Miscoding for reimbursement or downcoding due to lack of information? Analysis of unspecific ICD-9-CM codes distribution and reimbursement in Portuguese hospitals

Abstract

Introduction: Missing information in health records can lead to "unspecified" codes. However, there is evidence of miscoding practices worldwide. This study aims to assess hospital morbidity miscoding by evaluating the presence of "unspecified" codes when representing different reimbursements. Materials and methods: Each ICD-9-CM "unspecified" code defined a clinical group. For each clinical group, "peer" (specified) codes were identified. They were further divided into six categories according to the reimbursement (relative weights) in comparison to their "peer" codes, including "higher", "equal" or "lower". Using the principal diagnosis of non-surgical hospital episodes of Portuguese public hospitals in 2015, we calculated the main outcome, i.e. proportion of "unspecified codes" for each clinical group and reimbursement category. The reimbursement categories were compared using weighted mean proportion, as well as the interquartile range, of the proportion of "unspecified" codes. Results: Although there is a heterogeneity of the number of clinical groups by the 6 reimbursement categories (Min=5; Max=1476), there was no evidence that those "unspecified" codes representing "higher" reimbursement showed higher proportion of such codes. The proportion of "unspecified" codes was lower in the "higher" and "equal or higher" categories, compared with the "lower" category. Conclusions: In Portugal, there is no evidence of upcoding using an "unspecified" codes as principal diagnosis. Instead, there might be "downcoding", potentially associated with the quality of health records.