A cross-sectional study of a convenience sample of 107 gay  men was conducted. The study ascertained sexual bahavior history along with syphillis/HIV serostatus. Of the 107 men, 33 were HIV seropisitive and 30 VDRL(syphillis) reactive. Of the 33 who were HIV seropositive,20 were also syphillis reactive. Stage II The HIV postive men were monitored prospectively for 3 years. They were all at almost  similar stage, and received AZT therapy  followed by Protease Inhibitors. After the three year period, 15 of the 20  who  were both HIV & Syphillis positive had  died compared to  3 who were HIV positive only. With the excepton of sexual behavior history, no other information is available. ANY IDEAS FOR ANALYSIS AND INTERPRETATION

 < snip, detail : ...  After the three year period, 15 of the 20  who  were both HIV & : Syphillis positive had  died compared to  3 who were HIV positive only. : With the excepton of sexual behavior history, no other information is : available. : ANY IDEAS FOR ANALYSIS AND INTERPRETATION  - In the early days, there was the suspicion that every infection your immune system had to deal with  might speed the progression of HIV to AIDS. There was also a suggestion that syphillis was related to AIDS, not least due to the prevalence of Kaposi's carsonoma [sp?]. I'm not certain if this was pre-HIV or not, but a post-HIV you might say that the theory was that AIDS help bring out late-stage syphillis symptoms early. In part, there was concern that the [purported] US practice of one-shot treatment of VD, rather than the european VD-clinic with follow-up, was leaving some people relatively symptom-free but uncured. If they than went on to catch-HIV/develop-AIDS they would have a relatively poor prognosis. Saw it in the [UK] New Scientist many years ago. Think one of the people talking about the idea was in a London hospital or school of medicine. Peter

Since this is an observational study, one should certainly investigate the data on sexual behaviour history to see if there were any factors that were associated with either condition or with the observed outcome.   You said that the HIV positive subjects were all at almost  similar stage and that they received the same kind of treatment (AZT plus Protease Inhibitors.) Are more precise measurements  available on either the stage or on the treatments?  If so, could small difference in stages  (or treatments) explain any part of the observed association? And since this is a convenience sample, one might wonder whether/how the method of obtaining the subjects might explain the observed association. Having examined all the factors measured in this observational study, and having considered what factors might relate to the subjects who participated in this convenience sample, one should consider what unmeasured factors seem most likely to be potential confounders of the observed association between the death rate and the presence or absence of syphillis among gay men who are HIV positive. .