Development and Alpha Testing of Specifications for a Prolonged Opioid Prescribing Electronic Clinical Quality Measure (eCQM)

Abstract

Brigham and Women’s Hospital (BWH) has received funding from the Centers for Medicare and Medicaid Services (CMS) to design and implement an electronic clinical quality measure (eCQM) assessing the rate of prolonged opioid prescribing practices following Total Hip Arthroplasty (THA) and Total Knee Arthroplasty (TKA). Utilizing an existing guideline, ‘prolonged prescribing’ has been defined as opioid prescriptions that exceed 42 days (6 weeks) following surgery. This measure was tested on 12,803 Partners’ Healthcare (PHS) patients. Findings demonstrated that after 42 days, meeting the criteria for ‘prolonged prescribing’ as defined by the proposed measure, 3.7% of THA patients and 12.1% of TKA patients were still receiving opioids. With a better understanding of how specific clinician group post-operative prescribing practices compare with their peers and incorporating monetary incentives through the MIPS participation pathway of the Quality Payment Program (QPP), this measure will motivate orthopedic practices to improve their prescribing patterns, ultimately driving evidence-based quality improvement.

Introduction

The United States is in the midst of a profound opioid crisis. Since 1999, opioid prescribing has quadrupled, and opioid overdoses have increased six-fold¹. On average, 130 Americans die every day due to an opioid overdose and an estimated 40% of those deaths involve a prescription opioid². In response to the opioid epidemic, a public health emergency was declared in 2017 by the United States Department of Health and Human Services².

Developing opioid dependence is a concern for previously opioid naïve THA and TKA patients. A national database study examined 30,938 patients undergoing TKA and 13,744 patients undergoing THA who had not taken opioids in the 6 months prior to surgery. After one year, 10% of THA patients and 13% of TKA patients were still receiving opioids³. This demonstrates that a significant proportion of patients, previously naïve to opioids, developed dependence in part due to their Total Joint Arthroplasty (TJA).

Orthopedic surgery is the third highest opioid-consuming specialty⁴, with THA and TKA resulting in large quantities of opioids prescribed post-operatively relative to other common orthopedic surgeries⁴. This is partially due to the fact that these are common, painful procedures, and that existing guidelines tend to be inconsistent regarding the recommended duration of opioid prescriptions to manage post-operative pain^5,6. Therefore, surgeons are challenged with balancing the risk of under-treating pain or placing their patients at a greater risk of developing opioid dependence or opioid use disorder. As a result, post-operative prescriptions vary widely in both their duration and the total number of pills prescribed⁴, with the evidence suggesting that orthopedic surgeons tend to overprescribe opioids following surgery^7,8,9,10.

Both THA and TKA procedures are expected to see substantial increases in demand over the coming decade. If the number of primary TKAs performed continues to increase at the current rate, demand is expected to reach approximately 3.48 million procedures by 2030, up 673% since 2005¹¹. Demand for THA is estimated to grow by 174%, reaching 572,000 procedures by 2030¹¹. The expected rapid increase in the frequency of these procedures, and the fact that surgeons tend to overprescribe opioids after surgery, highlights the need for continuous quality improvement in this field.

Therefore, the objective of this research was to develop draft specifications for a new electronic clinical quality measure (eCQM) that will assess the percentage of patients that are prescribed opioids for a prolonged duration following elective primary THA or TKA. The proposed measure is a potential candidate for the Merit-based Incentive Payment System (MIPS). The MIPS pathway provides an opportunity for clinicians to receive performance-based pay based on the quality of care provided to their patients. This manuscript summarizes the steps that were taken to specify, develop, and alpha test the measure.

Methods Environmental scan

Structured review of the published literature and clinical guidelines

In collaboration with the Countway Harvard Librarian, the BWH research team conducted a literature review to inform the initial specifications and inclusion/exclusion criteria for this eCQM. A focus of the review was to understand how opioids are prescribed to various patient populations to determine which patients would be included in the measure. A number of searches related to the prolonged use of opioids following TJA were executed using PubMed, and the records were exported to Covidence, a systematic review management software¹². Titles and abstracts were screened for relevance, and subsequently evaluated by full-text review by a single reviewer. Articles were considered relevant if they met the following inclusion criteria: 1) related to post-operative opioid prescribing practices, 2) focused on THA or TKA populations, and 3) were published in the last 5 years. Articles were excluded if: 1) the study went beyond the scope of the eCQM (e.g., child or youth populations), 2) the full text was not accessible, or 3) the paper was not published in English. Based on the articles that met the inclusion criteria, similar papers were located in PubMed using the “similar papers” filter until no new papers were returned.

