Patient Engagement in and Adaptations to Delivery of Outpatient Care for Opioid Use Disorder During the COVID-19 Pandemic

The AMNet Recruitment Survey is an online survey of medical practices and their providers who provide outpatient medication treatment focused on opioid use disorder within non–opioid treatment programs for substance use disorders. Practices eligible for the survey could have physicians, nurse practitioners, physician assistants, and nonprescribing mental health providers. The survey captured information on practice characteristics, practices’ adaptations to care delivery for opioid use disorder since the declaration of the COVID-19 public health emergency, as well as the impact of the pandemic (18) on patient demand and engagement. The survey was implemented after the DEA issued regulatory easements regarding telehealth MOUD management.

Potential respondents completed an online survey, via Alchemer (https://www.alchemer.com), that was distributed broadly by the American Psychiatric Association (APA) and by the American Society of Addiction Medicine through their promotional channels (i.e., newsletters, e-mails, social media, membership Listservs, and advertisements in APA’s e-newsletter, Psychiatric News). Postcards advertising the survey were mailed to buprenorphine providers certified by the Substance Abuse and Mental Health Services Administration who were eligible to treat ≥100 patients. AMNet investigators also publicized the survey during presentations at professional conferences. For this study, we purposively sampled physician and nonphysician (nurse practitioners and physician assistants) prescribers. The eligible study sample (N=569) was drawn from a total of 643 respondents as of November 22, 2020 (see Figure S1 in the online supplement to this article).

The AMNet Recruitment Survey (https://bit.ly/3gSHOHq) comprised two sections. Section 1 inquired about general practice characteristics, including location (urban, rural, or suburban), practice type (solo, group, community health or mental health center, state-certified substance use treatment program, or other), types of clinicians (physician, nurse practitioner, physician assistant, therapist, or counselor), whether the practice was covered by federal alcohol and drug confidentiality regulations (part 2 of Title 42 of the Code of Federal Regulations [42 CFR]: yes vs. no), and whether the practice provided buprenorphine prescriptions (yes vs. no), extended-release naltrexone (XR-NTX) for opioid use disorder (yes vs. no), and XR-NTX for alcohol use disorder (yes vs. no). The last question in section 1 of the survey asked respondents whether they were willing (yes vs. no) to answer questions about the practice’s experience during the COVID-19 pandemic. Only respondents who answered yes to the last question were asked to complete section 2.

Section 2 of the survey was adapted from a similar survey (18). Respondents were asked whether MOUD adaptations had been made (yes vs. no) in response to the COVID-19 pandemic and, if so, what the adaptations were (i.e., changes to urine drug screening [UDS], visit type, buprenorphine prescription duration [e.g., weekly or more], refill frequency, and injectable naltrexone administration). Respondents also were asked whether the practice’s patients had been “impacted by COVID-19” (yes vs. no) and, if so, what the impacts were (increase or decrease in demand for visits, preference for virtual vs. in-person MOUD appointments, difficulty engaging or retaining patients in care, and changes in treatment retention). The adaptation and impact questions allowed respondents to select multiple responses.

For this study, the primary dependent variables were MOUD adaptation, UDS adaptation, and impact on patients. Practice characteristics (e.g., practice location and type, prescriber type, and 42 CFR part 2) were independent variables that were used to examine factors associated with MOUD adaptation, UDS adaptation, and patient impact. Prescriber type was physician versus nonphysician prescribers (nurse practitioners and physician assistants). The survey was approved by the APA Institutional Review Board.

Frequency distributions were calculated for practice and prescriber characteristics, adaptations to care delivery, and changes in patient demand and engagement. Further descriptive analyses were conducted to characterize changes to MOUD visits, MOUD management, UDS practices, and patient demand and engagement. Bivariate analyses were used to assess differences in general practice characteristics by MOUD adaptation (yes vs. no), UDS adaptation (yes vs. no), and patient impact (yes vs. no). These bivariate results, including unadjusted logistic models, informed the decision which variables to include in stepwise multivariable logistic regression models in order to explore whether general practice and prescriber characteristics predicted MOUD adaptation, UDS adaptation, and patient impact. Because of high correlation between practices prescribing XR-NTX for alcohol use disorder and those prescribing XR-NTX for opioid use disorder (Spearman correlation coefficient ρ=0.83, p<0.001), these two variables were not included in the same multivariable adjusted models. Using a 26:1 ratio of those who prescribed buprenorphine (N=494) relative to those who did not (N=19), we conducted post hoc sensitivity analyses with case-control designs (i.e., 1:2 and 1:3 ratios of cases [nonprescribers] to controls [prescribers]) on the basis of buprenorphine prescribing (no vs. yes) to assess the robustness of findings across all outcomes. All analyses were conducted with SAS, version 9.4, with α=0.05 as the cutoff value for statistical significance.

