Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Oct 1.
Published in final edited form as: Ann Emerg Med. 2018 Jul 24;72(4):389–400.e1. doi: 10.1016/j.annemergmed.2018.06.003

National Variation in Opioid Prescribing and Risk of Prolonged Use for Opioid-Naïve Patients Treated in the Emergency Department for Ankle Sprains

M Kit Delgado 1,3,4,5,6, Yanlan Huang 2,6, Zachary Meisel 1,5,6, Sean Hennessy 3,6, Michael Yokell 1, Daniel Polsky 2,6,7, Jeanmarie Perrone 1,6
PMCID: PMC6319263  NIHMSID: NIHMS989897  PMID: 30054152

Abstract

Objective:

To inform opioid stewardship efforts, we describe the variation in emergency department (ED) opioid prescribing for a common minor injury, ankle sprains, and determine the association between initial opioid prescription intensity and transition to prolonged opioid use.

Methods:

We analyzed 2011–2015 U.S. private insurance claims (Optum Clinformatics DataMart) for ED treated ankle sprains among opioid-naive patients aged ≥ 18 years. We determined: 1) the patient- and state-level variation in the opioid prescription rate and characteristics; and 2) the risk-adjusted association between total morphine milligram equivalents (MME) of the prescription and transition to prolonged use (filling 4 or more opioid prescriptions 30–180 days after the index visit).

Results:

30,832 patients met inclusion criteria. Of these, 25.1% received an opioid prescription with a median total MME of 100 (IQR 75–113), tablet quantity of 15 (IQR 12–20), and days supplied of 3 (IQR 2–4). State-level prescribing rates ranged from 2.6% in North Dakota to 40.0% in Arkansas. Among patients who received a total MME of > 225 (equivalent to > 30 tabs of oxycodone 5 mg), the adjusted rate of prolonged opioid use was 4.9% (95% CI: 1.8–8.1%) compared with 1.1% (95% CI: 0.7–1.5%) among those who received at total MME of 75 or and 0.5% (95% CI: 0.4–0.6%) among those who did not fill an opioid.

Conclusion:

Opioid prescribing for ED patients treated for ankle sprains is common and highly variable. Although infrequent in this population, prescriptions greater than 225 MME were associated with higher rates of prolonged opioid use. This is concerning since these prescriptions could still fall within 5- or 7-day supply limit policies aimed at promoting safer opioid prescribing.

INTRODUCTION

Background

The U.S. opioid epidemic claimed 42,249 lives in 2016.1 In response, many states in the country and some health insurers and pharmacy chains have enacted policies limiting the duration of new opioid prescriptions for acute pain conditions.2–4 In the early part of 2018, Congress introduced the Comprehensive Addiction and Recovery Act (CARA) 2.0, a bipartisan bill that includes a three-day supply limit on all new opioid prescriptions. The goal of opioid stewardship and prescription limit policies is to reduce the potential for transition to prolonged use, potential misuse, as well as the diversion of unused tablets. Despite recent policy action, the scientific evidence base on the threshold prescription intensity associated increased risk of transition to prolonged use and misuse is limited.

Importance

Prior studies have found an association between increasing opioid prescription duration or intensity and transition to prolonged use, but were not designed to account for differences in the indication for the initial prescription.5–7 This is problematic because the onset of some pain conditions, such as acute lumbar radiculopathy, often leads to chronic pain. Therefore, it cannot be determined whether prolonged use is due to the development of opioid dependence following the initial prescription or to the independent development of increased pain sensitivity (hyperalgesia), a chronic pain syndrome, or other factors. A second issue with current prescription limits is that basing policy on days supplied is imprecise. A common dosing order for oxycodone 5 mg is to take 1–2 tablets every 4–6 hours as needed for pain. With these instructions a 7-day prescription could vary anywhere from 1 tablet to 84 tablets or 7.5 to 630 morphine milligram equivalents.8 Therefore, the impact of policies to limit the days supplied on reducing the total number of opioid tablets and morphine milligram equivalents entering the community remains unclear, particularly if clinicians respond by writing for the same number of tablets but to be taken over fewer days.

Goals of this Investigation

To guide opioid stewardship efforts by benchmarking variation in prescribing and to generate clearer evidence on the risks of features of initial opioid prescriptions, we analyzed commercial insurance claims from opioid-naïve patients treated in the emergency department (ED) for ankle sprains. We chose this indication and setting for three reasons. First, ankle sprains are common minor injuries for which pain rapidly improves within two weeks.9 Therefore, any risk of subsequent opioid misuse is likely to outweigh the treatment benefit given that non-opioid analgesics have been found to be equally effective for acute pain management.10,11 Second, although initial opioid prescriptions written in the ED are more consistent with current prescribing guidelines than those from non-ED settings and are typically for smaller quantities, (and thus less likely to progress to long-term use), EDs remain a common source of acute opioid prescriptions.12,13 Finally, there is still significant variability in opioid prescribing by emergency physicians and patient assignment is essentially random.6,13 This both represents an opportunity for quality improvement and also reduces the potential bias of downstream prescriptions being associated with characteristics of the initial prescriber.

