Care Redesign

Tackling the Opioid Crisis with Clear Prescription Guidelines, Accurate Monitoring, and Provider Education

Case Study · May 23, 2018

More than 1,000 people in New York City died from an overdose involving an opioid in 2016, the most recent year available and most deadly year for these types of deaths on record. More New Yorkers died from opioid overdoses last year than from car accidents and homicides combined. At NewYork-Presbyterian Hospital, emergency department visits for opioid use disorder more than doubled between 2015 and 2017.

For many patients, opioid addiction can be traced to prescriptions for the treatment of acute and chronic pain. Evidence shows that patients filling high-dosage opioid prescriptions have much higher rates of opioid use disorder than those filling low-dose prescriptions. Therefore, one promising approach for addressing the opioid crisis lies in identifying high-risk prescription patterns that may lead to addiction.

At NewYork-Presbyterian Medical Groups, we developed straightforward clinical guidelines and reference tools, analyzed and benchmarked opioid prescriptions, and provided feedback to clinicians about their prescribing patterns. Results show improved adherence to clinical standards, a reduction in opioid prescription dosages, and fewer high-risk prescription combinations.

Key Takeaways

  1. Identify an essential set of best practices that clinicians can adopt and perform consistently and reliably to have the greatest impact on outcomes.

  2. Distribute best practice guidelines widely and discuss at a variety of forums to focus the attention of busy clinicians.

  3. Identify key metrics that can be measured over time and communicate findings to leadership as well as to providers.

  4. Identify providers requiring additional education and attention in a systematic way, controlling for confounding factors that would otherwise misidentify outliers.

  5. Provide constructive feedback to providers confidentially and take time to review charts to understand clinical context and opportunities for improvement.

The Challenge

NewYork-Presbyterian Medical Groups is a multi-specialty physician organization in the New York City metropolitan area. The Medical Groups have offices in Hudson Valley, Westchester, Queens, and Brooklyn and employ over 800 physicians and 300 additional licensed professionals. The size and geographic distribution of the medical groups make it challenging to establish uniform clinical standards and best practices.

The implementation of electronic health records across the organization has helped with standardization. However, the sheer volume of patient data along with numerous opioid prescription recommendations by the state, the Centers for Disease Control and Prevention (CDC), specialty societies, and other stakeholders was initially overwhelming.

The Goal

There were three main goals in standardizing opioid prescription best practices:

  • To establish straightforward clinical standards and best practices for opioid prescriptions in the ambulatory setting. We sought to distill the numerous opioid recommendations into five to seven key best practices that could be easily communicated across a large, diverse physician organization. We believed that adhering to these guidelines consistently would reduce high-risk controlled substance prescriptions and improve patient outcomes.
  • To develop a decision aid to help clinicians identify potentially high-risk prescriptions at a glance.
  • To implement methodologies to monitor compliance and identify high-risk prescribers across specialties. With thousands of electronic prescriptions submitted daily, identifying providers needing educational interventions had been challenging, as patient panel size and provider specialty were key factors affecting the rate of opioid prescriptions. Controlling for these two variables was essential to targeting the efforts.

The Execution

Evidence-based guidelines and best practices were reviewed and distilled into seven key recommendations. Clinical leaders across the medical groups provided input, including regional executive medical directors and physician champions. The concept of “morphine milligram equivalents” (MME) emerged as the backbone of the guidelines. CDC recommendations identify prescription strengths of 50 and 90 MME/day as key inflection points in decision-making. Prescriptions of more than 50 MME/day should be carefully considered, while those over 90 MME/day should be limited in most cases. On this basis, prescription strengths were defined as Level 1, 2, or 3 to aid in evaluating risk.

Although additional recommendations were considered, such as encouraging naloxone prescriptions to mitigate overdose risk and screening and treating patients for opioid use disorder, the initial seven recommendations were prioritized in the first phase of the guidelines rollout.

Opioid Prescription Best Practices

The following recommendations were distributed to all providers and discussed at physician meetings and forums to promote adoption and buy-in across the medical groups:

1.  In general, do not prescribe opioids as first-line treatment for chronic pain (excluding active cancer, palliative care, or end-of-life care).

Before prescribing opioids, consider non-opioid therapies such as NSAIDs, tricyclics, SNRIs, topical agents, exercise, physical therapy, or cognitive behavioral therapy. If opioids are required, they should be combined with non-opioid therapy, as appropriate.

2.  For Schedule II-IV controlled substances, the New York State Prescription Monitoring Program (PMP) website must be reviewed.

Providers must review the patient’s prescription history for evidence of similar medications prescribed by other providers during the past 6 months.

