QUALITY OF BUPRENORPHINE CARE FOR INSURED ADULTS WITH OPIOID USE DISORDER

Abstract

Aims:

To characterize quality of buprenorphine care for opioid use disorder by quantifying buprenorphine initiation, engagement, and maintenance for individuals in a large, diverse, real-world cohort in the United States.

Design:

Retrospective cohort analysis.

Setting:

Opioid use disorder treatment in the outpatient setting.

Participants:

45,210 commercially insured and Medicare Advantage enrollees 18 years or older in the OptumLabs® Data Warehouse with an index diagnosis of opioid use disorder between 1/1/2018 and 12/31/2018.

Interventions:

Treatment with buprenorphine.

Measurements:

We calculated 6 measures of buprenorphine treatment quality. We conducted survival analyses to characterize treatment duration and logistic regressions to evaluate the association between clinical and sociodemographic characteristics and quality.

Findings:

Of 45,210 eligible individuals with opioid use disorder, approximately 1 in 10 (n=4,600, 10.2%) initiated buprenorphine within 365 days following diagnosis (Measure #1) and 2,850 individuals (6.3%) initiated buprenorphine within 14 days of diagnosis (Measure #2). Of individuals initiating treatment within 14 days of diagnosis, 1,769 (62.1%) had 2 or more buprenorphine claims within 34 days of initiation (Measure #3). Of the 4,600 individuals who received buprenorphine, 2,300 (50.0%) were maintained in care with 180 days or more of covered buprenorphine treatment during 365 days after diagnosis (Measure #4). Finally, of the 4,600 individuals who received buprenorphine, 2,543 (55.3%) did not fill any other concurrent opioid analgesic (Measure #5) and 2,951 (64.2%) did not fill any concurrent benzodiazepine (Measure #6). Quality was generally lower for individuals with Medicare Advantage compared to commercial coverage and among Hispanic and Black adults compared to white adults.

Conclusions:

Widespread gaps exist in quality of buprenorphine treatment initiation, engagement, and maintenance among commercially insured and Medicare Advantage enrollees with opioid use disorder.

INTRODUCTION

Opioid overdose from prescription and illegal opioids is now the leading cause of unintentional fatal injury in persons ages 25 through 64,¹ and more than two million Americans are estimated to have past-year opioid use disorder (OUD).² There are three FDA-approved medicines to treat OUD. Of these products, buprenorphine has unique advantages that facilitate its use in ambulatory settings.³ In contrast to naltrexone, an opioid antagonist, buprenorphine does not require complete opioid abstinence before initiation. In contrast to methadone, an opioid agonist, buprenorphine’s use is not restricted under federal law to specialized opioid treatment programs. However, clinicians must undergo training and receive a waiver to prescribe buprenorphine. High-quality, effective buprenorphine treatment includes continuous medication use and the careful consideration of any pharmacotherapy for pain or anxiety, as it may require careful management or a pause in buprenorphine treatment.^4,5

Despite the unequivocal value of medications for opioid use disorder, there are widespread and enduring gaps in their use.^6,7,8,9 Challenges initiating and continuing buprenorphine treatment have been further exacerbated by the COVID-19 pandemic.¹⁰ In one study examining OUD treatment in a single state between 2012–2014, fewer than one third of individuals with a non-fatal overdose received pharmacologic treatment for OUD over the subsequent year.⁸ Many other investigations support these findings and demonstrate the widespread underutilization of medications for OUD.^{11,12,13,14,15} Even when initiated, treatment courses are variable, interruptions are frequent, and OUD treatment is often punctuated by use of other opioids.^16,17,18

During the past several years, significant investments have been made to expand the treatment infrastructure for OUD in the United States, yet it is unclear if these have improved OUD quality of care. Also, most prior investigations have examined limited dimensions of quality, yet care for OUD requires a “cascade of care” that can be impacted by strategies at clinical, organizational, and policy levels.¹⁹ Additionally, there is limited evidence on the relationship between factors such as race/ethnicity, education level, presence of comorbidities, and insurance type on continued engagement with buprenorphine treatment.

We used recent data to calculate six measures along the cascade of care for a diverse cohort of individuals with OUD with commercial and Medicare Advantage (MA) insurance. In addition to characterizing treatment initiation, engagement and maintenance, we also quantified the association between these measures of quality and individuals’ demographic and clinical characteristics.

