1082 — An evaluation of the supporting chart documentation of incident opioid use disorder (OUD) diagnoses via chart review
Lead/Presenter: Benjamin Howell,
COIN - West Haven
All Authors: Howell BA (VA Connecticut, Yale University School of Medicine), Abel E (VA Connecticut, Yale University School of Medicine), Edmond SN (VA Connecticut, Yale University School of Medicine) Becker WC (VA Connecticut, Yale University School of Medicine)
Improvement in the proportion of veterans with OUD receiving medications for addiction treatment (MAT) is a priority within the Veterans Health Administration (VHA). Our objective is to evaluate the documentation supporting an incident OUD diagnosis via chart review and to assess characteristics associated with misclassification.
We identified all patients at 3 VISN1 medical centers with an incident OUD diagnosis, defined as encounter an ICD-10 code of opioid abuse or opioid dependence (F11.11* or F11.2*) between October 1, 2016 and June 1, 2018, and no such code in the prior year. We then randomly selected 30 patients from each medical center to perform chart review with a standardized extraction tool. Two extractors per chart will evaluate clinical notes and data within a 30-day pre- and 90-day post diagnosis window. Variables of interest include clinic location, diagnosing specialty, opioid prescriptions, and referral to substance use disorder treatment. We will assess the sufficiency and internal consistency of supporting documentation and the likelihood of an accurate diagnosis.
In the 20-month observation period, we identified 1,357 Veterans (Connecticut: 358, Bedford: 493, Manchester: 206) with an incident diagnosis of OUD: 94% male with a mean age of 50. Initial diagnoses were made in a variety of settings: the most common being substance use treatment (26%), mental health treatment (22%), inpatient (19%), primary care (10%), and emergency/urgent care (6%). Preliminary chart review suggests a source of misclassification may be clinicians using opioid dependence (F11.2*) to indicate physiologic opioid dependence and not OUD.
Incident OUD diagnoses occur in a variety of settings within VHA, many where clinicians are well-trained in identifying diagnostic criteria (ie substance use treatment), but a majority in settings where clinicians may be less familiar with diagnostic criteria. We anticipate chart review will provide insight into how clinicians in different settings document OUD diagnoses, likelihood of an accurate diagnosis, and treatment patterns after incident diagnoses.
We anticipate better understanding the characteristics and settings of encounters of new diagnoses of OUD that inform VHA quality metrics. We anticipate this information will lead to potential systems changes that improve the accuracy of OUD diagnoses.