Results: Most reports used probability samples of users and non-users and employed cross sectional
designs. Among variables considered in 5 or more reports, race, proximity to health
centers, medical consultations, sleep complaints, and health perception were virtually always
associated to drug use. Gender, mental health, and physical health status were associated in about
two-thirds of reports. Associations with age, marital status, medication coverage, socioeconomic
status, and social support were usually not observed.