Postoperative Restrictive Opioid Protocols and Durable Changes in Opioid Prescribing and Chronic Opioid Use

Key Points

Question  Can postoperative pain be managed appropriately after hospital discharge with 3 or fewer days of an opioid supply?

Findings  In this cohort study of 4068 surgical patients in a large tertiary cancer center, implementation of a restrictive opioid prescription protocol specifying an opioid supply of 3 or fewer days was feasible, and no compromises in patient recovery were detected. The change led to a significant reduction in opioids dispensed postoperatively and was associated with significantly decreased conversion to chronic opioid use in postsurgical patients.

Meaning  The findings suggest that prescription of an opioid supply of 3 or fewer days is feasible for most postsurgical patients and may be associated with decreased chronic opioid use.

Importance  Changes in postsurgical opioid prescribing practices may help reduce chronic opioid use in surgical patients.

Objective  To investigate whether postsurgical acute pain across different surgical subspecialties can be managed effectively after hospital discharge with an opioid supply of 3 or fewer days and whether this reduction in prescribed opioids is associated with reduced new, persistent opioid use.

Design, Setting, and Participants  In this prospective cohort study with a case-control design, a restrictive opioid prescription protocol (ROPP) specifying an opioid supply of 3 or fewer days after discharge from surgery along with standardized patient education was implemented across all surgical services at a tertiary-care comprehensive cancer center. Participants were all patients who underwent surgery from August 1, 2018, to July 31, 2019.

Main Outcomes and Measures  Main outcomes were the rate of compliance with the ROPP in each surgical service, the mean number of prescription days and refill requests, type of opioid prescribed, and rate of conversion to chronic opioid use determined via a state-run opioid prescription program. Postsurgical complications were also measured.

Results  A total of 4068 patients (mean [SD] age, 61.0 [13.8] years; 2528 women [62.1%]) were included, with 2017 in the pre-ROPP group (August 1, 2018, to January 31, 2019) and 2051 in the post-ROPP group (February 1, 2019, to July 31, 2019). The rate of compliance with the protocol was 95%. After implementation of the ROPP, mean opioid prescription days decreased from a mean (SD) of 3.9 (4.5) days in the pre-ROPP group to 1.9 (3.6) days in the post-ROPP group (P < .001). The ROPP implementation led to a 45% decrease in prescribed opioids after surgery (mean [SD], 157.22 [338.06] mean morphine milligram equivalents [MME] before ROPP vs 83.54 [395.70] MME after ROPP; P < .001). Patients in the post-ROPP cohort requested fewer refills (367 of 2051 [17.9%] vs 422 of 2017 [20.9%] in the pre-ROPP cohort; P = .02). There was no statistically significant difference in surgical complications. The conversion rate to chronic opioid use decreased following ROPP implementation among both opioid-naive patients with cancer (11.3% [143 of 1267] to 4.5% [118 of 2645]; P < .001) and those without cancer (6.1% [19 of 310] to 2.7% [16 of 600]; P = .02).

Conclusions and Relevance  In this cohort study, prescribing an opioid supply of 3 or fewer days to surgical patients after hospital discharge was feasible for most patients, led to a significant decrease in the number of opioids prescribed after surgery, and was associated with a significantly decreased conversion to long-term opioid use without concomitant increases in refill requests or significant compromises in surgical recovery.

Prescription of opioids remains an important driver of the opioid epidemic in the US. Each year, 48 million US individuals undergo surgical procedures,¹ and 5% to 9% of these patients may develop new, persistent opioid use.^2,3 Opioids continue to be the primary management of acute postoperative pain after hospital discharge.⁴ Presently, legislative policies that limit opioid prescriptions for acute postsurgical pain do not exist in many states.⁵ States with statutes curbing prescriptions commonly allow for prescriptions of up to 7 days of opioids for patients without cancer and no restrictions for patients with cancer, even when undergoing the same surgery.^5,6 Thus, current restrictions may be insufficient, considering data that show the chance of chronic opioid use begins to increase after the third day of opioid consumption and rapidly rises after the fifth day.⁷ Shorter-term opioid prescriptions following surgery may be associated with reduced new, chronic opioid use by limiting the duration of opioid exposure during the critical period after surgery for patients with cancer as well.⁷ However, there are presently limited data on the amounts of opioids necessary for adequate pain management and optimal recovery⁸ and on feasibility of implementing such strategies.

