Use of in-network insurance benefits is critical for improving retention in telehealth-based buprenorphine treatment

Abstract

An empiric evidence base is lacking regarding the relationship between insurance status, payment source, and outcomes among patients with opioid use disorder (OUD) on telehealth platforms. Such information gaps may lead to unintended impacts of policy changes. Following the phase-out of the COVID-19 Public Health Emergency, states were allowed to redetermine Medicaid eligibility and disenroll individuals. Yet, financial barriers remain a common and significant hurdle for patients with OUD and are associated with worse outcomes. We studied 3842 patients entering care in 2022 at Ophelia Health, one of the nation's largest OUD telehealth companies, to assess associations between insurance status and 6-month retention. In multivariable analyses, in-network patients who could use insurance benefits were more likely to be retained compared with cash-pay patients (adjusted risk ratio [aRR]: 1.50; 95% CI: 1.40–1.62; P < .001). Among a subsample of 882 patients for whom more detailed insurance data were available (due to phased-in electronic health record updates), in-network patients were also more likely to be retained at 6 months compared with insured, yet out-of-network patients (aRR: 1.86; 95% CI: 1.54–2.23; P < .001). Findings show that insurance status, and specifically the use of in-network benefits, is associated with superior retention and suggest that Medicaid disenrollment and insurance plan hesitation to engage with telehealth providers may undermine the nation's response to the opioid crisis.

Introduction

More than 100 000 annual overdose deaths are now attributed to the opioid and drug addiction crisis in the United States, the great majority among untreated individuals with opioid use disorder (OUD).^1,2 For 20 years, efforts to scale access to affordable, evidence-based care have struggled to meet rising demand.^3,4 Several significant regulatory changes under the COVID-19 Public Health Emergency (PHE) contributed to increasing access to medication for OUD (MOUD), which is the gold standard of care for OUD due to overdose protection and mortality benefits.⁵ Most notably, allowances for prescribing controlled substances without first having an in-person visit,^6,7 as well as a moratorium on Medicaid redeterminations of status eligibility and plan disenrollment, allowed for the rapid expansion of affordable telehealth services for MOUD. As the prevalence of OUD among Medicaid beneficiaries is approximately 4-fold that among the general population, Medicaid is the number-1 payer of addiction care in the United States.^8-10 Following the passage of the Affordable Care Act in 2014, this is especially true in Medicaid-expansion states.¹¹ Therefore, the pause on Medicaid disenrollments under the COVID-19 PHE likely allowed for uninterrupted coverage to disproportionately marginalized patient populations in most need of treatment amid the opioid crisis.

However, following the end of the PHE, beginning March 2023, states were allowed to once again determine Medicaid eligibility and disenroll individuals, which could result in an estimated 8 to 24 million individuals losing coverage.¹² Financial barriers have been widely shown to undermine clinical outcomes across a variety of health conditions.^13-15 While studies have shown that higher out-of-pocket costs will suppress medication use, including for OUD,^16,17 an empiric evidence base is lacking regarding the relationship between insurance status, payment source, and outcomes among patients entering care for OUD on telehealth platforms. Such concerns are especially timely as new regulations for mental health parity enforcement to ensure adequate functional network access for substance use disorder treatment services were proposed in August 2023.

Ophelia Health, Inc, founded in 2020, is one of the nation's largest telehealth-based opioid treatment (TBOT) platforms. As of July 2023, Ophelia had treated over 10 000 patients across 14+ states, with two-thirds of active in-network patients (ie, those who could use their insurance benefits to pay for care) covered by Medicaid (vs commercial or Medicare plans). Ophelia's relative nascency and the fact that it can take several years for new providers to finalize contracts with insurance companies resulted in a broad patient case mix of in-network (ie, could pay for care with insurance benefits), out-of-network (ie, while insured, could not use benefits and had to pay cash), and uninsured (ie, a priori cash-pay) patients. We were therefore able to evaluate associations between both overall payment source (in-network benefits vs cash-pay), as well as more detailed insurance status and payment source (in-network vs out-of-network [cash-pay] vs uninsured [cash-pay]), and clinical outcomes. Given that out-of-pocket costs will generally be lowest for in-network patients, followed by out-of-network (cash-pay), then uninsured (cash-pay) patients, these comparisons also shed light on the relationship between patient costs and retention in a telehealth setting. In the present study, we analyzed a cohort of individuals entering care in 2022 to assess 6-month buprenorphine retention rates, a quality measure adopted by the US Centers for Medicare and Medicaid Services (CMS) as a marker for care quality at the population level¹⁸ by insurance status and payment source. We hypothesized that patients who could pay for care with insurance benefits would have superior retention and care outcomes compared with those required to pay out of pocket.