Review of the clinical guidelines

A similar search was conducted to capture relevant clinical guidelines related to post-operative opioid prescribing practices. The guidelines were used to inform selection of an appropriate definition for ‘prolonged use’ of opioids following surgery. Guidelines were examined if they were published in the last 5 years and were intended to guide management of acute to chronic post-surgical pain.

Review of related clinical quality measures

Using the CMS Measure Inventory Tool (CMIT) and Quality Positioning System (QPS), existing measures as well as those currently under development relating to opioid prescribing practices or dispensation patterns were examined. Additionally, a member of our Technical Expert Panel (TEP) provided information about opioid prescribing measures being developed by the Pharmacy Quality Alliance (PQA). Measures were reviewed to determine if they were “related” or “competing.” National Quality Forum (NQF) defines related measures as having “either the same measure focus or the same target population,” and competing measures as having “both the same measure focus and the same target population”¹³. This was done to harmonize specifications where appropriate, and to ensure that the proposed eCQM was substantially different from other measures that currently exist or are under development.

Key stakeholder engagement

Patient, provider and payer interviews

As part of the measure development process, BWH partnered with Massachusetts Health Quality Partners (MHQP) to engage key stakeholders using structured interviews to inform the design and development of the eCQM. In order to be eligible to participate in the interview, the following criteria had to be met for each type of stakeholder:

• Patients: English-speaking adult patients who had THA and/or TKA at Brigham and Women’s Physicians Organization (BWH and Brigham and Women’s Faulkner Hospital) between January 1, 2017 and October 9, 2019. The BWH research team identified patients meeting these criteria and shared only their names, mailing addresses, and phone numbers with MHQP. MHQP mailed invitation letters to these patients, and interested patients contacted MHQP’s Project Manager to participate.

• Providers: Orthopedic clinicians including surgeons, physician assistants, nurse practitioners, physical therapists, occupational therapists, and medical assistants who practiced at a PHS hospital or affiliated practice were eligible to participate. Dr. David Bates, the study principal investigator, sent emails to eligible orthopedic providers. Interested clinicians contacted MHQP’s Project Manager to schedule an interview.

• Payers: Payers from Massachusetts health plans that cover orthopedic surgery were eligible to participate. MHQP contacted members of its Health Plan Council via email informing them about the project. Payers responded to MHQP’s Project Manager if they were interested in participating.

Technical Expert Panel engagement

Additional stakeholder input was provided during a TEP meeting held on January 15, 2020. BWH sought stakeholder feedback regarding the initial specifications, appropriateness of the inclusion/exclusion criteria, preliminary testing, and validation that the proposed eCQM was substantially different from the other measures that currently exist or are under development.

Alpha testing

Based on the environmental scan and stakeholder engagement, the eCQM specifications were established, and alpha testing was conducted using PHS data. Records for the target population of patients, aged 18 or older, undergoing an elective primary THA and/or TKA from January 1, 2016 to December 31, 2018 were extracted from the PHS Electronic Health Record (EHR) system, EPIC, using Clarity, EPIC’s database. Information documented in the EHR (e.g., health and medication history) was used to determine which patients were excluded from the measure denominator. The number of patients and reasons for exclusion were calculated (Table 3).

Table 3:

eCQM exclusion patients

Initial Population = 12,803
Exclusion	Number of patients
Patient received a non-elective THA/TKA	n = 953
Patient was prescribed opioids within 90 days prior to the index admission or received a diagnosis for opioid use disorder 365 days prior to the index admission	n = 1,408
Patient received a separate THA- or TKA-related procedure within the 90 days prior to the index admission or 90 days after surgical discharge	n = 355
Patient received additional general or major surgery within 90 days of surgical discharge	n = 126
Patient had a cancer diagnosis within the 365 days prior to the index admission or 90 days following surgical discharge	n = 1,352
Patient received hospice or palliative care within the 365 days prior to the index admission or 90 days following surgical discharge	n = 12
Patient had a diagnosis of sickle cell disease within the 365 days prior to the index admission or 90 days following surgical discharge	n = 6
Patient was discharged Against Medical Advice	n = 0
Final Denominator Population = 8,591

