RATIONALE AND OBJECTIVE: To review response rates published in medical imaging journals, and to analyze potential factors which contributed to a low response rate.
MATERIALS AND METHODS: A literature search was performed in MEDLINE and Embase to identify and assess published medical imaging survey studies. Variables assessed were response rate, incentives such as reminders and remuneration, and rationales provided for a potential low response rate. Statistical significance was calculated using unpaired t tests, ANOVA, Mann-Whitney, and Kruskal-Wallis tests.
RESULTS: Three hundred and fifty-six unique surveys were included for analysis. The mean survey response rate in the current age of predominately electronic surveys was 45%. Factors which statistically significantly demonstrated a difference in response rate were survey location (European countries: 52%, Canada: 47%, United States: 42%; p < 0.05), survey topic (musculoskeletal: 69%, nuclear medicine: 64%, and education: 47%; p < 0.05), survey delivery method (telephone: 76%, email: 41%; p < 0.0001), and survey question type (short answer: 62%, multiple choice: 43%; p < 0.01). Statistically significant linear correlations were observed between the response rate compared to the number of reminders sent (r = 0.27; p < 0.01) and the number of participants (r = -0.26; p < 0.0001).
CONCLUSION: The survey response rate serves as a surrogate marker for nonresponse bias. Survey response controlled for intrinsic nonadjustable characteristics offer achievable research goals. Adjustable factors to low response, including survey delivery method, question type, and number of reminders demonstrated statistical difference in response rate, and can be utilized by researchers to prospectively minimize nonresponse bias.