comorbidity: An R package for computing comorbidity scores

Abstract

Comorbidity scores are extensively used in observational medical research studies to avoid potential bias when the burden of disease could be confounding the association of interest. This is of primary importance, given the increasing availability and use of administrative data (such as medical records and insurance claims) for research purposes. Several comorbidity scores have been proposed throughout the years; however, two of them are used most frequently in practice: the Charlson comorbidity index (Charlson et al. 1987) and the Elixhauser comorbidity index (Elixhauser et al. 1998). The Charlson comorbidity index defines a set of comorbid conditions using International Classification of Disease (ICD) diagnostic codes. Each comorbid condition has an associated weight, and the sum of all weights results in a single comorbidity score per patient. The current version of the Charlson score includes 17 comorbidities. Similarly, the Elixhauser comorbidity index is based on ICD diagnostic codes and includes 31 comorbidities. In origin, the Elixhauser index score was based on the cumulative number of conditions present; since then, several weighting systems accounting for the increase or decrease in mortality risk associated with each condition have been proposed and used in practice (van Walraven et al. 2009; Moore et al. 2017).