A preview of this full-text is provided by American Psychological Association.
Content available from Psychology of Addictive Behaviors
This content is subject to copyright. Terms and conditions apply.
Shorter and Proximal Timeline Followback Windows Are Representative
of Longer Posttreatment Functioning
Christopher J. Gioia, Linda Carter Sobell, Mark B. Sobell, and Edward R. Simco
Nova Southeastern University
Very little research has been conducted on what time window provides a representative picture of daily
drinking. With respect to pretreatment drinking, one study that used the Timeline Followback (TLFB)
with problem drinkers found that a 3-month window is generally representative of annual pretreatment
drinking. The objective of the present study was to determine the shortest representative time window for
reports of annual posttreatment drinking. A second objective was to determine which of two time
windows, 90 days from the end of treatment or 90 days prior to the end of follow-up, was the most
representative proxy for annual posttreatment drinking. TLFB reports from 467 problem drinkers who
participated in a randomized controlled trial of a mail-based intervention were used in the present
analysis. The results show that a 3-month posttreatment window (i.e., first 90 days after the intervention)
is sufficiently representative (r⫽.94) of annual posttreatment drinking for problem drinkers (i.e., less
severely dependent alcohol abusers). In addition, although there were no clinically significant differences
in drinking behavior between the two 90-day posttreatment windows, the use of proximal windows (i.e.,
closer to the end of treatment) would minimize participant attrition. In addition, a 3-month versus
12-month TLFB follow-up time frame resulted in a much higher percentage of participants completing
the full TLFB (89% vs. 71%). Further research is needed to determine if these findings will generalize
to more severely dependent alcohol abusers.
Keywords: alcohol, drinking, Timeline Followback, short follow-up windows, alcohol treatment out-
comes, problem drinkers
The percentage of participants found for follow-up in random-
ized clinical trials is important because several studies have shown
that participants who are difficult to locate or lost to follow-up
generally have poorer outcomes and a worse premorbid status
prior to treatment (Edwards & Rollnick, 1997; Sobell, Sobell, &
Maisto, 1984). Clearly, in such cases, loss of participant data
would positively bias outcome results and limit the external va-
lidity of a study’s findings.
Although investigators have used varying time intervals (e.g., 3
to 24 months) to collect posttreatment drinking data, there is no
optimal or agreed on interval length over which posttreatment data
should be collected or about what percentage of a sample needs to
be interviewed in order for meaningful data analyses to occur
(reviewed in Hansten, Downey, Rosengren, & Donovan, 2000).
The frequency of follow-up and the length of the follow-up inter-
val are important because both could affect attrition rates, partic-
ularly with more labor-intensive outcome measures such as the
Timeline Followback (TLFB). Although there are many advan-
tages to collecting daily drinking data using the TLFB (e.g., a
variety of descriptive variables can be generated, detailed data
about the pattern of and variability in drinking is obtained) com-
pared to aggregate measures (e.g., quantity-frequency measures),
some participants at follow-up have initially refused to complete
lengthy follow-up interviews (Cunningham, Ansara, Wild, Tone-
atto, & Koski-Ja¨nnes, 1999; Miller & Del Boca, 1994; Sobell et al.,
2002). It is interesting that many of these participants, when
offered the alternative of answering a short set of questions in lieu
of a lengthy interview, provided limited information about their
drinking (Sobell et al., 2003). Clearly, based on these studies, there
is a tradeoff between the percentage of participants from whom
data are collected and the quality of data collected (TLFB vs.
aggregate measures). In this regard, using treatment outcome in-
terviews that are as short as possible, without compromising data
quality, is an important objective. Shorter versus longer time
windows also would have the advantage of decreasing memory
errors (Lemmens, Knibbe, & Tan, 1988).
One study recently compared 825 problem drinkers’ self-
reported annual drinking data using the TLFB with annual drink-
ing extrapolated from shorter drinking windows using the same
dataset (Vakili, Sobell, Sobell, Simco, & Agrawal, 2008). Results
indicated that. with aggregated data and large samples, a 3-month
time window provides a reasonable estimate of drinking during the
year preceding treatment, with little to no loss in data accuracy.
Because the Vakili et al. study only examined pretreatment data,
their conclusion cannot be extended to posttreatment time win-
dows.
This article was published Online First January 30, 2012.
Christopher J. Gioia, Linda Carter Sobell, Mark B. Sobell, and Edward
R. Simco, Center for Psychological Studies, Nova Southeastern University.
Supported, in part, by National Institute on Alcohol Abuse and Alco-
holism (Grant AA08593). The secondary data analysis described in this
article was conducted as part of Christopher Gioia’s doctoral dissertation.
Correspondence concerning this article should be addressed to Christo-
pher J. Gioia, Center for Psychological Studies, Nova Southeastern Uni-
versity, 3301 College Avenue, Fort Lauderdale, FL 33314. E-mail:
cgioia@nova.edu
Psychology of Addictive Behaviors © 2012 American Psychological Association
2012, Vol. 26, No. 4, 880– 887 0893-164X/12/$12.00 DOI: 10.1037/a0027027
880
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.