Making Measurement Based Care for Addictions a Reality in Primary Care

Abstract

Three factors are important to consider when integrating measurement-based care for opioid use disorders into primary care: integration with other behavioral health and substance use disorders care, the availability of a brief, valid measure that is responsive to change, and implementation in a manner that fosters accurate reporting.

We applaud Dr. Marsden and colleagues’ thoughtful paper on measurement-based care (MBC) for opioid use disorders (OUDs).¹ Development of systems that support MBC for substance use disorders (SUDs) is critical to improving SUD outcomes.² We agree that systems are needed that can practically monitor response to treatment not only in specialty settings, but also in primary care (PC) where SUDs are increasingly treated and followed.³ Although there is currently no validated, practical measure for monitoring change over time in PC settings, prior experiences implementing other substance use measures in PC suggest 3 important lessons.

First, sustainable implementation of MBC for OUDs in PC will require that it be integrated with MBC for common behavioral health conditions² and assess SUDs broadly rather than OUD alone. For example, when Kaiser Permanente (KP) Washington decided to implement routine alcohol screening and assessment of DSM-5 alcohol use disorder (AUD) symptoms in an implementation trial, clinical leaders asked to incorporate screenings for depression, cannabis and other drug use, and to assess suicidality and DSM-5 symptoms of SUDs in high risk patients.^4,5 As a result, MBC for SUDs was integrated into routine workflows with other behavioral health screenings and assessments and was widely embraced by PC clinicians. Medical assistants enter results of screenings and assessments into EHRs before PC providers enter exam rooms, and PC providers are trained to ask open-ended questions about any reported SUD symptoms to initiate patient-centered discussions. Of note, it was not feasible to have separate DSM-5 assessments for each substance, so there are two DSM-5 “symptom checklists” to support diagnosis—one for AUD and one for other SUDs. Both assess symptoms (yes/no) over the past-year, consistent with DSM-5.⁶ Efforts to identify symptoms associated with specific substances were unsuccessful; many patients use multiple substances and cannot readily link a single substance to their symptoms. While these assessments are not optimal for treatment monitoring due to their 1-year timeframe, they suggest that routine use of SUD measures in PC is feasible if integrated with other behavioral health care.

Second, the optimal instrument for MBC of SUDs in PC will likely be brief and responsive to change over time, like the 9-item Patient Health Questionnaire (PHQ-9) used for depression MBC, which asks about symptoms in the past 2 weeks.^7–9 We suspect that a DSM-5 SUD symptom checklist with a 3-month timeframe, as recommend by Marsden et.al.¹ and others,¹⁰ may be too long and not responsive enough to change to be useful clinically. For example, KP Washington uses a single behavioral health “Monitoring Tool” in PC, mental health clinics, and urgent care that includes the PHQ-9, GAD-2¹¹ and 5 questions about substance use,^12–14 but it was not feasible to include the 22 questions—11 each for AUD and SUD symptoms—that are used for initial AUD and SUD assessment. Recognizing the need for brevity, as well as for a short timeframe for monitoring patients early in treatment, a recent expert panel on substance use management in PC proposed two 5-item instruments along with 5 screening questions about substance use¹⁰ to monitor AUD and SUD symptoms in patients treated for SUDs. A Short Alcohol Monitor and Short Drug Use Monitor¹⁰ were built on a global SUD patient-reported outcome measure¹⁵ as well as common symptoms, like craving, from the rigorously curated PROMIS item databanks,^16,17 using PHQ-9 response options. These or similar brief scaled measures asking about recent symptoms warrant more study for monitoring treatment of SUDs in PC.

Finally, implementing MBC in ways that promote accurate patient reporting will be important. After VA’s implementation of routine alcohol screening in PC nationwide, our research suggested substantial underreporting: 61% of patients who screened positive on confidential VA surveys had a negative screen documented in their EHRs within 90 days of the survey.¹⁸ Many factors likely contributed, including the use of verbal interviews for screening, variability in screening procedures, and stigma.¹⁹ Patients who are experiencing OUD symptoms may similarly under-report due to stigma or shame. Using standardized methods that reduce bias in MBC reporting, such as having patients directly report their substance use and SUD symptoms onto paper or tablets, may improve the accuracy and value of monitoring over time.

In conclusion, research is needed to understand whether a DSM-5 symptom checklist with a 3-month timeframe is practical and useful in routine PC practice. Research is also needed to explore briefer measures, with shorter timeframes which might provide as much useful information for MBC both early in treatment and late —while decreasing the burden of MBC implementation.