Abstract

PURPOSE There is a critical need to treat opioid use disorder (OUD) in primary care. We describe the incidence of OUD medication treatment among primary care patients who reported opioid use and moderate or severe symptoms of substance use disorder (SUD), as defined by the Diagnostic and Statistical Manual of Mental Illnesses, Fifth Edition, Text Revision (DSM-5-TR), during routine care.

METHOD This retrospective cohort study used electronic health record and insurance claims data from 33 primary care clinics in Washington that routinely screen for substance use and ask patients who report daily cannabis use or any past-year drug use to complete a DSM-5-TR Substance Use Symptom Checklist (Checklist). The sample included 1,502 adult primary care patients (from March 1, 2015 to January 1, 2023) who completed a Checklist, reported past-year opioid use, and had no recent OUD treatment. Primary outcomes were OUD medication treatment within 14 days of completing the Checklist (ie, initiation), and in the following 34 days (ie, engagement).

RESULTS Among 80 (5%) patients with moderate symptoms, 8 (10%) initiated and 6 (8%) remained engaged with medication treatment. These patients were significantly more likely to initiate (P < .001) and remain engaged (P = .003) compared with the 746 (50%) reporting no SUD symptoms. Among 542 (36%) patients with severe symptoms, 141 (26%) initiated and 108 (20%) engaged. These patients were also significantly more likely to initiate (P <.001) and remain engaged (P <.001) compared with those with no SUD symptoms (P = .003) or moderate SUD symptoms (P = .009).

CONCLUSION Most primary care patients reporting opioid use and moderate or severe SUD symptoms did not initiate OUD treatment, but most who initiated remained engaged. Screening and assessment alone is insufficient to result in adequate OUD medication treatment.

Key words:INTRODUCTION

The estimated prevalence of opioid use disorder (OUD) in the United States is 2% which translates to 5.7 million people.1 In primary care, the prevalence of documented OUD has been as high as 5%.2,3 The National Academies of Science, Engineering, and Medicine have recently highlighted the critical importance of OUD medication treatment in saving lives,4 yet less than one-third of people with OUD receive evidence-based treatment with medications (eg, buprenorphine, methadone, injectable naltrexone).5,6 Previous studies suggest that those with more severe OUD symptoms may be more likely to receive treatment with medication.5 It is unknown how often primary care patients receive medication treatment after reporting OUD symptoms on measures integrated into routine care.

Kaiser Permanente Washington, a large integrated health system, implemented a behavioral health screening questionnaire in primary care clinics beginning in 2015 that included 2 items for use of cannabis and other drugs. Patients who report high risk substance use, defined as daily use of cannabis or any other drug, are asked to complete a Diagnostic and Statistical Manual Fifth Edition, Text Revision (DSM-5-TR)7 Substance Use Symptom Checklist (Checklist) that assesses substance use disorder (SUD) criteria.8,9 This process of routinely assessing for substance use and SUD symptoms offers the potential to increase the identification of patients with SUDs, including those with OUD, who may benefit from treatment with medication.

The purpose of this study was to understand how often primary care patients reporting opioid use and moderate or severe SUD symptoms on the Checklist received OUD treatment with medication. Specifically, we described the incidence of initiation and engagement in treatment with medication among primary care patients who report opioid use and SUD symptoms, and the association between the severity of SUD symptoms and OUD medication treatment. Our hypothesis was that patients who reported more severe SUD symptoms would be more likely to receive treatment with medication than those with less severe SUD symptoms.

METHODS

This study was approved by the Kaiser Permanente Washington Institutional Review Board.

