Abstract
Background:
Opioid misuse and overdose in the US remain a public health emergency. Overprescribing has been recognized as a significant contributor to the epidemic. Opioids are the mainstay for pain management after burn; however, to date, no large-scale nationally representative study has evaluated outpatient opioid prescribing practices in this population.
Methods:
A retrospective study of patients up to 65 years old with burn between 2007 and 2017 using national commercial insurance data. The primary outcome was initial opioid prescribing after burn injury. Secondary outcomes were total days’ supply, oral daily morphine milligram equivalents (MME) and number of refills.
Results:
Of the 140,753 patients with burn, 34,685 (24.6%) received an opioid prescription. The odds of prescription opioid use were lower in 2015, 2016 and 2017 compared to 2007. Interactions with age, severity (p<.0001) and region (p=0.003) showed significant variation in rates of decline from 2007 to 2017, with the steepest decline in those aged <20 and in residents of Northeast US. Prescribing rates remained stable over time among those with more severe burn injury. The significant decline in daily opioid MMEs after 2013 was paralleled by increase in days of supply (p-values <.005). The odds of refill declined in 2016 and 2017.
Conclusions:
While opioid prescribing after burn has declined in the past decade, significant variation remains among regions and age groups, suggesting a need to develop uniform guidelines to improve the quality of opioid prescribing and pain management protocols in burn patients.
Article Summary:
Opioid prescribing following burn injury has declined over the past decade, however significant variation remains among different regions and age groups. There is a need to develop and implement uniform guidelines to improve the quality of opioid prescribing and pain management in burn patients
Introduction
Prescription opioid use in the United States remains a significant problem in health care, particularly in the last two decades. According to the Centers for Disease Control and Prevention (CDC), opioid prescriptions peaked in 2012 at a rate of 81.3 per 100 persons, and gradually declined to 51.4 prescriptions per 100 persons in 20181. Despite progress in prescribing rates, opioid misuse remains a pressing public health concern as opioid-related hospitalizations and deaths have not followed a similar trajectory. Inpatient stays due to opioids have increased by 72% between 2007 and 20162, while opioid-induced overdose deaths have doubled during the same period reaching 46,802 deaths in 20183,4. Of these deaths, 32% are attributed specifically to prescription opioids3. Long-term use of prescription opioids frequently starts with opioids prescribed after surgery5.
Severe burn is a debilitating injury that causes significant pain acutely and also throughout recovery and rehabilitation6. For a considerable number of patients, burns can result in chronic pain that interferes with sleep, daily activities, and quality of life. As adequate pain management is fundamental to recovery, it is not surprising that opioids are the first line treatment for burn injuries. In a survey of 133 US burn centers, 95% reported that they always or almost always use long-acting opioids throughout hospitalization7. However, development of opioid tolerance is not uncommon, indicating even higher doses to achieve pain control8,9. Despite the high rate of opioid prescription for hospitalized burn patients, little is known about the rate, patterns, and predictors of prescription opioid use upon hospital discharge. Data on rate and duration of opioid prescription after discharge are limited to single-center reports10–13. Moreover, the prevalence and patterns of opioid prescribing in the general patient population across a wide range of burn injuries has not been examined, including patients with less severe burns who receive treatment in the outpatient setting. According to the 2016 CDC guidelines, providers were advised to avoid prescribing long-term opioid therapy (>90 days) or high-dose opioids (>90 oral morphine milligram equivalents, MME) in the setting of non-cancer pain14. In order to balance the potential adverse effects of opioid over-prescription and the challenges of pain management in the burn population to guide prescribers and policy makers, it is crucial to understand the patterns of opioid use in this patient population at a national level. In this study, we used data from a large US commercial insurance database to examine the 10-year trend and predictors of opioid prescribing in patients with burn injury.
Methods
Data source
We performed a retrospective cohort study using administrative claims from Optum’s De-identified Clinformatics Data Mart (CDM, Optum Insight, Eden Prairie, MN) for the years 2006 through 2018. CDM is a national commercial and Medicare Advantage claims database that covers at least 15 million patients each year across all age groups. The database is representative of the privately insured US population less than 65 years old. For each enrollee we used data from inpatient, outpatient, and pharmacy claims files. The University of Texas Institutional Review Board approved this study.
Cohort selection
All patients with a primary burn injury diagnosis between 2007 and 2017 and aged from 2 to 64 years old at diagnosis were included in the study. Burns were identified using International Classification for Diseases (ICD) version 9 and 10 codes (ICD-9: 940.xx - 949.xx; ICD-10: T20.xx – T32.xx). Patients were excluded if they had less than one-year continuous enrollment before the first burn diagnosis or less than 30 days after diagnosis or hospital discharge. Patients were also excluded if in the year prior to primary diagnosis they had burn injury coded in any secondary diagnosis position or had unknown gender information. For those with subsequent new burns during the study period, only the first injury was included (Supplemental Figure 1). A subset of patients with at least 6 months continuous enrollment after diagnosis or hospital discharge was used to evaluate the number of refills post-diagnosis.
Outcome definition
Opioid prescriptions were identified using pharmacy claims and National Drug Codes with product and therapeutic class names from the 2019 Red-book Select database15. The primary outcome variable was initial opioid prescription after burn injury and was defined as having at least one opioid pharmacy claim filled within 7 days from outpatient burn diagnosis or within 3 days before or 7 days after hospital discharge for inpatients. For patients with consecutive hospital admissions less than 2 days apart the latest discharge date was used. Secondary outcomes were days of supply, total oral Morphine Milligram Equivalents (MME) per day in the initial opioid prescription, and number of refills within 6 months from diagnosis. MME/day was calculated using the Centers for Medicare & Medicaid Services recommendation as strength per unit X (Number of Units/ Days Supply) X MME conversion factor, where conversion factors reflect each drug’s potency compared to morphine16. Subsequent opioid prescriptions were considered as refills if the new prescription was filled within 14 days of the previous prescription’s last supply day.
