Agreement between electronic medical records and self-reported urologic domains in the National Spina Bifida Patient Registry (NSBPR): Implications for future research

SummaryBackgroundSelf-report (SR) is an efficient data collection method. However, SR data have been shown to be discrepant with medical record (MR) documentation, which raises questions about using SR to supplement retrospective chart review in research. In this study, pediatric spina bifida (SB) patients who completed SR interviews about continence status and personal bladder/bowel management were identified. We examined agreement between SR data and Urology provider notes in MRs.ObjectiveThis study aimed to (1) identify demographic, medical, or methodological factors that might contribute to SR/MR disagreement; (2) postulate how these findings might be significant clinically; and (3) recommend improvements to SR data collection and MR documentation.Study designOur institution participates in the National Spina Bifida Patient Registry (NSBPR). NSBPR-enrolled subjects typically complete annual interviews about their urologic outcomes; we consider this to be a form of SR. After identifying patients who interviewed within 1 month of an encounter with a urology provider, we systematically reviewed and compared their SR responses to the MR. Overall SR/MR agreement (no. of agreeing data pairs/no. of complete data pairs) and strength of agreement (kappa, κ) were assessed. Agreement about daytime continence status was assessed for children ≥5 years or in younger children who were toilet trained. Analyses were also stratified by diagnosis, type of bladder management, and ethnicity.ResultsEleven urologic domains were analyzed for 176 patients. Overall SR/MR agreement was ≥90% for nine out of 11 domains (figure). Daytime urinary and stool incontinence (DUSI) domains demonstrated the lowest overall agreement, at 69% and 74% respectively. Patients with myelomeningocele (MM) and those on clean intermittent catheterization demonstrated twice as much SR/MR disagreement about DUSI than patients without MM and those who void. There was no significant difference in rates of SR/MR agreement about DUSI when analyzed by ethnicity, race, and ambulatory function status. Among cases of SR/MR disagreement about DUSI, the SR and MR had a roughly equal percentage of better outcomes reported for both UI and SI.DiscussionThere was strong SR/MR agreement for the majority of urologic data we analyzed. Medically complex patients faced lower SR/MR agreement, which is consistent with findings in other patient populations. Biased reporting by patients/families or providers was not found.ConclusionMinimizing SR/MR disagreement through standardized data collection methods and tools, improved definitions of patient outcomes, and documentation of respondent identity will improve large, multisite studies that utilize SR and MR concurrently.294Figure. Overall SR/MR agreement by urologic domain.