For the purposes of these guidelines, all opioid prescriptions will be evaluated as morphine milligram equivalents (MME) per day.

The opioid strength of each medication unit (e.g. tablet, milliliter, patch) will be converted to the equivalent dosage in milligrams of morphine, which will be multiplied by the number of units dispensed and divided by the duration of the prescription (e.g., 30 days).

Opioid prescription dosages will be classified as Level 1, 2, and 3 as follows:

  • Level 1: 0–49 morphine milligram equivalents/day
  • Level 2: 50–89 MME/day
  • Level 3: Over 90 MME/day

3.  Patients receiving Level 2 and 3 opioid prescriptions should receive in-person follow-up every 3 months. Patients without an upcoming appointment should be scheduled.

4.  Level 3 prescriptions should be avoided when possible. Consider pain management consultation for patients requiring Level 3 opioid prescriptions.

5.  Patients on chronic opioid therapy (> 3 months) should be encouraged to review and sign a controlled substances agreement. The agreement details the responsibilities of the patient and prescriber with respect to controlled substances and is signed by both parties. The agreement establishes the need for a single prescriber of opioids for the patient, regular follow up, and limits on early refills. The patient also agrees not to sell or transfer the medications and to store the medication securely.

6.  Concurrent prescription of opioids and benzodiazepines should be avoided whenever possible. Both drug classes are central nervous system depressants, and combined usage greatly increases the risk of respiratory collapse, hospitalization, and death.

7.  Clinical rationale for each controlled substance prescription must be documented clearly. Each prescription must be accurately linked to the appropriate diagnosis in the clinical note, patient case, or order group.

The best practice guidelines were discussed at numerous town hall meetings and also emailed to providers. The guidelines were well-received by nearly all providers. Several outlier providers, however, believed their prescription patterns were not unusual and were slow to modify their practices until data was reviewed with them along with an assessment of patient charts.

A decision aid for clinicians was designed to allow easy calculation of morphine milligram equivalents for commonly prescribed opioid formulations. Offices were encouraged to print and distribute this Opioid Strength Conversion Table for reference.

Opioid Prescriptions Strength Conversion Table

  Click To Enlarge.


Thanks to the opioid prescription best practices, providers now had clear expectations and a framework on which to base their decisions.

The electronic health record was leveraged to monitor compliance and outcomes and to identify providers requiring additional attention. A simple tally of opioid prescriptions by provider was not an effective metric to identify outliers, as patient volumes, dosage, and prescription duration varied considerably among clinicians. Normalizing for panel size, the total morphine equivalents prescribed per patient managed was calculated. Values tended to cluster within each clinical specialty, and this information was used to establish opioid outlier thresholds.

The figure below illustrates the specialty variation in opioid prescribing observed within the medical group. For example, hematologist/oncologists prescribed on average 330 MME per patient managed during the time period. For those prescribing more than one standard deviation above the specialty mean — in this case more than 720 MME per patient — selected chart reviews were conducted. If significant variation from best practice was identified, feedback and targeted education was provided to the physician. Interventions included individualized feedback to review best practice guidelines and discuss specific patient charts as needed. Nearly all providers receiving feedback and education were able to improve adherence to the guidelines. If improvement was not observed despite educational efforts, the issue was escalated to department chairs for further coaching and evaluation of the provider’s clinical practice.

Opioid Prescriptions Total MME Prescribed Per Patient Managed - Select Specialties

  Click To Enlarge.


Between January and September 2017, we observed a 24% decrease in total morphine equivalents prescribed per patient managed across all specialties. During the same time period, the number of referrals for pain management consultation nearly doubled.

Opioid Prescriptions Total MME Prescribed Per 100 Patients Managed Per Month

  Click To Enlarge.

Opioid Prescriptions Number of Pain Management Referrals Requested Per Month

  Click To Enlarge.

Concomitant benzodiazepine and opioid prescriptions were also identified, and providers were given feedback on the numbers of patients receiving these prescriptions. There was a trend toward fewer concomitant opioid-benzo prescriptions during the time period. The initiative has brought greater clarity to controlled substance prescriptions and management of chronic pain across a large physician organization.

Concomitant Benzodiazephine and Opioid Prescriptions per 1000 Patients Managed

  Click To Enlarge.

Where to Start

Organizations seeking to implement clinical standards and best practices should take a systematic approach to increase the likelihood of success. Once a focus area has been selected, a thorough review of the clinical data must be undertaken to identify areas needing improvement and standardization.

