Article Text
Abstract
Objective: Using meta-analytic procedures, we compare the effectiveness of recent controlled trials of worksite smoking cessation during the 1990s with a previous meta-analysis of programmes conducted in the 1980s.
Data sources: ABI/Inform, BRS, CHID, Dissertation Abstracts International, ERIC, Medline, Occupational Health and Safety Database, PsycInfo, Smoking and Health Database, SSCI, and Sociological Abstracts.
Study selection: Controlled smoking cessation interventions at the workplace with at least six months follow up published from 1989 to 2001 and reporting quit rates (QRs).
Data extraction: Two reviewers independently scanned titles/abstracts of relevant reports, and we reached consensus regarding inclusion/exclusion of the full text reports by negotiation. A third reviewer resolved disagreements. Two reviewers extracted data according to a coding manual. Consensus was again reached through negotiation and the use of a third reviewer.
Data synthesis: 19 journal articles were found reporting studies conforming to the study’s inclusion criteria. Interventions included self help manuals, physician advice, health education, cessation groups, incentives, and competitions. A total of 4960 control subjects were compared with 4618 intervention subjects. The adjusted random effects odds ratio was 2.03 (95% confidence interval 1.42 to 2.90) at six months follow up, 1.56 (95% CI 1.17 to 2.07) at 12 months, and 1.33 (95% CI 0.95 to 1.87) at more than 12 months follow up. Funnel plots were consistent with strong publication bias at the first two follow ups but not the third. In Fisher et al’s 1990 study, the corresponding ORs were 1.18, 1.66, and 1.18.
Conclusions: Smoking cessation interventions at the worksite showed initial effectiveness, but the effect seemed to decrease over time and was not present beyond 12 months. Compared to the Fisher (1990) analysis, the effectiveness was higher for the six month follow up. Disappointingly, we found methodological inadequacies and insufficient reporting of key variables that were similar to those found in the earlier meta-analysis. This prevented us from determining much about the most effective components of interventions. It is advisable for researchers conducting studies in the future to report data on attrition and retention rates of participants who quit, because these variables can affect QRs.
- worksite
- cessation
- systematic review
- meta-analysis