Effect of Changing Electronic Health Record Opioid Analgesic Dispense Quantity Defaults on the Quantity Prescribed: A Cluster Randomized Clinical Trial | Electronic Health Records | JAMA Network Open | JAMA Network
[Skip to Navigation]
Sign In
Visual Abstract. Effect of changing electronic health record opioid analgesic dispense quantity defaults on the quantity prescribed
Effect of changing electronic health record opioid analgesic dispense quantity defaults on the quantity prescribed
Figure 1.  Consolidated Standards of Reporting Trials Flow Diagram
Consolidated Standards of Reporting Trials Flow Diagram
Figure 2.  Unadjusted Primary Outcome Measures by Study Month
Unadjusted Primary Outcome Measures by Study Month

A, Percentage of prescriptions for 10 or fewer tablets. B, Tablets prescribed. C, Morphine milligram equivalents prescribed.

Table 1.  Characteristics of Clinical Sites and Prescribers
Characteristics of Clinical Sites and Prescribers
Table 2.  Characteristics of Patients Before and After Intervention
Characteristics of Patients Before and After Intervention
Table 3.  Primary and Secondary Outcomes
Primary and Secondary Outcomes
1.
Bohnert  AS, Valenstein  M, Bair  MJ,  et al.  Association between opioid prescribing patterns and opioid overdose-related deaths.   JAMA. 2011;305(13):1315-1321. doi:10.1001/jama.2011.370 PubMedGoogle ScholarCrossref
2.
Fulton-Kehoe  D, Sullivan  MD, Turner  JA,  et al.  Opioid poisonings in Washington State Medicaid: trends, dosing, and guidelines.   Med Care. 2015;53(8):679-685. doi:10.1097/MLR.0000000000000384 PubMedGoogle ScholarCrossref
3.
Lewis  ET, Cucciare  MA, Trafton  JA.  What do patients do with unused opioid medications?   Clin J Pain. 2014;30(8):654-662. doi:10.1097/01.ajp.0000435447.96642.f4 PubMedGoogle ScholarCrossref
4.
Centers for Disease Control and Prevention (CDC).  Adult use of prescription opioid pain medications—Utah, 2008.   MMWR Morb Mortal Wkly Rep. 2010;59(6):153-157.PubMedGoogle Scholar
5.
Kennedy-Hendricks  A, Gielen  A, McDonald  E, McGinty  EE, Shields  W, Barry  CL.  Medication sharing, storage, and disposal practices for opioid medications among US Adults.   JAMA Intern Med. 2016;176(7):1027-1029. doi:10.1001/jamainternmed.2016.2543 PubMedGoogle ScholarCrossref
6.
McCabe  SE, West  BT, Boyd  CJ.  Leftover prescription opioids and nonmedical use among high school seniors: a multi-cohort national study.   J Adolesc Health. 2013;52(4):480-485. doi:10.1016/j.jadohealth.2012.08.007 PubMedGoogle ScholarCrossref
7.
Hall  AJ, Logan  JE, Toblin  RL,  et al.  Patterns of abuse among unintentional pharmaceutical overdose fatalities.   JAMA. 2008;300(22):2613-2620. doi:10.1001/jama.2008.802 PubMedGoogle ScholarCrossref
8.
Center for Behavioral Health Statistics and Quality. Behavioral Health Trends in the United States: Results From the 2014 National Survey on Drug Use and Health (HHS Publication No. SMA 15-4927, NSDUH Series H-50). Center for Behavioral Health Statistics and Quality; 2015.
9.
Voepel-Lewis  T, Wagner  D, Tait  AR.  Leftover prescription opioids after minor procedures: an unwitting source for accidental overdose in children.   JAMA Pediatr. 2015;169(5):497-498. doi:10.1001/jamapediatrics.2014.3583 PubMedGoogle ScholarCrossref
10.
Bailey  JE, Campagna  E, Dart  RC; RADARS System Poison Center Investigators.  The underrecognized toll of prescription opioid abuse on young children.   Ann Emerg Med. 2009;53(4):419-424. doi:10.1016/j.annemergmed.2008.07.015 PubMedGoogle ScholarCrossref
11.
Brat  GA, Agniel  D, Beam  A,  et al.  Postsurgical prescriptions for opioid naive patients and association with overdose and misuse: retrospective cohort study.   BMJ. 2018;360:j5790. doi:10.1136/bmj.j5790 PubMedGoogle ScholarCrossref
12.
Vélez-Díaz-Pallarés  M, Pérez-Menéndez-Conde  C, Bermejo-Vicedo  T.  Systematic review of computerized prescriber order entry and clinical decision support.   Am J Health Syst Pharm. 2018;75(23):1909-1921. doi:10.2146/ajhp170870 PubMedGoogle ScholarCrossref
13.
Garg  AX, Adhikari  NK, McDonald  H,  et al.  Effects of computerized clinical decision support systems on practitioner performance and patient outcomes: a systematic review.   JAMA. 2005;293(10):1223-1238. doi:10.1001/jama.293.10.1223 PubMedGoogle ScholarCrossref
14.
Kruse  CS, Ehrbar  N.  Effects of computerized decision support systems on practitioner performance and patient outcomes: systematic review.   JMIR Med Inform. 2020;8(8):e17283. doi:10.2196/17283 PubMedGoogle Scholar
15.
Chiu  AS, Freedman-Weiss  MR, Jean  RA, Cohen  E, Yoo  PS.  No refills: the durable impact of a multifaceted effort by surgical trainees to minimize the prescription of postoperative opioids.   Surgery. 2019;166(5):758-763. doi:10.1016/j.surg.2019.05.021 PubMedGoogle ScholarCrossref
16.
Chiu  AS, Jean  RA, Hoag  JR, Freedman-Weiss  M, Healy  JM, Pei  KY.  Association of lowering default pill counts in electronic medical record systems with postoperative opioid prescribing.   JAMA Surg. 2018;153(11):1012-1019. doi:10.1001/jamasurg.2018.2083 PubMedGoogle ScholarCrossref
17.
Delgado  MK, Shofer  FS, Patel  MS,  et al.  Association between electronic medical record implementation of default opioid prescription quantities and prescribing behavior in two emergency departments.   J Gen Intern Med. 2018;33(4):409-411. doi:10.1007/s11606-017-4286-5 PubMedGoogle ScholarCrossref
