Elsevier

Addictive Behaviors

Volume 29, Issue 1, January 2004, Pages 51-60
Addictive Behaviors

A comparison of four self-report smoking cessation outcome measures

https://doi.org/10.1016/S0306-4603(03)00084-4Get rights and content

Abstract

Interventions for smoking cessation have been developed by researchers in a number of distinct disciplines. As a result, a variety of different outcome measures have been developed and employed. This paper will report an analysis that compares four smoking cessation outcome measures on data gathered from three population-based studies: (1) 24-hour point prevalence abstinence, (2) 7-day point prevalence abstinence, (3) 30-day prolonged abstinence, and (4) 6-month prolonged abstinence. The three studies provided a total of 41 mean estimates for the first three measures but only 28 mean estimates for the 6-month prolonged abstinence measure. The data demonstrate an extremely high relationship between all four measures. The first three measures (24-hour point prevalence, 7-day point prevalence, and 30-day prolonged abstinence) all correlated in excess of .98 with each other. The only measure that did not demonstrate the same degree of almost perfect equivalence was 6-month prolonged abstinence, but even here the lowest correlation with the other three measures was .82. For practical purposes, the first three measures will result in the same conclusions when used as outcome measures in smoking cessation studies.

Introduction

Interventions for smoking cessation have been developed by researchers in a number of distinct disciplines. As a result, a variety of different outcome measures have been developed and employed. The measures can be been broadly classified as self-report and biochemical. Velicer, Prochaska, Rossi, & Snow (1992) provide a summary and evaluation of these measures. This paper will focus only on the self-report measures. A number of attempts have been made to develop a consensus about a single measure that would be employed by all investigators (Hughes et al., 2003) but with little success. The decision about which measure to employ has typically relied on logical arguments rather than any empirical evaluations of the specific measures. In this paper, data from three recently published studies will be employed to empirically compare four self-report measures. All three studies involve representative samples from a general population.

Self-report measures can be classified into one of three broad classes of measures (Velicer et al., 1992): (1) point prevalence abstinence, the proportion of subjects not smoking at a point in time; (2) continuous abstinence, the proportion of people not smoking at all since the occurrence of intervention or a critical event; and (3) prolonged abstinence, the proportion abstinent for some specified interval of extended duration. Each type of measure characterizes former smokers using a different perspective and has different potential strengths and weaknesses and uses of the measures.

Point prevalence abstinence is defined as the percentage of former smokers who are not smoking at a particular point in time, typically at the time of assessment. These rates have been most commonly derived from surveys assessing the relationship between smoking cessation and health and disease (e.g., Friedman, Siegelaub, Dales, & Seltzer, 1979). They are also used in epidemiological survey research and the evaluation of interventions (e.g., Rose & Hamilton, 1978). Point prevalence measures can include a broad range of former smokers, ranging from individuals who have not smoked for years to those who have only stopped for a matter of days. The measure can result in a group that is very heterogeneous compared to groups defined by continuous or prolonged abstinence measures.

The length of abstinence is often not specified but when specified, the most common minimum periods of abstinence used for point prevalence rates are 24 hours or 7 days. The National Interagency Council on Smoking and Health (1974) recommended using a minimum of 7 days of abstinence for defining cessation. However, many studies do not specify a minimum amount of time individuals need to be abstinent to be included as former smokers. For example, in the reports of the London Civil Servants Study Rose & Hamilton, 1978, Rose et al., 1982, no indications are given as to the criterion used for defining abstinence.

Point prevalence abstinence measures have several advantages. First, point prevalence abstinence of 24 hour has the advantage of potential validation by biochemical measures. Second, point prevalence rates, if measured sometime after the event or intervention, such as 6 or 12 months later, can include smokers who take delayed action and quit. This smoking cessation measure therefore captures the dynamic process of quitting and reflects better how people change in their natural environments than does a continuous abstinence measure. Third, when used in intervention studies, point prevalence rates also allow lapses (brief returns to smoking) or relapses (extended returns to smoking) to occur following treatment without making it necessary to categorize the smoker as a permanent failure. Fourth, the immediacy of the measure avoids the problems inherent in measures that rely on recall of past events. Fifth, the question represents the natural language of the lay public so the results of a study employing point prevalence abstinence is easily communicated to nonscientists.

Point prevalence rates have several disadvantages. First, given the high rates of relapse during the first 3 months following quitting Hughes et al., 1981, Hunt et al., 1971, Hunt & Bespalec, 1974, it is predictable that many individuals who are counted as former smokers at one point in time will be current smokers at the next point in time. Second, point prevalence abstinence rates are not as stable as continuous abstinence rates since they depend on the minimum duration of abstinence used to define former smokers and the point in time that participants are assessed. Third, point prevalence rates may be difficult to interpret in relation to the health effects of smoking cessation since they often include those former smokers who have quit for only 24 hour or l week along with those who quit for much longer periods of time. Smokers who have not smoked for l or 7 days may demonstrate only the more immediate health benefits of quitting smoking while those not smoking for several years allow us to look at the longer-term consequences of cessation.

Continuous abstinence reflects the percentage of former smokers who have not smoked at all since the occurrence of an intervention or some critical event. Usually, a smoking cessation intervention is the focus of study when continuous abstinence rates are used. Naturally occurring events such as changes in health status or changes in smoking policies may be treated as an intervention.

