Administrative codes for Behçet's, Still's disease, Mediterranean fever 'highly accurate' for diagnosis
Click Here to Manage Email Alerts
The administrative codes for Behçet's disease, adult-onset Still’s disease and familial Mediterranean fever are “very good to excellent” in terms of estimates of specificities and positive predictive value, according to findings published in the Journal of Clinical Rheumatology.
“Autoinflammatory syndromes (AIS) are a group of rare rheumatic diseases driven by the innate immune system’s inflammasome leading to recurrent and disabling attacks of fever, rash, arthritis, serositis and orogenital ulcers,” Aleksander Lenert, MD, MS, FRCPC, of the University of Iowa, told Healio Rheumatology. “Additionally, patients with AIS can present with severe complications at onset including macrophage activation syndrome, and severe cardiac and pulmonary disease. Due to a lack of longitudinal cohorts, we are limited in our ability to conduct clinical and pharmacoepidemiologic research and describe key long-term outcomes of patients with AIS.”
“The use of “big data” in rheumatology research includes tapping into the power of large administrative health databases, which contain information on all patient contacts with the health care system and capture provider billing codes for diseases, procedures and treatments,” he added. “Administrative databases represent an opportunity to study rare rheumatic diseases at a population level. However, there are no studies validating billing codes for the accurate diagnosis of AIS.”
To assess the accuracy of case definitions for autoinflammatory syndromes based on administrative claims codes, compared with electronic medical records, Lenert and colleagues analyzed data from the University of Kentucky HealthCare. The researchers reviewed data from July 2004 through February 2019 on all adult patients with at least one outpatient rheumatology visit and one or more the ICD-9/ICD-10 codes 277.31, 136.10, 711.2x, 714.20, 714.30, M04.1, M04.2, M04.8, M04.9, M06.1, or M35.2.
In addition, the researchers identified additional patients with potential autoinflammatory syndromes by searching the National Drug Codes for anakinra, canakinumab (Ilaris, Novartis) and rilonacept (Arcalyst, Regeneron). Lenert and colleagues determined the sensitivity, specificity, positive predictive values, negative predictive values and the area under the receiver operating characteristic curve for each of the specific diagnosis codes.
According to the researchers, there were 273 individuals with possible autoinflammatory syndrome, of whom 26.4% had a true diagnosis. Among the patients with a true diagnosis, 24 had adult-onset Still’s disease, 32 demonstrated Behçet’s disease and nine had familial Mediterranean fever. For all three conditions, the estimates of specificities and negative predictive values for specific administrative codes were excellent, or greater than 95%. Sensitivity estimates were excellent, or greater than 89%, for Behçet’s disease and familial Mediterranean fever codes, but lower (46% to 50%) for adult-onset Still’s disease.
In addition, positive predictive values were excellent for Behçet’s disease — greater than 99% — and adult-onset Still’s disease — greater than 86% — but lower for familial Mediterranean fever — greater than 53%. The area under the receiver operating characteristic curve estimates were excellent for Behçet’s disease (97% to 98%) — and familial Mediterranean fever (93%) —and very good for adult onset Still’s disease (75%).
“We successfully identified specific billing codes for AOSD (M06.1, 714.20), BD (M35.2, 136.10) and FMF (277.31, M04.1) which are highly accurate in identifying these diseases,” Lenert said. “As a next step, we plan to test our findings in additional electronic medical records (EMRs) and large administrative databases. Future application of our study's findings would enable clinical outcomes and epidemiologic research to advance the understanding of autoinflammatory syndromes and help lead to improved care for our patients.” – by Jason Laday
Disclosure: The researchers report no relevant financial disclosures.