18.
Zivin  K, White  JO, Chao  S,  et al.  Implementing electronic health record default settings to reduce opioid overprescribing: a pilot study.   Pain Med. 2019;20(1):103-112. doi:10.1093/pm/pnx304 PubMedGoogle ScholarCrossref
19.
Montoy  JCC, Coralic  Z, Herring  AA, Clattenburg  EJ, Raven  MC.  Association of default electronic medical record settings with health care professional patterns of opioid prescribing in emergency departments: a randomized quality improvement study.   JAMA Intern Med. 2020;180(4):487-493. doi:10.1001/jamainternmed.2019.6544 PubMedGoogle ScholarCrossref
20.
Smalley  CM, Willner  MA, Muir  MR,  et al.  Electronic medical record-based interventions to encourage opioid prescribing best practices in the emergency department.   Am J Emerg Med. 2020;38(8):1647-1651. doi:10.1016/j.ajem.2019.158500PubMedGoogle ScholarCrossref
21.
Blutinger  EJ, Shofer  FS, Meisel  Z, Perrone  J, Engel-Rebitzer  E, Delgado  MK.  Variability in emergency department electronic medical record default opioid quantities: a national survey.   Am J Emerg Med. 2019;37(10):1963-1964. doi:10.1016/j.ajem.2019.03.023 PubMedGoogle ScholarCrossref
22.
Meisenberg  BR, Grover  J, Campbell  C, Korpon  D.  Assessment of opioid prescribing practices before and after implementation of a health system intervention to reduce opioid overprescribing.   JAMA Netw Open. 2018;1(5):e182908. doi:10.1001/jamanetworkopen.2018.2908 PubMedGoogle Scholar
23.
Santistevan  JR, Sharp  BR, Hamedani  AG, Fruhan  S, Lee  AW, Patterson  BW.  By default: the effect of prepopulated prescription quantities on opioid prescribing in the emergency department.   West J Emerg Med. 2018;19(2):392-397. doi:10.5811/westjem.2017.10.33798 PubMedGoogle ScholarCrossref
24.
Schwartz  GD, Harding  AM, Donaldson  SR, Greene  SL.  Modifying emergency department electronic prescribing for outpatient opioid analgesia.   Emerg Med Australas. 2019;31(3):417-422. doi:10.1111/1742-6723.13192 PubMedGoogle ScholarCrossref
25.
Zwank  MD, Kennedy  SM, Stuck  LH, Gordon  BD.  Removing default dispense quantity from opioid prescriptions in the electronic medical record.   Am J Emerg Med. 2017;35(10):1567-1569. doi:10.1016/j.ajem.2017.04.002 PubMedGoogle ScholarCrossref
26.
McCarthy  DM, Kim  HS, Hur  SI,  et al.  Patient-reported opioid pill consumption after an ED visit: how many pills are people using?   Pain Med. 2021;22(2):292-302. PubMedGoogle ScholarCrossref
27.
Shi  R, Quinones  A, Bair  J,  et al.  Patient utilization of prescription opioids after discharge from the emergency department.   Am J Emerg Med. 2020;38(8):1568-1571. PubMedGoogle ScholarCrossref
28.
Daoust  R, Paquet  J, Cournoyer  A,  et al.  Quantity of opioids consumed following an emergency department visit for acute pain: a Canadian prospective cohort study.   BMJ Open. 2018;8(9):e022649. doi:10.1136/bmjopen-2018-022649 PubMedGoogle Scholar
29.
Bicket  MC, Long  JJ, Pronovost  PJ, Alexander  GC, Wu  CL.  Prescription opioid analgesics commonly unused after surgery: a systematic review.   JAMA Surg. 2017;152(11):1066-1071. doi:10.1001/jamasurg.2017.0831 PubMedGoogle ScholarCrossref
30.
Von Korff  M, Saunders  K, Thomas Ray  G,  et al.  De facto long-term opioid therapy for noncancer pain.   Clin J Pain. 2008;24(6):521-527. doi:10.1097/AJP.0b013e318169d03b PubMedGoogle ScholarCrossref
31.
Hooten  WM, St Sauver  JL, McGree  ME, Jacobson  DJ, Warner  DO.  Incidence and risk factors for progression from short-term to episodic or long-term opioid prescribing: a population-based study.   Mayo Clin Proc. 2015;90(7):850-856. doi:10.1016/j.mayocp.2015.04.012 PubMedGoogle ScholarCrossref
32.
Bachhuber  MA, Nash  D, Southern  WN,  et al.  Reducing the default dispense quantity for new opioid analgesic prescriptions: study protocol for a cluster randomised controlled trial.   BMJ Open. 2018;8(4):e019559. doi:10.1136/bmjopen-2017-019559 PubMedGoogle Scholar
33.
US Department of Health and Human Services.  National Pain Strategy: A Comprehensive Population Health-Level Strategy for Pain. US Department of Health and Human Services; 2016.
34.
Healthcare Cost and Utilization Project (HCUP). Beta Clinical Classifications Software (CCS) for ICD-10-CM/PCS. Agency for Healthcare Research and Quality. Accessed January 29, 2018. https://www.hcup-us.ahrq.gov/toolssoftware/ccs10/ccs10.jsp
35.
Tian  TY, Zlateva  I, Anderson  DR.  Using electronic health records data to identify patients with chronic pain in a primary care setting.   J Am Med Inform Assoc. 2013;20(e2):e275-e280. doi:10.1136/amiajnl-2013-001856 PubMedGoogle ScholarCrossref
36.
Greevy  R, Lu  B, Silber  JH, Rosenbaum  P.  Optimal multivariate matching before randomization.   Biostatistics. 2004;5(2):263-275. doi:10.1093/biostatistics/5.2.263 PubMedGoogle ScholarCrossref
37.
Huber  PJ. The behavior of maximum likelihood estimates under nonstandard conditions. Proc Fifth Berkeley Symp Math Statist Prob. 1967;1:221-223.
38.
White  H.  A Heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity.   Econometrica. 1980;48(4):817-838. doi:10.2307/1912934 Google ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Views 2,409
    Citations 0
    Original Investigation
    Substance Use and Addiction
    April 22, 2021

