Diagnosis-based and external cause-based criteria to identify adverse drug reactions in hospital ICD-coded data: application to an Australia population-based study

Abstract

OBJECTIVES: External cause International Classification of Diseases (ICD) codes are commonly used to ascertain adverse drug reactions (ADRs) related to hospitalisation. We quantified ascertainment of ADR-related hospitalisation using external cause codes and additional ICD-based hospital diagnosis codes. METHODS: We reviewed the scientific literature to identify different ICD-based criteria for ADR-related hospitalisations, developed algorithms to capture ADRs based on candidate hospital ICD-10 diagnoses and external cause codes (Y40-Y59), and incorporated previously published causality ratings estimating the probability that a specific diagnosis was ADR related. We applied the algorithms to the NSW Admitted Patient Data Collection records of 45 and Up Study participants (2011-2013). RESULTS: Of 493 442 hospitalisations among 267 153 study participants during 2011-2013, 18.8% (n = 92 953) had hospital diagnosis codes that were potentially ADR related; 1.1% (n = 5305) had high/very high-probability ADR-related diagnosis codes (causality ratings: A1 and A2); and 2.0% (n = 10 039) had ADR-related external cause codes. Overall, 2.2% (n = 11 082) of cases were classified as including an ADR-based hospitalisation on either external cause codes or high/very high-probability ADR-related diagnosis codes. Hence, adding high/very high-probability ADR-related hospitalisation codes to standard external cause codes alone (Y40-Y59) increased the number of hospitalisations classified as having an ADR-related diagnosis by 10.4%. Only 6.7% of cases with high-probability ADR-related mental symptoms were captured by external cause codes. CONCLUSION: Selective use of high-probability ADR-related hospital diagnosis codes in addition to external cause codes yielded a modest increase in hospitalised ADR incidence, which is of potential clinical significance. Clinically validated combinations of diagnosis codes could potentially further enhance capture.