Editorial Comment: Questionnaire survey-based research: Is there a need for consensus?

Research based on questionnaire surveys has become an attractive method of gathering data in scientific studies and an increasing trend has been seen recently in clinical medicine. It is a quick and cost-effective way of assessing a particular practice and the variation in its adoption across disciplines. The quality, rigor, and scientific standards of formulating instruments (questionnaires) to conduct surveys for obtaining information vary. There is a need to have agreed guidelines. First, the design of questionnaires should have inherent methods of validation. The validity of questionnaire is an essential element as it provides assurance that information gathered is robust and conclusions drawn are reliable.[12] The steps of validation, including pilot testing, should bring to the fore final scientific punctilious details.[3] Second, research studies should also reflect attempts to improve response rate to questionnaire-based surveys.[4] It is disheartening to see a poor response rate after significant efforts have been put into a well-designed research study. Underlying reasons for poor response are often lack of motivation or time constraints that may differ depending on the study population selected and questionnaires used. The Association of Medical Education in Europe has produced a document enlisting strategies to address some of the issues of poor response rate;[5] however, this can further be tailored to specialty-specific or research question-oriented way of addressing a nonresponse bias. Contacting participants before, or after, sending them questionnaires, personalizing invitation letters, incentivizing responses, interval reminders, and keeping questionnaires short and punchy are some of the well-described strategies,[6] and these should reflect in survey studies to provide a measure of confidence in research conclusions.

Should there be a restriction on publication of low-response rate questionnaire-based research? If yes, what should be the target response rate? The Journal of the American Medical Association is explicit in its policy and states “Survey studies should have sufficient response rates (generally at least 60 percent) and appropriate characterization of nonresponders to ensure that nonresponse bias does not threaten the validity of the findings.”[7] Similarly, The Office of Management and Budget responsible for funding of survey protocols in the USA clarifies in their statement as “submit a plan for a nonresponse bias analysis if the expected unit response rate is below 80% (see Section 3.2.9).”[7]

This may sound stringent and perhaps too fixated on response rate, but an equally important emphasis has been placed on plans for a nonresponse bias analysis and this should be the focus of scientific publications.[7] Regardless of response rate, a plan of handling nonresponse bias before and after administration of survey questionnaires should be part of submission to journals for publication as a quality assurance to health-care policy makers and medical practitioners. Halbesleben and Whitman[8] have described a framework to achieve this and it should be a basis for further implementation of this practice.