Variables
The main independent variable of interest in this study is year of burn diagnosis. Demographic factors included age at diagnosis, gender, and region of residence (South, West, Northeast, or Midwest). Prior opioid use was defined as having at least one filled opioid prescription within the 12 months preceding injury. Comorbidities that are known risk factors for opioid use were identified from claims data and ICD codes using a one-year retrospective period, and included tobacco, alcohol or drug abuse, cancer, arthritis (rheumatoid or osteoarthritis) and depression5. A modified Elixhauser comorbidity score, that did not include the aforementioned comorbidities, was calculated as previously described17. Patients were categorized as inpatients with any hospitalization due to burn injury within 10 days from the initial diagnosis; otherwise, they were classified as outpatients. For those whose claims contained ICD codes specific enough to determine the extent and depth of injury, we used the American Burn Association’s grading and referral criteria to classify them as more severe (moderate or major), or less severe (minor) injuries18,19. Among those not initially classified (34%), all inpatients as well as those outpatients who had any grafting, wound care, or debridement procedure (Supplemental Table 1) were classified as more severe; otherwise these were considered as minor.
Statistical analysis
Demographic and clinical characteristics were summarized by year of diagnosis as counts and percentages or means and standard deviations and compared using chi-square or analysis of variance, as appropriate. Multivariable logistic regression was used to assess the effect of diagnosis year on opioid use while adjusting for all other covariates. The impact of diagnosis year on days of supply and MME/day among those prescribed opioids was determined with multivariable Poisson and lognormal models, respectively. The number of refills in the first 6 months was categorized as 0, 1 or 2+ refills and modeled with ordinal logistic regression. Interactions between year of diagnosis with age, region, inpatient status, severity, or prior opioid use were tested and significant terms were selected via a stepwise method. In addition, similar analyses were conducted for the subset of more severe burn patients to determine any differences in opioid prescribing patterns. A sensitivity analysis among patients with known burn depth information was also performed. All analyses were conducted using SAS 9.4 (Cary, NC, US).
Results
A total of 140,753 persons with burns were included in the study (Figure 1). Of these, 34,685 (24.6%) filled an opioid prescription within 7 days from injury or hospital discharge. The majority in the opioid cohort were females (54.9%), 20–50 years old (48.4%) followed by >=50 years (25.9%) old, from the South (45.8%) or Midwest (24.9%), and 27.5% of these had been prescribed opioids in the past 12 months. 4,079 (2.9%) were hospitalized due to burn, and 49.8% were classified as minor (less severe) burns. The percentage of patients with higher comorbidity score, arthritis, depression, drug, or tobacco use increased considerably over the study period. Population characteristics are summarized in Table 1.
Figure 1:
Opioid prescriptions after burn injury, 2007–2017 A) Total rate of initial opioid prescriptions after burn injury B) Rate of initial opioid prescriptions after burn injury by inpatient or outpatient status C) Mean Morphine Milligram Equivalents per day by inpatient/outpatient status D) Mean days of supply by inpatient/outpatient status
Table 1:
Patient and injury characteristics for selected diagnosis years
| Characteristics | Category | Totala | 2007 | 2010 | 2013 | 2017 | p-valueb |
|---|---|---|---|---|---|---|---|
| N (%) | N (%) | N (%) | N (%) | ||||
| Age, cont. mean (SD) | 32.9 (17.6) | 34.3 (17.8) | 34.8 (17.7) | 36.5 (17.8) | <.0001 | ||
| Age at diagnosis, categorical | < 10y | 14696 | 1695 (12.11) | 1486 (11.15) | 1225 (9.75) | 1046 (8.54) | <.0001 |
| 10 – 20y | 21571 | 2469 (17.64) | 2146 (16.1) | 1844 (14.68) | 1580 (12.9) | ||
| 20 – 50y | 68062 | 6786 (48.47) | 6388 (47.94) | 6153 (48.98) | 5911 (48.26) | ||
| >= 50y | 36424 | 3049 (21.78) | 3306 (24.81) | 3340 (26.59) | 3712 (30.3) | ||
| Gender | Female | 77299 | 7502 (53.59) | 7352 (55.17) | 6955 (55.37) | 6624 (54.08) | 0.0178 |
| Male | 63454 | 6497 (46.41) | 5974 (44.83) | 5607 (44.63) | 5625 (45.92) | ||
| Region | Midwest | 35047 | 3689 (26.35) | 3121 (23.42) | 3216 (25.6) | 3144 (25.67) | <.0001 |
| Northeast | 14782 | 1434 (10.24) | 1326 (9.95) | 1372 (10.92) | 1259 (10.28) | ||
| South | 64474 | 6494 (46.39) | 6444 (48.36) | 5596 (44.55) | 5390 (44.0) | ||
| West | 25821 | 2330 (16.64) | 2377 (17.84) | 2317 (18.44) | 2395 (19.55) | ||
| Unknown | 629 | 52 (0.37) | 58 (0.44) | 61 (0.49) | 61 (0.5) | ||
| Opioid use in the last 12 months | No | 102088 | 10310 (73.65) | 9620 (72.19) | 8976 (71.45) | 8989 (73.39) | <.0001 |
| Yes | 38665 | 3689 (26.35) | 3706 (27.81) | 3586 (28.55) | 3260 (26.61) | ||
| Burn severity | Less severe | 70108 | 6420 (45.86) | 6386 (47.92) | 5808 (46.23) | 7611 (62.14) | <.0001 |
