Session Information
Date: Tuesday, October 23, 2018
Title: Epidemiology and Public Health Poster III: SLE, SSc, APS, PsA, and Other Rheumatic Diseases
Session Type: ACR Poster Session C
Session Time: 9:00AM-11:00AM
Background/Purpose: Psoriatic arthritis (PsA) is an inflammatory arthritis that can present variably in patients with or without psoriasis. The ability to accurately identify PsA in the large electronic healthcare database is critical for epidemiological studies of this disease. This study aimed to develop and validate a claims-based algorithm to identify patients with PsA.
Methods: We used data from the Partners Healthcare electronic medical record linked to Medicare claims in 2012. We developed 4 claims-based algorithms to identify cases of PsA: 1) ≥2 International Classification of Diseases, Ninth Revision (ICD-9) codes for PsA (696.0), at least one by a rheumatologist; 2) ≥2 ICD-9 codes for PsA by any physician and ≥1 claims for PsA-related medication, including biologic and non-biologic disease modifying anti-rheumatic drugs; 3) ≥2 ICD-9 codes for PsA and ≥1 ICD-9 codes for psoriasis (696.1) by any physician; and 4) ≥2 ICD-9 codes for PsA and no more than one ICD-9 code for rheumatoid arthritis (RA; 714.0). The ICD-9 codes were all separated by ≥7 days but <365 days. In algorithms 1, 3, and 4, the index date was defined as the date of the diagnosis code for PsA that occurred second; in algorithm 2, the index date was defined as whichever event (ICD-9 code or medication claim) occurred third. PsA cases defined by the algorithms were confirmed by medical record review, with diagnosis of PsA documented in the clinical record by the treating physician considered as the gold standard. Positive predictive value (PPV) and 95% confidence intervals (CI) of the algorithms were calculated.
Results: The 4 algorithms identified 281, 261, 224, and 216 records respectively, however around 40% had adequate data (defined below) and were included. The PPV of the algorithms ranged from 86.6-90.5% (Table).
Conclusion: Our records suggest that a claims-based algorithm utilizing two or more diagnosis codes for PsA alone, or in combination with a diagnosis for psoriasis, no more than one diagnosis code for RA, or linked to a PsA-related medication, can accurately identify cases of PsA in the claims database.
Table. Positive predictive value of algorithms.
Algorithm |
Records identified |
Records with adequate data* |
Confirmed PsA per treating physician |
PPV of algorithm |
|
n |
n (%) |
n |
% (95% CI) |
≥2 ICD-9 codes for PsA, at least one by rheumatologist |
281 |
134 (47.6) |
116 |
86.6 (79.6, 91.8) |
≥2 ICD-9 codes for PsA AND ≥1 medication claim |
261 |
113 (43.3) |
98 |
86.7 (79.1, 92.4) |
≥2 ICD-9 codes for PsA AND ≥1 ICD-9 code for psoriasis |
224 |
95 (42.4) |
86 |
90.5 (82.8, 95.6) |
≥2 ICD-9 codes for PsA AND no more than 1 ICD-9 code for RA |
216 |
104 (48.1) |
93 |
89.4 (81.9, 94.6) |
*Defined as clinic notes that discuss or address diagnosis of PsA within 6 months of the index date.
To cite this abstract in AMA style:
Ford J, MacFarlane LA, Tong A, Kim SC. Identification of Psoriatic Arthritis Using an Administrative Claims-Based Algorithm [abstract]. Arthritis Rheumatol. 2018; 70 (suppl 9). https://acrabstracts.org/abstract/identification-of-psoriatic-arthritis-using-an-administrative-claims-based-algorithm/. Accessed .« Back to 2018 ACR/ARHP Annual Meeting
ACR Meeting Abstracts - https://acrabstracts.org/abstract/identification-of-psoriatic-arthritis-using-an-administrative-claims-based-algorithm/