- Email: [email protected]
One-Year Mortality After Emergency Department Visit for Nonfatal Opioid Poisoning: A Population-Based Analysis
TOXICOLOGY/ORIGINAL RESEARCH
One-Year Mortality After Emergency Department Visit for Nonfatal Opioid Poisoning: A PopulationBased Analysis Pamela Leece, MD, MSc*; Cynthia Chen, MSc; Heather Manson, MD, MHSc; Aaron M. Orkin, MD, MSc; Brian Schwartz, MD, MScCH; David N. Juurlink, MD, PhD; Tara Gomes, PhD *Corresponding Author. E-mail: [email protected], Twitter: @PublicHealthON.
Study objective: We aim to characterize the incidence and risk factors for opioid-related and all-cause mortality in the year after an emergency department (ED) visit for nonfatal opioid poisoning by conducting a population-based study. Methods: We used linked health care databases in Ontario, Canada, to identify individuals who attended an ED for nonfatal opioid poisoning between January 1, 2015, and December 31, 2016. Using Cox proportional hazards regression, we examined predictors of mortality in the year after discharge (ED or hospital, if admitted). Results: In this cohort (n¼6,140), 327 individuals (5.3%) died of any cause and 118 (1.9%) died of opioid-related causes within 1 year. Adjusting for other covariates, we found that health service use in the first week was not protective for opioid-related death (hazard ratio [HR] 0.70; 95% confidence interval [CI] 0.47 to 1.06) or all-cause mortality (HR 0.98; 95% CI 0.78 to 1.24). In exploring other covariates, predictors of opioid-related mortality included male sex (HR 1.98; 95% CI 1.32 to 2.97) and using opioid agonist therapy (HR 1.79; 95% CI 1.15 to 2.80) or benzodiazepine (HR 1.54; 95% CI 1.02 to 2.31) in the 12 months before the index event. Assessment by a family physician in the previous 12 months was associated with a lower risk of opioid-related and all-cause mortality (HR 0.58, 95% CI 0.39 to 0.86; and HR 0.63, 95% CI 0.49 to 0.82, respectively). Conclusion: We identified predictors of opioid-related and all-cause mortality after ED presentation for opioid poisoning. Several predictors of mortality may facilitate targeted interventions. [Ann Emerg Med. 2019;-:1-9.] Please see page XX for the Editor’s Capsule Summary of this article. 0196-0644/$-see front matter Crown Copyright © 2019 Published by Elsevier, Inc on behalf of the American College of Emergency Physicians. https://doi.org/10.1016/j.annemergmed.2019.07.021
INTRODUCTION Background The global burden of disease attributable to opioid dependence has increased since 1990, with an estimated 9.2 million disability-adjusted years of life lost in 2010,1 exceeding the years of life lost attributable to pneumonia, HIV/AIDS, influenza, and alcohol use disorder in some regions.2 Since 2010, the burden of opioid-related disease has escalated significantly, particularly in North America, owing to the emergence of fentanyl and its analogues in the illicit drug supply.3,4 Opioid-related emergency department (ED) visits nearly doubled in the United States between 2005 and 2014, to a rate of 177.7 per 100,000.5 Similarly, ED visits in Canada for opioid poisoning increased substantially in Canada, reaching rates of 34.7 and 88.2 per 100,000 in the 2016 fiscal year in the provinces of Ontario and Alberta, respectively.6 Volume
-,
no.
-
:
-
2019
Importance Patient encounters with health care providers, such as ED visits for opioid poisoning, present opportunities to prevent subsequent death among people experiencing opioid-related harms. Patterns of health care use before and after an ED visit for nonfatal poisoning can also be assessed to identify system-level opportunities for improvement. There is little information on the rates and risk factors for death after treatment for nonfatal opioid poisoning.7-15 Specific to the ED setting, 2 studies conducted in a large health care plan in the United States (records for up to 1.2 million unique patients)16,17 estimated between 7% and 9.4% all-cause mortality within 1 year after treatment in the ED for opioid poisoning. However, we are not aware of studies that assess specifically for opioid-related mortality greater than 48 hours after nonfatal opioid poisoning. A small number of previous studies have examined risk factors for all-cause mortality among patients treated for opioid Annals of Emergency Medicine 1
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
Editor’s Capsule Summary
What is already known on this topic The emergency department (ED) has an important role in detecting and starting care for opioid use disorder patients. What question this study addressed How often do patients die after being treated in an ED for a nonfatal opioid use event? What this study adds to our knowledge Of a total of 6,140 Ontario patients treated in 2015 to 2016, 5.3% died within a year and 1.9% of deaths were directly related to opioid use. Men, those using an opioid agonist, and those also receiving a benzodiazepine had a higher frequency of death. How this is relevant to clinical practice This reinforces the opportunities to identify and try to help a group at a higher risk of death after ED discharge.
poisoning.8,9,13,14,16 One previous study in the ED setting found higher risk associated with higher comorbidity and medication use (benzodiazepine and opioid prescriptions) and lower risk with mental health disorders, having a primary care provider within the health system, commercial or private insurance, or a prescription for a muscle relaxant.16 The limited available literature reveals important gaps in regard to outcomes after ED treatment for opioid poisoning. Goals of This Investigation We sought to characterize the incidence of and risk factors for opioid-related and all-cause mortality in the year after an ED visit for nonfatal opioid poisoning. We hypothesized that better follow-up care may be associated with lower mortality, according to analogy with other health conditions.18 MATERIALS AND METHODS Study Design and Setting We conducted a retrospective population-based cohort study of Ontario residents presenting to an ED for nonfatal opioid poisoning between January 1, 2015, and December 31, 2016 (accrual period). We used a followup period of 12 months to identify subsequent all-cause and opioid-related deaths, and a Cox proportional 2 Annals of Emergency Medicine
hazards model to examine predictors of mortality within 1 year. Ontarians have universal access to single-payer publicly funded hospital and physician services (including ED care); however, prescription drug coverage varies, whereby some residents have public or private drug coverage and others pay for medications out of pocket. The study was approved by the Research Ethics Boards of Public Health Ontario, Toronto, Ontario, and Sunnybrook Health Sciences Centre, Toronto, Ontario. The reporting of this study follows the Reporting of Studies Conducted Using Observational RoutinelyCollected Health Data statement (Appendix E1, available online at http://www.annemergmed.com).19 Selection of Participants In the analysis, we included individuals aged 10 to 95 years who were alive at discharge from either the ED or related hospital admission after an ED visit for opioid poisoning during the accrual period (International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Canada [ICD-10-CA] codes T40.0 to 40.6 and T40.6). We defined the index date as the hospital discharge date of the episode of care resulting from the ED visit for opioid poisoning. If there were multiple ED visits for opioid poisoning, we selected the first such poisoning for inclusion in our study. We excluded individuals who received palliative care in the past 12 months and those who had any previous cancer diagnosis because they have a unique set of clinical circumstances compared with the group that was the target of our study. We defined an opioid-related death as any death recorded in the Drug and Drug-Alcohol Related Death database of confirmed opioid-related deaths. Methods of Measurement We used comprehensive provincial health care administrative data sets to conduct this analysis, which are linked at the individual level in an anonymous fashion by unique identifiers and analyzed at the Institute for Clinical Evaluative Sciences (https://www. ices.on.ca/; also see their Data Dictionary). Each health care record has a unique identifier assigned, an encrypted version of each patient’s unique 10-digit health card number. The databases are linked at the Institute for Clinical Evaluative Sciences with the identifier, and the databases used for this study have greater than 95% of records with a valid identifier. We identified ED visits for opioid poisoning by using the Canadian Institute for Health Information (CIHI) Volume
-,
no.
