Differences in the contents of two randomized surveys of GPs' prescribing intentions affected response rates

We investigated the differences in response rates and the presence of response bias in two randomized surveys of prescribing intentions for statins and asthma.

We conducted the surveys of British general practitioners (GPs) in 2002. The two surveys had similar designs, formats, administration time, administration methods, and target populations. We compared the response rates to the two surveys while controlling for the characteristics of respondents with nonrespondents. We also compared early respondents with late respondents and assessed heterogeneity in the answers of early and late respondents to two key questions.

The response rates to the two surveys were significantly different (statins: 27%; asthma: 19%; P=0.002). We found no interaction between the survey type and any of the GP and practice characteristics we examined. The GPs' answers to the key questions did not differ regardless of the timing of the responses.

We demonstrated that the surveys' contents significantly influenced the response rates. We found no evidence that the nonrespondents would have answered the key questions differently. Future studies should investigate the mechanisms by which contents of surveys may influence response rate.