Background In Latin America (LA), men who have sex with men (MSM) are the population most affected by HIV, crack/cocaine production and trafficking routes are well-established, and local consumption has risen sharply in recent years. Yet, drug use and associations with sexual risk behaviours among MSM remain largely unexamined LA despite being recognised as important drivers of HIV transmission in the USA. Variation in crack/cocaine consumption within LA is also unknown.

Methods Data from participants self-identifying as MSM and aged 18 years and older in recent (2006−2009) respondent-driven sampling (RDS) studies using similar survey methodologies in nine high-population cities in Argentina, Bolivia, Costa Rica, El Salvador and Honduras (N=2081) were pooled and analysed in a multilevel statistical framework to increase statistical power to evaluate past 12-month crack/cocaine consumption (CC). CC prevalence was estimated by study site, adjusting for sampling weights incorporating personal network and population sizes. Logistic, multilevel models identified correlates of CC, adjusted for intraclass correlation within levels defined by study site and recruitment chain, and partitioned variance in CC among levels.

Results CC prevalence was 21.7% (95% CI 17.4 to 26.7) in the pooled sample and varied significantly across cities from 0.1% in Cochabamba, Bolivia to 44.5% in Ceiba, Honduras. In multivariate analysis, daily alcohol consumption (adjusted OR [AOR]=4.0), sexual debut of ≤ 12 (AOR=3.1) and 13−17 (AOR=1.9) years old, self-identifying as transexual/transgender/transvestite (TTT) (AOR=2.2), age 18−22 (vs >40) years (AOR=2.0), having a stable female partner (AOR=1.8) and ever purchasing sex (AOR=1.6) were associated with CC (p<0.05). Significant city-level random effects expressed as AOR were 5.4 (SE=1.3) and 2.7 (SE=1.3) in Ceiba and San Pedro Sula, Honduras, respectively and 3.4 (SE=1.2) in Buenos Aires, Argentina. Of the total variance in CC, 29.3% was associated with unobserved differences among cities. Associations between CC and condom use at last sex, and recruitment chain, were non-significant (p>0.1).

Conclusions CC prevalence among MSM, and variation among cities in LA, are considerable. Associations with frequent drinking and sexual risk Behaviours likely compound HIV transmission risk among MSM and to female partners. Substance abuse prevention and treatment for MSM are needed, particularly among TTT, and in Honduras and Argentina.