Buprenorphine is a partial opioid agonist indicated for the treatment of opioid use disorder (OUD). It reduces symptoms of craving and withdrawal among individuals habituated to opioid use. Buprenorphine for OUD is available on a prescription basis from clinicians who possess a federal waiver. In late 2016, nearly 700,000 individuals in the United States filled prescriptions for buprenorphine, more than doubling the number from the early 2010s.1 The rising use of buprenorphine corresponds with a national increase in addiction and overdose related to opioids. Federal and state efforts have attempted to expand the pool of clinicians prescribing buprenorphine and other medications for addiction treatment (MAT).
Patients treated with buprenorphine have an elevated prevalence of comorbid psychiatric and somatic chronic illness.2 One of the benefits for providing buprenorphine in primary care settings is that patients can receive MAT from the same clinicians that manage other chronic conditions, potentially improving care coordination.3 There are also well-documented benefits of buprenorphine to patient health. For example, compared with counseling without medication, patients on buprenorphine experience improved abstinence and reduced overdose risk.4,5 Patients also tend to fare better when they are maintained on buprenorphine for longer durations rather than shorter-term, tapered treatments. Given the increasing buprenorphine use for MAT and the prevalence of chronic diseases among individuals with OUD, it is important to understand what kind of impact the buprenorphine treatment has on the treatment of unrelated, chronic conditions.
We used longitudinal, individual-level commercial claims data to examine whether initiation of buprenorphine is associated with changes in adherence to medications to treat such conditions. We focused on treatments used for conditions such as cardiovascular disease, diabetes, depression, and bipolar disorder. We hypothesized that adherence to these treatments would improve with buprenorphine treatment, as individuals experience improved self-efficacy and functioning when they manage an addiction, which might better enable them to obtain and adhere to treatments for other conditions.6
We conducted a population-based, retrospective, cohort study among patients with a diagnosis of OUD and incident use of buprenorphine from 2010 to 2015 using data from the Truven Health MarketScan Commercial Claims and Encounters database (Truven Health Analytics, Ann Arbor, MI).7 Marketscan is one of the largest commercial claims databases in the United States and captures medical and pharmacy information for >25 million individuals annually. These data consist of individual-level demographic characteristics such as patient age and sex, enrollment indicators for health plan and prescription drug coverage, diagnosis information based on International Classification of Diseases, Ninth Revision, Clinical Modification codes (ICD-9-CM), and pharmacy information such as the quantity, days of supply, and National Drug Code (NDC) identification numbers of each prescription.
We initially identified 428,766 patients with a diagnosis of an OUD between 2010 and 2015, of whom 124,926 patients also had a prescription of buprenorphine. We identified buprenorphine medications using NDCs provided by the Centers for Disease Control and Prevention (CDC) and all NDCs associated with buprenorphine were included.8 We considered an individual’s first buprenorphine prescription date as their index date; we derived some information, such as baseline use of each of the 5 therapeutic classes of interest, from the 180-day period before the index date (preindex period), and we extended the study to 180 days after each patients’ index date. We then restricted our study sample to patients who had complete medical and pharmacy records during the 360-day period of observation and who filled buprenorphine for 60 or more days (allowing for gaps of ≤10 d), increasing the likelihood that such use was for long-term maintenance rather than a short-term taper among individuals rapidly discontinuing opioids as part of treatment for OUD. Lastly, we also required an outpatient visit within the 30-day period before the index date to ensure the incident buprenorphine prescription was associated with a clinical visit.
Among the remaining 12,719 eligible subjects, we identified patients receiving chronic disease medications who had: (1) at least 1 prescription for 1 or more of the 5 classes of chronic medications dated no later than the start of the preindex period; and (2) at least 1 prescription of such medication during the study period. The products we used to identify these medications are listed in eTable 1 (Supplemental Digital Content 1, http://links.lww.com/MLR/B814). Thus, our final sample included patients with: (1) antilipids (N=771/6.06%); (2) antipsychotics (N=704/5.54%); (3) antiepileptics (N=2376/18.68%); (4) antidiabetics (N=226/1.78%); and (5) antidepressants (N=4647/36.54%). Figure 1 depicts the exclusion and inclusion criteria used to derive our sample.
Measure of Adherence
We combined information on days’ supply, which we truncated at 90 days, and the dispensing date of each prescription, to create 2 binary daily indicators for whether an individual had: (1) buprenorphine; and (2) one of 5 classes of chronic disease medications on hand for each day of the 360-day study period. In addition to these daily indicators, we also generated adherence rates for buprenorphine and each of the 5 therapeutic classes during the 180-day preindex and the 180-day postindex period. To do so, we first calculated days with drugs on hand, a simple count variable (0–180 d) that reflects the number of days with specific drugs on hand for each patient over the 180-day preindex or postindex period. Proportion of days covered (PDC) with medications on hand (0%–100%) was just days with drugs on hand divided by 180 days. As we limited our sample to incident buprenorphine patients, the preindex adherence rate of buprenorphine was equal to 0 by design.
