Early Discontinuation of Buprenorphine Therapy for Opioid Use Disorder Among Privately Insured Adults
Abstract
Objective:
This study examined the temporal relationship between early discontinuation of buprenorphine treatment and health care expenditures before and after treatment initiation.
Methods:
MarketScan commercial claims for patients who initiated buprenorphine for opioid use disorder in 2013 and had continuous insurance for the subsequent 12 months (N=6,444) were used to examine the relationship between treatment retention and health care expenditures before and after buprenorphine initiation. Analysis of covariance and generalized linear models (with gamma distribution/log link) were used to compare expenditures across four buprenorphine-retention groups (0–3, 3–6, 6–12, and 12 or more months).
Results:
Average total health care expenditures in the 3 months prior to buprenorphine initiation ranged from a high of $7,588 among those with the shortest retention to $4,929 among those with the longest retention (p<0.001). In the 12 months after buprenorphine initiation, total health care expenditures averaged $26,332 per year, with $2,916 (11.1%) in out-of-pocket expenditures. Average annual expenditures for medication were highest among patients with the longest buprenorphine retention, and total health care expenditures were highest among those with the shortest retention. Expenditures for health care services other than medication were highest among those with early discontinuation both before the initiation of buprenorphine and during the initial period after initiation but not in subsequent quarters.
Conclusions:
Poorer treatment retention among privately insured adults was associated with greater clinical and financial burdens that preceded and continued during the period shortly following treatment initiation, suggesting that cost burdens may contribute to poor retention among privately insured adults.
HIGHLIGHTS
In a sample of privately insured adults with opioid use disorder, 6,444 initiated buprenorphine treatment.
Of these, 28% discontinued treatment within 3 months, 10% within 6 months, 16% within 9 months, and 46% within 12 months.
Annual health expenditures were highest for those with the poorest buprenorphine retention.
However, this group had high expenditures for health care before initiation of buprenorphine treatment.
Opioid use disorder represents a major public health crisis, affecting more than 2 million individuals in the United States (1). It has contributed to overdose mortality, spread of infectious diseases such as hepatitis C and HIV (2–5), and increased hospitalization and emergency department expenditures (6). With passage of the Drug Addiction Treatment Act of 2000, medication-assisted treatment (MAT) with buprenorphine became available in office practices funded through private health insurance (7).
Buprenorphine has proven effective in relieving withdrawal symptoms, improving remission rates, minimizing opioid abuse and diversion, decreasing infectious disease prevalence, and improving survival (8–10). However, it is associated with retail costs of $100 to $200 per month, potentially impeding its use among privately insured adults (11), a barrier highlighted recently in a National Academies report (12). Higher mortality and relapse rates are among the consequences of buprenorphine discontinuation (13). Studies have shown poorer retention in buprenorphine treatment among privately insured adults with copay obligations than among Veterans Health Administration patients, who pay little—if anything—for their medications (14, 15). Whether being burdened with greater expenditures at the time of treatment initiation is associated with poor retention among privately insured adults has not been examined.
Previous studies of health plan spending and consumer expenditures for buprenorphine have been based on small samples, focused narrowly on total buprenorphine expenditures alone, or compared expenditures among patients prescribed buprenorphine or methadone with those of patients who were not prescribed medication at all (16–20). Mohlman et al. (16) used Vermont Medicaid data to compare health care expenditures between those receiving MAT, primarily buprenorphine, and those not receiving MAT. They found that those in the MAT group had lower health care expenditures, but Medicaid requires no out-of-pocket cost sharing, and expenditure time trends were not reported.
Tkacz et al. (18) examined claims data from a small sample of privately insured adults (N=455) and found that the patients with greater buprenorphine treatment adherence during the 12 months after initiation had greater pharmacy expenditures than patients with less adherence ($6,156 versus $3,581). They also found lower expenditures for health care other than drugs among the more adherent patients ($22,409 versus $45,381), driven by the group’s reduced hospital admissions. However, the study did not examine the relationship between preinitiation costs, temporal expenditure trends, or out-of-pocket expenditures and treatment discontinuation. To our knowledge, no studies have examined the association of buprenorphine retention with preinitiation expenditures.
In this study, we used a database of private insurance claims to identify a cohort of patients newly started on buprenorphine for opioid use disorder in 2013. We examined data from 15 months of continuous insurance coverage, including 3 months before and 12 months after buprenorphine initiation, and compared expenditures for medications and all other health services among patients retained in treatment for varying periods of time. Like Tkacz et al. (18), we hypothesized that, other things being equal, out-of-pocket expenditures for drugs would be modest and that more prolonged retention in buprenorphine treatment would be associated with greater buprenorphine-related expenditures but also with a more rapid decline over time in expenditures for health care other than drugs. We also sought to explore the magnitude and timing of out-of-pocket expenditures.