    Effect of Changing Electronic Health Record Opioid Analgesic Dispense Quantity Defaults on the Quantity Prescribed: A Cluster Randomized Clinical Trial

    Author Affiliations
    • 1Division of General Internal Medicine, Department of Medicine, Montefiore Medical Center/Albert Einstein College of Medicine, Bronx, New York
    • 2Now with Section of Community and Population Medicine, Louisiana State University Health Sciences Center–New Orleans
    • 3Institute for Implementation Science in Population Health, City University of New York, New York
    • 4Graduate School of Public Health and Health Policy, Department of Epidemiology and Biostatistics, City University of New York, New York, New York
    • 5Division of Hospital Medicine, Montefiore Medical Center/Albert Einstein College of Medicine, Bronx, New York
    • 6Now with College of Behavioral, Social and Health Sciences, Department of Public Health Sciences, Clemson University, Clemson, South Carolina
    • 7Montefiore Information Technology, Montefiore Medical Center, Bronx, New York
    • 8College of Nursing and Health Sciences, University of Massachusetts Boston, Boston
    JAMA Netw Open. 2021;4(4):e217481. doi:10.1001/jamanetworkopen.2021.7481
    Key Points

    Question  Does reducing the default number of tablets for opioid analgesic prescriptions in the electronic health record influence prescribing or other outcomes?

    Findings  In this cluster randomized clinical trial of 32 primary care and 4 emergency department sites, with a total of 21 331 prescriptions, a reduced uniform default dispense quantity of 10 tablets led to an increase in the percentage of prescriptions for 10 tablets or fewer (7.6 percentage points). No significant difference in health service use was noted.

    Meaning  Findings of this study suggest that implementation of a reduced uniform opioid analgesic prescribing default is a feasible intervention that can modestly reduce prescribing.

    Abstract

    Importance  Interventions to improve judicious prescribing of opioid analgesics for acute pain are needed owing to the risks of diversion, misuse, and overdose.

    Objective  To assess the effect of modifying opioid analgesic prescribing defaults in the electronic health record (EHR) on prescribing and health service use.

    Design, Setting, and Participants  A cluster randomized clinical trial with 2 parallel arms was conducted between June 13, 2016, and June 13, 2018, in a large urban health care system comprising 32 primary care and 4 emergency department (ED) sites in the Bronx, New York. Data were analyzed using a difference-in-differences method from 6 months before implementation through 18 months after implementation. Data were analyzed from January 2019 to February 2020.

    Interventions  A default dispense quantity for new opioid analgesic prescriptions of 10 tablets (intervention) vs no change (control) in the EHR.

    Main Outcomes and Measures  The primary outcome was the quantity of opioid analgesics prescribed with the new default prescription. Secondary outcomes were opioid analgesic reorders and health service use within 30 days after the new prescription. Intention-to-treat analysis was conducted.

    Results  Overall, 21 331 patients received a new opioid analgesic prescription from 490 prescribers. Comparing the intervention and control arms, site, prescriber, and patient characteristics were similar. For the new prescription, compared with the control arm, patients in the intervention arm had significantly more prescriptions for 10 tablets or fewer (7.6 percentage points; 95% CI, 6.1-9.2 percentage points), a lower number of tablets prescribed (−2.1 tablets; 95% CI, −3.3 to −0.9 tablets), and lower morphine milligram equivalents (MME) prescribed (−14.6 MME; 95% CI, −22.6 to −6.6 MME). Within 30 days after the new prescription, significant differences remained in the number of tablets prescribed (−2.7 tablets; 95% CI, −4.8 to −0.6 tablets), but not MME (−15.8 MME; 95% CI, −33.8 to 2.2 MME). Within this 30-day period, there were no significant differences between the arms in health service use.

    Conclusions and Relevance  In this study, implementation of a uniform reduced default dispense quantity of 10 tablets for opioid analgesic prescriptions led to a modest reduction in the quantity prescribed initially, without significantly increasing health service use. However, during 30 days after implementation, the influence on prescribing was mixed. Reducing EHR default dispense quantities for opioid analgesics is a feasible strategy that can be widely disseminated and may modestly reduce prescribing.

    Trial Registration  ClinicalTrials.gov Identifier: NCT03003832

    Introduction

    Opioid analgesics prescribed for acute noncancer pain are associated with risks such as diversion, misuse, and overdose.1-11 Efforts to promote judicious opioid prescribing for acute pain have included prescriber education, clinical guidelines, and legislative mandates. However, with widespread use of electronic health records (EHRs) and the demonstrated efficacy of EHR-based interventions in influencing care,12-14 technology offers new scalable opportunities, such as modifying opioid analgesic prescribing defaults.

