Statement of purpose In the US, over 130 lives are lost from opioid-related drug overdoses each day. Accurate and prompt diagnosis of opioid use disorders (OUD) may contribute substantially to prevention of overdose deaths. However, OUD research is limited because ICD codes used to identify OUD are specific but not sensitive, resulting in underestimation OUD prevalence. Further, the specificity and sensitivity of the OUD ICD codes is unknown due to the lack of a gold standard.

Methods/Approach In this study, eight pain and addiction medicine clinical experts across four large healthcare systems in two southern United States (two experts per healthcare system) reviewed 166 electronic medical records (EMR) from 2014–2017 which may or may not have been assigned OUD ICD codes. The experts relied on Diagnostic and Statistical Manual of Mental Disorders-5 criteria for making OUD diagnoses. These ‘gold standard’ charts were used to 1) develop and validate a highly sensitive OUD identification algorithm over four iterations and 2) examine the sensitivity and specificity of OUD ICD codes.

Results Of the 166 EMR charts, 100 (60%) were reviewed by two experts and exhibited 83% agreement. After four iterations, the algorithm identified OUD with >90% sensitivity and 67% specificity. The OUD ICD codes had 59% sensitivity and 93% specificity, underscoring the extent underestimation.

Conclusions This is the first study to measure the sensitivity and specificity of OUD ICD codes and develop an algorithm to address OUD underestimation. This work can be used in future research on OUD prevention and management.

Significance Strategies to address the opioid epidemic have primarily focused on prevention of overdose deaths. Underdiagnoses of OUD may contribute to frequent failure in connecting patients to treatment. From an injury pyramid perspective, preventing OUD development or escalation would reduce overdoses and improve quality of life for patients with OUD.