Drug and Alcohol Dependence 105 (2009) 164–167
Contents lists available at ScienceDirect
Drug and Alcohol Dependence
journal homepage: www.elsevier.com/locate/drugalcdep
Validity of the timeline followback among treatment-seeking smokers in
Susan E. Collins1, Sandra Eck, Iris Torchalla2, Martina Schröter, Anil Batra∗
Department of Psychiatry and Psychotherapy, University Hospital Tübingen, Osianderstr. 24, D-72076 Tübingen, Germany
a r t i c l e i n f o
Received 19 February 2009
Received in revised form 23 May 2009
Accepted 27 May 2009
Available online 14 July 2009
a b s t r a c t
Few studies have assessed the convergent validity of the timeline followback for smoking (TLFB-S)
among treatment-seeking smokers, and there are no studies assessing its discriminant validity. Further,
no studies to date have documented the psychometric properties of the TLFB in the German population.
The aim of this study was therefore to test the convergent and discriminant validity of the TLFB-S among
adult, treatment-seeking smokers in Germany. In the context of a smoking cessation trial, participants
(N=268) completed sociodemographic, TLFB-S and other, self-aggregated, single-item (SA) smoking
measures at pretreatment. During the 12-month posttreatment period, participants completed TLFB-S
and biochemically validated SA measures of smoking intensity and abstinence at regular intervals. Dis-
criminant validity analyses showed no significant associations for substance-use and sociodemographic
variables and measurement discrepancy between TLFB-S and SA measures of pretreatment smoking
intensity. Convergent validity analyses indicated that TLFB-S and SA measures of smoking intensity were
in high agreement at both pretreatment and posttreatment. Finally, there was high concordance between
TLFB-S and SA measures of abstinence at both 1- and 12-month posttreatment. This study replicated
and extended the current literature on the TLFB and showed that, even across cultures, it can be a valid
measure of various smoking-related variables.
© 2009 Elsevier Ireland Ltd. All rights reserved.
Valid measurement of smoking variables is vital to the
assessment of treatment and natural recovery outcomes. Rec-
ommendations in the literature (Hughes et al., 2003; Ossip-Klein
et al., 1986; Velicer et al., 1992; West et al., 2005) typically
focus on “self-aggregated” summary scores, which participants
calculate and estimate themselves at discrete timepoints (e.g.,
12-month continuous abstinence, mean smoking intensity in the
past month). Although such single-item measures are considered
to be the “gold-standard,” they assess limited aspects of smoking
behavior and fail to provide the detail necessary for more dynamic
and longitudinal evaluations of smoking behavior (Colder et al.,
2006; Piasecki et al., 2002).
is a calendar that can assess daily smoking intensity (Sobell and
∗Corresponding author. Tel.: +49 7071 29 82685; fax: +49 7071 29 5384.
E-mail addresses: email@example.com (S.E. Collins), firstname.lastname@example.org
(I. Torchalla), email@example.com (A. Batra).
Box 351629, Seattle, WA 98195, United States. Tel.: +1 206 543 4460;
fax: +1 206 685 1310.
2Iris Torchalla is now at the BC Centre of Excellence for Women’s Health,
E311 – 4500 Oak Street, Vancouver, BC V6H 3N1. Tel. 604-875-2633.
ing behavior, it provides more detail for longitudinal analyses than
tion by the researcher than self-aggregated, single-item measures.
That said, the TLFB yields more or less valid data depending on the
substance and population with which it is used (Hersh et al., 1999;
Searles et al., 2000). Thus, the psychometric properties of the TLFB
must be established for specific substances and populations to fully
realize its measurement potential.
Psychometric properties typically include assessments of relia-
bility (i.e., measurement consistency) and validity (i.e., evidence a
measure represents the underlying construct of interest with min-
imal error and bias). Two studies have shown that the TLFB-S can
validly measure aggregate smoking intensity (Gariti et al., 1998),
and daily use (Toll et al., 2005). Beco˜ na and Míguez (2006) doc-
umented concordance between posttreatment TLFB-S reports of
abstinence and carbon monoxide measures; however, this analysis
only involved a small subset (26≤n≤34) of the overall study sam-
ple. There are also no studies of discriminant validity of the TLFB-S.
Finally, although the TLFB has been used to measure alcohol use
in German populations (Arndt et al., 2002; Croissant et al., 2008),
there are presently no studies featuring the TLFB-S in a German
sample. Thus, despite its promising psychometric properties, there
are gaps in the international literature on the validity of the TLFB-S
in assessing pre- and posttreatment smoking outcome variables.
0376-8716/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved.