Setting

This retrospective cohort study used data from electronic health records (EHR) and insurance claims from 33 Kaiser Permanente primary care clinics in Washington state. In these clinics, primary care patients complete an annual previsit behavioral health questionnaire (via the patient EHR portal before the visit, or in person on paper or a tablet) that includes a single-item assessment instrument for past-year cannabis use (medical or nonmedical)10 and another for past-year use of any other drug.11 Both have the response options of never, less than monthly, monthly, weekly, and daily or almost daily.12 Patients are asked to complete a Checklist if they report past-year daily use of cannabis or any other drug (automatically via the EHR portal, or in person on a tablet, or by a medical assistant). If a paper Checklist was completed, the Checklist was given to the clinician. If a Checklist was completed electronically (via the patient EHR portal), there was no EHR prompt for clinicians to review results of the Checklist. If a clinician documented an SUD diagnosis with International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) code, they received a best-practice alert in the EHR, prompting them to initiate treatment within 14 days.

Sample

This study includes all adult patients with a primary care visit from March 1, 2015 to January 1, 2023 who reported past-year opioid use on the Checklist as part of routine medical care on the day of a primary care visit or within 14 days before the visit. Patients were excluded if they received OUD medication within 30 days before completing the Checklist (ie, already engaged in treatment), and if they were not enrolled in the health plan for at least 75% of the 48-day period after completing the Checklist (to allow for capture of outcomes). We did not exclude patients who reported using substances in addition to opioids, therefore symptoms reported on the Checklist could not be attributed to a specific substance. If patients completed multiple eligible Checklists, the Checklist closest to the primary care visit was selected.

MeasuresSUD Severity

Severity of SUD was assessed based on the number of DSM-5-TR SUD criteria that were reported on the Checklist. The Checklist of 11 SUD symptoms asks patients which they have experienced in the past year and yields scores of 0 through 11. The Checklist also asks what substances a patient uses but does not ask patients to attribute SUD symptoms to any specific substance. Previous studies of the Checklist have demonstrated its reliability and unidimensionality consistent with DSM-5-TR SUD9,13 and high test-retest reliability (ICC = 0.8 for number of symptoms endorsed) in primary care.8 Consistent with DSM-5-TR, SUD severity was classified based on the number of SUD symptoms reported: no SUD (0-1), mild (2-3), moderate (4-5), and severe (6-11). Patients were also categorized into those who reported only opioid use and those who reported cannabis or other drugs in addition to opioid use on the Checklist.

OUD Medication Treatment

The primary outcome measure was treatment with medication which including buprenorphine, methadone from an opioid treatment program, or injectable naltrexone, based on pharmacy dispense records, insurance claims for specialty care received outside Kaiser Permanente Washington, and procedure codes within 48 days after the Checklist was completed. Medication initiation was defined as treatment with medication within 14 days of the completing the Checklist, and engagement was defined as additional treatment with medication in the 34 days following initiation, consistent with Healthcare Effectiveness Data and Information Set timeframes.

Other Measures

Demographic characteristics (ie, age, sex, race, and ethnicity) were used to approximate the social conditions that shape lived experiences rather than biologic factors. Demographic information was collected from patients and documented in the EHR by the health care team. Insurance status was obtained from enrollment records. Past-year diagnoses for pain, mental health, and SUD were obtained from ICD-10-CM codes in the EHR and insurance claims. Long-term opioid therapy was based on pharmacy dispensing, and defined as 60 days or more of prescribed opioids in a 90-day period.

Analyses

We described demographics, clinical characteristics, and the incidence of OUD medication treatment initiation and engagement by SUD symptom severity category using descriptive statistics. We used logistic regression to characterize the association between SUD severity and OUD medication treatment initiation and engagement, comparing patients reporting moderate or severe SUD symptoms to patients reporting no SUD symptoms and patients with severe symptoms to patients with moderate symptoms; P values for these associations were obtained from Wald tests. Logistic regression was selected a priori as the modeling approach because the outcome was binary.14 We chose to focus on patients with moderate and severe SUD symptoms because these are the patients for whom medication treatment is specifically indicated. From regression models, we estimated predicted probabilities of treatment initiation and engagement with 95% CIs using recycled predictions. Main analyses were unadjusted while secondary analyses adjusted for sociodemographics (ie, age, sex, Hispanic ethnicity, and insurance status). Unadjusted models were considered primary because adjustment for sociodemographic characteristics can control for lived experiences in the causal pathway to receipt of OUD medication.15 Secondary analyses stratified findings by patient report on the Checklist of only opioid use or opioid use with cannabis or other drugs. We conducted sensitivity analyses of main models using robust SEs to understand whether inferences were robust when using SEs that are not dependent on correct model specification.

