A short-duration telementoring pain management programme for Medicaid: impact on clinician outcomes

Abstract

Previous evaluations of the pain care-related Extension for Community Healthcare Outcomes (ECHO) telementoring programmes found that long-term programmes (16–52 weeks) improve clinician knowledge, self-efficacy, and prescribing practices. We evaluated a 6- to 7-week Pain Management ECHO in Nevada Medicaid clinician networks. We collected pre- and post-knowledge and self-efficacy scores from 15 of 18 unique ECHO participants (83% response rate). We derived opioid prescribing outcomes from 44 894 Medicaid pharmacy claims records from 11 ECHO participants and 10 comparison clinicians. The three outcomes included any opioid (binary), non-opioid pain medication (binary), and opioid dose (continuous). Logistic regressions using difference-in-difference (DID) estimated the ECHO treatment effects. Knowledge scores (75% to 82%) and self-efficacy scores (3.4–4.1) increased after ECHO participation. After ECHO participation, opioid prescribing decreased, and non-opioid prescribing increased; changes in both outcomes were above and beyond changes in the comparison group (any opioid DID treatment effect: −0.6 percentage points; non-opioid pharmacologic: 1.1 percentage points). Incremental changes across three domains of Moore’s Framework for continuing medical education provide evidence supporting a short-duration ECHO intervention in partnership with Medicaid managed care. Promulgation of this less resource-intensive approach can sustainably aid clinicians in managing pain experienced by Medicaid beneficiaries.

Introduction

In previous decades, opioid analgesic prescribing without weighing the risks and benefits of opioids or considering less addictive alternatives contributed to high rates of opioid addiction and overdose deaths [1–4]. The 2016 and 2022 Centers for Disease Control and Prevention (CDC) guidelines, along with the implementation of state Prescription Drug Monitoring Programs and state opioid prescribing regulations, led to reductions in opioid prescribing rates and doses. However, uncertainty remained among clinicians about how to best care for patients with chronic pain [5, 6], a condition experienced by over a fifth of adults in the USA [7]. A survey found that close to 90% of clinician respondents required retraining on the use of non-opioid pain medication (e.g. Nonsteroidal anti-inflammatory drugs, Acetaminophen, and Topical Agents) [8]. Indeed, even as the pain care pendulum swings away from the extreme of overprescribing opioids for pain [9, 10], clinicians may need additional support for safe and effective pain management [11].

The Extension for Community Healthcare Outcomes (ECHO) programme, a telementoring initiative that enables clinicians to learn new clinical practices and resolve challenging patient cases, provided a venue for this training. Initially found to be clinically effective in expanding access to Hepatitis C treatment in primary care, this programme’s hub and spoke model has also been applied to pain management by ECHO programmes across the USA and the world [12, 13].

The University of Nevada, Reno School of Medicine’s Office of Statewide Initiative’s Project ECHO Nevada (PEN) established a Pain Management ECHO programme to improve clinician knowledge, self-efficacy, and prescribing behaviour. PEN marketed the programmes to clinicians in a Medicaid managed care network. Medicaid is the public insurance programme in the USA for low-income individuals, and all but 10 US states offer Medicaid benefits either in full or in part through private managed care companies [14]. Historically, Medicaid patients experienced higher rates of opioid prescribing and overdose than privately insured patients [15–17].

Prior evaluations of other Pain Management ECHO programmes have found that ECHO participation increased clinicians’ knowledge and self-efficacy of safe and effective pain care practices in community health centres in the USA [18], the Veterans Health Administration [19], and among primary care clinicians in Ontario, Canada [20], and in Chhattisgarh, India [21, 22]. Several studies, including one set on US Military Bases, have used quasi-experimental study designs to determine that Pain Management ECHO programmes reduce opioid prescriptions [13, 18] and opioid dose levels [13, 23, 24].