We sought to determine the degree of patient- and state-level variation in opioid prescribing in this setting and the association between the variation in morphine milligram equivalents prescribed and transition to prolonged opioid use. We hypothesized variation in prescribing would be substantial and that larger opioid prescriptions would be associated with higher likelihood of prolonged use.

METHODS

Study Population

We employed a cohort study design in a retrospective analysis of the Clinformatics Data Mart Database (OptumInsight), which is comprised of administrative claims of 13 million privately insured enrollees throughout the U.S.14–16 This population is similar to the US population of commercially insured people in age, race or ethnicity, and sex and has a median continuous enrollment period of 24 months.16

We identified all index ED encounters for ankle sprains in patients aged ≥ 18 years old that occurred from January 1, 2011 to December 31, 2015. The index visit was defined as the earliest visit in which the enrollee had a provider claim by an emergency physician with a place of service designation in the emergency department and a diagnosis code of ankle sprain (ICD-9 code 845.00–09 and ICD10 S93.4). We excluded patients who had any other injury diagnoses as defined by State and Territorial Injury Prevention Directors Association consensus definitions,17–19 and those who had recurrent visits for ankle sprain during the study period to decrease the likelihood that future prescriptions were associated with a more severe initial presentation. To restrict the study population to those who are likely to be prescription opioid-naïve, we excluded all patients who were not continuously enrolled for 6 months prior to the index ED visit and those who had filled an opioid prescription in these prior 6 months.6,12

Definition of Opioid Prescriptions

We identified prescription claims corresponding to an opioid according to National Drug Codes (excluding methadone and non-tablet formulations) filled within 3 days of the index ED visit. We excluded any encounters for patients who had intervening medical claims for ankle sprains within 3 days of the ED visit.

Outcomes

The primary outcome was opioid prescription rate defined as the proportion of adult opioid-naïve patients treated for ankle sprains that filled an opioid prescription with 3 days of the ED visit. Secondary outcomes included characteristics of the initial prescription and continued opioid use beyond the initial prescription.

Characteristics of the initial prescription included number of tablets, days supplied, and morphine milligram equivalents (MME) per prescription.20 The days supplied on the pharmacy prescription claims are entered at the point of sale by the pharmacist based on dosing instructions and number of tablets dispensed.21

We measured prolonged opioid use in the 30–180 days following the index ED visit given pain from ankle sprains rapidly improves within 2 weeks.9 We defined prolonged use as filling 4 or more subsequent opioid prescriptions during this time period,22,23 and used alternative specifications in our sensitivity analyses. Limiting the follow-up period to six months captures high-risk individuals who drop out of the dataset by virtue of losing employment, and thus insurance coverage, due to an opioid use disorder if longer follow up periods were used.24,25 Indeed, in our final study sample the insurance enrollment attrition rate between 6 and 12 months of follow-up was 1.5% (95% CI: 0.4%, 2.5%) higher among those who were prescribed opioids (16.7%) compared with those who did not fill opioids (15.2%). To further validate our outcome, we tabulated the most common first listed medical diagnosis codes within the 7 days prior to new opioid prescription fills 30–180 days after the index encounter. This analysis allowed us to explore the extent to which new prescriptions could be attributed to persistent ankle pain rather than another condition completely unrelated to the index ankle sprain.

Patient Covariates

We extracted information on patients’ age, sex, highest level of education, and race. We also identified the patients’ Elixhauaser comorbidities, as well as diagnosis codes for drug abuse, alcohol abuse, depression, and psychoses as gathered from diagnosis codes on the index ED visit and any medical claims filed in the previous 6 months.26

Statistical Analysis

First, we tabulated patient characteristics according to whether an opioid prescription was filled or not. Among those who filled opioids, we stratified patient characteristics according to whether a prescription was above or below the median total MME prescribed in the sample. We compared the magnitude of differences in patient and prescription characteristics between groups using 95% confidence intervals for the differences.

Second, we described the variation in the opioid prescription rate and characteristics at the patient- and state-level and over time. Prescriptions were aggregated by state according the state associated with the prescribers’ de-identified National Provider Identifier (NPI) number.

Next, we calculated the patients’ expected probability of receiving an opioid prescription using a logistic regression model adjusting for age, sex, highest level of education, race, the total number of Elixhauser comorbidities, a history of drug abuse, alcohol abuse, depression, or psychosis, and year. We then estimated observed-to-expected state-level prescribing ratios with 95% confidence intervals, with values over 1 indicating patients in that state were more likely than expected to fill opioid prescriptions and less than 1 indicating patients in states that were less likely than expected to fill opioid prescriptions.27 For these state-level analyses we excluded states with less than 25 patients in the study sample.

We then quantified the number of tablets that would be prevented from entering the community from reducing excessive variation by: 1) reducing all states with above-median opioid prescribing rates to the median rate for ankle sprains; and 2) leveling all above-median opioid prescriptions supplies for ankle sprains to the median.

