Factors affecting the validity of a Timeline Follow-Back interview.

OBJECTIVE: The Timeline Follow-Back (TLFB) interview is a calendar-prompted, retrospective measure of alcohol consumption. This report examines limitations of the TLFB's validity by examining change in reported consumption going back in time.
METHOD: This report analyzes data from a case-control study. Cases (N = 2,517; 56.9% men) were patients presenting for care of an acute injury to one of three emergency departments in Boone County, MO. Two control groups were recruited. Community controls (N = 1,856; 51.1% men) were recruited by random-digit dialing and interviewed by telephone (response rate, 46.5%); medical patients (N = 2,103; 50.9% men) presenting for care of a noninjury illness were interviewed in person and, a few months later, by telephone (complete data obtained on 2,082). A 28-day TLFB interview was conducted with cases and community controls and an 8-day TLFB was done twice with medical controls.
RESULTS: A linear regression analysis was done on each individual's drinking over the 28 or 8 days. Averaging participants' regression slopes, cases as a group showed a significant decay in self-reported consumption (0.011 drink per day for each day going back in time). Among community controls, the decay was significantly greater (0.018 drink per day). Analyzing only the 8 days prior to the day of interview, medical controls showed more evidence of bias than either cases or community controls.
CONCLUSIONS: The smaller decay in cases' reporting is consistent with an effect of motivation (e.g., engagement in the interview process). Whether the interview is conducted in person (cases and first interviews with medical controls) or by telephone (community controls and second interviews with medical controls) may be relatively less important.