The present study builds on the existing research in three ways: First, it describes a novel partnership between an ECHO programme based at an academic institution and a Medicaid managed care plan. Second, it considers non-opioid pain medication prescribing, an outcome that addresses control of pain symptoms but is not currently reported in the ECHO evaluation literature. Third, and most importantly, it evaluates an ECHO programme that was substantially shorter than the programmes evaluated in the literature (i.e. lasting 6–7 week versus 16−52 week programmes in the literature). Since the shorter duration may be a more affordable format, it is essential to know how it performs on four domains from the Moore’s framework of continuing medical education: participation, knowledge, self-efficacy, and behaviour change [25].

Methods

Overview

We used a difference-in-difference (DID) framework to examine changes in opioid prescribing by ECHO-participating clinicians after differencing out changes in opioid prescribing by non-ECHO-participating clinicians. We supplemented the DID analysis with attendance data and before and after responses from knowledge and self-efficacy questionnaires completed by the ECHO-participating clinicians. This allowed us to evaluate multiple steps in Moore’s framework for continuing medical education. Moore’s framework is a hierarchical model for assessing and improving educational outcomes among healthcare professionals [25]. In this framework, self-efficacy, or confidence in a clinical topic connects declarative knowledge to an even more impactful outcome: behaviour change [25]. The University of Nevada, Reno Institutional Review Board reviewed and determined that this study met the criteria for Exemption Category #4.

Project ECHO Nevada Pain Management ECHO description

Dr. Sanjeev Arora developed the ECHO model of telementoring at the University of New Mexico in the early 2000s, and the Project ECHO Nevada (PEN) Pain Management ECHO used this model to improve care for Nevada Medicaid pain patients. The University of New Mexico disseminated the ECHO model worldwide in the 2010s [26]. The iteration of the PEN Pain Management ECHO being evaluated here started in the Fall of 2020 in partnership with a Nevada Medicaid Managed Care Organization (MCO). The partnership with the MCO allowed the PEN programme to target the Medicaid population covered by the MCO in an attempt to reduce opioid prescribing and overdose in this population [17, 27].

A previous iteration of the PEN Pain ECHO started in 2017, with support from the State Targeted Response grant funding from the Department of Health and Human Services, as legislated in the 21st Century Cures Act. After collecting positive participant satisfaction data from this initial iteration, the original hub team was reconvened for the 2020 PEN Pain Management ECHO. The hub team included a Physical Medicine and Rehabilitation Doctor with Board Certification in Pain Medicine, a licensed psychologist, and a licensed Clinical Alcohol and Drug Counsellor who is also a licensed Clinical Professional Counsellor. The programme selected these practitioners based on their expertise in their respective fields and professional recommendations from other colleagues. The hub team also included a programme coordinator who maintained regular e-mail communication with participants and assisted with background technological infrastructure (i.e. Zoom™ videoconferencing technology).

Two cohorts of clinicians completed the curriculum: Cohort 1 in Fall 2020 over six sessions and Cohort 2 in Fall 2021 over seven sessions. The sessions for both cohorts were held on the second and fourth Wednesday of each month. The first cohort held hour-long morning sessions, and the second cohort held hour-and-15-minute-long sessions to ensure more time for case review.

The MCO recruited participating clinicians based on their opioid prescribing patterns before ECHO participation. They recruited Medical Doctors (MD, DO), Advanced Practice Registered Nurses (APRN), and Physician Assistants (PA). The MCO paid ECHO participants $500 per ECHO session they attended, with an additional bonus of $500 if they presented a case for review. The MCO only compensated participating clinicians if they attended all but one of the ECHO sessions and completed the pre-test and post-test.

The hub team oriented ECHO participants to the PEN Pain ECHO format in the first session. During the five/six remaining meetings, the hub team delivered short didactics covering:

1. The history of the opioid epidemic and understanding and assessing popular drugs

2. Introduction to mental health and substance use treatment (Cohort 2 only)

3. CDC guidelines for prescribing opioids

4. 12 steps to proper opioid prescribing

5. Motivational interviewing for patients with chronic pain, and

6. Cognitive behavioural therapy approach to pain management.

The hub team selected these topics to provide participating clinicians with a range of guideline-based approaches to caring for patients presenting with pain, using tools from various clinical disciplines. Additionally, the hub team designed didactic content to provide participants with a skill base that can be applied even as the landscape of pain management challenges and drug availability evolves.

In close fidelity to the ECHO model, the PEN Pain Management ECHO format relied on case presentations to motivate learners and engage their interests by asking spoke participants to bring cases from their clinical practice [10, 20, 28]. This element of the ECHO model was founded in Situated Learning Theory [29] and the concept of a community of practice [30]. Cases presented an environment for participants to develop adaptive expertise, a form of knowledge characterized by flexibility that results in efficient innovation [31]. The ECHO model promoted discussion of these real-life challenges in hopes that practical recommendations could be made to the presenting participant, and the other participants would gain information applicable to their patients.

Before each meeting, one or two participant/s completed a case review form summarizing their patient’s medical history. Participants completed the forms on Research Electronic Data Capture (RedCAP), a HIPAA-compliant online data collection programme [32], including information about the patient’s age, race/ethnicity, body mass index, pain diagnoses, and treatment. The form included a section where participants described a patient care challenge they would like to discuss with the ECHO hub and other spokes. During the case review, the presenting participant introduced the case and the question, while the other participants and the hub team asked clarifying questions, and the other participants and the hub team offered ideas for how to address the challenges of the case. The programme coordinator compiled written recommendations, sent the recommendations to the hub team for edits/approval, and finally, shared the recommendations with the case presenter(s).

Data

We collected participant attendance records, including name, practice location, and credentials in iECHO, an online attendance database enabled for any ECHO programme affiliated with the University of New Mexico [33]. We also collected knowledge and self-efficacy questionnaire responses from ECHO participants before and after programme completion. The hub team provided clinical input to develop questions with clinical relevancy and alignment with the material delivered throughout the programme.

The knowledge questionnaire contained 16 ‘multiple-choice’ and ‘fill in the blank’ questions, and the self-efficacy questionnaire contained 24 Likert scale (1–5, with 5 being the highest) items. The participants completed both questionnaires on RedCAP. The knowledge and self-efficacy questionnaires can be found in Supplementary Appendix A.

The Nevada Medicaid MCO provided de-identified pharmacy claims data from January to August 2020–2022, covering 8 months before and after all PEN Pain Management ECHO programmes happening in the Fall of 2020 and 2021. The claims represented all Medicaid MCO claims submitted by the ECHO participants during the study period and all claims for clinicians who were invited to participate in ECHO but did not ultimately do so. The pharmacy claims included member and clinician identification numbers, drug name and national drug code, fill date, days’ supply, and quantity dispensed.

Study samples

We administered pre- and post-knowledge and self-efficacy surveys to clinicians during their first cohort. We obtained responses from 15 of the 18 unique participating clinicians.

To be included in the claims analysis, we required that treatment group clinicians had records in both the pre- and post-periods of their respective cohort/s . The MCO claims data extract excluded salaried clinicians who did not submit claims to the MCO, so these clinicians were also excluded from our claims analysis. We created a single comparison group comprised of clinicians who did not participate in the Pain Management ECHO but who did have claims during the pre- and post-periods for both Cohort 1 and Cohort 2. We further required that clinicians in the comparison group had an average number of prescriptions per patient in the range observed among the treatment group clinicians (between 2.5 and 18.5 prescriptions per patient). These criteria resulted in a sample of 11 clinicians in the treatment group (two clinicians participating in Cohorts 1 and 2, three clinicians participating in Cohort 1 only, and six clinicians participating in Cohort 2 only) and 10 clinicians in the comparison group. The 21 clinicians in the claims analysis contributed 44 894 total prescription claims records.

Claims-derived prescribing outcomes

We reported results for three claims-based outcome variables: any opioid prescribing (binary), prescribing of non-opioid pain medication (binary), and opioid dose (continuous). We coded opioid prescriptions as = 1 for qualifying opioid prescriptions (excluding buprenorphine for pain) and = 0 otherwise. We calculated the daily morphine milligram equivalent (MME) by dividing the quantity dispensed by the days’ supply and multiplying the result by the CDC-reported conversion factor and strength per unit. MME is a continuous variable.

The third prescribing outcome, non-opioid pain medication, identified the use of alternative pharmacologic pain therapies, including Pregabalin and Gabapentin (see Supplementary Appendix B for complete list), which we coded = 1 for cases of non-opioid pain medication prescriptions and = 0 otherwise. Other outcomes that we created from the claims data but did not analyse included the use of physical therapy, a CDC-recommended non-pharmacological alternative to opioids for pain treatment [34], psychotherapy, which can also be therapeutic for physical pain, and naloxone prescribing. Naloxone, a critical tool for harm reduction, can revive patients who have overdosed [35, 36]. We did not perform analyses on these outcomes due to their infrequent appearance in the available claim records.

Claims-analysis predictors

Three binary indicators identified clinicians participating in 1 ECHO cohort (hereafter referred to as the ‘single-cohort’ participants), those participating in both Cohort 1 and Cohort 2 (hereafter referred to as the ‘double-cohort’ participants), and those in the comparison group. We distinguished between the ‘single-cohort’ and the ‘double-cohort’ participants because the double dose of treatment that the ‘double-cohort’ clinicians experienced may have resulted in a more substantial treatment effect. A time variable indicated claims drawn from before and after the ECHO, respectively (Post-ECHO = 1; Pre-ECHO = 0).

Clinician covariates

We obtained clinician information, including gender, enumeration year associated with the national clinician identifier, license, and speciality from the National Plan and Clinician Enumeration System. We calculated years in practice from the enumeration year to 2021. A clinician-type variable combined license and speciality information to create the following groups: primary care physician (PCP), PA, APRN, and pain physician.

Statistical analysis

We reported clinician-level changes in average knowledge (where each participant can score between 0% and 100%) and self-efficacy scores between the pre- and post-periods. Three of the knowledge test questions were ‘select all that apply’. We treated each correct answer as a possible point for 31 points on the knowledge test. The number of ECHO participants did not provide sufficient power to conduct statistical tests with a low probability of a Type 2 error (i.e. concluding no differences between groups when significant differences did exist).

We reported clinician-level frequencies for the two binary claims-based outcomes, before and after the ECHO, for single-cohort participants. We also reported adjusted odds ratios from logistic regression models on the claims-level data. These models implement the DID study design using two interaction terms: Post × Single cohort and Post × Double cohort. When the interaction term was significant, we estimated the predicted probabilities of having each outcome in the relevant treatment group and the comparison group before and after the ECHO. The differences in the (post−pre) changes in the predicted probabilities between the treatment and the comparison group estimate the ECHO programme’s treatment effects. Analyses were conducted in Stata Version 17 [37].

Results

Participation

Each session had an average of 9 attendees, and 88% of the 18 unique enrolees attended 4 or more sessions (data not in a table).

Description of clinician samples

Table 1 describes the 15 unique clinicians who responded to the questionnaire. Among them, 60% were female, 40% were PCPs, 40% were Physician Assistants, and 20% were APRNs. Clinicians were slightly more concentrated in Southern Nevada (where Las Vegas is located) than in Northern Nevada (where Reno is located) (60% versus 40%). Table 1 also describes the clinicians represented by the claims sample (21 clinicians, 44 894 prescription claims records). The 8 PCPs represented 30% of claims, the pain physician represented 7% of claims, the 6 APRNs represented 41% of claims, and the 6 PAs represented 22% of claims (data not in table).

Table 1.

