Healthcare utilization and costs associated with treatment for opioid dependence

Abstract

Objective: Opioid use disorder (OUD) can be managed with medication assisted therapy (MAT) (methadone [MET], buprenorphine [BUP], or extended-release naltrexone [XR-NTX]) or counseling alone (non-pharmacological therapy [NPT]). The objective of this study was to evaluate healthcare resource utilization and costs associated with XR-NTX compared with alternative treatments for opioid dependence.

Methods: Adults with a diagnosis of opioid dependence who initiated treatment with XR-NTX, BUP, MET, or NPT between January 1, 2011 and December 31, 2014 were identified in the Truven Health MarketScan Commercial administrative claims database. Healthcare resource utilization, costs (inpatient [IP], emergency department [ED], outpatient [OP], and pharmacy) and adherence were evaluated for each cohort during 12-month baseline and follow-up periods.

Results: A total of 29,235 patients were included in the analysis; 1,041, 20,566, 745, and 6,883 received XR-NTX, BUP, MET, and NPT, respectively. Patients in the XR-NTX cohort were significantly younger and had more comorbidities compared with the other cohorts. Patients in the XR-NTX group had the largest percentage decrease in IP and ED utilization and costs from baseline to follow-up. OP and pharmacy costs increased significantly from baseline to follow-up for all cohorts. Overall, there was no significant change in total healthcare costs for the XR-NTX group, whereas the costs increased significantly for other groups (BUP = +43%, MET = +47.7%, NPT = +38.8%).

Conclusions: Healthcare resource utilization and costs increased from baseline to follow-up in BUP, MET, and NPT patients, whereas patients receiving XR-NTX experienced no such increase. This analysis suggests there may be economic value in the use of XR-NTX for OUD.

Keywords:

• Naltrexone
• opioid dependence
• buprenorphine
• methadone
• healthcare costs
• healthcare resource utilization

Introduction

The opioid epidemic in the US is considered a major public health crisis and continues to grow¹. Opioid use disorder (OUD) is characterized as a problematic pattern of opioid use, leading to clinically significant impairment or distress². Approximately 2 million adults aged 18 years and older had an OUD diagnosis in 2015³. According to the Centers for Disease Control and Prevention (CDC), overdose deaths related to opioids (including prescription opioids, heroin, fentanyl, and other synthetic opioids) tripled between 2000 and 2014⁴. There were more than 52,000 drug overdose deaths in 2015, and ∼63% of these deaths involved an opioid⁴. Additionally, deaths from opioid- and heroin-related overdoses are under-reported and may actually be up to 24% higher⁵.

Along with the increasing prevalence of and mortality rates from OUD, healthcare resource utilization due to the disease has increased considerably and is associated with a significant socioeconomic burden^6–12. Based on data from the Agency for Healthcare Research and Quality (AHRQ) Healthcare Cost and Utilization Project (HCUP), opioid-related inpatient stays and emergency department visits increased by 64.1% and 99.4%, respectively, between 2005 and 2014¹². The total economic burden of OUD in the US was estimated to be greater than $500 billion in 2015, which is equivalent to 2.8% of the US GDP in 2015¹³. Studies have reported average annual healthcare costs for patients diagnosed with OUD to be between $10,000 and $18,000 higher than for similar patients without an OUD diagnosis^6,10,11.

Patients with OUD can be treated using psychosocial, or “non-pharmacological treatment” (NPT), alone or with medication-assisted therapy (MAT) in conjunction with NPT¹⁴. Current treatment guidelines encourage the use of MAT combined with NPT^14–18. The three medications currently approved by the US Food and Drug Administration (FDA) for the treatment of opioid dependence are buprenorphine, with or without naloxone (BUP), methadone (MET), and extended-release injectable naltrexone (XR-NTX).

MET is a full µ-opioid receptor agonist and BUP is a partial µ-opioid receptor agonist^19–21. Both medications are effective in reducing the use of other opiate drugs; buprenorphine is also associated with reduced cravings for opiates^14,22. These treatments are associated with the potential for diversion, and can only be dispensed through specially licensed treatment facilities (MET) or by providers who have undergone special training and obtained a waiver for treating OUD patients (BUP)^{14,16,23–25}. XR-NTX, a µ-opioid receptor antagonist, is indicated for the prevention of relapse to opioid dependence following opioid detoxification²⁶. XR-NTX received US FDA approval as a treatment option for opioid dependence in 2010, and is one of the most recent approved treatments for opioid dependence. XR-NTX is also US FDA approved for treatment of alcohol dependence disorder²⁷. Opioid detoxification is required prior to initiating treatment with XR-NTX to avoid precipitating withdrawal²⁶. XR-NTX is a treatment option that may reduce cravings and assist patients in abstaining from opioids by blocking opioid receptors²⁸. There is no abuse potential or risk for diversion with XR-NTX, and the once-monthly dosing via injection prevents the need for daily administration and can be convenient for both patients and providers^16,28,29.

Few studies have compared the real-world healthcare resource utilization and costs associated with XR-NTX relative to other treatment options for OUD^30–32. Existing studies have been limited by the fact that they were conducted prior to FDA approval of XR-NTX³⁰, were done within a framework of a model³¹, or estimated healthcare costs as part of a clinical trial³². An improved understanding of the treatment patterns and costs associated with the use of various treatment modalities for patients with a diagnosis of OUD in real world settings can help managed care organizations, behavioral health organizations, and state payers better address the disease burden associated with OUD. The objective of this study was to evaluate healthcare resource utilization and costs (both OUD and non-OUD related) among patients diagnosed with opioid dependence treated with XR-NTX compared with BUP, MET, or NPT using data from administrative commercial claims.