    Prescribing defaults refer to configuration settings within the EHR that prepopulate certain fields in new prescription orders, such as the number of tablets to dispense. Although fully modifiable by the prescriber, these prepopulated numbers can guide prescribers toward a lower quantity than they otherwise might have chosen. In 10 recent studies, implementing a new or lower default dispense quantity for opioid analgesics, either alone or as part of a package of interventions, was associated with reductions in the quantity of new prescriptions.15-24 In 2 studies, removing defaults, and therefore requiring prescribers to make an active choice with each prescription, yielded mixed results.23,25

    Given evidence that many patients do not use the full amount of opioid analgesic tablets dispensed and retain the remainder,5,26-29 reducing the quantity prescribed may better approximate the amount used. However, a reduced default may result in unintended consequences. If the default dispense quantity is too low, for example, some patients may need additional medication or seek medical attention for uncontrolled pain. Furthermore, given that patterns of opioid analgesic prescribing are changing, the use of rigorous, controlled, study designs is necessary. Uncontrolled analyses (eg, pre-post analyses) may suggest an intervention is efficacious when changes in prescribing would have happened regardless of the intervention, for example, due to policy or regulatory changes.

    The goal of this study was to assess the effect of a uniform reduced default dispense quantity for new opioid analgesic prescriptions on the quantity prescribed. We hypothesized that, compared with control, reducing the default dispense quantity would lead to a higher percentage of prescriptions written for the new reduced default number of tablets or fewer. In addition, compared with the control (ie, no intervention), we hypothesized that reducing the default dispense quantity would not lead to a significant increase in subsequent opioid analgesic prescription reorders or primary care visits, emergency department (ED) visits, or hospitalizations. For this study, we selected a cluster randomized design (ie, at the level of the clinical site) to mitigate the risk of contamination and take advantage of possible peer effects to increase the effectiveness of the intervention.

    Methods
    Study Design and Participants

    Between June 13, 2016, and June 13, 2018, we conducted a cluster randomized clinical trial with 2 parallel arms among all primary care (n = 32) and ED (n = 4) sites within Montefiore Medical Center. Data were analyzed from January 2019 to February 2020. Montefiore Medical Center is the largest health care system in the Bronx (a borough of New York City) and provides comprehensive primary, specialty, surgical, and emergency care with more than 3 million patient visits annually. Montefiore Medical Center uses the Epic EHR system (Epic Systems Corp), which was the primary source of study data. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline extension for cluster randomized clinical trials.

    To determine the effect of the intervention, we conducted an intention-to-treat analysis of outcomes for patients who (1) received a new opioid analgesic prescription at a study site, defined as no other opioid analgesic prescription of any type in the preceding 6 months (a definition used in previous cohort studies)30,31; (2) were aged 18 years or older; and (3) had no International Statistical Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) diagnosis code for cancer within 1 year before the new opioid analgesic prescription. For patients receiving more than 1 new opioid analgesic prescription during the study period, we only included the first prescription.

    The protocol of this trial has been previously published32 and is included in its original form in Supplement 1. The protocol was approved, with a waiver of informed consent per the standard criteria as described in 45 CFR 46.116, by the Montefiore Medical Center/Albert Einstein College of Medicine Institutional Review Board. There were no changes to the methods after the trial was initiated.

    Intervention and Control Conditions

    The intervention condition was a site-level change to the EHR to implement a uniform, reduced, default dispense quantity of 10 tablets for new opioid analgesic prescriptions. We chose this default dispense quantity because it represents a 3- to 5-day supply for most people, based on our clinical judgment and a review of available literature at the time.32 This default number of tablets prepopulated in all new opioid prescription orders in the intervention condition and remained fully modifiable. The intervention included all short-acting opioid analgesics commonly used to treat acute pain: immediate-release oxycodone, immediate-release hydrocodone, tramadol, and codeine. We included all brand and generic formulations and all tablet strengths and coformulations with acetaminophen.

    The usual EHR served as the control condition. Depending on the exact medication product, the preexisting default number of tablets was typically 30 or blank, with several outliers.32 Of 44 products, the intervention reduced the default dispense quantity for most products (n = 36 for primary care, n = 37 for EDs) and created a default dispense quantity for several products (n = 8 for primary care, n = 7 for EDs).

    To determine the effect of the intervention, we analyzed patient-level outcomes from 6 months before implementation (baseline period) to 18 months after implementation. Our primary outcome was the quantity of opioid analgesic prescribed in the new prescription, consisting of 3 measures: (1) 10 or fewer tablets (primary measure, dichotomous), (2) number of tablets prescribed (continuous), and (3) total morphine milligram equivalents (MME) prescribed (continuous). Secondary outcomes included opioid analgesic prescription reorders of either the same or a different opioid and health service use during the 30-day period after the new prescription. We measured reorders as (1) any reorder (dichotomous), (2) total number of tablets prescribed during the 30-day period after the new prescription and inclusive of the new prescription (continuous), and (3) total MME prescribed during the 30-day period after the new prescription and inclusive of the new prescription (continuous). We measured health service use as any (1) primary care visit, (2) ED visit, or (3) hospitalization (all dichotomous) during the 30-day period after the new prescription. There were no changes to trial outcomes after trial initiation.

    Participant Characteristics and Randomization

    In addition to primary and secondary outcomes, we collected data on prescriber and patient characteristics that were potential confounders. For prescribers, we collected sex and years since graduation from medical school. For patients, we collected demographic information (age, sex, and race/ethnicity as recorded in the EHR). We also noted the pain diagnosis at the visit in which the new opioid analgesic was prescribed (ie, the indication for the opioid analgesic) in addition to the documented presence or absence of a history of psychiatric illness and substance use disorder within 1 year before the new prescription. We identified pain diagnoses, psychiatric illness, and substance use disorder through grouping ICD-10-CM diagnosis codes into clinically meaningful categories.33-35

    The unit of randomization was the site (ie, cluster randomization) and we randomized in matched pairs. First, we stratified sites by type (ie, primary care vs ED). Next, within primary care sites, we stratified by specialty (ie, internal medicine and family medicine) and whether the setting is a training site for resident physicians. In addition, within strata, we used optimal nonbipartite matching to pair sites based on the number of new opioid analgesic prescriptions, number of visits, and percentage of patients with commercial insurance.36 Emergency department sites had large differences in visit volume and so we divided these sites into a pair consisting of the largest ED vs the 3 other smaller EDs combined. Randomization of sites was conducted by the study statistician (M.H.) and provided directly to the health information technology department. Other study investigators were blinded to randomization assignment.

