Implementing Routine Screening for Cannabis and Other Drug Use Disorders in Primary Care: Impact on Diagnosis and Treatment in a Randomized Pragmatic Trial in 22 Clinics
Implementing Routine Screening for Cannabis and Other Drug Use Disorders in Primary Care: Impact on Diagnosis and Treatment in a Randomized Pragmatic Trial in 22 Clinics
Our proposed evaluation study is designed to evaluate the impact of a recently completed stepped wedge cluster randomized trial, conducted at Kaiser Permanente Washington (KPWA), of an intervention to improve care and management of patients with drug use disorders (DUDs) in primary care (Aim 1). We will also explore the reasons for any apparent gaps in DUD care by analyzing clinicians' free-text encounter notes using manual chart review, natural language processing (NLP), and/or NLP-assisted manual chart review, as appropriate (Aim 2).
Specific Project Aims are as follows:
Aim 1 The primary research question we address in Aim 1 is whether routine screening for drug use disorders in primary care (PC) settings increases DUD treatment. We define DUDs as including opioid use disorders (OUD), cannabis use disorders (CUD), and other non-alcohol drug use disorders (OTH). Previously published analyses indicate that the 22 PC clinics in this trial sustained very high rates of screening (88%) and a 3-clinic DUD pilot study suggested that this screening resulted in increased diagnosis of CUD and increased treatment of DUDs in general, even at relatively low observed rates of PC-based screening and assessment.
Aim 2
The overall goal of Aim 2 is to expand our understanding of gaps in DUD diagnosis and treatment that persist-despite implementation of high rates of PC screening and assessment for SUDs-using rich information available only in free-text chart notes. Through analysis of relevant chart notes Aim 2 of this project will descriptively characterize gaps in DUD diagnosis and DUD treatment (i.e., instances where information in a patient's record suggests a DUD could be diagnosed but no diagnosis is present, or a new diagnosis suggest treatment is indicated but no evidence of treatment is present), and characterize reasons for DUD care gaps.
Prior to 2015, KPWA's Behavioral Health Service (BHS) leaders decided to implement an intervention to improve the quality of population-based primary care of several behavioral health conditions including depression and unhealthy use of alcohol, opioids, cannabis, and other drugs. Additionally, these leaders decided to transition primary care social workers to become integrated behavioral health clinicians in 2015.
Pilot testing of quality improvement implementation strategies in 2015 was led by BHS leaders in collaboration with other KPWA departments. State-of-the-art implementation methods were used to integrate evidence-based alcohol-related care into 3 pilot primary care clinics. Implementation strategies included: participatory design, clinical champions, practice facilitation, performance monitoring and feedback, and clinical decision support in the electronic health record (EHR). To encourage care team members to routinely engage patients in discussions of unhealthy substance use while avoiding stigmatized language, a video and informational handout was designed and used in an effort to shift caregiver attitudes and practices. Screening and follow-up assessment for symptoms of all substance use disorders are conducted on paper and entered into the KPWA Epic EHR by medical assistants. The implementation strategy was refined through the use of ongoing formative evaluation.
Timing of the implementation of the quality improvement intervention was staggered to 1) allow adequate support of care teams by available care practice facilitators, and 2) facilitate rigorous evaluation of the impact on patient care of implementing the intervention using secondary quality improvement data.
Research teams at Kaiser Permanente Washington Health Research Institute (KPWHRI) supported the implementation effort and have lead or are leading various components of the evaluation. All evaluations are based on a pragmatic stepped-wedge quality improvement randomized trial design for 22 KPWA clinics. Three pairs of these clinics were combined for purposes of implementation and evaluation, resulting in a set of 19 implementation "sites." Implementation was staggered in 7 waves over a three-year period, with each wave including 2 or 3 sites (3 waves in the first year and 4 waves in each of the second and third years). Randomization was also stratified by year. For pragmatic reasons, KPWA clinical leaders designated 9 clinics that were randomized in the first year (as 9 sites). The remaining 13 clinics were randomized in the second and third years (as 10 sites).
The implementation period in each clinic lasted 4 months.
Our definitions of drug use disorder treatment are modified versions of definitions published in the Healthcare Effectiveness Data and Information Set (HEDIS).
Inclusion Criteria:
Exclusion Criteria: