Puntos claveEnglish 中文 (chinese) Pregunta
Entre los individuos a los que se les ha recetado terapia con opiáceos a largo plazo, ¿se relaciona la variación en la dosis de opiáceos con un mayor riesgo de sobredosis?
Conclusiones
En este estudio de caso control anidado que incluyó a 228 pacientes de casos que sufrieron una sobredosis de opiáceos y a 3547 pacientes de control que no la sufrieron, una alta variación en la dosis de opiáceos se relacionó con un incremento de más del triple en el riesgo de sobredosis de opiáceos, incluso después del realizar el control de la dosis.
Significado
Las conclusiones sugieren que, al controlar la terapia con opiáceos a largo plazo, los médicos deberían considerar el riesgo de sobredosis relacionada con la variabilidad de la dosis.
Importance
Attempts to discontinue opioid therapy to reduce the risk of overdose and adhere to prescribing guidelines may lead patients to be exposed to variability in opioid dosing. Such dose variability may increase the risk of opioid overdose even if therapy discontinuation is associated with a reduction in risk.
Objective
To examine the association between opioid dose variability and opioid overdose.
Design, Setting, and Participants
A nested case-control study was conducted in a large Colorado integrated health plan and delivery system from January 1, 2006, through June 30, 2018. Cohort members were individuals prescribed long-term opioid therapy.
Exposures
Dose variability was defined as the SD of the milligrams of morphine equivalents across each patient’s follow-up and categorized based on the quintile distribution of the SD in the cohort (0-5.3, 5.4-9.1, 9.2-14.6, 14.7-27.2, and >27.2 mg of morphine equivalents).
Main Outcomes and Measures
Opioid overdose cases were identified using International Classification of Diseases, Ninth Revision and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes. Each case patient with overdose was matched to up to 20 control patients using risk set sampling. Conditional logistic regression models were used to generate matched odds ratios and 95% CIs, adjusted for age, sex, race/ethnicity, drug or alcohol use disorder, tobacco use, benzodiazepine dispensings, medical comorbidities, mental health disorder, opioid dose, and opioid formulation.
Results
In a cohort of 14 898 patients (mean [SD] age, 56.3 [16.0] years; 8988 [60.3%] female) prescribed long-term opioid therapy, 228 case patients with incident opioid overdose were matched to 3547 control patients. The mean (SD) duration of opioid therapy was 36.7 (33.7) months in case patients and 33.0 (30.9) months in control patients. High-dose variability (SD >27.2 mg of morphine equivalents) was associated with a significantly increased risk of overdose compared with low-dose variability (matched odds ratio, 3.32; 95% CI, 1.63-6.77) independent of opioid dose.
Conclusions and Relevance
Variability in opioid dose may be a risk factor for opioid overdose, suggesting that practitioners should seek to minimize dose variability when managing long-term opioid therapy.
Epidemiologic studies1-5 have demonstrated that individuals prescribed high-dose opioid therapy are at increased risk for opioid overdose. These observational findings informed guidelines that encourage practitioners to minimize opioid prescribing for acute pain, to avoid initiating opioid therapy for chronic pain, and to consider tapering or discontinuing long-term opioid therapy when the risks of opioids outweigh the benefits.6-8 Although practice guidelines have led to substantial reductions in opioid prescribing across the United States,6-10 significant decreases in pharmaceutical opioid overdose have not been documented.11 It is thus possible that unexamined prescribing practices or unintended consequences of prescribing policies are contributing to persistently elevated pharmaceutical opioid overdose rates.12
Management of opioid therapy is a dynamic process that may result in a range of medication exposure patterns, including stable doses to changes in dose over time and therapy discontinuation. Changes in dose may include an increase, a decrease, or fluctuating increases and decreases over time. Studies of people initiating extended-release opioid treatment,13 initiating and discontinuing opioid agonist treatment for opioid use disorder,14-17 being released from prison,18 and being discharged from inpatient opioid use disorder treatment19 or hospitalization20 suggest that changes in dose may be associated with an increased risk of opioid overdose. The mechanisms that lead to these periods of risk are likely to be biological and behavioral, such as resuming opioid use after a period of reduced tolerance. To date, the association between dynamic changes in opioid dose or dose variability and opioid overdose has not been examined in the context of long-term opioid therapy.
The primary aim of this study was to examine the association between dose variability and the risk of opioid overdose. We also sought to quantify the association of sustained (≥3 months) opioid therapy discontinuation with opioid overdose. We hypothesized that increasing dose variability would be independently associated with opioid overdose and that dose variability would be more strongly associated with overdose at higher opioid doses. We also hypothesized that sustained opioid therapy discontinuation would be associated with a reduction in overdose risk.