METHODS

Data

We performed a retrospective cohort study of data from the OptumLabs® Data Warehouse (OLDW), which includes de-identified claims data for commercially insured and MA enrollees. The database contains sociodemographic and longitudinal information regarding inpatient, outpatient, and prescription drug utilization, representing a diverse mixture of ages, races/ethnicities, and geographical regions across the United States (eAppendix 1).

Study Cohort

We identified 98,720 individuals with an OUD diagnosis (“index diagnosis”) in the inpatient, emergency department, or outpatient setting between 1/1/18 and 12/31/18 (eFigure 1). From this cohort, we excluded 36,647 individuals who had a gap longer than 45 days in medical or pharmacy coverage during the 90 days before and 12 months after their index diagnosis. We excluded 11,567 cases because an individual had a diagnosis of OUD during a 90-day “washout period” prior to their index date. We also excluded 5,098 cases where the index diagnosis occurred in the outpatient setting without a subsequent outpatient OUD claim.^20,21 Finally, we excluded 198 cases where the individual was less than 18 years. Our final analytic sample included 45,210 individuals. For this sample, we extracted claims from 1/1/2018 through 12/31/2019 to ensure 12 months of data following index diagnosis.

Analysis

We first examined the sociodemographic and clinical characteristics of our study cohort. We then characterized the quality of buprenorphine care based on six measures depicted in Table 1 and described below:

• Measure #1 (365-Day Initiation): Percentage of analytic sample with one or more claims for buprenorphine within 365 days of index diagnosis

• Measure #2 (14-Day Initiation): Percentage of analytic sample with one or more claims for buprenorphine within 14 days of index diagnosis

• Measure #3 (Engagement): Percentage of analytic sample initiating buprenorphine within 14 days of index diagnosis who had 2 or more claims for buprenorphine within 34 days of initiation

• Measure #4 (Maintenance): Percentage of analytic sample with 180 days or more of covered buprenorphine treatment in the 365 days following index diagnosis among those who received buprenorphine

• Measure #5 (No Opioid Co-Prescribing): Percentage of analytic sample who initiated buprenorphine within 365 days of index diagnosis who do not fill a prescription for another opioid analgesic within 180 days following buprenorphine treatment

• Measure #6 (No Benzodiazepine Co-Prescribing): Percentage of analytic sample who initiated buprenorphine within 365 days of index diagnosis who do not fill a prescription for a benzodiazepine within 180 days following buprenorphine treatment

Table 1.

Quality of Buprenorphine Treatment for OUD in Commercial and Medicare Advantage (MA) Enrollees (n=45,210)

	Numerator	Results for All Individuals w/ OUD	Results for Individuals who Receive Any Buprenorphine	Results for Individuals who Initiate Buprenorphine within 14 Days of Diagnosis
Buprenorphine Treatment Phase (%) – Overall		(n=45,210)	(n=4,600)	(n=2,850)
Measure 1: 365-Day Initiation (Any buprenorphine within 365 days of index diagnosis)	4,600	10.2%	---	---
Measure 2: 14-Day Initiation (Buprenorphine within 14 days of index diagnosis)	2,850	6.3%	62.0%	---
Measure 3: Engagement (Two or more buprenorphine claims within 34 days of initiation)	1,769	3.9%	38.5%	62.1%
Measure 4: Maintenance (180 days or more of buprenorphine treatment)	2,300	5.1%	50.0%	---
Measure 5: No Opioid Co-Prescribing (No other opioid analgesic within 180 days following buprenorphine treatment)	2,543	---	55.3%	---
Measure 6: No Benzodiazepine Co-Prescribing (No benzodiazepine within 180 days following buprenorphine treatment)	2,951	---	64.2%	---
Buprenorphine Treatment Phase (%) – Commercial Enrollees		(n=14,466)	(n=2,454)	(n=1,579)
Measure 1: 365-Day Initiation (Any buprenorphine within 365 days of index diagnosis)	2,454	17.0%	---	---
Measure 2: 14-Day Initiation (Buprenorphine within 14 days of index diagnosis)	1,579	10.9%	64.3%	---
Measure 3: Engagement (Two or more buprenorphine claims within 34 days of initiation)	1,057	7.3%	43.1%	66.9%
Measure 4: Maintenance (180 days or more of buprenorphine treatment)	1,304	9.0%	53.1%	---
Measure 5: No Opioid Co-Prescribing (No other opioid analgesic within 180 days following buprenorphine treatment)	1,712	---	69.8%	---
Measure 6: No Benzodiazepine Co-Prescribing (No benzodiazepine within 180 days following buprenorphine treatment)	1,733	---	70.6%	---
Buprenorphine Treatment Phase (%) – MA Enrollees		(n=30,744)	(n=2,146)	(n=1,271)
Measure 1: 365-Day Initiation (Any buprenorphine within 365 days of index diagnosis)	2,146	7.0%**	---	---
Measure 2: 14-Day Initiation (Buprenorphine within 14 days of index diagnosis)	1,271	4.1%**	59.2%**	---
Measure 3: Engagement (Two or more buprenorphine claims within 34 days of initiation)	712	2.3%**	33.2%**	56.0%**
Measure 4: Maintenance (180 days or more of buprenorphine treatment)	996	3.2%**	46.4%**	---
Measure 5: No Opioid Co-Prescribing (No other opioid analgesic within 180 days following buprenorphine treatment)	831	---	38.7%**	---
Measure 6: No Benzodiazepine Co-Prescribing (No benzodiazepine within 180 days following buprenorphine treatment)	1,218	---	56.8%**	---