Methods
Data Source and Sample
Using data from the IBM MarketScan Commercial Claims and Encounters Database, we included patients ages 19 years or older if they were diagnosed as having opioid use disorder (ICD-9 codes 304.0x, 305.5x, and 304.7x) from January 1 to December 31, 2013, and had a prescription drug claim for buprenorphine or buprenorphine-naloxone tablets. Claims for a buprenorphine transdermal patch, which is indicated for pain management, were excluded. Furthermore, patients had to be continuously enrolled for at least 3 months before and for 1 full year after initiating buprenorphine treatment (i.e., their index buprenorphine prescription date). To define new starts, we included only patients who were covered by insurance and had not filled any buprenorphine prescription for 3 months prior to their index buprenorphine prescription date. (A consort diagram of sample selection is available in the online supplement.)
Measures
Retention in buprenorphine treatment.
We calculated the total duration of buprenorphine treatment as the number of days between the date of the first buprenorphine prescription in 2013 (the index date) and the date of the last buprenorphine prescription documented in the MarketScan files for prescription drug claims. Although medication received through the last prescription may offer drug coverage for an additional 30 days, we do not know whether treatment adherence continued during that time and took the conservative approach of not including that time in the retention calculation. Because opioid use disorder is a chronic, relapsing disease requiring long-term treatment, we regarded both uninterrupted and interrupted participation in treatment as part of a continuous episode of care. We grouped the resultant sample into four mutually exclusive groups on the basis of duration of retention in buprenorphine treatment: 0–3 months (group 1), 3–6 months (group 2), 6–12 months (group 3), and ≥12 months (group 4).
Patient characteristics.
Demographic characteristics included age, sex, urban versus rural residence as based on metropolitan statistical area, buprenorphine dose, and time (in days) to index buprenorphine prescription. Information about psychiatric and substance use disorder diagnoses for each patient was based on ICD-9 codes. The burden of medical comorbidity was represented by the Charlson index (21), an aggregate measure calculated on the basis of ICD-9 medical codes and shown to predict 1-year mortality.
Health service utilization, medication prescription fills, and expenditures.
Service utilization codes from insurance claims were used to identify health service use, prescription drug fills, and related expenditures. Total health care expenditures were grouped into subsets for pharmacy services, all health services, mental health services, and nonmental health services expenditures. Mental health services expenditures were further grouped into subsets of expenditures linked to opioid use disorder and other mental disorders on the basis of ICD-9 codes associated with each claim, as well as a subset of expenditures for mental health services provided in an inpatient setting. Buprenorphine expenditures were separated out from total pharmacy expenditures. Claims data also allowed determination of the out-of-pocket expenditures charged to patients for each type of service. Expenditure data were summarized by 3-month periods corresponding to the period of retention.
Analysis
We used chi-square tests and analysis of variance to compare the four retention groups on sociodemographic characteristics, diagnoses, and health care expenditures in the 3 months before buprenorphine initiation. Claims following buprenorphine initiation were aggregated to determine the total annual health care and medication expenditures in each category, along with total out-of-pocket expenditures. Next, we compared annual expenditures between the four retention groups on the aforementioned cost categories, using analysis of covariance to test for the significance of differences between groups while controlling for expenditures incurred prior to buprenorphine initiation and for sociodemographic and diagnostic variables that were significantly different between the retention groups. Adjusted costs were calculated by using the sample with complete dose information.
To better understand the basis for differences in expenditures, we further examined differences between retention groups in health service expenditures over time. The same 3-month periodization used to characterize buprenorphine retention was applied to the longitudinal evaluation of health care expenditures. Because of the skewed nature of expenditure data, we used generalized linear models (with gamma distribution/log-link function) to estimate the differences among groups in expenditures at the various time points (22). These analyses examined the significance of differences in expenditures over time, of differences between retention groups, and of the interaction of time × retention group membership (i.e., differences in longitudinal trends), controlling for characteristics that were significantly different between retention groups. Statistical analyses were conducted by using SAS, version 9.4. The Penn State College of Medicine institutional review board approved this study.
Results
Sample Selection and Correlates of Retention
Altogether, 32,305 people ages 19 years or older in the MarketScan database were diagnosed as having opioid use disorder and received a prescription for buprenorphine in 2013. Of these, 12,988 (40.2%) were considered as having newly started buprenorphine, and 6,444 (49.6%) of the new starts were continuously enrolled in the insurance claims database for at least 3 months before starting buprenorphine and for at least 12 months after. Overall, 1,809 (28.1%) stayed on buprenorphine for 0 to 3 months (group 1); 617 (9.6%) for 3 to 6 months (group 2); 1,060 (16.5%) for 6 to 12 months (group 3); and 2,958 (45.9%) for 12 or more months (group 4).
Patients who continued buprenorphine for at least 12 months were older than patients in the other groups and less likely to reside in urban areas, with no significant gender differences between groups (Table 1). Patients in group 4 also had significantly fewer concurrent psychiatric and substance use disorder diagnoses (both p<0.001). There were similar differences on the Charlson index, which was lowest in group 4. Buprenorphine dose comparisons showed significant differences between groups, with a monotonic trend toward higher doses with longer buprenorphine retention (range 11.80 mg/day [group 1] to 14.05 mg/day [group 4]). In addition, group 4 started buprenorphine significantly sooner than the other groups (24.4 versus 35.0–43.8 days).