S.E. Collins et al. / Drug and Alcohol Dependence 105 (2009) 164–167
Baseline sample description (N=268).
Variables Descriptive statistics, %/M(SD)
Highest education level 47% Apprenticeship
30% University degree
19% Technical/associates degree
5% High school
Employment 60% Full-time
5% Stay-at-home parents/homemakers
Daily smoking intensity
23.7 (9.7) cigarettes
28.5 (9.8) years
4.7 (5.0) times
ican, Austrian, Belgian, Croatian, Greek, Italian, Polish, and Serbian. Due to rounding
error, percentages do not always add up to 100%.
The current study aimed to augment the international literature
on the validity of the TLFB-S and served as the first psychome-
tric study of the TLFB in the German population. First, we the
tested discriminant validity (i.e., lack of correlation between tar-
get variable and unrelated constructs) of TLFB-S assessments of
pretreatment smoking intensity. Next, we tested convergent valid-
ity (i.e., significant, positive correlation between the target variable
as well as posttreatment abstinence. It was expected that the TLFB-
S would evince acceptable psychometric properties, as shown in
the English-language literature.
Participants (N=268) were adult, regular smokers who had provided written,
informed consent to participate in a smoking cessation research trial conducted at a
university hospital in Germany (Batra et al., submitted for publication) (see Table 1
for sample description).
The Sociodemographic and Smoking Baseline Questionnaire (SSBQ) consists of
single items assessing sociodemographic information and self-aggregated baseline
smoking intensity; duration; quit attempts; and lifetime, regular, illegal drug use.
Sociodemographic variables were used for sample description and discriminant
validity tests. Smoking variables were used for sample description, and baseline
smoking intensity was used for convergent validity. Drug use was used for discrim-
inant validity tests.
The Fagerström Test of Nicotine Dependence (FTND; current ˛=.63) measured
participants’ pretreatment nicotine dependence for the sample description and dis-
established reliability and validity of the FTND (Heatherton et al., 1991; Schumann
et al., 2003).
The Alcohol use disorders inventory checklist (AUDIT; ˛=.75) was used in dis-
criminant validity analyses (Babor et al., 1991). The German-language version of
this questionnaire has evinced reliability and validity (Dybek et al., 2006).
Native English and German speakers adapted and translated the TLFB-S (Sobell
and Sobell, 1992) into German. The TLFB-S consists of monthly calendars for ret-
rospective evaluation of daily smoking behavior. Written instructions included
encouragement to use “best estimates” for daily smoking intensity, suggestions for
better recollection (i.e., use of daily routines, holidays, special occasions as memory
hooks), and an example of a completed TLFB-S calendar (Sobell and Sobell, 2000).
Reliability and validity have been established for up to 26-week reports of smok-
ing intensity (Brown et al., 1998; Toll et al., 2005). The TLFB-S assessed smoking at
1-month pretreatment and in 3-month segments during posttreatment to create a
SSBQ: Sociodemographic and Smoking Baseline Questionnaire; FTND: Fagerström
Test of Nicotine Dependence; AUDIT: Alcohol use disorders inventory checklist;
timeline followback-smoking (TLFB-S); FUSQ: Follow-up Smoking Questionnaire;
CO measurement: carbon monoxide measurements.
daily, 12-month record of posttreatment smoking intensity (see Fig. 1). The TLFB-S
yielded various smoking variables, including 1-month, pretreatment, mean smok-
ing intensity; 24-h posttreatment smoking intensity; point-prevalence abstinence
and continuous abstinence (CA; defined as being abstinent since treatment end and
having consistently reported CA throughout the 12 months posttreatment). These
researcher-aggregated variables were used in discriminant and convergent validity
tests with corresponding, self-aggregated variables from the SSBQ and FUSQ.
The Follow-up Smoking Questionnaire (FUSQ) consists of single items establish-
ing posttreatment smoking intensity and abstinence, and was distributed at 1-, 6-
and 12-month follow-ups. FUSQ smoking variables were used in convergent and
discriminant validity tests.
ically validate abstinence self-reports on the FUSQ. The piCO Smokerlyzer (Bedfont
Scientific Ltd., Kent, England) measured CO parts per million in exhaled air. Values
over 9ppm were considered to be evidence of smoking (West et al., 2005).
Data were collected during a smoking cessation intervention study (Batra et al.,
submitted for publication), in which two types of 6-week, group-based treatments
S.E. Collins et al. / Drug and Alcohol Dependence 105 (2009) 164–167
macotherapy) smoking cessation tailored to at-risk smokers. No overall treatment
group differences were found, and thus groups were collapsed for these analyses.