Finally, we quantified the association between the initial MME prescribed among those who filled a prescription and prolonged use. MME cutoffs were selected based on common tablet quantities dispensed for the most potent opioid prescribed, oxycodone 5 mg: 1–75 MME (<10 tabs), 76–150 (11–20 tabs), 151–225 MME (21–30 tabs), and >225 MME (>30 tabs). We quantified the association using a logistic regression model adjusting for age, sex, highest level of education, race, the total number of Elixhauser comorbidities, a history of drug abuse, alcohol abuse, depression, or psychosis, year, and state. We then performed sensitivity analyses to assess the robustness of our findings, including: 1) using tablets dispensed (adjusted for MME/tablet), and days supplied as the independent measure of interest; and 2) limiting the exposure to just prescriptions for hydrocodone and oxycodone; and 3) considering alternative measures of prolonged use based on the number of prescriptions filled in the follow-up window. The University of Pennsylvania Institutional Review Board (IRB) determined that this study was eligible for IRB exemption. Stata 14.1 (StataCorp, College Station, TX) and SAS 9.4 (SAS Institute, Cary, NC) were used to conduct the statistical analyses.

RESULTS

Characteristics of the Study Subjects

Our sample consisted of 30,832 patients with an ED visit for an isolated injury of ankle sprain who had not filled an opioid prescription in the prior 6 months (Figure 1). The overall opioid prescribing rate in this sample was 25.1% (n=7,739). Among the 25,849 patients who had continuous enrollment for 6 months after the ED visit, the prescribing rate was similar (25.0%, n=6,463). Demographic and clinical characteristics were similar between those were prescribed an opioid vs. those who were not as well as those who received above vs. below median (100) total MME prescriptions (Table 1).

Figure 1.

Figure 1

Open in a new tab

Study Sample

Table 1:

Characteristics of Opioid-Naïve Patients Treated for Ankle Sprains

Patient
Characteristics		   No opioid
(n =19,386)		 Filled opioid
(n= 6,463)		 Filled opioid,
< 100 MME
(n = 4,576) 		Filled opioid,
> 100 MME
(n = 1,887)
Median Age (IQR) 39 (25–56) 38 (28–52) 38 (25–52) 40 (29–52)
Female (%) 10,915 (56.3) 3,392 (52.5) 2,492 (54.5) 900 (47.7)
Race/ethnicity (%)
 Non-Hispanic White 13,165 (67.9) 4,595 (71.1) 3,235 (70.7) 1,360 (72.1)
 Hispanic White 2,073 (10.7) 652 (10.1) 468 (10.2) 184 (9.8)
 Black 2,483 (12.8) 826 (12.8) 599 (13.1) 227 (12.0)
 Asian 554 (2.9) 116 (1.8) 90 (2.0) 26 (1.4)
 Unknown 470 (2.4) 104 (1.6) 70 (1.5) 34 (1.8)
Highest education level (%)
 <12th grade 88 (0.5) 33 (0.5)  23 (0.5) 10 (0.5)
 12th grade 5,818 (30.0) 1,960 (30.3) 1,389 (30.4) 571 (30.3)
 <Bachelor’s 9,723 (50.2) 3,346 (51.8) 2,392 (52.3) 954 (50.6)
 ≥ Bachelor’s 3,287 (17.0) 1,020 (15.8) 702 (15.3) 318 (16.9)
Unknown 470 (2.4) 104 (1.6) 70 (1.5) 34 (1.8)
Mean No. of Elixhauser Comorbidities (SD) 0.98 (1.62) 0.94 (1.52) 0.93 (1.50) 0.97 (1.57)
Hypertension (%) 4,206 (21.7) 1,459 (22.6) 1,004 (21.9) 455 (24.1)
Depression (%) 1,770 (9.1) 629 (9.7) 452 (9.9) 177 (9.4)
Diabetes, Uncomplicated (%) 1,708 (8.8) 579 (9.0) 401 (8.8) 178 (9.4)
Chronic Pulmonary Disease (%) 1,829 (9.4) 563 (8.7) 414 (9.0) 149 (7.9)
Hypothyroidism (%) 1,484 (7.7) 454 (7.0) 321 (7.0) 133 (7.0)
Obesity (%) 943 (4.9) 367 (5.7) 252 (5.5) 115 (6.1)
Cardiac Arrhythmias (%) 941 (4.9) 250 (3.9) 172 (3.8) 78 (4.1)
Diabetes, Complicated (%) 449 (2.3) 147 (2.3) 101 (2.2) 46 (2.4)
Fluid and Electrolyte Disorders (%) 503 (2.6) 139 (2.2) 92 (2.0) 47 (2.5)
Renal Failure (%) 471 (2.4) 136 (2.1) 89 (1.9) 47 (2.5)
Alcohol Abuse (%) 180 (0.9) 73 (1.1) 51 (1.1) 22 (1.2)
Drug Abuse (%) 203 (1.0) 73 (1.1) 55 (1.2) 18 (1.0)
         