RESULTS

A total of 1,502 primary care patients completed the Checklist, reported past-year opioid use and were eligible for the study (Figure 1). Table 1 describes the demographic and clinical characteristics of patients included in the sample, overall and across substance use severity levels based on Checklist scores. The sample was 52.9% male, 76.8% White, 6.1% Hispanic, and 57.6% commercially insured. Most patients reported other substance use in addition to opioid use: 858 (57.1%) reported past-year cannabis use, and 327 (21.8%) reported past-year stimulant use. Only 466 patients (31.0%) reported past-year opioid use and no other substance use. Over one-half of the sample had a past-year pain diagnosis. The prevalence of past-year pain was highest in the group who reported no SUD symptoms and was progressively lower with increased severity of reported SUD symptoms. About one-half of the sample had a past-year mental health disorder that was consistent across SUD severity categories. One-third of the sample had a past-year SUD diagnosis based on clinician documentation with ICD-10-CM codes that was progressively higher with severity of SUD symptoms with prevelence ranging from 15.8% for those reporting no SUD symptoms to 60.1% for those reporting severe SUD symptoms. Just over one-fourth of the sample had a past-year OUD diagnosis based on clinician documented ICD-10-CM codes that was progressively higher with severity of SUD symptoms with prevelence ranging from 10.7% in those reporting no SUD symptoms to 50.7% in those reporting severe SUD symptoms. The Checklist was completed via the EHR portal by 515 (34.3%) of patients.

Figure 1.

Study Sample

Checklist = DSM-5-TR Substance Use Symptom Checklist; DSM-5-TR = Diagnostic and Statistical Manual of Mental Illnesses, Fifth Edition, Text Revision; OUD = opioid use disorder; SUD = substance use disorder.

Table 1.

Characteristics of Primary Care Patients Who Completed Checklist and Reported Opioid Use by Symptom Category

Most patients who reported opioid use and symptoms consistent with moderate to severe SUD did not receive OUD medication treatment, but most patients who initiated treatment remained engaged the next month. In the total sample, 167 patients (11.1%) received medication within 14 days, and of those 131 (8.7% of the total sample, 78.4% of those who initiated medication) remained engaged in the following 34 days. Among patients who received medication, 95.8% received buprenorphine and 4.2% received injectable naltrexone. None of the patients in the sample had received methadone, although 22 of 23 (95.7%) of the patients who were excluded from the study sample due to prior OUD medication treatment had received methadone in the 30 days before completing the Checklist.

Initiation and engagement in medication treatment differed markedly across SUD symptom severity categories (Table 2, Figure 2). A total of 80 patients (5.3% of total sample) reported moderate SUD symptoms. The patients with moderate SUD symptoms, of whom only 10.0% initiated and 7.5% remained engaged (75.0% of those initiating) (Table 2), were significantly more likely to initiate OUD medication treatment (P < .001) and remain engaged (P = .003) compared with patients with no SUD symptoms. A total of 542 patients (36.1% of total sample) reported severe SUD symptoms. The patients with severe SUD symptoms, of whom 25.8% initiated and 20.3% remained engaged (78.6% of those initiating) (Table 2), were significantly more likely to initiate OUD medication treatment (P <.001) and remain engaged (P <.001) compared with patients with no SUD symptoms (P = .003, Supplemental Table 1), and compared with patients with moderate SUD symptoms (P = .009).