Reports on individual institutions’ experience to curb opioid prescriptions have focused on prescriber education and changes in electronic orders and have shown variable reductions in opioid prescriptions.^9,10 Several professional societies have sought to provide guidance on the number of prescribed narcotic pills,¹¹ but many still recommend generous opioid prescriptions based on anecdotal evidence and are not harmonized with state-level opioid restriction policies. Some of us previously reported the development of a restrictive opioid prescription protocol (ROPP) piloted in our gynecologic oncology department and showed that most women undergoing gynecologic surgery required either no opioids or only a 3-day supply without complications.¹² We sought to build on this success by expanding the ROPP to all surgical services at our cancer center, which performs complex cancer surgeries in western New York. Prior to implementation of a standardized postsurgical pain management protocol, our cancer center lacked guidelines on pain management after hospital discharge and relied on prescriber discretion. Our goal was to evaluate whether the care of patients undergoing complex surgical procedures could be managed safely and effectively with an ROPP based on treatment length. We hypothesized that universal implementation of an ROPP stipulating an opioid supply of 3 or fewer days at the time of hospital discharge from surgery would be feasible for most postsurgical patients, would result in fewer opioids dispensed, and would be associated with reduced conversion to chronic use without adverse effects on surgical recovery. In this article, we report on feasibility of implementation and clinical outcomes of the ROPP when applied across all surgical services at our tertiary cancer center.

In this cohort study, all participating surgical services (breast, gastrointestinal, head and neck, neurosurgery, soft tissue and melanoma, thoracic, and urology) at a National Cancer Institute–designated comprehensive cancer center developed their own ROPP based on multidisciplinary input (Figure 1). Surgical services established default prescription amounts for their common procedures that maximized nonnarcotic analgesia (eg, acetaminophen, nonsteroidal anti-inflammatory drugs) and set a maximum 3-day opioid supply at the time of discharge (12-24 tablets, depending on dosage instruction) unless there was a compelling reason for a longer prescription as determined by the prescriber. Patients could receive 1 electronic refill of an opioid prescription, but additional opioid prescriptions required a clinical evaluation. The list of procedures for each surgical service and the recommended number of opioid prescription days are given in eTable 1 in the Supplement. The Roswell Park Comprehensive Cancer Center institutional review board provided approval for clinical data collection, but because the change in opioid prescribing practice was considered a quality improvement study, it was deemed exempt from approval and patient consent as no intervention was performed. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Prior to implementation, health care professional education sessions about the ROPP were presented to prescribers and other staff (eg, pharmacists, nurses), with reference material for each surgical specialty protocol distributed to prescribers to guide prescriptions. Custom order sets were added to the electronic medical record that provided default prescriptions compliant with the ROPP (ie, maximum 3-day opioid supply).

A standardized video about postsurgical pain management and our ROPP was developed by our media team and presented to patients during preoperative visits to help set postsurgical pain management expectations.¹³ All patients undergoing surgery (except for endoscopic cases) were required to watch this video during preoperative teaching starting February 1, 2019.

Following implementation, dedicated pharmacists reviewed all discharge prescriptions to assess compliance with the ROPP (eTable 2 in the Supplement) and contacted health care professionals who deviated from the ROPP during the first 6 months of implementation to document the reason for deviation. Compliance data were shared monthly with surgical services to provide feedback for prescribing practices and compare adherence between services. After the first 6 months of monitoring by pharmacy, prescribing information was automated using data from the New York State Internet System for Tracking Over-Prescribing (I-STOP) database, a prescription drug monitoring program.