Methods

Data source

We conducted a retrospective cohort analysis using the Truven Health MarketScan^® Commercial Claims and Encounters Database from January 1, 2010 to December 31, 2015³³. The database contains the pooled healthcare experience of over 120 million individuals encompassing employees, their spouses and dependents who are covered by employer-sponsored private health plans in the US. Data on inpatient, outpatient, and emergency department encounters, as well as outpatient prescription drug claims are captured and are linked by a unique patient identifier. The database includes a variety of fee-for-service, preferred provider organizations (PPO), and capitated health plans. The data contain de-identified information and conforms to the Health Insurance Portability and Accountability Act (HIPAA) confidentiality requirements. Therefore, no institutional review board (IRB) approval was required.

Study sample

Patients were included in the study if they had ≥2 claims for XR-NTX, BUP, or MET, or ≥3 claims for NPT within a period of 45 days between January 1, 2011 and December 31, 2014. This requirement is in alignment with the HEDIS quality measure for the initiation and engagement in addiction treatment³⁴. The date of the first claim for treatment was defined as the index date. Patients were screened for XR-NTX claims before other treatments claims. Hence, if patients had ≥2 claims for XR-NTX within 45 days, then they were classified in the XR-NTX cohort, regardless of if they were prescribed other treatments before initiating XR-NTX. This was because patients prescribed XR-NTX need to be detoxified before they initiate XR-NTX³⁵, and agonist medications are often used for induction onto XR-NTX. Also, XR-NTX was approved very recently; hence, most of the patients prescribed XR-NTX would have been switched from other treatments. Furthermore, patients were required to be at least 18 years old on the index date and to have at least one medical claim with a diagnosis of opioid dependence (International Classification of Diseases 9th Revision, Clinical Modification [ICD-9 CM] codes: 304.0x, 304.7x) in the 6 months prior to or on the index date.

Patients treated with XR-NTX and BUP were identified by Healthcare Common Procedure Coding System (HCPCS) codes (J2315 for XR-NTX and J0571–J0575 for BUP) in the medical claims database and National Drug Codes (NDC) in the pharmacy claims database. Patients treated with MET were identified by HCPCS codes (H0020) only. Patients in the NPT cohort were identified by substance abuse counseling claims, detoxification facility claims, or substance abuse treatment facility claims.

Patients were required to have continuous enrollment with medical and pharmacy benefits for at least 12 months prior to the index date (baseline period) and 12 months after the index date (follow-up period). The follow-up period starts from the date of the first claim of the treatment. Patients with any use of pharmacological treatment for opioid dependence in 1 month prior to the index date were excluded from the BUP, MET, and NPT cohorts. This exclusion criterion was not applied to the XR-NTX cohort for the reasons stated above.

Study variables

Baseline demographics, including age, gender, geographic region, and health plan type were assessed for all patients. Comorbidity burden was estimated using the Quan adaptation of the Charlson Comorbidity Index (CCI) score³⁶, Elixhauser Comorbidity Index score³⁷, psychiatric comorbidities, other medical comorbidities, and number of psychiatric medications in the baseline period. Healthcare resource utilization and costs, including inpatient admissions, inpatient length of stay (LOS), emergency department (ED) visits, physician office visits, outpatient hospital services, and medication use were assessed during the baseline and follow-up periods for each cohort. All-cause, OUD-related, and non-OUD-related healthcare resource utilization and costs outcomes were calculated. Treatment patterns for the MAT cohorts were measured in terms of proportion of days covered (PDC), medication possession ratio (MPR), and persistent days during the follow-up period. PDC was defined as the ratio of number of days in the follow-up period covered by the index medication to the total number of days in the follow-up period. MPR was defined as the ratio of day’s supply of the index medication to the total number of days in the follow-up period. Inpatient days were excluded while estimating PDC and MPR, as medication received during an inpatient stay is not visible in administrative claims data. Persistence was defined as the number of days from the index date until the date of discontinuation. Discontinuation of therapy was defined as a gap in coverage of the index medication of 45 days or more in the follow-up period. For XR-NTX, the day’s supply was assumed to be 28 days for each medical or pharmacy claim. We did not analyze switching because of the study sample definition. XR-NTX patients were defined prior to other cohorts and could have other treatments in the baseline period, so patients switching to XR-NTX are already included as the study cohort.

Statistical analysis

Descriptive and bivariate statistics were used to evaluate differences in demographics and clinical characteristics between XR-NTX and other cohorts. Chi-square/Fisher’s exact tests and t-tests were used to compare categorical and continuous variables between the cohorts. Differences in healthcare resource utilization and costs between the baseline and follow-up periods were assessed within each cohort as well as between XR-NTX and the other cohorts. McNemar’s test was used to determine statistical significance of within-group changes in the proportion of patients with specific healthcare resource utilization, and the paired t-test was used to determine statistical significance of within-group changes in mean measures of healthcare resource utilization and costs between the baseline and follow-up periods. Logistic regression analyses were performed to assess between-group differences from the baseline to follow-up period in the proportion of patients with healthcare resource utilization. Negative binomial regression analyses and generalized linear models (GLM) with a log-link function and gamma distribution were performed to assess between-group differences in changes in mean healthcare resource utilization and costs, respectively. Linear regression analyses were performed to assess differences in PDC, MPR, and persistence between cohorts during the follow-up period.

