Using the capture-recapture technique to estimate the ascertainment-corrected incidence of musculoskeletal injuries during Marine Corps recruit training

Abstract: Introduction: Injury epidemiology research with military populations typically utilizes data obtained through medical chart review (MCR) or injury self-reports (ISRs). MCR data will not capture musculoskeletal injury (MSKI) data for which medical care was not sought, which is common during military recruit training. Injury self-report is affected by issues with recall, especially for MSKIs perceived as less severe. U.S. Marine Corps (USMC) recruits participate in an intense 13-week recruit training program during which they are susceptible to MSKIs. The purpose of the current analysis was to utilize a novel statistical method, the capture-recapture (CRC) technique, to account for the undercounting inherent in MSKI data sources and estimate the ascertainment-corrected cumulative incidence of MSKIs during USMC recruit training. Materials and Methods: Data for the current study were derived from a larger study, the USMC Gender-Integrated Recruit Training Study, which was initiated to provide data-driven recommendations to increase gender integration in USMC recruit training. The estimated cumulative incidence of MSKIs during 13-weeks of USMC recruit training was calculated from the 2 sources of MSKI data (MCR, ISR) and using CRC analysis. Medical charts were reviewed to extract data about MSKIs that occurred during recruit training. Self-reported MSKI data for the same period were obtained from recruits at the end of recruit training. MSKIs were classified according to their anatomical location and type. The Chapman modification of the Lincoln-Peterson estimator was utilized to conduct the CRC analysis. Results: Medical chart review and ISR MSKI data were available for 464 USMC recruits (age: 19.1 ± 1.9 years; gender: men 70.0%). The observed 13-week cumulative incidence of MSKI in the sample was 21.8% in the MCR and 28.4% in the ISR, while the CRC incidence was much higher (62.0%). The MCR and ISR ascertainment were 35.1% and 45.9%, respectively, while the overall ascertainment or completeness of MSKI data when 2 sources were used was moderate (65.0%). When stratified by MSKI anatomical location, the overall ascertainment varied by anatomical location of the MSKI. It was highest for lower extremity MSKIs (64.8%), but lower for upper extremity (38.9%) and spine (33.3%) MSKIs. The overall ascertainment also varied by MSKI type; it was highest for sprain (55.1%), followed by strain (54.8%), and the pain/spasm/ache (43.3%). Conclusions: This was the first study to utilize the CRC technique to access the ascertainment-corrected incidence of MSKIs among USMC recruits. There was significant undercounting in both sources of the data analyzed, and the extent of undercounting varied by both MSKI anatomical location and type. When 2 sources of data were utilized simultaneously, the percent of CRC-estimated MSKIs observed from 2 sources of data was more complete. There is a need for further application of the CRC technique to MSKI data in military populations to provide a more complete assessment of MSKIs. Identification of modifiable factors that influence completeness of MSKI data obtained during military recruit training is also warranted.