Retention in Treatment after Emergency Department-Initiated Buprenorphine

Abstract

Background:

Emergency department-initiated buprenorphine (EDIB) has been shown to be effective in connecting patients with opioid use disorder (OUD) to outpatient treatment. Five diverse emergency departments (EDs) have successfully implemented EDIB programs.

Objectives:

1) To measure attendance at the first referral appointment and 30-day retention in treatment rates for patients receiving EDIB; 2) To describe demographic and opioid use characteristics of patients receiving EDIB; and 3) To determine average length of time in treatment after EDIB at the five participating EDs.

Methods:

All patients receiving EDIB at the participating EDs (n = 522) were seen by a peer recovery specialist in the ED and demographic and opioid use characteristics were recorded. Patients were followed prospectively. The referral site was contacted and information regarding attendance at the first referral appointment and 30-day retention in treatment was obtained. All patients still in treatment at 30 days were continually followed at subsequent 30-day intervals until the referral site indicated the patient had ended treatment at their facility.

Results:

The rate of attendance at the first referral appointment was 77.0% for patients receiving EDIB. At 30-day follow-up, 43.1% of patients were retained in treatment. The mean age of patient enrollment was 36.7 years, 58% of enrollees were male, 90.5% were white, and 73.4% had no medical insurance. Seventy-seven percent reported no substance use other than opioids. The mean time in treatment was 158 days.

Conclusions:

EDIB programs across diverse ED settings are effective at promoting attendance at the first referral appointment for OUD treatment; however, additional work is warranted to increase 30-day treatment retention rates, particularly among patients with nonprescription-only use profiles.

Introduction

In 2018, over 47,000 (nearly 70%) of the overdose deaths in the United States involved opioids (1). In South Carolina, drug overdose deaths involving opioids have continued to rise from 628 (age-adjusted rate of 13.1 per 100,000) in 2016 to 835 (rate of 17.1) in 2018 (2). Recent data have also suggested an increase in opioid overdoses during the COVID-19 pandemic. Kentucky Emergency Medical Services (EMS) calls for opioid overdoses increased 17% between January 14 and April 26, 2020, with a 50% increase in deaths at the scene related to suspected opioid overdoses (3). Likewise, South Carolina EMS responses for suspected overdoses increased by almost 50% between January and May of 2020, compared with the previous year (4). Of concern, approximately 10% of individuals resuscitated with naloxone by EMS die within the subsequent year (5).

Efforts to identify individuals with opioid use disorders (OUD) and engage them in treatment is critical to accelerate declines in opioid-related deaths. As the opioid epidemic has worsened in recent years, emergency departments (EDs) have moved to the forefront of providing care to affected individuals. EDs are uniquely situated to identify patients at risk for, or with, an OUD via complications of opioid use, including infection, trauma, opioid withdrawal, or overdose. For example, in 2017, 305,623 nonfatal opioid overdoses were treated in EDs in the United States, a 3.1% increase from the previous year (6). The rate of opioid-related ED visits per capita has increased 2.5 times between 2010 and 2018 (7).

In recognition of the potential role of the ED improving access to treatment during the opioid crisis, D’Onofrio et al. examined the effect of buprenorphine initiation for opioid-dependent patients in an ED setting on retention in treatment (8). Results from this group indicate that almost 80% of patients initiated on buprenorphine in the ED with facilitated referral for follow-up were still engaged in treatment at 30 days, which was approximately twice as many when compared with the simple referral to treatment (p < 0.001) (8).

Through a legislative proviso, funds were granted to the South Carolina (SC) Department of Health and Human Services in consultation with the SC Department of Alcohol and Other Drugs of Abuse Services and the Medical University of South Carolina, to expand access to care through ED-initiated buprenorphine (EDIB) in conjunction with peer recovery support. Peer recovery support services are delivered by Peer Recovery Specialists (PRS), individuals in recovery from substance use disorder(s) that are well suited to support others currently suffering substance use-related problems and disorders. Peer recovery support services across a variety of settings have been found to reduce substance use, decrease relapse rates, and improve treatment retention (9).