We conducted a matched case-control study that was nested within a cohort of Kaiser Permanente Colorado (KPCO) patients. KPCO is an integrated health care plan and delivery system with approximately 630 000 members. Cohort members were 18 years or older and were prescribed long-term opioid therapy, with 3 or more opioid dispensings of 10 mg or more of daily morphine equivalents in a 90-day period. Opioid dispensings were identified from KPCO pharmacy records and external pharmacy claims using National Drug Codes. Prescriptions had to be dispensed in the ambulatory setting from January 1, 2006, through December 31, 2017, with no more than 10 days without any opioid coverage. For each cohort member, milligrams of morphine equivalents were calculated for every month of follow-up using established methods.4 This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. The KPCO Institutional Review Board approved the study before data collection, with a waiver of Health Insurance Portability and Accountability and informed consent. Data were not deidentified because this was a limited data set.
For entry into the long-term opioid cohort, buprenorphine-containing products were excluded because they are generally used for opioid use disorder treatment at KPCO. Methadone was included if it was dispensed at a pharmacy or if a pharmacy claim was generated; methadone dispensed in an addiction treatment center was excluded. Patients could have entered the cohort if they were already taking long-term opioid therapy in 2006 or if they newly started undergoing long-term opioid therapy from January 1, 2006, to December 31, 2017. The third opioid dispensing date of the 3 eligibility prescriptions represented the index dose. To be included in the cohort, patients needed at least 30 days of enrollment after their index dose. Patients were excluded from the cohort if they were in hospice, were in a nursing home, or had evidence of cancer during the follow-up. Patients were followed up until disenrollment from the health care plan, death, or June 30, 2018.
Across each patient’s follow-up, we assessed variability in dose by calculating the SD of the monthly milligrams of morphine equivalents over time from the index dose to the end of follow-up. A similar approach has been used to assess the association of heart rate variability and the risk of cardiovascular deaths.21,22 The dose variability exposure categories for the primary case-control analysis were based on the quintiles of the distribution of the SD in the cohort. Examining quintiles allowed us to examine a dose response between variability and overdose.
Case and Control Patients
In the cohort, we identified cases of fatal and nonfatal opioid overdose using the International Classification of Diseases, Ninth Revision (ICD-9) and the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes (eTable 1 in the Supplement). These codes have an estimated positive predictive value of 81%23 and were identified in inpatient and emergency department settings or using state vital statistics. The date of the first overdose represented the index date.
Using risk set sampling,24,25 we randomly selected up to 20 control patients for each case patient who experienced overdose. This number was based on a sample size calculation that indicated the need for 100 case patients and 2000 matched control patients in the top and bottom quintiles and 2000 matched control patients to detect a matched odds ratio (mOR) of 2.05, with 80% statistical power, assuming an α of .05 and a probability of high variability in dose exposure of 0.2.
Eligible control patients did not have an ICD-9 or ICD-10 code for opioid overdose before the index date. Control patients were matched to case patients by milligrams of morphine equivalents at the index dose (±20%), calendar time (±60 days), and length of follow-up (±60 days). Length of follow-up was defined as the time between the index dose and index date. Each case patient and his or her set of matched control patients represented a separate stratum.
Dose Variability, Sustained Opioid Therapy Discontinuation, and Dose for the Case-Control Analyses
For each case and control patient stratum, dose variability (SD), sustained opioid therapy discontinuation, and dose were assessed retrospectively from the index date. Dose variability was measured as the SD in dose between the index dose and the index date. For the primary analysis, the SD was categorized into the following groups based on the quintile distribution of SD in the cohort: 0 to 5.3, 5.4 to 9.1, 9.2 to 14.6, 14.7 to 27.2, and more than 27.2 mg of morphine equivalents.
Sustained therapy discontinuation was defined as 0 mg of morphine equivalents in the 3-month period before the index date and was coded as a dichotomous variable (yes or no). Dose was similarly calculated as the mean milligrams of morphine equivalents in the 3 months before the index date to capture the magnitude of dose in proximity to the overdose event. The doses were categorized as 0 to 20, 21 to 50, 51 to 100, and more than 100 mg of morphine equivalents.