Descriptive statistics were calculated to describe the type of surgery received, sociodemographic characteristics, and receipt of post-operative opioids by the patients that met the inclusion criteria. Prescribing data, including the type of opioid and the days’ supply of each post-operative opioid prescription, were extracted to calculate the duration of each patient’s post-operative opioid use. The duration of post-operative opioid use was calculated from the day of discharge to the end date of the last prescription given to the patient within the 90 days following surgery. The proportion of patients who were receiving opioid prescriptions was tracked for 90 days after surgical discharge. The percentage of patients who met the criteria for ‘prolonged prescribing’ was reported, stratified by type of surgery (i.e., THA or TKA).

Results

Environmental scan

Structured review of the published literature and clinical guidelines

The literature review returned 309 records, and 28 articles met the inclusion criteria. The literature was synthesized to inform the inclusion and exclusion criteria for the measure as well as to provide the BWH research team with an overview of post-operative opioid prescribing trends. A summary of the literature supporting the exclusion criteria is presented in Table 2.

Table 2:

Summary of exclusion criteria & exclusion source

Exclusion	Source
Patient received a non-elective THA/TKA	NQF 1550: Hospital-Level Risk-Standardized Complication Rate Following Elective Primary THA and/or TKA17
Patient was prescribed opioids within 90 days prior to the index admission or received a diagnosis for opioid use disorder 365 days prior to the index admission	Literature review6,9,10
Patient received a separate THA- or TKA-related procedure within the 90 days prior to the index admission or 90 days after surgical discharge	NQF 1550: Hospital-Level Risk-Standardized Complication Rate Following Elective Primary THA and/or TKA17
Patient received additional general or major surgery within 90 days of surgical discharge	Internal chart reviews
Patient had a cancer diagnosis within the 365 days prior to the index admission or 90 days following surgical discharge	NQF 2940: Use of Opioids at High Dosages in Persons Without Cancer15
Patient received hospice or palliative care within the 365 days prior to the index admission or 90 days following surgical discharge	NQF 2940: Use of Opioids at High Dosages in Persons Without Cancer15
Patient had a diagnosis of sickle cell disease within the 365 days prior to the index admission or 90 days following surgical discharge	Literature review16
Patient was discharged Against Medical Advice	NQF 1550: Hospital-Level Risk-Standardized Complication Rate Following Elective Primary THA and/or TKA17

Review of the clinical guidelines

After reviewing nine existing post-operative opioid prescribing guidelines from a variety of developers, the Washington State Agency Medical Directors’ Group (2018) guideline entitled: Interagency Guideline on Prescribing Opioids for Pain was selected to inform measure development. The guideline was developed in collaboration with a broad group of academic leaders, pain experts, and clinicians using an evidence-based approach. This guideline is current and has demonstrated reliability and effectiveness¹⁴. Release of the first edition (2007) and the second edition (2010) contributed to a 29% decrease in the rate of prescription opioid-related deaths between 2008 and 2013 in the state of Washington14. The guideline suggests prescribing opioids for ≤ 14 days (i.e., 2 weeks) following THA or TKA for uncomplicated cases. However, in exceptional cases, where further prescriptions may be needed, it is recommended that opioid use be tapered off by 42 days (i.e., 6 weeks) following surgery¹⁴. An exploratory analysis was performed using both recommended thresholds (i.e., 14 days and 42 days) for prolonged prescibing to calculate the respective rates. These results were presented to the TEP for discussion and feedback.

Review of related clinical quality measures

Scans of CMIT, QPS, and PQA showed there is one related measure under development, Potential Opioid Overuse, which examines the percentage of patients who receive opioid prescriptions for 90 days or longer with an average daily dosage exceeding 90 morphine milligram equivalents for management of chronic non-cancer pain. The eCQM proposed by BWH focuses on opioid-naïve patients (i.e., patients who were not prescribed any opioids in the 90 days before surgery) undergoing an elective primary THA and/or TKA, with a focus on acute to chronic post-surgical pain.

Another related measure from PQA, identified by a TEP member, Initial Opioid Prescribing for Long Duration, examines the days’ supply of initial opioid prescriptions. This measure uses claims data and includes all patients starting opioid therapy. The eCQM proposed by BWH focuses on a subset of patients undergoing elective primary THA and/or TKA. Additionally, the PQA measure does not consider refill prescriptions; the proposed eCQM does.