Data and methods

Overview

The aim of the study was to assess differences in 180-day retention in telehealth-based treatment with buprenorphine for OUD by overall payment source (in-network benefits vs cash-pay) and, among a subsample of patients for whom more detailed insurance data were available after May 26, 2022, by detailed insurance status and payment source (in-network [could use benefits] vs out-of-network [cash-pay] vs uninsured [cash-pay]). Comparing retention outcomes between in- and out-of-network patients permits evaluating the association between retention and payment source among patients with insurance, thus controlling for systematic differences between insured and uninsured patients.

Study setting and patient sample

We analyzed data from a cohort of individuals with OUD treated at Ophelia,^19-21 a virtual-first TBOT platform operating in 14 states (although the patient sample used for this analysis was mostly located in 2 states—New York and Pennsylvania—as they were the first 2 in operation and thus the most represented in the sample). The care model, medical visits and protocol, organizational structure, custom electronic health records (EHRs), and care coordination services were all designed explicitly for remote care without requiring any in-person visits.

Patients were typically prescribed buprenorphine at intake, seen weekly during the stabilization phase, then stepped down to biweekly and then monthly visits under a nurse care manager model adapted to a TBOT setting via real-time face-to-face video visits. To be eligible for care, patients had to be 18 years or older and meet OUD diagnostic criteria. Patients were considered ineligible (ie, to initiate treatment at Ophelia) if they required a higher level of care, such as for unstable psychiatric conditions (eg, active suicidality or psychosis), physical dependency on high doses of sedative hypnotics (eg, lorazepam ≥8 mg/d, clonazepam ≥4 mg, alprazolam ≥4 mg, diazepam ≥60 mg), or severe alcohol use requiring medical detoxification, or were currently enrolled in a methadone maintenance program with a dose of more than 80 mg per day.

Patients were reached directly through online advertisements (eg, Google and Facebook ads) or word of mouth. In-network patients, predominantly Medicaid beneficiaries, used insurance benefits to pay for care. Patients who were insured but out-of-network or uninsured paid a flat $195 monthly fee. Prescription costs associated with filling buprenorphine prescriptions at local pharmacies were in addition to service costs and not observable for our analysis.

The study sample included all patients who completed an intake visit between January 1, 2022, and September 8, 2022 (the last date on which patients would have at least 180 days of follow-up at the time of analysis), and received their first Ophelia buprenorphine prescription within 7 days of their intake (in order to exclude patients who may have declined to initiate treatment). If patients had multiple intakes meeting these criteria during the study period, only the first was included. The small minority of patients who transitioned between payment types (eg, those who were initially cash-pay but whose plan then became in-network due to contracting activity; see below for details) while in care were excluded from the analysis.

Although the parent sample for analysis categorized patients as either (1) using in-network benefits or (2) cash-pay, for a subset of patients in the sample with an intake visit on or after May 26, 2022—when EHR updates were implemented to collect more detailed self-reported insurance status—we could further categorize cash-pay patients as insured but out-of-network or uninsured.

The study was reviewed and granted a waiver of informed consent by the WCG Institutional Review Board and reported in accordance with STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines for cohort studies.

Discharge policies

The majority of treatment discontinuations involved patients self-determining their end of care (ie, they notified Ophelia that they would not be continuing care or were lost to follow-up without notice), but patients could also be discharged and referred elsewhere for clinical or financial reasons. Clinical discharge, although rare, occurred when a patient misused or was nonadherent to prescribed medications, was heavily dependent on non-opioid substances threatening safety, or had complex clinical needs better suited for a higher level of care, such as residential or inpatient treatment settings. Financial discharges may have occurred when a patient was unable to pay for their care for more than 2 months in a row. Typically, if a patient does not pay an invoice by the due date, they will be contacted to discuss payment options. If a patient has still not paid 30 days after the original invoice due date, they will be notified that they have 30 days to pay their invoice before being referred out. During this time, clinicians may increase visit frequency to work with the patient on payment issues, connect patients with adjunctive support, or work with the Ophelia care coordination team to establish local referrals. If a patient has still not paid 60 days after the original invoice due date, they were transferred to an outside provider. A patient could also be discharged if they were verbally aggressive, abusive, or violently threatened Ophelia staff, although this was rare.