Prescription characteristics
Drug name (%)
 Hydrocodone (%) 4,197 (64.9) 3,267 (71.4) 930 (49.3)
 Oxycodone (%) 928 (14.4) 255 (5.6) 673 (35.7)
 Codeine (%) 353 (5.5) 272 (5.9) 81 (4.3)
 Tramadol (%) 1,050 (16.2) 782 (17.1) 268 (14.2)
 Other (%) 8 (0.1) 2 (0.0) 6 (0.3)
Median number of tablets (IQR) 16 (12–20) 15 (12–20) 24 (20–30)
Median days supplied (IQR) 3 (2–4) 3 (2–4) 4 (3–5)
Open in a new tab

Main Results

Hydrocodone was the most commonly prescribed opioid (64.9%), followed by tramadol (16.2%), oxycodone (14.4%), and codeine (5.5%). The median number of tablets prescribed for all opioid prescriptions was 16 with an interquartile range of 12 to 20 tablets. The median total MME was 100 (IQR 75–112.5) and median days supplied was 3 (IQR 2–4). However, only 5% of those who were prescribed opioids were given more than 30 tablets or greater than 200 MME total (Figure 2). The overall prescription rate declined modestly over the study period from 28.1% (95% CI: 27.1–29.1%) in 2011 to 20.4% (95% CI: 19.2–22.6%) in 2015 (Figure 3).

Figure 2.

Figure 2.

Open in a new tab

These histograms display the patient-level variation in the quantity prescribed for opioid-naïve patients with ankle sprains.A) Displays the total morphine milligram equivalents (MME) per prescription of all patients who received opioid prescriptions (n = 6,464, median = 100). B) Displays the number of tablets prescribed for those who received hydrocodone (n = 4,197, median = 15). C) Displays the number of tablets prescribed for those who received oxycodone (n = 928, median = 16).

Figure 3.

Figure 3.

Open in a new tab

This figure displays the proportion of opioid-naïve patients treated in the ED for ankle sprains who were prescribed opioids according to year of ED visit. he whiskers indicate 95% confidence intervals.

State-specific prescription rates varied by a factor of 14.2, ranging from 40.0% in Arkansas to 2.8% in North Dakota with a national median prescribing rate of 24.1% (Figure 4). There was less state-level variation in the MME per prescription (median 98.9 [range 77.5–136,3]) and quantity supplied per patient (median 18.2 [range 13.6–40.0]) for those who received prescriptions. After adjustment for case-mix and volume, the prescription rate variation persisted in 2014–15. States in the southern U.S. accounted for most of the higher than expected prescribing rates (Figure 4, eFigure 1).

Figure 4.

Figure 4.

Open in a new tab

The map displays the state-level variation in the emergency department opioid prescribing rate for ankle sprains in 2014–15 among patients who were opioid-naive.The median state-level prescribing rate during these years was 21.3%. The observed prescribed rate is displayed within each state.States with higher than expected prescribing rates based on case mix are highlighted in red and states with lower than expected prescribing rates are shown in blue (see eFigure 1 for caterpillar plot). Case-mix was adjusted for age, sex, race, ethnicity, education, comorbid conditions, and year using multivariate logistic regression.

There were a total of 143,064 opioid tablets prescribed for those who filled prescriptions in the study sample. Just reducing the excess variation in the state-level prescription rates by bringing states with above average prescribing rates to the median of 24.1% would have resulted in 18,260 fewer opioid tablets being prescribed. On the other hand, reducing the excess variation in prescription quantities by bringing prescriptions above the median to the median of 16 tablets would have resulted in an even bigger reduction, 32,177 less tablets prescribed.

Prolonged opioid use (4 or more new opioid prescriptions 30 days after the initial prescription) was slightly higher among those prescribed any opioid, 0.73% (95%: 0.63%−0.84%), versus those who were not prescribed opioids, 0.50% (95% CI: 0.40–0.59%). Only small proportion of subsequent prescriptions among those who went on to high-risk prolonged use were due to first listed diagnoses of non-traumatic joint disorders (7.5%) and sprains and strains (5.8%) of any kind. In contrast, subsequent prescriptions were more likely to be associated with other diagnosis codes (>80%) such as connective tissue disease (7.8%), back problems (5.8%), medical examination (3.6%), headache (2.6%), abdominal pain (2.0%), urinary tract infections (2.0%), and dozens of other conditions (eTable 1).

After multivariate risk adjustment, the associated probability of high-risk prolonged use among those who received a total of MME of 75 or less was 1.10% (95% CI: 0.72–1.53%) and remained similar for those who received up to 225 MME. However, above that threshold there was increased association with prolonged use 4.9% (95% CI: 1.8–8.1%) (Figure 5a).

Figure 5.

Figure 5.

Open in a new tab

A) Association between total morphine milligram equivalents (MME) prescribed and the probability of filling 4 or more new opioid prescriptions 30–180 days after the index visit for ankle sprain among those who were previously opioid-naive. The whiskers indicate 95% confidence intervals. The logistic regression model adjusted for age, sex, race, ethnicity, educational level, Elixhauser comorbidity score, history of drug abuse, depression, psychoses, state, and year (model area under the curve [AUC] = 0.78). In B) the population is limited to just those prescribed hydrocodone and oxycodone (model area under the curve [AUC] = 0.81). For reference, 225 MME is the equivalent of 30 tablets of oxycodone 5 mg and the baseline rate of prolonged use among those who were not prescribed opioids, was 0.5% (95% CI: 0.4–0.6%). See eAppendix for full model estimates and sensitivity analyses of model specifications and outcomes.

