Theresa E. Perlis

A number of studies suggest females may be more likely to engage in injection and sex risk behavior than males. Most data on gender differences come from industrialized countries, so data are needed in developing countries to determine how well gender differences generalize to these understudied regions.

Background: Successful cross-national research requires methods that are both standardized across sites and adaptable to local conditions. We report on the development and implementation of the methodology underlying the survey component of the WHO Drug Injection Study Phase II – a multi-site study of risk behavior and HIV seroprevalence among Injecting Drug Users (IDUs). Methods: Standardized operational guidelines were developed by the Survey Coordinating Center in collaboration with the WHO Project Officer and participating site Investigators. Throughout the duration of the study, survey implementation at the local level was monitored by the Coordinating Center. Surveys were conducted in 12 different cities. Prior rapid assessment conducted in 10 cities provided insight into local context and guided survey implementation. Where possible, subjects were recruited both from drug abuse treatment centers and via street outreach. While emphasis was on IDUs, non-injectors were also recruited in cities with substantial non-injecting use of injectable drugs. A structured interview and HIV counseling/testing were administered. Results: Over 5,000 subjects were recruited. Subjects were recruited from both drug treatment and street outreach in 10 cities. Non-injectors were recruited in nine cities. Prior rapid assessment identified suitable recruitment areas, reduced drug users' distrust of survey staff, and revealed site-specific risk behaviors. Centralized survey coordination facilitated local questionnaire modification within a core structure, standardized data collection protocols, uniform database structure, and cross-site analyses. Major site-specific problems included: questionnaire translation difficulties; locating affordable HIV-testing facilities; recruitment from drug treatment due to limited/selective treatment infrastructure; access to specific sub-groups of drug users in the community, particularly females or higher income groups; security problems for users and interviewers, hostility from local drug dealers; and interference by local service providers. Conclusion: Rapid assessment proved invaluable in paving the way for the survey. Central coordination of data collection is crucial. While fully standardized methods may be a research ideal, local circumstances may require substantial adaptation of the methods to achieve meaningful local representation. Allowance for understanding of local context may increase rather than decrease the generalizability of the data.

Aims  To assess audio computer‐assisted self‐interviewing (A‐CASI) as a mode of data collection with injecting drug users (IDUs) entering two drug treatment programs in New York City. A‐CASI has been found to increase reporting of sensitive items among a variety of population subgroups.Design  A field test of A‐CASI data collection conducted within an ongoing cross‐sectional study of drug use and HIV risk behaviors among IDUs entering drug treatment. Participants were assigned without bias to either a computer‐assisted interviewer‐administered personal interview (CAPI) or to a mixed CAPI/A‐CASI interview. In the latter, ‘sensitive’ portions (dealing with stigmatized behavior) of the questionnaire were self‐administered through A‐CASI, while the remaining portions were interviewer‐administered.Setting  The Detoxification Program and the Methadone Maintenance Treatment Program (MMTP) at Beth Israel Medical Center in New York City.Participants  Seven hundred and eighty‐three IDUs entering drug treatment.Measurements  Odds ratios and adjusted odds ratios (controlling for demographic differences) for comparison of A‐CASI versus CAPI responses on 111 sensitive questions.Findings  Twenty‐three statistically significant differences (each at P < 0.05), all in the direction of more reporting of the behaviors by the A‐CASI group. Forty‐one per cent of A‐CASI participants said they would prefer any subsequent interviews to be fully A‐CASI and 46% said they would prefer the mixed CAPI/A‐CASI mode.Conclusions  A‐CASI was associated with greater reporting of potentially stigmatized drug, sex and HIV risk behaviors on a moderate number of questions. Moreover, a large majority of participants who used A‐CASI would like to be assigned to this method of data collection in future interviews.