All the above regression analyses were adjusted for baseline demographics and clinical characteristics such as age, gender, health plan type, geographic region, year of index date, Elixhauser comorbidity score, number of prior opioid dependence diagnoses, psychiatric comorbidities, other medical comorbidities, and number of psychiatric medications. There were no adjustments made for multiplicity. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC). Differences with a p-value of <.05 were considered statistically significant.

Results

Demographics and clinical characteristics

A total of 29,235 patients were included in the analysis: 1,041, 20,566, 745, and 6,883 received XR-NTX, BUP, MET, and NPT, respectively (Figure 1). The demographic and clinical characteristics of the patients in the study are presented in Table 1. Patients in the XR-NTX cohort were significantly younger than patients in the other three cohorts (29.9 years for XR-NTX vs 35.8, 37.9, and 33.2 years for BUP, MET, and NPT, respectively). The majority of patients in the MAT cohorts were male, whereas the proportion of females was higher in the NPT cohort. Baseline Elixhauser comorbidity index scores, which are indicative of a patient’s overall disease severity, were significantly higher for patients in the XR-NTX cohort compared with patients in the other cohorts (2.6 for XR-NTX vs 2.0, 1.5, and 2.4 for BUP, MET, and NPT, respectively). Similarly, patients in the XR-NTX group had higher rates of psychiatric comorbidities, including alcohol dependence, anxiety disorder, bipolar disorder, and depression, and were prescribed more psychiatric medications at baseline than the other patients.

Figure 1. Patient attrition.

Table 1. Baseline demographics and clinical characteristics.

Characteristics	XR-NTX (n = 1,041)	BUP (n = 20,566)	MET (n = 745)	NPT (n = 6,883)	p-value XR-NTX vs BUP	p-value XR-NTX vs MET	p-value XR-NTX vs NPT
Age (years), mean (SD)	29.9 (10.93)	35.8 (12.46)	37.9 (13.03)	33.2 (13.43)	<.001	<.001	<.001
Gender, n (%)
Male	718 (69.0)	13,150 (63.9)	444 (59.6)	2,679 (38.9)	.001	<.001	<.001
Female	323 (31.0)	7,416 (36.1)	301 (40.4)	4,204 (61.1)
Region, n (%)
Northeast	265 (25.5)	5,675 (27.6)	158 (21.2)	2,264 (32.9)	.028	.036	<.001
North Central	220 (21.1)	4,304 (20.9)	142 (19.1)	1,421 (20.7)
South	328 (31.5)	6,836 (33.2)	242 (32.5)	1,937 (28.1)
West	225 (21.6)	3,666 (17.8)	202 (27.1)	1,235 (17.9)
Unknown	3 (0.3)	85 (0.4)	1 (0.1)	26 (0.4)
Health plan type, n (%)
Comprehensive	27 (2.59)	632 (3.07)	16 (2.15)	228 (3.3)	<.001	.013	<.001
PPO	636 (61.10)	12,316 (59.89)	490 (65.77)	4,174 (60.6)
HMO	96 (9.22)	2,585 (12.57)	57 (7.65)	910 (13.2)
POS	112 (10.76)	1,880 (9.14)	100 (13.42)	646 (9.4)
Other	150 (7.59)	1,826 (8.88)	72 (9.66)	689 (10.0)
Unknown	20 (1.48)	1,327 (6.45)	10 (1.34)	236 (3.4)
Quan-Charlson comorbidity score, mean (SD)	0.3 (0.78)	0.3 (0.84)	0.3 (1.03)	0.4 (0.90)	.850	.521	.080
Elixhauser comorbidity score, mean (SD)	2.6 (1.57)	2.0 (1.56)	1.5 (1.48)	2.4 (1.72)	<.001	<.001	<.001
Psychiatric comorbidities, n (%)
Alcohol dependence	293 (28.1)	1,559 (7.6)	21 (2.8)	1,345 (19.5)	<.001	<.001	<.001
Anxiety disorder	440 (42.3)	5,821 (28.3)	129 (17.3)	2,394 (34.8)	<.001	<.001	<.001
Bipolar disorder	141 (13.5)	1,536 (7.5)	31 (4.2)	740 (10.8)	<.001	<.001	.008
Cannabis dependence	166 (15.9)	886 (4.3)	13 (1.7)	922 (13.4)	<.001	<.001	.026
Cocaine dependence	72 (6.9)	405 (2.0)	9 (1.2)	332 (4.8)	<.001	<.001	.004
Dementias	3 (0.3)	21 (0.1)	0 (0.0)	7 (0.1)	.079	.143	.114
Depression	496 (47.6)	6,599 (32.1)	133 (17.9)	2,814 (40.9)	<.001	<.001	<.001
Nicotine dependence	311 (29.9)	3,998 (19.4)	111 (14.9)	1,549 (22.5)	<.001	<.001	<.001
PTSD	50 (4.8)	432 (2.1)	21 (2.8)	293 (4.3)	<.001	.034	.420
Schizophrenia	4 (0.4)	103 (0.5)	4 (0.5)	64 (0.9)	.601	.634	.075
Sedative, hypnotic or anxiolytic dependence	150 (14.4)	964 (4.7)	17 (2.3)	726 (10.5)	<.001	<.001	.002
Sleep disorder	35 (3.4)	866 (4.2)	19 (2.6)	278 (4.0)	.181	.323	.296
Other comorbidities, n (%)
Hepatitis B	3 (0.3)	31 (0.2)	2 (0.3)	7 (0.1)	.275	.938	.114
Hepatitis C	39 (3.7)	590 (2.9)	40 (5.4)	217 (3.2)	.100	.100	.313
HIV	4 (0.4)	57 (0.3)	4 (0.5)	25 (0.4)	.525	.634	.917
Opiates and related narcotic poisoning	77 (7.4)	469 (2.3)	10 (1.3)	250 (3.6)	<.001	<.001	<.001
Number of psychiatric medications, mean (SD)	5.8 (4.86)	4.8 (4.3)	4.1 (3.91)	3.1 (2.07)	<.001	<.001	<.001