Table 2.

Initiation and Engagement in Opioid Use Disorder Medication by SUD Symptom Severity Reported by Patients on Checklist

Figure 2.

Percent of Patients Who Initiated and Engaged in Medication for Opioid Use Disorder Stratified by Severity of SUD Symptoms

SUD = substance use disorder.

Adjustment for sociodemographic characteristics did not meaningfully change results (Table 2). Sensitivity analyses that used robust SEs did not change inference (Supplemental Table 1). In secondary analyses, stratified on whether patients reported use of cannabis or other substances in addition to opioids, OUD medication treatment initiation appeared higher in patients with severe symptoms who reported only opioid use (31% vs 23%, Supplemental Table 2).

DISCUSSION

In a sample of primary care patients who reported opioid use (with or without other substance use) on the Checklist, most patients who reported moderate or severe SUD symptoms did not receive OUD medication treatment. Those who initiated OUD medication treatment, however, tended to remain engaged in the following month which is consistent with other studies of OUD medication retention in primary care.16 This finding suggests that screening and assessment of SUD symptoms and associated substance use alone is insufficient to result in adequate OUD treatment of patients reporting opioid use and moderate to severe SUD symptoms. In order in increase uptake of medication for OUD in primary care, we must understand why more patients are not starting OUD medication. We must examine and address barriers to starting medication, especially for patients who access primary care services and report SUD symptoms associated with opioid use to their primary care clinician.

Initiation and engagement in OUD medication treatment was lower than other rates of OUD treatment reported in the literature.2,17-19 This may reflect the fact that our sample was derived from patient-reported SUD symptoms and opioids use rather than clinician diagnosed OUD. This could indicate that epidemiologic studies that rely on documented ICD-10-CM codes for OUD underestimate the number of patients with OUD, and therefore overestimate rates of medication treatment.

Previous literature indicates patients with more severe OUD symptoms have a higher likelihood of receiving treatment with medication.5 Our results suggest that this is also true in a primary care sample which includes non–treatment seeking patients who may be excluded in randomized controlled trials or other primary data collection studies. These findings could reflect the fact that patients with severe symptoms are more likely to want treatment for OUD, and are more likely to initiate medications. The Checklist may facilitate discussions between patients and clinicians about OUD symptoms and treatment options for OUD, which could increase rates of medication initiation. At the same time, in the absence of prompts, this study could not determine whether clinicians saw and addressed the results of screening and assessment, especially when it was completed via the online patient EHR portal. Future research should explore the role of a Substance Use Symptom Checklist in patient and clinician decisions to initiate OUD treatment with medications, and whether clinician prompts in the EHR could increase OUD treatment.

While screening for OUD in primary care has not been shown to meaningfully increase rates of OUD diagnosis in 1 study,20 screening followed by SUD symptom checklists was implemented to support primary care diagnosis of SUD and as an engagement tool.12,21 Despite routine integration of checklists into primary care, multiple patient, clinician, and health care system barriers may contribute to low rates of initiation of OUD treatment medications. Patient level barriers may include internalized stigma, incorrect information about OUD medication, negative previous treatment experiences, and care that is not patient-centered or does not use shared decision making.22-25, 26 Clinicians may also experience barriers that prevent them from starting patients on medication for OUD such as stigma toward people who use drugs, inadequate education and training, time constraints, and lack of institutional support.22 Health care system factors such as long wait times and lack of multidisciplinary support for patients with OUD may also play a role.22 Larger societal structural issues leading to health care inequities such as racism and criminalization of substance use also likely contribute to low rates of initiation of OUD treatment medications27 Further, treatment of OUD with medications may not be in line with all patients’ values and preferences, and may not occur at the first visit after completion of the Checklist. Future research should explore approaches to overcoming patient, clinician, and health care system barriers to initiating OUD medications.