Measures were administered as shown in Fig. 1. Research assistants scanned
questionnaires for completeness. During in-person sessions, participants were
prompted to fill in skipped items; however, all participants retained the right of
refusal. When incomplete mailed questionnaires were returned, research assistants
collected the missing data over the telephone. If questionnaires were not returned
within 14 days, research assistants reminded participants via mail or telephone.
When this strategy was unsuccessful, TLFB-S and FUSQ were administered over the
telephone. If questionnaires were not returned by 1 month after the target date, par-
ticipants were classified as smoking according to intent-to-treat criteria (West et al.,
At the 12-month follow-up, one participant who reported being abstinent in
each of the 31 treatment groups was randomly selected to provide a CO measure-
ment for which he/she was paid 20 Euros. Otherwise, participants were not offered
were waived. Participants who completed all follow-up assessments were entered
into a drawing to receive two prizes of 250 Euros.
2.4. Data analysis plan
Kolmogorov–Smirnoff) were conducted to determine data distributions and
detect outliers. Due to nonnormal distributions for baseline smoking intensity,
nonparametric tests were used where possible. Convergent validity tests involving
continuous variables were conducted using bivariate Spearman correlations,
and absolute agreement was established using 2-way, mixed model intraclass
correlation coefficients (ICC). Convergent validity tests involving dichotomous
variables were conducted using kappa coefficients and percent agreement for-
mulas suggested by Cicchetti and Feinstein (1990). Discriminant validity tests
were conducted using Spearman correlations and Mann–Whitney tests to assess
associations for: (a) discrepancy (i.e., absolute difference) between self-aggregated
and TLFB-S measures of pretreatment, mean smoking intensity and (b) variables
hypothesized to be unrelated to the potential discrepancy (i.e., sociodemographic
and substance-dependence measures).
data analysesusing plots andnormalitystatistics(e.g.,
Discriminant validity analyses indicated no significant asso-
ciations for measurement discrepancy between SSBQ and TLFB
assessments of pretreatment smoking intensity and gender, age,
educational status, employment status, nationality (German vs.
other), alcohol dependence (AUDIT), level of nicotine dependence
(FTND), or lifetime, regular use of illegal drugs (n=253; ps>.11).
Convergent validity analyses indicated that TLFB-S and SSBQ
mean smoking intensity measures for 1-month pretreatment cor-
related significantly (N=263, ?=.89, p<.001), and showed a high
level agreement (N=263, ICC=.93, p<.001). Likewise, TLFB-S and
FUSQ measures of smoking intensity for the same 24-h period at 1-
month posttreatment were significantly correlated (N=179, ?=.71,
p<.001), and showed a high level of agreement (ICC=.84, p<.001).
Findings showed 94% (173/184) overall agreement for FUSQ and
TLFB PPA assessments at 1-month posttreatment (N=184, ?=.67,
p<.001), including 97% true positive (abstinence) agreement and
itive agreement, and 96% negative agreement for FUSQ and TLFB
assessments of CA at 12-month posttreatment, with an overall
agreement of 95% (251/265) (N=265, ?=.87, p<.001).
This study tested discriminant and convergent validity of the
TLFB-S in a sample of adult, treatment-seeking German smokers
participating in a smoking cessation trial. Discriminant validity
was supported in the current findings: there were no significant
associations for measurement discrepancy between TLFB-S and
self-aggregated measures of pretreatment smoking intensity and
the sociodemographic and substance-use variables hypothesized
to be unrelated to this measurement discrepancy. Findings also
supported convergent validity for TLFB-S measures of pre- and
posttreatment smoking intensity and posttreatment PPA and CA.
These items evinced strong agreement with the self-aggregated,
biochemically validated items that represent the widely accepted
The only concordance that fell below the nineties was the 70%
agreement between self-aggregated and TLFB-S 24-h PPA at 1-
month posttreatment. Because clinical observation and research
smoking than abstinence (Gariti et al., 2002), it was surprising
that abstinence agreement (97%) was notably higher than smok-
ing agreement. One explanation is that participants may shift the
occurrence of a lapse or resumed smoking by one or two days in
their retrospective recall of events. This would explain why this
month CA, which evinced very high agreement for both abstinence
and smoking reports. Despite the lower concordance for smoking,
however, 70% agreement was still relatively high. Further, because
the goal of assessing posttreatment smoking is often to ascertain
abstinence, the high agreement for biochemically validated self-
reports of PPA and CA abstinence was encouraging and perhaps
more clinically relevant for future studies.
Limitations of this study deserve mention. First, this study did
not address reliability of self-reported smoking on the TLFB-S.