When restricting the population to just those prescribed the higher potency drugs, hydrocodone and oxycodone, the dose-response relationship between total MME of the initial prescription and high-risk prolonged use was stronger (Figure 5b). Those prescribed >225 total MME in this group had a 6.3% (95% CI: 2.1–10.5%) probability of developing high-risk prolonged use vs. those prescribed <75 MME (1.2 %; 95% CI: 0.7–1.6%)

Female sex, age 35–44 years old, higher comorbidity burden, and previous history of drug abuse were also independently associated with transition to high-risk prolonged use (see full model results in eTable 1). Other sensitivity analyses varying alternative ways to measure prolonged use and initial prescription intensity confirmed the robustness of our findings (eAppendix).

LIMITATIONS

There are several limitations to this study. First, unmeasured differences between patients could have accounted for the observed associations. Arguing against this possibility is the observation that, among patients who were prescribed an opioid, measured patient characteristics were not significantly different according to prescription intensity. Furthermore, limiting the population to ankle sprains reduces the possibility that continued use was due to the development of chronic pain since the vast majority of ankle sprains resolve within 2–3 weeks regardless of initial pain severity.9 This is supported by the fact we found that <14% of new prescriptions among those with prolonged use were associated with sprains, strains, and non-traumatic joint pain of any type.

Second, we made the assumption that prescriptions were written for ankle sprains by limiting this analysis to claims with ankle sprains as the only injury diagnosis. It is possible that there were prescriptions for other indications that were not coded. Relatedly, data limitations preclude us from unequivocally attributing the prescription to ED clinicians. We addressed this by excluding patients who had other intervening medical claims for ankle sprains within the 3-day attribution window.

Third, we likely underestimated the opioid prescription rate if some patients filled prescriptions by paying out of pocket instead of using health insurance.28 Finally, if data from 2017 were available and used to replicate this analysis, there may be a lower measured association with prolonged use to due to the increase in prescription drug monitoring programs (PDMPs),29,30 prescribing guidelines,31 and the explosion in the availability and use of low-cost illicit opioids.32,33 All these factors increase the likelihood of an earlier transition to illicit opioids rather than seeking repeated prescription fills for individuals developing a new opioid use disorder after exposure to a prescription opioid.34–36 Given the recent explosion in deaths from illicit opioids,37 this further highlights the importance of keeping opioid-naïve patients opioid-naïve when possible.

DISCUSSION

In this national study of 30,832 commercially insured, opioid-naive patients treated in the ED for ankle sprains in 2011–15 we found opioid prescribing was common and that there was significant patient- and state-level variation in prescribing patterns. Although declining over time, 20% of opioid-naïve ankle sprain patients were prescribed opioids in 2015, and the prescribing rate was over 10 times higher in some states versus others. This is concerning because ankle sprains are a minor, self-limited condition for which there is likely to be little clinical benefit from opioids.10

While there was significant variation in the intensity of prescriptions written, less than 5% of prescriptions were written above a clinically significant threshold of more than 225 MME. We found that prescriptions above this threshold were associated with nearly a fivefold increased probability of transition to prolonged use. The interpretation of this finding is that for approximately every 26 patients exposed to greater than 225 MME (the equivalent of more than 30 tabs of oxycodone 5 mg) instead of 75 MME or less (equivalent to 10 tablets of oxycodone 5 mg) 1 additional patient transitioned to prolonged use. While there is room for improvement in promoting opioid stewardship for this minor condition in the ED, the majority of prescriptions written were concordant with guidelines for a 3-day supply12,13,31 and had a low total MME of 100 or less. Therefore, more research is urgently needed to examine the relationship between total MME, prolonged use, and adverse events in other contexts, such as for post-operative pain, where prescriptions are much larger.12

Our findings make several novel contributions to the growing understanding of opioid prescribing and downstream risks. First, this analysis addresses limitations of previous analyses5–7 in which prolonged opioid use could be caused by the development of a new chronic pain condition at the time of index prescription. By focusing on ankle sprains, an isolated, acute injury that is readily apparent on exam and that rapidly improves within two weeks,9 we decreased the likelihood that future opioid prescriptions beginning one month after injury are related to the index condition for which the original opioid was prescribed. Nonsteroidal anti-inflammatory drugs, rather than opioids are first-line treatment for ankle sprains38,39 and are just as effective as opioids in pain reduction.10 We confirmed that the majority of subsequent prescriptions were unlikely to be related to the initial ankle sprain or chronic ankle pain. This suggests that association between larger prescriptions and increased likelihood of prolonged use could be due to other factors such as patients requesting opioids as default pain control, or the development of dependence or misuse. More research is needed to better understand the underlying mechanisms between exposure to larger prescriptions and increased downstream prescription fills. Furthermore, our study population with a median age of 39 is also younger than previous populations that have been studied.5–7 For example, a prior rigorously conducted study using Medicare data6 may underestimate the relationship between initial prescription and long term use since opioid prescriptions carry increased acute side effects for elderly patients and are generally prescribed with more caution in this population.40 Additionally, younger and middle-aged populations are at highest risk of developing a new opioid use disorder after exposure to opioid prescriptions,41,42 and for diverting unused tablets.43

While our study was not designed to evaluate the association between being prescribed any opioid and prolonged use, our findings that those prescribed opioids were more likely to go on to prolonged use compared with those who did not fill opioids is consistent with previous research,6,7,44,45 and supports the importance of keeping opioid-naïve patients opioid-naïve.