-
:
-
2019
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
National Ambulatory Care Reporting System, which contains comprehensive records of ED visits in Ontario. To describe the demographic characteristics of the cohort, we used the Registered Persons Database. To identify comorbidities and health services use, we used the diagnostic code (ICD-10) associated with health service visits in the past 12 months, recorded in the CIHI Discharge Abstract Database, CIHI National Ambulatory Care Reporting System, Ontario Health Insurance Plan Database, and the CIHI Ontario Mental Health Reporting System databases. We also described patient comorbidities and traumatic injury with the Ontario Asthma data set and Ontario Chronic Obstructive Pulmonary Disease data set of the Ontario Trauma Registry. We excluded individuals with a cancer diagnosis by using the Ontario Cancer Registry. The Aggregated Diagnosis Group system was used to score comorbidity in the preceding year (version 10.0; The Johns Hopkins ACG System, Baltimore, MD). We compared health services use in the 12 months before and 12 months after the index date, and used the Ontario Narcotics Monitoring System to characterize the dispensing of opioids, opioid agonist therapy (eg, methadone, buprenorphine), and benzodiazepines (referred to here as “monitored medications”). We compared monitored medication use in the previous 12 months with that in the 12 months after the index date. We defined patient sociodemographic characteristics (age, sex, urban/rural location of residence, and income quintile), elements of the index event (admission reason, triage level, and discharge disposition), comorbidities (weighted Aggregated Diagnosis Group quintile and specific diagnoses), health services use, and monitored medication use. The main exposure for our primary analysis was any physician or hospital health care visit that occurred within 7 days of the index date. A complete description of these variables is provided in Table E1 (available online at http://www.annemergmed.com).
Outcome Measures Confirmed opioid-related deaths were identified with the Drug and Drug-Alcohol Related Death database, which contains records abstracted from the Office of the Chief Coroner for Ontario. The methods for this abstraction have been described previously, and the database includes deaths for which the pathologist or coroner conducting the death investigation determined that opioids directly contributed to death, according to postmortem toxicology and other findings.20 We identified deaths from any cause by using the Registered Persons Database. Volume
-,
no.
-
:
-
2019
Primary Data Analysis The variables in our analysis were chosen to investigate our hypothesis in regard to health care access in the first 7 days, as well as important characteristics related to risk of subsequent mortality. We included characteristics of the individual (age and sex), comorbidities (Aggregated Diagnosis Group and mental health and substance use), recent prescription medication use (opioids, opioid agonist therapy, and benzodiazepines), and health service use (ED, hospital, and family physician visits). We used frequencies, percentages, medians, and interquartile ranges to describe cohort demographic characteristics, comorbidities, health service use, and prescribed medications. We described missing values with a separate category. We used Cox proportional hazards regression to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between opioid-related or all-cause death and receipt of health care services in the 7 days after the index date. The length of follow-up in our study was either 365 days or the number of days between death date and index date, whichever was shorter. In exploring other covariates in the model, we examined the association for demographic characteristics, comorbidities, health service use, and use of monitored medications. We adjusted all models for potential risk factors, including baseline characteristics (age and sex), health service use in the 12 months before the index event (1 or more visits to an ED or a hospital or mental health admission, or family physician visits), use of monitored medications (yes/no) in the 12 months before the index event (opioids, opioid agonist therapy, and benzodiazepines), comorbidities (Aggregated Diagnosis Group) in the previous 12 months, or mental health disorder (psychotic or nonpsychotic), substance use disorder, or previous opioid poisoning. All analyses used a 2-sided type I error rate of 0.05 as the threshold for statistical significance and were performed with SAS (version 9.3; SAS Institute, Inc., Cary, NC). Our analytic plan is accessible in Appendix E2 (available online at http:// www.annemergmed.com). We assessed 3 variables as interaction terms in our final model: Aggregated Diagnosis Group and mental health or substance use diagnosis, Aggregated Diagnosis Group and health service use in the past 12 months, and mental health or substance use diagnosis and opioid agonist therapy. To assess multicollinearity, we examined the variance inflation factor for all variables in our model. Furthermore, we used the Akaike information criterion statistic to examine the model fit. We calculated standardized mortality ratios for all-cause and opioid-related mortality, using the Ontario 2016 Annals of Emergency Medicine 3
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
population as a reference. We used the Drug and DrugAlcohol Related Death database to calculate the expected opioid-related death rate in the population within 3 age categories selected a priori (10 to 24, 25 to 54, and 55 to 95 years) and by sex. These categories are informed by known epidemiology in Ontario for age of opioid-related death,21,22 with most deaths occurring in individuals aged 25 and 54 years. In provincial data, few if any opioidrelated deaths occur in individuals younger than 10 years or older than 95 years. We used the Registered Persons Database to calculate the expected mortality rate from all causes in the population, using the same categories. The standardized mortality ratios were calculated with the ratio of observed deaths in our cohort to expected deaths in the population. We also calculated 95% CIs for each standardized mortality ratio, using the following formula: standardized mortality ratio1.96standard error standardized mortality ratio. RESULTS Characteristics of Study Subjects During the 2-year study period, we identified 6,904 patients who attended an ED in Ontario for opioid poisoning. After exclusions, 6,140 individuals met our study criteria (Figure). The characteristics of the cohort are shown in Table 1, and Table 2 describes health service use and use of monitored medications at baseline. The median age of