Eligible patients included individuals undergoing minor or major surgical procedures with the participating surgical services during the study period (August 1, 2018, to July 31, 2019) who would have routinely received an opioid prescription prior to the ROPP implementation. The pre-ROPP (unrestricted opioid use) cohort consisted of all patients undergoing surgery at our cancer center between August 1, 2018, and January 31, 2019; the post-ROPP (restricted opioid use) cohort consisted of all patients undergoing surgery between February 1, 2019, and July 31, 2019. Endoscopic procedures, outpatient biopsy cases, and patients with multiple surgeries within 30 days were excluded. Surgical routes were open or minimally invasive (all laparoscopic and robotic cases). Surgical complexity, described as minor (minimal alteration of body parts) or major (extensive alteration of body parts and/or reconstructions), was assigned based on disease site–specific factors and input from the associated surgical service. Cancer diagnoses were extracted manually by using New York State Cancer Registry data. Demographic information was obtained from the electronic medical record. Race and ethnicity are reported in Table 1 to describe the characteristics of the study population but were not included in the analysis.

The primary outcomes were (1) compliance with the ROPP in each surgical service from the time of implementation; (2) length of the opioid prescription after surgery, total number of opioid medications prescribed during the 60-day perioperative period (30 days before surgery and 30 days after surgery) through 180 days after surgery, and the type of prescribed opioids tracked in I-STOP; (3) requests for opioid prescription refills within 30 days of surgery; and (4) surgical complications, measured by the American College of Surgeons’ National Surgical Quality Improvement Program (NSQIP).

The New York State I-STOP was used to gather data about opioid prescriptions filled by patients from 90 days before surgery through 180 days after surgery. The database provided date dispensed, generic medication name, dose, and number of days prescribed. Prescriptions filled within 14 days of hospital discharge were attributed to surgery.

Our secondary outcome was the number of patients with new, chronic opioid use. Determination of opioid use (ie, naive, exposed, or tolerant) was based on all verified opioid prescriptions reported in I-STOP. Opioid-naive status was defined as patients who had no opioid prescription within 90 days before surgery; patients classified as opioid exposed received at least 1 opioid prescription during this time. Opioid tolerant was the classification for patients who received at least 60 mg morphine milligram equivalents (MME) of opioids daily in at least 7 of the 14 days prior to surgery.¹⁴

Chronic opioid use has been defined variably in the literature, so we modified one of the most commonly used criteria. Patients were considered new, persistent opioid users after surgery if they were not opioid tolerant before surgery and met 1 of the following definitions: (1) 1 or more additional opioid prescription(s) dispensed 91 to 180 days after surgery based on modified Brummett criteria² or (2) 2 or more additional opioid prescriptions dispensed 91 to 180 days after surgery. The second definition was included to delineate patients receiving multiple prescriptions from those with 1 prescription.

Compliance, postsurgical outcomes, and chronic opioid use were monitored beyond the initial study period. These data are presented to show durability of our interventions in the 14 months after our protocol initiation. Additional patients who underwent surgery during this period (August 1, 2019, to September 30, 2020) had their prescription data added to the initial post-ROPP cohort to calculate the rate of conversion to chronic opioid use in a larger population (extended post-ROPP cohort) to ensure ROPP accuracy and sustainability of the data. Patients with incomplete opioid prescribing data in I-STOP or those who died within 180 days of surgery were excluded from calculations of new chronic opioid use, as opioids were used to manage end-of-life care for these patients.

Patients’ demographic and clinical characteristics were summarized by cohort (pre-ROPP vs post-ROPP and extended post-ROPP opioid use) using the mean and SD for continuous variables and frequencies and relative frequencies for categorical variables. Comparisons were made using the Mann-Whitney U, Pearson χ², and Fisher exact tests as appropriate.

The length of opioid prescriptions and total dispensed MME (both 14 and 30 days after surgery) were summarized by cohort in the overall sample and within each surgical service by using the mean and SD and were graphically illustrated using bar charts. Comparisons were made using the Mann-Whitney U test. The refill and conversion rates were summarized by cohort and compared using the Fisher exact test. The NSQIP complications were summarized by cohort using frequencies and relative frequencies, with comparisons made using the Fisher exact test. All analyses were conducted in SAS, version 9.4 (SAS Institute) at a 2-sided significance level of P = .05.