Buprenorphine retention group | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Group 1: 0–3 months | Group 2: 3–6 months | Group 3: 6–12 months | Group 4: ≥12 months | |||||||||
Characteristic | N | % | N | % | N | % | N | % | Test statistic | df | p | Pairwise comparisons |
Total | 1,809 | 28.1 | 617 | 9.6 | 1,060 | 16.5 | 2,958 | 46.0 | ||||
Sociodemographic | ||||||||||||
Age at diagnosis | 30.75±12.19 | 30.62±11.55 | 31.52±11.78 | 34.58±11.83 | F=50.41 | 3, 6,440 | <.001 | 4>1,2,3 | ||||
Sex | t=6.82 | 3 | .078 | |||||||||
Female | 665 | 36.8 | 202 | 32.7 | 373 | 35.2 | 1,119 | 37.8 | ||||
Male | 1,144 | 63.2 | 415 | 67.3 | 687 | 64.8 | 1,839 | 62.2 | ||||
Urban area | 1,602 | 88.6 | 532 | 86.2 | 912 | 86.0 | 2,493 | 84.0 | t=17.07 | 3 | .001 | |
Time to buprenorphine initiation (days) | 43.78±75.63 | 34.99±82.28 | 37.88±86.47 | 24.43±80.96 | F=23.38 | 3, 6,440 | <.001 | 1,2,3>4 | ||||
Buprenorphine dose (mg/day)b | 11.80±6.21 | 12.64±5.59 | 13.05±5.57 | 14.05±5.98 | F=51.59 | 3, 6,060 | <.001 | 4>1,2,3; 3,2>1 | ||||
Psychiatric diagnosis | ||||||||||||
Total N | 1.08±1.23 | 1.14±1.18 | 1.10±1.19 | .94±1.11 | F=10.32 | 3, 6,440 | <.001 | 1,2,3>4 | ||||
Any psychiatric diagnosis | 1,460 | 80.7 | 511 | 82.8 | 879 | 82.9 | 2,238 | 75.7 | t=38.28 | 3 | <.001 | |
Bipolar disorder | 202 | 11.2 | 63 | 10.2 | 106 | 10.0 | 234 | 7.9 | t=15.27 | 3 | .002 | |
Major depression | 343 | 19.0 | 122 | 19.8 | 213 | 20.1 | 501 | 16.9 | t=7.38 | 3 | .061 | |
Other depression | 451 | 25.0 | 184 | 29.8 | 302 | 28.5 | 680 | 23.0 | t=20.85 | 3 | <.001 | |
PTSD | 59 | 3.3 | 21 | 3.4 | 33 | 3.1 | 74 | 2.5 | t=3.23 | 3 | .356 | |
Anxiety disorder | 677 | 37.4 | 245 | 39.7 | 383 | 36.1 | 996 | 33.7 | t=12.05 | 3 | .007 | |
Adjustment disorder | 116 | 6.4 | 42 | 6.8 | 86 | 8.1 | 209 | 7.1 | t=3.01 | 3 | .390 | |
Personality disorder | 24 | 1.3 | 9 | 1.5 | 12 | 1.1 | 27 | .9 | t=2.49 | 3 | .477 | |
Schizophrenia | 15 | .8 | 4 | .7 | 7 | .7 | 12 | .4 | t=3.61 | 3 | .307 | |
Other | 69 | 3.8 | 16 | 2.6 | 24 | 2.3 | 46 | 1.6 | t=24.59 | 3 | <.001 | |
Substance use disorder | ||||||||||||
Total N | .37 ±.68 | .39 ±.69 | .36 ±.66 | .21 ±.51 | F=37.38 | 3, 6,440 | <.001 | 1,2,3>4 | ||||
Any substance use disorder | 894 | 49.4 | 321 | 52.0 | 535 | 50.5 | 1,035 | 35.0 | t=152.19 | 3 | <.001 | |
Alcohol use disorder | 292 | 16.1 | 102 | 16.5 | 168 | 15.9 | 295 | 10.0 | t=52.63 | 3 | <.001 | |
Cocaine use disorder | 78 | 4.3 | 20 | 3.2 | 34 | 3.2 | 67 | 2.3 | t=15.85 | 3 | .001 | |
Cannabis use disorder | 99 | 5.5 | 39 | 6.3 | 69 | 6.5 | 97 | 3.3 | t=27.01 | 3 | <.001 | |
Sedative use disorder | 146 | 8.1 | 59 | 9.6 | 84 | 7.9 | 127 | 4.3 | t=44.36 | 3 | <.001 | |
Amphetamine use disorder | 48 | 2.7 | 17 | 2.8 | 25 | 2.4 | 31 | 1.1 | t=20.87 | 3 | .001 | |
Hallucinogen use disorder | 3 | .2 | 1 | .2 | 4 | .4 | 3 | .1 | —c | .261 | ||
Other | 813 | 45.0 | 294 | 47.7 | 487 | 45.9 | 911 | 30.8 | t=151.52 | 3 | <.001 | |
Charlson indexd | .40 ±.94 | .39±1.12 | .35 ±.86 | .31 ±.80 | F=4.26 | 3, 6,440 | .005 | 1>4 | ||||
Total health care expenditures during the 3 months before initiating buprenorphine (M±SD US $) | 7,588±15,986 | 5,405±9,818 | 5,468±10,574 | 4,929±11,545 | F=17.11 | 3, 6,440 | <.001 | 1>2,3,4 | ||||
Pharmacy | 558±1,554 | 530±2,933 | 444±1,093 | 504±1,336 | F=1.21 | 3, 6,440 | .306 | NS | ||||
All health services other than pharmacy | 7,030±15,509 | 4,875±9,028 | 5,024±10,285 | 4,425±11,223 | F=17.43 | 3, 6,440 | <.001 | 1>2,3,4 | ||||
Mental health services | 4,275±10,568 | 3,200±7,219 | 2,847±6,642 | 2,425±6,999 | F=19.76 | 3, 6,440 | <.001 | 1>2,3,4; 2>4 | ||||
Opioid use disorder services | 2,242±6,383 | 1,742±4,358 | 1,412±3,768 | 1,347±4,402 | F=13.22 | 3, 6,440 | <.001 | 1>2,3,4 | ||||
Services for other mental disorders | 2,032±6,747 | 1,458±5,374 | 1,435±4,931 | 1,078±4,387 | F=12.03 | 3,6440 | <.001 | 1>2,3,4 | ||||