Case patients and matched control patients were analyzed using conditional logistic regression to calculate mORs and 95% CIs. Two separate models were built: one for dose variability and one for sustained therapy discontinuation. In the first model, the dependent variable was overdose, and the main independent variable (exposure) was dose variability measured as a 5-level categorical variable (quintiles). The model was adjusted for the following variables, which have been reported in prior studies2,4,5,13,26-29 to be associated with overdose risk: opioid dose in the 3 months before the index date, age, sex, receipt of an extended-release and/or long-acting opioid formulation, mental health disorder diagnosis, drug or alcohol use disorder, benzodiazepine dispensings, medical comorbidity (Quan-Deyo Modified Charlson Comorbidity Index30,31), and tobacco use documented in the social history or tobacco use disorder as a diagnosis. Race/ethnicity was also included; data on race/ethnicity are populated in the KPCO electronic health record from patient self-report. Missing tobacco, race, and Hispanic ethnicity values were replaced using multiple imputations based on the available information from the matched sample. A dose-response association was tested for dose variability by modeling the variability categories (quintiles 1-5) as a continuous variable. To examine the opioid dose independent of dose variability, we analyzed dose without the dose variability measure in the model. To evaluate whether dose variability varied by the magnitude of the dose, dose variability and dose were also analyzed as an interaction. Model fit was assessed with the Hosmer-Lemeshow goodness-of-fit test on unmatched data,32 and multicollinearity was evaluated using the variance inflation factor.
The second multivariable conditional logistic regression model was built with overdose as the outcome and sustained therapy discontinuation as the main exposure variable, adjusting for the same covariates as the first model except for opioid dose in the 3 months before the index date.
We conducted 4 secondary analyses on dose variability, which all controlled for the same variables as the primary variability analysis. First, dose variability (SD) was modeled as a continuous variable. Second, to evaluate the association of the most recent opioid dose change with overdose, we modeled the net change in milligrams of morphine equivalents in the month immediately preceding the index date. Third, to account for increasing and decreasing trends in dose, we fit a linear regression of opioid dose for the case patients during their follow-up time. On the basis of the dose trend, we created 2 groups: decreasing or flat (coefficient for the linear trend ≤0) and increasing (coefficient for linear trend >0). We then modeled the dose trend as a dichotomous variable (decreasing or flat vs increasing). Fourth, to assess the timing of dose variability relative to the index date, we conducted a separate regression analysis that evaluated the association between dose variability in 3-, 6-, 9-, and 12-month periods before the index date and overdose. Dose variability was modeled as a continuous variable for these latter analyses.
We also conducted a secondary analysis on therapy discontinuation that controlled for the same variables as the primary therapy discontinuation analysis. In this secondary analysis, we examined the association between therapy discontinuation and overdose when use of opioids was discontinued only 1 month before the index date.
Sensitivity Analyses for Differential and Nondifferential Outcome Misclassification
A multiple-imputation method was used to address the uncertainty introduced by potential outcome (case) misclassification.33-35 On the basis of the literature, we assumed an overall 19% misclassification rate for the ICD-9 or ICD-10 overdose diagnoses. We used Monte Carlo simulation (5000 replications) to determine the levels of outcome misclassification that would have affected our conclusion on the association between dose variability and overdose. We simulated both nondifferential and differential outcome misclassification by dose variability (exposure) status.
All analyses were conducted using SAS statistical software, version 9.4 (SAS Institute Inc). All statistical tests were 2-sided, and P < .05 was considered statistically significant.
A cohort of 14 898 patients (mean [SD] age, 56.3 [16.0] years; 8988 [60.3%] female) was prescribed long-term opioid therapy (Table 1). The Figure describes the cohort inclusions and exclusions. During the follow-up, 10 885 individuals (73.1%) had a mental health disorder diagnosis, 4816 (32.3%) had a drug or alcohol use disorder, 5089 (34.2%) used tobacco or had a tobacco use disorder, and 12 887 (86.5%) had a chronic pain diagnosis. The index dose was greater than 50 mg of morphine equivalents for 4064 individuals (27.3%) in the cohort.
In the cohort of patients prescribed long-term opioid therapy, we identified 305 opioid overdose events (incidence rate, 456 per 100 000 person-years). Of these events, 57 were recurrent (in 42 individuals), 26 were fatal, and 39 occurred during a suicide attempt. Of 248 case patients with incident overdoses, 228 could be matched to at least 1 control. Of 14 650 potential control patients, case patients were matched to 3547 control patients. Table 1 gives the characteristics of the case and control patients. Four patients had overdoses that involved heroin, and 224 had overdoses primarily attributable to pharmaceutical opioids. The mean (SD) time undergoing opioid therapy before the index date was 36.7 (33.7) months among case patients and 33.0 (30.9) months among control patients, and the mean (SD) opioid doses before the index date were 110.6 (133.8) mg of morphine equivalents for case patients and 73.0 (112.9) mg of morphine equivalents for control patients.