patients was 38 years (interquartile range 27 to 53 years) and 52.8% were men. The most common diagnostic codes in the past 12 months were mental health disorders (58.2%), substance use diagnoses (40.5%), and low back pain (16.2%). In the past 12 months, 1.5% had been treated for nonfatal opioid poisoning, although 74.3% had attended an ED overall. Opioid, opioid agonist therapy, and benzodiazepine prescribing in the past 12 months occurred among 50.8%, 23.3%, and 41.1% of patients, respectively. ED visits, hospital admissions, and opioid use decreased when 12 months before and after the index event were compared. Table E2 (available online at http://www. annemergmed.com) describes the characteristics of our cohort after inclusion of 83 individuals who died on or before the index date (eg, did not survive to discharge; n¼6,223). There were no meaningful differences in characteristics between this group and the group included in our main analysis, other than death after ED arrival. Main Results Among the index ED visits for opioid toxicity, 70.0% were classified as unintentional opioid poisonings. The 4 Annals of Emergency Medicine
triage severity score was high for most patients: 18.7% were classified as resuscitation and 62.4% were classified as emergency. More than a third of patients were hospitalized (39.5%), whereas 54.8% were discharged from the ED to home or their place of residence, and 5.8% left the ED before being discharged by a physician. In comparison of health service use before and after the index event, there were meaningful decreases in hospital admissions (42.2% versus 26.2%; standardized difference –0.34) (Table 2). In the 12-month follow-up, 118 subjects (1.9%) died of opioid-related causes, whereas 327 (5.3%) died from any cause (see Figure E1, available online at http://www. annemergmed.com, for histogram of all-cause mortality; a histogram is not available for opioid-related mortality because of small cells). Adjusting for other variables, contact with the health care system within 7 days after the index date was not associated with opioid-related mortality (HR 0.70; 95% CI 0.47 to 1.06) (Table 3; see Table E3, available online at http://www.annemergmed.com, for univariate analysis) or all-cause mortality (HR 0.98; 95% CI 0.78 to 1.24) (Table 4; see Table E4, available online at http://www.annemergmed.com, for univariate analysis). The standardized mortality ratio for all-cause mortality was
6904
6787
6548
6223
6140
Figure. Inclusion and exclusion criteria applied to populationbased analysis. Index date refers to the discharge date of the health care episode involved in the ED visit for opioid poisoning. Volume
-,
no.
-
:
-
2019
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
Table 1. Characteristics of individuals with nonfatal opioid poisoning at the index ED visit. Cohort (N[6,140)
Characteristic Median age (IQR), y
38 (27–53)
Age, No. (%), y
Table 1. Continued. Cohort (N[6,140)
Characteristic Mental health or substance use diagnosis (including poisoning with any intent)
4,507 (73.4)
10–24
1,122 (18.3)
Alcohol-related
595 (9.7)
25–54
3,700 (60.3)
Substance use
2,484 (40.5)
1,318 (21.5)
Mental health
3,571 (58.2)
3,239 (52.8)
Opioid poisoning
55–95 Men, No. (%) Rurality, No. (%) Rural
726 (11.8)
Urban
5,407 (88.1)
Missing
7 (0.1)
Income quintile, No. (%)*
Accidental
64 (1.0)
Intentional
28 (0.5)
Pain Osteoarthritis
613 (10.0)
Rheumatoid arthritis
103 (1.7)
1,942 (31.6)
Fibromyalgia
2
1,364 (22.2)
Low back pain
3
1,056 (17.2)
Posttraumatic pain
1 (lowest)
4
911 (14.8)
5 (highest)
806 (13.1)
Missing Accidental
4,297 (70.0)
Intentional
1,843 (30.0)
Triage level, No. (%) Resuscitation
1,147 (18.7)
Emergency
3,833 (62.4)
Urgent
1,015 (16.5)
Less urgent/semiurgent Nonurgent/unknown
131 (2.1) 14 (0.2)
Disposition, No. (%) Discharged from ED to home or place of residence Registered but left ED
3,363 (54.8) 355 (5.8)
Hospital admission
1,140 (18.6)
Admitted to special care unit
919 (15.0)
Mental health admission
363 (5.9)
Weighted ADG quintile, No. (%)* 1 (lowest)
1,228 (20.0)
2
1,237 (20.2)
3
1,339 (21.8)
4
1,122 (18.3)
5 (highest)
1,214 (19.8)
Diagnoses, No. (%)* Chronic lung disease
127 (2.1)
Chronic liver disease
477 (7.8)
Chronic kidney disease
221 (3.6)
Heart disease (coronary artery disease or arrhythmia)
839 (13.7)
-,
39 (0.6) 992 (16.2) 35 (0.6)
IQR, Interquartile range; ADG, Aggregated Diagnosis Group. *In the 12 months before the index date.
61 (1.0)
Reason
Volume
92 (1.5)
no.
-
:
-
2019
10.0 (95% CI 9.0 to 11.1), and the standardized mortality ratio for opioid-related mortality was 238.7 (95% CI 195.6 to 281.7). All interactions were nonsignificant, so we did not include interaction terms in the final models for allcause or opioid-related mortality. Examination of other model covariates revealed that predictors of opioid-related mortality were male sex (HR 1.98; 95% CI 1.32 to 2.97) and using opioid agonist therapy (HR 1.79; 95% CI 1.15 to 2.80) or a benzodiazepine (HR 1.54; 95% CI 1.02 to 2.31) in the 12 months preceding the index event (Table 3; see Table E3, available online at http://www.annemergmed.com, for univariate analysis). All-cause mortality was associated with age groups 25 to 54 and 55 to 95 years (reference age 10 to 24 years) (HR 1.81, 95% CI 1.13 to 2.91; and HR 4.13, 95% CI 2.51 to 6.79, respectively), male sex (HR 1.62; 95% CI 1.28 to 2.03), highest comorbidity (Aggregated Diagnosis Group quintile 4 or 5) (HR 1.92; 95% CI 1.50 to 2.46), benzodiazepine use (HR 1.44; 95% CI 1.13 to 1.84), and hospital or ED admission (HR 1.69; 95% CI 1.10 to 2.61) (Table 4; see Table E4, available online at http://www.annemergmed.com, for univariate analysis). For all-cause mortality, an addiction or mental health diagnosis was associated with a lower risk (HR 0.69; 95% CI 0.53 to 0.89). Furthermore, 1 or more visits to a family physician in the previous 12 months was a protective factor for opioid-related and all-cause mortality (HR 0.58, 95% CI 0.39 to 0.86; and HR 0.63, 95% CI 0.49 to 0.82,
Annals of Emergency Medicine 5
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning Table 2. Health service use and controlled medication use among individuals with nonfatal opioid poisoning. Health Service and Medication Use Health service use (‡1 visit), No. (%)
Past Subsequent Standardized 12 Months* 12 Months* Difference N¼6,140
N¼6,140 4,432 (72.2)
–0.36
ED
4,563 (74.3)
4,069 (66.3)
–0.18
Hospital admission
2,592 (42.2)
1,609 (26.2)
–0.34
836 (13.6)
772 (12.6)
–0.03
4,620 (75.24) 4,436 (72.2)
–0.07
Family physician
HR,* Adjusted Model (95% CI)
Characteristic (Cohort N[6,140) Hypothesis
ED or hospital admission 5,309 (86.5)
Mental health admission
Table 3. Characteristics associated with opioid-related mortality in the year after nonfatal opioid poisoning.