Data collection and measures

Payment source (in-network benefits vs cash-pay) for each patient visit was documented as part of routine billing processes and patients were categorized as either in-network or cash-pay corresponding to the payment source of their visits. Basic demographic information (date of birth, sex, gender, race/ethnicity, address) was collected via self-report in an online form prior to intake visits. Race and ethnicity categories were modeled after those used by the US Census Bureau and grouped as follows: Hispanic/Latino, non-Hispanic African American, non-Hispanic White, and, due to small sample sizes of additional groups (non-Hispanic Asian, non-Hispanic Native American, non-Hispanic Pacific Islander, non-Hispanic multiple races), non-Hispanic other or multiple races. Patient urbanicity (ie, urban or rural) was defined using US Department of Agriculture rural-urban commuting area (RUCA) codes associated with the patient's zip code, with codes of 1–3 denoting urban and codes of 4–10 denoting rural locations.²² Patient buprenorphine status at intake (ie, negative for patients newly initiating buprenorphine and positive for patients already using buprenorphine, whether by prescription or another source) was documented as part of routine clinical care.

Beginning on May 26, 2022, a new process was launched for prospective patients seeking care at Ophelia. Specifically, patients who clicked “Am I a candidate?” on the Ophelia website would provide basic contact information, information about their opioid use, and whether they have insurance, and would schedule a welcome call to learn more about the Ophelia platform and to schedule an intake visit if appropriate. This new process was gradually rolled out to an increasing and random allocation of web traffic until being launched for all prospective patients on July 21, 2022. Among the subsample of patients who completed this new process, insurance status and payment source could be further defined as in-network, out-of-network (thus, cash-pay), and uninsured (cash-pay) (see Figure 1). Specifically, patients who paid for their care with insurance were considered in-network, patients who self-reported having insurance but paid for their care with cash were considered out-of-network, and patients who self-reported not having insurance and paid for their care with cash were considered uninsured. It was possible for an “out-of-network” patient to have been in-network but to have paid cash for other reasons (eg, anonymity, patient co-pays that would exceed the cash-pay rate of $195), but we estimated this to be extremely rare based on clinician and staff feedback. Unlike overall payment source (ie, in-network benefits vs cash-pay), which could be tracked longitudinally for all patients, the delineation between uninsured and out-of-network cash-pay patients was only available at intake.

Figure 1.

Detailed insurance status and payment source groups. Source: Patient data from Ophelia Health, Inc. *Only includes a subset of 882 patients for whom more detailed insurance data were available after May 26, 2022 (see text for details).

The primary outcome of interest was 6-month (ie, 180-day) retention, a continuity of pharmacotherapy for OUD quality measure developed with the National Committee on Quality Assurance, endorsed by the National Quality Forum, and incorporated into measure sets by CMS.¹⁸ Treatment retention was defined using buprenorphine prescription data, with discontinuation defined as a more than 30-day gap in medication coverage, consistent with previous literature.^23-25 Retention was defined as continuous use of medication without a gap in days’ supply exceeding 30 days (per structured prescription data in the EHR): patients were considered retained from the date of their first Ophelia prescription through their last date covered by medication before a gap of more than 30 days. This prescription-based definition was applied consistently for all patients, regardless of why the patient discontinued treatment (eg, loss to follow-up, transferred to another provider, death). See the Appendix Methods for additional details regarding the prescription data.

As a marker of care quality, buprenorphine adherence was also assessed using the proportion of days covered (PDC) as prior studies have shown a buprenorphine PDC greater than 0.8 is associated with superior outcomes. The PDC was calculated for each patient as the number of unique days covered over the number of days in treatment from a patient's first prescription date through their discontinuation date or their 180th day in care, whichever was sooner. Specifically, patients were considered covered from the date of each prescription onwards according to the indicated days’ supply prescribed and inclusive of the prescription date. If multiple prescriptions overlapped (ie, if a new prescription was written prior to the end of the days’ supply of a previous prescription), each calendar date covered by multiple prescriptions would only be counted as a single covered day (ie, days covered by multiple prescriptions would not be double-counted).