A total of 4068 patients (mean [SD] age, 61.0 [13.8] years; 1538 men [37.8%], 2528 women [62.1%], and 2 [0.1%] whose sex was unknown) underwent a surgical procedure during the study period (initial 12 months). The control (pre-ROPP, unrestricted opioid use) and intervention (post-ROPP, restricted opioid use) cohorts consisted of 2017 and 2051 patients, respectively. No significant differences in age, body mass index, sex, or race were observed between cohorts (Table 1). The number of cases performed by each surgical service was comparable. The route of surgery (minimally invasive or open) and the complexity of the procedures (major or minor) showed no significant differences between the 2 cohorts; length of hospital stay was the same. Approximately 82% of all patients in each cohort had a diagnosis of cancer (1642 of 2017 [81.4%] pre-ROPP and 1683 of 2051 [82.1%] post-ROPP), with the rest of the patients having premalignant or benign lesions removed. The overall compliance rate with the ROPP during the first 6 months of implementation was 95% (eTable 2 in the Supplement). After implementation, the mean duration of opioid prescriptions decreased from a mean (SD) of 3.9 (4.5) days in the pre-ROPP group to 1.9 (3.6) days in the post-ROPP group (P < .001) (Figure 2A).

The extended post-ROPP cohort consisted of 3978 patients (1927 additional patients). The extended post-ROPP cohort had similar demographics and surgical procedures performed as the pre-ROPP cohort (eTable 3 in the Supplement). A total of 78 patients in the pre-ROPP cohort (3.7%) and 128 in the extended post-ROPP cohort (3.1%) were excluded from calculations of new chronic opioid use because of incomplete opioid prescribing data or death within 180 days of surgery. The mean (SD) duration of prescriptions continued to decrease to 0.9 (2.8) days during the additional 12 months of monitoring in the poststudy period (extended post-ROPP cohort) (eFigure 1 in the Supplement). The ROPP led to a 45% decrease in prescribed opioids after surgery (mean [SD], 157.22 [338.06] MME before ROPP vs 83.54 [395.70] MME after ROPP; P < .001) (eFigure 2 in the Supplement). This decrease continued in the poststudy period monitoring to a mean (SD) of 43.01 (287.08) MME in the subsequent year in the extended cohort (eFigure 3 in the Supplement). There was a significant decrease in the most prescribed short-acting opioids and an increase in prescriptions of tramadol, which was often prescribed for patients with contraindications to nonsteroidal anti-inflammatory drugs or acetaminophen (eFigure 4 in the Supplement).

The change in opioid prescribing practices remained stable beyond the first 6 months of ROPP implementation (Figure 2B). Changes in prescribing patterns were observed prior to the official implementation of the ROPP, which corresponded to the start of health care professional education around November 2018 (Figure 2B) to orient staff to the program. Despite patients receiving fewer opioids after surgery, patients in the post-ROPP cohort requested fewer opioid refills than patients in the control group (367 of 2051 [17.9%] vs 422 of 2017 [20.9%]; P = .02). There was no difference in surgical complications in the pre-ROPP and post-ROPP periods as measured by the NSQIP (eTable 4 in the Supplement).

Among the patients with a cancer diagnosis, 308 of 1575 in the pre-ROPP group (19.6%) and 521 of 3163 in the extended post-ROPP group (16.5%) met the criteria for being opioid exposed prior to surgery (P = .009). A total of 59 of 369 patients without a cancer diagnosis (16.0%) were classified as opioid exposed in the pre-ROPP group vs 102 of 702 (14.5%) in the extended post-ROPP group (P = .53) (Table 2). When using the Brummett definition for new, persistent opioid use (≥1 opioid prescription dispensed 91-180 days after surgery), 143 of 1267 opioid-naive patients with cancer in the pre-ROPP cohort (11.3%) developed new, persistent opioid use compared with 118 of 2645 after ROPP implementation (4.5%) (P < .001) (Table 2). In patients without a cancer diagnosis, the rate of new persistent opioid use also significantly decreased after ROPP implementation (19 of 310 [6.1%] vs 16 of 600 [2.7%]; P = .02). Even when applying the stricter definition for new persistent opioid use (≥2 opioid prescriptions 90-180 days after surgery), we observed a decreased trend in the rate of new persistent opioid use in both patients with a cancer diagnosis and those without a cancer diagnosis (Table 2).