0.70 (0.47–1.06)
Any health service use in the week after index date Exploratory analyses Age, y
Monitored medication use, No. (%) Opioid
3,116 (50.8)
2,625 (42.8)
–0.16
OAT
1,431 (23.3)
1,657 (27.0)
0.08
Benzodiazepine
2,523 (41.1)
2,413 (39.3)
–0.04
OAT, Opioid agonist therapy. *Relative to the index date.
respectively). In accordance with this finding, we estimated that 20 opioid-related deaths (118 observed, 98 expected) and 12 all-cause deaths (327 observed, 315 expected) could have been prevented if all members of the cohort had been assessed by a family physician in the previous 12 months. Tables E5 and E6 (available online at http://www. annemergmed.com) present results stratified by whether individuals were discharged directly from the ED or from related hospital admission. Results were generally consistent, except that use of opioid agonist therapy was associated with risk of opioid-related death among patients admitted to the hospital but not for those discharged from the ED. LIMITATIONS Our study has several limitations. First, because we were using administrative data, there were no clinical details available that may have allowed us to be more specific in our description of patient characteristics or health care at the nonfatal opioid poisoning. Second, we ended our follow-up in 2017, and the current risk factors among patients presenting to the ED for nonfatal opioid poisoning may have shifted as the epidemic evolved. We have framed our hypothesis as care within 1 week, although immediate access is preferable for patients with opioid use disorder who need treatments such as opioid agonist therapy. There is risk of bias in our study through censoring; for example, if individuals died after discharge and before a health care visit. In the first week, there were 18 deaths, and 13 of 6 Annals of Emergency Medicine
10–24
1.0 [Reference]
25–54
1.43 (0.80–2.56)
55–95
1.21 (0.59–2.51)
Men
1.98 (1.32–2.97)
Weighted ADG quintile† 1–3 (lowest)
1.0 [Reference]
4–5 (highest)
1.43 (0.96–2.13) 0.83 (0.51–1.36)
Mental health or substance use diagnosis (including poisoning with any intent)† Monitored medication use†,‡ Opioid
0.96 (0.64–1.46)
OAT
1.79 (1.15–2.80)
Benzodiazepine
1.54 (1.02–2.31)
Health service use† ED or hospital admission
1.34 (0.75–2.41)
Family physician
0.58 (0.39–0.86)
*Adjusted for potential risk factors in the past 12 months, including baseline characteristics (age and sex), comorbidities (weighted ADG and mental health or substance-use-related diagnoses, including opioid poisoning), monitored medication use (in the previous 12 months) (opioid, methadone or buprenorphine, and benzodiazepine), and health service use (1 or more ED, hospital, or mental health admissions, or family physician visits). † In the 12 months before the index date. ‡ Monitored medications refer to opioids, opioid agonist therapies, and benzodiazepines.
these individuals did not have a health care visit before their death. This could affect the test of our hypothesis because death occurred before the opportunity to visit a health care professional. Additionally, we were unable to detect nonfatal opioid poisoning events treated in the community (eg, emergency medical services, community-based naloxone use without transfer to the hospital). Although some ethnic and cultural groups may be at higher risk for opioid poisoning, this aspect was out of scope for our initial exploration of these data. Databases included in our study capture the majority of physician services in Volume
-,
no.
-
:
-
2019
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
Table 4. Characteristics associated with all-cause mortality in the year after nonfatal opioid poisoning. HR* Adjusted Model (95% CI)
Characteristic (Cohort N[6,140) Hypothesis Any health service use in the week after index date
0.98 (0.78–1.24)
Exploratory analyses Age, y 10–24
1.0 [Reference]
25–54
1.81 (1.13–2.91)
55–95
4.13 (2.51–6.79)
Men
1.62 (1.29–2.03)
Weighted ADG quintile† 1–3 (lowest)
1.0 [Reference]
4–5 (highest)
1.92 (1.50–2.46)
Mental health or addictions diagnosis (including poisoning with any intent)†
0.69 (0.53–0.89)
Monitored medication use†,‡ Opioid
1.20 (0.92–1.57)
OAT
1.09 (0.80–1.49)
Benzodiazepine
1.44 (1.13–1.84)
Health service use† ED or hospital admission
1.69 (1.10–2.61)
Family physician
0.63 (0.49–0.82)
*Adjusted for potential risk factors in the past 12 months, including baseline characteristics (age and sex), comorbidities (weighted ADG and mental health or substance-use-related diagnoses, including opioid poisoning), monitored medication use (in the previous 12 months) (opioid, methadone or buprenorphine, and benzodiazepine), and health service use (1 or more ED, hospital, or mental health admissions, or family physician visits). † In the 12 months before the index date. ‡ Monitored medications refer to opioids, opioid agonist therapies, and benzodiazepines.
Ontario, although there are a few hundred family physicians who work in community health centers in Ontario, where the funding model does not require submission of individual service claims. We were unable to adjust our analyses for movement in and out of the province, although this likely represents a very small proportion. Patients treated outside of the ED may have important differences compared with those treated in the ED. The generalizability of our results to the Ontario population is limited to people who visit the ED in for opioid toxicity and survive to discharge, but our findings are not generalizable to those who may experience poisoning and never come to the hospital. Our study Volume
-,
no.