Searles et al. (2000) showed that memory for daily substance-use
ies should conduct test-retest reliability assessments to determine
over time. That said, the current study showed that 12-month con-
tinuous abstinence reports ascertained from both the TLFB-S and a
single-item question validated with CO measures were highly cor-
related. Thus, there is some initial support for the 3-month data
collection intervals used in the current study.
of nonrandom selection and exposure to treatment (Kazdin, 1998),
these findings should be replicated in a randomly selected, non-
treatment sample. That said, this study established a new way of
in a treatment context, which is clinically and scientifically rele-
vant for future treatment studies. Finally, biochemical validation
was only assessed at 1- and 12-month posttreatment, and no other
measures of tobacco/nicotine exposure were utilized (e.g., urine
cotinine, plasma nicotine). Given the short half-life of CO in expired
air (Velicer et al., 1992), other biomarkers could enhance future
findings for the validity of the TLFB.
rately assessing treatment and natural recovery outcomes. This
study therefore added to the international literature in support-
ing the validity of the TLFB-S in pretreatment and posttreatment
assessments of smoking intensity and abstinence. Further, to the
authors’ knowledge, this study represents the first research on psy-
should replicate these findings, establish reliability, and regularly
incorporate TLFB-S to obtain more detailed smoking behavior data.
Role of funding source
Funding for this study was provided by grant #01 EB 0110 from
the German Federal Ministry of Education and Research and the
Ministry of Science, Research and the Arts of Baden-Württemberg
to Anil Batra. Susan E. Collins’ time was largely supported by
an NIAAA Institutional Training Grant (T32AA007455) awarded to
Mary E. Larimer at the University of Washington.
S.E. Collins et al. / Drug and Alcohol Dependence 105 (2009) 164–167 Download full-text
The NIAAA, the German Federal Ministry of Education and
Research, and the Ministry of Science, Research and the Arts of
Baden-Württemberg had no further roles in study design; in the
collection, analysis, and interpretation of the data; in the writing of
the report; or in the decision to submit the paper for publication.
Susan E. Collins, Ph.D., developed the current study idea, design
and methodology. She recruited and trained research assistants to
collect and enter the data and conducted the main statistical analy-
Sandra Eck, a research assistant on this project, helped collect
and enter these data and oversaw other research assistants work-
ing on the project in this capacity. She also conducted the literature
review and contributed valuable insights on the direction of the
manuscript. She conducted the analyses for and drafted the partic-
ipant description section. Ms. Eck critically reviewed and revised
the original manuscript.
sures and contributed insights as to their appropriate delivery. She
drafted a paragraph of the introduction addressing the use of the
measure in smoking studies. She critically reviewed and revised
multiple drafts of the current manuscript.
Martina Schröter, one of the study therapists, administered the
measures and provided insights as to their appropriate delivery
as well as on limitations of the current manuscript. She critically
reviewed and revised multiple drafts of the current manuscript.
Anil Batra, M.D., the head of the Smoking Cessation Research
and was involved in the current study conceptualization. Dr. Batra
critically reviewed and revised multiple drafts of the manuscript.
Conflict of interest
In the past, Anil Batra has accepted funding for smoking cessa-
tion research and contributions to regional advisory boards from
GlaxoSmithKline, Pfizer Consumer Health Care, Pfizer, Pharmacia
GmbH, and Sanofi Aventis. However, the current study was not
funded by any of these companies. All other authors declare that
they have no conflicts of interest.
The authors would like to thank Maria Caspary, Andrea Doebler,
M.A., Martina Epple, M.A., Evelyn Kick, M.A., Matthias Pforr, M.A.,
who has facilitated departmental support for and provided general
guidance to the Smoking Cessation Research Group.
Arndt, T., Korzec, A., Bär, M., Kropf, J., 2002. Further arguments against including
trisialo-Fe2-transferrin in carbohydrate-deficient transferrin (CDT): a study on
male alcoholics and hazardous drinkers. Med. Sci. Monit. 8, CR411-418.
Babor, T.F., Higgins-Biddle, J.C., Saunders, J.B., Monteiro, M.G., 1991. The Alcohol Use
Disorders Identification Test: Guidelines for Use in Primary Care. World Health
Organization, Geneva, Switzerland.
Batra, A., Collins, S.E., Schröter, M., Eck, S., Torchalla, I., Buchkremer, G. A cluster-
randomized trial of smoking cessation tailored to multidimensional smoker
profiles, submitted for publication.
Beco˜ na, E., Míguez, M.C., 2006. Concordance of self-reported abstinence and mea-
surement of expired air carbon monoxide in a self-help smoking cessation
treatment. Psychol. Rep. 99, 125–130.