    Statistical Analysis

    First, we used descriptive statistics to compare site, prescriber, and patient characteristics between the intervention and control arms. Next, we analyzed the change in outcomes using a difference-in-differences (DID) analysis. One of the key assumptions of DID analysis is parallel trends before intervention. To test this assumption, we estimated the significance of an interaction term between study allocation (intervention/control) and time (month) in the preintervention period for all primary outcomes.

    For the main DID analysis, we used generalized linear mixed regression models. For dichotomous outcomes, we used linear probability models. We chose this method to facilitate estimation given the hierarchical nature of the data and provide for easily interpretable results that would be useful for clinicians. For continuous outcomes, we used models with a γ distribution and a log link, selected based on the distributions of the outcome variables.32 To adjust for potential changes in composition over time, we included relevant site characteristics (number of new opioid analgesic prescriptions, the number of visits, and percentage of patients with commercial insurance), prescriber characteristics (sex and years since medical school graduation), and patient characteristics (age, sex, race/ethnicity, pain diagnosis, history of substance use disorder, and history of psychiatric disorder) as covariates (fixed effects) in all models. To account for the nesting of patients within prescribers and prescribers within sites, for dichotomous outcomes we included random intercepts both at the prescriber level and at the matched site pair level. For continuous outcomes, models with random intercepts both at the prescriber level and at the matched site pair level did not converge and so we used models with a random intercept at the prescriber level only. For all estimates, we calculated heteroscedasticity robust (empirical) SEs.37,38

    In addition to the main analysis, we conducted several exploratory analyses. First, we analyzed the durability of the intervention after 6 months by comparing the outcomes after that time period (6-18 months post intervention) with outcomes 0 to 6 months post intervention. Next, we analyzed the impact of the intervention stratified by site type (ie, primary care vs ED), medication type (eg, schedule II vs schedules III and IV), and by whether the 10 tablets represented a new default vs a reduction in a preexisting default for opioid prescribing. We conducted these analyses using a triple differences method, which estimates the difference in the outcome from before to after intervention in the intervention arm minus the difference in the outcome from before to after intervention in the control arm in one subgroup minus the comparable difference in the comparison subgroup. We conducted all analyses with SAS, version 9.4 (SAS Institute Inc) and Stata, version 15.1 (StataCorp LLC). A 2-sided value of P ≤ .05 was considered significant.

    From preliminary data analyses, we estimated that the planned study would be powered to detect a change in primary outcome in the intervention arm over and above any change in the control arm of 4.4 to 4.7 percentage points.32 We believed this to represent a clinically meaningful difference owing to the scalable nature of the intervention.

    Results

    All primary care sites and ED sites were included, randomized, and analyzed with no losses to follow-up resulting in a total of 490 prescribers writing new opioid analgesic prescriptions for 21 331 patients (Figure 1). Compared with the control arm, sites in the intervention arm had a lower median number of visits (median, 8070; interquartile range [IQR], 3078-17 699 vs 9676; IQR, 4914-15 366) and new opioid analgesic prescriptions (median, 71; IQR, 17-103 vs 85; IQR, 49-182) before intervention as well as a lower percentage of patients with commercial insurance (median, 20.4; IQR, 18.0-29.1 vs 24.3; IQR, 19.8-35.4) (Table 1). For prescribers, characteristics between the intervention and control arm were similar. Of all prescribers, 26 (5.3%) prescribed an opioid analgesic at both an intervention and a control site after intervention, for a total of 981 prescriptions (7.1% of all analyzed prescriptions). In both the preintervention and postintervention periods, patients in the intervention and control arms had similar characteristics except, compared with the control arm, the intervention arm had a higher percentage of Hispanic/Latinx patients, a lower percentage of White patients, and a higher percentage of those with limb or extremity pain or arthritis in both the preintervention and postintervention periods (Table 2). We did not find any evidence to reject the parallel trends assumption.

    Compared with the control arm, patients in the intervention arm had a significantly increased percentage of prescriptions for 10 tablets or fewer (adjusted DID, 7.6 percentage points; 95% CI, 6.1-9.2 tablets), lower number of tablets prescribed (adjusted DID, −2.1 tablets; 95% CI, −3.3 to −0.9 tablets), and lower total MME prescribed (adjusted DID, −14.6 MME; 95% CI, −22.6 to −6.6 MME) (Figure 2; Table 3). For secondary outcomes, during the 30-day period after the new prescription, compared with the control arm, patients in the intervention arm did not have a significant difference in opioid analgesic reorders (adjusted DID, 0.5 percentage points; 95% CI, −0.7 to 1.8), had a significantly lower number of tablets prescribed (adjusted DID, −2.7 tablets; 95% CI, −4.8 to −0.6), but showed no significant difference in the total MME prescribed (adjusted DID, −15.8 MME; 95% CI, −33.8 to 2.2). There were no significant differences between the arms in postprescription health service use. Comparing 6 to 18 months post intervention with 0 to 6 months post intervention between the intervention and control arms (ie, durability), the percentage of prescriptions for 10 tablets or fewer significantly increased in the intervention arm (adjusted DID, 2.0 percentage points; 95% CI, 0.9-3.1) with no significant differences in other outcomes (eTable 1 in Supplement 2).