Inpatient mental health services | 2,284±6,504 | 1,806±4,802 | 1,539±4,232 | 1,299±4,481 | F=14.20 | 3, 6,440 | <.001 | 1>2,3,4; 2>4 | ||||
Nonmental health services | 2,755±10,498 | 1,675±4,349 | 2,177±7,053 | 2,000±8,298 | F=3.89 | 3, 6,440 | .009 | 1>2,4 | ||||
Total out-of-pocket expenditures | 761±1,918 | 621±902 | 641±1,049 | 607±1,223 | F=4.70 | 3, 6,440 | .003 | 1>2,3,4 |
Health Care and Medication Expenditures Before Buprenorphine Initiation
The average total health care expenditures prior to buprenorphine initiation ranged from a high of $7,588 in group 1 to $4,929 in group 4 (p<0.001) (Table 1). Similarly, expenditures for mental health services ranged, on average, from $4,275 (group 1) to $2,425 (group 4) (p<0.001) and accounted for the majority of health care expenditures other than medication in the 3 months prior to the start of treatment with buprenorphine. Out-of-pocket expenditures before buprenorphine initiation were significantly higher in group 1 than among all three groups with greater retention. There was no statistical difference between the groups for drug expenditures before treatment initiation (Table 1).
Annual Health Care and Medication Expenditures
After patients started buprenorphine, their total health care expenditures averaged $26,332 per year, with out-of-pocket expenditures amounting to 11.1% of the total or $2,916 per patient per year (Table 2). Total medication expenditures accounted for 19.0% ($5,007) of total average health care expenditures and 23.0% ($671) of all out-of-pocket expenditures. Annual buprenorphine expenditures averaged $2,618, with out-of-pocket expenditures averaging $387 per year ($32 per month) for the entire sample and $576 per year ($48 per month) for those who stayed on the medication for all 12 months (data not shown).
Total expenditures | Out-of-pocket expenditures | ||||||
---|---|---|---|---|---|---|---|
Expenditure category | M | SD | % of total expenditures | % of expenditures for pharmacy or all other health services | M | SD | % of expenditure category |
Total health care expenditures (pharmacy and health services) | 26,332 | 45,418 | — | — | 2,916 | 4,041 | 11.1 |
Pharmacy | 5,007 | 8,343 | 19.0 | — | 671 | 705 | 13.4 |
Buprenorphine | 2,618 | 2,369 | 9.9 | 52.3a | 387 | 472 | 14.8 |
All health services other than pharmacy | 21,326 | 44,269 | 81.0 | — | 2,245 | 3,939 | 10.5 |
Mental health servicesb | 14,228 | 37,186 | 54.0 | 66.7c | 1,594 | 3,705 | 11.2 |
Opioid use disorder services | 7,647 | 18,073 | 29.0 | 35.9c | 1,028 | 2,324 | 13.4 |
Services for other mental disorders | 6,581 | 26,643 | 25.0 | 30.9c | 567 | 2,594 | 8.6 |
Nonmental health services | 7,098 | 21,225 | 27.0 | 33.3c | 650 | 1,342 | 9.2 |
Unadjusted annual expenditures for pharmacy and other health care services among privately insured adults with opioid use disorder, in US $
Expenditures for health services other than drugs averaged $21,326 per patient per year and accounted for 81.0% of total health care expenditures. Mean total and out-of-pocket expenditures were highest in the category of mental health services ($14,228 and $1,594, respectively) (Table 2).
Health Care and Medication Expenditures by Retention Group
Overall, adjusted average total health care expenditures in the year after buprenorphine initiation were higher among those with the briefest buprenorphine retention (group 1) than for each of the three groups with greater retention (Table 3). Total medication expenditures, in contrast, increased across retention groups, as expected, with the average ranging from $3,119 in group 1 (shortest retention) to $6,495 in group 4 (longest retention). This was especially true for buprenorphine expenditures, which increased monotonically from lower to higher retention groups (Table 3).
