A randomised controlled trial to determine the effect on response of including a lottery incentive in health surveys [ISRCTN32203485]

Randomised controlled trial. Setting: North and West Birmingham. 8,645 patients aged 18 or over randomly selected from registers of eight general practices (family physician practices). Intervention: Inclusion of a flyer and letter with a health questionnaire informing patients that returned questionnaires would be entered into a lottery-style draw for ￡100 of gift vouchers. Control: Health questionnaire accompanied only by standard letter of explanation. Main outcome measures: Response rate and completion rate to questionnaire.5,209 individuals responded with identical rates in both groups (62.1%). Practice, patient age, sex and Townsend score (a postcode based deprivation measure) were identified as predictive of response, with higher response related to older age, being female and living in an area with a lower Townsend score (less deprived).This RCT, using a large community based sample, found that the offer of entry into a lottery style draw for ￡100 of High Street vouchers has no effect on response rates to a postal health questionnaire.Self-completed postal questionnaires are an economical and simple method of data collection for both research studies and audit activity. They are cheaper than telephone or personal interviews [1] and may be particularly useful in medical research as the response rate to sensitive questions may be greater than from other methods of data collection [2]. However, questionnaire based studies are subject to non-response bias [3]. If there are differential response rates from certain groups in the sample population, then generalisability of results to the target population is questionable. The lower the response rate, the more open to criticism the conclusions drawn from the research will be. High response rates are important not only because they reduce the risk of non-response bias but also because they increase the precision of parameter estimates [4]. Ensuring a high response rate to initial mailings has the additional benefit o