Open AccessIdentifying perianal fistula complications in pediatric patients with Crohn’s disease using administrative claims
Author(s) -
Jeremy Adler,
Hannah K. Jary,
Sally J. Eder,
Shiming Dong,
E. M. Brandt,
Jessica K Haraga,
Kevin J. Dombkowski
Publication year - 2019
Publication title -
plos one
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.99
H-Index - 332
ISSN - 1932-6203
DOI - 10.1371/journal.pone.0219893
Subject(s) - medicine , fistula , medicaid , retrospective cohort study , receiver operating characteristic , medical record , cohort , gold standard (test) , surgery , pediatrics , general surgery , health care , economics , economic growth
Background Although perianal fistulas occur commonly in pediatric Crohn’s disease (CD), evaluations of health services have been limited since no validated claims-based methods exist for identifying cases. Objective To develop and validate accurate case definitions for perianal fistulas among pediatric patients with CD from administrative claims. Methods Retrospective cohort study in which we developed and tested candidate case definitions for perianal fistula. Patients (age 5–21 years between 2005–2012) with CD enrolled in Michigan Medicaid with healthcare at University of Michigan were identified via claims. Medical records were obtained from all identified patients, whose entire records were abstracted. Medical record evidence for perianal fistula was considered the “gold standard” against which candidate case definitions were compared. The reference case definition of perianal fistula (ICD9 565.1) and candidate case definitions were evaluated. Results Of 843 patients identified via claims, 274 (33%) met CD criteria for inclusion. The true perianal fistula rate among CD patients was 18% (n = 49). The top-performing candidate case definition identified 15% (n = 42), had sensitivity of 77.6%, specificity of 98.2%, positive predictive value (PPV) 90.5%, negative predictive value (NPV) 95.3%, and area under receiver operator characteristic curve (ROC) of 0.88. In contrast, the reference case definition identified 9% (n = 26), sensitivity 51.0%, specificity 99.6%, PPV 96.2%, NPV 90.3%, and had an area under ROC of 0.75. Conclusions We demonstrated that it is feasible to use administrative claims data to accurately identify pediatric patients with perianal fistula complications. Claims-based case definitions were found to be highly accurate through medical record review, providing a high degree of confidence for future studies where chart review is not feasible. These claims-based methods can be applied to claims data in other settings for the evaluation of health services utilization as well as to assess the comparative effectiveness of prevention and treatment strategies.