| More severe | 70645 | 7579 (54.14) | 6940 (52.08) | 6754 (53.77) | 4638 (37.86) | ||
| Inpatient stay | No | 136674 | 13618 (97.28) | 12918 (96.94) | 12201 (97.13) | 11860 (96.82) | 0.437 |
| Yes | 4079 | 381 (2.72) | 408 (3.06) | 361 (2.87) | 389 (3.18) | ||
| Modified Elixhauser score | 0 | 84475 | 8985 (64.18) | 7968 (59.79) | 7496 (59.67) | 6823 (55.7) | <.0001 |
| 1 | 28256 | 2848 (20.34) | 2812 (21.1) | 2503 (19.93) | 2335 (19.06) | ||
| 2 | 12531 | 1126 (8.04) | 1200 (9.0) | 1154 (9.19) | 1149 (9.38) | ||
| >=3 | 15491 | 1040 (7.43) | 1346 (10.1) | 1409 (11.22) | 1942 (15.85) | ||
| Alcohol Abuse | No | 138111 | 13809 (98.64) | 13095 (98.27) | 12310 (97.99) | 11908 (97.22) | <.0001 |
| Yes | 2642 | 190 (1.36) | 231 (1.73) | 252 (2.01) | 341 (2.78) | ||
| Drug Abuse | No | 137304 | 13816 (98.69) | 13077 (98.13) | 12221 (97.29) | 11788 (96.24) | <.0001 |
| Yes | 3449 | 183 (1.31) | 249 (1.87) | 341 (2.71) | 461 (3.76) | ||
| Depression | No | 121138 | 12303 (87.88) | 11504 (86.33) | 10781 (85.82) | 10288 (83.99) | <.0001 |
| Yes | 19615 | 1696 (12.12) | 1822 (13.67) | 1781 (14.18) | 1961 (16.01) | ||
| Tobacco Use/Dependence | No | 129454 | 13274 (94.82) | 12457 (93.48) | 11606 (92.39) | 10542 (86.06) | <.0001 |
| Yes | 11299 | 725 (5.18) | 869 (6.52) | 956 (7.61) | 1707 (13.94) | ||
| Cancer | No | 137584 | 13708 (97.92) | 13015 (97.67) | 12296 (97.88) | 11974 (97.75) | 0.319 |
| Yes | 3169 | 291 (2.08) | 311 (2.33) | 266 (2.12) | 275 (2.25) | ||
| Arthritis (RA/OA) | No | 125830 | 12876 (91.98) | 12046 (90.39) | 11139 (88.67) | 10585 (86.42) | <.0001 |
| Yes | 14923 | 1123 (8.02) | 1280 (9.61) | 1423 (11.33) | 1664 (13.58) | ||
| Initial opioid prescription | No | 106068 | 10505 (75.04) | 9867 (74.04) | 9347 (74.41) | 9713 (79.3) | <.0001 |
| Yes | 34685 | 3494 (24.96) | 3459 (25.96) | 3215 (25.59) | 2536 (20.7) | ||
|
| |||||||
| Days of supply among opioid prescriptions, mean (SD) | 5.8 (6.2) | 6.0 (6.5) | 6.4 (7.3) | 6.9 (7.6) | <.0001 | ||
| MME/day among opioid prescriptions, mean (SD) | 41.9 (46.2) | 42.5 (46.8) | 41.2 (48.5) | 39.0 (50.3) | 0.031 | ||
Total reflects all years from 2007 – 2017, not only the selected years shown
p-values calculated by Chi-square or Analysis of Variance test for all years between 2007 and 2017
The yearly percentage of burned patients prescribed opioids remained stable from 2007 to 2013 with a maximum of 26% in 2010; then gradually declined to 20.7% in 2017 (Figure 1A). Opioid prescriptions for outpatients declined at a faster rate than for inpatients (Figure 1B) (2007: inpatients 63.5, outpatients 23.8; 2017: inpatients 60.4, outpatients 19.4). Average prescribed MME/day followed a similar pattern, gradually reducing from 39.9 units in 2007 to 36.1 in 2017 for outpatients, while for inpatients average MME/day peaked in 2012 with 79.5 and declined to 67.4 in 2017 (Figure 1C). While opioid strength remained stable or declined, the average days of supply for outpatients continuously increased from 5.6 in 2007 to 6.7 in 2017 (Figure 1D). Among those who filled an opioid prescription, we identified 29,870 meeting the continuous enrollment criterion for 6 months after injury. Of these, 76.3% did not receive any refill within 14 days from the end of the initial prescription, while 10.6% received one refill and 13.1% two or more refills. At 90 days after the initial prescription 7.1% of eligible patients were still prescribed opioids, while at 180 days the percent was 5.8%. Of those that had opioid prescriptions 90 and 180 days after injury, ≥ 98% were prior opioid users. (Supplemental Table 2).
Table 2 summarizes the results of the multivariable logistic model assessing the association between year of burn diagnosis and opioid prescribing. After adjusting for patient and injury characteristics, the odds of receiving a prescription were lower in 2015 (Odds Ratio [OR] 0.92, 95% Confidence Interval [CI] 0.86–0.97), 2016 (OR 0.84, 95% CI 0.79–0.89) and 2017 (OR 0.79, 95% CI 0.74–0.84) compared to 2007. Compared to those >= 50 years, children and adolescents were less likely to receive opioids (< 10 years OR 0.63, 95% CI 0.59–0.67; 10–20 years OR 0.94, 95% CI 0.90–0.99], while adults 20–50 years old were significantly more likely to receive a prescription (OR 1.52, 95% CI 1.47–1.57). Female gender was negatively associated with opioid prescribing (OR 0.77, 95% CI 0.75 – 0.78) compared to male, as well as being from the Midwest (OR 0.91, 95% CI 0.88 – 0.94), Northeast (OR 0.57, 95% CI 0.54 – 0.59) or West (OR 0.83, 95% CI 0.80 – 0.86) compared to the South. Inpatients and more severe burns were 4.13 (95% CI 3.86 – 4.42) and 1.52 (95% CI 1.48 – 1.56) times more likely to have an opioid prescription than outpatients or minor injuries, respectively. Having received an opioid prescription for another reason, having prior drug or tobacco use, or history of depression or arthritis were also positively associated with increased likelihood of opioid prescription. Alternatively, increased Elixhauser comorbidity score and a history of cancer were associated with reduced odds of opioid prescription (Table 2).