Description of the clinicians included in the questionnaire and claims analyses

	Questionnaire (N = 15 clinicians)	Claims (N = 21 clinicians)
Cohort
Cohort 1	10	5a
Cohort 2	5	8a
Comparison group	NA	10a
Female	9	11
Male	6	10
Clinician type
PCP	6	8
Pain MD	0	1
APRN	3	6
Physician assistant	6	6
Years of experience
1−5 years	6	5
6−10 years	3	5
>10 years	6	11
Location
Northern Nevada	6	4
Southern Nevada	9	17

PCP: Primary Care Physician; MD: Medical Doctor; APRN: Advanced Practice Registered Nurse.

^aThe sum of these values exceeds 21 because 2 clinicians participated in Cohort 1 and Cohort 2 (i.e. the double-cohort participants).

Changes in knowledge scores

On average, survey respondent knowledge test scores increased by 7 percentage points (75% to 82%; Cohen’s d = 0.42) across the two cohorts. Both cohorts had score increases, although the magnitude was greater in Cohort 2 than in Cohort 1 (14 percentage points versus 3 percentage points). In Cohort 2, 68% of the responding clinicians had increases in knowledge scores. Knowledge questions about motivational interviewing, cognitive behaviour therapy, CDC guidelines, and substance use addiction showed improvement (incorrect before ECHO participation, correct afterwards) for five or more participants (data not in table).

Changes in self-efficacy

On average, across the two cohorts, survey respondents’ self-efficacy scores increased from 3.4 to 4.1 (out of 5). Likewise, participants in both cohorts experienced similar magnitudes of change across the three domains addressed in the questionnaire: awareness, understanding, and confidence. Increases were observed within both cohorts. Average score increases were observed among 14 of the 15 respondents. These score increases ranged in magnitude from 0.1 to 2 and had a median of 0.7. The largest numerical increases were seen among APRNs and among clinicians in Northern Nevada (data not in table).

Changes in prescribing based on claims data

Among single-cohort participants, after participating in the ECHO programme, unadjusted average proportions of any opioid prescribed as a proportion of all prescriptions decreased (1.7% to 1.0%), while the average proportions of non-opioid pain medication prescriptions as a proportion of all prescriptions increased (2.6% to 3.7%) (Fig. 1). Also, among single-cohort participants, when opioids were prescribed, the average opioid dose levels were the same before and after participating in the ECHO programme (19.3 daily MME).

Figure 1.

Unadjusted proportions of prescriptions with the outcome, before and after ECHO participation, among medications prescribed by single-cohort ECHO participants.

Table 2 reports adjusted odds ratios, which indicate whether the changes in prescription claims for ECHO participants differed significantly from those for non-ECHO participants. The statistically significant coefficient on the single-cohort interaction term for ‘any opioid’ confirmed that changes in this outcome were significantly different from changes among the non-ECHO participants (aOR: 0.6, 95%CI: 0.4, 0.9) (Table 2). Similarly, the statistically significant coefficient on the single-cohort interaction term for non-opioid pain medication confirmed that changes in this outcome were significantly different from changes among the non-ECHO participants (aOR: 1.4, 95%CI 1.1, 2.1) (Table 2). We did not observe the hypothesized dose−response effect, with greater treatment effects among double-cohort participants (Table 2).

Table 2.

Associations between ECHO participation and three prescribing outcomes (N = 44 894 prescription claims)

	Any opioid aORa (95% CI)	Non-opioid pain medication aORa (95% CI)	Opioid dose linear coefficientb (95% CI)
Post × Single cohortc	0.6 (0.4, 0.9)	1.4 (1.1, 2.1)	5.3 (−2.9, 13.5)
Post × Double cohortc	0.6 (0.1, 2.3)	0.7 (0.5, 1.0)	−35.9 (−35.8, 45.7)
Post ECHO participation	0.9 (0.8, 1.1)	1.1 (0.9, 1.2)	2.3 (−2.0, 2.6)
Single cohort	1.3 (0.9, 1.9)	1.4 (1.1, 1.9)	8.4 (−8.3, 9.1)
Double cohort	0.2 (0.1, 0.4)	1.5 (1.1, 2.1)	29.5 (10.2, 48.9)

Bold text indicates a significant 95% CI.

aOR: Adjusted odds ratio; CI: confidence interval; MME: Morphine milligram equivalents.