Second, our study contributes actionable evidence for policy makers, health systems, and clinicians.The significant statewide variation in ED ankle sprain opioid prescribing rates with a concentration of higher prescribing in the southern U.S. is consistent with geographic variation in prescribing for all indications.46 By limiting this analysis to a single minor condition for which opioids are not first line treatment, and adjusting for case-mix, our study suggests ample opportunity to reduce excessive prescribing. Over 140,000 opioid tablets could have been prevented from entering the community if opioids had not been prescribed for our study sample.Substantial reductions in tablets being prescribed could be accomplished with efforts aimed at decreasing excessive variation in the prescribing rate and quantity supplied.This study demonstrates how well defined prescribing indications can be used to promote benchmarks for opioid stewardship efforts.More research is needed to understand the patient, clinician, and environmental causes of this geographic variation and its contributions to the opioid epidemic.

Despite most current guidelines and prescription limit policies being written around days supplied, there is a lack of specificity as to how many tablets and morphine milligram equivalents of common opioid prescriptions such as hydrocodone or oxycodone constitutes a day’s supply.This is problematic since we demonstrate higher risk prescriptions of >225 MME could still fall within 5- or 7-day supply limit policies aimed at promoting safer opioid prescribing. The days supplied can be specified by the prescriber based on the start and end date of the prescription, or if this information is not provided, the pharmacist calculates this before claim submission based on the brand label dosing instructions.Therefore, the number of tablets dispensed per prescription is subject to manipulation and can be quite variable for the same number of days supplied.Furthermore, the number of tablets is more clinically relevant.Large numbers of tablets are commonly left over after prescriptions for acute pain and surgical procedures and are poorly secured.47–49 Decreasing leftover tablets is critical to reducing diversion and overdoses.43 A promising approach is implementing lower electronic medical record order default opioid quantities (e.g. 10 tablets), which was recently shown to significantly shift ED discharge prescribing patterns towards the default quantity.50 More research is needed to investigate whether prescribing limit policies should incorporate total MME or quantity limits.

In summary, our findings continue to support efforts to keep opioid-naïve patients opioid-naïve and to use the smallest quantities of opioid possible when clinically indicated for treatment of acute pain. Prescriptions exceeding 30 tablets of oxycodone 5mg or a total of 225 MME were associated with transition to prolonged use 30–180 days after an initial encounter for ankle sprain among those who had not previously filled an opioid prescription. Opioid prescribing for ankle sprains remains common and is highly variable at the state- and patient-level even though opioids are not first-line treatments for this condition. Given deaths from prescription and illicit opioids continue to increase, this suggests multiple opportunities for clinical, health system, and state-level interventions.

Supplementary Material

appendix

Grants and Acknowledgments

Research reported in this manuscript was supported by the National Institute on Drug Abuse and the National Institute of Child Health and Human Development of the National Institutes of Health under award numbers P30DA040500 (MKD, JM, ZM, DP) K23HD090272001 (MKD).The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.This work was also supported with funding from the Leonard Davis Institute of Health Economics at the University of Pennsylvania. The authors thank Jennifer Love, MD for her thoughtful suggestions and David Karp, MUSA for assistance with mapping software.

Meetings

This work was presented in the Plenary Session of the 2017 Society for Academic Emergency Medicine Annual Meeting in Orlando, FL on May 17, 2017.

Footnotes

Conflict of Interest Summary

Dr. Delgado reports receiving an honorarium for participating in an Expert Roundtable on Innovative Solutions for Pain Management convened by United Health Group. All other authors declare no conflicts of interest.