Table 2:
Multivariable Logistic Regression Model for Initial Opioid Prescriptions after Burn Injury
| Multivariate Logistic Regression Model for Initial Opioid Prescribing after Burn Injury | |||||
|---|---|---|---|---|---|
| Characteristics | Opioid prescription* N (%) | No Opioid prescription N (%) | Odds Ratio** | 95% Confidence Limits | |
| Year of diagnosis | |||||
| 2007 | 3494 (24.96) | 10505 (75.04) | Ref | ||
| 2008 | 3709 (25.85) | 10641 (74.15) | 1.055 | 0.998 | 1.115 |
| 2009 | 3504 (25.63) | 10169 (74.37) | 1.044 | 0.987 | 1.104 |
| 2010 | 3459 (25.96) | 9867 (74.04) | 1.050 | 0.992 | 1.111 |
| 2011 | 3393 (25.9) | 9709 (74.1) | 1.051 | 0.993 | 1.112 |
| 2012 | 3430 (25.86) | 9836 (74.14) | 1.028 | 0.971 | 1.088 |
| 2013 | 3215 (25.59) | 9347 (74.41) | 1.008 | 0.952 | 1.068 |
| 2014 | 2795 (24.57) | 8580 (75.43) | 0.959 | 0.903 | 1.017 |
| 2015 | 2655 (23.69) | 8554 (76.31) | 0.917 | 0.863 | 0.974 |
| 2016 | 2495 (21.43) | 9147 (78.57) | 0.837 | 0.788 | 0.890 |
| 2017 | 2536 (20.7) | 9713 (79.3) | 0.789 | 0.742 | 0.838 |
| Age at diagnosis | |||||
| >= 50y | 9093 (24.96) | 27331 (75.04) | Ref | ||
| < 10y | 2145 (14.6) | 12551 (85.4) | 0.626 | 0.592 | 0.662 |
| 10 – 20y | 4002 (18.55) | 17569 (81.45) | 0.939 | 0.895 | 0.985 |
| 20 – 50y | 19445 (28.57) | 48617 (71.43) | 1.520 | 1.469 | 1.573 |
| Gender | |||||
| Male | 17408 (27.43) | 46046 (72.57) | Ref | ||
| Female | 17277 (22.35) | 60022 (77.65) | 0.764 | 0.745 | 0.784 |
| Region | |||||
| South | 17462 (27.08) | 47012 (72.92) | Ref | ||
| Midwest | 8727 (24.9) | 26320 (75.1) | 0.906 | 0.879 | 0.935 |
| Northeast | 2439 (16.5) | 12343 (83.5) | 0.566 | 0.540 | 0.594 |
| West | 5898 (22.84) | 19923 (77.16) | 0.830 | 0.802 | 0.860 |
| Opioid use in the last 12 months | |||||
| No | 21294 (20.86) | 80794 (79.14) | Ref | ||
| Yes | 13391 (34.63) | 25274 (65.37) | 1.723 | 1.673 | 1.774 |
| Inpatient stay | |||||
| No | 32138 (23.51) | 104536 (76.49) | Ref | ||
| Yes | 2547 (62.44) | 1532 (37.56) | 4.132 | 3.861 | 4.422 |
| Burn severity | |||||
| Less severe | 14709 (20.98) | 55399 (79.02) | Ref | ||
| More severe | 19976 (28.28) | 50669 (71.72) | 1.516 | 1.474 | 1.558 |
| Total Modified Elixhauser score | |||||
| Score 0 | 19892 (23.55) | 64583 (76.45) | Ref | ||
| Score 1 | 6933 (24.54) | 21323 (75.46) | 0.927 | 0.897 | 0.959 |
| Score 2 | 3275 (26.14) | 9256 (73.86) | 0.882 | 0.841 | 0.924 |
| Score 3+ | 4585 (29.6) | 10906 (70.4) | 0.878 | 0.838 | 0.920 |
| Arthritis (RA/OA) | |||||
| No | 30112 (23.93) | 95718 (76.07) | Ref | ||
| Yes | 4573 (30.64) | 10350 (69.36) | 1.073 | 1.028 | 1.120 |
| Cancer | |||||
| No | 33873 (24.62) | 103711 (75.38) | Ref | ||
| Yes | 812 (25.62) | 2357 (74.38) | 0.863 | 0.792 | 0.940 |
| Alcohol Abuse | |||||
| No | 33747 (24.43) | 104364 (75.57) | Ref | ||
| Yes | 938 (35.5) | 1704 (64.5) | 0.958 | 0.875 | 1.050 |
| Drug Abuse | |||||
| No | 33312 (24.26) | 103992 (75.74) | Ref | ||
| Yes | 1373 (39.81) | 2076 (60.19) | 1.258 | 1.161 | 1.361 |
| Depression | |||||
| No | 28710 (23.7) | 92428 (76.3) | Ref | ||
| Yes | 5975 (30.46) | 13640 (69.54) | 1.178 | 1.135 | 1.223 |
| Tobacco Use/Dependence | |||||
| No | 30529 (23.58) | 98925 (76.42) | Ref | ||
| Yes | 4156 (36.78) | 7143 (63.22) | 1.318 | 1.258 | 1.380 |
Unadjusted rates
Adjusted for all covariates
We found significant interactions between year of diagnosis and age (p<.0001), region of residence (p = 0.003) and severity (p<0.001). Figure 2 shows predicted probabilities of opioid prescription for each year between 2007 and 2017 by interaction term, adjusting for all other covariates in the model. While the likelihood of opioid prescription in those over 50 years of age remained relatively constant over the study period, it declined for all other age groups, with younger children experiencing the steepest decline (16.3% probability of opioid prescription in 2007 – 9.4% in 2017 – Figure 2A). The Northeast was the region with the steepest decline in the likelihood of opioid prescriptions (17.4% in 2007 to 10.3% in 2017), followed by the West (24.4 % in 2007 to 18.2% in 2017 - Figure 2B) while reductions in the South and Midwest were less pronounced. For patients with more severe burns, the probability of opioid prescribing was constant throughout the study period but was significantly reduced for those with less severe injuries (20.4% in 2007 to 14.4% in 2017 - Figure 2C).
Figure 2:
Adjusted probabilities of opioid prescription by year of diagnosis and A) age at diagnosis B) region of residence C) burn severity
Tables 3 and 4 show the results of the multivariable models for the secondary outcomes among opioid users. After adjusting for patient and clinical characteristics, we found a statistically significant increase in days of supply prescribed in 2013 (p=0.03) and 2014 (p=0.04), but in later years results remained comparable to 2007 (Table 3). On the contrary, in 2013 (p=0.015) and from 2015 to 2017 (p-values <.005) we found a steady decrease in prescribed MMEs when compared to 2007 (Table 3). The odds of getting two or more refills compared to none were 0.81 (95% CI 0.68 – 0.98), 0.77 (95% CI 0.63 – 0.93) and 0.67 (95% CI 0.55 – 0.81) times lower in 2014, 2016 and 2017 compared to 2007, but otherwise remained similar over the study period. No interactions with diagnosis year were significant for the secondary outcomes.