Results for the denominator specified in the measure are in bold; results for other populations are provided for context. See eTable 1 for more details on measure specifications.

^*p≤0.05

^**p≤0.01 statistically significant difference between individuals with MA vs commercial coverage.

We provide additional details on measure specifications and their sources in eTables 1 and 2. We stratified all analyses based on whether an individual was enrolled in a commercial or MA plan to account for differences by insurance market in the coverage of buprenorphine during the period of analysis.^22,23

We plotted Kaplan-Meier survival curves to examine whether duration of buprenorphine treatment varied based on sociodemographic and clinical characteristics. Consistent with the Healthcare Effectiveness Data and Information Set (HEDIS) Pharmacotherapy for Opioid Use Disorder measure,²⁴ we considered individuals to have continuous buprenorphine treatment if they had no longer than a 7-day gap between when a prescription would have been exhausted and when the next prescription was filled. In order to assess the hazard (risk) of discontinuation of buprenorphine treatment, we used multivariate cox regression models. We tested multiple models including sociodemographic and clinical covariates separately for individuals with commercial and MA coverage. We selected the Cox regression models with the lowest Akaike information criteria (AIC) and Bayesian information criteria (BIC) scores and confirmed that the selected models did not violate the proportional hazard assumptions.

Finally, we analyzed the adjusted association of sociodemographic and clinical characteristics with quality of buprenorphine care by calculating adjusted risk ratios using generalized linear models with binomial family and log link function.

RESULTS

Study Population

Table 2 depicts characteristics of the 14,466 commercial enrollees and 30,744 MA enrollees in the study. For example, 47.7% of the commercial enrollees and 58.7% of the MA enrollees were female, and 69.4% of commercial enrollees and 61.7% of MA enrollees were white. Elixhauser comorbidity index scores²⁵ were higher for MA enrollees (6.7, standard deviation [SD] 3.5) compared to commercial enrollees (3.7, SD 2.6). The most common comorbid conditions included hypertension (39.4% of commercial enrollees, 80.5% of MA enrollees), hyperlipidemia (29.1%, 66.4%), arthritis (39.2%, 69.1%), depression (39.6%, 47.7%), anxiety disorder (45.0%, 45.8%), and chronic pain (55.0%, 78.0%).

Table 2.

Characteristics of Study Cohort with OUD Diagnosis (n=45,210)

