Validation of Rash and Nonspecific Eruption Diagnoses Using ICD-9 and ICD-10 Codes

Abstract

Despite being common cutaneous findings in the hospital setting, cutaneous drug eruptions have not been thoroughly validated for administrative claims data. The purpose of this study was to validate primary discharge diagnoses for the ICD code rash or other nonspecific eruptions by finding the positive predictive value (PPV) and to determine how often it was used to indicate an eruption that was eventually determined to be a drug eruption. Two dermatologists independently examined each of 39 hospital patient encounters to assess the validity of the ICD code used by stating whether the code used was appropriate or inappropriate, per table 1 criteria. Since in most patients, a drug eruption was suspected, the Naranjo Adverse Drug Reaction Probability Scale was utilized to externally validate our findings. Our results yielded a PPV for nonspecific rashes of 94.87% (83.11-98.58%) for appropriateness of use. The mean Naranjo score for appropriately versus inappropriately coded cases was 5.7 versus 2.5, respectively, with a mean difference of 3.17568 (0.18833-6.16302), (pooled P-value = 0.0378). Out of the cases analyzed, 29/39 or 74% were confirmed to be drug-related. Our high PPV supports the validity of non-specific eruptions found in national patient databases and the high Naranjo criteria suggest that often, these eruptions are drug related, yet not confirmed at the point of discharge.