    In exploratory analyses, there were no significant differences between primary care sites and ED sites for all outcomes (eTable 2 in Supplement 2). Among those prescribed schedule II opioid analgesics, compared with those receiving schedules III and IV opioid analgesics, the intervention increased prescription opioid analgesic reorders (adjusted triple difference, 4.8 percentage points; 95% CI, 2.4-7.1), total MME prescribed (adjusted triple difference, 38.6 MME; 95% CI, 4.2-73.1) and the percentage with an ED visit (adjusted triple difference, 1.7 percentage points; 95% CI, 1.1-2.3) (eTable 3 in Supplement 2). Results were similar comparing medications when the intervention represented a new default vs a reduction of an existing default (eTable 4 in Supplement 2).

    Discussion

    In this cluster randomized clinical trial including 36 clinical sites, 490 prescribers, and 21 331 patients receiving a new opioid analgesic prescription, implementing a uniform default dispense quantity of 10 tablets led to a significant reduction in the quantity of opioids prescribed initially, without significantly increasing health services use. However, findings 30 days after the initial prescription were mixed. Together, our results indicate that reducing the default dispense quantity of opioid analgesics in an EHR modestly reduces opioid analgesic prescribing without increasing subsequent health services use.

    The predominance of evidence indicates that defaults can influence opioid analgesic prescribing,15-25 and our study supports that conclusion. However, to date, few experimental studies have examined the association between prescribing defaults on subsequent prescriptions. In an observational study of postsurgical patients, a reduced default dispense quantity of 12 tablets was associated with a reduction in the quantity prescribed without a corresponding increase in prescription reorders.16 Although we had a similar finding, in one of our secondary outcomes, we also found no significant difference between the arms in the total MME prescribed during the 30-day period after the new prescription. Possible explanations for this finding include inadequate statistical power for this measure or that the reduced quantity of opioid analgesics received initially by intervention arm patients was offset by higher quantity prescription reorders later. For a more complete picture, our results suggest that future studies should include the total opioid analgesic prescriptions received over a follow-up period as an outcome.

    Our findings also suggest that a default dispense quantity of 10 tablets may have unintended consequences for some patients. For those receiving schedule II opioids, we found that patients in the intervention arm were more likely than those in the control arm to receive a prescription reorder, receive prescriptions for a modestly higher quantity of opioids (ie, equivalent to approximately 5 oxycodone 5-mg tablets), and to visit the ED. For prescriptions, this result echoes a recent study that reported a default quantity of 10 tablets led to a higher mean quantity initially prescribed than a default quantity of 12 tablets.19 For ED visits, our finding was unanticipated and may be due to some patients seeking medical attention for undertreated pain. Overall, these results may be the result of, at least in part, the differences in pain condition and severity between patients receiving stronger vs weaker opioid analgesics. Further study is needed to identify optimal default quantities and our results additionally suggest that stratification by schedule is warranted.

    Limitations

    This study has several limitations. First, although the trial was conducted at multiple sites, they were all within a single medical center and so the results may not be generalizable to other settings. In particular, the preexisting EHR configuration for opioid analgesic orders was set with a mixture of defaults and no defaults and so the control group may not reflect what a control group might be at another institution. Second, we were only able to obtain data from within our medical center and so outside prescriptions and visits were not captured, resulting in missingness of outcome data that may be nonrandom given imbalances in patient and site characteristics by study arm. Therefore, we may have underestimated the frequency of opioid analgesic reorders and health service use. This limitation may also bias the study findings if patients in one arm are more likely to obtain follow-up care within our medical center than patients in the other arm. Third, because our main data source was the EHR, we do not have information on whether prescriptions were dispensed or on patient-oriented outcomes, such as pain, functioning, or quality of life. Fourth, we used cluster randomization and a DID method with covariate adjustment to account for measured and unmeasured differences between the arms; however, there may be residual differences that bias our results. Fifth, a small number of prescribers (5.3%) wrote a prescription at both an intervention site and at a control site, resulting in contamination. However, this factor would be expected to bias our results toward a negative finding. Sixth, we selected a single, uniform, default dispense quantity of 10 tablets; a lower or higher default quantity or multiple default quantities depending on the medication order or setting, may have led to different results.

    Conclusions

    In this randomized clinical trial, a uniform reduced default dispense quantity of 10 tablets for new opioid analgesic prescriptions decreased the quantity prescribed. Although the effects were modest, our findings support the efficacy of modifying EHR prescribing defaults in reducing opioid analgesic prescribing, which remains appealing owing to ease of implementation.

    Back to top
    Article Information

    Accepted for Publication: March 4, 2021.

    Published: April 22, 2021. doi:10.1001/jamanetworkopen.2021.7481

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Bachhuber MA et al. JAMA Network Open.

    Corresponding Author: Marcus A. Bachhuber, MD, MSHP, Section of Community and Population Medicine, Louisiana State University Health Sciences Center–New Orleans, 533 Bolivar St, 5th Flr, New Orleans, LA 70112 (marcus.bachhuber@gmail.com).

    Author Contributions: Dr Bachhuber 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: Bachhuber, Nash, Southern, Berger, Schepis, Thakral, Cunningham.

    Acquisition, analysis, or interpretation of data: Bachhuber, Nash, Heo, Berger, Cunningham.

    Drafting of the manuscript: Bachhuber, Nash.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Bachhuber, Nash, Heo.

    Obtained funding: Bachhuber.

    Administrative, technical, or material support: Bachhuber, Schepis.

    Supervision: Bachhuber, Nash, Berger, Cunningham.

    Conflict of Interest Disclosures: Dr Bachhuber reported receiving grants from the National Institutes of Health during the conduct of the study. Dr Heo reported receiving personal fees from HealthCorps. Dr Cunningham reported receiving grants from National Institute of Health during the conduct of the study. No other disclosures were reported.

    Funding/Support: This study was supported by the National Institute on Drug Abuse (grant K08DA043050) and the National Center for Advancing Translational Sciences (CTSA grant UL1TR001073) of the National Institutes of Health. Dr Cunningham is also supported by the National Institute on Drug Abuse (K24DA036955). Drs Cunningham and Nash are supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health (grant P30AI124414).