Statistical analysis

We examined associations between baseline demographic and clinical characteristics and overall payment source (in-network benefits vs cash-pay) using chi-square and Fisher's exact tests. We examined associations between overall payment source and 6-month retention using a Poisson regression model with robust standard errors,²⁶ controlling for patient age, race/ethnicity, gender, urbanicity, buprenorphine status at intake, whether this was the patient's first care episode with Ophelia (the analysis only included patients’ first care episode during the study period, but patients could have had earlier care episodes that started prior to the study period), and US state. Multivariable associations with 6-month retention are presented as risk ratios (RRs) with 95% confidence intervals (CIs). Patients missing values for any covariate included in the multivariable model were excluded from the multivariable analysis.

Mean PDC was calculated among in-network and cash-pay patients and compared using a t test. Among the subsample of patients who completed the new prospective patient process and could thus be categorized as in-network, out-of-network (cash-pay), or uninsured (cash-pay), we fit a model similar to the multivariable Poisson model described above, but which included this 3-group categorization of insurance status and payment source instead of the 2-group categorization (in-network benefits vs cash-pay). The variable capturing whether this was the patient's first care episode was excluded from this model because this subsample included very few subsequent care episodes. In addition, mean PDC was calculated separately among in-network, out-of-network (cash-pay), or uninsured (cash-pay) patients and compared using 1-way analysis of variance.

For visualizing retention among patients by insurance status and payment source, we generated separate Kaplan-Meier curves showing time to discontinuation for the 2-group (in-network benefits vs cash-pay) and 3-group (in-network, out-of-network [cash-pay], or uninsured [cash-pay]) categorizations. To assess whether patients included in this subsample (ie, those who completed the new prospective patient process and could thus be categorized as in-network, out-of-network [cash-pay], or uninsured [cash-pay]) were representative of all patients starting treatment during the same time frame, we assessed differences in baseline patient characteristics between patients included and not included in this subsample among those with an intake on or after May 26, 2022 (the date that the new prospective patient process collecting self-reported insurance status was launched).

To better understand the association between insurance status and payment source among incident users, we included sensitivity analyses in which we fit the 2 main multivariable Poisson regression models (ie, one comparing in-network with cash-pay patients and one comparing in-network with out-of-network [cash-pay] and uninsured [cash-pay] patients) among patients who were buprenorphine negative at intake.

Limitations

Our study had several limitations. First, our sample only included patients receiving care from a single TBOT provider and mostly residing in New York and Pennsylvania, and thus our findings may not be generalizable to other OUD treatment settings or regions. Second, our delineation of cash-pay patients as out-of-network or uninsured among a subsample of patients relied on self-reported insurance status at the time of care initiation, which could be vulnerable to reporting inaccuracies or undetected changes in insurance status over time. Third, our retention and PDC definitions relied on prescriptions written, not prescriptions dispensed, and thus could overestimate both measures if prescriptions were either never dispensed or dispensed significantly later than when written. Fourth, we only had access to buprenorphine prescriptions from Ophelia providers and thus our retention measure only captures retention at Ophelia and cannot account for patients who may discontinue treatment at Ophelia yet receive buprenorphine prescriptions from other providers. Fifth, to be consistent with prior studies in this population²⁰ and with other retention studies,^27-29 we used a 30-day gap in buprenorphine coverage to define discontinuation; however, no clear gold standard exists for this methodological parameter and use of shorter gaps (eg, 14 or 7 days) may lead to higher rates of discontinuation. Sixth, we did not have access to type of insurance (ie, Medicaid, Medicare, commercial) for all patients during the study period and thus were unable to evaluate differential associations by insurance type; however, we have no reason to expect the composition of the sample to deviate significantly from the composition of active patients at the time of analysis (67% Medicaid), and thus expect results to be driven largely by Medicaid patients. In addition, we did not have access to out-of-pocket costs. Utilization management policies (eg, prior authorizations) and out-of-pocket costs vary by insurance type and plan provider and have been shown to be associated with buprenorphine treatment retention,^14,16,30-32 highlighting the need for future research to further investigate the relationship between insurance type (and corresponding policies and out-of-pocket costs) and retention in telehealth settings. Seventh, we did not have access to other measures that may be associated with both insurance status and retention (eg, socioeconomic status, general life stability) that could potentially bias our findings; however, comparing retention outcomes between in- and out-of-network patients permitted evaluating the association between payment source and retention among patients with insurance, which may better control for systematic differences between insured and uninsured patients.