Implementation of the ED-initiated buprenorphine programs followed Chamberlain’s Stages of Implementation Completion (10). Preimplementation activities included engagement, consideration of feasibility, and readiness planning. Sites were selected utilizing community needs assessments along with state-level data on prevalence of OUD, opioid overdose rates, and treatment availability (11). Feasibility was assessed through collaboration with hospital and ED leadership at each of the participating hospitals. Readiness planning included a review of funding, development of a plan for hiring and training staff, creation of a recruitment plan, and finalization of an individualized implementation plan at each site. Each site received funding from South Carolina’s Department of Health and Human Services and had an ED workflow analysis and assessment of current electronic health record (EHR) substance use screening workflow. The National Institute on Drug Abuse (NIDA) Quick Screen, an evidence-based substance use screening tool, was added into nursing triage. Additionally, EHR documentation was created for patient navigators, a Clinical Opioid Withdrawal Scale (COWS) calculator was integrated into the EHR for nurses or physicians, depending on the site, and order sets were created for buprenorphine and naloxone kit ordering (12). Additionally, providers capable of providing next-day medication for opioid use disorder (MOUD) treatment were identified near each ED site and the needs of these referral sites to provide next-day treatment were discussed. These “fast-track” sites received grant funding from the state to support the potential increase of uninsured ED patients seeking care, and the following information was requested to be sent to fast-track sites from the ED: the ED provider note, dose and time of buprenorphine administration, if the patient was provided with a prescription of buprenorphine, urine drug screen results, pregnancy test (if applicable), and demographic information. Additionally, a program champion was identified at each site.

Implementation activities included training of staff, adherence monitoring processes, services provided, and continual feedback with quality improvement. Patient navigators were hired at all sites to assist with referrals for patients with substance use disorders in the ED setting. They received 40 h of live education regarding the role of the patient navigator; ethical responsibilities and boundaries; motivational interviewing; screening, brief intervention, and referral to treatment (SBIRT) process; stages of change; and MOUD. In 2019, we switched from a patient navigator model to a peer recovery specialist model at all sites. Training also occurred for nurses at program sites, including an overview of SBIRT, buprenorphine pharmacology and dosing, calculation of the COWS and its importance in buprenorphine initiation, and the process of naloxone kit dispensing and education. Physicians and advanced practice providers were also trained through in-person training by the site champion with regards to opioid use disorder, MOUD, SBIRT, regulations surrounding buprenorphine, buprenorphine pharmacology, administration, and prescription writing. After starting the ED-initiated buprenorphine program, all sites receive monthly feedback regarding number of patients screened, number of patients referred to treatment, number of naloxone kits dispensed, and number of patients on initiated buprenorphine. Additionally, the lead team obtains retention in treatment outcomes from the referral center at the first referral (typically next day) appointments and 30 days after the initial ED visit (13).

The workflow at sites is as follows: patients who screen positive for substance misuse at triage via the NIDA Quick Screen are referred to an onsite peer recovery specialist who conducts an evidence-based SBIRT (14). SBIRT aims to identify and intervene with individuals who are at moderate to high risk of health or psychosocial problems related to substance use, followed by a short awareness-raising conversation and a referral to patients who would benefit from substance use disorder treatment. Individuals identified with an OUD who are interested in treatment are started on buprenorphine in the ED with a facilitated close follow-up appointment (all within 72 h, in most cases < 24 h) with a community partner, typically a state-licensed and nationally accredited county alcohol and drug abuse authority, for ongoing medication management and other substance use disorder treatment as needed. The formulation of buprenorphine used varied by site, but was either the sublingual tablet or film. The buprenorphine monoproduct or the combination product (buprenorphine/naloxone) was used. Funds administered by the SC Department of Alcohol and Other Drug Abuse Services via the Federal State Opioid Response and State Targeted Response to the Opioid Crisis grants are available at county authorities to provide financial assistance for ongoing treatment, including medication for the uninsured.

Initially piloted at an academic medical center, the program has expanded to four additional EDs targeting counties with the highest overdose rates in the state. Here we aim to:

Measure attendance at the first referral appointment and 30-day retention in treatment rates for patients receiving EDIB.

Describe demographic and opioid use characteristics for patients receiving EDIB.

Determine average length of time in treatment for EDIB.