An important limitation of this study was that for patients who report using multiple drugs, symptoms reported on the Checklist cannot be attributed to a specific substance; thus, we were unable to determine if symptoms were due to opioids or another substance. For example, some patients who reported symptoms consistent with moderate or severe SUD may not have had moderate or severe OUD specifically (ie, some symptoms could be related to another co-occurring SUD). On analyses that stratified by only opioid use vs opioid and other substance use, patients who reported using opioids and other substances were less likely to initiate and engage in treatment, which could reflect some of them having SUD symptoms attributable to another drug. On the other hand, it could also indicate people who use multiple substances experience greater barriers to starting medication for OUD in primary care. We chose not to exclude patients who reported other substance use in addition to opioid use from the sample because they represent the majority of the sample with OUD, and are a population at high risk for overdose. In many situations, it is not possible for patients to isolate which SUD symptoms are due to opioids vs other substances. In addition, it is possible that patients who reported opioid use on the Checklist were reporting prescription opioid use only. While patients are prompted to complete the Checklist if they report daily cannabis use, other past-year drug use, or prescription drug use for nonmedical reasons, it remains possible that patients who used prescription opioids were triggered to complete the checklist due to another substance use and subsequently marked opioid use in addition to that substance.

This study had several other important limitations. The sample included only Kaiser Permanente Washington patients, so results may not generalize to other populations. In this health system, patients could be treated with buprenorphine in primary care, in an addiction recovery services clinic, or in specialty treatment outside of Kaiser Permanente Washington primary care. While this study captured outside specialty treatment using insurance claims, any treatment that did not require insurance reimbursement (eg, free clinic, patients paid for out-of-pocket) would have been missed. Our study took place during a period when there was a shift in the non-prescribed drug supply to predominantly high potency synthetic opioids such as fentanyl which may have impacted buprenorphine prescribing practices as well as patient experiences. Our study ended at the time the federal waiver to prescribe buprenorphine was removed so we have not captured buprenorphine prescribing that occurred after this policy change. The Checklist is reliable, discriminates SUD severity consistent with current definitions, and performs equitably across age, sex, race, and ethnicity,8,9,13 but does not ask about recurrence of symptoms, required for DSM-5-TR SUD diagnosis. While the questions on the Checklist map onto the 11 DSM-5-TR SUD criteria used to determine SUD severity, the Checklist’s sensitivity and specificity for detecting specific levels of DSM-5-TR OUD severity (eg, moderate, severe) on a gold standard measure has not been defined. Patients may have underreported symptoms on the Checklist—particularly those on long-term opioids (27%) who may have been concerned that reporting symptoms would impact their access to medications.

This study showed that most patients who reported opioid use and symptoms consistent with moderate or severe DSM-5-TR SUD via routine assessment before a primary care visit did not receive OUD medication treatment, while most who initiated OUD medication treatment remained engaged in the following 34 days. While patients are willing to report use of opioids and SUD symptoms on the Checklist, use of the Checklist will likely need to be paired with robust implementation strategies28,29 and other proactive, patient-centered, population-based systems to engage patients in medication treatment for OUD. Future research should further examine additional interventions to increase initiation and engagement with OUD medication treatment in primary care.

Footnotes

Conflicts of interest: authors report none.

Read or post commentaries in response to this article.

Funding support: Research reported in this abstract was supported by the National Institute on Drug Abuse of the National Institutes of Health under Award Numbers UG1DA040314 (CTN-0113) and R25DA033211, and by the National Institute of Mental Health of the National Institutes of Health under Award Number T32 MH020021. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Previous presentations: 2023 North American Primary Care Research Group Annual Meeting; October 30–November 3, 2023; San Francisco, California, and the 2024 College on Problems of Drug Dependence Annual Meeting, June 15-19, 2024; Montreal, QC, Canada.

Supplemental materials

Received for publication September 7, 2024.Revision received January 4, 2025.Accepted for publication February 10, 2025.© 2025 Annals of Family Medicine, Inc.