Buprenorphine retention group ($) | |||||||||
---|---|---|---|---|---|---|---|---|---|
Group 1: 0–3 months(N=1,707; 28.2%) | Group 2: 3–6 months(N=582; 9.6%) | Group 3: 6–12 months(N=1,001; 16.5%) | Group 4: ≥12 months(N=2,774; 45.8%) | ||||||
Expenditure category | M | SE | M | SE | M | SE | M | SE | Pairwise comparisonsb |
Total health care expenditures (pharmacy and medical services) | 29,521 | 965 | 25,854 | 1,632 | 25,601 | 1,243 | 25,621 | 758 | 1>3,4 |
Pharmacy | 3,119 | 174 | 3,446 | 295 | 4,837 | 225 | 6,495 | 137 | 4>1,2,3; 3>1,2 |
Buprenorphine | 691 | 35 | 1,548 | 60 | 2,686 | 46 | 3,966 | 28 | 4>1,2,3; 3,2>1;3>2 |
All health services other than pharmacy | 26,397 | 943 | 22,355 | 1,596 | 20,772 | 1,215 | 19,138 | 741 | 1>2,3,4 |
Mental health services | 17,742 | 801 | 14,801 | 1,356 | 14,570 | 1,033 | 12,519 | 630 | 1>3,4 |
Opioid use disorder services | 8,853 | 409 | 8,446 | 693 | 7,884 | 528 | 6,970 | 322 | 1>4 |
Services for other mental disorders | 8,991 | 598 | 6,307 | 1,013 | 6,500 | 771 | 5,563 | 471 | 1>2,3,4 |
Inpatient mental health services | 4,470 | 240 | 3,480 | 406 | 4,055 | 309 | 2,924 | 189 | 1>2,4; 3>4 |
Nonmental health services | 8,645 | 463 | 7,554 | 784 | 6,236 | 597 | 6,612 | 364 | 1>3,4 |
Total out-of-pocket expenditures | 2,690 | 93 | 3,088 | 158 | 2,936 | 120 | 3,055 | 73 | 2,4>1 |
Analysis of postinitiation expenditures (adjusted for preinitiation expenditures and sociodemographic and diagnostic variables identified previously) showed that the group with the shortest retention (group 1) had significantly higher annual expenditures than groups with more prolonged retention for all health service categories other than medications and for total health care (p<0.05)(Table 3).
Differences in Temporal Patterns of Expenditures
To further evaluate 12-month temporal expenditure patterns, we evaluated the significance of the interaction between retention group and time, again controlling for variables in which there were significant differences between the retention groups (Table 4 and Figure 1). In almost all expenditure categories, there were significant main effects for time, in which longer retention in buprenorphine treatment was associated with declining expenditures and significant differences by retention group in every cost category, except total health care expenditures (Table 4). Significant group × time interactions indicate that all expenditures except those for nonmental health services declined more steeply among those who stayed on buprenorphine for only 0 to 3 months. Analysis of total expenditures, for example, showed that patients with the shortest retention (group 1) had the highest total expenditures during the first 3 months of treatment (Table 4), while they were (however briefly) receiving buprenorphine, just as they had before treatment was initiated. These early expenditure levels stand in stark contrast to those of patients retained for at least 12 months (group 4), who had significantly lower expenditures than group 1 for total health care and mental health services both before buprenorphine initiation (Table 1) and during the first quarter of buprenorphine treatment (Table 4).