	Overall (n=45,210)	Commercial Enrollees (n=14,466)	MA Enrollees (n=30,744)
Female, %	55.2%	47.7%	58.7%
Age, %
18–34 Years	9.0%	26.3%	0.8%
35–49 Years	15.8%	32.8%	7.8%
50–64 Years	35.8%	35.9%	35.8%
65–74 Years	24.9%	3.9%	34.8%
≥ 75 Years	14.5%	1.1%	20.8%
Race/ethnicity, %
Non-Hispanic, White	64.1%	69.4%	61.7%
Non-Hispanic, Black	12.2%	7.9%	14.3%
Hispanic	8.0%	7.3%	8.3%
Asian	0.8%	1.2%	0.7%
Missing/unknown	14.8%	14.2%	15.1%
Education, %
≤ High School	35.2%	24.7%	40.1%
Some College	44.5%	50.7%	41.%
≥ Bachelor Degree	7.9%	13.9%	5.1%
Missing/unknown	12.5%	10.7%	13.3%
Household Income, %
<$40,000	29.5%	16.4%	35.7%
$40,000-$74,999	21.5%	20.3%	22.1%
$75,000-$124,999	16.9%	22.2%	14.4%
$125,000-$199,000	6.4%	11.8%	3.9%
$200,000+	3.1%	7.4%	1.1%
Missing/unknown	22.6%	22.0%	22.9%
Elixhauser Comorbidity IndexA, %
Low (0–4)	43.3%	71.7%	30.0%
Medium (5–8)	36.1%	22.4%	42.5%
High (≥ 9)	20.6%	5.9%	27.5%
Elixhauser Comorbidity Index (mean and sd)	5.6 (3.5)	3.7 (2.6)	6.7 (3.5)
Chronic Conditions, %
1+ chronic conditionsB	96.8%	91.0%	99.5%
1+ mental health conditionsC	61.7%	57.6%	63.6%
Hypertension	67.3%	39.4%	80.5%
Hyperlipidemia	54.5%	29.1%	66.4%
CAD	18.9%	5.8%	25.1%
CVA	3.2%	1.3%	4.1%
Asthma	12.9%	10.2%	14.2%
COPD	28.8%	10.3%	37.5%
Arthritis	59.5%	39.2%	69.1%
Diabetes	31.0%	13.7%	39.1%
Depression	45.1%	39.6%	47.7%
Anxiety Disorder	45.5%	45.0%	45.8%
Bipolar Disorder	11.3%	10.9%	11.5%
Hepatitis C	4.3%	3.1%	4.8%
HIV/AIDS	0.6%	0.4%	0.8%
Alcohol Use Disorder	10.9%	17.0%	8.1%
Chronic Pain	70.6%	55.0%	78.0%

^AElixhauser scores range from 0–30, where a higher score indicates an individual has more comorbid conditions

^BOne or more diagnoses in the categories listed below

^COne or more diagnoses within the categories of depression, bipolar disorder, and anxiety

Quality of Buprenorphine Care

Table 1 depicts the measures of quality of care among individuals receiving buprenorphine:

• Measure #1, reflecting initiation of buprenorphine within 365 days of index diagnosis, was fulfilled among 4,600 of 45,210 eligible individuals with a diagnosis of OUD (10.2%). For the 40,610 individuals who did not receive buprenorphine during the 365 days following index diagnosis, only 2.7% alternatively received naltrexone and/or methadone and 23.6% had claims for psychosocial treatment.²⁶

• Measure #2, reflecting buprenorphine initiation within 14 days of an index OUD diagnosis, was fulfilled among 2,850 individuals, or 6.3% of the eligible sample of 45,210 individuals with a diagnosis of OUD.

• Measure #3, capturing engagement in care and reflected by two or more buprenorphine claims within 34 days of the initiation of buprenorphine, was met among 1,769 of 2,850 individuals who initiated buprenorphine within 14 days of diagnosis (62.1%).

• Measure #4, maintenance of treatment, defined as 180 days or more of buprenorphine treatment within the 365 days following index diagnosis, was fulfilled by 2,300 of 4,600 individuals who received buprenorphine (50.0%). Results were similar when examining whether individuals received 180 days or more of buprenorphine in the 365 days after initial buprenorphine treatment (2,683 of 4,600; 58.3%).

• Measure #5, which assesses the absence of other opioid analgesics within 180 days of receiving buprenorphine, was fulfilled among 2,543 of 4,600 eligible individuals who initiated buprenorphine within 365 days of index diagnosis (55.3%).

• Measure #6, which assesses the absence of benzodiazepines within 180 days of receiving buprenorphine, was fulfilled among 2,951 of 4,600 eligible individuals who initiated buprenorphine within 365 days of index diagnosis (64.2%).

Individuals with commercial coverage were statistically significantly more likely to meet each quality measure compared to those with MA coverage (Table 1). Quality of buprenorphine care also varied by setting of index diagnosis, with worse performance on the quality measures for individuals diagnosed in the inpatient and emergency department settings compared to the outpatient setting (eTable 3).

Treatment Duration

Figure 1 depicts survival curves, which represent the duration of buprenorphine treatment among the 4,600 individuals who initiated buprenorphine within 365 days following an index diagnosis of OUD. The median duration of the first buprenorphine treatment episode was 71 days (interquartile range [IQR]: 29 days to 243 days).

Figure 1. Kaplan Meier Survival Curves of Buprenorphine Treatment Duration for Commercial and Medicare Advantage (MA) Enrollees (n=4,600).