Harmonization with the measures Hospital-level Risk-standardized Complication Rate following THA/TKA and Use of Opioids at High Dosage in Persons Without Cancer was used to inform exclusion criteria (Table 2), as well as to determine which opioids would be included in the eCQM (Table 1). These claims-based measures were selected for harmonization because of their similar target population or measure focus.

Table 1:

Opioids used to calculate duration of use

Opioid
Butorphanol	Hydrocodone	Methadone	Oxymorphone
Codeine	Hydromorphone	Morphine	Pentazocine
Dihydrocodeine	Levorphanol	Opium	Tapentadol
Fentanyl	Meperidine	Oxycodone	Tramadol

Key stakeholder engagement

Patient, provider and payer interviews

A total of 18 THA and/or TKA patients, 10 healthcare providers, and 5 payers were interviewed about the measure concept and its development through the partnership with MHQP. The majority of patients, providers, and payers supported development of this eCQM, stating that it would be effective in assessing the overprescription of opioids following THA or TKA and could result in meaningful improvements to the care process. The main concern raised by patients and providers was that such a measure may result in the under-prescription of opioids when their use is warranted. This concern has been considered throughout the development process by applying appropriate inclusion/exclusion criteria. Additionally, this concern was brought to the TEP and considered when deciding whether to use the 14-day or 42-day threshold promoted by the Washington State guideline.

Technical Expert Panel engagement

The environmental scan, stakeholder engagement, and the results from preliminary testing were presented to the TEP on January 15, 2020. The TEP supported all of the inclusion/exclusion criteria as well as the use of Washington State’s guidelines. They noted that no clear established definition for ‘prolonged prescriptions’ exists, but that the thresholds from the Washington State guideline are well regarded and serve as an effective benchmark for assessing prescribing practices. In considering all the research team’s findings, the TEP recommended using the 42-day threshold to calculate the rate for this measure. Therefore, the following specifications have been proposed:

Denominator: All patients, aged 18 years or older, who received an elective primary THA or TKA procedure and do not meet any exclusion criteria.

Numerator: The subset of patients from the denominator who were prescribed post-operative opioids for > 42 days after surgical discharge following an elective primary THA or TKA.

Alpha testing

The target population consisted of 12,803 patients that underwent a THA and/or TKA procedure from January 2016 to December 2018 at PHS. Of these patients, 8,591 met the inclusion criteria (Table 3) and were included for alpha testing.

Of the 8,591 patients (3,533 THA; 5,058 TKA), nearly all (99.1%) received an opioid prescription at surgical discharge. The majority of patients who met the inclusion criteria were white (90.4%), female (57.5%), and English speaking (95.0%). There was a significant variation in age, ranging from 18 – 96 years, with a median age of 67 years (Table 4).

Table 4:

Characteristics of patients who met the inclusion criteria

n = 8,591
Characteristic	Percentage
Procedure
THA	41.9%
TKA	58.1%
Age
≥ 65 years	54.7%
< 65 years	45.3%
Primary Language
English	95.0%
English as a second language	5.0%
Sex
Male	42.5%
Female	57.5%
Race
White	90.4%
Black of African American	3.54%
Asian	1.40%
Hispanic or Latino	0.68%
Received an Opioid Prescription at Surgical Discharge
Yes	99.1%
No	0.9%

On average, patients who received a THA were prescribed opioids for shorter durations, indicating more rapid tapering following surgery compared to TKA patients. After 2 weeks, 20.8% and 47.1% of THA and TKA patients were still receiving opioids, respectively. After 42 days, meeting the criteria for “prolonged prescribing” as defined by the proposed measure, 3.7% of THA patients and 12.1% of TKA patients were still being prescribed opioids (Figure 1). Characteristics of the prescribing duration for various cohorts are presented in Table 5.