^aAdjusted odds ratios were estimated in a logistic regression controlling for the parameters in the table and the following covariates: clinician type, self-reported gender, and years in practice.

^bLinear coefficients estimated in an ordinary least squares regression controlling for the parameters in the table and the following covariates: clinician type, self-reported gender, and years in practice.

^cSingle cohort refers to claims from clinicians who participated in one cohort (e.g. either Cohort 1 or Cohort 2), and double cohort refers to claims from clinicians who participated in both Cohort 1 and Cohort 2.

Figure 2a reports that the treatment effect (i.e. the DID estimate) of single-cohort participation on any opioid prescription was −0.6 percentage points. Figure 2b reports that the associated treatment effect of single-cohort participation on any non-opioid pain prescriptions was 1.1 percentage points.

Figure 2.

(a) Adjusted predicted probabilities of opioid prescriptions, by ECHO participation and time. (b) Adjusted predicted probabilities of non-opioid pain medication prescriptions, by ECHO participation and time. DID: Difference-in-difference; %pts: Percentage points. Notes: The DID treatment effect was calculated by subtracting the change in the comparison group (ECHO expected) from the change in the single-cohort group (ECHO observed). Predicted probabilities were calculated using the -Margins- command in Stata Version 17.

Discussion

This evaluation of the Medicaid managed care, cohort-based, PEN Pain Management ECHO programme provides new evidence about a short-duration ECHO intervention for pain care. We observed significantly greater reductions in opioid prescribing and significantly greater increases in non-opioid pain medication prescriptions among ECHO participants than clinicians in the comparison group. We also observed improved accuracy across knowledge test questions and higher self-reported self-efficacy scores. These findings suggest that the ECHO programme may have benefited clinicians and, ultimately, patients with pain. However, the changes in opioid prescribing were small, and ECHO participation was not associated with significant improvements in the reduction of opioid doses.

Other Pain Management ECHO evaluations report similar knowledge score increases as the present study as well as similar improvements in self-efficacy scores [18–20]. The PEN Pain Management programme was substantially shorter in duration than the interventions previously reported in the literature; the finding that increases in knowledge and self-efficacy were comparable to longer interventions is striking.

Our evaluation adds to a growing literature that the ECHO format can be an effective way to change clinician opioid prescribing behaviour. Katzman et al. also observed significantly greater decreases in opioid prescribing rates among clinicians participating in a 48-week ECHO compared with a comparison group [13]. However, the treatment effect observed by Katzman et al. was more substantial than the treatment effect reported here. It is possible that while the short-duration intervention evaluated in the present analysis has a statistically significantly detectable effect, the magnitude of the effect would have been larger with a long-duration intervention. Alternatively, physicians practising on a military base may interact differently with the intervention than clinicians participating in a Medicaid managed care network, resulting in different treatment effect magnitudes. Interestingly, the double-cohort group did not have the dose−response predicted based on the Katzman study.

The present analysis presents new evidence about the prescribing of non-opioid pain medication in conjunction with limiting opioid prescriptions. Our findings may indicate that the ECHO programme successfully conveyed information about how to weigh the pros and cons of opioid prescribing, resulting in fewer opioid prescriptions and more reliance on non-opioid pain medications. This may reduce the number of Medicaid patients starting opioid pain therapy, which can lead to addiction, opioid misuse, and ultimately, overdose [38–40]. Critically, while the present analysis does not provide information on whether pain symptoms of pain patients were adequately controlled, the small increase in non-opioid pain medication drugs is encouraging. Still, it is not possible to tell if the reductions in opioid prescribing came at the expense of uncontrolled and unaddressed pain, or if gains were made through deprescribing for patients who were ready to be tapered off opioids or through less initiation of opioids where alternatives were available.

