Potential barriers to filling buprenorphine and naltrexone prescriptions among a retrospective cohort of individuals with opioid use disorder

Abstract

Introduction:

Medications for opioid use disorder (MOUD) are highly effective, but barriers along the cascade of care for opioid use disorder (OUD) from diagnosis to treatment limit their reach. For individuals desiring MOUD, the final step in the cascade is filling a written prescription, and fill rates have not been described.

Methods:

We used data from a large de-identified database linking individuals’ electronic medical records (EMR) and administrative claims data and employed a previously developed algorithm to identify individuals with a new diagnosis of OUD. We included individuals with a prescription for buprenorphine or naltrexone recorded in the EMR. The outcome was a prescription fill within 30 days as reported in claims data. We compared demographic and clinical characteristics between those who did and did not fill the prescription and used a Kaplan-Meier curve to assess whether fill rates differed based on patient copay.

Results:

We identified 264 individuals with a new diagnosis of OUD who had a prescription written for buprenorphine or oral naltrexone. Of these, 70% (184) filled the prescription within 30 days, and more than half (57%) filled the prescription on the day it was written. Individuals with prescription copay at or below the mean had a 75% fill rate at 30 days compared with 63% for those with copay above the mean (p<0.05) and this difference was consistent across fill times (log rank p-value <0.05).

Conclusions:

It is alarming that nearly 1 in 3 MOUD prescriptions go unfilled. More research is needed to understand and reduce barriers to this final step of the OUD cascade of care.

Introduction

Highly effective medications for opioid use disorder (MOUD) are a reason for optimism amid the ongoing opioid overdose crisis; however, challenges of access, use, and retention undermine the potential of MOUD (J. Morgan, Schackman, Weinstein, Walley, & Linas, 2019; Wakeman et al., 2020). Using a cascade of care framework, receipt of MOUD requires patients to navigate several sequential steps from obtaining a diagnosis of opioid use disorder (OUD) and accessing a provider who prescribes MOUD (Williams, Nunes, Bisaga, Levin, & Olfson, 2019). The final step of filling a written prescription is a known barrier for other chronic conditions, where 20–30% of prescriptions go unfilled (Viswanathan et al., 2012); however, whether this phenomenon applies to MOUD is unknown. In this study, we used linked electronic medical record (EMR) and claims data to assess fill rates for mucosal buprenorphine and oral naltrexone. We characterized the cost-sharing, clinical, and demographic factors associated with filling a prescription.

Methods

We used de-identified administrative claims data from the OptumLabs^® Data Warehouse (OLDW), which includes medical and pharmacy claims, laboratory results, and enrollment records for commercial and Medicare Advantage enrollees. Using a shared identifier, we linked these claims to EMR data containing details on written prescriptions and demographic characteristics, retaining only individuals with both claims and EMR data. We identified individuals with incident OUD evident in claims data between October 2015 and July 2019 using a previously developed algorithm (Wakeman et al., 2020) and identified the presence of a written prescription using EMR data. We excluded individuals without pharmacy cost-sharing benefit design information. The Boston University Medical Center Institutional Review Board deemed this research exempt.

We analyzed the first prescription written for buprenorphine or naltrexone after the initial OUD diagnosis (Wakeman et al., 2020). The outcome was filling a prescription within 30 days of writing. We censored individuals before 30 days if they received a subsequent MOUD prescription. The plan design exposure of interest was the pharmacy copay. We used the tier 2 pharmacy copay as we found examples of large plans categorizing generic MOUD as tier 2 (American Association of Retired Persons, 2019), and examined the effect of having a tier 2 copay above the mean copay, or at or below the mean copay of the sample. We included demographic and clinical characteristics as potential confounders. We included the calendar quarter of the prescription which may be associated with individuals reaching their out-of-pocket maximum later in the year, after which copayment is not required by most insurance. Demographic characteristics included age, sex, and region of residence, defined as metropolitan or non-metropolitan. Clinical characteristics included type of MOUD prescribed, the time from initial OUD diagnosis to the MOUD prescription, co-occurring depression and anxiety, and a modified Elixhauser score that excluded substance use (which is an inclusion criterion for the sample) and depression and anxiety, which are captured separately.

We present descriptive characteristics and used Chi-square tests to assess bivariate associations between individual characteristics and filling a prescription within 30 days. Next, we assessed the timing of the fill with a Kaplan-Meier curve and used a log-rank test to assess the difference between those with and without above average copays. For the survival analysis we examined the 30 days after the date the first prescription was written, censoring at 30 days, where the event was prescription fill and time to event was the number of days from writing to prescription fill.

Results

We identified 264 individuals with a written prescription for buprenorphine (146 individuals, 55%) or naltrexone (118 individuals, 45%) following a new OUD diagnosis. Overall, 184 (70% ) individuals filled the prescription within 30 days. Table 1 details the characteristics of the cohort. We found that having an above average copay was associated with a lower fill rate (p<0.05, Table 1). While our small sample size limited power to detect any statistically significant differences in demographic or clinical characteristics, we note that individuals with commercial insurance had lower fill rates (64%) than for Medicare Advantage <65 (76%) or Medicare Advantage ≥65 (74%). On average, the time between OUD diagnosis and first MOUD prescription was nominally shorter among those who filled the prescription within 30 days (153 days) compared to those who did not (224 days; p=0.06). Other demographic and clinical covariates were well balanced between the fill and no fill groups, and we did not detect an effect of calendar quarter. Most individuals who filled their prescription did so the day it was written (150 individuals, representing 82% of those who filled a prescription and 57% of the sample). We additionally detected a level difference in fill rates over time by copay amount (Figure 1; log-rank p=0.03). At day 30, 75% of those with a copay at or below the mean and 63% of those above the mean filled prescriptions.