Table 3:
Multivariable models for the effect of year of diagnosis on days of supply and MME/day in the initial opioid prescription
| Days of supply | MME/day | |||
|---|---|---|---|---|
| RRa | P-value | RRb | P-value | |
| Year of diagnosis | ||||
| 2007 | 1.000 | 1.000 | ||
| 2008 | 0.955 | 0.015 | 0.984 | 0.509 |
| 2009 | 0.971 | 0.129 | 0.996 | 0.881 |
| 2010 | 0.990 | 0.595 | 0.980 | 0.407 |
| 2011 | 1.000 | 0.988 | 0.994 | 0.824 |
| 2012 | 0.995 | 0.804 | 0.975 | 0.300 |
| 2013 | 1.042 | 0.033 | 0.940 | 0.015 |
| 2014 | 1.042 | 0.037 | 0.956 | 0.082 |
| 2015 | 1.033 | 0.106 | 0.926 | 0.004 |
| 2016 | 1.014 | 0.499 | 0.864 | <0.001 |
| 2017 | 1.008 | 0.703 | 0.826 | <0.001 |
| Age at diagnosis | ||||
| >= 50y | 1.000 | 1.000 | ||
| < 10y | 0.987 | 0.520 | 0.276 | <0.001 |
| 10 – 20y | 0.762 | <0.001 | 0.791 | <0.001 |
| 20 – 50y | 0.817 | <0.001 | 1.009 | 0.517 |
| Gender | ||||
| Male | 1.000 | 1.000 | ||
| Female | 0.951 | <0.001 | 0.902 | <0.001 |
| Region | ||||
| South | 1.000 | 1.000 | ||
| Midwest | 0.917 | <0.001 | 1.070 | <0.001 |
| Northeast | 0.958 | 0.010 | 1.007 | 0.759 |
| West | 0.991 | 0.449 | 1.155 | <0.001 |
| Opioid use in the last 12 months | ||||
| No | 1.000 | 1.000 | ||
| Yes | 1.465 | <0.001 | 1.155 | <0.001 |
| Inpatient stay | ||||
| No | 1.000 | 1.000 | ||
| Yes | 1.401 | <0.001 | 1.703 | <0.001 |
| Burn severity | ||||
| Less severe | 1.000 | 1.000 | ||
| More severe | 0.960 | <0.001 | 1.103 | <0.001 |
| Total Modified Elixhauser | ||||
| Score 0 | 1.000 | 1.000 | ||
| Score 1 | 1.024 | 0.043 | 0.983 | 0.275 |
| Score 2 | 1.109 | <0.001 | 0.993 | 0.720 |
| Score 3+ | 1.289 | <0.001 | 1.108 | <0.001 |
| Arthritis (RA/OA) | ||||
| No | 1.000 | 1.000 | ||
| Yes | 1.332 | <0.001 | 1.120 | <0.001 |
| Cancer | ||||
| No | 1.000 | 1.000 | ||
| Yes | 1.041 | 0.082 | 1.312 | <0.001 |
| Alcohol Abuse | ||||
| No | 1.000 | 1.000 | ||
| Yes | 0.861 | <0.001 | 0.839 | <0.001 |
| Drug Abuse | ||||
| No | 1.000 | 1.000 | ||
| Yes | 1.178 | <0.001 | 1.339 | <0.001 |
| Depression | ||||
| No | 1.000 | 1.000 | ||
| Yes | 1.078 | <0.001 | 1.103 | <0.001 |
| Tobacco Use/Dependence | ||||
| No | 1.000 | 1.000 | ||
| Yes | 1.054 | <0.001 | 1.078 | <0.001 |
Risk Ratio estimated from Poisson Model, adjusted for all variables
Risk Ratio estimated from Log-normal model, adjusted for all variables
Table 4:
Multivariable ordinal regression model for the effect of year of diagnosis on number of refills
| Odds Ratio Estimatesa | ||||||
|---|---|---|---|---|---|---|
| 1 vs 0 refillsb | >= 2 vs 0 refillsb | |||||
|
| ||||||
| Characteristics | Estimate | 95% Wald Confidence Limits | Estimate | 95% Wald Confidence Limits | ||
| Year of diagnosis | ||||||
| 2008 vs 2007 | 1.028 | 0.870 | 1.214 | 0.833 | 0.694 | 1.001 |
| 2009 vs 2007 | 1.025 | 0.866 | 1.213 | 0.923 | 0.771 | 1.105 |
| 2010 vs 2007 | 0.986 | 0.832 | 1.169 | 0.894 | 0.747 | 1.069 |
| 2011 vs 2007 | 0.983 | 0.828 | 1.167 | 0.953 | 0.798 | 1.139 |
| 2012 vs 2007 | 0.940 | 0.793 | 1.115 | 0.874 | 0.732 | 1.043 |
| 2013 vs 2007 | 0.910 | 0.763 | 1.085 | 0.911 | 0.760 | 1.091 |
| 2014 vs 2007 | 0.926 | 0.774 | 1.107 | 0.814 | 0.676 | 0.981 |
| 2015 vs 2007 | 0.962 | 0.802 | 1.154 | 0.843 | 0.699 | 1.017 |
| 2016 vs 2007 | 0.969 | 0.805 | 1.167 | 0.767 | 0.631 | 0.932 |
| 2017 vs 2007 | 0.886 | 0.734 | 1.068 | 0.667 | 0.549 | 0.810 |
| Age at diagnosis | ||||||
| < 10y vs >= 50y | 0.247 | 0.190 | 0.321 | 0.106 | 0.062 | 0.182 |
| 10 – 20y vs >= 50y | 0.656 | 0.561 | 0.768 | 0.238 | 0.184 | 0.307 |
| 20 – 50y vs >= 50y | 1.051 | 0.953 | 1.158 | 0.780 | 0.710 | 0.856 |
| Gender | ||||||
| Female vs Male | 0.714 | 0.660 | 0.773 | 0.790 | 0.727 | 0.858 |
| Region | ||||||
| Midwest vs South | 1.028 | 0.937 | 1.128 | 1.015 | 0.920 | 1.119 |
| Northeast vs South | 0.881 | 0.753 | 1.031 | 0.863 | 0.733 | 1.017 |
| West vs South | 1.113 | 1.001 | 1.237 | 1.187 | 1.065 | 1.323 |
| Opioid use in the past 12 months | ||||||
| Yes vs No | 1.661 | 1.527 | 1.806 | 6.090 | 5.519 | 6.719 |
| Inpatient stay | ||||||
| Yes vs No | 2.343 | 2.065 | 2.659 | 3.630 | 3.181 | 4.143 |
| Burn severity | ||||||
| More severe vs Less severe | 1.716 | 1.572 | 1.874 | 1.601 | 1.461 | 1.755 |
| Elixhauser score | ||||||
| 1 vs 0 | 1.032 | 0.933 | 1.141 | 1.142 | 1.023 | 1.276 |
| 2 vs 0 | 1.117 | 0.977 | 1.278 | 1.451 | 1.276 | 1.65 |
| >=3 vs 0 | 1.141 | 0.998 | 1.304 | 1.818 | 1.615 | 2.046 |
| Arthritis (RA/OA) | ||||||
| Yes vs No | 1.076 | 0.950 | 1.219 | 2.255 | 2.049 | 2.481 |
| Cancer | ||||||
| Yes vs No | 1.111 | 0.871 | 1.417 | 1.083 | 0.887 | 1.321 |
| Alcohol Abuse | ||||||
| Yes vs No | 1.170 | 0.936 | 1.464 | 0.792 | 0.646 | 0.972 |
| Drug Abuse | ||||||
| Yes vs No | 1.344 | 1.088 | 1.661 | 2.163 | 1.839 | 2.544 |
| Depression | ||||||
| Yes vs No | 1.125 | 1.011 | 1.253 | 1.466 | 1.333 | 1.612 |
| Tobacco Use | ||||||
| Yes vs No | 1.303 | 1.154 | 1.471 | 1.698 | 1.533 | 1.881 |
Adjusted for all variables
The ordinal regression model contrasts each response level against the reference level (refills = 0)
Results from the subset analyses of severe burn patients showed comparable annual trends for supply days, MME/day and refills and a non-significant decline in the yearly opioid prescribing rate (Supplemental Tables 3, 4). Interactions with age and region were significant; in patients with more severe burns the declines in the younger age group and the Northeast over the past decade, were countered by the increasing trend in the 20–50 years group and the Midwest (Supplemental Figures 2, 3). In addition, we performed a sensitivity analysis using the available information on degree of burn instead of severity classification with similar results (not shown).