Surgical procedures frequently result in acute pain management with narcotics, but exposure to opioids can predispose patients to chronic use. While individual interventions have been reported to be associated with decreased opioid prescriptions (eg, changing default electronic orders, health care professional–level education), comprehensive integration of these approaches at an institutional level remains elusive partially due to ingrained practices and inherent challenges of engaging all health care professionals in a large initiative.

Prior to the implementation of our ROPP, our cancer center already used enhanced-recovery-after-surgery protocols and maximized nonopioid medications during hospitalization across all surgical services. While these efforts impacted patients during their hospitalization, we lacked any standardized approach to postdischarge pain management. We wanted to build on improvements made for admitted patients and implement a protocol to minimize opioids after discharge, when risk of chronic use is proportional to the duration of exposure.⁵ A prior experience in our gynecologic oncology department demonstrated that most patients did not need more than a 3-day supply of opioids after surgery to control pain effectively.¹² Consequently, we believed that a supply of 3 days or less would be a reasonable prescription for most surgical patients, thus mitigating the risk of opioid-related adverse events, including unintentional chronic use.

Our results indicated the implementation of a ROPP with multidisciplinary interventions was feasible within all surgical services, had a high compliance rate, and was durable over time. Standardizing the postoperative pain management of patients led to significant decreases in the mean amount and duration of opioid prescriptions. These reductions did not compromise postsurgical complications, as reflected in hospital-wide NSQIP data. We also found fewer opioid prescription refills were provided to patients after implementing the ROPP. This result may have been influenced by standardized patient education and providing prescribers with a structured approach to postoperative pain management.

In addition to the successful implementation of the ROPP, we found an association of the ROPP with opioid use in patients 91 to 180 days after surgery. Fewer opioid-naive patients from the post-ROPP cohort had an opioid prescription dispensed during this period according to the prescription drug monitoring program. Minimizing opioid exposure in this group following surgery may help prevent new, persistent opioid use. Interestingly, there was no difference in long-term opioid prescriptions between the 2 cohorts for patients categorized as being opioid exposed prior to surgery. While opioid-naive patients had their exposure limited by the ROPP, the opioid-exposed cohort had already experienced a period of exposure that may have affected their risk for long-term use.

Guidelines on opioid stewardship emphasize the importance of a multifaceted, team-based approach to affect change. We incorporated this approach in the design and implementation of our protocol. Staff and health care professionals at every level were engaged through education and meetings to communicate the importance of the ROPP. Patient education, changes to electronic prescribing defaults, and monitoring of discharge medications were all modified to support the initiative. While significant effort and investment were needed to align our postoperative pain management, the program changed the culture at our institution and has had a lasting effect on clinical practice.

While opioid-related acute adverse events (eg, respiratory depression, constipation, and nausea) negatively affect postsurgical recovery, they are mostly short-term and reversible. However, long-term complications such as new, persistent opioid use is associated with an increased risk of morbidity and mortality and poor clinical outcomes in patients with cancer.^15,16 New, persistent opioid use after surgery has been documented following 5% to 9% of surgeries,² yet this outcome is not routinely tracked as a quality metric.

The main strength of our study was demonstration of the feasibility of implementing an ROPP across a broad surgical population. The use of a statewide drug monitoring database also provided reliable information about the number of prescriptions filled by patients. Comparing hospital records with dispensed prescriptions from the database demonstrated a substantial difference in the number of prescriptions filled.