Continuous abstinence rates have the advantage of being more stable over time and across studies than point prevalence rates. The longer the period of continuous abstinence being measured, the less the probability that relapse will occur and the more stable the cohort of former smokers will remain. Another advantage of continuous abstinence measures is that they are easier to interpret in relation to health effects of cessation.

Continuous abstinence measures also suffer from several disadvantages. First, the measure requires that the point of intervention can be clearly defined. This is reasonable in a pharmacological study but is typically impossible to define in a self-help study where the timing of the cessation attempt is under the control of the subject. Second, only a small minority of smokers actually changes in a linear manner from smoking to nonsmoking without experiencing any lapses or relapses. Cohen et al. (1989) indicate, for example, that only 4.3% of the participants shifted immediately from being current smokers to being former smokers with nonlapses. Therefore, continuous abstinence measures do not reflect how the vast majority of people change in their natural environments, making it an inappropriate descriptive tool for most cessation behavior. Third, when a continuous abstinence measure is used in an intervention study, it can only decline as more quitters lapse or relapse. While a continuous abstinence rate declines following an intervention, the percentage of current smokers becoming former smokers is likely to increase as participants take delayed action to quit or recycle back to not smoking following relapse. Fourth, continuous abstinence generally cannot be validated biochemically or by reports of other informants. Fifth, a problem for all abstinence measures, but especially for continuous abstinence, is the determination of the criterion to use to consider someone as having relapsed or no longer being abstinent. It is not clear how many lapses should be considered a relapse or how closely spaced lapses need to be to be considered a relapse.

The problems with this measure are extensive. We could not develop a meaningful way to define continuous abstinence when the point of treatment could not be unambiguously defined for the intervention studies employed in this paper.

Prolonged abstinence reflects a combination of continuous and point prevalence abstinence rates. Individuals are counted as former smokers if they have been continuously abstinent for a prolonged period of time, such as 1, 6, or 12 months. This time period does not necessarily include the initiating event or intervention. Prolonged abstinence rates are, in effect, a point prevalence rate of long duration. This rate could be a measure gathered at a distant point in time following a critical event, such as at an l8- or 24-month assessment. Prolonged abstinence rates are able to include people who make delayed or repeated quit attempts following an intervention.

Similar to continuous abstinence, a prolonged abstinence rate can be appropriate for assessing the long-term health benefits of smoking cessation. Given the well-known relapse curves in which the degree of relapse is inversely related to the time since cessation Hunt et al., 1971, Hunt & Bespalec, 1974, a prolonged abstinence rate would be more stable than a shorter point prevalence rate. As with continuous abstinence rates, there are no methods available for biochemically validating individual self-reports of not having smoked even a puff for the past 6 months, with the exception of repeated random testing during the whole time period.

This paper will report an analysis that compares four smoking cessation outcome measures on data gathered from three studies. The four measures are (1) 24-hour point prevalence abstinence, (2) 7-day point prevalence abstinence, (3) 30-day prolonged abstinence, and (4) 6-month prolonged abstinence. (It is somewhat arbitrary whether to classify the 30-day measure as prolonged abstinence or point prevalence.) Since the focus of this paper is on outcome measures, the aggregated outcome measures typically reported in outcome studies will be employed as the data for the study. The three studies were selected because they were readily available, they report all four measures for all assessment occasions, they are population-based, and they had relatively large sample sizes. Many smoking cessation studies have relied on volunteer samples. Population-based studies permit the accurate estimation of the relation between the four variables without the potential biasing effects of nonrepresentative samples.

Section snippets

Method

The data for this analysis were taken from three published studies Prochaska et al., 2001, Prochaska et al., 2001, Velicer et al., 1999. The studies were selected because they were both population-based smoking cessation studies and all four outcome measures were reported. The basic data were the mean of each measure for each group on each occasion. (The mean of a dichotomous variable is also the proportion not smoking. All input data could be also described as proportions.) The three studies

Results

Table 1 presents the means and the correlations between the four measures. All means and correlations were based on the total available data for that pair of variables. The highest correlation was found between 24-h point prevalence abstinence and 7-day point prevalence abstinence (r=.99) and the lowest correlation was found between 24-h point prevalence and 6-month prolonged abstinence (r=.82). An examination of the bivariate scatterplots showed that all relationships were linear.

In addition

Discussion

The data demonstrate an extremely high relationship between all four measures. The first three measures (24-h point prevalence, 7-day point prevalence, and 30-day prolonged abstinence) all correlated at .98 and above with each other. This extremely high relationship was also reflected in the measurement model, where the paths for all three measures approached one. The only measure that did not demonstrate the same degree of almost perfect equivalence was 6-month prolonged abstinence, but even

Acknowledgements

Grants CA27821, CA63045, CA71356, and CA50087 from the National Cancer Institute supported this work.

References (15)

There are more references available in the full text version of this article.

Cited by (113)

  • Specialized tobacco quitline and basic needs navigation interventions to increase cessation among low income smokers: Study protocol for a randomized controlled trial

    2019, Contemporary Clinical Trials
    Citation Excerpt :

    The primary study outcomes are cessation-related. Participants are asked if they smoked at all in the past 7 days [57] and past 30 days to assess point prevalence abstinence [58]. Those who report any smoking in the past 7 days are asked if they quit for at least 24 h since the previous survey.

View all citing articles on Scopus

Further information can be obtained from the following Web site: http://www.uri.edu/research/cprc.

View full text