Accuracy of self-reported injuries compared to medical record data.

Background: Self-reported injury data are frequently used in epidemiologic investigations. These data provide useful information about the activities and mechanisms of injuries because injury cause-coding is often not required for outpatient medical visits.

Objectives: The purpose of this evaluation is to determine the accuracy of self-reported military injuries when compared to injuries in outpatient medical records.

Method: Injuries reported by survey were compared to diagnoses for injuries (International Classification of Diseases (ICD-9-CM 800-999)) and injury-related musculoskeletal disorders (selected ICD-9-CM 710-739) obtained from medical records. Self-reported injury responses from military personnel were matched to diagnoses by date and body part. A new methodology for including secondary matching body parts was proposed and implemented.

Results: Infantry Soldiers (n = 5490) completed surveys that requested details about their most recent injury. About one-quarter (24%, n = 1336) reported injuries on the survey and had an injury diagnosis in their medical record in a six month period. Seventy-five percent of the self-reported injuries (n = 996 of 1336) were confirmed by medical records with a date match within 3 months and an identical or nearby body part. Common self-reported injuries were ankle sprains (10%), knee sprains (9%), lower back strains (4%), shoulder strains (3%), and lower back pain (3%).

Conclusions: A high percentage of self-reported injuries were accurate when compared with medical records, substantiating the use of survey data for the evaluation of injury outcomes. This is the first effort to validate self-reported injuries and musculoskeletal disorders with medical records in a large military population.