Results

A total of 4216 patients (1734 in-network patients, 2482 cash-pay patients) had an intake during the study period and received a buprenorphine prescription within 7 days (see Appendix Table S1 for a breakdown of all intake visits completed during the study period and corresponding reasons for inclusion/exclusion). Of these patients, 3842 (91.1%; 1613 in-network patients, 2229 cash-pay patients) maintained a consistent payment type while in care and were therefore included in the analysis. In-network patients were more likely to be female (P < .001), non-Hispanic White (P < .001), reside in a rural zip code (P < .001), reside in the Northeast United States (P < .001), and to already be receiving buprenorphine at the start of care relative to cash-pay patients (P < .001) (Table 1).

Table 1.

Baseline characteristics by overall payment source (in-network benefits vs cash-pay).

Characteristic	All patients, n (%)a	In-network-benefits patients, n (%)a	Cash-pay patients, n (%)a	P b
Total	3842	1613	2229
Age at intake visit				.466
Under 30 y	633 (16.5%)	260 (16.1%)	373 (16.7%)
30–39 y	1678 (43.7%)	730 (45.3%)	948 (42.5%)
40–49 y	977 (25.4%)	404 (25.0%)	573 (25.7%)
50–59 y	418 (10.9%)	169 (10.5%)	249 (11.2%)
60 y and older	134 (3.5%)	50 (3.1%)	84 (3.8%)
Gender				<.001
Male	2056 (53.5%)	799 (49.5%)	1257 (56.4%)
Female	1697 (44.2%)	787 (48.8%)	910 (40.8%)
Trans/nonbinary/other	38 (1.0%)	16 (1.0%)	22 (1.0%)
Race/ethnicity				<.001
Non-Hispanic White	3083 (80.2%)	1337 (82.9%)	1746 (78.3%)
Hispanic/Latino	301 (7.8%)	91 (5.6%)	210 (9.4%)
Non-Hispanic African American	186 (4.8%)	82 (5.1%)	104 (4.7%)
Non-Hispanic other or multiple races	178 (4.6%)	74 (4.6%)	104 (4.7%)
Urbanicity				<.001
Urban	3182 (82.8%)	1274 (79.0%)	1908 (85.6%)
Rural	648 (16.9%)	337 (20.9%)	311 (14.0%)
US region				<.001
Northeast	2389 (62.2%)	1427 (88.5%)	962 (43.2%)
South	1038 (27.0%)	141 (8.7%)	897 (40.2%)
West	415 (10.8%)	45 (2.8%)	370 (16.6%)
Buprenorphine status at intake				<.001
Negative	2114 (55.0%)	778 (48.2%)	1336 (59.9%)
Positive	1652 (43.0%)	797 (49.4%)	855 (38.4%)
Care episode				.439
First	3755 (97.7%)	1580 (98.0%)	2175 (97.6%)
Subsequent	87 (2.3%)	33 (2.0%)	54 (2.4%)

Source: Patient data from Ophelia Health, Inc.

^aColumn percentages are presented.

^bP values calculated using chi-square tests.

There were 3676 (95.7%) patients (1548 in-network patients, 2128 cash-pay patients) who had values for all covariates and were thus included in the multivariable analysis. In the multivariable Poisson model evaluating 6-month retention, in-network patients were more likely to be retained at 6 months compared with cash-pay patients (adjusted RR [aRR]: 1.50; 95% CI: 1.40–1.62; P < .001), controlling for covariates (Table 2). Full regression results (including regression coefficients, robust standard errors, and P values) for all independent variables are presented in Appendix Table S2. The Kaplan-Meier curve comparing time to discontinuation between in-network and cash-pay patients is presented in Appendix Figure S1. The mean PDC was 0.93 (SD = 0.11) among all patients and there was no difference between in-network (mean = 0.93, SD = 0.11) and cash-pay (mean = 0.93, SD = 0.10) patients (P = .179).

Table 2.

Six-month retention rates among patient subgroups and results of multivariable Poisson regression model evaluating associations between overall payment source (in-network benefits vs cash-pay), other patient characteristics, and retention in buprenorphine treatment at 6 months.