References

  • 1.Hedegaard HWM, Miniño AM. Drug overdose deaths in the United States, 1999–2016 Hyattsville, MD: National Center for Health Statistics;2017. [Google Scholar]
  • 2.Zezima K With drug overdoses soaring, states limit the length of painkiller prescriptions Washington Post; 2017. [Google Scholar]
  • 3.Independence Blue Cross. Independence limits new opioid prescriptions to five days [press release] June 27, 2017.
  • 4.CVS.CVS Health Responds to Nation’s Opioid Crisis [press release] September 21, 2017.
  • 5.Shah A Characteristics of Initial Prescription Episodes and Likelihood of Long-Term Opioid Use—United States, 2006–2015. MMWR Morbidity and Mortality Weekly Report 2017;66. [DOI] [PMC free article] [PubMed]
  • 6.Barnett ML, Olenksi AR, Jena AB. Opioid Prescribing by Emergency Physicians and Risk of Long-Term Use. The New England Journal of Medicine 2017;376(19):1896. [DOI] [PubMed] [Google Scholar]
  • 7.Deyo RA, Hallvik SE, Hildebran C, et al. Association between initial opioid prescribing patterns and subsequent long-term use among opioid-naïve patients: A statewide retrospective cohort study. Journal of General Internal Medicine 2017;32(1):21–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Days Supply Question: Norco 7.5 - Take 1–2 tablets Every 4–6 Hours #40. Pharmacy-tech Resources.Com. https://www.pharmacy-tech-resources.com/days-supply-question-norco-75-take-12-tablets-every-46-hours-40.html. Accessed June 15, 2017.
  • 9.van Rijn RM, Van Os AG, Bernsen RM, Luijsterburg PA, Koes BW, Bierma-Zeinstra SM. What is the clinical course of acute ankle sprains? A systematic literature review. The American Journal of Medicine 2008;121(4):324–331. e327. [DOI] [PubMed] [Google Scholar]
  • 10.Chang AK, Bijur PE, Esses D, Barnaby DP, Baer J. Effect of a Single Dose of Oral Opioid and Nonopioid Analgesics on Acute Extremity Pain in the Emergency Department: A Randomized Clinical Trial. JAMA 2017;318(17):1661–1667. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Dukakis A Colorado ER Docs Are Testing Non-Addictive Alternatives To Opioids. Colorado Public Radio 2017.
  • 12.Jeffery MM, Hooten WM, Hess EP, et al. Opioid Prescribing for Opioid-Naive Patients in Emergency Departments and Other Settings: Characteristics of Prescriptions and Association With Long-Term Use. Annals of Emergency Medicine 2017. [DOI] [PMC free article] [PubMed]
  • 13.Hoppe JA, Nelson LS, Perrone J, Weiner SG, in the Emergency POS. Opioid prescribing in a cross section of US emergency departments. Annals of emergency medicine 2015;66(3):253–259. e251. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Wallace PJ, Shah ND, Dennen T, Bleicher PA, Crown WH. Optum Labs: building a novel node in the learning health care system. Health Affairs 2014;33(7):1187–1194. [DOI] [PubMed] [Google Scholar]
  • 15.Optum.com. Clinformatics Data Mart. 2014; https://www.optum.com/content/dam/optum/resources/productSheets/Clinformatics_for_Data_Mart.pdf. Accessed October 15, 2017.
  • 16.Optum.com. Data Assets Chart Sheet. [Presentation]. 2015; https://www.optum.com/content/dam/optum/resources/productSheets/5302_Data_Assets_Chart_Sheet_ISPOR.pdf. Accessed September 27, 2017.
  • 17.Workgroup IS. Consensus Recommendations for Using Hospital Discharge Data for Injury Surveillance Marietta, GA: 2003. [Google Scholar]
  • 18.Quality AfHRa. Injury ICD-9-CM diagnosis reported on record. Healthcare Cost and Utilization Project (HCUP) 2008; http://www.hcup-us.ahrq.gov/db/vars/siddistnote.jsp?var=injury. Accessed October 1, 2017.
  • 19.Quality AfHRa. Injury ICD-10-CM diagnosis reported on record. HCUP Central Distributor SASD Description of Data Elements - All States 2008; https://www.hcup-us.ahrq.gov/db/vars/sasddistnote.jsp?var=i10_injury. Accessed October 1, 2017.
  • 20.Centers for Disease Control and Prevention. CDC compilation of opioid analgesic formulations with morphine milligram equivalent conversion factors, 2015 version. Available at: http://pdmpassist.org/pdf/BJA_performance_measure_aid_MME_conversion.pdf. Accessed September 17, 2016.
  • 21.Personal communication, Schaeffer Jamie, Vice President, Health Informatics, Optum July 31, 2017. [Google Scholar]
  • 22.Clarke H, Soneji N, Ko DT, Yun L, Wijeysundera DN. Rates and risk factors for prolonged opioid use after major surgery: population based cohort study. BMJ 2014;348:g1251. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Johnson SP, Chung KC, Zhong L, et al. Risk of prolonged opioid use among opioid-naive patients following common hand surgery procedures. The Journal of Hand Surgery 2016;41(10):947–957. e943. [DOI] [PubMed] [Google Scholar]
  • 24.Krueger AB. “Where Have All the Workers Gone? unpublished, Princeton University and NBER, October 2016;4. [Google Scholar]
  • 25.Gillespie P. The opioid crisis is draining America of workers. CNN Money.
  • 26.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Medical Care 1998;36(1):8–27. [DOI] [PubMed] [Google Scholar]
  • 27.Hosmer DW, Lemeshow S. Confidence interval estimates of an index of quality performance based on logistic regression models. Statistics in medicine 1995;14(19):2161–2172. [DOI] [PubMed] [Google Scholar]