The log-rank tests for the equality of the survival functions indicated there were statistically significant differences based on insurance type (χ²: 3.92, p-value: 0.0478), gender (χ²: 21.90, p-value: <0.0001), race (χ²: 21.98, p-value: 0.0001), and Elixhauser Comorbidity Index Score (χ²: 103.18, p-value: <0.0001). The vertical line at 180 days corresponds with Measure 4: Maintenance (180 days or more of buprenorphine treatment).

In unadjusted models, there were significant differences in the duration of buprenorphine treatment based on individuals’ sociodemographic and clinical characteristics. For example: the median duration of treatment was shorter for women (63 days) than men (79 days); individuals who were Black (59 days) or Hispanic (61 days) compared with their white counterparts (71 days); those covered by Medicare (66 days) compared to individuals covered by a commercial plan (75 days); and for those with higher rather than lower Elixhauser scores. For each category of individuals, there was a substantial drop in retention at 30 days. The most common number of days supply for a buprenorphine prescription in our study was 30 days. The large drop in retention at 30 days reflects individuals who filled an initial buprenorphine prescription who do not access or receive a second prescription.

These differences persisted in the multivariate Cox regressions, where we adjusted for insurance type, gender, race/ethnicity, education, household income, age, Elixhauser comorbidity index score, and the presence of a mental health condition or alcohol use disorder (Table 3). For example, the hazard of treatment discontinuation was 13.4% greater for Black individuals, when compared to white individuals (95% confidence intervals [CI] 1.01–1.27). The hazard of treatment discontinuation was also higher for individuals with more comorbid conditions, mental illness, or alcohol use disorder compared to their counterparts. Among MA enrollees, the hazard of treatment discontinuation was again greater among those with more comorbid conditions.

Table 3.

Cox Proportional Hazards Results for Likelihood of Discontinuing Buprenorphine Treatment (n=4,600)

	Hazard Ratio
Insurance
Commercial	Referent
Medicare Advantage	0.91 (0.84–1.00)
Gender
Male	Referent
Female	1.12 (1.05–1.20)
Race/ethnicity
Non-Hispanic, White	Referent
Non-Hispanic, Black	1.13 (1.01–1.27)
Hispanic	1.09 (0.95–1.25)
Asian	1.40 (0.98–2.00)
Missing/unknown	0.82 (0.68–0.98)
Education
≤ High School	0.96 (0.89–1.04)
Some College	Referent
≥ Bachelor Degree	0.98 (0.88–1.10)
Missing/unknown	1.05 (0.85–1.30)
Household Income
<$40,000	0.98 (0.88–1.07)
$40,000 - <$75,000	Referent
$75,000-$124,999	1.04 (0.94–1.16)
$125,000-$199,000	0.99 (0.86–1.15)
$200,000+	1.25 (1.04–1.50)
Missing/unknown	0.99 (0.89–1.12)
Age
18–34 Years	1.20 (1.08–1.33)
35–49 Years	0.89 (0.82–0.98)
50–64 Years	Referent
65–74 Years	1.13 (1.02–1.26)
≥ 75 Years	1.30 (1.10–1.53)
Elixhauser Comorbidity Index
Low (0–4)	Referent
Medium (5–8)	1.25 (1.16–1.36)
High (≥9)	1.47 (1.29–1.68)
One or More Mental Health ConditionsA
No	Referent
Yes	1.10 (1.02–1.18)
Alcohol Use Disorder
No	Referent
Yes	1.31 (1.21–1.43)

^AOne or more diagnoses within the categories of depression, bipolar disorder, and anxiety

Association Between Demographic and Clinical Characteristics and Treatment Quality