Abbreviations. SD, Standard deviation; XR-NTX, extended-release naltrexone; BUP, Buprenorphine; MET, Methadone; NPT, Non-pharmacological therapy; PPO, Preferred Provider Organization; HMO, Health Maintenance Organization; POS, Point of Service; PTSD, Post-traumatic Stress Disorder; HIV, Human Immunodeficiency Virus.

p-values based on t-test and Chi-square/Fisher’s exact test, where applicable.

Healthcare resource utilization

Table 2 contains healthcare resource utilization in the baseline and follow-up periods for each cohort. Patients in the XR-NTX group had significantly higher baseline healthcare resource utilization compared with patients in the other treatment groups. Among XR-NTX patients, both the proportion of patients hospitalized (not shown) and the mean number of inpatient hospitalizations decreased significantly between the baseline and the follow-up periods (by 44.9% and 46.6%, respectively). A significant reduction in mean number of inpatient visits was observed for patients in the BUP and NPT cohorts, although the percentage decrease was approximately over one-half of the declines observed in the XR-NTX cohort (20.8% and 15.1% for BUP and NPT, respectively). There was no significant difference in the number of inpatient admissions from baseline to follow-up for the MET group. The total inpatient admission days decreased significantly from baseline to follow-up for patients in the XR-NTX group; the 56.8% decrease was more than 6-times the decrease observed in the BUP (8.8%) cohort. No significant changes were observed in total inpatient admission days from baseline to follow-up for the MET and NPT cohorts. For OUD-related hospitalizations, there was a significant decrease from baseline to follow-up across all of the MAT cohorts in mean number of hospitalizations (53.9% for XR-NTX vs 20.0% and 59.0% for BUP and MET, respectively), but not NPT.

Table 2. Healthcare resource utilization during baseline and follow-up periods.

Utilization category	XR-NTX (n = 1,041)			BUP (n = 20,566)			MET (n = 745)			NPT (n = 6,883)
	Baseline	Follow-up	Change	Baseline	Follow-up	Change	Baseline	Follow-up	Change	Baseline	Follow-up	Change
Inpatient admissions, mean (SD)	1.15 (1.34)	0.61 (1.14)	−46.6%*	0.43 (0.90)	0.34 (0.87)	−20.8%*	0.25 (0.70)	0.20 (0.61)	−23.2%	0.75 (1.07)	0.63 (1.13)	−15.1%*
OUD-related	0.54 (0.86)	0.25 (0.64)	−53.9%*	0.15 (0.47)	0.12 (0.51)	−20.0%*	0.08 (0.37)	0.03 (0.19)	−59.0%*	0.27 (0.58)	0.27 (0.69)	+0.7%
Non-OUD related	0.61 (1.07)	0.36 (0.86)	−40.2%*	0.28 (0.73)	0.22 (0.66)	−21.5%*	0.18 (0.54)	0.16 (0.54)	−8.0%	0.48 (0.92)	0.36 (0.87)	−24.3%*
Detox and/or rehab	0.91 (1.16)	0.45 (0.93)	−50.6%*	0.25 (0.60)	0.19 (0.61)	−25.3%*	0.13 (0.51)	0.06 (0.33)	−50.0%*	0.48 (0.77)	0.44 (0.92)	−8.1%*
Inpatient, days, mean (SD)	11.55 (16.43)	4.99 (10.38)	−56.8%*	3.02 (7.72)	2.75 (8.61)	−8.8%*	1.65 (6.04)	1.41 (5.64)	−14.8%	6.05 (11.00)	6.01 (11.96)	−0.6%
OUD-related	6.40 (12.55)	2.34 (7.21)	−63.5%*	1.24 (4.81)	1.25 (5.82)	+0.8%	0.66 (4.19)	0.41 (2.78)	−38.9%	2.32 (6.57)	2.96 (8.25)	+27.7%*
Non-OUD related	5.15 (11.59)	2.65 (7.16)	−48.4%*	1.78 (5.71)	1.50 (5.82)	−15.5%*	0.99 (3.57)	1.00 (4.32)	+1.4%	3.73 (9.07)	3.05 (8.67)	−18.1%*
Detox and/or rehab	10.13 (15.42)	4.12 (9.53)	−59.3%*	2.01 (6.10)	1.89 (7.36)	−5.9%*	1.01 (5.10)	0.67 (3.84)	−33.8%	4.43 (9.32)	4.92 (10.91)	+11.1%*
ED visits, mean (SD)	1.27 (2.22)	0.94 (2.39)	−26.1%*	1.06 (2.68)	0.92 (2.54)	−13.3%*	0.89 (2.14)	0.82 (1.53)	−8.6%	1.33 (2.54)	1.12 (2.96)	−15.5%*
OUD-related	0.18 (0.90)	0.12 (1.37)	−34.4%*	0.07 (0.36)	0.06 (0.97)	−10.0%	0.04 (0.23)	0.02 (0.16)	−65.1%*	0.09 (0.40)	0.12 (1.60)	+30.0%
Non-OUD related	1.09 (2.02)	0.82 (1.94)	−24.7%*	0.99 (2.63)	0.86 (2.30)	−13.5%*	0.85 (2.10)	0.80 (1.49)	−5.8%	1.24 (2.49)	1.01 (2.39)	−18.7%*
Outpatient visits, mean (SD)	35.55 (32.21)	44.32 (41.21)	+24.7%*	18.36 (19.90)	30.87 (26.95)	+68.1%*	22.52 (34.89)	66.48 (78.01)	+195.2%*	22.83 (23.95)	44.02 (37.89)	+92.8%*
Office visits	15.39 (15.74)	20.94 (19.20)	+36.1%*	11.69 (13.10)	19.19 (16.81)	+64.2%*	12.56 (23.55)	39.10 (60.40)	+211.2%*	12.73 (15.60)	20.19 (22.60)	+58.6%*
Outpatient hospital	12.33 (19.30)	13.41 (22.19)	+8.7%	3.88 (8.14)	6.05 (12.43)	+55.8%*	4.79 (13.72)	14.34 (41.38)	+199.1%*	6.26 (10.93)	15.07 (21.43)	+140.7**
Laboratory	3.78 (7.92)	5.65 (13.01)	+49.3%*	1.34 (3.36)	3.57 (7.14)	+166.3%*	0.80 (2.21)	1.40 (4.45)	+75.9%*	1.71 (4.34)	4.22 (10.24)	+147.0%*
Other outpatient	4.05 (9.15)	4.33 (11.46)	+7.0%	1.45 (4.09)	2.06 (6.21)	+42.6%*	4.37 (18.85)	11.64 (41.22)	+166.4%*	2.13 (5.50)	4.54 (10.43)	+112.8%*
Pharmacy fills, mean (SD)	28.65 (26.30)	32.44 (25.64)	+13.2%*	25.90 (27.65)	38.97 (28.86)	+50.5%*	21.75 (30.24)	64.89 (76.99)	+198.4%*	24.33 (30.39)	25.26 (30.57)	+3.9%*