Discussion
In this nationally representative cohort study of 140,753 burned patients, we evaluated opioid prescriptions after injury from 2007 to 2017. We found that opioid prescription rates declined by 17.2%, from 25% in 2007 to 21% in 2017 in line with national trends. Patients treated in the outpatient setting received 18.8% fewer opioid prescriptions, while inpatients upon hospital discharge received 4.8% less prescriptions in this period. However, the rate of decline was not as evident in some age groups and regions, even after adjusting for relevant patient and clinical characteristics. The decline was more pronounced for persons under 20 years of age and those residing in the Northeast and West. Also, for patients with more severe injury, opioid prescriptions remained stable over the study period. This is encouraging, as it indicates that patients who were more likely to experience severe or prolonged pain were not declined opioids; instead opioid prescribing reduction has been focused on patients with less severe injury. While these findings suggest a potential positive effect of clinical advancements and efforts in education and policy development aimed at mitigating opioid overprescribing, however, the effect was not equally generalized to all parts of the population.
Our study is the first to characterize opioid use after burn injury at a national level, accounting for patient and clinical factors. Prior single-center studies at regional burn centers indicated rates of opioid prescription at hospital discharge ranging from 86% to 90% for adults10,11 and 77% for children13. Our study showed a generally lower percentage of inpatients (62.4%) filled opioid prescriptions within a week after discharge. Studies from burn centers reflect data from patients admitted for major or more complicated burns, unlike our study that includes all facilities participating in a commercial insurance program and covers opioid prescribing for burn patients across all ranges of severity. Thus, some epidemiologic characteristics in our study may appear different from prior research; for instance, the percentage of females in our cohort is 54.9%, whereas in the National Burn Repository (NBR) is 31.8%20. However, NBR data are comprised mostly from burn centers and include predominantly hospitalized patients. Indeed, in our study the percent of females among hospitalized patients was similar at 34.2%. To our knowledge, the only studies evaluating opioid prescribing were those of Shahi et al, who reported a 16.2% opioid prescription rate in the outpatient setting for children ≥ 7 years old with burns21, and a descriptive cross-sectional study using insurance claims data, where 20% of adults and 11% of pediatric opioid-naïve burn patients received an opioid prescription22. In our study, 23.5% of outpatient children or adults ≤ 65 years were given opioid prescriptions for pain management.
While no other study assessed temporal trends in opioid prescription rate after burns, a study of adults discharged from a Midwestern burn hospital showed a significant increase in the amount of total oral MMEs in the initial prescription from 2008 to 201512. At first glance, this study may appear contradictory to our finding of an overall decline of MME/day prescribed from 2007 to 2017. However, in our cohort the reduction in prescribing doses stabilized after 2015, while the days of supply prescribed significantly increased in 2013 and 2014. In regard to prescribed doses, we found that the median prescribed MME/day was 58 (IQR:33 – 90) units for inpatients, while the respective dose at an Iowa adult burn center between 2009 and 2012 was double this value (median: 114, IQR:90–180)11, with most patients receiving prescription doses far exceeding the current CDC recommendation to “avoid prescriptions of ≥ 90 MME/day”14. Variation among hospitals and physicians in the amounts prescribed is expected, as no consensus has been reached for the appropriate amount to balance pain control and potential adverse health effects as the patients themselves are quite variable in this regard. Given the variation in opioid prescription rates and potential for toxicity, more research would be of benefit on cost-effectiveness of non-opioid and non-drug approaches as alternative strategies for managing pain, while preserving function and quality of life in patients with burn injury. Another area of needed future research is an ecological study of temporal trends in frequency and severity of pain, and pain-related disability, to explore if the decreasing trends in outpatient opioid prescribing have any substantial impact on post-burn pain and health. Past research in non-burn settings however showed no association between longitudinal change in opioid prescribing and level of disability, or pain over time among the community-dwelling population23 or among patients who underwent joint replacement surgery24.
The distribution of patients undergoing hospitalization or being treated as outpatients was stable throughout the study period, although we found a decline in the percentage of patients with more severe injury; however, even when accounting for these factors the evident decline in opioid prescribing after 2015 remained statistically significant in the overall burn population. It is possible that the decline reflects the effect of the 2014 Drug Enforcement Administration’s law rescheduling hydrocodone from schedule III to schedule II, as previously shown in other studies25,26. In the years after 2011, when the CDC, in a series of publications, declared prescription opioids as an “epidemic” and a public health crisis, increased awareness of associated risks by both physicians and patients is likely to have also contributed to this decline, particularly among the younger age groups. In addition to the 2014 rescheduling of hydrocodone products, increased adoption of statewide prescription monitoring programs and other state-mandated opioid-prescribing regulations, as well as the 2016 CDC guidelines, may have also contributed to the steeper observed decline in later years25,27–29. Since the scope of each state’s response and opioid-related guidelines vary, it is not surprising that we found significant differences in the rate of decline among regions.