Characteristic	Not retained at 180 days, n (%)a	Retained at 180 days, n (%)a	Adjusted risk ratio (95% CI)b
Total	1754 (47.7%)	1922 (52.3%)
Payment source
Cash-pay	1247 (58.6%)	881 (41.4%)	Reference
In-network benefits	507 (32.8%)	1041 (67.2%)	1.50 (1.40–1.62)
Age at intake visit
Under 30 y	374 (62.1%)	228 (37.9%)	Reference
30–39 y	741 (46.1%)	867 (53.9%)	1.35 (1.21–1.5)
40–49 y	416 (44.3%)	522 (55.7%)	1.43 (1.28–1.59)
50–59 y	165 (41.7%)	231 (58.3%)	1.52 (1.34–1.72)
60 y and older	58 (43.9%)	74 (56.1%)	1.52 (1.28–1.81)
Gender
Male	964 (48.2%)	1038 (51.8%)	Reference
Female	773 (47.2%)	864 (52.8%)	0.98 (0.93–1.04)
Trans/nonbinary/other	17 (45.9%)	20 (54.1%)	1.14 (0.85–1.53)
Race/ethnicity
Non-Hispanic White	1365 (45.2%)	1657 (54.8%)	Reference
Hispanic/Latino	186 (62.6%)	111 (37.4%)	0.81 (0.7–0.93)
Non-Hispanic African American	113 (62.1%)	69 (37.9%)	0.77 (0.64–0.92)
Non-Hispanic other or multiple races	90 (51.4%)	85 (48.6%)	0.95 (0.83–1.1)
Urbanicity
Urban	1485 (48.6%)	1569 (51.4%)	Reference
Rural	269 (43.2%)	353 (56.8%)	0.99 (0.92–1.06)
Buprenorphine status at intake
Negative	1215 (59.0%)	844 (41.0%)	Reference
Positive	537 (33.3%)	1075 (66.7%)	1.51 (1.42-1.61)
Care episode
First	1704 (47.4%)	1891 (52.6%)	Reference
Subsequent	50 (61.7%)	31 (38.3%)	0.69 (0.53–0.89)

Source: Patient data from Ophelia Health, Inc.

^aRow percentages are presented to allow direct presentation of 6-month retention rates.

^bAdjusted risk ratios and 95% CIs calculated using a multivariable Poisson regression model with robust standard errors and including all presented characteristics, as well as US state as independent variables.

Among the 3842 patients included in the main analysis, 1926 (50.1%; 868 in-network patients, 1058 cash-pay patients) had intake visits on or after May 26, 2022, when the new prospective patient process collecting self-reported insurance status first was launched. Among these 1926 patients, 882 (45.8%; 404 in-network patients, 478 cash-pay patients) completed the new prospective patient process and thus had data available to be categorized as in-network, out-of-network (cash-pay), or uninsured (cash-pay). Differences in baseline characteristics between in-network, out-of-network (cash-pay), and uninsured (cash-pay) patients are presented in Table 3. In the multivariable Poisson model evaluating 180-day retention among this subsample, in-network patients (aRR: 1.86; 95% CI: 1.54–2.23; P < .001) and uninsured (cash-pay) patients (aRR: 1.25; 95% CI: 1.02–1.54; P = .035) were more likely to be retained in treatment at 6 months compared with out-of-network (cash-pay) patients, controlling for covariates (Table 4). Full regression results (including regression coefficients, robust standard errors, and P values) for all independent variables are presented in Appendix Table S3. The Kaplan-Meier curve comparing time to discontinuation between in-network, out-of-network (cash-pay), and uninsured (cash-pay) patients is presented in Appendix Figure S2. The mean PDC was 0.93 (SD = 0.11) among all patients in this subsample and there were no differences between in-network (mean = 0.93, SD = 0.11), out-of-network (cash-pay; mean = 0.94, SD = 0.11), and uninsured (cash-pay; mean = 0.93, SD = 0.11) patients (P = .745).

Table 3.

Baseline characteristics by detailed insurance status and payment source (in-network vs out-of-network [cash-pay] vs uninsured [cash-pay]).