Abbreviations. SD, Standard deviation; XR-NTX, extended-release naltrexone; BUP, Buprenorphine; MET, Methadone; NPT, Non-pharmacological therapy.

*p < .05.

p-values for difference between baseline and follow-up periods based on paired t-test.

The number of ED visits decreased by over 26% from baseline to follow-up for the XR-NTX group. The percentage decreases in number of ED visits were approximately one-half this amount for patients in the BUP (13.3%) and NPT (15.5%) cohorts. No significant change was observed in ED visits for the MET cohort. For OUD-related ED visits, the XR-NTX and MET cohorts experienced decreases of 34.4% and 65.1%, respectively. There was no significant decrease in the BUP cohort. Patients in the NPT group experienced an increase in OUD-related ED visits that was not statistically significant.

There was a significant increase in the number of outpatient visits for all cohorts. The percentage increase in outpatient visits for the XR-NTX was lowest compared with other cohorts (24.7% vs 68.1%, 195.2% and 92.8% for BUP, MET, and NPT, respectively). A similar trend was observed for the number of outpatient pharmacy fills (13.2% for XR-NTX vs 50.5%, 198.4%, and 3.9% for BUP, MET, and NPT respectively).

Healthcare costs

Table 3 contains healthcare costs for patients in each cohort during the baseline and follow-up periods. Patients in the XR-NTX group had significantly higher baseline total healthcare costs as well as individual resource category costs compared with patients in the other treatment groups. Patients in the XR-NTX group experienced a decrease in inpatient costs that was substantially greater than patients in the other cohorts (46.8% for XR-NTX compared with 9% for BUP, and no significant change for the MET and NPT cohorts). The XR-NTX group also experienced the highest percentage decrease in OUD-related hospitalization costs from baseline to follow-up (63.9% for XR-NTX compared with 56.2% for MET, a significant increase for NPT by 43.2% and no significant change for BUP).

Table 3. Healthcare costs during baseline and follow-up periods.