One limitation of this study was that it was a single cancer center experience with results reflecting the patient characteristics of western New York; thus, further study in other regions is warranted, although we believe our data would likely be applicable to a broader group of patients undergoing benign elective surgery. Another limitation is that we did not collect data on postsurgical adjuvant treatment among these patients. However, our cancer registry data do not show a difference in the patient population for which we provided care during the study period in western New York, and one of our major goals was to look at the effects of change in opioid prescribing at the time of surgery, which is typically one of the most acute painful interventions for patients and thus may be associated with a high risk for conversion to chronic opioid use. While we support the use of opioids in managing cancer-related pain when necessary, many of the patients with cancer at our center are opioid naive at the time of surgery and are long-term survivors. Reduction in conversion to chronic opioid use due to surgical pain management could potentially further extend their lives. Thus, the accumulation of evidence on feasibility of ROPPs could lead to a new standard of care for managing postsurgical pain after hospital discharge.

In this cohort study, prescribing an opioid supply of 3 or fewer days to surgical patients after hospital discharge was feasible for most patients, led to a significant decrease in the number of opioids prescribed postoperatively, and was associated with decreased conversion to long-term opioid use without concomitant increases in refill requests or significant compromises in surgical recovery. Changes in established clinical practices can be a daunting prospect, and unlearning conventional, anecdotal wisdom may be difficult. Our findings suggest that success at reducing opioid prescribing on an institutional level is possible through interprofessional collaborations, simple guidelines, and regular feedback on performance to establish a high-quality postsurgical recovery strategy and may be associated with decreased risk for unintended, long-term complications from opioid use.

Accepted for Publication: October 3, 2022.

Published Online: January 5, 2023. doi:10.1001/jamaoncol.2022.6278

Corresponding Author: Emese Zsiros, MD, PhD, Department of Gynecologic Oncology, Roswell Park Comprehensive Cancer Center, Scott Bieler Clinical Sciences Center, Elm and Carlton Streets, Buffalo, NY 14263 (emese.zsiros@roswellpark.org).

Author Contributions: Dr Zsiros had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Zsiros, Ricciuti, Riebandt, Yendamuri, Hicks, Odunsi, Johnson, de Leon-Casasola.

Acquisition, analysis, or interpretation of data: Zsiros, Ricciuti, Gallo, Argentieri, Attwood, Ji, Hutson, Visco, Coffey, Mark, Varghese, Hess, Furlani, Fabiano, Hennon, Yendamuri, Kauffman, Wooten, Young, Takabe, Case, Segal, Kuvshinoff, Nurkin, Paragh.

Drafting of the manuscript: Zsiros, Ricciuti, Nurkin, de Leon-Casasola.

Critical revision of the manuscript for important intellectual content: Zsiros, Ricciuti, Gallo, Argentieri, Attwood, Ji, Hutson, Visco, Coffey, Riebandt, Mark, Varghese, Hess, Furlani, Fabiano, Hennon, Yendamuri, Kauffman, Wooten, Hicks, Young, Takabe, Odunsi, Case, Segal, Johnson, Kuvshinoff, Paragh.

Statistical analysis: Zsiros, Gallo, Attwood, Ji, Hutson, Visco.

Obtained funding: Zsiros.

Administrative, technical, or material support: Zsiros, Ricciuti, Gallo, Argentieri, Visco, Hess, Furlani, Yendamuri, Wooten, Hicks, Takabe, Odunsi, Nurkin.

Supervision: Zsiros, Visco, Case, Johnson, Nurkin, de Leon-Casasola.

Conflict of Interest Disclosures: Dr Yendamuri reported receiving grants from Lumeda Inc outside the submitted work. Dr Odunsi reported receiving nonfinancial support from Tactiva Therapeutics as cofounder and grants from AstraZeneca and Tesaro outside the submitted work. Dr Segal reported receiving grants from Apellis and personal fees from NextCure outside the submitted work. No other disclosures were reported.

Funding/Support: This work was supported by grant P30CA016056 (Dr Johnson) from the National Cancer Institute, National Institutes of Health involving the use of Roswell Park Comprehensive Cancer Center’s Scientific Communications and Research Communications Core and by the Roswell Park Alliance Foundation.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.