In the first multivariable conditional logistic regression analysis, dose variability and dose were independently associated with overdose (Table 2). Individuals exposed to the highest category of dose variability (SD >27.2 mg of morphine equivalents) had an mOR of 3.32 (95% CI, 1.63-6.77) for experiencing an overdose compared with individuals exposed to the lowest category of dose variability (SD ≤5.3 mg of morphine equivalents). Individuals prescribed high doses (>100 mg of morphine equivalents) in the 3 months before the index date had an mOR of 2.37 (95% CI, 1.41-3.98) for experiencing an overdose compared with individuals prescribed lower doses (0-20 mg of morphine equivalents). The analysis evaluating variability in dose as dose response was statistically significant (β = 0.30; 95% CI, 0.14-0.45). Without dose variability in the model, an opioid dose greater than 100 mg of morphine equivalents was associated with a higher risk of overdose (mOR, 3.10; 95% CI, 1.85-5.19) compared with a dose of 0 to 20 mg of morphine equivalents. An interaction was not found between dose and variability in dose (χ212 = 8.81; P = .72), suggesting that the association of dose variability with overdose risk did not vary by the magnitude of dose. The Hosmer-Lemeshow test on the unmatched data indicated that model fit was adequate (χ28 = 5.49; P = .70). The tolerance was 0.74, and the variance inflation factor was 1.35, indicating an absence of multicollinearity between dose variability and dose.36
Individuals with sustained opioid therapy discontinuation (defined as 3 continuous months with 0 mg of morphine equivalents before the index date) were 51% less likely to have experienced an overdose than individuals who had not discontinued opioid therapy (mOR, 0.49; 95% CI, 0.26-0.93) (Table 3).
In the secondary analyses, dose variability measured as a continuous variable was associated with overdose (mOR, 1.01; 95% CI, 1.00-1.01; P = .02) (eTable 2 in the Supplement). The change in opioid dose 1 month before the index date was not associated with overdose (mOR, 1.00; 95% CI, 1.00-1.00; P = .36). An increasing dose trend was not associated with overdose compared with a flat or decreasing dose trend (mOR, 1.11; 95% CI, 0.77-1.58). In the analyses examining dose variability in the 3-, 6-, 9-, and 12-month periods before the index date, dose variability measured as a continuous variable was associated with overdose (mOR, 1.01; 95% CI, 1.00-1.01 for each; 3 months, P = .003; 6 months, P = .001; 9 months, P = .002; 12 months, P = .005). The estimate for therapy discontinuation 1 month before the index date was similar to the estimate for discontinuation 3 months before the index date (mOR, 0.51; 95% CI, 0.30-0.87).
The sensitivity analyses assuming 19% nondifferential misclassification of outcome status (overdose classification) changed the main effect estimates by 2.7% or less and did not change the main conclusion that high-dose variability (SD, >27.2) was positively associated with overdose (eTable 3 in the Supplement). When assuming 19% differential misclassification of outcome status by dose variability status, the estimate for dose variability was reduced by approximately 30% under the most extreme differential misclassification scenarios. The main conclusion did not change under any of the misclassification scenarios.
In this case-control study of patients prescribed long-term opioid therapy, there was an apparent dose-response association between variability in opioid dose and overdose risk. Opioid dose also demonstrated an association with short-term overdose risk, whereas sustained opioid therapy discontinuation was associated with an approximate 50% reduction in risk of overdose. Together, these findings suggest that tapering patients off long-term opioid therapy may be beneficial if it ultimately leads to discontinuation of opioid therapy, but they also suggest that attempts to modify opioid doses could lead to dose variability that may be associated with increased risk of opioid overdose.
Although we hypothesized that dose variability would be associated with higher risk in patients receiving high opioid doses, the interaction between dose variability and dose was not statistically significant. This finding suggests that even among patients receiving low doses of opioids, dose variability is a risk factor for overdose. The results of our secondary analyses suggest that our primary results were robust to measuring variability as a continuous variable, the timing of the variability, the overall dose trend, and the timing of the discontinuation. Future research could focus on how to use measures of dose variability to guide prescribing practices.