Cost category	XR-NTX (n = 1,041)			BUP (n = 20,566)			MET (n = 745)			NPT (n = 6,883)
	Baseline	Follow-up	Change	Baseline	Follow-up	Change	Baseline	Follow-up	Change	Baseline	Follow-up	Change
Inpatient admissions, mean (SD)	$12,099 (19,818)	$6,437 (15,462)	−46.8%*	$4,518 (13,196)	$4,110 (13,136)	−9.0%*	$3,034 (11,196)	$3,142 (13,825)	+3.6%	$7,579 (15,633)	$7,634 (16,650)	+0.7%
OUD-related	$5,507 (13,179)	$1,986 (7,231)	−63.9%*	$1,004 (4,345)	$980 (4,992)	−2.4%	$514 (3,276)	$225 (1,611)	−56.2%*	$1,929 (5,711)	$2,763 (8,485)	+43.2%*
Non-OUD related	$6,592 (15,106)	$4,451 (13,538)	−32.5%*	$3,514 (12,230)	$3,130 (11,967)	−10.9%*	$2,519 (10,427)	$2,917 (13,606)	+15.8%	$5,649 (14,633)	$4,871 (14,161)	−13.8%*
Detox and/or rehab	$9,306 (16,669)	$4,257 (11,811)	−54.2%*	$1,766 (5,975)	$1,636 (6,969)	−7.4%*	$850 (4,355)	$507 (3,071)	−40.3%	$3,759 (8,314)	$4,747 (11,835)	+26.3%*
ED visits, mean (SD)	$1,329 (2,815)	$959 (2,399)	−27.8%*	$1,012 (3,058)	$897 (2,803)	−11.3%*	$763 (1,973)	$806 (2,190)	+5.7%	$1,352 (3,045)	$1,130 (3,073)	−16.4%*
OUD-related	$181 (814)	$113 (625)	−37.9%*	$65 (408)	$47 (384)	−28.5%*	$43 (293)	$14 (162)	−67.2%*	$95 (515)	$67 (467)	−29.4%*
Non-OUD related	$1,147 (2,667)	$847 (2,296)	−26.2%*	$946 (3,004)	$850 (2,748)	−10.1%*	$720 (1,911)	$792 (2,178)	+10.1%	$1,258 (2,979)	$1,063 (3,009)	−15.5%*
Outpatient visits, mean (SD)	$14,926 (22,482)	$19,003 (25,300)	+27.3%*	$5,650 (10,882)	$8,737 (14,096)	+54.6%*	$4,290 (9,818)	$8,278 (10,152)	+93%*	$8,251 (14,214)	$15,679 (24,191)	+90.0%*
Office visits	$2,689 (4,280)	$6,399 (7,456)	+137.9%*	$1,820 (3,060)	$3,082 (4,289)	+69.3%*	$1,664 (4,028)	$3,972 (3,662)	+138.7%*	$2,051 (3,644)	$3,178 (5,995)	+54.9%*
Outpatient hospital	$8,309 (15,905)	$7,862 (15,601)	−5.4%	$2,682 (7,517)	$3,582 (9,443)	+33.6%*	$1,844 (6,351)	$2,758 (6,980)	+49.5%*	$4,391 (9,870)	$8,281 (15,827)	+88.6%*
Laboratory	$1,408 (4,891)	$2,411 (8,683)	+71.3%*	$350 (1,702)	$1,091 (3,543)	+211.6%*	$151 (823)	$278 (1,430)	+84.9%*	$515 (2494)	$1,682 (7,318)	+226.2%*
Other outpatient	$2,520 (8,276)	$2,331 (7,633)	−7.5%	$798 (3,627)	$982 (4,534)	+23.1%*	$631 (3,053)	$1,270 (5,331)	+101.4%*	$1,294 (5,032)	$2,537 (8,267)	+96.1%*
Pharmacy costs, mean (SD)	$2,464 (4,323)	$5,973 (8,146)	+142.4%*	$2,617 (4,960)	$5,987 (6,764)	+128.8%*	$1,134 (4,528)	$1,395 (3,925)	+23%	$2,149 (5,742)	$2,397 (7,113)	+11.5%*
Total costs, mean (SD)	$30,817 (35,240)	$32,372 (33,756)	+5.0%	$13,797 (22,123)	$19,731 (23,953)	+43.0%*	$9,220 (18,479)	$13,621 (19,935)	+47.7%*	$19,331 (26,383)	$26,839 (34,263)	+38.8%*

Abbreviations. SD, Standard deviation; XR-NTX, extended-release naltrexone; BUP, Buprenorphine; MET, Methadone; NPT, Non-pharmacological therapy.

*p < .05.

p-values for difference between baseline and follow-up periods based on paired t-tests.

The overall trend was similar for costs of ED visits. The costs for ED visits went down from baseline to follow-up by almost 28% for XR-NTX, compared with 11.3% and 16.4% for BUP and NPT, respectively. There was no significant change observed among patients in the MET cohort. The costs for outpatient visits increased significantly from baseline to follow-up across all cohorts; however, the percentage increase was lowest for the XR-NTX cohort compared with the other cohorts (27.3% for XR-NTX, compared with 54.6%, 93.0%, and 90.0% for BUP, MET, and NPT, respectively). Costs for pharmacy fills costs also increased significantly from baseline to follow-up for all the cohorts, with XR-NTX experiencing the greatest increase (142.4% for XR-NTX, compared with 128.8%, 23.0%, and 11.5% for BUP, MET, and NPT, respectively).

Finally, with respect to aggregate healthcare costs (which represent the sum of all cost categories), XR-NTX patients experienced no significant change in total costs from the baseline to follow-up period. In contrast, total healthcare costs increased significantly from baseline to follow-up for all of the other cohorts (BUP +43%, MET +47.7%, and NPT cohort +38.8%).

Treatment patterns

Table 4 displays the treatment patterns for patients in each cohort. Medication adherence as measured by PDC for patients in the MAT cohorts was lower for XR-NTX compared with BUP (0.48 ± 0.28 for XR-NTX, compared with 0.69 ± 0.33 for BUP). PDC for XR-NTX and MET were not significantly different. Persistence as measured by the number of days of continuous index medication was also significantly lower for XR-NTX compared with the other MAT groups (161.5 days for XR-NTX vs 249.3 and 235.0 days for BUP and MET, respectively).

Table 4. Treatment patterns during the follow-up period.

