Impact of targeted wording on response rates to a survey of general practitioners on referral processes for suspected head and neck cancer: an embedded randomised controlled trial.

Abstract

Introduction Low response rates to surveys can lead to non-response bias, limiting generalisability of findings. When survey topics pertain to uncommon conditions, the decision of general practitioners (GPs) to complete a questionnaire may be swayed by the perceived relevance of the questionnaire content to their practice. Aim To explore whether targeted wording of a questionnaire for GPs about head and neck cancer referral patterns affects response rates. Methods A randomised controlled trial was embedded into a larger survey on referral practices for head and neck cancer among GPs in New South Wales, Australia. GPs were randomly allocated to receive versions of the study material with explanatory text written using either a 'symptom-frame' or a 'cancer-frame'; however, the questions and responses were the same in both groups. Results The overall response rate was 10.9% (196/1803). The response rate to the 'cancer-frame' version was 10.6% and 11.1% for the 'symptom-frame' version. After adjusting for practice location and GP gender, the difference in response rate based on wording was not significant (difference 0.5% [95%CI: -2.4, 3.4%]). A sub-analysis showed that GPs practicing in regional New South Wales were more likely to respond to the survey compared to those practicing in metropolitan New South Wales, independent of the intervention group or participant sex (AOR 1.61 [95%CI: 1.12, 2.31]; P = 0.01). Discussion The wording 'frame' of the survey did not appear to impact response rates in a survey of referral practices for suspected head and neck cancer; however, the significantly higher response rate from regional GPs warrants further investigation as to whether the content was considered more salient to their practice.