At the age 32 assessment, we calculated prevalence using data from all participants who had their sera tested. We calculated incidence rates (per 1000 person?years) for each age period, including only those who tested negative for HSV?2 at the previous assessment and had experienced coitus, taking into account years since first coitus. For those of whom age at first heterosexual intercourse was unavailable, we estimated age of first coitus on the basis of the first assessment at which sexual contact was acknowledged. Those who had had coitus but reported no such partners in a subsequent time period were included in the analysis, as HSV?2 may be transmitted through other sexual activity, and information on opposite sex partners without intercourse was not collected.
We calculated incidence rates and 95% confidence intervals (95% CI), associated incidence rate ratios and 95% CI, using robust Poisson regression in Stata V.8, including years since the first coitus as an exposure variable. For analyses that compared rates across assessments, we clustered the study members by their study identification number to account for repeated measures. To assess whether rates across assessments differed by sex, the model was refitted, including an interaction term for sex and age. To test for linear trend, analyses were repeated using number of sexual partners as a continuous variable.
At age 32 years, we tested sera from 884 participants (432 women and 452 men, 87.1% of the surviving cohort) for HSV?2 antibodies. The overall prevalence of HSV?2 infection was 18.4%, higher among women (22.5%) than among men (14.6%; p = 0.003). As reported previously, the prevalence at age 21 years was 4.3% and 2.7% for women and men, respectively, and at age 26 years, 15.3% and 7.1% (fig 1 ? 1 ). 13 ,14
Table 1 ? 1 shows the incidence rates for the two age periods; p values resulting from the comparison of these rates are given in the text. For the period from first coitus to age 26 years, the incidence rate for men (6.5 per 1000 person?years) was less than half that for women (14.3 per 1000 person?years; p = 0.001). For the age period 26–32 years, the rates were similar for men and women: 14.3 and 15.8 per 1000 person?years, respectively (p = 0.69). The incidence rates increased between these age periods for men (p = 0.002) but not for women (p = 0.64). Among the women, there was a linear trend for the incidence to increase with number of partners both up to age 26 years (p = 0.005), and from age 26 to 32 years (p = 0.001). For the men, this relationship was not so clear (p = 0.11 up to age 26 years, and p = 0.13 from age 26–32 years). Within each stratum of the number of sexual partners (except women with no or one partner), the incidence was higher from age 26–32 years. Men with same?sex contact from age 26–32 years were more likely to become infected (p = 0.035), but this relationship was not found when younger, or for women.
Table 2 ? 2 shows the comparisons of incidence by age period and sexparing age periods, the unadjusted incidence rate ratio was 2.2 (95% CI 1.3 to 3.6) for men, but only 1.1 (95% CI 0.73 to 1.7) for women; the test of interaction indicated that men and women had different patterns of incidence over time (p = 0.039). When adjusted for number of sexual partners and same?sex contact, the incidence rate ratio increased for men to 2.9 (95% CI 1.7 to 4.9) and for women to 2.0 (95% CI 1.2 to 3.4). The marked change, particularly for the women, is because the pattern of sexual partnerships is very different in the two age periodsparing incidence by sex in the two age periods, the incidence rate ratio was 2.2 (95% CI 1.4 to 3.4) for women compared with men up to age 26 years, but was only 1.1 (95% CI 0.69 to 1.8) for the age period 26–32 years. When adjusted for number of sexual partners and same?sex contact, the incidence rate ratio increased for women compared with men to 2.5 (95% CI 1.6 to 4.0) up to age 26 years, but only to 1.3 (95% CI 0.78 to 2.2) for the age period 26–32 years.