-
:
-
2019
results may not be generalizable outside of Ontario, where the dynamics of the opioid epidemic and health care systems differ.
DISCUSSION In this population-based study, we found that among people discharged from the hospital after opioid poisoning, 1 in 52 died of opioid-related causes and 1 in 19 died from any cause within 1 year. The opioid-related and all-cause mortality were higher among this group compared with the general population, particularly for opioid-related mortality. We also found that male sex and use of opioid agonist therapy or benzodiazepines in the previous 12 months were associated with an increased risk of opioidrelated death among patients in this cohort. Furthermore, risk factors for all-cause mortality included age, male sex, comorbidity, benzodiazepine use, and hospital or ED admissions in the previous 12 months. Visits to a family physician in the year before the ED visit for nonfatal poisoning were protective for opioid-related and all-cause mortality. Our study advances the literature by, to our knowledge, providing the largest available population-based analysis of individuals treated in the ED for opioid poisoning, and characterizing for the first time the risk factors for opioid-related deaths. The incidence of all-cause mortality at 1 year in our study (5.3%) was lower than in 2 previous studies conducted within a US health care plan,16,17 and similar to that in a recent statewide study in Massachusetts.23 The rate of opioid-related mortality is higher in the United States than in Canada, which may contribute to differences compared with our study findings.4,24 However, the variation between studies in the United States may be due to different methods in assessing the outcome. The study in Massachusetts could have produced a lower estimate because of challenges with data linkage.23 As in the previous US health care plan study, we found an increased risk for all-cause mortality among patients with benzodiazepine use.16 We found discordant results, with an association between mental health and substance-use-related diagnoses and lower all-cause mortality, but higher opioid-related mortality for individuals prescribed opioid agonist therapy. In previous literature, substance use disorders were a significant risk factor for all-cause mortality after opioid poisoning.16 Our results may reflect the decision to combine outcomes for mental health and substance use diagnoses, and the use of opioid agonist therapy in the 12 months before the index event may be an indicator for opioid use disorder as a risk factor. Our finding in regard to mental Annals of Emergency Medicine 7
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
health and substance use is similar to that in a US study that found a protective association with mental health disorders and all-cause mortality after opioid poisoning.16 The finding that opioid agonist therapy in the previous 12 months was associated with opioidrelated mortality for patients discharged from the hospital but not for those discharged from the ED could indicate that these patients had additional complexities or were at risk because of potential interruptions in treatment. Likewise, our finding that family physician visits were protective could reflect that patients with additional disease complexity may not consult their physician as often. In summary, we found several predictors of mortality that may facilitate targeted interventions to reduce risk of subsequent death among people who attended an ED for nonfatal opioid poisoning. Particularly, our findings suggest that deaths may be averted by improving access to primary care for individuals at risk of ED visits because of opioid poisoning. Further research is needed to assess the effectiveness of interventions to prevent mortality after nonfatal opioid poisoning. Supervising editor: Donald M. Yealy, MD. Specific detailed information about possible conflict of interest for individual editors is available at https://www.annemergmed.com/editors. Author affiliations: From Health Promotion, Chronic Disease, and Injury Prevention, Public Health Ontario, Toronto, Ontario, Canada (Leece, Chen, Manson, Schwartz); the Department of Family and Community Medicine (Leece, Orkin, Schwartz), Dalla Lana School of Public Health (Leece, Manson, Orkin, Schwartz), Institute of Health Policy, Management, and Evaluation (Juurlink, Gomes), and Leslie Dan Faculty of Pharmacy (Gomes), University of Toronto, Ontario, Canada; The Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada (Chen, Juurlink, Gomes); the School of Public Health and Health Systems, University of Waterloo, Waterloo, Ontario, Canada (Manson); Schwartz/Reisman Emergency Medicine Institute, Sinai Health System, Toronto, Ontario, Canada (Orkin); the Department of Medicine, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada (Juurlink); The Sunnybrook Research Institute, Toronto, Ontario, Canada (Juurlink); and the Li Ka Shing Knowledge Institute, St. Michael’s Hospital, Toronto, Ontario, Canada (Gomes). Author contributions: PL, HM, AMO, BS, DNJ, and TG conceptualized the study. HM acquired funding. PL undertook project administration. TG supervised the conduct of investigation. All authors contributed to the investigation and methodology. CC conducted the analysis, and PL, CC, AMO, and TG validated the results. PL and CC completed the visualization and drafted the article. All authors contributed to revising and approving the final article. PL takes responsibility for the paper as a whole. All authors attest to meeting the four ICMJE.org authorship criteria: (1) Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the
8 Annals of Emergency Medicine
work; AND (2) Drafting the work or revising it critically for important intellectual content; AND (3) Final approval of the version to be published; AND (4) Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Funding and support: By Annals policy, all authors are required to disclose any and all commercial, financial, and other relationships in any way related to the subject of this article as per ICMJE conflict of interest guidelines (see www.icmje.org). This project was supported by funds from Public Health Ontario (PHO), which is funded by the Ontario Ministry of Health and Long-Term Care (MOHLTC). Dr. Orkin receives salary support from the Schwartz/ Reisman Emergency Medicine Institute, the University of Toronto Department of Family and Community Medicine, and the Canadian Institutes of Health Research Postdoctoral Fellowship program. Drs. Leece, Manson, and Schwartz and Ms. Chen report receiving employment income from PHO during the conduct of the study. Drs. Leece, Manson, and Schwartz report receiving grants from the Canadian Institutes of Health Research outside the submitted work. Dr. Gomes reports receiving grants from the Ontario MOHLTC outside the submitted work. Drs. Leece and Orkin report receiving nonfinancial support from Adapt Pharma for a study on related subject matter, outside the submitted work. Publication dates: Received for publication December 21, 2018. Revisions received May 30, 2019, and July 10, 2019. Accepted for publication July 15, 2019. Presented at the Canadian Centre on Substance Use and Addiction Issues of Substance meeting, November 2017, Calgary, Alberta, Canada; the Ontario Public Health Convention, March 2018, Toronto, Ontario, Canada; and the American Society of Addiction Medicine, April 2018, San Diego, CA. The sponsor had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the article. The opinions, results, and conclusions reported in this article are those of the authors and are independent from the funding source. No endorsement by PHO or the Ontario MOHLTC is intended or should be inferred.
