Background/Purpose:

Gout is a common inflammatory arthritis characterized by repeated acute flares. The ability to accurately identify gout flares is critical for comparative effectiveness studies of gout treatments. Prior studies use claims data to identify gout flares, however, these algorithms have not been validated. This study aimed to develop and validate a claims-based algorithm to identify gout flares.

Methods:

We identified patients receiving care at an academic medical center between 2006 and 2010 with a diagnosis of gout or hyperuricemia using an electronic medical record-Medicare claims linked dataset. We developed 3 algorithms to identify gout flares: 1) International Classification of Diseases, Ninth Revision (ICD-9) code for gout (274.X) and ≤ 1 dispensing for a gout-related medication including colchicine, nonsteroidal anti-inflammatory drugs, COX-2 selective inhibitors, and oral glucocorticoids ≤ 7 days from the date of gout ICD-9 code. The algorithm was run for any gout-related medication and for the individual medication categories. 2) ICD-9 code for gout and a J code for injectable glucocorticoid or a current procedural terminology (CPT) code for arthrocentesis or joint injection ≤ 7 days from the date of gout ICD-9 visit code. 3) ICD-9 code for gout and a J code for injectable glucocorticoid or a CPT code for arthrocentesis or joint injection on the same day.  Gout flares defined by the algorithms were confirmed through medical record review. Physician documentation of gout flare in the record was considered the gold standard. Positive predictive value (PPV) and 95 % confidence intervals (CI) of the algorithms were calculated. A set of 100 patients with a visit coded for gout but without any gout-related medication, arthrocentisis or injectable glucocorticoid claims was used to identify gout without flare and to calculate the negative predictive value (NPV).

Results:

503 flares were identified using the medication algorithm, and 290 were identified using the procedure ≤ 7 days algorithm. The mean age of the patients in the medication algorithm was 75 (±8) years, and 61% were male. The mean age for the procedure algorithm was 76 (±8) years and 68% were male.  The PPV of medication claims ranged from 50-54%. The PPV of the procedure claims ≤ 7 days was 59%, the same day procedure claim was stronger with a PPV of 68% (Table).  The NPV of the algorithm identifying gout without flare was 88% (95% CI 82, 94).

Conclusion:

Our results suggest that a claims-based algorithm utilizing a combination of diagnosis and procedure codes as well as medications may misclassify patients as having a gout flare and caution should be used in interpreting data using claims-based definition of flares. However, as the NPV was high, the claims-based algorithm may be useful to assess the absence of gout flare or to identify a cohort of gout patients with low disease activity or disease remission.

« Back to 2015 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/identification-of-gout-flare-using-an-administrative-claims-based-algorithm/