HIV incidence statistics can give an idea of whether prevention strategies are being successful in reducing the number of new infections. A society that shows regularly declining incidence figures is one that is experiencing fewer and fewer new infections, which is certainly desirable.

Trends in HIV prevalence are less easy to interpret.

In the early years of a typical HIV epidemic, prevalence increases rapidly because more and more people are becoming infected and few are dying. But prevalence cannot increase forever - eventually the death rate (number of deaths per year) rises to equal the incidence rate (number of new infections per year), and so prevalence reaches a peak.

In some African countries, HIV prevalence appears to have stabilized at a very high level. This means that two things are happening at the same time: many new infections are occurring and many people are dying. And if a country's total population continues to grow then the number of people living with HIV increases even while the prevalence rate remains stable.

A rise in HIV prevalence is not necessarily a sign of failing prevention campaigns. Besides a rise in incidence, it could result from any of the following:

  • The death rate has fallen because of improvements in treatment and care (this has happened in high-income countries).
  • The death rate has fallen because fewer infected people are dying as a result of war, famine or other causes that had disproportionately affected people living with HIV.
  • The death rate has fallen as a result of an earlier drop in incidence (on average, people survive for a number of years after becoming infected, so incidence trends have a delayed effect on death trends).
  • More people living with HIV are imigrating than are emigrating (this affects a number of high-income countries).
  • The survey bias has changed.

Equally, a fall in HIV prevalence is not necessarily a sign of effective prevention campaigns, as it could result from an increase in the number of deaths.

It is even possible for HIV prevalence to increase at a time when HIV incidence is decreasing - for example, in a society that is rapidly scaling-up antiretroviral treatment provision while making only small improvements to prevention activities. The drop in the number of new infections might then be outweighed by the effect of people living longer.

Glossary

HIV and AIDS epidemic statistics use particular terms in particular ways, and sometimes with special meanings. Here, you can find explanations of the terms used in our pages.

Adults

Adults in most reports are defined as men and women aged 15-49. This age range captures those in their most sexually active years. While the risk of HIV infection continues beyond the age of 50, the vast majority of people with substantial risk behaviour are likely to have become infected by this age. Since population structures differ greatly from one country to another, especially for children and the upper adult ages, the restriction of 'adults' to 15-49 has the advantage of making different populations more comparable.

When a report refers to 'men' or 'women', it is usually referring to males and females within these age ranges.

Children

Most reports define children as males and females aged between 0 and 14 years.

Cumulative figures

A 'cumulative' figure gives a total number, from the time that recording of data first began, until a specified date.

Exposure category

This term is used when talking about the ways in which people were infected with HIV. It is often very difficult to say for certain how someone became infected with HIV - for example, an injecting drug user might be infected sexually, or a man who has sex with other men might become infected by a woman. The term 'exposure category' refers to the assumed most probable means of HIV transmission.

Orphans

In UNAIDS/WHO reports, and for most others, an 'orphan' is defined as being someone aged 0 to 17 years, who has lost one or both parents to AIDS. Most other studies also use this definition.

People living with AIDS

Statistics giving numbers of people living with AIDS can sometimes make confusing reading because different countries and agencies have different definitions of what AIDS actually is. For example, in Europe an AIDS diagnosis must be based on the diagnosis of an AIDS-related illness, but in the USA it may also be based on a low CD4 cell count. However, most people who have a low CD4 count will go on to develop an AIDS-related illness anyway.

When testing is unavailable, the presence of AIDS, as opposed to HIV, is determined on the basis of a number of clinical symptoms or signs associated with immune deficiency.

Estimates of the number of people living with AIDS are usually based on the estimated number of people living with HIV and estimated survival times, but minimum estimates may be derived from case reports.

People living with HIV

Many reports and tables give figures for the 'number of people living with HIV'. This number represents all people living with HIV infection, whether or not they have developed symptoms of AIDS, who are alive at the time given.

Estimates of the number of people living with HIV are usually based on the estimated HIV prevalence and total population size, but minimum estimates may be derived from case reports.

Reporting delays

A problem with reported data is that it is sometimes hard to detect trends over time, especially when looking at recent years. Reports of HIV or AIDS diagnoses can sometimes take months, or a few years, to reach a central agency that compiles the total figure.

Statistics experts know from experience roughly what proportion of reports will arrive late. They are therefore able to produce estimates in which the most recent figures have been adjusted to cancel out the effect of reporting delays. The table below illustrates the way in which reported and estimated data can differ, especially in recent years.

Year Reported cases
of HIV		 Estimated cases
of HIV
2000 154 155
2001 171 173
2002 250 260
2003 259 302
2004 169 333

You can see from the table that the apparent drop in 2004 is probably due to reporting delays, and that the true number of new diagnoses probably continued to rise.

Page written by Steve Berry and Rob Noble, and amended by Annabel Kanabus.

Sources:

References:

  1. WHO, Report on Global Surveillance of Epidemic-prone Infectious Diseases - Human Immunodeficiency Virus and Acquired Immune Deficiency Syndrome (HIV/AIDS), 2000

Last updated December 04, 2009