2011 HSR&D National Meeting Abstract
1014 — Validity of Diagnosis Codes in VA Administrative Data for Identifying OEF/OIF Veterans with History of Traumatic Brain Injury
Carlson KF (Center for the Study of Chronic, Comorbid Mental and Physical Disorders, Portland VA Medical Center), Barnes JE
(Center for Chronic Disease Outcomes Research (CCDOR), Minneapolis VA Medical Center), Hagel EM
(CCDOR), Cifu DX
(Physical Medicine & Rehabilitation Program Office, Veterans Health Administration), Sayer NA
Prior non-VA research has identified limitations in using medical records diagnosis codes for enumerating cases of traumatic brain injury (TBI). The validity of TBI diagnosis codes in VA records is unknown. This study examined the sensitivity and specificity of VA-approved TBI diagnosis codes in VA administrative data.
VA Directive 2010-012 requires that clinicians with requisite TBI expertise use a structured TBI Comprehensive Evaluation template to complete evaluations of OEF/OIF Veterans who screened positive for TBI. TBI Comprehensive Evaluation data for OEF/OIF Veterans evaluated through June 16, 2010 were linked with inpatient and outpatient ICD-9 (International Classification of Diseases – 9th Revision) diagnosis codes from national datasets. We used clinician-confirmed cases and non-cases of TBI as indicated in the TBI Comprehensive Evaluation as the criterion standard. We searched for TBI-related diagnosis codes that were assigned plus or minus seven days from the entry date of the TBI Comprehensive Evaluation. We calculated measures of concordance and examined variations in concordance by VISN and by time since TBI Comprehensive Evaluation template implementation in October, 2007.
Participants included 49,962 OEF/OIF Veterans with completed TBI Comprehensive Evaluation templates who had VA administrative records during the specified time frame. Most Veterans were male (94%) and reported having been exposed to blasts (75%). Mean age was 31.7 years. Of the 29,534 (59%) clinician-confirmed TBI cases, 70% had been assigned a TBI diagnosis code and 30% had not. Of 20,428 (41%) confirmed non-cases, 18% had been assigned a TBI diagnosis code while 82% had not. Thus, sensitivity of TBI diagnosis codes was 70%, specificity was 82%, and overall concordance was 75%. Concordance varied significantly across VISNs (p < 0.0001). No trends in concordance were observed over time.
Similar to prior non-VA research, our findings show that reliance on VA medical records diagnosis codes for TBI case ascertainment may result in substantial false-positive and false-negative cases.
This is the first study to examine validity of TBI-related diagnosis codes in VA administrative data. Results help quantify potential bias when using VA-approved codes to enumerate OEF/OIF Veterans with TBI. Findings may also be useful for efforts directed at quality improvement of coding.