Methods

The sample included a cohort of patients that were started on buprenorphine in the EDs of five hospitals in South Carolina. The study sites include two academic medical centers, two nonacademic community hospitals, and a private teaching hospital. EDIB was initially offered at one academic medical center on December 11, 2017, with other sites beginning EDIB on the first of the month in March 2018, April 2019, and August 2019. All patients receiving EDIB prior to June 30, 2020 were included for analysis, and retention in treatment was followed for all patients through August 31, 2020. After EDIB, patients at all sites were referred to outpatient referral treatment facilities, also called “fast-track” sites, within their local community, who agreed to provide close follow-up care. The fast-track sites included office-based treatment programs, opioid treatment programs, and state-funded, nonprofit drug and alcohol treatment agencies.

The peer recovery specialist (PRS) in the ED collected data on demographics and self-reported substance use at the time of the ED visit (Figure 1). Sex and race data were not initially collected at two hospitals, resulting in missing data for these variables for 144 and 145 patients, respectively. An individual was classified as a polysubstance user if they self-reported misusing at least one other substance in addition to opioids. After buprenorphine initiation, the PRS directly linked the patient with a referral by arranging for follow-up within 24 h at the fast-track site for further medication dosing. Due to limited operating hours of fast-track sites over weekends, holidays, and severe weather events, occasionally, patients were dosed with buprenorphine on sequential days or were given a prescription to bridge for those few days. Such cases were treated as a single buprenorphine initiation. All other patients receiving additional doses of buprenorphine were considered to be a new buprenorphine initiation. Depending on the fast-track site, EDIB patients were provided the opportunity at their referral appointment to continue their treatment with buprenorphine or switch to a different medication such as methadone or naltrexone. All patients started on buprenorphine between December 2017 and June 2020 (n = 522), regardless of whether they continued on buprenorphine at their referral appointment, were included in analyses.

Figure 1.

Overview of ED-initiated buprenorphine workflow at all sites.

OUD = opioid use disorder; NIDA = National Institute on Drug Abuse; ED = emergency department.

Primary outcomes were collected by contacting fast-track sites monthly and included: 1) arrival to the first referral appointment (Yes/No); 2) retention in formal addiction treatment at 30 days (Yes/No); and 3) length of time in formal addiction treatment. Sites reported whether patients presented to their first appointment, the date of their most recent visit, and if they had disengaged from treatment regardless of reason. Patients were considered “in treatment at 30 days” if their most recent visit was at least 30 days after their date of EDIB. Patients were considered to have left treatment if the fast-track sites reported they had disengaged with treatment, characterized by reporting treatment termination or not conducting a visit in the past 30 days. We calculated the proportion of patients that arrived at their first referral appointment and that remained in treatment at 30 days.

We compared variables among individuals that attended and those that did not attend their first referral appointment, and those that were and were not in treatment at 30 days after the index ED visit. We assessed categorical variables by chi-squared tests and continuous variables by t-tests. Multiple regression was utilized to predict the probability that a patient arrives to their first referral appointment and remains in treatment at 30 days using the previously mentioned variables as predictors.

Time in treatment was evaluated using Kaplan–Meier survival curves from the time of the ED visit to the time of reported disengagement at the fast-track sites. For patients that did not disengage from treatment by the end of data collection (August 21, 2020), the time in treatment was censored, meaning the total time in treatment for that subject could not be accurately determined. Censoring occurred when the individual was lost to follow-up or was engaged in treatment at the end of data collection. Log-rank tests were used to compare the probability of disengaging in treatment between groups. A Bonferroni correction was used to adjust for multiple comparisons. Data analysis was performed using SPSS 25.0 and 27.0 (IBM, Armonk, NY) (15).

Results

Demographic characteristics of all EDIB patients are shown in Table 1. Of the individuals where sex and race were recorded (n = 378 and 377, respectively), 58.2% were male and 90.5% were white. The mean age was 36.7 ± 10.5. The 30–39-year age class was the largest (39.5%). Of all patients, 61.3% self-reported using solely nonprescription opioids and 77.4% reported no other substance use. Most patients received EDIB at the private teaching hospital (37.5%) followed by academic medical centers (32.4%) and community medical centers (30.1%). A majority of patients were uninsured (73.4%). There were 30 patients that were dosed multiple times outside of those who were expected to return to the ED for additional doses over weekends and holidays. Of these 30, 28 were dosed two times, 2 were dosed three times (range between doses: 25–666 days, average number of days between doses: 193 ±150), and these cases were considered to be a new buprenorphine initiation, as all of these patients were no longer engaged in outpatient treatment.