Figure 1:

Percentage of patients who were receiving opioid prescriptions over the 90 days after surgical discharge, stratified by THA and TKA

Table 5:

Patient characteristics over the course of opioid episode

Characteristic	Total (n)	0-14 Days	>14 & ≤ 42 Days	> 42 Days
Procedure
Hip	3533	79.16%	17.13%	3.71%
Knee	5058	52.94%	35.01%	12.05%
Sex
Male	3667	66.62%	25.72%	7.66%
Female	4924	61.54%	29.20%	9.26%
BMI
≥ 30	4125	59.68%	30.16%	10.16%
< 30	4466	67.47%	25.42%	7.11%
Primary Language
English	8158	63.95%	27.75%	8.30%
Non-English	433	58.16%	27.66%	14.18%
Race
White	7735	64.38%	27.58%	8.04%
Black of African American	308	52.92%	30.52%	16.56%
Asian	129	58.91%	31.01%	10.08%
Hispanic or Latino	56	66.07%	17.86%	16.07%
Age
≥ 65 years	4831	64.54%	27.92%	7.53%
< 65 years	3760	62.63%	27.45%	9.92%

The BWH research team then examined the rates at the clinician group levels across the 6 PHS sites. There is notably variability in the rates, ranging from 6.25%, to 14.85%. Such variability emphasizes the importance of the proposed measure, which will allow clinicians to compare their prescribing practices to that of their peers in order to drive evidence-based care. A statistical risk adjustment was then performed to account for patient-specific risk factors (e.g. BMI, race, language, etc). The results from this risk adjustment are presented in Table 6 below. Again, even with the risk adjustment applied, rates vary notably between sites, ranging from 7.43% to 12.46%. Across PHS, the overall risk-adjusted prolonged opioid prescribing rate is 8.58% for THA and TKA patients.

Table 6:

PHS overall risk-adjusted and unadjusted prolonged opioid prescribing rate at clinician group level

Clinician Group (Blinded)	Unadjusted Rate	Adjusted Rate
A	9.81%	7.43%
B	6.25%	7.88%
C	7.38%	7.97%
D	7.14%	8.32%
E	14.58%	9.24%
F	6.85%	12.46%
Overall Adjusted Rate: 8.58%

Discussion and Implications

This study reports on the alpha testing of a novel eCQM. The specifications were developed through an environmental scan and engagement of key stakeholders. The eCQM measures the percentage of patients who were prescribed opioids beyond 42 days after surgery and therefore meet the criteria for “prolonged prescribing.” Based on 2016-2018 PHS data, 3.7% of THA and 12.1% of TKA patients were prescribed opioids for more than 6 weeks, producing an overall risk-adjusted rate of 8.58%, demonstrating that this eCQM is meaningful for PHS patients and providers. By developing an eCQM that measures potentially inappropriate postsurgical prescribing, orthopedic practices will be better equipped to assess the quality of care provided to their patients against a standardized benchmark. Given that THA and TKA are expected to increase in frequency in the future¹¹, coupled with evidence that orthopedic surgeons tend to overprescribe opioids following surgery^7,8,9,10, it is imperative that clinicians have the tools to improve their practice.

The proposed eCQM will enhance the MIPS pathway by providing clinicians with a less burdensome method to compute their performance score that minimizes provider time spent collecting and submitting data to CMS. With a better understanding of how specific clinician group post-operative prescribing practices compare with their peers and incorporating monetary incentives through the MIPS participation pathway of the QPP, orthopedic practices will be motivated to improve their prescribing practices, ultimately driving quality improvement.

Limitations

This measure has only been tested on PHS patients. Further testing outside of PHS is needed to ensure measure generalizability and to determine if there is variation in quality throughout the United States. Additionally, stakeholder interviews suggest that BWH will need to assess the impact of including patients who receive care at multiple sites, which could bias results since these patients would not have all their health and medication information documented in a single EHR system.

Next steps

BWH has partnered with Cerner to conduct testing at a geographically different hospital system outside of PHS using a different EHR system. This testing will provide valuable information about how the measure performs on a different patient cohort and will provide insight into any potential new challenges.

In order to address and better understand the impact of including patients who receive care at multiple sites, the BWH research team is testing the eCQM against the full PHS database as well as against a subset of the population living within a 25-mile radius from each site. Here, the assumption is that those who live within 25 miles of the site will receive all their care from a PHS-affiliated hospital or clinic, providing the research team with a more complete medical history. An analysis comparing the results between the two populations will be conducted. Cerner will also be testing the eCQM against their full database and against patients within a 25-mile radius of specified sites. In performing these tests, BWH hopes to develop a better understanding of how to approach the cohort of patients receiving care through multiple EHR vendors.

BWH will continue to review the literature, meet with stakeholders, and report for public comment on this eCQM in order to receive further feedback and refine the measure wherever necessary. The overarching goal of this measure development process is to obtain NQF endorsement on this eCQM and implement the measure in CMS’s QPP to meaningfully impact quality of care.