A Medicare claims-based algorithm using both psoriasis (PsO) diagnosis codes and PsO-specific treatment dispensing had a significant positive predictive value (PPV) for identifying patients with psoriasis effectively, according to a study published in Pharmacoepidemiology and Drug Safety.
A research group study sought to accurately identify patients with PsO through a number various combinations of factors to guide future real-world evidence on disease course and treatment utilization. Nine claims-based algorithms for PsO were developed using a combination of the International Classification of Diseases, Ninth Revision (ICD-9) codes, specialist visit, and medication dispensing factors from Medicare-linked electronic health records (EHRs) data (2013-2014) in two health care provider networks in Boston, MA. The PPV and 95% CI were calculated for each algorithm using the treating physician's diagnosis of PsO via chart review as the standard. Disease activity was assessed among confirmed PsO cases.
The nine developed claims-based algorithms identified 990 unique patient records, of which 918 (92.7%) with adequate information were reviewed. PPV varied among the algorithms, ranging from 65.1% to 82.9%. The highest PPV (82.9%) was the algorithm defined as one or more ICD-9 diagnotic code for PsO plus one or more prescription claim for topical vitamin D agents. Additionally, the PPV of the algorithm defined as two or more ICD-9 diagnosis codes plus one or more prescription claim for PsO treatment (excluding topical steroids) was 81.1%. This algorithm’s PPV improved to 82.5% when further defined by one or more diagnosis from a dermatologist. Of the 411 patients with PsO with adequate information on disease activity in EHRs, 1.5% to 5.8% had no disease activity, 31.3% to 36.8% had mild activity, and 26.9% to 35.1% had moderate to severe activity ranges across the algorithms.
The authors concluded that claims-based algorithms based on a combination of PsO diagnosis codes and dispensing of PsO-specific treatments had a moderate to high PPV. They argue that these algorithms may be useful in identify appropriate real-world data for pharmacoepidemiologic studies. –Jessica Garlewicz
Lee H, He M, Cho SK, et al. Validation of claims-based algorithms to identify patients with psoriasis. Pharmacoepidemiol Drug Saf. Published online March 13, 2021. doi:10.1002/pds.5229