Although there could be numerous reasons why patients are exposed to dose variability, one reason is an attempt to taper or discontinue opioid therapy. An attempted dose reduction may be associated with opioid withdrawal or increased short-term pain,37-39 prompting patients to request a subsequent dose increase. Because of loss of tolerance after a period of reduced opioid exposure,40-42 such dose adjustments may be associated with increased risk of opioid overdose. Alternatively, after a dose reduction, patients may seek pharmaceutical opioids from other sources (eg, friends) without being aware of the potency of the opioid formulation they are ingesting, increasing the risk of overdose. Future research should specifically examine the effects of tapering on overdose.
Although our results corroborate prior epidemiologic studies2,4,5,26,36 demonstrating that opioid doses above 50 mg of morphine equivalents are associated with estimated 2- to 6-fold increased short-term risks of overdose among patients prescribed long-term opioid therapy, the dose association was attenuated (30.8%) when dose variability was included in the analyses. This finding suggests that prior analyses assessing the association between opioid dose and overdose may have been confounded by dose variability. Prior research27 has also found that opioid dose is not associated with the long-term (2-year) risk of overdose. Given that dose is likely to fluctuate over time, the poor predictive value of dose for 2-year overdose risk may be partially explained by dose variability.
Strengths and Limitations
This population-based study has several strengths. We assembled a cohort of longitudinally followed up patients receiving long-term opioid therapy in a large integrated health plan that is demographically representative of Colorado. In the cohort, overdose events were identified in the inpatient and emergency department settings and in vital records, and the estimated incidence rate of overdose was similar to or higher than rates among patients prescribed long-term opioids in prior studies.4,43 With use of a risk set sampling approach, randomly selected control patients were matched to case patients who experienced overdose by calendar time and length of follow-up time to help control for time-varying factors, such as secular trends in opioid prescribing policies and practices, potency and availability of illicit opioids, and health care system–specific overdose reduction interventions implemented across the follow-up period. In addition, the conditional logistic regression models were adjusted for several important potential confounding variables that are known risk factors for overdose.
This study also has limitations. Although the ICD-9 and ICD-10 overdose codes are associated with a high positive predictive value, up to 19% of the overdose cases may have been misclassified as false-positive. This misclassification could have been nondifferential or differential with respect to exposure status, leading to biased OR estimates. However, in our sensitivity analyses, we found that a wide range of exposure misclassification would not have changed our conclusion that dose variability may be associated with an increased risk of overdose among patients prescribed long-term opioid therapy.
Another potential limitation is loss to follow-up. It is possible that the patients who experienced dose variability may have sought illicit opioids such as heroin, lost their insurance, and experienced an overdose during a period of disenrollment from the health care plan. In our data, this would have appeared to be an exposed noncase when in fact it was an exposed case, implying that our observed OR was an underestimate of the true association. There could be multiple reasons why people experienced dose variability, including worsened pain, practitioner-initiated tapers, changing physicians, missed appointments, poor adherence to urine toxicology screening, or travel. If one of these reasons was driving the association, the observed association may have been diluted by combining them. Future research should examine specific reasons for dose variability and other potential outcomes, such as suicide.44
This study suggests that dose variability is a risk factor for opioid overdose independent of dose alone, whereas sustained therapy discontinuation may be protective of overdose. Additional studies are needed to better understand the pathways by which patients undergoing long-term opioid therapy can safely discontinue opioid therapy. Until such pathways can be elucidated, policymakers and physicians should consider the risks that may be associated with dose variability when designing and implementing new policies to reduce opioid prescribing.
Accepted for Publication: March 4, 2019.
Published: April 19, 2019. doi:10.1001/jamanetworkopen.2019.2613
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2019 Glanz JM et al. JAMA Network Open.
Corresponding Author: Jason M. Glanz, PhD, Institute for Health Research, Kaiser Permanente Colorado, 2550 S Parker Rd, Ste 200, Aurora, CO 80014 (jason.m.glanz@kp.org).
Author Contributions: Drs Glanz and Binswanger served as co–first authors, each with equal contribution to this work. Drs Glanz and Binswanger had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Glanz, Binswanger, Xu.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Glanz, Binswanger, Xu.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Glanz, Binswanger, Xu.
Obtained funding: Glanz, Binswanger.
Supervision: Glanz, Xu.
Conflict of Interest Disclosures: Dr Binswanger reported receiving royalties for educational content on the health of incarcerated persons from UpToDate. No other disclosures were reported.
Funding/Support: Research reported in this publication was supported by award 1R56DA044302 from the National Institute on Drug Abuse, National Institutes of Health.