Table 4 depicts the relationship between sociodemographic and clinical characteristics and the likelihood that individuals met each quality measure relevant to buprenorphine receipt (Measures #1–4). Insurance type, gender, race/ethnicity, age, Elixhauser comorbidity score, education, household income, and presence of a mental health condition or alcohol use disorder had a statistically significant relationship with quality of buprenorphine care. For example, compared to men, women had a 18.3% lower relative risk of initiating buprenorphine within 365 days of index diagnosis (risk ratio [RR] 0.82, CI: 0.77–0.86) (Measure #1) and had similar disparities for the subsequent measures. Compared to white individuals, Black individuals had a 22.3% lower relative risk of initiating buprenorphine within 365 days of index diagnosis (RR: 0.78, CI: 0.70–0.86) (Measure #1) and 26.2% lower relative risk of initiating buprenorphine within 14 days of diagnosis (RR 0.74, CI: 0.65–0.84) (Measure #2). Similarly, Hispanic individuals were less likely to receive high quality buprenorphine care compared to white individuals. Compared to individuals without mental health conditions or alcohol use disorder, individuals with mental health conditions or alcohol use disorder were more likely to receive buprenorphine within 365 days of diagnosis (Measure #1) and initiate buprenorphine within 14 days of index diagnosis (Measure #2), but less likely to receive 180 days of buprenorphine (Measure #4). Compared to individuals with commercial coverage, individuals with Medicare Advantage coverage had lower quality of buprenorphine care, including a 11.6% lower relative risk of receiving buprenorphine within 365 days of index diagnosis (RR: 0.88, CI: 0.82–0.95) (Measure #1) and a 11.0% lower relative risk of receiving buprenorphine within 14 days of index diagnosis (RR: 0.89, CI: 0.81–0.98) (Measure #2).

Table 4.

Adjusted Association Between Sociodemographic and Clinical Characteristics and Buprenorphine Quality of Care for Commercial and Medicare Advantage (MA) Enrollees (n=45,210)

	Measure #1 (365-Day Initiation)	Measure #2 (14-Day Initiation)	Measure #3 (Engagement)A	Measure #4 (Maintenance)B
	Adjusted Risk Ratio	Adjusted Risk Ratio	Adjusted Risk Ratio	Adjusted Risk Ratio
Insurance
Commercial	Referent	Referent	Referent	Referent
Medicare Advantage	0.88 (0.82–0.95)	0.89 (0.81–0.98)	0.96 (0.89–1.04)	0.97 (0.90–1.05)
Gender
Male	Referent	Referent	Referent	Referent
Female	0.82 (0.77–0.86)	0.76 (0.70–0.81)	0.93 (0.88–0.99)	0.90 (0.85–0.95)
Race/ethnicity
Non-Hispanic, White	Referent	Referent	Referent	Referent
Non-Hispanic, Black	0.78 (0.70–0.86)	0.74 (0.65–0.84)	0.99 (0.89–1.11)	0.92 (0.82–1.03)
Hispanic	0.70 (0.62–0.79)	0.67 (0.57–0.78)	0.99 (0.89–1.13)	1.00 (0.89–1.14)
Asian	0.77 (0.56–1.05)	0.72 (0.47–1.09)	0.85 (0.57–1.26)	0.62 (0.37–1.03)
Missing/unknown	1.04 (0.90–1.21)	1.12 (0.93–1.35)	1.06 (0.92–1.22)	1.09 (0.95–1.25)
Education
≤ High School	0.91 (0.85–0.97)	0.94 (0.87–1.03)	1.04 (0.97–1.11)	1.01 (0.94–1.08)
Some College	Referent	Referent	Referent	Referent
≥ Bachelor Degree	1.04 (0.95–1.15)	1.12 (0.97–1.27)	0.94 (0.84–1.05)	1.07 (0.97–1.18)
Missing/unknown	0.69 (0.58–0.82)	0.65 (0.52–0.81)	0.99 (0.83–1.17)	1.05 (0.90–1.23)
Household Income
<$40,000	1.07 (0.99–1.16)	1.16 (1.04–1.29)	0.95 (0.87–1.03)	0.98 (0.90–1.07)
$40,000 - <$75,000	Referent	Referent	Referent	Referent
$75,000-$124,999	0.93 (0.85–1.02)	0.92 (0.82–1.04)	0.95 (0.86–1.04)	1.00 (0.92–1.10)
$125,000-$199,000	0.88 (0.78–0.99)	0.96 (0.83–1.12)	0.94 (0.83–1.06)	0.94 (0.82–1.06)
$200,000+	0.78 (0.67–0.91)	0.67 (0.54–0.84)	0.86 (0.71–1.04)	0.94 (0.80–1.11)
Missing/unknown	1.22 (1.12–1.34)	1.23 (1.08–1.38)	0.96 (0.88–1.06)	0.96 (0.87–1.06)
Age
18–34 Years	1.86 (1.70–2.02)	1.79 (1.60–2.01)	1.03 (0.94–1.13)	0.89 (0.81–0.98)
35–49 Years	1.68 (1.56–1.80)	1.83 (1.67–2.00)	1.08 (1.00–1.16)	1.07 (0.99–1.15)
50–64 Years	Referent	Referent	Referent	Referent
65–74 Years	0.58 (0.53–0.64)	0.55 (0.48–0.62)	0.91 (0.80–1.02)	0.96 (0.86–1.07)
≥ 75 Years	0.34 (0.29–0.40)	0.33 (0.27–0.41)	0.56 (0.42–0.75)	0.83 (0.68–1.00)
Elixhauser Comorbidity Index
Low (0–4)	Referent	Referent	Referent	Referent
Medium (5–8)	0.77 (0.72–0.82)	0.65 (0.59–0.71)	0.90 (0.84–1.04)	0.83 (0.76–0.89)
High (≥9)	0.60 (0.53–0.67)	0.43 (0.36–0.51)	0.85 (0.94–1.11)	0.79 (0.69–0.92)
One or More Mental Health ConditionsC
No	Referent	Referent	Referent	Referent
Yes	1.29 (1.22–1.37)	1.21 (1.12–1.31)	0.98 (0.92–1.04)	0.93 (0.88–0.99)
Alcohol Use Disorder
No	Referent	Referent	Referent	Referent
Yes	1.14 (1.06–1.22)	0.97 (0.88–1.08)	1.02 (0.94–1.11)	0.76 (0.69–0.83)

