HIV diagnosis leads to significant reduction in risky behaviour in African sex workers

A large, prospective study of nearly 1,600 female sex workers in Mombasa, Kenya has found that the women reduced the amount of unprotected sex they had by 44% after being diagnosed with HIV.

This reduction was independent of age, time spent in the study or the use of other contraceptive methods, but it was affected by the state of health: it was only statistically significant in women with CD4 cell counts under 500 cells/mm³.

The study recruited 1,597 initially HIV-negative female sex workers from 1993 and finished follow-up in March 2004 when antiretrovirals (ARVs) became available to those in need of treatment under WHO guidelines. The researchers explain that they did this because they wanted to establish the influence of HIV diagnosis on behaviour in a vulnerable group as a baseline before they did additional studies on the effect of antiretroviral therapy on behaviour.

During the course of the study the women paid monthly visits to a clinic where they were tested for HIV, completed a questionnaire about their sexual behaviour over the previous week, and had a pelvic examination. From April 1998 those who became HIV-positive also had CD4 cell counts checked.

During the course of the study 265 (one in six) of the women seroconverted, yielding an HIV incidence rate of 7.7% a year. The women who seroconverted contributed a median of 481 days of follow-up pre-diagnosis and 1,392 days post-diagnosis, meaning that women were followed for an average of five years.

At enrolment, the women had an average age of 26 and nearly all had education until the age of 14.

The researchers defined ‘unsafe sex’ during the study as the person having had at least one episode of sex without a condom in the week before their visit. Of note is that though the visits were monthly, the women were defined as ‘abstinent’ if they had not had sex during the previous week, so clearly some sexual behaviour, unsafe or otherwise, was not captured by the study and it may have underestimated sexual frequency because of this definition.

Nonetheless, the observation that women reduced their risk behaviour post-diagnosis seems robust. Overall 27% of occasions of sexual intercourse were unsafe pre-diagnosis and 15% afterwards, a reduction of 44%. Some of this reduction was influenced by factors such as increasing age, so when confounders were accounted for, the adjusted odds ratio for unsafe sex post-diagnosis was 0.69: in other words diagnosis accounted for about two-thirds of the reduction in observed unsafe sex.

Concomitant with a reduction in unsafe sex, abstinence and condom use increased. The proportion of women who were abstinent in the previous week increased from 35% to 54% (a more than 50% increase, adjusted odd ratio 1.38) and the proportion of women reporting 100% condom use increased from 59% to 67%.

The number of both sexual partners and sexual encounters also went down. The proportion reporting more than one sex partner in the previous week went down from 20% to 9% (a one-third decline when adjusted) and the proportion reporting more than two sexual encounters in the previous week went down from 27% to 16%. These figures make one wonder if ‘sex worker’ was an appropriate description for the women involved in the study, certainly post-diagnosis, even if there was a transactional element to the sex they had.

CD4 cell count was related to the reduction in unsafe sex in a strongly ‘dose-related’ fashion. The odds ratio for unsafe sex was only 0.93 or a 7% reduction in women with CD4 cell counts over 500 cells/mm³, which was not statistically significant after adjusting for confounders. The only variable that was significant in women with CD4 cell count over 500 cells/mm³ was the number of sexual partners, where even after adjustment there was a 58% reduction in the number reporting more than one sex partner a week.

However the reduction in unsafe sex was statistically significant in women with CD4 counts between 200 and 500 cells/mm³ (42% reduction after adjusting) and under 200 cells/mm³ (a 55% reduction, though clearly in this case sex would be being affected by physical illness in many of the women.)

The lengthy and prospective nature of the study and the short interval between questionnaires make the researchers confident that they have, for the first time in a resource-limited world study, documented a real, though not dramatic, decline in sexual risk behaviour following HIV diagnosis.