		Arrived to 1st Referral Appointment				In Treatment at 30 Days
		Overall	True	False		True	False
Demographics	N = 522	n = 402 (77.0%)	n = 120 (23.0%)	p-Value	n = 225 (43.1%)	n = 297 (56.9%)	p-Value
Age* mean (SD)36.70 (10.49)		36.74 (10.60)	36.60 (10.16)	0.90	36.93 (10.68)	36.53 (10.36)	0.67
Sex,† n (%)
	Male	220 (58.2)	166 (57.4)	54 (60.7)	0.59	96 (59.3)	124 (57.4)	0.72
	Female	158 (41.8)	123 (42.6)	35 (39.3)		66 (40.7)	92 (42.6)
Race,† n (%)
	White	341 (90.5)	260 (90.6)	81 (90.0)	0.87	147 (91.3)	194 (89.8)	0.63
	Non-white	36 (9.5)	27 (9.4)	9 (10.0)		14 (8.7)	22 (10.2)
Opioid use, n (%)
	Prescription	124 (23.8)	97 (24.1)	27 (22.5)	0.75	63 (28.0)	61 (20.5)	< 0.01‡
	Nonprescription	320 (61.3)	243 (60 .4)	77 (64.2)		116 (51.6)	204 (68.7)
	Prescription & nonprescription	78 (14.9)	62 (15.4)	16 (13.3)		46 (20.4)	32 (10.8)
Polysubstance use,§ n (%)
	Yes	118 (22.6)	82 (20.4)	36 (30.0)	0.03‡	50 (22.2)	68 (22.9)	0.86
	No	404 (77.4)	320 (79.6)	84 (70.0)		175 (77.8)	229 (77.1)
Hospital type, n (%)
	Academic medical center	169 (32.4)	119 (29.6)	50 (41.7)		82 (36.4)	87 (29.3)
	Community medical center	157 (30.1)	119 (29.6)	38 (31.7)		59 (26.2)	98 (33.0)
	Private teaching hospital	196 (37.5)	164 (40.8)	32 (26.7)		84 (37.3)	112 (37.7)
Insurance type (%)
	Private	37 (7.1)	30 (7.5)	7 (5.8)	0.79	14 (6.2)	23 (7.7)	0.18
	Public	102 (19.5)	77 (19.2)	25 (20.8)		52 (23.1)	50 (16.8)
	Uninsured	383 (73.4)	295 (73..4)	88 (73.3)		159 (70.7)	224 (75.4)

^*Date of birth not recorded for 4 individuals.

^†Sex (n = 378) and Race (n = 377) data were not initially collected at two hospitals resulting in missing data.

^‡Chi-squared test shows significance at the p = 0.05 level.

^§Polysubstance use was classified as “yes” if the individuals self-reported misusing at least one substance other than opioids.

Next-Day and 30-Day Treatment Retention

Overall, 77.0% of patients arrived at their first referral appointment at fast-track sites and 43.1% remained in treatment at 30 days (Table 1). Patients who reported polysubstance use were significantly less likely to attend their first referral appointment (χ² = 4.87, p = 0.03), but were equally likely to be engaged or disengaged in treatment at 30 days (χ² = 0.33, p = 0.86). First referral appointment arrival was similar regardless of opioid type used (χ² = 0.58, p = 0.75), however, at 30 days there was a difference in treatment retention depending on opioid use type (χ² = 17.1, p < 0.01). At 30 days post EDIB, patients who reported using both nonprescription and prescription opioids had the highest retention rate (59.5%), followed by those reporting solely prescription opioid use (50.8%), and solely nonprescription use (36.3%).

Multiple logistic regression analysis indicated a significant difference in treatment retention at 30 days post induction between opioid type adjusted for the other predictor variables (Table 2). Individuals reporting solely prescription opioid use were 2 times more likely to remain in treatment at 30 days, compared with those reporting solely nonprescription use (odds ratio 2.09, 95% confidence interval 1.24–3.54) and those reporting use of both prescription and nonprescription opioids were three times more likely to remain in treatment at 30 days than those reporting solely nonprescription use (odds ratio 3.26, 95% confidence interval 1.67–6.36).

