Participation and factors associated with late or non‐response to an online survey in primary care

Abstract

Online surveys have several advantages, but a low response rate is common and it is uncertain how results are affected. Response inducement techniques can be used to overcome this problem. The objectives of this study were to describe the percentage of change in the response rate after reminders and to analyse the characteristics associated with non-response and late response based on the survey results, as well as by applying archival and extrapolation techniques. In the Autonomous Community of Madrid, an online questionnaire about cancer prevention was sent to a random sample of primary care health professionals (3586 physicians and nurses). Two reminders were sent later. The percentage of change in response rates after reminders, global and by demographic and health care characteristics of participants; and factors associated with non-response and late response were analysed using response rates and odds ratios (ORs). After the reminders, the response rate increased from 22.6% to 32.9% and to 39.4%. Non-response was associated with age [OR: 3.14; confidence interval (CI) 95%: 2.23–4.42 for aged >60 years], gender and functional area. Further, a higher response rate after reminders was observed in professionals with heavier workloads (OR: 1.46; CI 95%: 1.08–1.97) and in those who stated a lower relevance of cancer prevention in primary care. After electronic reminders, the response rate increased, especially among professionals with the highest workloads and a minor interest in the survey topic. However, possible bias associated with non-response remains and the factors behind this should be examined in future research.