Can lifestyle surveys survive a low response rate?

Primary care trusts (PCTs) need local lifestyle data to plan and monitor the effectiveness of public health interventions. Several sources of data are available, such as synthetic estimates, primary care data, and national or local surveys. Many PCTs across the country have opted to use local lifestyle surveys to get up-to-date, accurate health behaviour data. There is a national trend towards lower response rates from postal surveys, posing a risk to the accuracy of survey data in many areas. As a result, it is not unusual for a PCT to find itself pondering how best to deal with a disappointing response rate and the consequent selection bias this often means. This was the situation in which Stockport PCT found itself after conducting a survey in 2004. The Stockport Health Survey was sent to a simple random sample of 15 000 residents aged 18 years (7% of the total population aged 18 years) in 2004. The response rate to the lifestyle survey was low (36%). It is generally agreed that a response rate of 60% for a survey is acceptable, but this is the real world and a quick review of response rates achieved in similar surveys in the UK revealed that 60% is rarely achieved (of seven surveys, the average rate was 45%, ranging from 27% to 70%).