BACKGROUND: The Internet has introduced an alternative approach to traditional methods for conducting multinational cohort studies. However, the validity of Internet studies is not well established. The ATTEMPT study was designed to describe the natural history and patterns of smoking cessation, as well as the short-term health outcomes associated with smoking cessation. OBJECTIVES: The objective of this sub-analysis was to evaluate the validity, loss to follow-up, and missing data within a cohort of smokers (ATTEMPT). METHODS: ATTEMPT participants (n¼2009), were recruited by email from existing Internet panels in four countries, Canada (CA) (n¼208), France (FR) (n¼201), United Kingdom (U.K.) (n¼200), and the United States (U.S.) (n¼1400), in spring 2003. Participants had to be aged 35–65 years, smoke at least five cigarettes per day, and intend to quit smoking within the next 3 months. Online assessments were conducted at baseline and quarterly thereafter for 1 year during the ATTEMPT pilot phase. Assessments took approximately 25 minutes to complete and included questions about smoking, health and medical resources, and weight. Study participants were mailed standardized weight scales. Self-reported body weight was validated in a random sample of U.S. participants who agreed to an in-home visit by a health professional. Agreement between self-reported weight and actual weight was tested using the Bradley-Blackwood method. RESULTS: The overall response rate was 76% at month 3 and 52% at month 12 and varied by country (CA-55%; FR-36%; U.K.-39%; U.S.-56%). Duration of baseline data collection ranged from 3 days (CA) to 16 days (FR), and duration of month 12 data ranged from 12 days (CA) to 21 days (FR, U.K.). There was no difference in age, sex, race, or selected health conditions between baseline and month 12 respondents. Mean number of missing items per respondent was 2.2 items at month 3 and decreased after minor questionnaire revision at month 6. Eighty-two per cent of participants agreed to an in-home visit. Of 815 randomly sampled participants, 64% completed an in-home assessment, 23% could not be reached, and 13% refused. There was no difference between self-reported and in-home visit weight at month 3 (0.24 kg NS). Only a slight difference was seen at months 6 and 9 (1.2 kg [p¼0.02], and 1.01 kg [p¼0.01]), respectively. CONCLUSIONS: This analysis demonstrated the feasibility of using an Internet panel to enroll a cohort of smokers. There was no indication of bias related to non-response associated with demographic and health characteristics.