A recent metanalysis on efficacy and pain control of various wound dressing types, determined that burn patients treated with hydrogel dressings reported significantly lower pain scores than those treated with any other traditional medicament30. It is likely that changes in medical care over time and exploration of adjunctive pain control strategies may have contributed as well to the observed decline31.
Other findings from our study are consistent with prior research including higher rates of opioid prescribing among males, those aged between 20 and 50 years, and those with a history of drug or tobacco dependence, depression, arthritis, or any use of opioids in the past year32,33. Also, we found that prolonged opioid use after burn injury was lower than previously reported10, but was disproportionately higher among prior opioid users compared to opioid naïve patients. These findings are important for clinical practice as previous studies suggest that opioid misuse and overdose are associated with mental health conditions, substance addiction, chronic opioid use, male sex and ages 41–64 years32–35.
Of concern are the three groups of burn patients with high-risk prescription opioid use according to the 2016 CDC criteria. These are: the 13% of burn patients who received two or more opioid refills; the 7% of patients receiving opioid prescriptions 90 days or more after the initial prescription; and the 5% still receiving opioid prescriptions 180 days or more from injury. This is concerning given their increased risk of long-term opioid use, dependence and overdose5,32,33.
We could not address some issues in the analysis. First, insurance claims data capture only those prescriptions that were ultimately filled; it is not possible to differentiate between those that adhered to the prescribed regimen, those that may have not consumed the entire supply or those that could have misused it or diverted it to others. Further, our source of data does not include information on race/ethnicity or socioeconomic status, pain scores or prescriber specialty and is not fully representative of the population over 65 years old, therefore we limited our analysis to those younger than 65. Also, some patients have missing data on detailed injury severity information (such as the percent of total body surface area burned). We sought to account for the effect of burn severity by using a composite measure based on criteria set by the American Burn Association as well as hospitalization (versus only outpatient treatment) or the type of procedures received. To confirm that this measure captures burn severity adequately, we conducted a sensitivity analysis including only the subset of patients with available burn degree information with similar results. Our definition of initial opioid prescription may underestimate overall burn-related use; a previous study found that 5% of burn adults did not receive an opioid prescription upon hospital discharge, but were given one at a later outpatient clinic visit11. Finally, our study was not designed to address specifically whether changes in health policy or medical care contributed to the observed prescribing trends.
Our study is the first to use a large, nationally representative cohort of US children and adults with burns to evaluate opioid prescriptions over a long period, —2007 to 2017— a time spanning implementation of numerous state and federal policies regulating opioid prescription. Another major strength of this study is that we captured each patient’s opioid prescriptions across multiple providers and facilities, allowing for a more complete evaluation of prescribing practices in this population.
Conclusions
Opioid prescriptions following burn injury have declined over the past decade, however significant variation remains among different regions, suggesting an opportunity to develop and implement uniform guidelines to improve the quality of opioid prescribing and pain management in burn patients. Of concern, burn patients at higher risk for prolonged use are more likely to receive opioid prescriptions. In the absence of universal guidelines for the management of burn pain, providers have developed different protocols or have adapted to prescribing recommendations at a different pace. Further research on pain management of burn patients is indicated to establish safe and effective opioid prescribing strategies (as well as non-opioid alternatives) that balance pain control, preserve function, and avoid opioid-related toxicities.
Supplementary Material
Acknowledgments:
Funding: This study was supported by grant R01DA039192 (National Institute on Drug Abuse). The sponsor 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.
Footnotes
Conflict of Interest: The authors have no conflict of interest to disclose.
References
- 1.Centers for Disease Control and Prevention. U.S. Opioid Prescribing Rate Maps. https://www.cdc.gov/drugoverdose/maps/rxrate-maps.html.Published 2020. Accessed April 30, 2020.
- 2.Agency for Health Care Research and Quality. Trends in the Rate of Opioid-Related Hospitalizations. https://www.ahrq.gov/opioids/map/index.html.Accessed April 30, 2020.
- 3.Centers for Disease Control and Prevention. Overdose Death Maps. https://www.cdc.gov/drugoverdose/data/prescribing/overdose-death-maps.html.Published 2020. Accessed April 30, 2020.
- 4.Seth P, Rudd RA, Noonan RK, Haegerich TM. Quantifying the epidemic of prescription opioid overdose deaths. Am J Public Health. 2018;108(4):500–502. doi: 10.2105/AJPH.2017.304265 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Neuman MD, Bateman BT, Wunsch H. Inappropriate opioid prescription after surgery. Lancet (London, England). 2019;393(10180):1547–1557. doi: 10.1016/S0140-6736(19)30428-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Summer GJ, Puntillo KA, Miaskowski C, Green PG, Levine JD. Burn Injury Pain: The Continuing Challenge. J Pain. 2007;8(7):533–548. doi: 10.1016/j.jpain.2007.02.426 [DOI] [PubMed] [Google Scholar]
- 7.Myers R, Lozenski J, Wyatt M, et al. Sedation and Analgesia for Dressing Change: A Survey of American Burn Association Burn Centers. J Burn care Res 2017;38(1):e48–e54. doi: 10.1097/BCR.0000000000000423 [DOI] [PubMed] [Google Scholar]
- 8.Wibbenmeyer L, Eid A, Kluesner K, et al. An evaluation of factors related to postoperative pain control in burn patients. J Burn Care Res 2015;36(5):580–586. doi: 10.1097/BCR.0000000000000199 [DOI] [PubMed] [Google Scholar]
- 9.Holtman JR, Jellish WS. Opioid-induced hyperalgesia and burn pain. J Burn Care Res 2012;33(6):692–701. doi: 10.1097/BCR.0b013e31825adcb0 [DOI] [PubMed] [Google Scholar]
- 10.Yenikomshian HA, Curtis EE, Carrougher GJ, Qiu Q, Gibran NS, Mandell SP. Outpatient opioid use of burn patients: A retrospective review. Burns 2019;45(8):1737–1742. doi: 10.1016/j.burns.2019.05.019 [DOI] [PubMed] [Google Scholar]
- 11.Wibbenmeyer L, Oltrogge K, Kluesner K, Zimmerman MB, Kealey PG. An evaluation of discharge opioid prescribing practices in a burn population. J Burn Care Res 2015;36(2):329–335. doi: 10.1097/BCR.0000000000000110 [DOI] [PubMed] [Google Scholar]
- 12.Tully A, Anderson L, Adams W, Mosier MJ. Opioid creep in burn center discharge regimens: Doubled amounts and complexity of narcotic prescriptions over seven years. Burns 2019;45(2):328–334. doi: 10.1016/j.burns.2018.08.004 [DOI] [PubMed] [Google Scholar]
- 13.Khan A, Parikh M, Minhajuddin A, et al. Opioid prescribing practices in a pediatric burn tertiary care facility: Is it time to change? Burns 2019;33(6):692–701. doi: 10.1016/j.burns.2019.07.016 [DOI] [PubMed] [Google Scholar]
- 14.Dowell Deborah, M Tamara. Haegerich RC. CDC Guideline for Prescribing Opioids for Chronic Pain-United States 2016. MMWR Recomm Rep 2016;65:1. https://www.cdc.gov/mmwr/volumes/65/rr/rr6501e1.htm.Accessed May 3, 2020. [DOI] [PubMed] [Google Scholar]
- 15.RED BOOK Select Extracts. Truven Health Analytics; 2019. Ann Arbor, MI. [Google Scholar]
- 16.Opioid Oral Morphine Milligram Equivalent (MME) Conversion Factors. Center for Medicare and Medicaid Services. https://www.cms.gov/Medicare/Prescription-Drug-Coverage/PrescriptionDrugCovContra/Downloads/Oral-MME-CFs-vFeb-2018.pdf.Published 2018. Accessed March 1, 2020.