Table 2.

Multiple Logistic Regression Analysis of First Referral Appointment Arrival and 30-Day Treatment Retention of EDIB Individuals

		Arrived to 1st Referral Appointment		In Treatment at 30 Days
Variable		Odds Ratio	95% CI	Odds Ratio	95% CI
Age		1.01	0.98–1.04	0.99	0.97–1.01
Gender
	Female	1.13	0.67–1.90	0.90	0.57–1.40
Race
	Non-white	1.03	0.45–2.35	0.81	0.38–1.70
Polysubstance use
	No	1.65	0.92–2.99	1.52	0.86–2.69
Insurance type
	Private	1.57	0.60–4.06	0.98	0.45–2.12
	Public	1.23	0.62–2.44	1.39	0.78–2.47
Opioid
	Prescription &	1.30	0.60–2.82	3.26	1.67–6.36
	nonprescription Prescription	1.22	0.65–2.30	2.09	1.24–3.54

EDIB = emergency department-initiated buprenorphine; CI = confidence interval.

Length of Time in Treatment

For all patients receiving EDIB, the estimated mean time until treatment dropout was 168 days (± 14 days), with half of patients leaving treatment by 15 days (Figure 2). Duration of treatment was shorter for patients taking only prescription opioids, compared with patients also using nonprescription opioids (mean 137 days vs. 236 days, p < 0.01; Figure 3). There was no difference in treatment length between sexes, races, age class, poly-substance vs. solely opioid users, or insurance type.

Figure 2.

Survival function showing time to treatment drop-out of all patients receiving emergency department-initiated buprenorphine.

Figure 3.

Survival function showing time to treatment drop-out dependent on opioid type use of all emergency department-inducted buprenorphine individual.

Discussion

Implementation

EDIB was successfully implemented at all sites, with a range of 157–196 patients receiving EDIB at each site over the study period. Attendance rate at the first referral appointment ranged between 70.4% and 83.7%. To date, the majority of studies and reports on EDIB have been from academic medical centers, and these findings are encouraging that EDIB is effective in other ED settings as well (8,16).

30-Day Retention in Treatment

The average rate of first referral appointment attendance was 77.0% for patients receiving EDIB at all sites, and 43.1% at 30 days. The 30-day retention in treatment rate from our cohort was lower (43.1%) than the initial study of EDIB (78%) (17). This may be due to differences in EDIB protocols for this project and the protocol utilized for the initial studies. Our model relies heavily on next-day follow-up for patients receiving EDIB, whereas D’Onofrio et al. ensured outpatient appointments within 72 h of the ED visit, and study patients were provided with a prescription of buprenorphine until the follow-up appointment (17). In this analysis, only 10 patients (1.9%) received a prescription for buprenorphine. The inability of physicians to write prescriptions for buprenorphine without additional training is a significant barrier to accessing treatment for OUD. Under the Drug Addiction Treatment Act of 2000, practitioners must obtain a waiver to treat opioid dependency with buprenorphine, which requires an 8-h training for physicians and 24 h of training for advanced practice providers (17,18). This requirement may currently be under revision, but was in place throughout the study period (19). Next-day follow-up was arranged as a way to overcome this barrier; however, it does limit access to care when the following day is a weekend or holiday. The inability to prescribe buprenorphine also limits access to patients who are actively in withdrawal while they are in the ED. A majority (57%) of the patients in D’Onofrio et al. received a prescription and instructions for an unobserved induction due to inability to initiate in the ED because withdrawal was not severe enough upon ED presentation (X). Due to the limited number of waivered emergency physicians, only a small minority of the patients included in this cohort (8 patients) received a prescription of buprenorphine for a home induction due to the COWS being < 8 during the ED visit (20). It is also possible that a one-time buprenorphine dose in the ED does not adequately treat a patient’s withdrawal through the next day, resulting in decreased rates of first referral appointment attendance and subsequently lower 30-day retention in treatment rates. Future studies evaluating treatment retention for patients receiving a one-time ED dosing compared with ED dosing plus a prescription may be helpful to evaluate the efficacy of these different approaches to EDIB. Of note, our first referral appointment attendance rate was similar to the 30-day retention in treatment rate reported by D’Onofrio et al. (8). There also may be differences at the fast-track treatment sites that affect longer-term retention.

