Human reviewers could identify roughly 70 percent more patients with an opioid use disorder (OUD) than an EHR algorithm.
Human reviewers can identify patients with an opioid use disorder (OUD) more consistently than EHR natural language processing, even if EHR diagnostic codes cannot define the diagnosis, according to a study published in JAMA Network Open.
Additionally, researchers found the benefit of coding for Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition) (DSM-5) criteria from EHRs to generate a DSM-5 severity score that is further associated with OUD patients.
According to researchers, opioid addiction rates and opioid-related deaths have been increasing since 2012. However, OUD is commonly undiagnosed or misdiagnosed during patient-physician consultation.
Want to publish your own articles on DistilINFO Publications?
Send us an email, we will get in touch with you.
Physicians commonly utilize questionnaires from DSM-5 OUD criteria or dialogue, based on impairment, opioid cravings, and high levels of usage. If a patient hits two of the 11 criteria, then the patient is diagnosed with an OUD. However, these criteria typically revolve around self-reports from patients.
“The goal of this research is to use the comprehensive EHR data of patients who are prescribed opioids to develop proxy measures for the DSM-5 criteria for OUD,” explained the study authors.
Researchers from Geisinger utilized EHR data from over 16,000 patients who were prescribed opioids between 2000 and 2017, and a contract-based, Geisinger-specific medication monitoring program (GMMP) to find a connection between those who maintained or violated the contract.
Of the more 16,253 patients enrolled in GMMP, OUD diagnoses defined by EHR diagnostic codes were only observed at 2 percent, meaning researchers need to find alternative strategies.
On the contrary, researchers who assessed OUD through a manual EHR review, identified severe OUD 73 percent of the time.
“We found that patients enrolled in a contract-based, health system–specific drug monitoring program showed higher rates of OUD based on a medical record review procedure that adapted DSM-5 interview criteria,” wrote the study authors.
“We also observed that when patients are appropriately documented as having violated the terms of the contract with an EPIC code, this code can be a useful proxy for OUD diagnosis. This finding is consistent with previous work that demonstrated the utility of a prescription monitoring program.”
Researchers identified inconsistent EHR documentation as a primary cause of the EHR algorithm’s inaccuracies. Additionally, the research group noted a human reviewer could dissect the difference between search times, while an algorithm has difficulty doing the same.
For example, a human reviewer can assess the terms of “abuse” and “high” as separate words, while an algorithm is searching for the term “substance abuse mentioned.”
“These search terms could serve as the basis of future natural language processing algorithms and would improve the scalability of this method,” explained the Geisinger researchers. “Future work may also benefit from combining search terms and ICD codes, as Carell et al reported that the combined use of ICD codes and natural language processing data were more effective in identifying OUD than either method alone.”
Researchers also observed additional substance and psychiatric codes in the EHR prescription drug monitoring program. These codes can evaluate definitions and identify risk factors. Using these terms, human reviewers can notify OUD associated with overdose and misuse.
“Thus, this study contributes to the growing body of knowledge that emphasizes the utility of EHRs to evaluate a patient’s status or potential for opioid or other substance misuse,” concluded the study authors.
“Precision medicine within integrated health systems such as Geisinger could be a major associated factor in developing more efficient pain treatments with less risk for addiction, and studies of this potential could be helped by establishing more effective proxy measures for OUD using EHR data.”
Source: EHR Intelligence