    Role of the Funder/Sponsor: The funding organizations 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.

    Data Sharing Statement: See Supplement 3.

    Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

    References
    1.
    Bohnert  AS, Valenstein  M, Bair  MJ,  et al.  Association between opioid prescribing patterns and opioid overdose-related deaths.   JAMA. 2011;305(13):1315-1321. doi:10.1001/jama.2011.370 PubMedGoogle ScholarCrossref
    2.
    Fulton-Kehoe  D, Sullivan  MD, Turner  JA,  et al.  Opioid poisonings in Washington State Medicaid: trends, dosing, and guidelines.   Med Care. 2015;53(8):679-685. doi:10.1097/MLR.0000000000000384 PubMedGoogle ScholarCrossref
    3.
    Lewis  ET, Cucciare  MA, Trafton  JA.  What do patients do with unused opioid medications?   Clin J Pain. 2014;30(8):654-662. doi:10.1097/01.ajp.0000435447.96642.f4 PubMedGoogle ScholarCrossref
    4.
    Centers for Disease Control and Prevention (CDC).  Adult use of prescription opioid pain medications—Utah, 2008.   MMWR Morb Mortal Wkly Rep. 2010;59(6):153-157.PubMedGoogle Scholar
    5.
    Kennedy-Hendricks  A, Gielen  A, McDonald  E, McGinty  EE, Shields  W, Barry  CL.  Medication sharing, storage, and disposal practices for opioid medications among US Adults.   JAMA Intern Med. 2016;176(7):1027-1029. doi:10.1001/jamainternmed.2016.2543 PubMedGoogle ScholarCrossref
    6.
    McCabe  SE, West  BT, Boyd  CJ.  Leftover prescription opioids and nonmedical use among high school seniors: a multi-cohort national study.   J Adolesc Health. 2013;52(4):480-485. doi:10.1016/j.jadohealth.2012.08.007 PubMedGoogle ScholarCrossref
    7.
    Hall  AJ, Logan  JE, Toblin  RL,  et al.  Patterns of abuse among unintentional pharmaceutical overdose fatalities.   JAMA. 2008;300(22):2613-2620. doi:10.1001/jama.2008.802 PubMedGoogle ScholarCrossref
    8.
    Center for Behavioral Health Statistics and Quality. Behavioral Health Trends in the United States: Results From the 2014 National Survey on Drug Use and Health (HHS Publication No. SMA 15-4927, NSDUH Series H-50). Center for Behavioral Health Statistics and Quality; 2015.
    9.
    Voepel-Lewis  T, Wagner  D, Tait  AR.  Leftover prescription opioids after minor procedures: an unwitting source for accidental overdose in children.   JAMA Pediatr. 2015;169(5):497-498. doi:10.1001/jamapediatrics.2014.3583 PubMedGoogle ScholarCrossref
    10.
    Bailey  JE, Campagna  E, Dart  RC; RADARS System Poison Center Investigators.  The underrecognized toll of prescription opioid abuse on young children.   Ann Emerg Med. 2009;53(4):419-424. doi:10.1016/j.annemergmed.2008.07.015 PubMedGoogle ScholarCrossref
    11.
    Brat  GA, Agniel  D, Beam  A,  et al.  Postsurgical prescriptions for opioid naive patients and association with overdose and misuse: retrospective cohort study.   BMJ. 2018;360:j5790. doi:10.1136/bmj.j5790 PubMedGoogle ScholarCrossref
    12.
    Vélez-Díaz-Pallarés  M, Pérez-Menéndez-Conde  C, Bermejo-Vicedo  T.  Systematic review of computerized prescriber order entry and clinical decision support.   Am J Health Syst Pharm. 2018;75(23):1909-1921. doi:10.2146/ajhp170870 PubMedGoogle ScholarCrossref
    13.
    Garg  AX, Adhikari  NK, McDonald  H,  et al.  Effects of computerized clinical decision support systems on practitioner performance and patient outcomes: a systematic review.   JAMA. 2005;293(10):1223-1238. doi:10.1001/jama.293.10.1223 PubMedGoogle ScholarCrossref
    14.
    Kruse  CS, Ehrbar  N.  Effects of computerized decision support systems on practitioner performance and patient outcomes: systematic review.   JMIR Med Inform. 2020;8(8):e17283. doi:10.2196/17283 PubMedGoogle Scholar
    15.
    Chiu  AS, Freedman-Weiss  MR, Jean  RA, Cohen  E, Yoo  PS.  No refills: the durable impact of a multifaceted effort by surgical trainees to minimize the prescription of postoperative opioids.   Surgery. 2019;166(5):758-763. doi:10.1016/j.surg.2019.05.021 PubMedGoogle ScholarCrossref
    16.
    Chiu  AS, Jean  RA, Hoag  JR, Freedman-Weiss  M, Healy  JM, Pei  KY.  Association of lowering default pill counts in electronic medical record systems with postoperative opioid prescribing.   JAMA Surg. 2018;153(11):1012-1019. doi:10.1001/jamasurg.2018.2083 PubMedGoogle ScholarCrossref
    17.
    Delgado  MK, Shofer  FS, Patel  MS,  et al.  Association between electronic medical record implementation of default opioid prescription quantities and prescribing behavior in two emergency departments.   J Gen Intern Med. 2018;33(4):409-411. doi:10.1007/s11606-017-4286-5 PubMedGoogle ScholarCrossref
    18.
    Zivin  K, White  JO, Chao  S,  et al.  Implementing electronic health record default settings to reduce opioid overprescribing: a pilot study.   Pain Med. 2019;20(1):103-112. doi:10.1093/pm/pnx304 PubMedGoogle ScholarCrossref
    19.