REFERENCES 1. Degenhardt L, Charlson F, Mathers B, et al. The global epidemiology and burden of opioid dependence: results from the Global Burden of Disease 2010 study. Addiction. 2014;109:1320-1333. 2. Gomes T, Mamdani MM, Dhalla IA, et al. The burden of premature opioid-related mortality. Addiction. 2014;109:1482-1488. 3. Ahmad FB, Escobedo LA, Rossen LM, et al; National Center for Health Statistics. Provisional drug overdose death counts. Available at: https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm. Accessed May 28, 2019. 4. Special Advisory Committee on the Epidemic of Opioid Overdoses. National report: apparent opioid-related deaths in Canada (January 2016 to September 2018). Available at: https://health-infobase. canada.ca/datalab/national-surveillance-opioid-mortality.html#AORD. Updated April 10, 2019. Accessed May 28, 2019. 5. Weiss AJ, Elixhauser A, Barrett ML, et al. Opioid-related inpatient stays and emergency department visits by state, 2009-2014. HCUP statistical brief 219. Available at: https://www.hcup-us.ahrq.gov/
Volume
-,
no.
-
:
-
2019
Leece et al
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
One-Year Mortality After Nonfatal Opioid Poisoning
reports/statbriefs/sb219-Opioid-Hospital-Stays-ED-Visits-by-State.jsp. Accessed May 28, 2019. Canadian Institute for Health Information (CIHI). Opioid-related harms in Canada. Available at: https://secure.cihi.ca/free_products/opioidharms-chart-book-en.pdf. Published September 14, 2017. Accessed January 2, 2019. Leece P; Ontario Agency for Health Protection and Promotion (Public Health Ontario). Evidence brief: mortality following non-fatal opioid overdose. Available at: https://www.publichealthontario.ca/en/ eRepository/EB_Mortality_following_non-fatal_opioid_overdose.pdf. Published September 15, 2017. Accessed December 3, 2018. Lund C, Bjornaas MA, Sandvik L, et al. Five-year mortality after acute poisoning treated in ambulances, an emergency outpatient clinic and hospitals in Oslo. Scand J Trauma Resusc Emerg Med. 2013;21:65. Wichmann S, Nielsen S, Siersma VD, et al. Risk factors for 48-hours mortality after prehospital treatment of opioid overdose. Emerg Med J. 2013;30:223-225. Boyd JJ, Kuisma MJ, Alaspää AO, et al. Recurrent opioid toxicity after pre-hospital care of presumed heroin overdose patients. Acta Anaesthesiol Scand. 2006;50:1266-1270. Rudolph SS, Jehu G, Nielsen SL, et al. Prehospital treatment of opioid overdose in Copenhagen—is it safe to discharge on-scene? Resuscitation. 2011;82:1414-1418. Wampler DA, Molina DK, McManus J, et al. No deaths associated with patient refusal of transport after naloxone-reversed opioid overdose. Prehosp Emerg Care. 2011;15:320-324. Stoové MA, Dietze PM, Jolley D. Overdose deaths following previous non-fatal heroin overdose: record linkage of ambulance attendance and death registry data. Drug Alcohol Rev. 2009;28:347-352. Nielsen K, Nielsen S, Siersma V, et al. Treatment of opioid overdose in a physician-based prehospital EMS: frequency and long-term prognosis. Resuscitation. 2011;82:1410-1413. O’Brien BP, Murphy D, Conrick-Martin I, et al. The functional outcome and recovery of patients admitted to an intensive care unit following
Volume
-,
no.
-
:
-
2019
16.
17.
18.
19.
20.
21.
22.
23.
24.
drug overdose: a follow-up study. Anaesth Intensive Care. 2009;37:802-806. Boscarino JA, Kirchner HL, Pitcavage JM, et al. Factors associated with opioid overdose: a 10-year retrospective study of patients in a large integrated health care system. Subst Abuse Rehabil. 2016;7:131-141. Maeng DD, Han JJ, Fitzpatrick MH, et al. Patterns of health care utilization and cost before and after opioid overdose: findings from 10year longitudinal health plan claims data. Subst Abuse Rehabil. 2017;8:57-67. Health Quality Ontario. Effect of early follow-up after hospital discharge on outcomes in patients with heart failure or chronic obstructive pulmonary disease: a systematic review. Ont Health Technol Assess Ser. 2017;17:1-37. Benchimol EI, Smeeth L, Guttmann A, et al. The Reporting of Studies Conducted Using Observational Routinely-Collected Health Data (RECORD) statement. PLoS Med. 2015;12:e1001885. Dhalla IA, Mamdani MM, Sivilotti MLA, et al. Prescribing of opioid analgesics and related mortality before and after the introduction of long-acting oxycodone. CMAJ. 2009;181:891-896. Gomes T, Greaves S, Martins D, et al; Ontario Drug Policy Research Network. Latest trends in opioid-related deaths in Ontario: 1991 to 2015. Available at: https://odprn.ca/wp-content/uploads/2017/04/ ODPRN-Report_Latest-trends-in-opioid-related-deaths.pdf. Published April 20, 2017. Accessed May 28, 2019. Ontario Agency for Health Protection and Promotion (Public Health Ontario). Interactive opioid tool: opioid-related morbidity and mortality in Ontario. Available at: https://www.publichealthontario.ca/en/dataand-analysis/commonly-used-products/interactive-opioid-tool#/drug. Updated March 20, 2019. Accessed May 28, 2019. Weiner SG, Baker O, Bernson D, et al. One-year mortality of patients after emergency department treatment for nonfatal opioid overdose. Ann Emerg Med. 2019; In press. Centers for Disease Control and Prevention. Drug overdose deaths. Available at: https://www.cdc.gov/drugoverdose/data/statedeaths. html. Updated December 19, 2018. Accessed May 28, 2019.
Annals of Emergency Medicine 9
One-Year Mortality After Emergency Department Visit for Nonfatal Opioid Poisoning: A PopulationBased Analysis Pamela Leece, MD, MSc*; Cynthia Chen, MSc; Heather Manson, MD, MHSc; Aaron M. Orkin, MD, MSc; Brian Schwartz, MD, MScCH; David N. Juurlink, MD, PhD; Tara Gomes, PhD *Corresponding Author. E-mail: [email protected], Twitter: @PublicHealthON.