Table 1:

Characteristics of individuals with a new OUD diagnosis receiving a written prescription for buprenorphine or naltrexone between 2015–2019, by prescription fill status at 30 days.

		Filled Rx in 30 days	Did not fill within 30 days		p-value*
	n	Row %	n	Row %
TOTAL	184	70%	80	30%
Cost-sharing (Tier 2 copay)
At or below mean copay	106	75%	35	25%	0.04
Above mean copay	78	63%	45	37%
Demographic covariates
Sex
Male	95	68%	44	32%	0.61
Female	89	71%	36	29%
Insurance and age
Commercial insurance	84	64%	47	36%	0.14
Medicare <65	68	76%	22	24%
Medicare 65+	32	74%	11	26%
Region
Metropolitan	163	<71%	>69	>30%	0.21
Non-metro	21	>66%	<11	<34%
Clinical covariates
Medication
Mucosal buprenorphine	100	68%	46	32%	0.65
Oral naltrexone	84	71%	34	29%
Time from OUD diagnosis
Time in days (SD)	153	(245)	224	(334)	0.06
Mental health
Depression	65	72%	25	28%	0.52
Anxiety	71	72%	27	28%	0.45
Modified Elixhauser
0	80	67%	39	33%	0.73
1	41	72%	16	28%
2+	63	72%	25	28%

Percentages are row percent and are the proportion of a given variable filled or not filled. Cell counts under 11 are suppressed

SD=standard deviation, used in place of % for continuous variables

^*Chi-square test for categorical variables and t-test for continuous variables

Figure 1: Comparing probability of prescription fill over time between those a pharmacy copay of less than or equal to $20 and those with a copay of more than $20 in a population with opioid use disorder between 2015–2019.

Curve is the inverse of the Kaplan Meier curve to depict probability of prescription fill where each point is 1-(Kaplan Meir estimate).

Discussion

In this novel analysis of 264 insured individuals with a new diagnosis of OUD who were prescribed buprenorphine or naltrexone, nearly 1 in 3 did not fill the prescription. This fill rate is in line with other chronic conditions (Viswanathan et al., 2012), but the stigma associated with OUD make this fill rate uniquely concerning. The OUD cascade of care is defined by barriers at every step of the process, including diagnosis, engagement in care, and access to MOUD, with less than a quarter of those who would benefit from MOUD being engaged in care and in a position to receive a prescription (J. R. Morgan, Schackman, Leff, Linas, & Walley, 2018; Williams et al., 2019). That 30% of prescriptions go unfilled after such a challenging cascade represents an urgent point for intervention. We find that patients most often fill on the same day the prescription was written, so reducing any barriers to same day fills may be an important part of increasing the fill rate, including pharmacy stock and pre-authorization challenges. The shorter time between OUD diagnosis and MOUD prescription among those who filled the prescription, while not statistically significant in this study, warrants further exploration. A longer gap between diagnosis and MOUD prescription may identify barriers to provider access, or seeking and receipt of other OUD services and treatments first such as detoxification, psychotherapy, or methadone.

While differences in the small sample should be interpreted cautiously, our Kaplan-Meier analysis suggests an approximately 10 percentage point difference in fill rates between those with an above average copay compared to those without. Previous study of the HIV care cascade has revealed that relatively small (compared to the price of healthcare interventions) increases in patient out of pocket costs could dramatically affect medication fills and retention, and interventions aimed specifically at copay support have been successful (Sood et al., 2014). While not statistically significant, our finding that those on Medicare (with an average copay of $10) nominally filled prescriptions more often than those with commercial insurance (average copay of $30) suggests that more research into medication subsidization and public insurance mechanisms would be fruitful. There is a pressing need to understand how costs to patients affect MOUD uptake and whether interventions targeting patient out-of-pocket costs would be effective. Other research has identified that many pharmacies do not stock buprenorphine (Hill et al., 2020), and delays from the pharmacy having to order it, or the patient finding an alternative pharmacy to fill, may decrease likelihood of filling. Notably, a recent study demonstrated that provision of buprenorphine at clinic visits rather than filling through pharmacies led to improved retention and outcomes (Khan, Khan, & Kolb, 2021). The scalability of that model is uncertain, but points to the importance of reducing barriers to accessing medication once prescribed.

The largest limitation of this work is the small sample size, which constrains our ability to conduct adjusted analyses and identity specific factors affecting fill rates with the necessary degree of certainty. However, data linking electronic medical records and MOUD outpatient pharmacy claims data are rare, and we were not able to identify any other published research examining medication fill rates in an OUD population. This study represents a first step and a call for more research motivated by our finding that under 70% of initial MOUD prescriptions were filled. Barriers to prescription fill are an important area of research given that there may be opportunity for relatively inexpensive and effective interventions to overcome barriers to filling prescriptions. Improving access to MOUD requires attention to every aspect of the cascade of care from diagnosis of OUD through provision of medication.

Highlights.

• Effective medications for opioid use disorder exist, but are underutilized

• Only 70% of new prescriptions for medications for opioid use disorder were filled within 30 days

• Future research should further investigate this stage of thet reatment cascade