Sources of information

Thailand established a sentinel surveillance system for HIV in 1987, early in the epidemic. Beginning in 14 provinces and rapidly expanded across the country, the system was able to determine levels of infection in various vulnerable and general population groups: sex workers, male STD patients and IDUs as well as pregnant women and blood donors.

Another important source of epidemiological information is the anonymous-unlinked testing of 60,000 21 -year-old men chosen each year as military recruits. Selection of recruits is by lottery, so the group is relatively representative of young men across the country.

Data about other STDs also offers useful corroborating information about HIV, since these diseases are largely transmitted by sexual activity. If people are curtailing or modifying the behaviours that expose them to HIV - principally unprotected sex with partners whose sexual history may include risk behaviour - we would expect to see a drop in new cases of other STDs as well as of HIV.

Data on HIV gathered through cross-sectional monitoring - that is, testing of a certain population at single points in time - can be validated with longitudinal cohort studies that follow individuals over time to determine how many become infected within a certain period. In Thailand, such studies are available for a number of different groups, including repeat blood donors.

What these sources show

· Overall levels of infection

Sentinel surveillance and other sources of cross-sectional information give a picture of HIV prevalence - the proportion of people testing positive at any one time. Because HIV is an incurable chronic disease, infected people may stay in the tested population for long periods of time. That means that prevalence figures will include people who were infected as many as 10 years ago, and therefore will be slow to reflect changes in incidence, or new infections.

Figure 1 shows HIV prevalence in various groups followed in sentinel surveillance. HIV infection in female sex workers in brothels has climbed steadily until by 1995 nearly one-third of brothel workers tested were HIV positive. Sex workers working "indirectly" from restaurants, bars, karaoke lounges, etc., have seen infection level off at around 12%, about the same rate as for male STD patients, who generally report that they are clients of sex workers. In blood donors and pregnant women, rates have remained relatively low, but even so, over 2% of pregnant women nationally were testing positive in mid- 1995.

Infection rates remain largely unchanged in IDUs, over one-third of whom were already infected when sentinel surveillance began. When a prevalence rate is stable it does not mean there are no new infections, but rather that there is one new infection for every person dying or dropping out of the testing group. Since the epidemic among IDUs is a decade old, it is reasonable to assume that many of those infected early in the epidemic are now dying. The stable prevalence figures may, therefore, conceal considerable levels of new infection, and indeed studies over the years have found 5% to 10% of IDUs become infected with HIV each year.

Figure 1. National HIV prevalence from Thailand's sentinel surveillance system

Source: Epidemiology Division. Ministry of Public Health. Thailand

· First signs of a decline in HIV

Although the data gathered from annual testing of military conscripts are also cross-sectional, they tell us rather more than ordinary surveillance data do. A different group of 21 -year-olds is tested each year; from this we can easily determine trends in prevalence in that age group.

The average age at first sex for young Thai men is 18, so it is reasonable to assume that many of the conscripts have been sexually active for only a few years. In this young group, HIV prevalence figures will therefore reflect sexual behaviour in the few years before the testing date, and changes over time are likely to reflect changes in risk behaviour in recent years.

Figure 2 shows that nationally fewer 21-year-old conscripts were infected in 1995 than in the peak years of 1992-93, and that this was especially true of the heavily-infected northern region.

Figure 2. Mean national and regional HIV infection levels in military conscripts

Source: Jugsudee et al., 1996. Royal Thai Army

Since other sexually transmitted diseases are curable, reported cases of new STD are a more reliable indicator of sexual risk behaviour in the recent past than HIV prevalence data, which may reflect risk behaviour of a decade or more in the past. In Thailand, STD prevalence data are available only for people who seek treatment at government clinics; the sizeable private sector client base is not included. While the data shown in Figure 3 are therefore incomplete, they suggest strongly that unprotected sex with high-risk partners is on the decline.

Figure 3. Number of STDs reported from government clinics by gender

Source: VD Division, Ministry of Public Health

· New infections declining

The most convincing evidence of recent changes in levels of infection comes from longitudinal cohort studies that measure the rate of new infections, or incidence. Such studies follow people over time, calculating how many people become infected for each year (or month) they are followed. Incidence rates are normally expressed in terms of the number of new infections per 100 "person years" or 100 "person months" of follow-up. Among conscripts in the north, new HIV infections fell from 3.2 per 100 person years in 1991 to just 0.3 per 100 person years in 1995. Other STDs also fell in this group, by a margin similar to that reported by government clinics as seen in Figure 3. This finding suggests that the drop in STD rates seen at government clinics is real and not a result of people's seeking treatment elsewhere.

Among sex workers, HIV seroconversion rates were far higher, although several studies have shown that new infections among sex workers are also declining. In one rural area in the north, the rate of new infections rose from 12 per 100 person months in 1989 and 1990 to 17 in 1991, before falling back to 9 per 100 person months in 1992 and 1993. Note that incidence rates for sex workers were often given in person months because new infections, especially during the first years of the Thai epidemic, were occurring rapidly. These continuing infections show the importance of consistent condom use. While overall 90% of sexual contacts may be protected, those sex workers who do not use condoms with all of their clients and other sexual partners may still quickly become infected with HIV.

Encouragingly, new STD infections among sex workers have shown an even sharper drop than HIV, and are again in line with government clinic data.

Figure 1 showed significant differences in overall levels of HIV infection between brothel-based and indirect sex workers. These are confirmed by large differences in recorded new infection in these two groups - 29 new infections per 100 person years of follow-up for brothel workers in one northern study in 1993, against 0.9 new infections among indirect sex workers.

Information for other groups is less readily available - HIV incidence among male sex workers in gay bars in the north was shown to be consistently high at around 12 per 100 person years between 1989 and 1994. However this may not be generalized to all men who have sex with men because gay bar workers are at elevated risk of HIV infection and also are predominantly heterosexual in Thailand, often visiting female sex workers. Repeat testing of individual blood donors has shown a drop from 1.7 infections per 100 person years between 1989 and 1990 to 0.5 by 1994. This may, however, reflect a growing reluctance to give blood among people whose behaviour puts them at risk of HIV infection.