No room for complacency
While the new estimates for the HIV-infected have brought a sense of relief, it is important to understand where the numbers add up and where they didn’t, writes Sujatha Rao.india Updated: Jul 20, 2007 00:31 IST
The recent estimates published by the National Aids Control Organisation (Naco) of a much lower level of prevalence of HIV infections in the country than previously believed generated relief as well as disbelief. It might help if we review what made the prevalence levels plummet so drastically.
For efficient programme implementation, two sets of information are critical — information about the number of people, as a proportion to the population, who are already infected (prevalence), and the number of persons getting added every year (incidence). In countries with universal access to healthcare and where reporting systems are streamlined, estimations in actual numbers are possible. In India, we have scanty information of prevalence and hardly any of incidence of any public health programmes — be it malaria, maternal morbidity, cardiac problems, hypertension or diabetes. In countries where data systems are weak, as in India, estimations are made by making assumptions on whatever data is collected and extrapolating this to the entire population.
In the past, prevalence estimates for HIV-infected were based on 400 unlinked, anonymous samples of blood collected from sentinel sites, which were 155 in 1999, and increased to 1,122 in 2006. The sentinel sites were located in antenatal clinics in medical colleges, district hospitals and some in clinics frequented by high-risk groups. The samples that tested positive would be sent to a national reference laboratory for confirmation. These results would then be extrapolated to the population groups. The prevalence level among antenatal groups was taken as ‘proxy’ for the general population. The objective of sentinel sites is to stand guard and inform of the direction the virus is taking — where, among whom, and how. However, in the absence of any other source of data, the prevalence figure has always been taken to arrive at the actual number of persons infected. Thus 0.9 per cent of the population prevalence level was translated into 5.2 million people of the adult population between 15 and 49 years.
Under the National Family Health Survey (NFHS) III, the HIV factor was also added. Blood samples of about 110,000 persons in the age group 15 to 54 years of the sample households were collected. Since 70 per cent of the infections are in six high-prevalence states, nearly 75 per cent of the samples were collected from here and Uttar Pradesh. This data showed a prevalence level of 0.28 per cent.
There are strengths and weaknesses of both systems of survey. The sentinel surveys were inadequate as, given the health-seeking behaviour pattern, the survey missed out on those visiting private clinics. Also, in the case of the poor and sick attending government facilities, there was potential for double counting of persons with STDs who could also be clients of sex workers. Finally, antenatal mothers were not typical groups as they were young and sexually active and not using any barrier method such as condoms. The NFHS had similar shortcomings — the sample was too small; it excluded the high risk groups since ordinary households (the population not infected) did not have sex workers, or prison inmates at home.
These methodological imperfections are not peculiar to India. Since a country had to have one estimate, external experts from UNAids, WHO, Atlanta’s CDC and Imperial College, London, with experience of such estimation processes, were invited. In addition to the two surveys, we also had data from Voluntary Counselling and Testing Centres.
New tools of analysis developed recently to arrive at estimations were introduced. Essentially, this meant factoring for the groups that could have been left out by the NFHS but were most critical to HIV infections. To make the data comparable, similar methods of calculation were employed to the previous five years. Accordingly, we have today new estimates for the last 5 years, which show a stationary level of 0.4 in 2002 as against 0.38 in 2006.
There has been no decline in the prevalence rate — for adult population, the range is between 2 to 3.1 million HIV-infected. This is far more reliable than the earlier UNAids estimate of between 2 and 9 million. However, there is no finality to these numbers either. New data may change them again. Even if not, there will be a steady increase in prevalence as with treatment, more HIV-infected will live longer; and due to prevention efforts, incidence of new cases will get reduced.
Sujatha Rao is Director General, Naco