Diagnostic work-up and treatment in patients with pyoderma gangrenosum: retrospective analysis of US insurance claims-based data

Pyoderma gangrenosum (PG) is a rare, and often challenging to diagnose, inflammatory disorder with relatively high rates of morbidity and mortality. Central to the diagnosis of PG is histologic evaluation and exclusion of other entities. Large-scale studies investigating the proportion of patients receiving a thorough diagnostic work-up, as well as prevalence studies regarding comorbidities and systemic treatment in PG using claims-based data, are sparse. Our objective was to identify patients diagnosed with PG and describe the diagnostic work-up and prevalence of common comorbidities and therapies in this population using claims-based data in a retrospective cohort study. In order to better understand practices of diagnostic work-up, we captured rates of skin biopsy, tissue culture, and/or surgical debridement prior to initial diagnosis. We also identified the prevalence of PG-associated comorbidities and initial immunosuppressive therapy given for PG. Of the 565 patients diagnosed with PG, 9.4% underwent skin biopsy, 8% tissue culture, and 1.4% both skin biopsy AND tissue culture prior to diagnosis. Inflammatory bowel disease was the most prevalent comorbidity (16.3%). The most common treatment administered was systemic corticosteroids (17%). Although practice guidelines explicitly delineate histology and exclusion of infection as important diagnostic criteria, only a minority of patients in this study underwent skin biopsy and/or tissue culture prior to receiving a diagnosis of PG, suggesting that patients may receive a diagnosis of PG without having tissue evaluation. Such discordance between practice guidelines and "real-world" practice inevitably increases the risk for misdiagnosis of PG and misdirected treatment with immunosuppressants for presumptive PG in cases of PG mimickers. Moreover, comorbidities associated with PG may occur, or be identified in, a lower proportion of patients as compared with what is reported in the existing literature. Study limitations include a population restricted to < 65 years with commercial insurance and the reliance upon ICD diagnostic coding to capture the population.