Describing The Methodology To Estimate Accelerometer Return Percentage From A Study Utilizing Mail-based Data Collection

Abstract

Recent advances in accelerometer design and battery life have improved the practicality of mail-based delivery and return for data collection in large epidemiologic studies. Mail-based administration can significantly reduce the cost and logistic challenges associated with in-person delivery/recovery, but requires other resources (e.g. phone calls, direct mailing) to optimize protocol compliance and monitor return. To date, there is no available method to forecast the number of monitors expected to be returned within a certain time-frame. This information is critical to appropriate allocate study staff and resources. PURPOSE: To describe a novel methodology to estimate accelerometer return percentage in the Houston TRAIN (Transit-Related Activity In Neighborhoods) Study. METHODS: TRAIN is a prospective natural experiment of diverse, low-resource adults (aged ≥ 18 years), residing within a 3-mile radius of new light rail transit lines in Houston, TX. Continuous enrollment took place from January 2014-October 2015. Accelerometers were delivered via first class U.S. mail and participants were compensated upon return of the monitor. Accelerometer retrieval times were used to calculate percentiles of “days to return”. These percentiles were then used to determine the number of participants falling within each percentile range for “days to return” among those who have not yet returned their monitor. Finally, the expected number of accelerometers to be returned was calculated for participants who have not yet returned their monitor, based on the calculated probability of returning the accelerometer as a function of the total number of days outstanding. RESULTS: Return data from participants in the first 19 months of data collection for the TRAIN study were analyzed (n = 553). To date, 426 (77%) were returned. The 95th and 99th percentiles of “days to return” was 60 and 127 days, respectively. Currently, 127 accelerometers have not yet been returned, of which 64 have been outstanding for over 127 days. Among those outstanding, 19.1 accelerometers are expected to be returned. CONCLUSIONS: This empirical-based methodology can be useful for implementing cost-effectiveness strategies in continuous-enrollment studies utilizing mail-based retrieval of data collection materials. Funded by NIH R01 DK101593.