Average expenditure ($) | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
0–3 months | 3–6 months | 6–9 months | 9–12 months | Paired comparisons | ||||||||
Expenditure category and group | M | SE | M | SE | M | SE | M | SE | 0–3 months | 3–6 months | 6–9 months | 9–12 months |
Total health careb,c | 1>2,3,4; 3>2 | 1,2>3 | ns | 3,4>1; 3>2 | ||||||||
Group 1: 0–3 months | 9,113 | 1.05 | 4,773 | 1.06 | 4,450 | 1.07 | 4,089 | 1.07 | ||||
Group 2: 3–6 months | 5,550 | 1.05 | 5,151 | 1.07 | 5,269 | 1.11 | 4,223 | 1.11 | ||||
Group 3: 6–12 months | 6,489 | 1.06 | 4,062 | 1.05 | 4,920 | 1.06 | 5,426 | 1.07 | ||||
Group 4: ≥12 months | 6,149 | 1.03 | 4,498 | 1.03 | 4,528 | 1.03 | 4,995 | 1.04 | ||||
Total pharmacyb,c,d | 2,3,4>1; 4>3 | 2,3,4>1; 3,4>2; 4>3 | 1,3,4>2; 3,4>1; 4>3 | 3,4>1,2; 4>3 | ||||||||
Group 1: 0–3 months | 869 | 1.03 | 355 | 1.08 | 426 | 1.12 | 474 | 1.14 | ||||
Group 2: 3–6 months | 1,430 | 1.05 | 866 | 1.04 | 307 | 1.10 | 399 | 1.10 | ||||
Group 3: 6–12 months | 1,335 | 1.02 | 1,118 | 1.06 | 1,054 | 1.07 | 759 | 1.07 | ||||
Group 4: ≥12 months | 1,568 | 1.02 | 1,387 | 1.04 | 1,414 | 1.04 | 1,540 | 1.04 | ||||
Mental health servicesb,c,d | 1>2,3,4; 3>2 | 2>3,4; 1>3 | 1,2,3>4 | 3>1,2,4 | ||||||||
Group 1: 0–3 months | 5,285 | 1.06 | 2,034 | 1.09 | 1,831 | 1.10 | 1,645 | 1.11 | ||||
Group 2: 3–6 months | 2,482 | 1.08 | 2,290 | 1.10 | 2,507 | 1.16 | 1,822 | 1.14 | ||||
Group 3: 6–12 months | 3,469 | 1.11 | 1,600 | 1.08 | 2,199 | 1.09 | 2,616 | 1.11 | ||||
Group 4: ≥12 months | 2,868 | 1.05 | 1,722 | 1.06 | 1,449 | 1.06 | 1,593 | 1.06 | ||||
Inpatient mental health servicesc,d | 1>2,3,4 | 1,2>3,4 | 1,2,3>4 | 3>1,2,4 | ||||||||
Group 1: 0–3 months | 1,324 | 1.12 | 557 | 1.18 | 436 | 1.15 | 425 | 1.24 | ||||
Group 2: 3–6 months | 321 | 1.23 | 645 | 1.20 | 513 | 1.23 | 372 | 1.28 | ||||
Group 3: 6–12 months | 465 | 1.16 | 314 | 1.22 | 591 | 1.17 | 789 | 1.17 | ||||
Group 4: ≥12 months | 314 | 1.18 | 284 | 1.24 | 230 | 1.31 | 292 | 1.27 | ||||
Nonmental health servicese | na | na | na | na | ||||||||
Group 1: 0–3 months | 1,518 | 1.12 | 1,363 | 1.15 | 1,160 | 1.08 | 1,295 | 1.12 | ||||
Group 2: 3–6 months | 1,059 | 1.13 | 1,174 | 1.13 | 1,459 | 1.19 | 1,417 | 1.25 | ||||
Group 3: 6–12 months | 992 | 1.11 | 894 | 1.12 | 1,037 | 1.12 | 1,264 | 1.15 | ||||
Group 4: ≥12 months | 1,132 | 1.07 | 895 | 1.06 | 1,092 | 1.10 | 1,220 | 1.09 | ||||
Out-of-pocket expendituresb,c,d | 1>2,3,4 | 2>1,3,4; 4>1 | 3,4>1 | 2,3,4>1 | ||||||||
Group 1: 0–3 months | 1,114 | 1.05 | 499 | 1.07 | 475 | 1.06 | 405 | 1.07 | ||||
Group 2: 3–6 months | 869 | 1.07 | 819 | 1.07 | 568 | 1.10 | 549 | 1.12 | ||||
Group 3: 6–12 months | 925 | 1.05 | 570 | 1.04 | 668 | 1.06 | 636 | 1.06 | ||||
Group 4: ≥12 months | 912 | 1.03 | 622 | 1.03 | 627 | 1.03 | 662 | 1.03 |
There were no significant differences among groups in total expenditures in the third quarter. By the fourth quarter, the groups with the poorest retention had the lowest total health care expenditures (Table 4 and Figure 1). Patients in group 4 had significantly higher total pharmacy expenditures than those retained for shorter periods at all time points, reflecting the additional cost of buprenorphine (Table 4 and Figure 1).
Adjusted total expenditures for mental health services also remained significantly lower for group 4 compared with some or all of the other retention groups at each time point (Table 4 and Figure 1). That was also true for inpatient mental health expenditures (Table 4). There were no significant main effects of time or significant time × group interactions for expenditures for nonmental health services. Expenditures in the year following buprenorphine initiation thus continued the pattern observed among groups before buprenorphine initiation, with the highest expenditures among those with the poorest subsequent retention and persistently lower expenditures among those with the longest retention.
Out-of-pocket expenditures for group 1 were the highest of all the groups before initiation of buprenorphine (Table 1) and continued to be highest during the first quarter (Table 4 and Figure 1), while they were taking buprenorphine. In subsequent quarters, out-of-pocket expenditures for group 1 were lower than for other groups.
Discussion
This study of a large, privately insured cohort of patients who were diagnosed as having opioid use disorder and who had newly started buprenorphine found that direct buprenorphine expenditures represented a modest proportion of all expenditures, averaging $48 per month among those who stayed on treatment for a full 12 months and accounting for 14.8% of all annual out-of-pocket expenditures. However, total out-of-pocket health care expenditures averaged $2,916 per year or $243 per month, a substantial sum, and were higher for patients in the poorest retention group in the first quarter.
