August 21, 2006

Jaya Shreedhar

Researchers involved in the project warn against complacence.

• Countrywide survey needed for accurate picture
• India has millions of HIV infections irrespective of estimation method

THE ACTUAL number of People Living with HIV (PLHIV) in India might be a third lower than the present estimate of 5.2 million, according to a new population-based study by Indian researchers. The study was done in Guntur district of Andhra Pradesh and was based on antenatal care patterns from the recent National Reproductive and Child Health Survey. The study cautiously extrapolates the findings to make indicative estimates of the national HIV burden. According to the researchers, extrapolation of their findings to the country overall await confirmation. That might well happen later this year when the results of a countrywide National Family Health Survey become known.

The study gives no cause for complacency, warned the researchers. A generous downward revision of HIV prevalence would still leave the country with a massive smouldering epidemic of about 3.5 million people, and hotspots of infection with the imminent danger of explosive spread. Financial, technical, and human resources will remain a high priority to enable a rapid countrywide scale-up of services to prevent new infections and provide care and treatment to the millions of people living with HIV.

Lalit Dandona, Professor of Health Studies from the Administrative Staff College (ASCI), Hyderabad whose team collaborated with the Nizam's Institute of Medical Sciences to conduct the study, said: "The method NACO [National AIDS Control Organisation] follows to gather and process HIV data is doubtless useful, but population-based surveys like the Guntur study probably give us a picture closer to reality. India's HIV figures are likely [to be] a gross overestimation."

India's HIV prevalence rate is presently calculated from data collected each year by the various State AIDS Control Societies through HIV Sentinel Surveillance (HSS). For a few months each year, blood samples are collected from incoming patients at `sentinel' sites in the public health sector. These are located in the clinics for pregnant women coming to the large government hospitals (Ante-Natal Centres or ANCs) and in the departments treating people with sexually transmitted infections (STI). The data are then collated and processed to come up with estimates for the general population.

HSS 2005 indicated that Tamil Nadu, Andhra Pradesh, Karnataka, and Maharashtra together accounted for 3.7 million infections — close to three fourths of India's present estimate of about 5.2 million infections. The extrapolation of the patterns found in the Guntur study would lower the total of HIV infections in the four Southern States to 2 million and bring down India's total to 3.5 million.

Reacting to the new findings, Sujata Rao, Director-General of NACO, said that "the Guntur study is a well researched one."

What explains the higher burden that was estimated last year? The HSS-based estimate relies on the assumption that 5-6 per cent of the general population each year get STIs, says Professor Dandona. "When the high HIV rates among STI patients attending large government hospitals are used for this assumption, the HIV estimates for the general population gets inflated. Inclusion of the STI data in the HIV estimation method is a major distortion factor."

Though the Guntur study adjusted the HIV burden upwards for groups usually left out of HIV surveys such as hostel inmates, prisoners and military recruits all of who may have higher risk of HIV exposure, it adjusted the HIV burden downward based on the finding that ANC sites were picking up a higher number of HIV positive women because private practitioners shunt their HIV positive patients to government hospitals, and because there was overrepresentation of poor women who are likely to have a higher HIV rate.

The see-saw swing of "sudden drops" and "steep increases" in the HIV burden estimates in India points to the need for good science and accurate estimation methods. Robert Bollinger of the Johns Hopkins School of Medicine said: "Comparing estimates from varying methods is like comparing apples and oranges. If India wants more accurate estimates of where the epidemic is heading, it needs to apply similar estimation methods in a consistent way over time." Perhaps the Guntur study and the National Family Health Surveys this year will give India its first real snapshot of how many people are likely to be living with HIV in the country.

(Jaya Shreedhar of Internews Network specialises in areas related to HIV/AIDS.)