Classification of non-respondents in a population-based tobacco cessation contest—"Quit and Win"

Abstract

In most questionnaire studies there are some subjects who do not respond, and the response rate is viewed as an indicator representative of the sample population (1). The critical issue is how the results should be interpreted because of non-response bias. In a comprehensive review of characteristics of respondents and non-respondents in medical surveys, it has been found that current smoking is associated with non-response (2), a connection that is recurrent in different types of surveys (3-6). There are recommendations that non-respondents should be classified as smokers (7). A great number of smoking cessation methods have been developed over the years (8). Abrams has described various types by their level of intervention: minimum, moderate, and intensive intervention (9). Minimum intervention comprises self-help material and other types of support. Moderate intervention comprises self-help material with access to brief advice and support from, for example, healthcare personnel. Intensive intervention comprises treatment at smoking cessation clinics with specially trained personnel.One example of a minimum intervention method for aiding tobacco-users to quit is the &dquo;Quit and Win&dquo;contest, a method based on a number of different theories and models, such as self-efficacy, locus of control, stages-of-change, and persuasion (10). The contest was first introduced in the Nordic countries, in 1985 in Finland and in 1988 in Sweden. During the 1990s the method has been diffused rapidly and applied worldwide in several countries by the WHO's Cindy Program (11).In this study, which is based on a one-year followup of participants from a national &dquo;Quit and Win&dquo; contest in Sweden in 1994 (12), we studied the bias in smoking prevalence because of the non-response. The participants in this contest were recruited mainly via mass-media efforts and at the workplace and had to remain tobacco-free for four weeks. The participants signed up for the contest by sending in a registration form.The statistical analysis has been completed, based on three approaches to assess the non-response bias: in the first approach, a univariate analysis, the participants were studied based on their response behavior and socio-demographic factors along with tobacco habits. The second approach, also a univariate analysis, was carried out to study the extent to which the participants had remained tobacco-free, and compared respondents who returned questionnaires after the initial request with those who responded after the first or second reminder letter, or who were interviewed by telephone at a later date. Finally, the third approach dealt with the rate of tobacco-free persons, and involved a comparison between the participants who responded to the questionnaire and those who were interviewed by telephone.The significance of non-response on the rate of tobacco-free individuals during a twelve-month period has been estimated through various assumptions. A pre-and post-test design using data from the participants entry forms and twel...

Are non-responders in a quitline evaluation more likely to be smokers?

“…However, the possibility exists that non-responders in studies with lower response rate may differ from the present study population and our results may only apply for studies with a similar or higher response rate. The present results are in line with a study based on a one-year follow-up of participants from a national "Quit and Win" contest where bias in smoking prevalence because of non-response was studied [19]. …”

Relapse prevention in a national smoking cessation contest: Effects of coping planning

“…In order to minimize potential bias, we have presented intention to treat estimates of abstinence. These conservative estimates may, however, underestimate actual abstinence rates and trivialize the effects of the planning intervention (Tillgren, Ainetdin, & Stjerna, 2000; Tomson, Björnström, Gilljam, & Helgason, 2005).…”

Beyond motivation : an exploration of pre- and postmotivational determinants of cancer-related behaviors

“…In order to minimize potential bias, we have presented intention to treat estimates of abstinence. These conservative estimates may, however, underestimate actual abstinence rates and trivialize the effects of the planning intervention (Tillgren et al, 2000;Tomson et al, 2005).…”