Characteristic	All patients, n (%)a	In-network patients, n (%)a	Out-of-network (cash-pay) patients, n (%)a	Uninsured (cash-pay) patients, n (%)a	P b
Total	882	404	275	203
Age at intake visit					.018
Under 30 y	141 (16.0%)	60 (14.9%)	61 (22.2%)	20 (9.9%)
30–39 y	399 (45.2%)	185 (45.8%)	112 (40.7%)	102 (50.2%)
40–49 y	209 (23.7%)	92 (22.8%)	67 (24.4%)	50 (24.6%)
50–59 y	100 (11.3%)	52 (12.9%)	28 (10.2%)	20 (9.9%)
60 y and older	33 (3.7%)	15 (3.7%)	7 (2.5%)	11 (5.4%)
Gender					.056
Male	464 (52.6%)	201 (49.8%)	144 (52.4%)	119 (58.6%)
Female	389 (44.1%)	197 (48.8%)	119 (43.3%)	73 (36.0%)
Trans/nonbinary/other	4 (0.5%)	1 (0.2%)	2 (0.7%)	1 (0.5%)
Race/ethnicity					.121
Non-Hispanic White	678 (76.9%)	321 (79.5%)	205 (74.5%)	152 (74.9%)
Hispanic/Latino	73 (8.3%)	23 (5.7%)	29 (10.5%)	21 (10.3%)
Non-Hispanic African American	53 (6.0%)	29 (7.2%)	17 (6.2%)	7 (3.4%)
Non-Hispanic other or multiple races	41 (4.6%)	21 (5.2%)	11 (4.0%)	9 (4.4%)
Urbanicity					.188
Urban	730 (82.8%)	325 (80.4%)	231 (84.0%)	174 (85.7%)
Rural	148 (16.8%)	77 (19.1%)	44 (16.0%)	27 (13.3%)
US region					<.001
Northeast	490 (55.6%)	325 (80.4%)	107 (38.9%)	58 (28.6%)
South	286 (32.4%)	61 (15.1%)	123 (44.7%)	102 (50.2%)
West	106 (12.0%)	18 (4.5%)	45 (16.4%)	43 (21.2%)
Buprenorphine status at intake					<.001
Negative	493 (55.9%)	190 (47.0%)	179 (65.1%)	124 (61.1%)
Positive	371 (42.1%)	207 (51.2%)	91 (33.1%)	73 (36.0%)

Table only includes a subset of patients for whom more detailed insurance data were available; see text for details. Source: Patient data from Ophelia Health, Inc.

^aColumn percentages are presented.

^bP values calculated using chi-square tests or Fisher's exact tests where expected cell sizes <5.

Table 4.

Six-month retention rates among patient subgroups and results of multivariable Poisson regression model evaluating associations between detailed insurance status and payment source (in-network vs out-of-network [cash-pay] vs uninsured [cash-pay]), other patient characteristics, and retention in buprenorphine treatment at 6 months.

Characteristic	Not retained at 180 days, n (%)a	Retained at 180 days, n (%)a	Adjusted risk ratio (95% CI)b
Total	363 (44.1%)	461 (55.9%)
Insurance status and payment source
Out-of-network (cash-pay)	162 (63.0%)	95 (37.0%)	Reference
Uninsured (cash-pay)	94 (51.9%)	87 (48.1%)	1.25 (1.02–1.54)
In-network	107 (27.7%)	279 (72.3%)	1.86 (1.54–2.23)
Age at intake visit
Under 30 y	73 (56.2%)	57 (43.8%)	Reference
30–39 y	167 (45.1%)	203 (54.9%)	1.15 (0.95–1.40)
40–49 y	85 (43.1%)	112 (56.9%)	1.25 (1.02–1.54)
50–59 y	24 (25.3%)	71 (74.7%)	1.59 (1.29–1.96)
60 y and older	14 (43.8%)	18 (56.2%)	1.28 (0.94–1.73)
Gender
Male	196 (44.0%)	249 (56.0%)	Reference
Female	165 (44.0%)	210 (56.0%)	0.94 (0.84–1.05)
Trans/nonbinary/other	2 (50.0%)	2 (50.0%)	1.12 (0.57–2.18)
Race/ethnicity
Non-Hispanic White	272 (41.2%)	388 (58.8%)	Reference
Hispanic/Latino	46 (63.9%)	26 (36.1%)	0.76 (0.57–1.00)
Non-Hispanic African American	33 (63.5%)	19 (36.5%)	0.69 (0.49–0.97)
Non-Hispanic other or multiple races	12 (30.0%)	28 (70.0%)	1.16 (0.96–1.39)
Urbanicity
Urban	303 (44.2%)	382 (55.8%)	Reference
Rural	60 (43.2%)	79 (56.8%)	0.95 (0.81–1.11)
Buprenorphine status at intake
Negative	267 (57.4%)	198 (42.6%)	Reference
Positive	96 (26.7%)	263 (73.3%)	1.54 (1.36–1.74)

Table only includes a subset of patients for whom more detailed insurance data were available; see text for details. Source: Patient data from Ophelia Health, Inc.