As reported in other studies (18, 23), annual expenditures for medication were highest among those retained in buprenorphine therapy the longest, whereas annual expenditures for all other health services were lower in groups that were retained longer. At first glance, these data might seem to support the hypothesis that longer retention lowered total health care expenditures because of the benefits of buprenorphine therapy. However, longitudinal analysis of this sample showed that although patients with poorer retention had the highest total health care expenditures, these higher levels of expenditures occurred primarily in the 3 months before initiation of buprenorphine treatment and continued during the first 3 months after treatment initiation, apparently reflecting some degree of acute clinical instability that may have led to both high pretreatment service use and greater out-of-pocket expenditures, which continued after buprenorphine initiation. These elevated expenditures may have stressed patient finances and thereby contributed to early buprenorphine discontinuation due to actual or perceived inability to afford further services.
Our results also showed that persons who discontinued buprenorphine early received significantly lower doses of buprenorphine per day, doses that may have been subtherapeutic for some patients, than those who were retained longer, a finding consistent with previous research (24–26). The fact that these patients also had the longest time to first buprenorphine prescription further suggests less readiness, poorer motivation, or reduced financial ability to initiate and sustain buprenorphine treatment. Consistent with these findings, previous studies of polysubstance use disorder have found that use of multiple addictive substances is associated with more total days of substance use per month (27) as well as with more serious psychosocial dysfunction and higher levels of service use (28).
Our initial findings on expenditures over the entire year were similar to those of Tkacz et al. (18), who also used observational data from private insurance claims data to examine the impact of buprenorphine adherence on service use and expenditures and who concluded that more extensive use of buprenorphine was associated with relatively lower outpatient and inpatient annual health care expenditures. Tkacz et al. relied on the medication possession ratio (proportion of days in a year during which patients were covered by filled prescriptions) rather than duration of retention to characterize adherence to buprenorphine treatment, as in this study. But the critical difference between the methods of Tkacz et al. and of our study is that in addition, we examined preinitiation costs and the timing of expenditure differences. Even after adjustment for pretreatment values, expenditures were highest in the first part of the treatment year. The association of higher overall expenditures, higher out-of-pocket expenditures, and more extensive comorbidities with poorer retention suggests that early discontinuation and poor retention might be explained, at least in part, by financial considerations, although other factors—such as illness severity and relapse—may explain these patterns, too.
Longitudinal analysis can be a critical tool for uncovering alternative interpretations in observational studies. Ultimately, a multisite randomized clinical trial would ideally be needed to evaluate the cost and cost-effectiveness of buprenorphine therapy for opioid use disorder and the benefits of extended retention. However, withholding this treatment, well known as an effective therapy for a serious condition, is likely to be ethically unacceptable. This observational study provides a unique window on distinctive clinical and financial antecedents of early discontinuation, suggesting that underlying differences between the subgroups in clinical variables and expenditures before buprenorphine initiation may influence retention.
The study had several methodological limitations that require comment. First, we assumed, but could not prove, that the requirement for patients to have a 3-month window with no buprenorphine use indicated that they were initiating a new episode of buprenorphine treatment, specifically, and more generally, of MAT. Although it is possible that some patients had been taking methadone, claims for methadone are uncommon in private insurance claims data, and it seems likely that few patients, if any, would have switched from methadone to buprenorphine. Second, this study, like others before it, was based on a convenience sample of patients initiating buprenorphine treatment. We had hoped to find that there was equivalence in health care spending across the sample at the time of buprenorphine initiation. However, our hope was not realized. In fact, we found marked nonequivalence in expenditures, suggesting that financial stress, along with the high levels of concurrent behavioral diagnoses, may have been an important reason for early discontinuation, a factor not previously noted in the literature.
Third, the data for our study came from private insurance claims and as such, do not include uninsured services or nonprescribed drugs, such as illicit opioids, which might affect treatment discontinuation. The cost consequences of illicit opiate use are thus not addressed, given that this study focuses only on the cost perspective of the privately funded health care system and not of society as a whole. Furthermore, we considered both interrupted and uninterrupted treatment as a continuous episode of care because interruptions are hard to model analytically. In addition, our results might not be generalizable to people with other types of insurance, such as Medicaid and Medicare, or to the uninsured. We used p<0.05 as our criterion for significance, in spite of making multiple comparisons, given that this was a descriptive rather than a hypothesis-testing study. Finally, as with any study based on administrative data, we did not have detailed clinical data or information on the reasons for discontinuation (e.g., relapse, successful detoxification, and financial stress) (14, 15).
Conclusions
Among patients prescribed buprenorphine for opioid use disorder, expenditures for buprenorphine medication were a modest proportion of all health care expenditures. Expenditures for other services were far more substantial, especially among patients in the poorest retention group, both before they began buprenorphine treatment and early in the postinitiation period. Thus, although our study could not address whether, other factors being equal, greater buprenorphine retention leads to lower health care expenditures other than drugs, it is the first to suggest that poor retention in buprenorphine treatment among privately insured adults may be the result, at least partially, of both clinical and financial burdens that precede treatment initiation. The high cost of health care, even to those with private insurance, has long been recognized as a major political challenge in the United States. This study suggests that one additional benefit of reducing such costs may be improved adherence to treatment.
