The ATTEMPT cohort: a multi-national longitudinal study of predictors, patterns and consequences of smoking cessation; introduction and evaluation of internet recruitment and data collection methods



This article is corrected by:

  1. Errata: Errata Volume 101, Issue 10, 1528, Article first published online: October 2006

Robert West, University College London, Gower Street, London WC1E 6BT, UK. E-mail:


Aims  The ATTEMPT study was designed to chart the natural history of smoking cessation and associated short-term health outcomes and effects on medical resource utilization among a cohort recruited across multiple countries. This paper describes the methods for recruitment and follow-up, the baseline population characteristics of the enrolled population and 1-year response rates.

Design  ATTEMPT is a multi-national prospective cohort study that used the internet for subject recruitment and online assessments every 3 months for 2.5 years.

Setting  Subjects were recruited via e-mail from existing internet panels [Canada (n = 208), France (n = 201), the United Kingdom (n = 200) and the United States (n = 1400].

Subjects  Panel members who were aged 35–65 years, smoked at least five cigarettes per day and at initiation stated an intention to quit smoking within the next 3 months were eligible for this study.

Measurements  Measures included: attempts to quit, smoking status, smoking history, nicotine dependence and craving, methods used to quit smoking, reasons for quitting or failing to quit smoking, short-term health effects, health resource utilization, wellbeing, concern over weight gain, confidence in preventing weight gain, body weight and demographics. In addition, in-home assessments of weight were undertaken by field staff for a random sample of US participants.

Findings  Country-specific recruitment was completed within 17 days. The recruitment method produced a sample with characteristics broadly similar to those found in national surveys of smokers except for higher prevalence of obesity in the US and Canadian samples and higher educational level. At the end of 1 year the response rate was 52%, and there was little evidence of differential loss to follow-up by key subject characteristics. Weight reported in the survey was found to correlate highly with weight measured during in-home visits.

Conclusion  This paper demonstrates the feasibility of enrolling and following a diverse cohort of smokers for self-reported health and behaviour measures via the internet.