- 17.Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130–1139. doi: 10.1097/01.mlr.0000182534.19832.83 [DOI] [PubMed] [Google Scholar]
- 18.Wald DA. Burns. In: Garmel GM, Mahadevan SV, eds. An Introduction to Clinical Emergency Medicine. 2nd ed. Cambridge: Cambridge University Press; 2012:207–220. doi:DOI: 10.1017/CBO9780511852091.022 [DOI] [Google Scholar]
- 19.American Burn Association. Burn Center Referral Criteria. http://ameriburn.org/wp-content/uploads/2017/05/burncenterreferralcriteria.pdf.Published 2006.
- 20.American Burn Association. National Burn Repository Report of Data from 2005–2014.; 2015. [Google Scholar]
- 21.Shahi N, Meier M, Phillips R, et al. Pain Management for Pediatric Burns in the Outpatient Setting: A Changing Paradigm? J Burn Care Res 2020;(In Press):1–6. doi: 10.1093/jbcr/iraa049 [DOI] [PubMed] [Google Scholar]
- 22.Ross E, Meyer WHD. Opiate Prescribing Following Burn and Trauma Care in Adults and Children. In: Academic Surgical Congress.; 2018. [Google Scholar]
- 23.Sites BD, Beach ML, Davis MA. Increases in the use of prescription opioid analgesics and the lack of improvement in disability metrics among users. Reg Anesth Pain Med 2014;39(1):6–12. doi: 10.1097/AAP.0000000000000022 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Shah R, Kuo YF, Westra J, Lin YL, Raji MA. Opioid Use and Pain Control After Total Hip and Knee Arthroplasty in the US, 2014 to 2017. JAMA Netw open. 2020;3(7):e2011972. doi: 10.1001/jamanetworkopen.2020.11972 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Raji MA, Kuo YF, Adhikari D, Baillargeon J, Goodwin JS. Decline in opioid prescribing after federal rescheduling of hydrocodone products. Pharmacoepidemiol Drug Saf 2018;27(5):513–519. doi: 10.1002/pds.4376 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Kuo Y-F, Raji MA, Liaw V, Baillargeon J, Goodwin JS. Opioid Prescriptions in Older Medicare Beneficiaries After the 2014 Federal Rescheduling of Hydrocodone Products. J Am Geriatr Soc 2018;66(5):945–953. doi: 10.1111/jgs.15332 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.National Conference of State Legislatures. Prescribing Policies: States Confront Opioid Overdose Epidemic. https://www.ncsl.org/research/health/prescribing-policies-states-confront-opioid-overdose-epidemic.aspx.Published 2018. Accessed May 5, 2020.
- 28.Bao Y, Pan Y, Taylor A, et al. Prescription Drug Monitoring Programs Are Associated With Sustained Reductions In Opioid Prescribing By Physicians. Heal Aff 2016;35(6):1045–1051. doi: 10.1377/hlthaff.2015.1673 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.CDC. CDC Guidelines for Prescribing Opioids for Chronic Pain. https://www.cdc.gov/drugoverdose/resources/data.html.
- 30.Zhang L, Yin H, Lei X, et al. A Systematic Review and Meta-Analysis of Clinical Effectiveness and Safety of Hydrogel Dressings in the Management of Skin Wounds. Front Bioeng Biotechnol 2019;7(November):1–16. doi: 10.3389/fbioe.2019.00342 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Kim DE, Pruskowski KA, Ainsworth CR, Linsenbardt HR, Rizzo JA, Cancio LC. A Review of Adjunctive Therapies for Burn Injury Pain during the Opioid Crisis. J Burn Care Res 2019;40(6):983–995. doi: 10.1093/jbcr/irz111 [DOI] [PubMed] [Google Scholar]
- 32.Connolly J 3rd, Javed Z, Raji MA, Chan W, Kuo Y-F, Baillargeon J. Predictors of Long-term Opioid Use Following Lumbar Fusion Surgery. Spine (Phila Pa 1976). 2017;42(18):1405–1411. doi: 10.1097/BRS.0000000000002133 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Shah R, Chou L-N, Kuo Y-F, Raji MA. Long-Term Opioid Therapy in Older Cancer Survivors: A Retrospective Cohort Study. J Am Geriatr Soc 2019;67(5):945–952. doi: 10.1111/jgs.15945 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Campbell CI, Bahorik AL, VanVeldhuisen P, Weisner C, Rubinstein AL, Ray GT. Use of a prescription opioid registry to examine opioid misuse and overdose in an integrated health system. Prev Med (Baltim). 2018;110:31–37. doi: 10.1016/j.ypmed.2018.01.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Bohnert ASB, Ilgen MA. Understanding links among opioid use, overdose, and suicide. N Engl J Med 2019;380(1):71–79. doi: 10.1056/NEJMra1802148 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.