^ATo align with the measure denominator, the logistic regression for Measure #3 (Engagement) is specific to the 2,850 individuals who initiate buprenorphine within 14 days of index diagnosis.

^BTo align with the measure denominator, the logistic regression for Measure #4 (Maintenance) is specific to the 4,600 individuals who received buprenorphine within 365 days of index diagnosis.

^COne or more diagnoses within the categories of depression, bipolar disorder, and anxiety

Similarly, for individuals with MA coverage, race/ethnicity, age, Elixhauser comorbidity score, and presence of a mental health condition or alcohol use disorder had a statistically significant relationship with quality of buprenorphine care. With the exception of alcohol use disorder, the direction of effects was consistent for individuals with commercial coverage and MA coverage.

DISCUSSION

We characterized the quality of buprenorphine care for OUD among a large, diverse, real-world cohort of commercially insured and MA enrollees in the United States. While more than half of individuals who initiated treatment within 365 days of diagnosis were further engaged in care, overall rates of treatment initiation and maintenance in care were low. Of the 45,210 individuals with diagnosed opioid use disorder, 10.2% initiated buprenorphine within 365 days of index diagnosis and among these 4,600 individuals who received buprenorphine 50.0% received 180 days or more of buprenorphine treatment in the year following diagnosis. Among the 4,600 individuals who received buprenorphine only 55.3% did not fill a prescription for another opioid analgesic and 64.2% did not fill a prescription for a benzodiazepine within 180 days following buprenorphine treatment.

The median treatment duration among those initiating buprenorphine treatment was 71 days. The probability of treatment discontinuation, as well as other quality gaps, was greater among Black and Hispanic adults than non-Hispanic, white adults and among those with greater comorbid or psychiatric illness compared with their counterparts. Performance on the quality measures was also worse for individuals with MA compared to commercial coverage.

Our findings are important because of continued concerns about quality of care for individuals with OUD as well as uncertainty regarding how recent initiatives to reduce access barriers and to expand treatment infrastructure have impacted care delivery. For this study, we adapted measures that have been developed and validated for use as part of HEDIS and the OptumLabs Comprehensive Opioid Use Quality Measure Framework.²⁸ By assessing measures along the cascade of care, our approach provides payers and providers with data regarding where quality lapses may occur. These insights may inform quality improvement activities and iterative assessment of measures to understand the effectiveness of interventions over time.

If payers or providers are seeking to prioritize quality measures, given the very low rates on the buprenorphine initiation measures, these are extremely important to evaluate but likely not as useful as other measures. Clinical practice guidelines recommend against use of opioids or benzodiazepines concurrent to buprenorphine when possible, given significant risks.²⁷ However, they may be appropriate in uncommon situations, such as necessary opioid use for an acute painful condition, or continuing current benzodiazepines short-term or intermittent use (e.g., for panic attacks). Additionally, recent research has found that concomitant opioid and benzodiazepine prescribing may be more likely coming from other clinicians (not those prescribing buprenorphine),²⁹ so these may also potentially not be the most useful measures for quality specifically, but may be useful to evaluate coordination of care. Given that the goal is long-term buprenorphine use, and that performance on the 180-day measure (a standard measure of retention) was met for half of individuals and was not too much lower than receiving a 2nd prescription, this study supports the use of the 180-day measure as the quality measure most likely to be of clinical use in the insured population.