In addition to patient age and sex, we used diagnosis information from the preindex period and Clinical Classification Software (CCS) from the Agency for Healthcare Research and Quality (AHRQ) to evaluate morbidity. The AHRQ CCS aggregates patient diagnoses and procedures in a number of clinically meaningful categories9,10; we calculated the number of CCS categories each patient had and used this CCS Score as a proxy for individuals’ overall morbidity.11–13 Higher CCS scores suggest patients with higher levels of comorbidity.9,14,15
Each patient contributed 180 records to the analysis, one for each day of observation after buprenorphine initiation. Each record contained fixed variables for age, sex, preindex chronic disease medication adherence, a binary indicator for having buprenorphine on hand on a given day, and an additional daily indicator representing one of the five chronic disease medications of interest; we excluded observations with missing information on any variable. The association between buprenorphine and each of the 5 therapeutic classes was examined separately, so 5 separate cohorts were constructed, one for each therapeutic class of interest. Individuals were allowed to contribute to >1 cohort; for example, 3174 (56.85%) of 5583 eligible subjects included contributed to 1 cohort only, 1754 (31.42%) contributed to 2 cohorts, and the remainder contributed to 3 or more cohorts.
We first described the characteristics of study subjects who had fills with each therapeutic class. Next, we derived the 2-by-2 classification of having buprenorphine and having chronic medications on hand for each class of medications in the study period and evaluated whether there was association between these 2 variables. Lastly, we performed logistic regression to assess whether having buprenorphine on hand on a given day (independent variable) was statistically associated with having chronic medications on hand on the same day (outcome variable), controlling for demographics (sex, age, age squared), morbidity (5 levels of CCS scores: 0–5, 6–10, 11–15, 16–20, and 21+) and baseline adherence using the PDC. We incorporated generalized estimating equations into the logistic regression models to adjust for the correlation between observations, since each individual contributed 180 observations to the analysis. Finally, we used the results of the generalized estimating equations regressions and the mean values of the control variables in the statistical models to derive the predicted probability of having chronic medications available during periods when buprenorphine was and was not on hand.
Characteristics of Study Sample
Among 12,719 eligible subjects, approximately two-thirds were male (62.8%) and the mean age was 35 years old (Table 1). Approximately half of the subjects had 5 or fewer chronic conditions, while approximately one fourth had 6–10 and the remainder had >10 chronic conditions. Adherence to buprenorphine was relatively high following its initiation; the proportion of buprenorphine days covered during the study period was 84.5%.
A total of 5583 subjects were in at least one of the 5 cohorts, each cohort representing a different therapeutic class (eTable 2, Supplemental Digital Content 2, http://links.lww.com/MLR/B815). There were moderate differences across the cohorts with respect to some characteristics. For example, the mean age was greater for individuals using antilipids or antidiabetic products than among their counterparts, and in general, individuals using antidiabetics had more comorbid conditions than those using the other chronic medications examined.
Trends in Adherence
The unadjusted association between buprenorphine initiation and changes in adherence varied across the 5 therapeutic classes examined. For example, the PDC pre-post buprenorphine initiation decreased for antilipids (61.6%–56.3%) and antiepileptics (54.0%–48.9%), remained stable for antipsychotics (∼47%) and antidiabetics (∼63%), and increased for antidepressants (55.8%–59.1%). Rates of buprenorphine adherence were similar across the different cohorts examined, ranging from 80.7% in the antipsychotics cohort to 85.0% in the antilipids cohort.
Co-occurrence of Buprenorphine and Chronic Disease Medications
Table 2 presents the bivariate association between daily adherence rates to the 5 types of medications on days with buprenorphine on hand versus days without buprenorphine. Among days where buprenorphine was on hand, the percentage of days with chronic medication ranged from 48.4% for antipsychotics to 65.0% for antidiabetics; among days where buprenorphine was not on hand, the percentage of days with chronic medication ranged from 45.6% for antipsychotics to 56.4% for antidiabetics. Across all classes of chronic medications, the percentage of days with medication was always higher during days buprenorphine was on hand; the absolute difference in the percentage, representing the association with having buprenorphine, ranged from 3.9% for antiepileptics to 8.8% for antidepressants.
