Diagnostic performance of an International Classification of Diseases, Tenth Revision-based definition of injection drug use status in US veterans with opioid and stimulant use disorder

Background and aims US veterans with a history of injection drug use (IDU) are at increased risk for drug-related consequences, such as overdose or injection-related infections (ID). The study of population rates and outcomes among veterans with IDU has been limited by the lack of an International Classification of Diseases, Tenth Revision (ICD-10) code for IDU and the lack of validated definitions appropriate for use in administrative data. We aimed to identify active IDU in veterans with opioid use disorder (OUD) or stimulant use disorder (StUD) by utilizing an ICD10 algorithmic approach. Methods Veterans who visited a VA facility between 2017 and 2022 with an ICD-10 diagnosis of OUD or StUD and one of seven IDU-associated infections were included. A random subsample from each cohort underwent manual chart verification for evidence of IDU. Demographic characteristics and comorbidities were collected. The diagnostic performance of each ICD10 algorithm was assessed. Results Among 41 961 veterans with OUD-ID and 46 936 with StUD-ID, the mean age was 62.2 and 61.4, respectively. A history of OUD plus hepatitis C virus and either acute upper extremity bacterial skin and skin structure infection or septic arthritis of the upper extremity most accurately identified active IDU with sensitivity (96%), specificity (92%), positive predictive value (73%–79%), negative predictive value (96%), and positive likelihood ratio (12.0). Conclusions We identified multiple algorithms with high diagnostic accuracy for identifying active IDU in veterans with either OUD or StUD. Use of ICD10 algorithms to identify active IDU has particular relevance for assessing ID outcomes and harm reduction efforts within the VA.