How do sports injury epidemiological outcomes vary depending on athletes' response rates to a weekly online questionnaire? An analysis of 39‐week follow‐up from 391 athletics (track and field) athletes

Abstract

AbstractObjectiveTo explore how sports injury epidemiological outcomes (i.e., prevalence, average prevalence, incidence, burden, and time to first injury) vary depending on the response rates to a weekly online self‐reported questionnaire for athletes.MethodsWeekly information on athletics injuries and exposure from 391 athletics (track and field) athletes was prospectively collected over 39 weeks (control group of the PREVATHLE randomized controlled trial) using an online self‐reported questionnaire. The data were used to calculate sports injury epidemiological outcomes (i.e., prevalence, average prevalence, incidence, burden, and time to first injury) for sub‐groups with different minimum individual athletes' response rates (i.e., from at least 100%, at least 97%, at least 95%, … to at least 0% response rate). We then calculated the relative variation between each sub‐group and the sub‐group with a 100% response rate as a reference. A substantial variation was considered when the relative variation was greater than one SD or 95% CI of the respective epidemiological outcome calculated in the sub‐group with a 100% response rate.ResultsOf 15 249 expected weekly questionnaires, 7209 were completed and returned, resulting in an overall response rate of 47.3%. The individual athletes' response rates ranged from 0% (n = 51) to 100% (n = 100). The prevalence, average weekly prevalence, and time to first injury only varied substantially for the sub‐groups below a 5%, 10% and 18% minimum individual response rate, respectively. The incidence and injury burden showed substantial variations for all sub‐groups with a response rate below 100%.ConclusionsEpidemiological outcomes varied depending on the minimum individual athletes' response rate, with injury prevalence, average weekly prevalence, and time to first injury varying less than injury incidence and injury burden. This highlights the need to take into account the individual response rate when calculating epidemiological outcomes, and determining the optimal study‐specific cut‐offs of the minimum individual response rate needed.