Correspondence between Interactive Voice Response and Time-Line Follow-Back Self-Reports of Alcohol Consumption among Problem Drinking Men-who-have-Sex-with-Men

Surveillance data has indicated that men-who-have-sex-with-men consume alcohol at elevated rates. Several data collection methods have been developed to enhance response accuracy specific to alcohol use behavior. A commonly used method is the Time-Line Follow-Back (TLFB) interview, a calendar-based technique which respondents report on their alcohol use each day during a specific recall period. More recently, Interactive Voice Response (IVR) technology, whereby participants respond to questions via a telephone daily, has become increasingly popular, and may provide a more real-time assessment of alcohol use. The present study examined the degree of correspondence between IVR and TLFB self-reports of alcohol use among problem-drinking MSM. Participants (N = 84) were part of an on-going clinical trial to reduce their alcohol consumption. Upon randomization, participants called into the IVR system daily for 90-days during treatment engagement. At the end of treatment involvement, the TLFB was administered which covered the same 90-day period. Results indicated moderate-to-high intraclass correlation coefficients (ICCs) between both methods for mean drinking per drinking day (MDDD; r = .41, p < .01), proportion of heavy drinking days (PHDD; r = .57, p < .01) and proportion of days abstinent (PDA; r = .85, p < .01). Paired sample t-tests revealed that reports generated from the IVR were greater for MDDD and PHDD (p's < .05), but not different for PDA (p > .05). Study findings indicate that both methods have a high degree of correspondence with one another with a tendency to report greater alcohol use on the IVR.