Changes in Substance Use-Related Health Risk Behaviors on the Timeline Follow-Back Interview as a Function of Length of Recall Period

Abstract

The timeline follow-back (TLFB) interview was adopted to collect retrospective data on daily substance use and violence from 598 youth seeking care in an urban Emergency Department in Flint, Michigan during 2009–2011. Generalized linear mixed models with flexible smooth functions of time were employed to characterize the change in risk behaviors as a function of the length of recall period. Our results suggest that the 1-week recall period may be more effective for capturing atypical or variable patterns of risk behaviors, whereas a recall period longer than 2 weeks may result in a more stable estimation of a typical pattern.

Assessment methods and schedules for collecting daily process data on substance use related health behaviors: A randomized control study

“…Remaining participants were sent a “welcome back postcard” and contacted (e.g., phone, home visit, social media) during the recruitment period. After providing consent for the daily process study, 331 participants self-administered a 30-minute computerized assessment including demographic information and conventional measures of substance use related risk behaviors/problems in past six months, followed by a 20–30 minute staff-administered timeline follow-back interview, which used a calendar and landmark events to facilitate participants’ recall of substance use related behaviors for each day in the past 90 days (see Buu et al, 2014). Participants received $20 cash for completing the baseline assessment and also offered the options to participate in urine drug screening ($5 cash) and HIV testing ($5 cash).…”

Statistical methods for evaluating the correlation between timeline follow-back data and daily process data with applications to research on alcohol and marijuana use

“…Specifically, we examine whether the correlation between these two types of data varies across substances (alcohol vs. marijuana), assessment schedules (daily vs. weekly), and recall windows . The first research hypothesis that the correlation between daily process and TLFB data would be higher for marijuana consumption than for alcohol consumption is based on previous research, which indicated that alcohol use behaviors are more likely to vary across days during a week than marijuana use behaviors (Buu et al, 2014). In fact, the empirical data collected from the M&M Study support this point, as alcohol use is more likely to occur on weekend days, whereas the likelihood of marijuana use does not change across days during a week.…”

“…Although the way TLFB is administered aims to encourage the access of episodic memory (defined as the retrieval of information about specific episodes of a behavior), participants may tend to switch to sematic memory (referring to generalization about behavior that is stored in memory) as the reported date goes back in time. Such a switch may already occur sometime during the first 30-day recall window (Buu et al, 2014) and thus results in no consistent change across the three recall windows.…”

“…A unique strength of the method proposed by Li and Chow (2005) for estimating ρ and ρ c in paired functional data is that the weight function w(•) allows researchers to assign more weight on the data collected during critical time points (given prior information). This is highly relevant to substance use related data, as it is well known that the consumption level tends to be higher on weekend days (Buu et al, 2014). When no prior information is available, however, common practices are to adopt equal weights across time or choose the weight function based on the patterns of empirical data.…”

Collecting Self-Reported Data on Dating Abuse Perpetration From a Sample of Primarily Black and Hispanic, Urban-Residing, Young Adults: A Comparison of Timeline Followback Interview and Interactive Voice Response Methods

“…The most informative study about the accuracy of the TLFB for the purpose of collecting any type of violence-related data from youth was conducted by Buu and colleagues (2014), who found that the length of the TLFB recall period is an important consideration when interpreting results. Specifically, Buu et al’s findings suggest that when the TLFB is used for collecting retrospective peer violence–related data from youth, the recall period of the past week may yield more reliable information about atypical or nonpatterned violent behavior (e.g., once on Tuesday morning and once on Friday night), whereas the recall period of 2 weeks or more may result in acceptable estimates of chronic, repeated patterns of violent behavior (e.g., every Saturday night for the past 6 weeks).…”