(In this section on the global epidemic, all references unless otherwise indicated come from the UNAIDS AIDS Epidemic Update for December 2004 seehttp://www.unaids.org/wad2004/report.html- and the individual-country UN Epidemiological Fact Sheets on HIV/AIDS and Sexually Transmitted Infections.)

As of December 2004, the United National Global AIDS Programme, UNAIDS, estimated that 39.4 million people were living with HIV worldwide. These included 2.2 million children under fifteen years of age and 17.6 million women. 

A total of 4.9 million people are thought to have been newly infected with HIV during 2004, 640,000 of whom were children aged under 15 years.

The total number of AIDS deaths in 2004 was 3.1 million exactly what it was two years previously. 2.6 million adults, almost half of whom were women, and 640,000 children under the age of 15 died of AIDS in 2004.

About 13,400 new HIV infections occurred every day in 2004 one every six and a half seconds, with an AIDS death every 10 seconds. 95% of them were in the developing world. 2,000 infections occurred in children under 15 years of age. Of the 12,000 new infections occurring each day in people aged 15-49, 50% are in women and 50% are in people aged 15-24.    

Since the beginning of the epidemic, more than 21 million people have died of AIDS.

The global prevalence appears to have since the AIDS Reference Manual was last revised in 2002 (see chart below). At that point the worldwide total was thought to be 40 million, with 3.2 million children and 19.2 million women infected.

In part this is a real slowdown. The epidemic in the most severely affected parts of the world has got to the mature point where as many people are dying of HIV as are catching it. Prevalence in southern Africa may be beginning to stabilise and in east Africa has already started to fall.

It is important to remember that the reason for the decline in prevalence is because mass dying is going on. Prevalence will rise again if universal treatment access becomes a reality, unless it goes hand-in-hand with effective and radical prevention measures.

Much of the decline is apparent, not real, however. UNAIDS has revised its figures downwards over the last few years because its estimates are getting more precise or rather, its estimate of exactly how imprecise they are is getting better. The 39.4 million is a best guess the true figure could be anything from 35.9 to 44.3 million.

Sentinel versus population surveillance

HIV and AIDS surveillance is a far less precise science in many countries than it is in the UK. Even the USA is unable to release accurate estimates of how many of its inhabitants have HIV, due to different reporting systems in different states. Some European countries like Italy and Spain do not compile HIV figures from every region. And when it comes to poor countries with inadequate health infrastructure, UNAIDS has to rely on ‘sentinel’ HIV surveillance among groups likely to come into contact with health workers and epidemiologists such as pregnant women, sex workers, STI clinic attendees and tuberculosis patients. They are also more likely to be city-dwellers, who in most (though not all) countries have higher HIV prevalence than the rural population. UNAIDS has come to realise that theses groups may tend to have higher HIV prevalence than the general population. If HIV surveillance in the UK relied on attendees at London GUM and antenatal clinics, for example, it would overestimate UK prevalence anything up to six-fold. Sampling pregnant women may in other cases underestimate HIV if the epidemic is concentrated among gay men and injecting drug users.

The other method is to use random population sampling to determine HIV prevalence. This risks bias in the opposite direction to sentinel surveillance. It will tend not to sample enough members of vulnerable communities such as sex workers and men who have sex with men, and is of little use in countries with tightly-focused epidemics. And voluntary sampling is likely to underestimate prevalence due to HIV stigma. A study comparing the two methods in Africa found that population-based sampling, compared with sentinel surveillance, tended to underestimate HIV prevalence in men and overestimate it in women – possibly because women are more often at home to be surveyed (Boerma 2003).

As a result, even in Uganda, for example, one of the most-researched and documented of African countries, UNAIDS estimates that about 530,000 people or 4.1% of the adult population is living with HIV, but gives high and low estimates of 350,000 and 880,000 (2.8% and 6.6%) as representing the spread of uncertainty. In neighbouring Kenya, a survey in January 2004 found a national prevalence of 6.7% (4.5% in men and nearly 9% in women); previous estimates had been as high as 15%.

In some countries – and in particular the two countries thought to have the highest numbers of people living with HIV, South Africa and India – the true number of people living with HIV has caused huge political controversy. Prominent anti-apartheid journalist Rian Malan questioned South Africa’s HIV figures in 2001. However he based his claim that prevalence was a great deal lower than the official figures on death certificates and other documented causes of death by AIDS. Subsequent investigation found that families were very reluctant to have HIV recorded on death certificates and in fact previous estimates had been up to 40% too low (Groenewald).

India bases its HIV figures on sentinel surveillance in publicly-funded antenatal clinics. As has previously been noted, surveillance among pregnant women may over- or underestimate HIV prevalence considerably, and India has the additional problem that 80% of its healthcare takes place in privately-run clinics. In addition India has an extremely variable HIV epidemic, with sentinel surveillance prevalence ranging from zero in Haryana state (which didn’t mean it was zero, it meant no HIV infections were found at antenatal clinics) to 2.25% in Andhra Pradesh.

As a result the Indian government was able to make the nonsensical claim in May 2005 that HIV incidence countrywide had fallen by 95% in a single year, from 520,000 in 2003 to 28,000 in 2004.

In fact the figures previous to 2004 were inflated; in 2003, a change in data collection methods included an increase in sentinel surveillance sites and an expansion of testing from mostly high-risk populations to include low-risk populations. Previously, the prevalence among the high-risk populations was reported as if it was the general prevalence in the population. New HIV infection figures for 2003 also included many cases that erroneously had not been recorded in previous years, making the number of new HIV infections between 2002 and 2003 appear larger than they were.

References

Boerma JT et al. Estimates of HIV-1 prevalence from national population-based surveys as a new gold standard? Lancet 362: 1929-1931, 2003. 

Groenewald, Pet al. Identifying deaths from AIDS in South Africa. AIDS 19(2): 193-201, 2005.

Malan Rian. AIDS in Africa: in Search of the Truth. Rolling Stone Magazine, November 22, 2001

Misra, Neelesh, HIV Cases Show Huge Drop in India, but It's a Glitch in the Statistics. Associated Press, 31 May 2005.

UNAIDS: AIDS Epidemic Update, December 2004.