Utility of observer reports of smoking status

Subjects in a smoking cessation trial named observers to verify their smoking status at 1 week and 1-, 6-, and 12-month follow-ups. Several results indicated observer reports of abstinence have little utility in the classification of smoking status: 1. observer data were available on only 50-70% of follow-ups, 2. observer data refuted claims of abstinence in less than 5% of cases, 3. return bias appears to have occurred (i.e., observer forms were more likely to be missing if a subject reported abstinence), and 4. observer and biochemical verification of abstinence were often discordant.