Clinical champions should be convened early in the process to begin writing best practice guidelines, targeting areas that have the greatest potential to impact outcomes. Organize clinical decision points into categories for easy recall, and provide decision aids and charts for maximum adoption. Best practice guidelines should be discussed at a variety of forums and continually reinforced. Monitor outcomes regularly with key metrics, and provide performance data to leadership as well as to individual clinicians. Systematically identify providers needing improvement based on review of the data, providing individualized educational interventions to encourage change over time.


The opioid crisis galvanized the team to think creatively about how to implement best practice guidelines, feedback, and monitoring to change provider behavior that had previously seemed intractable. Focusing physicians’ attention on what we believed were the most essential elements in the ambulatory setting proved to be highly successful. We are confident that our decision tool and analytics methodology for opioid prescriptions can be adopted by other health systems, as we all seek to collectively turn the tide on the opioid epidemic.


This publication reflects work conducted when Dr. Dalal was Medical Director for Physician Services at NewYork-Presbyterian and an Assistant Professor at Weill Cornell Medical College.

Acknowledgments: The author thanks Dr. Steven J. Corwin, MD, for his suggestion and support to write about the medical group opioid monitoring program. The author also acknowledges Joseph Cooke, MD, Stephen Rimar, MD, Parag Mehta, MD, Steven Silber, DO, Gerald Ridge, MD, James Trapasso, DO, William Higgins, MD, Michael Nochomovitz, MD, and David Alge, MBA.

New Call for Submissions ­to NEJM Catalyst


A weekly email newsletter featuring the latest actionable ideas and practical innovations from NEJM Catalyst.

Learn More »

More From Care Redesign
Consensus That Design Thinking Is Useful in Health Care

Survey Snapshot: Design Thinking Is Useful, So Why Aren’t More People Using It?

NEJM Catalyst Insights Council members agree that design thinking useful, but leadership buy-in and understanding of how to implement it may create barriers.

Artificial Intelligence AI Technologies - 5 Approaches to Augmentation of Health Care Decision-Making

Artificial Intelligence and the Augmentation of Health Care Decision-Making

Artificial intelligence is most likely to improve health care by augmenting the work of human clinicians.

Improving Scores Through Human-Centered Design

Designing and Implementing Better Patient Experiences

An ophthalmology provider with facilities in the greater Mexico City area is using Human-Centered Design to improve both the patient and staff experience.


Designing Hospitals That Heal as Well as Treat

Hospitals must provide wholesome food, sound sleeping conditions, and human connection to promote healing and wellness.

Health Care Organizations Only Occassionally Apply Design Thinking

Care Redesign Survey: How Design Thinking Can Transform Health Care

Health care leaders and frontline clinicians are eager to embrace design thinking. Yet its principles are not widely applied.

Chernew04_pullquote targeted supplemental data collection

Targeted Supplemental Data Collection — Addressing the Quality-Measurement Conundrum

Targeted supplemental data collection may be a valuable approach to balancing data needs with data-collection costs.

David Waters Community Servings CEO

Food as Medicine: Meeting the Needs of Complex Medical Diets

How Community Servings evolved from an HIV nutrition program to one that feeds people with any illness.

HaH-Plus Program Reduces Patient Days Spent in Health Care Facilities in Final 180 Days of Life

Hospital at Home–Plus Reduces Days Spent in Hospitals and Other Inpatient Facilities

Patients enrolled in Mount Sinai Health System’s HaH-Plus program spend about one-third as many days in health care facilities as inpatients in the last 180 days of life.

SLUCare General Internal Medicine Interactive Voice Response - Patient Access

Creating a Holistic Approach to Patient Access

To get beyond simplistic measures such as increased patient appointments or decreased lag days, operations administrators and physician champions must develop effective partnerships.

Bridgeport Hospital Senior Staff Safety Rounds - Improving Serious Safety Event Rate

Senior Staff Safety Rounds: A Commitment to Ensure Safety Is the Top Priority

Bridgeport Hospital leaders find that face-to-face feedback from physicians and staff leads to real-time improvements that have positively impacted safety.


A weekly email newsletter featuring the latest actionable ideas and practical innovations from NEJM Catalyst.

Learn More »


Moving from a Sickness Model of…

A public health expert discusses the scientific, technological, and financial foundations we need to modernize…

Quality Management

139 Articles

Better Physician Ratings from Discussing PROs…

A UPMC study demonstrates the value of PROs, but finds that physicians need to discuss…

Coordinated Care

113 Articles

Improving Patient Flow with Dynamic Work…

Staff make big improvements in post-operative care by changing small details.

Insights Council

Have a voice. Join other health care leaders effecting change, shaping tomorrow.

Apply Now