Adjusted Association Between Buprenorphine and Chronic Disease Medications
Table 3 presents the crude and adjusted association between buprenorphine utilization and adherence to treatments for chronic, unrelated conditions. For example, after adjusting for control variables, there was a statistically significant association between buprenorphine utilization and use of antilipids [odds ratio (OR), 1.27; 95% confidence interval (CI), 1.04–1.54]. Similar associations were observed when examining antiepileptics (OR, 1.22; 95% CI, 1.10–1.36) and antidepressants (OR, 1.42; 95% CI, 1.32–1.60), while the association between buprenorphine and the use of antipsychotics (OR, 1.16; 95% CI, 0.97–1.40) or antidiabetics (OR, 1.25; 95% CI, 0.93–1.79), while directionally positive, did not reach statistical significance.
Probability of Having Chronic Disease Medications on Hand
Table 3 also presents the predicted probability of having chronic medication on hand based on buprenorphine utilization. For example, the adjusted probability of having antilipids without buprenorphine was 0.52 (CI, 0.46–0.57), while with buprenorphine it was 0.57 (0.55–0.60), representing a 5 percentage point absolute and a ∼10% relative increase in the probability of having these medications on hand. In all cases, the adjusted probability was greater with buprenorphine than without it, with the largest effect of buprenorphine observed with respect to antidepressants, where the adjusted probability increased from 0.54% (0.52–0.56) to 0.62% (0.61–0.64), representing a 8 percentage absolute and a 15% relative increase in the probability associated with buprenorphine treatment.
With >2.1 million Americans with an OUD there remain pressing questions regarding how treatment access for OUD can be optimized, as well as the effect that such access may have on patient’s overall quality of care. In this analysis of commercially insured individuals diagnosed with OUD, we examined the effect of OUD treatment with buprenorphine on individuals’ adherence to medicine for chronic, unrelated conditions. Possession of buprenorphine was associated with a 42% higher odds of having an antidepressant on hand and an increased probability of medication possession across all 5 therapeutic areas examined, ranging from an absolute increase of ∼4 percentage points (antiepileptics) to ∼8 percentage points (antidepressants).
Our findings are of both clinical and economic importance because there is a large, and growing, population with OUD and a high burden of comorbid chronic somatic and psychiatric disease in this population.16 Policy efforts to increase uptake of buprenorphine have been primarily considered a mechanism for managing OUD, yet our study underscores potential spillover benefits to management of other chronic diseases. In general, adherence to medications for common conditions like depression and diabetes is low in the United States.17 Interventions designed to improve adherence have traditionally involved interventions such as patient education, improved dosing schedules and improved patient-physician communication.18 However, successful interventions are complex and labor intensive.19 One prior systematic review found that of 33 studies evaluating more traditional adherence interventions, just under half were associated with a statistically significant increase in adherence.19
To our knowledge, this is the first study assessing the effect of a medication for opioid addiction—buprenorphine—on adherence to treatments for common and costly cardiovascular and psychiatric disease. Our results are consistent with a prior analysis of the effect of another FDA-approved treatment for OUD, methadone, on adherence to treatments for unrelated conditions; in an open cohort of individuals co-infected with human immunodeficiency virus (HIV) and hepatitis C, methadone maintenance therapy was associated with positively associated with highly active antiretroviral therapy adherence, as well as suppression of HIV-1 RNA and increases in CD4 count.20
Although our study was not designed for causal inference, we can nevertheless speculate regarding several potentially complementary pathways that may explain the associations that we describe. First, and most directly, addiction is often characterized by dysregulation,21 diminished self-efficacy,22 and less future orientation,23 and thus when addiction is treated, individuals may experience improved organization and self-efficacy that could lead to better adherence to treatments for comorbid conditions. Second, access to buprenorphine requires engagement with the health care system, and thus the increased adherence to treatments such as antidepressants or antiepileptics may reflect the greater access to care possessed by those being actively treated for OUD. Finally, the greater adherence that is observed upon buprenorphine initiation may represent changing individual life circumstances in which treatment is initiated, such as the gaining of insurance, an intervention by friends or family, or a health shock such as overdose that independently is associated with the availability and acceptance of addiction treatment as well as treatments for other health conditions. Interestingly, the strongest association we identified was between buprenorphine initiation and treatment with antidepressants, which may reflect the greater experience of buprenorphine prescribers with the use of antidepressants, as well as a recognition of the important ways that untreated depression can interfere with adherence to unrelated conditions.24
Our analysis has several limitations. First, we may underestimate the true effects of buprenorphine treatment on adherence to treatments for unrelated conditions, as buprenorphine may be diverted25 and we are not able to assess the magnitude of such behavior, which would bias our results toward the null. Second, we derived cohorts of chronic medication users but we do not have data to identify the precise clinical indication for each medication use. Third, some of the patients we identified may have been prescribed buprenorphine (or antidepressants) for pain, although we attempted to limit the size of this population by requiring a diagnosis of OUD and an office visit prior to the 30-day of the buprenorphine initiation. Fourth, our results are not generalizable to all patients receiving buprenorphine for OUD, as we required individuals to have received at least a 60 day continuous supply, yet many patients receive buprenorphine treatment for shorter and more disrupted periods of time.26,27
Even though there is clear consensus that medications for OUD substantially reduce the likelihood of opioid and all-cause mortality,5 less is known regarding the effects that medication such as buprenorphine may have on treatment for unrelated chronic conditions. Our findings suggest that initiating buprenorphine may be associated with improvements in adherence for other conditions, such as depression, that are both common and costly among those with OUD.