Study objective: We aim to characterize the incidence and risk factors for opioid-related and all-cause mortality in the year after an emergency department (ED) visit for nonfatal opioid poisoning by conducting a population-based study. Methods: We used linked health care databases in Ontario, Canada, to identify individuals who attended an ED for nonfatal opioid poisoning between January 1, 2015, and December 31, 2016. Using Cox proportional hazards regression, we examined predictors of mortality in the year after discharge (ED or hospital, if admitted). Results: In this cohort (n¼6,140), 327 individuals (5.3%) died of any cause and 118 (1.9%) died of opioid-related causes within 1 year. Adjusting for other covariates, we found that health service use in the first week was not protective for opioid-related death (hazard ratio [HR] 0.70; 95% confidence interval [CI] 0.47 to 1.06) or all-cause mortality (HR 0.98; 95% CI 0.78 to 1.24). In exploring other covariates, predictors of opioid-related mortality included male sex (HR 1.98; 95% CI 1.32 to 2.97) and using opioid agonist therapy (HR 1.79; 95% CI 1.15 to 2.80) or benzodiazepine (HR 1.54; 95% CI 1.02 to 2.31) in the 12 months before the index event. Assessment by a family physician in the previous 12 months was associated with a lower risk of opioid-related and all-cause mortality (HR 0.58, 95% CI 0.39 to 0.86; and HR 0.63, 95% CI 0.49 to 0.82, respectively). Conclusion: We identified predictors of opioid-related and all-cause mortality after ED presentation for opioid poisoning. Several predictors of mortality may facilitate targeted interventions. [Ann Emerg Med. 2019;-:1-9.] Please see page XX for the Editor’s Capsule Summary of this article. 0196-0644/$-see front matter Crown Copyright © 2019 Published by Elsevier, Inc on behalf of the American College of Emergency Physicians. https://doi.org/10.1016/j.annemergmed.2019.07.021
INTRODUCTION Background The global burden of disease attributable to opioid dependence has increased since 1990, with an estimated 9.2 million disability-adjusted years of life lost in 2010,1 exceeding the years of life lost attributable to pneumonia, HIV/AIDS, influenza, and alcohol use disorder in some regions.2 Since 2010, the burden of opioid-related disease has escalated significantly, particularly in North America, owing to the emergence of fentanyl and its analogues in the illicit drug supply.3,4 Opioid-related emergency department (ED) visits nearly doubled in the United States between 2005 and 2014, to a rate of 177.7 per 100,000.5 Similarly, ED visits in Canada for opioid poisoning increased substantially in Canada, reaching rates of 34.7 and 88.2 per 100,000 in the 2016 fiscal year in the provinces of Ontario and Alberta, respectively.6 Volume
-,
no.
-
:
-
2019
Importance Patient encounters with health care providers, such as ED visits for opioid poisoning, present opportunities to prevent subsequent death among people experiencing opioid-related harms. Patterns of health care use before and after an ED visit for nonfatal poisoning can also be assessed to identify system-level opportunities for improvement. There is little information on the rates and risk factors for death after treatment for nonfatal opioid poisoning.7-15 Specific to the ED setting, 2 studies conducted in a large health care plan in the United States (records for up to 1.2 million unique patients)16,17 estimated between 7% and 9.4% all-cause mortality within 1 year after treatment in the ED for opioid poisoning. However, we are not aware of studies that assess specifically for opioid-related mortality greater than 48 hours after nonfatal opioid poisoning. A small number of previous studies have examined risk factors for all-cause mortality among patients treated for opioid Annals of Emergency Medicine 1
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
Editor’s Capsule Summary
What is already known on this topic The emergency department (ED) has an important role in detecting and starting care for opioid use disorder patients. What question this study addressed How often do patients die after being treated in an ED for a nonfatal opioid use event? What this study adds to our knowledge Of a total of 6,140 Ontario patients treated in 2015 to 2016, 5.3% died within a year and 1.9% of deaths were directly related to opioid use. Men, those using an opioid agonist, and those also receiving a benzodiazepine had a higher frequency of death. How this is relevant to clinical practice This reinforces the opportunities to identify and try to help a group at a higher risk of death after ED discharge.
poisoning.8,9,13,14,16 One previous study in the ED setting found higher risk associated with higher comorbidity and medication use (benzodiazepine and opioid prescriptions) and lower risk with mental health disorders, having a primary care provider within the health system, commercial or private insurance, or a prescription for a muscle relaxant.16 The limited available literature reveals important gaps in regard to outcomes after ED treatment for opioid poisoning. Goals of This Investigation We sought to characterize the incidence of and risk factors for opioid-related and all-cause mortality in the year after an ED visit for nonfatal opioid poisoning. We hypothesized that better follow-up care may be associated with lower mortality, according to analogy with other health conditions.18 MATERIALS AND METHODS Study Design and Setting We conducted a retrospective population-based cohort study of Ontario residents presenting to an ED for nonfatal opioid poisoning between January 1, 2015, and December 31, 2016 (accrual period). We used a followup period of 12 months to identify subsequent all-cause and opioid-related deaths, and a Cox proportional 2 Annals of Emergency Medicine
hazards model to examine predictors of mortality within 1 year. Ontarians have universal access to single-payer publicly funded hospital and physician services (including ED care); however, prescription drug coverage varies, whereby some residents have public or private drug coverage and others pay for medications out of pocket. The study was approved by the Research Ethics Boards of Public Health Ontario, Toronto, Ontario, and Sunnybrook Health Sciences Centre, Toronto, Ontario. The reporting of this study follows the Reporting of Studies Conducted Using Observational RoutinelyCollected Health Data statement (Appendix E1, available online at http://www.annemergmed.com).19 Selection of Participants In the analysis, we included individuals aged 10 to 95 years who were alive at discharge from either the ED or related hospital admission after an ED visit for opioid poisoning during the accrual period (International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Canada [ICD-10-CA] codes T40.0 to 40.6 and T40.6). We defined the index date as the hospital discharge date of the episode of care resulting from the ED visit for opioid poisoning. If there were multiple ED visits for opioid poisoning, we selected the first such poisoning for inclusion in our study. We excluded individuals who received palliative care in the past 12 months and those who had any previous cancer diagnosis because they have a unique set of clinical circumstances compared with the group that was the target of our study. We defined an opioid-related death as any death recorded in the Drug and Drug-Alcohol Related Death database of confirmed opioid-related deaths. Methods of Measurement We used comprehensive provincial health care administrative data sets to conduct this analysis, which are linked at the individual level in an anonymous fashion by unique identifiers and analyzed at the Institute for Clinical Evaluative Sciences (https://www. ices.on.ca/; also see their Data Dictionary). Each health care record has a unique identifier assigned, an encrypted version of each patient’s unique 10-digit health card number. The databases are linked at the Institute for Clinical Evaluative Sciences with the identifier, and the databases used for this study have greater than 95% of records with a valid identifier. We identified ED visits for opioid poisoning by using the Canadian Institute for Health Information (CIHI) Volume
-,
no.