Role of the Funder/Sponsor: The National Institutes of Health 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.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Drug Abuse or the National Institutes of Health. All information and materials in this article are original.
Additional Contributions: Jo Ann Shoup, PhD, and Kris Wain, MS, Institute for Health Research, Kaiser Permanente Colorado, Aurora, provided project management and data assistance. Their activities were supported by the grant.
3.Dasgupta
N, Funk
MJ, Proescholdbell
S, Hirsch
A, Ribisl
KM, Marshall
S. Cohort study of the impact of high-dose opioid analgesics on overdose mortality.
Pain Med. 2016;17(1):85-98.
PubMedGoogle Scholar 6.Dowell
D, Haegerich
TM, Chou
R. CDC guideline for prescribing opioids for chronic pain—United States, 2016.
MMWR Recomm Rep. 2016;65(1):1-49.
PubMed 7.US Department of Veterans Affairs/US Department of Defense. VA/DoD Clinical Practice Guideline for Opioid Therapy for Chronic Pain. Washington, DC: US Dept of Veterans Affairs/US Dept of Defense; 2017.
8.Rosenberg
JM, Bilka
BM, Wilson
SM, Spevak
C. Opioid therapy for chronic pain: overview of the 2017 US Department of Veterans Affairs and US Department of Defense clinical practice guideline.
Pain Med. 2018;19(5):928-941. doi:
10.1093/pm/pnx203PubMedGoogle ScholarCrossref 9.Bohnert
ASB, Guy
GP
Jr, Losby
JL. Opioid prescribing in the United States before and after the Centers for Disease Control and Prevention’s 2016 opioid guideline.
Ann Intern Med. 2018;169(6):367-375. doi:
10.7326/M18-1243PubMedGoogle ScholarCrossref 11.Centers for Disease Control and Prevention. 2018 Annual Surveillance Report of Drug-Related Risks and Outcomes—United States: Surveillance Special Report. Washington, DC: Centers for Disease Control and Prevention, US Dept of Health and Human Services; August 31, 2018.
12.Hedegaard
H, Miniño
AM, Warner
M. Drug Overdose Deaths in the United States, 1999–2017. Hyattsville, MD: National Center for Health Statistics; 2018.
14.Cornish
R, Macleod
J, Strang
J, Vickerman
P, Hickman
M. Risk of death during and after opiate substitution treatment in primary care: prospective observational study in UK General Practice Research Database.
BMJ. 2010;341:c5475. doi:
10.1136/bmj.c5475PubMedGoogle ScholarCrossref 16.Ma
J, Bao
Y-P, Wang
R-J,
et al. Effects of medication-assisted treatment on mortality among opioids users: a systematic review and meta-analysis [published online June 22, 2018].
Mol Psychiatry.
PubMedGoogle Scholar 23.Green
CA, Perrin
NA, Janoff
SL, Campbell
CI, Chilcoat
HD, Coplan
PM. Assessing the accuracy of opioid overdose and poisoning codes in diagnostic information from electronic health records, claims data, and death records.
Pharmacoepidemiol Drug Saf. 2017;26(5):509-517. doi:
10.1002/pds.4157PubMedGoogle ScholarCrossref 29.Zedler
B, Xie
L, Wang
L,
et al. Development of a risk index for serious prescription opioid-induced respiratory depression or overdose in Veterans’ Health Administration patients.
Pain Med. 2015;16(8):1566-1579. doi:
10.1111/pme.12777PubMedGoogle ScholarCrossref 32.Hosmer
DW
Jr, Lemeshow
S, Sturdivant
RX.
Applied Logistic Regression. Vol 398. New York, NY: John Wiley & Sons; 2013. doi:
10.1002/9781118548387 35.Rubin
DB. Multiple Imputation for Nonresponse in Surveys. Vol 81. New York, NY: John Wiley & Sons; 2004.
38.Kennedy
LC, Binswanger
IA, Mueller
SR,
et al. “Those conversations in my experience don’t go well”: a qualitative study of primary care provider experiences tapering long-term opioid medications.
Pain Med. 2018;19(11):2201-2211. doi:
10.1093/pm/pnx276PubMedGoogle ScholarCrossref 42.Young
JC, Lund
JL, Dasgupta
N, Jonsson Funk
M. Opioid tolerance and clinically recognized opioid poisoning among patients prescribed extended-release long-acting opioids.
Pharmacoepidemiol Drug Saf. 2019;28(1):39-47. doi:
10.1002/pds.4572PubMedGoogle ScholarCrossref