3. Alford DP, LaBelle CT, Kretsch N, et al. Collaborative care of opioid-addicted patients in primary care using buprenorphine
. Arch Intern Med. 2011;171:425–431.
4. Mattick RP, Kimber J, Breen C, et al. Buprenorphine
maintenance versus placebo or methadone maintenance for opioid dependence. Cochrane Database of Syst Rev. 2003:CD002207.
5. Sordo L, Barrio G, Bravo MJ, et al. Mortality risk during and after opioid substitution treatment: systematic review and meta-analysis of cohort studies. BMJ. 2017;357:j1550.
6. Simpson DD. A conceptual framework for drug treatment process and outcomes. J Subst Abuse Treat. 2004;27:99–121.
7. Truven Health Analytics Inc. MarketScan commercial claims and encounters Medicare supplemental user guide: data year 2015 edition. Ann Arbor, MI: Truven Health Analytics; 2016.
9. Cowen ME, Dusseau DJ, Toth BG, et al. Casemix adjustment of managed care claims data using the clinical classification for health policy research method. Med Care. 1998;36:1108–1113.
11. Chang H-Y, Singh S, Mansour O, et al. Association between sodium-glucose cotransporter 2 inhibitors and lower extremity amputation among patients with type 2 diabetes. JAMA Intern Med. 2018;178:1190–1198.
12. Chang HY, Zhou M, Tang W, et al. Risk of gastrointestinal bleeding associated with oral anticoagulants: population based retrospective cohort study. BMJ. 2015;350:h1585.
13. Berry SA, Fleishman JA, Moore RD, et al. Trends in reasons for hospitalization in a multisite United States cohort of persons living with HIV, 2001–2008. J Acquir Immune Defic Syndr. 2012;59:368–375.
14. Wei W-Q, Bastarache LA, Carroll RJ, et al. Evaluating phecodes, clinical classification software, and ICD-9-CM codes for phenome-wide association studies in the electronic health record. PLoS One. 2017;12:e0175508.
15. Tabak YP, Sun X, Nunez CM, et al. Using electronic health record data to develop inpatient mortality predictive model: Acute Laboratory Risk of Mortality Score (ALaRMS). J Am Med Inform Assoc. 2014;21:455–463.
17. Neiman AB, Ruppar T, Ho M, et al. CDC grand rounds: improving medication adherence
for chronic disease management—innovations and opportunities. MMWR Morb Mortal Wkly Rep. 2017;66:1248–1251.
18. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med. 2005;353:487–497.
19. Haynes RB, McDonald H, Garg AX, et al. Interventions for helping patients to follow prescriptions for medications. Cochrane Database Syst Rev. 2002;2:CD000011.
20. Palepu A, Tyndall MW, Joy R, et al. Antiretroviral adherence and HIV treatment outcomes among HIV/HCV co-infected injection drug users: the role of methadone maintenance therapy. Drug Alcohol Depend. 2006;84:188–194.
21. Koob G. Drug addiction, dysregulation of reward, and allostasis. Neuropsychopharmacology. 2001;24:97–129.
22. Hyde J, Hankins M, Deale A, et al. Interventions to increase self-efficacy in the context of addiction behaviours. J Health Psychol. 2008;13:607–623.
23. Bickel WK, Marsch LA. Toward a behavioral economic understanding of drug dependence: delay discounting processes. Addiction. 2001;96:73–86.
24. Grenard JL, Munjas BA, Adams JL, et al. Depression and medication adherence
in the treatment of chronic diseases in the United States: a meta-analysis. J Gen Intern Med. 2011;26:1175–1182.
25. Johanson C-E, Arfken CL, di Menza S, et al. Diversion and abuse of buprenorphine
: findings from national surveys of treatment patients and physicians. Drug Alcohol Depend. 2012;120:190–195.
26. Daubresse M, Saloner B, Pollack HA, et al. Non-buprenorphine
opioid utilization among patients using buprenorphine
. Addiction. 2017;112:1045–1053.
27. Saloner B, Daubresse M, Caleb Alexander G. Patterns of buprenorphine
-naloxone treatment for opioid use disorder
in a multistate population. Med Care. 2017;55:669–676.