As mentioned above, there was a significant drop in patient retention between attendance at the first referral appointment and 30 days (77.0% and 43.1% for the entire cohort, respectively). Hser et al. estimated buprenorphine retention rates at 30 days to be approximately 75%, with retention rates flattening out to around 50% at 140 days (21). The cause of this difference is unclear, but may be related to differences in the intervention and participating patients. Hser et al. had all patients return to an opioid treatment program for daily dosing of buprenorphine (21). Additionally, participants were recruited from patients already entering an opioid treatment program, which is likely a different patient population than those presenting to an ED.

Demographic Factors

There were no significant differences in post-ED retention in treatment based on any of the demographic factors: age, gender, and race. However, a notable difference in initial enrollment (patients receiving EDIB) was noted between white (90.5%) and non-white (9.5%) patients. Although the difference in enrollment is striking, this difference is consistent with documented racial disparities in MOUD access among other populations, and this difference across sites is being explored (22,23).

Opioid Use Characteristics

We hypothesized that prescription opioid users would be more likely to stay in treatment compared with patients using heroin or fentanyl. We found that patients who used prescription opioids were two times more likely to stay in treatment than the nonprescription-only group. We also hypothesized that patients with polysubstance use would be less likely to be retained in treatment compared with single-substance (opioid-only) use; however, this effect was not observed.

Limitations

Our outcome data are reflective of EDIB programs in one state, with unique barriers (rurality, a high number of uninsured patients) and facilitators (direct funding from the SC Department of Health and Human Services for program implementation and management) to creating EDIB programs. Additionally, our model may differ from other EDIB programs with regards to hours of availability and prescription-writing abilities (as previously highlighted), which may affect outcomes.

Additionally, we only contacted the site to which the patient was referred for treatment. No further contact was attempted with patients after ED discharge. It is possible that patients switched providers after their initial ED visit, meaning they would still be in treatment, just at a different treatment site, but would have been counted as “not in treatment” for this study. Also, because patients were not contacted after the initial ED visit, we do not have any way of assessing patient-centered outcomes such as quality of life, ability to work, or overall health.

Additionally, our study compares outcomes only among patients who received EDIB. We do not have a comparison group of patients who did not receive EDIB, as withholding this treatment would be unethical given its efficacy. For the retention in treatment analysis, patients whose data were censored may be systematically different than the rest of the group. For example, they would be more likely to have later start dates, so the system caring for them has become more experienced with handling their care. We did not have a means of adjusting for this potential difference.

Conclusion

EDIB was successfully implemented at all five sites, including state-funded academic centers, private hospital systems, and nonteaching community EDs. Attendance at the first referral appointment and 30-day retention in treatment rates were, on average, 76.7% and 43.1%, respectively. Further studies examining outcomes across different models of EDIB (next-day follow-up, prescription writing, dosing regimens) may be beneficial to determine the most effective EDIB strategy and improve 30-day retention in treatment rates.

ARTICLE SUMMARY.

1. Why is this topic important?

Emergency Department (ED) initiated buprenorphine has the potential to reduce mortality for patents with opioid use disorder (OUD) by connecting them to treatment. Retention in treatment of patients after ED-initiated buprenorphine across various types of EDs has not been well described and this information is important for future dissemination of this intervention.

2. What does this study attempt to show?

This study describes the short and long-term retention in treatment outcomes for patients receiving ED-initiated buprenorphine as part of routine clinical care (i.e. not part of a research study with randomization, participant reimbursement, etc) at five EDs including community, regional, and tertiary/academic lefts in both rural and urban areas.

3. What are the key findings?

ED-initiated buprenorphine was successfully implemented and performed at all five EDs, and retention in treatment at 30 days did not vary by ED setting. Attendance at the first referral appointment was 77.0%, and the 30 day retention in treatment rate was 43.1% averaged across all sites.

4. How is patient care impacted?

These results suggest that ED-initiated buprenorphine can be incorporated as part of routine clinical care across a range of ED settings, can impact a significant number of patients, and is associated with high short-term retention in treatment rates. These findings may encourage EDs to adopt ED-initiated buprenorphine as part of routine clinical practice.