Background Needle-exchange programmes (NEPs) are potentially a key str
ategy for containing the spread of HIV infection among injecting drug
users, but their implementation has been limited by uncertainty about
their effectiveness. We used an ecological study design to compare cha
nges over time in HIV seroprevalence in injecting drug users worldwide
, for cities with and without NEPs. Methods Published reports of HIV s
eroprevalence in injecting drug users were identified, and unpublished
information on HIV seroprevalence for injecting drug users entering d
rug treatment in the USA between 1988 and 1993 was obtained from the C
enters for Disease Control and Prevention. Details of the implementati
on of NEPs were obtained from published reports and experts. For each
of the 81 cities with HIV seroprevalence data from more than 1 year an
d NEP implementation details, the rate of change of seroprevalence was
estimated by regression analysis. The average difference in this rate
for cities with and without NEPs was calculated. Findings On average,
seroprevalence increased by 5.9% per year in the 52 cities without NE
Ps, and decreased by 5.8% per year in the 29 cities with NEPs. The ave
rage annual change in seroprevalence was 11% lower in cities with NEPs
(95% CI - 17.6 to -3.9, p = 0.004). Interpretation A plausible explan
ation for this difference is that NEPs led to a reduction in HIV incid
ence among injecting drug users. Despite the possibility of confoundin
g, our results, together with the clear theoretical mechanisms by whic
h NEPs could reduce HIV incidence, strongly support the view that NEPs
are effective.