Treatment patterns	XR-NTX (n = 1,041)	BUP (n = 20,566)	MET (n = 745)	p-value XR-NTX vs BUP	p-value XR-NTX vs MET
Adherence
MPR, Mean (SD)	0.48 (0.28)	0.69 (0.33)	0.47 (0.40)	<.001	.090
PDC, Mean (SD)	0.44 (0.25)	0.66 (0.32)	0.44 (0.37)	<.001	.240
Persistence (in days)
≥45-days gap, Mean (SD)	161.5 (106.48)	249.3 (134.64)	235.03 (135.89)	<.001	<.001

Abbreviations. SD, Standard deviation; XR-NTX, extended-release naltrexone; BUP, Buprenorphine; MET, Methadone; NPT, Non-pharmacological therapy; PDC, Proportion of days covered; MPR, Medication possession ratio.

p-values based on linear regression analyses after controlling for baseline covariates.

Between groups comparisons in healthcare costs

Between groups comparisons were assessed using regression models (Figure 2, estimates not shown). Findings from the models were consistent with results from descriptive analyses. Specifically, results suggest that patients treated with XR-NTX had significantly greater reductions in inpatient and ED costs relative to other treatments. For outpatient costs, the XR-NTX cohort exhibited a significantly lower percentage increase compared with other treatments. Pharmacy costs were significantly higher from baseline to follow-up for XR-NTX compared with MET and NPT; there was no difference compared with BUP. Finally, between groups comparisons demonstrated a significant difference in the change in total healthcare costs from baseline to follow-up between XR-NTX and other groups (BUP, MET, and NPT), where the other groups experienced an increase in costs (ranging from 38.8–47.7%) compared with XR-NTX (no change) after controlling for baseline covariates.

Figure 2. Difference in healthcare costs during baseline and follow-up periods: between groups comparisons. Abbreviations. XR-NTX, extended-release naltrexone; BUP, Buprenorphine; MET, Methadone; NPT, Non-pharmacological therapy; ED, Emergency department. Reference category: XR-NTX. *denote significant difference when compared with XR-NTX at p < .05. p-values were based on generalized linear models (GLM) with a log-link function and gamma distribution after controlling for baseline covariates.

Discussion

XR-NTX was approved by the FDA in 2010 for the prevention of relapse to opioid dependence. Few studies to date have examined the healthcare utilization and cost burden associated with this medication in comparison with other treatment options for opioid dependence. We used administrative claims data from geographically diverse commercial insurance plans to assess healthcare utilization and costs associated with XR-NTX, BUP, MET, and NPT. We found that, in this database, patients treated with XR-NTX were younger and had higher baseline clinical severity (more comorbidities and psychiatric medications) compared with other treatment groups. This is a similar finding to that in a previous study by Baser et al.³⁰. This increased clinical severity was reflected by the higher healthcare resource utilization and costs during the baseline period among XR-NTX patients relative to patients in the other cohorts. One potential explanation for this is that—as a newer opioid dependence medication—patients may not be prescribed XR-NTX until after they have tried the other medications. Newer medications can be subject to restrictions such as prior authorizations or “fail-first” requirements, meaning that insurers will not provide coverage for these medications until a patient has tried other (less costly) medications. Additionally, as patients initiated on XR-NTX require detoxification prior to initiation, patients and their providers may find it less convenient as a first-line therapy than the other therapies without such a requirement.

With respect to healthcare costs, patients in the XR-NTX group experienced significant reductions in inpatient and ED costs, and these reductions were greater than those for BUP, MET, and NPT patients. All cohorts experienced increased costs associated with outpatient visits and pharmacy from the baseline to follow-up periods, likely due to increased physician encounter and medication use once they commenced treatment for their opioid dependence. The increase in outpatient visit costs was greatest in the MET group, which may be because these patients must return to the treatment center daily to receive their medication. Pharmacy costs increased the most for XR-NTX patients, as it is the most expensive medication of the three MAT options. However, the increase in pharmacy costs from baseline to follow-up periods between XR-NTX and BUP was similar, which could partially be attributed to the differences in persistence between the medications (patients remained on BUP for more days than patients on XR-NTX). When all of the individual resource category costs were totaled, the increases in outpatient and pharmacy costs were offset in the XR-NTX group by a large decrease in inpatient and ED costs, and, as a result, there was no significant difference in the total healthcare costs from the baseline to the follow-up period. In contrast, total costs increased significantly for patients in the other treatment cohorts. While it is tempting to assume that high costs in the baseline period for the XR-NTX group reflect greater inpatient or outpatient detoxification visits just prior to initiating XR-NTX, XR-NTX patients had detoxification-related claims distributed uniformly across the entire baseline period. There was an increase in inpatient detox admissions in the 1 month prior to initiating XR-NTX; however, the percentage increases in inpatient detox admissions were greater in the BUP and the NPT cohorts than XR-NTX. Hence, patients initiating XR-NTX appear to be more severe than the patients in the other cohorts as they required detoxification for several months before they initiated XR-NTX. Also, patients in the XR-NTX had more baseline comorbidities. Therefore, the higher baseline healthcare utilization and costs are likely related to patient comorbidities and severity than the treatment option patients receive.

Between-group comparisons assessed by multivariate regression models suggested similar trends, with significant increases in total healthcare costs for all treatments groups compared with XR-NTX patients. Current findings regarding healthcare resource utilization and costs are consistent with prior research conducted among patients receiving XR-NTX^30,32. A study done by Baser et al.³⁰ demonstrated that total healthcare costs among XR-NTX patients were similar to BUP patients, and they were 49% lower than MET patients. However, this study was conducted prior to XR-NTX approval for opioid dependence; therefore, it represented off-label use of XR-NTX. In a retrospective exploratory analysis of clinical trial data, Soares et al.³² found that XR-NTX did not significantly increase rates of healthcare resource utilization compared with treatment as usual. Finally, cost-effectiveness analyses done by Jackson et al.³¹ showed XR-NTX to be cost-effective medication for treating opioid dependence. Our study expanded on prior research by addressing some of the limitations of these previous studies (e.g. off-label use, clinical trial data, economic modeling). To the best of our knowledge, this is the first study to compare real-world healthcare utilization and costs between XR-NTX and other treatment groups after approval of XR-NTX for opioid dependence using administrative claims data.