In total, 569 practitioners from outpatient opioid use disorder treatment settings in 48 U.S. states (excluding South Dakota and Wyoming) responded to section 1 of the AMNet Recruitment Survey (see the online supplement). Most eligible respondents were physician prescribers (92%), from urban (43%) or suburban areas (36%), and in group (32%) or solo practices (27%) (Table 1). Of these respondents, 533 (94%) agreed to complete the COVID-19 items in section 2, of whom 516 (97%) responded to one or more questions in this section. Respondents completing one or more COVID-19 items were similar to those who did not complete these items on all practice characteristics, except for practice type and buprenorphine prescription status (yes vs. no). Respondents who did not complete the COVID-19–related questions were more likely to be in the “other” practice type category (χ²=8.54, df=2, p=0.01), which included outpatient facilities of the U.S. Department of Veterans Affairs, academic centers, and hospitals. Moreover, the practices and their providers who did not prescribe buprenorphine were less likely to complete questions on impacts of the COVID-19 pandemic (χ²=11.35, df=1, p<0.001).

Of the 516 respondents to section 2 of the survey, 74% (N=381) reported MOUD adaptations in response to the COVID-19 pandemic (Table 2). Practice type or location or prescriber type did not significantly differ between respondents who reported MOUD adaptations and those who did not. However, 99% of respondents who made MOUD adaptations prescribed buprenorphine, compared with 90% of those who did not make MOUD adaptations. This difference remained statistically significant even after the analyses were adjusted for practice characteristics (adjusted OR [AOR]=7.70). We conducted post hoc sensitivity analyses to examine the robustness and reliability of the findings (Table 2). A case-control design with a 1:3 ratio of randomly selected buprenorphine prescribers (control group; N=57) and respondents who did not prescribe the medication (case group; N=19) showed similar differences (AOR=15.04), indicating robustness of the findings; the same was true when a 1:2 ratio was used (AOR=12.04) (Table 2).

Of the 381 respondents who reported any MOUD adaptations in response to the COVID-19 pandemic, 379 answered the question about UDS adaptations, 77% of whom made UDS adaptations (Table 3). Overall, 99% of buprenorphine prescribers who completed the UDS adaptation question made UDS adaptations. Similarly, 77% of respondents who prescribed XR-NTX for alcohol use disorder or opioid use disorder reported making UDS adaptations. After the analyses were adjusted for practice characteristics, these differences were no longer statistically significant. The sample of respondents who did not prescribe buprenorphine was too small for sensitivity analyses on UDS adaptation.

Of the 516 respondents to the COVID-19 component of the survey, 513 (99%) completed the question, “Were your patients impacted by COVID-19?” (Table 4). Of these respondents, 89% (N=458) indicated that their patients were affected by COVID-19. Buprenorphine prescribers were more likely to indicate an impact of the pandemic on patients compared with those who did not prescribe the medication (χ²=5.60, df=1, p=0.02). However, this difference was no longer statistically significant when the analyses were controlled for practice characteristics (AOR=2.84). Similar findings were observed in the sensitivity analyses.

Most respondents adapted their MOUD delivery by implementing virtual visits for initiation of care (67%, N=256 of 381) and follow-up visits (77%, N=292 of 381). The proportion of those who offered virtual visits only was similar for initiation and follow-up. Compared with pre–COVID-19, 30% of practices reported writing prescriptions for longer durations, with 15% writing prescriptions for more refills (Table 5). The providers reported small increases in buprenorphine prescribing (5%) and XR-NTX administration (4%).

Among 293 respondents who made UDS adaptations, 83% reduced the frequency of UDS for established patients, and 13% moved to conducting initial UDS on new patients only. A few practices (9%) indicated that they offered virtual UDS screening. Approximately 31% of practices reported increases in patient demand for MOUD visits, and 10% reported decreases. Overall, 30% of the practices reported more difficulty engaging patients, whereas 14% reported less difficulty. About 45% of practices observed that their patients preferred virtual visits for MOUD appointments. Only 29% of practices reported that patient retention was unchanged.