Objective: To measure trends in the incidence of HIV-1 infection among
drug users in treatment at Thailand's largest drug detoxification uni
t. Design: A retrospective cohort was established using computed, exis
ting HIV-1 test results of 26 396 inpatients and outpatients admitted
for 47 907 drug detoxification treatment courses from August 1987 to A
ugust 1992. Methods: Matching of patient record numbers showed that 10
050 (38.1%) patients had been admitted two or more times during the p
eriod. From these, we selected a cohort of 7807 initially HIV-negative
patients. Subsequent seroconversions among them were assumed to have
occurred with uniform probability throughout the interval between the
last HIV-negative and the first HIV-positive tests. Results: There wer
e 2311 (29.6%) seroconversions in the cohort. HIV-1 incidence among th
e 5974 (76.5%) who were injecting drug users (IDU) escalated from 20 n
ew infections per 100 person-years (PY) of observation in 1987 to a pe
ak of 57 per 100 PY in 1988, then gradually declining to a stable tate
of about 11 per 100 PY during 1991 and 1999. Non-IDU (smokers, inhale
rs) constituted 683 (8.8%) of the cohort patients, and had HIV-1 incid
ence rates varying from 0.2 to five per 100 PY. 'Mixed' drug users, de
fined as individuals reporting different routes of drug administration
on different admissions, composed 1150 (14.7%) of cohort patients and
had an HIV-1 incidence rate between that of IDU and non-IDU. Prevalen
ce of HIV-1 seropositivity among ail IDU increased rapidly, from about
1% in early 1988 to a peak or about 40% by early 1989, and has remain
ed stable through 1992. Conclusions: Prevention efforts must continue
for IDU, since recent annual HIV-1 incidence remains high at >10 per 1
00 PY. Such a high rate suggests that this group should be considered
for HIV-1 vaccine efficacy trials. Stable HIV-1 prevalence can mask su
bstantial incidence in a population with high turnover.