^aRow percentages are presented to allow direct presentation of 6-month retention rates.

^bAdjusted risk ratios and 95% CIs calculated using a multivariable Poisson regression model with robust standard errors and including all presented characteristics, as well as US state as independent variables.

Among the 1926 patients with intakes on or after May 26, 2022, there were no significant differences between patients included and not included in the subsample (P > .05) (Appendix Table S4).

In the sensitivity analyses evaluating the associations between insurance status and payment source and 180-day retention among patients who were buprenorphine negative at intake, results were similar to the main analyses (Appendix Tables S5 and S6).

Discussion

We found that patients able to use in-network benefits were 50% more likely to successfully retain at 6 months in care, a minimum duration of care recommended by CMS, compared with cash-pay patients (including both out-of-network and uninsured patients). This disparity persisted when restricting to insured patients and comparing in-network with out-of-network patients, suggesting that it is not likely to be driven only by unmeasured differences between insured and uninsured patients. These findings together indicate that the ability to use insurance benefits to cover service costs may have a meaningful impact on patient retention in TBOT settings. Patients who had to pay cash, even if insured, had significantly worse retention.

Analyses further determined that 6-month retention was actually higher among uninsured cash-pay patients in comparison with out-of-network cash-pay patients. Although unverifiable given that we only had access to Ophelia prescription data, this finding may reflect patient preference for using in-network benefits with providers who can accept their insurance (ie, out-of-network patients may actively search for in-network providers and then transfer care). Patients transferring care may experience treatment disruptions and resultantly diminished medication access, which is consistent with prior findings that provider continuity facilitates better retention and highlights the potential risks of health plans with sparse networks for patients with OUD.³³ Expanding networks to include a wide variety of OUD treatment providers, including telehealth providers, could facilitate greater functional network adequacy (ie, ensuring that beneficiaries truly have access to covered services in their vicinity without excessive wait times) and treatment access and thus continuity for patients with OUD.

While we were only able to track written prescriptions and unable to account for actual prescription fills at pharmacies, calculated PDCs between cohorts suggest that medication adherence while in care was equivalent between cohorts, irrespective of insurance status and payment source. Prior studies have suggested that higher prescription co-pays diminish timely fills (ie, the interval between when a prescription is written vs filled at a community pharmacy),^16,17,32 but replicating these findings was outside the scope of our analysis.

Although we did not have access to total out-of-pocket costs for patients in this analysis, our finding that patients with in-network benefits remained in care longer than uninsured patients and those with out-of-network benefits is consistent with multiple studies showing that higher out-of-pocket costs were associated with worse retention in buprenorphine treatment for OUD among commercially insured individuals.^14,16,32 However, another study among commercially insured patients found that high-deductible health plans increased out-of-pocket spending but had no effect on treatment retention, suggesting that demand for MOUD is relatively inelastic in certain settings and among certain populations.³⁴ These studies were restricted to commercially insured individuals who may have more disposable income than the average Ophelia patient, the majority of whom are on Medicaid or uninsured and may thus be more sensitive to higher health care costs. Indeed, in data collected since the end of the study period, 50.3% of patients reported facing recent difficulties with access to stable housing, paying bills, or reliable transportation (including gas costs).

Overall our findings suggest that the use of in-network benefits was significantly associated with successful retention on buprenorphine in TBOT settings, which has been shown in other studies to reduce the risk of overdose and death.^35-37 These findings may have significant implications for current efforts to enforce parity for mental health services across insurance plans and to improve functional network adequacy.

Supplementary material

Supplementary material is available at Health Affairs Scholar online.

Funding

S.M.M. was funded by a National Institute on Drug Abuse (NIDA) CHERI$H Center grant (P30 DA040500).