    Montoy  JCC, Coralic  Z, Herring  AA, Clattenburg  EJ, Raven  MC.  Association of default electronic medical record settings with health care professional patterns of opioid prescribing in emergency departments: a randomized quality improvement study.   JAMA Intern Med. 2020;180(4):487-493. doi:10.1001/jamainternmed.2019.6544 PubMedGoogle ScholarCrossref
    20.
    Smalley  CM, Willner  MA, Muir  MR,  et al.  Electronic medical record-based interventions to encourage opioid prescribing best practices in the emergency department.   Am J Emerg Med. 2020;38(8):1647-1651. doi:10.1016/j.ajem.2019.158500PubMedGoogle ScholarCrossref
    21.
    Blutinger  EJ, Shofer  FS, Meisel  Z, Perrone  J, Engel-Rebitzer  E, Delgado  MK.  Variability in emergency department electronic medical record default opioid quantities: a national survey.   Am J Emerg Med. 2019;37(10):1963-1964. doi:10.1016/j.ajem.2019.03.023 PubMedGoogle ScholarCrossref
    22.
    Meisenberg  BR, Grover  J, Campbell  C, Korpon  D.  Assessment of opioid prescribing practices before and after implementation of a health system intervention to reduce opioid overprescribing.   JAMA Netw Open. 2018;1(5):e182908. doi:10.1001/jamanetworkopen.2018.2908 PubMedGoogle Scholar
    23.
    Santistevan  JR, Sharp  BR, Hamedani  AG, Fruhan  S, Lee  AW, Patterson  BW.  By default: the effect of prepopulated prescription quantities on opioid prescribing in the emergency department.   West J Emerg Med. 2018;19(2):392-397. doi:10.5811/westjem.2017.10.33798 PubMedGoogle ScholarCrossref
    24.
    Schwartz  GD, Harding  AM, Donaldson  SR, Greene  SL.  Modifying emergency department electronic prescribing for outpatient opioid analgesia.   Emerg Med Australas. 2019;31(3):417-422. doi:10.1111/1742-6723.13192 PubMedGoogle ScholarCrossref
    25.
    Zwank  MD, Kennedy  SM, Stuck  LH, Gordon  BD.  Removing default dispense quantity from opioid prescriptions in the electronic medical record.   Am J Emerg Med. 2017;35(10):1567-1569. doi:10.1016/j.ajem.2017.04.002 PubMedGoogle ScholarCrossref
    26.
    McCarthy  DM, Kim  HS, Hur  SI,  et al.  Patient-reported opioid pill consumption after an ED visit: how many pills are people using?   Pain Med. 2021;22(2):292-302. PubMedGoogle ScholarCrossref
    27.
    Shi  R, Quinones  A, Bair  J,  et al.  Patient utilization of prescription opioids after discharge from the emergency department.   Am J Emerg Med. 2020;38(8):1568-1571. PubMedGoogle ScholarCrossref
    28.
    Daoust  R, Paquet  J, Cournoyer  A,  et al.  Quantity of opioids consumed following an emergency department visit for acute pain: a Canadian prospective cohort study.   BMJ Open. 2018;8(9):e022649. doi:10.1136/bmjopen-2018-022649 PubMedGoogle Scholar
    29.
    Bicket  MC, Long  JJ, Pronovost  PJ, Alexander  GC, Wu  CL.  Prescription opioid analgesics commonly unused after surgery: a systematic review.   JAMA Surg. 2017;152(11):1066-1071. doi:10.1001/jamasurg.2017.0831 PubMedGoogle ScholarCrossref
    30.
    Von Korff  M, Saunders  K, Thomas Ray  G,  et al.  De facto long-term opioid therapy for noncancer pain.   Clin J Pain. 2008;24(6):521-527. doi:10.1097/AJP.0b013e318169d03b PubMedGoogle ScholarCrossref
    31.
    Hooten  WM, St Sauver  JL, McGree  ME, Jacobson  DJ, Warner  DO.  Incidence and risk factors for progression from short-term to episodic or long-term opioid prescribing: a population-based study.   Mayo Clin Proc. 2015;90(7):850-856. doi:10.1016/j.mayocp.2015.04.012 PubMedGoogle ScholarCrossref
    32.
    Bachhuber  MA, Nash  D, Southern  WN,  et al.  Reducing the default dispense quantity for new opioid analgesic prescriptions: study protocol for a cluster randomised controlled trial.   BMJ Open. 2018;8(4):e019559. doi:10.1136/bmjopen-2017-019559 PubMedGoogle Scholar
    33.
    US Department of Health and Human Services.  National Pain Strategy: A Comprehensive Population Health-Level Strategy for Pain. US Department of Health and Human Services; 2016.
    34.
    Healthcare Cost and Utilization Project (HCUP). Beta Clinical Classifications Software (CCS) for ICD-10-CM/PCS. Agency for Healthcare Research and Quality. Accessed January 29, 2018. https://www.hcup-us.ahrq.gov/toolssoftware/ccs10/ccs10.jsp
    35.
    Tian  TY, Zlateva  I, Anderson  DR.  Using electronic health records data to identify patients with chronic pain in a primary care setting.   J Am Med Inform Assoc. 2013;20(e2):e275-e280. doi:10.1136/amiajnl-2013-001856 PubMedGoogle ScholarCrossref
    36.
    Greevy  R, Lu  B, Silber  JH, Rosenbaum  P.  Optimal multivariate matching before randomization.   Biostatistics. 2004;5(2):263-275. doi:10.1093/biostatistics/5.2.263 PubMedGoogle ScholarCrossref
    37.
    Huber  PJ. The behavior of maximum likelihood estimates under nonstandard conditions. Proc Fifth Berkeley Symp Math Statist Prob. 1967;1:221-223.
    38.
    White  H.  A Heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity.   Econometrica. 1980;48(4):817-838. doi:10.2307/1912934 Google ScholarCrossref
    ×