Brown, R.A., Burgess, E., Sales, S.D., Whiteley, J.A., Evans, D.M., Miller, I.W., 1998.
Reliability and validity of a smoking Timeline Followback Interview. Psychol.
Addict. Behav. 12, 101–112.
Cicchetti, D.V., Feinstein, A.R., 1990. High agreement but low kappa. II. Resolving the
paradoxes. J. Clin. Epidemiol. 43, 551–558.
Colder, C.R., Lloyd-Richardson, E.E., Flaherty, B.P., Hedeker, D., Segawa, E., Flay, B.R.,
The Tobacco Etiology Research Network, 2006. The natural history of college
smoking: trajectories of daily smoking during the freshman year. Addict. Behav.
Croissant, B., Hupfer, K., Loeber, S., Mann, K., Zober, A., 2008. Längsschnit-
tuntersuchung alkoholauffälliger Mitarbeiter in einem Grossbetrieb nach
werksärztlicher Kurzinterventionen (Long-term follow-up on alcohol use in a
big company after brief intervention by a company physician). Nervenarzt 79,
A., Hohagen, F., Rumpf, H.-J., 2006. The reliability and validity of the Alcohol Use
Disorders Identification Test (AUDIT) in a German general practice population
sample. J. Stud. Alcohol 67, 473–481.
Gariti, P., Alterman, A.I., Ehrman, R., Mulvaney, F.D., O’Brien, C.P., 2002. Detect-
ing smoking following smoking cessation treatment. Drug Alcohol Depend. 65,
of the aggregate method of determining number of cigarettes smoked per day.
Am. J. Addict. 7, 283–287.
Heatherton, T.F., Kozlowski, L.T., Frecker, R.C., Fagerström, K.O., 1991. The Fagerström
Test for Nicotine Dependence: a revision of the Fagerström Tolerance Question-
naire. Br. J. Addict. 86, 1119–1127.
cocaine use in two groups of cocaine abusers. J. Consult. Clin. Psychol. 67, 37–42.
Measures of abstinence in clinical trials: Issues and recommendations. Nicotine
Tob. Res. 5, 13–25.
Kazdin, A.E., 1998. Research Design in Clinical Psychology. Allyn and Bacon, Boston.
Ossip-Klein, D.J., Bigelow, G., Parker, S.R., Curry, S., Hall, S., Kirkland, S., 1986. Task
Force 1: classification and assessment of smoking behavior. Health Psychol. 5,
Piasecki, T.M., Fiore, M.C., McCarthy, D.E., Baker, T.B., 2002. Have we lost our way?
The need for dynamic formulations of smoking relapse proneness. Addiction 97,
Rumpf, H.J., Hapke, U., John, U., 1995. Deutsche Übersetzung des Fagerström Test for
Nicotine Dependence.German translation of the FTND, Institut für Epidemiolo-
gie und Sozialmedizin, Greifswald.
Schumann, A., Rumpf, H.-J., Meyer, C., Hapke, U., John, U., 2003. Deutsche Version
des Fagerström Test for Nicotine Dependence (FTND-G) und des Heaviness of
Smoking Index (HSI-G). In: Glöckner-Rist, A., Rist, F., Küfner, H. (Eds.), Elektro-
nisches Handbuch zu Erhebungsinstrumenten im Suchtbereich (EHES) Version
300. Zentrum für Umfragen. Methoden und Analysen, Mannheim.
Searles, J.S., Helzer, J.E., Walter, D.E., 2000. Comparison of drinking patterns mea-
sured by daily reports and timeline followback. Psychol. Addict. Behav. 14,
Sobell, L.C., Sobell, M.B., 1992. Timeline followback: a technique for assessing self-
reported ethanol consumption. In: Allen, J., Litten, R.Z. (Eds.), Measuring Alcohol
Consumption: Psychosocial and Biological Methods. Humana Press, Totowa, NJ,
Sobell, L.C., Sobell, M.B., 2000. Instructions for filling out the Timeline Cigarette Use
Calendar. http://www.nova.edu/gsc/forms/TLFBCIGinstructions.doc (accessed
on February 2, 2009).
Toll, B.A., Cooney, N.L., McKee, S.A., O’Malley, S.S., 2005. Do daily interactive voice
response reports of smoking behavior correspond with retrospective reports?
Psychol. Addict. Behav. 19, 291–295.
Velicer, W.F., Prochaska, J.O., Rossi, J.S., Snow, M.G., 1992. Assessing outcome in
smoking cessation studies. Psychol. Bull. 111, 23–41.
trials: proposal for a common standard. Addiction 100, 299–303.