Most clinical treatment guidelines and recommendations for the treatment of opioid use disorders suggest that all treatment options should be available to all patients. In 2016, the US government passed the Comprehensive Addiction Recovery Act (CARA), which includes a requirement that all healthcare practitioners treating opioid dependence must provide, either directly or by referral, access to all medications for opioid dependence, including XR-NTX, MET, and BUP³⁸. Managed care organizations are also taking an active role in combatting the opioid epidemic. In 2015, the Academy of Managed Care Pharmacy (AMCP) formed the Addiction Treatment Advisory Group (ATAG), which includes representatives from behavioral health organizations, treatment centers, advocacy groups, health plans, pharmacy benefit management companies, specialty pharmacies, employers, hospitals, and manufacturers. The goal of this group is to evaluate gaps in addiction treatment services and develop recommendations to enhance patient care, and one of the ATAG recommendations was to “Evaluate and update, as needed, managed care policies, processes, and benefit designs related to substance use disorders based on current evidence and evolving understanding of substance use disorders as chronic health conditions”³⁹. Large payers, such as the Centers for Medicare & Medicaid Services (CMS), announced that 2017 Part D formulary and plan benefit designs must not hinder access to MAT for opioid use disorder or the plans would not be approved, and commercial plans are also removing barriers to MAT^40–42. However, in our study, XR-NTX appeared to be reserved for patients who had more comorbid conditions, may have been further along in their disease course, and/or may have already progressed through other medications. As patients receiving XR-NTX had no significant change in healthcare costs before vs after treatment, study findings indicate that improving access to MAT options such as XR-NTX may result in lower healthcare costs and reduce health economic burden, especially if used earlier in the disease course.

While this study provides valuable insights into the real-world healthcare costs of initiating XR-NTX vs other treatment options for opioid dependence, some limitations should be considered. First, administrative claims are subject to data entry errors such as coding inaccuracies or incorrectly entered diagnoses that were coded for administrative processing rather than clinical completeness. Therefore, some clinical information may be unavailable or inaccurate. Also, the presence of a claim for a filled prescription does not indicate whether the medication was taken as prescribed, which is particularly relevant for BUP (not necessarily for XR-NTX, which is administered as injection, and MET, which is administered orally by medical personnel in a treatment facility). Variables such as disease severity, over-the-counter medication use, and patient health behavior are not captured and, therefore, could not be measured and included in our analyses. In addition, this study is based on data from a US commercial claims and encounters database; as such, the findings may not be generalizable to patients who lack commercial insurance, which is a significant proportion of the opioid dependence patients in the US^43,44. BUP in particular may be purchased with cash payments by patients, and these transactions are not captured⁴⁵. Another limitation is that we observed significant differences in baseline characteristics between the groups, including age, comorbidities, and healthcare costs. Additionally, patients may be prescribed different medications or treatments based on patient preferences, prescriber factors, or the point the patient is on in his/her recovery journey. We attempted matching patients using multiple statistical techniques; however, due to these inherent baseline differences, these patients could not be appropriately matched. Hence, we used a difference-in-differences approach to report percentage differences in healthcare costs from baseline to follow-up among the groups, instead of actual cost differences. In addition, we controlled for potential confounding variables in the regression model. However, there may be additional unobserved confounding variables for which we were not able to control. Further studies with more recent data or sub-groups in which patient differences during the baseline period are minimal should be conducted to assess the value of XR-NTX in patients with diagnosed opioid dependence. Finally, we could not analyze mortality or indirect costs due to lost productivity or caregiver burden in this study, which would have provided valuable insight in this population.

Conclusions

In summary, we found that XR-NTX patients, despite having more comorbidities at baseline, had the greatest percentage reduction in inpatient and ED costs from the baseline to follow-up period compared with BUP, MET, and NPT patients. Outpatient costs increased for BUP, MET, and NPT patients more so than for XR-NTX patients; pharmacy costs increased the most for XR-NTX patients. Overall, the increase in total healthcare costs between baseline and follow-up periods was significantly greater for BUP, MET, and NPT patients compared with XR-NTX patients; while the former group experienced increases, there was no change in the XR-NTX group. These findings highlight the value of XR-NTX, extended-release injectable naltrexone, in the treatment of opioid dependence.

Transparency

Declaration of funding

This research was funded by Alkermes, Inc.

Declaration of financial/other relationships

AS and NA are full-time employees and minor shareholders of Alkermes, Inc. MD and MG were employed at Alkermes, Inc. when the study was conducted. KST is an employee for Symlink, LLC, which has received research funds from Alkermes, Inc. in connection with conducting this study and development of this manuscript. Peer reviewers on this manuscript have received an honorarium from JME for their review work, but have no other relevant financial relationships to disclose.

Acknowledgments

The authors would like to thank Xiaowu Sun for assistance in performing statistical analyses for this study. Xiaowu Sun is an employee and minor shareholder of Alkermes, Inc.