Don't Be Misled On AIDS Numbers

[Guest guest]

Don't Be Misled On AIDS

Pallava Bagla

10 Jul 2007: It may be premature to start celebrating that number of

people infected with the dreaded human immunodeficiency virus (HIV)

has come down by half to 2.47 million as per the latest estimate

released by the government. There are more unknowns than knowns in

the methodology adopted by the National AIDS Control Organisation

(NACO) to arrive at this figure.

The government went in for a different approach in 2006. Earlier, it

would restrict itself to 'sentinel' surveys, thrust on India by

several UN agencies as being the best way to track the AIDS epidemic.

This approach was believed to be flawed, as it did not seem to

capture trends for the entire population.

This time, the government decided to supplement the results of

sentinel surveys with community data. Under this, an HIV/AIDS

component was included in the third National Family Health Survey

(NFHS) conducted in 2005.

Does a combined effort produce more accurate numbers? To figure this

out, one would have to look at how the NFHS survey compiled HIV/AIDS

data. The issue is complicated by the fact that the government has so

far been cagey about discussing its methodology or sharing its

disaggregated data. It has not released the results of sentinel

surveillance for the current year.

As for the performance of the national AIDS control programme, that

cannot be judged by comparing final figures derived through different

methods.

Health minister Anbumani Ramadoss expressed satisfaction over the new

data sets without telling us why. He only explained that sentinel

surveillance (in pregnant women) was expanded to 1,122 sites in 2006

from the earlier 703 sentinel sites. What about the results of this

expanded exercise?

According to NACO's latest estimate, the prevalence of HIV/AIDS in

India has to be scaled down from one in 100 or about 0.9 per cent to

one in 300 or 0.36 per cent, or to 2-3.1 million from the

government's 2006 estimate of 5.2 million. As Prabhat Jha of the

Center for Global Health Research, Toronto, says, " It is hard to know

how much of this drop is due to a new computer programme rather than

the efforts of the AIDS control programme " .

In fact, Ramadoss concedes that there has been a marginal reduction

in prevalence. This implies that if only the results of latest

sentinel survey were to be compared to the earlier one, as was the

norm before this year, the prevalence of HIV/AIDS is pretty much

where it was — at 0.9 per cent.

What are sentinel surveys? For about eight weeks every year clinics

frequented by pregnant women are intensively monitored and patients

tested for presence of HIV, the hope being that by tracking young

sexually active women trends may emerge on how the epidemic was

progressing at specific locations.

In addition, people visiting clinics treating sexually transmitted

diseases like syphilis and gonorrhoea were also monitored, which

would indicate how the disease was progressing in the so-called 'high

risk' groups. When collated and viewed over a period of time, the

sentinel surveillance gives a good indicator on whether HIV/AIDS

spread is on an upward curve or not.

However, Indian experts point out that it would be wrong to derive

absolute numbers of infections, as opposed to trends, from this

dataset. This led to a new approach this time round.

Under NFHS-3, the ministry of health and family welfare decided to

gather data on the prevalence of HIV/AIDS by carrying out an AIDS

test. NFHS is by design a random statistical survey done in the

community, meant to provide signals of what is happening in the

larger population. Because of this nature it under-represents certain

populations. In this nationwide exercise about 100,000 people in the

age group 15-54 voluntarily undertook an AIDS test.

This new data reveals that the prevalence of HIV/AIDS was 0.28 per

cent, which means that merely about 280 HIV positive cases were

actually detected by the surveyors.

This when broken down further for the 28 states and seven Union

territories (no AIDS testing was done in Nagaland which resented the

blood collection) results in an AIDS positivity which falls in single

digits. Meaningful results can only be interpreted when the number of

positive cases are several thousand in number. The NFHS sample size

should have been at least four times higher.

It was 'underpowered' as it under-represented high risk groups and

erred towards an underestimate. NACO did some nifty, intelligent

mathematical calibrations to adjust for weightages across states,

that resulted in the final HIV/AIDS prevalence figure of 0.36 per

cent.

Assisting NACO in producing feel-good data was the Integrated

Behavioral and Biologic Assessment (IBBA) system data. It was not the

government but UNAIDS which made this fact public.

According to UNAIDS, IBBA is a targeted surveillance system focusing

on groups at higher risk of HIV infection, located in the states with

high prevalence. IBBA data is believed to compare well with the

results of NACO's sentinel surveillance. IBBA was funded by Avahan,

the India AIDS control programme of the Bill and Melinda Gates

Foundation, Seattle.

Debates on methodology apart, the numbers are still large and

worrying. We cannot let down our vigil.

The writer is a commentator on science.

http://timesofindia.indiatimes.com/Opinion/Editorial/LEADER_ARTICLE_Do

nt_Be_Misled_On_AIDS/articleshow/2189624.cms