Background Considerable challenges exist for researchers attempting to monitor longitudinally the impact of any intervention on heavy drinkers, therefore they are often excluded from surveys. A particular challenge is the loss of validity through attrition.

Aim To describe issues encountered when recruiting and re-contacting difficult to reach heavy drinkers participating in a longitudinal study; and propose strategies to inform the design of future studies to minimise the effects of confounding factors.

Discussion Baseline recruitment exceeded targets, but attrition at first follow-up interview was considerable. Baseline alcohol consumption was not predictive of loss to follow-up. A variety of factors affected attrition including abstinence, severe intoxication at interview, deaths, selling of telephone, change of address and incarceration.

Conclusion Longitudinal studies that use personal telephones or address details in following up heavy drinkers face considerable challenges to minimise attrition. An important mitigating factor is the use of flexible and experienced interviewers.

Implications for practice The anticipated and reactive strategies documented in this paper provide important lessons for costing, designing and collecting data in future studies.