  • 28.Hartung DM, Ahmed SM, Middleton L, et al. Using prescription monitoring program data to characterize out‐of‐pocket payments for opioid prescriptions in a state Medicaid program. Pharmacoepidemiology and Drug Safety 2017;26(9):1053–1060. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Smith RJ, Kilaru AS, Perrone J, et al. How, why, and for whom do emergency medicine providers use prescription drug monitoring programs? Pain Medicine 2015;16(6):1122–1131. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Haffajee RL, Jena AB, Weiner SG. Mandatory use of prescription drug monitoring programs. JAMA 2015;313(9):891–892. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Weiner SG, Baker O, Poon SJ, et al. The Effect of Opioid Prescribing Guidelines on Prescriptions by Emergency Physicians in Ohio. Annals of Emergency Medicine 2017. [DOI] [PubMed]
  • 32.Jones CM, Logan J, Gladden RM, Bohm MK. Vital signs: demographic and substance use trends among heroin users-United States, 2002–2013. MMWR Morbidity and mortality weekly report 2015;64(26):719–725. [PMC free article] [PubMed] [Google Scholar]
  • 33.Guy GP. Vital signs: changes in opioid prescribing in the United States, 2006–2015. MMWR Morbidity and mortality weekly report 2017;66. [DOI] [PMC free article] [PubMed]
  • 34.Cicero TJ, Ellis MS, Surratt HL, Kurtz SP. The changing face of heroin use in the United States: a retrospective analysis of the past 50 years. JAMA Psychiatry 2014;71(7):821–826. [DOI] [PubMed] [Google Scholar]
  • 35.Compton WM, Jones CM, Baldwin GT. Relationship between nonmedical prescription-opioid use and heroin use. N Engl J Med 2016;2016(374):154–163. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Harocopos A, Allen B, Paone D. Circumstances and contexts of heroin initiation following non-medical opioid analgesic use in New York City. International Journal of Drug Policy 2016;28:106–112. [DOI] [PubMed] [Google Scholar]
  • 37.Rudd RA. Increases in drug and opioid-involved overdose deaths—United States, 2010–2015. MMWR Morbidity and mortality weekly report 2016;65. [DOI] [PubMed]
  • 38.Ivins D Acute ankle sprain: an update. American Family Physician 2006;74(10). [PubMed] [Google Scholar]
  • 39.Ogilvie-Harris D, Gilbart M. Treatment modalities for soft tissue injuries of the ankle: a critical review. Clinical Journal of Sport Medicine 1995;5(3):175–186. [DOI] [PubMed] [Google Scholar]
  • 40.Chau DL, Walker V, Pai L, Cho LM. Opiates and elderly: use and side effects. Clinical Interventions in Aging 2008;3(2):273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Edlund MJ, Steffick D, Hudson T, Harris KM, Sullivan M. Risk factors for clinically recognized opioid abuse and dependence among veterans using opioids for chronic non-cancer pain. Pain 2007;129(3):355–362. [DOI] [PubMed] [Google Scholar]
  • 42.Edlund MJ, Martin BC, Russo JE, Devries A, Braden JB, Sullivan MD. The role of opioid prescription in incident opioid abuse and dependence among individuals with chronic non-cancer pain: the role of opioid prescription. The Clinical journal of Pain 2014;30(7):557. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Hall AJ, Logan JE, Toblin RL, et al. Patterns of abuse among unintentional pharmaceutical overdose fatalities. Jama 2008;300(22):2613–2620. [DOI] [PubMed] [Google Scholar]
  • 44.Hoppe JA, Kim H, Heard K. Association of emergency department opioid initiation with recurrent opioid use. Annals of Emergency Medicine 2015;65(5):493–499. e494. [DOI] [PubMed] [Google Scholar]
  • 45.Alam A, Gomes T, Zheng H, Mamdani MM, Juurlink DN, Bell CM. Long-term analgesic use after low-risk surgery: a retrospective cohort study. Archives of Internal Medicine 2012;172(5):425–430. [DOI] [PubMed] [Google Scholar]
  • 46.Centers for Disease Control and Prevention.U.S. Prescrbing Rate Maps. https://www.cdc.gov/drugoverdose/maps/rxrate-maps.html. Accessed April 28, 2018.
  • 47.Hill MV, McMahon ML, Stucke RS, Barth RJ Jr. Wide variation and excessive dosage of opioid prescriptions for common general surgical procedures. Annals of Surgery 2017;265(4):709–714. [DOI] [PubMed] [Google Scholar]
  • 48.Bicket MC, Long JJ, Pronovost PJ, Alexander GC, Wu CL. Prescription opioid analgesics commonly unused after surgery: A systematic review. Jama Surgery 2017;152(11):1066–1071. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Maughan BC, Hersh EV, Shofer FS, et al. Unused opioid analgesics and drug disposal following outpatient dental surgery: a randomized controlled trial. Drug and Alcohol Dependence 2016;168:328–334. [DOI] [PubMed] [Google Scholar]
  • 50.Delgado M, Shofer FS, Patel MS, Halpern SD, Edwards C, Meisel ZF, Perrone J Association between Electronic Medical Record Implementation of Default Opioid Prescription Quantities on Prescribing Behavior in Two Emergency Departments Journal of General Internal Medicine 2018. (e-pub ahead of print). [DOI] [PMC free article] [PubMed]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

appendix
NIHMS989897-supplement-appendix.docx (240.5KB, docx)

RESOURCES