Our findings also have important implications for health care equity and disparities. Prior research, focused on individuals with Medicaid, has found that buprenorphine uptake and persistence is generally lower among racial/ethnic minorities compared to white individuals and women compared to men.^30–33 We extend this work by showing that these disparities also exist within MA and commercially insured populations, groups that are likely to have higher average socioeconomic status than Medicaid enrollees. We also find worse quality of buprenorphine care for individuals with MA coverage compared to commercial coverage, suggesting that there are age-based inequities in access to OUD treatment. Buprenorphine patients from minority groups report experiencing intersecting racial discrimination and stigma against addiction,³⁴ and these factors likely exist for commercially insured patients. Interestingly, we find that there were no significant differences by education after adjusting for other sociodemographic and clinical factors, which may suggest either that education is a relatively limited factor or that using area levels of education is too blunt a measure and that further research is needed to understand the role of education in quality of buprenorphine treatment.³⁵ Additionally, some caution is warranted since 12.5% of individuals had missing education data.

We also find that individuals with more diagnosed comorbidities, particularly other psychiatric conditions, have poorer outcomes including discontinuation of buprenorphine treatment compared to people without such diagnoses.³⁶ These disparities are of great concern because individuals in these groups are at higher risk of overdose and should be a target for enhanced care management to ensure that they are able to access and maintain buprenorphine treatment.³⁷

Payers and health insurance plans have an important role to play in improving quality of care. Cost-related barriers are a notable concern in commercial insurance and Medicare plans, as patients may skip refills because of copayments or deductibles.³⁸ Additionally, prior authorization may restrict or delay access to buprenorphine-naloxone.³⁹ These concerns should decrease, however, as generic formulations of buprenorphine-naloxone become increasingly available.⁴⁰ Further steps to increase access and quality could include creating centers of excellence, following the model pursued in several state Medicaid programs such as Rhode Island and Pennsylvania.⁴¹ Payers could also provide special incentives to providers to meet quality targets and ensure that providers who already dispense buprenorphine with high quality and oversight are included in plan networks.

Finally, federal policies play a role in access to buprenorphine for individuals with OUD. For example, the 21^st Century Cures Act provided states with $1 billion to invest in expanding access to OUD treatment.⁴² However, access remains constrained by the current regulations, including the requirement that providers receive a waiver in order to prescribe buprenorphine. There are many treatment deserts where individuals with OUD have little to no access to buprenorphine waivered providers.⁴³ Changing the waiver requirements or incentivizing more physicians to complete the waiver process may also improve buprenorphine treatment quality.^44–46

Our study has limitations. First, we limited our analysis to buprenorphine rather than examining other FDA-approved treatments for OUD such as naltrexone or methadone. Nonetheless, organizations that are undertaking initiatives to improve OUD quality of care should also consider these additional pharmacologic treatment options. Particularly for individuals with alcohol use disorder, providers may instead prescribe naltrexone as it can treat both alcohol use disorder and OUD.⁴⁷ Second, while our cohort reflected a large, diverse, real-world sample, it was drawn from a single database and it is unclear if it is fully generalizable to other commercial or MA plans. Third, because we use administrative claims data, we do not know if individuals are receiving buprenorphine treatment that is not reimbursed through their insurance or if discontinuation of buprenorphine treatment is due to a planned stop, for example due to reimbursement limitations. Fourth, by requiring 12 months of observation following diagnosis we increased our follow-up at the expense of decreasing our sample size and generalizability given the relatively high churn rates of individuals with OUD. Fifth, we excluded individuals with a single outpatient diagnosis of OUD, increasing the specificity of our diagnostic criteria at the cost of decreasing its sensitivity.

CONCLUSIONS

The large number of individuals in the United States with OUD combined with continuing investments to reduce addiction and overdose create newfound urgency and opportunity to assess and continuously monitor OUD quality of care. In this analysis of six measures spread along the care cascade, we found widespread gaps in quality of buprenorphine treatment initiation, engagement, and maintenance, highlighting opportunities for payers, providers, and health systems alike. Along the care cascade, buprenorphine uptake and persistence were generally lower among racial/ethnic minorities compared to white individuals and among older adults with Medicare Advantage coverage compared to working age adults with commercial coverage. Addressing health inequities and overall gaps in buprenorphine treatment quality are especially important given the clear scientific consensus that buprenorphine substantially reduces mortality and improves social and health outcomes among individuals with OUD.