1 Results from the 2016 National Survey on Drug Use and Health: Detailed Tables. Rockville, MD, Center for Behavioral Health Statistics and Quality, 2017. https://www.samhsa.gov/data/sites/default/files/NSDUH-DetTabs-2016/NSDUH-DetTabs-2016.htmGoogle Scholar
2 : Treatment of opioid dependence and coinfection with HIV and hepatitis C virus in opioid-dependent patients: the importance of drug interactions between opioids and antiretroviral agents. Clin Infect Dis 2005; 41(Suppl 1):S89–S95Crossref, Medline, Google Scholar
3 : Association between hepatitis C virus and opioid use while in buprenorphine treatment: preliminary findings. Am J Drug Alcohol Abuse 2015; 41:88–92Crossref, Medline, Google Scholar
4 HIV and Substance Use in the United States. Atlanta, Centers for Disease Control and Prevention, 2018. https://www.cdc.gov/hiv/risk/substanceuse.html. Accessed April 19, 2019Google Scholar
5 : Overdose deaths involving opioids, cocaine, and psychostimulants with abuse potential—United States, 2015–2016. MMWR Morb Mortal Wkly Rep 2018; 67:349–358Crossref, Google Scholar
6 : Prescription opioid abuse: a literature review of the clinical and economic burden in the United States. Popul Health Manag 2014; 17:372–387Crossref, Medline, Google Scholar
7 : New federal initiatives to enhance the medical treatment of opioid dependence. Ann Intern Med 2002; 137:688–692Crossref, Medline, Google Scholar
8 : Medication-assisted treatment with buprenorphine: assessing the evidence. Psychiatr Serv 2014; 65:158–170Link, Google Scholar
9 : Compliance with buprenorphine medication-assisted treatment and relapse to opioid use. Am J Addict 2012; 21:55–62Crossref, Medline, Google Scholar
10 : Medication-assisted therapies—tackling the opioid-overdose epidemic. N Engl J Med 2014; 370:2063–2066Crossref, Medline, Google Scholar
11 : Managing opioid addiction with buprenorphine. Am Fam Physician 2006; 73:1573–1578Medline, Google Scholar
12 : Medicine: Medications for Opioid Use Disorder Save Lives. Washington, DC, National Academies Press, 2019Google Scholar
13 : The evidence doesn’t justify steps by state Medicaid programs to restrict opioid addiction treatment with buprenorphine. Health Aff (Millwood) 2011; 30:1425–1433Crossref, Medline, Google Scholar
14 : Three-year retention in buprenorphine treatment for opioid use disorder nationally in the Veterans Health Administration. Am J Addict 2017; 26:572–580Crossref, Medline, Google Scholar
15 : Three-Year Retention in Buprenorphine Treatment for Opioid Use Disorder Among Privately Insured Adults. Psychiatr Serv 2018; 69:768–776Link, Google Scholar
16 : Impact of medication-assisted treatment for opioid addiction on Medicaid expenditures and health services utilization rates in Vermont. J Subst Abuse Treat 2016; 67:9–14Crossref, Medline, Google Scholar
17 : Cost analysis of clinic and office-based treatment of opioid dependence: results with methadone and buprenorphine in clinically stable patients. Drug Alcohol Depend 2009; 99:132–140Crossref, Medline, Google Scholar
18 : Relationship between buprenorphine adherence and health service utilization and costs among opioid dependent patients. J Subst Abuse Treat 2014; 46:456–462Crossref, Medline, Google Scholar
19 : Cost-effectiveness of long-term outpatient buprenorphine-naloxone treatment for opioid dependence in primary care. J Gen Intern Med 2012; 27:669–676Crossref, Medline, Google Scholar
20 : Cost and utilization outcomes of opioid-dependence treatments. Am J Manag Care 2011; 17(Suppl 8):S235–S248Medline, Google Scholar
21 : A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40:373–383Crossref, Medline, Google Scholar
22 : Estimating log models: to transform or not to transform? J Health Econ 2001; 20:461–494Crossref, Medline, Google Scholar
23 : Heterogeneity of nonadherent buprenorphine patients: subgroup characteristics and outcomes. Am J Manag Care 2017; 23:e172–e179Medline, Google Scholar
24 : Effect of buprenorphine dose on treatment outcome. J Addict Dis 2012; 31:8–18Crossref, Medline, Google Scholar
25 : One-year follow-up of heroin-dependent adolescents treated with buprenorfine/naloxone for the first time in a substance treatment unit. J Subst Abuse Treat 2016; 67:1–8Crossref, Medline, Google Scholar
26 : Treatment retention among patients randomized to buprenorphine/naloxone compared to methadone in a multi-site trial. Addiction 2014; 109:79–87Crossref, Medline, Google Scholar
27 : Polysubstance use among veterans in intensive PTSD programs: association with symptoms and outcomes following treatment. J Dual Diagn 2019; 15:36–45Crossref, Medline, Google Scholar
28 : Clinical epidemiology of single versus multiple substance use disorders: polysubstance use disorder. Med Care 2017; 55(Suppl 9 Suppl 2):S24–S32Crossref, Medline, Google Scholar