This work has several limitations. First, our study design assumes that the comparison group has outcomes with parallel trends to the treatment group in the pre-period. Supplementary Appendix C reports the monthly rates of the three outcomes in each pre-ECHO month and reveals some differences in the trends between the two groups. Although we used a comparison group of clinicians who were invited to participate in the ECHO based on their prescribing patterns, those who did enrol may have had different attitudes about opioids than the comparison group, and these attitudes may have driven some changes in the outcomes. Second, this study would have benefited from a larger sample size. This pragmatic evaluation was limited by the number of clinicians the ECHO programme can accommodate and the number of claims for Medicaid patients that participating clinicians billed. Another limitation to the pragmatic nature of this evaluation is that some knowledge questions use the ‘all of the above’ or ‘none of the above’ format, which has been critiqued in the literature [41]. Additionally, our self-reported self-efficacy results may have been inflated due to a sociability bias and were not compared with changes in a comparison group.

Implementing this PEN Pain Management ECHO in partnership with a Medicaid MCO demonstrated one model to offer ECHO programmes to clinicians in a managed care network. This partnership strategy for funding, participant recruitment strategy, and data sharing has been agreeable to both PEN and the MCO. Since completing the two cohorts described in this report, additional ECHO programmes have been made available to this and other Nevada Medicaid MCOs, further demonstrating the value of ECHO to these organizations. Further, a multi-site study of how the ECHO model has been implemented across the USA, funded by the Robert Wood Johnson Foundation, identified several other states that have used partnerships with Medicaid MCOs to make ECHO programming sustainable [42].

Although we used a Medicaid MCO as a funding source, other insurance sources, in addition to grants, public/private partnerships, healthcare system support, fee-for-service models, and direct government funding, are also all funding models used to support Project ECHO in the USA as well as in other countries [43]. Additionally, the programmatic details shared here may be applicable in countries where overprescribing of opioids is a challenge. The ECHO model is used in 207 countries and areas [44].

This evaluation also adds to the literature supporting the implementation of ECHO through a cohort model [45–47]. The cohort model promotes a learning community that, through regular attendance, learns together and from each other over a pre-determined number of sessions. We acknowledge, however, that limiting participation to 10–12 cohort participants may reduce the programme’s reach to clinicians who can commit to attending the entire programme. A ‘drop-in’ programme might be more accessible to a broader range of individuals who could pick and choose. Research quantifying the trade-offs between the cohort and drop-in model could offer additional guidance on this issue.

An additional operational element of the PEN Pain ECHO was a financial incentive granted to participants for weekly attendance, programme completion, and case presentations. The $500 incentives were intended to cover potential income lost by not seeing patients during the ECHO session or during case presentation preparation. However, the incentives may have made the ECHO more attractive to comparatively lower-wage or salaried clinicians than clinicians with higher wages or fees. This, in turn, may have implications for the generalizability of this evaluation; similar ECHO programme replications may see similar results among clinicians for whom the incentives provide the largest percent increase in remuneration.

Conclusion

The modest change in clinician opioid prescribing behaviour observed among ECHO clinicians participating in a Medicaid managed care network was consistent with increases in objectively measured knowledge scores and self-reported self-efficacy. The changes in prescribing behaviour were particularly notable given the comparatively shorter duration of the intervention compared with other Pain Management ECHO reported in the literature.

Supplementary data

Supplementary data are available at HEAL online.

Conflict of interest:

All authors worked full or part time for Project ECHO Nevada while this study was conducted. Denis Patterson receives grants or contracts, consulting fees or honoraria from Abbott, AIS, Allergan, Aurora Spine, CornerLoc, OrthoFundamentals, Nevro, PainTEQ, SI Bone, and Stryker.

Funding

None declared.

Data availability

The proprietary Medicaid managed care claims data used for this research are subject to a data use agreement that does not allow for data sharing.