-
:
-
2019
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
National Ambulatory Care Reporting System, which contains comprehensive records of ED visits in Ontario. To describe the demographic characteristics of the cohort, we used the Registered Persons Database. To identify comorbidities and health services use, we used the diagnostic code (ICD-10) associated with health service visits in the past 12 months, recorded in the CIHI Discharge Abstract Database, CIHI National Ambulatory Care Reporting System, Ontario Health Insurance Plan Database, and the CIHI Ontario Mental Health Reporting System databases. We also described patient comorbidities and traumatic injury with the Ontario Asthma data set and Ontario Chronic Obstructive Pulmonary Disease data set of the Ontario Trauma Registry. We excluded individuals with a cancer diagnosis by using the Ontario Cancer Registry. The Aggregated Diagnosis Group system was used to score comorbidity in the preceding year (version 10.0; The Johns Hopkins ACG System, Baltimore, MD). We compared health services use in the 12 months before and 12 months after the index date, and used the Ontario Narcotics Monitoring System to characterize the dispensing of opioids, opioid agonist therapy (eg, methadone, buprenorphine), and benzodiazepines (referred to here as “monitored medications”). We compared monitored medication use in the previous 12 months with that in the 12 months after the index date. We defined patient sociodemographic characteristics (age, sex, urban/rural location of residence, and income quintile), elements of the index event (admission reason, triage level, and discharge disposition), comorbidities (weighted Aggregated Diagnosis Group quintile and specific diagnoses), health services use, and monitored medication use. The main exposure for our primary analysis was any physician or hospital health care visit that occurred within 7 days of the index date. A complete description of these variables is provided in Table E1 (available online at http://www.annemergmed.com).
Outcome Measures Confirmed opioid-related deaths were identified with the Drug and Drug-Alcohol Related Death database, which contains records abstracted from the Office of the Chief Coroner for Ontario. The methods for this abstraction have been described previously, and the database includes deaths for which the pathologist or coroner conducting the death investigation determined that opioids directly contributed to death, according to postmortem toxicology and other findings.20 We identified deaths from any cause by using the Registered Persons Database. Volume
-,
no.
-
:
-
2019
Primary Data Analysis The variables in our analysis were chosen to investigate our hypothesis in regard to health care access in the first 7 days, as well as important characteristics related to risk of subsequent mortality. We included characteristics of the individual (age and sex), comorbidities (Aggregated Diagnosis Group and mental health and substance use), recent prescription medication use (opioids, opioid agonist therapy, and benzodiazepines), and health service use (ED, hospital, and family physician visits). We used frequencies, percentages, medians, and interquartile ranges to describe cohort demographic characteristics, comorbidities, health service use, and prescribed medications. We described missing values with a separate category. We used Cox proportional hazards regression to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between opioid-related or all-cause death and receipt of health care services in the 7 days after the index date. The length of follow-up in our study was either 365 days or the number of days between death date and index date, whichever was shorter. In exploring other covariates in the model, we examined the association for demographic characteristics, comorbidities, health service use, and use of monitored medications. We adjusted all models for potential risk factors, including baseline characteristics (age and sex), health service use in the 12 months before the index event (1 or more visits to an ED or a hospital or mental health admission, or family physician visits), use of monitored medications (yes/no) in the 12 months before the index event (opioids, opioid agonist therapy, and benzodiazepines), comorbidities (Aggregated Diagnosis Group) in the previous 12 months, or mental health disorder (psychotic or nonpsychotic), substance use disorder, or previous opioid poisoning. All analyses used a 2-sided type I error rate of 0.05 as the threshold for statistical significance and were performed with SAS (version 9.3; SAS Institute, Inc., Cary, NC). Our analytic plan is accessible in Appendix E2 (available online at http:// www.annemergmed.com). We assessed 3 variables as interaction terms in our final model: Aggregated Diagnosis Group and mental health or substance use diagnosis, Aggregated Diagnosis Group and health service use in the past 12 months, and mental health or substance use diagnosis and opioid agonist therapy. To assess multicollinearity, we examined the variance inflation factor for all variables in our model. Furthermore, we used the Akaike information criterion statistic to examine the model fit. We calculated standardized mortality ratios for all-cause and opioid-related mortality, using the Ontario 2016 Annals of Emergency Medicine 3
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
population as a reference. We used the Drug and DrugAlcohol Related Death database to calculate the expected opioid-related death rate in the population within 3 age categories selected a priori (10 to 24, 25 to 54, and 55 to 95 years) and by sex. These categories are informed by known epidemiology in Ontario for age of opioid-related death,21,22 with most deaths occurring in individuals aged 25 and 54 years. In provincial data, few if any opioidrelated deaths occur in individuals younger than 10 years or older than 95 years. We used the Registered Persons Database to calculate the expected mortality rate from all causes in the population, using the same categories. The standardized mortality ratios were calculated with the ratio of observed deaths in our cohort to expected deaths in the population. We also calculated 95% CIs for each standardized mortality ratio, using the following formula: standardized mortality ratio1.96standard error standardized mortality ratio. RESULTS Characteristics of Study Subjects During the 2-year study period, we identified 6,904 patients who attended an ED in Ontario for opioid poisoning. After exclusions, 6,140 individuals met our study criteria (Figure). The characteristics of the cohort are shown in Table 1, and Table 2 describes health service use and use of monitored medications at baseline. The median age of patients was 38 years (interquartile range 27 to 53 years) and 52.8% were men. The most common diagnostic codes in the past 12 months were mental health disorders (58.2%), substance use diagnoses (40.5%), and low back pain (16.2%). In the past 12 months, 1.5% had been treated for nonfatal opioid poisoning, although 74.3% had attended an ED overall. Opioid, opioid agonist therapy, and benzodiazepine prescribing in the past 12 months occurred among 50.8%, 23.3%, and 41.1% of patients, respectively. ED visits, hospital admissions, and opioid use decreased when 12 months before and after the index event were compared. Table E2 (available online at http://www. annemergmed.com) describes the characteristics of our cohort after inclusion of 83 individuals who died on or before the index date (eg, did not survive to discharge; n¼6,223). There were no meaningful differences in characteristics between this group and the group included in our main analysis, other than death after ED arrival. Main Results Among the index ED visits for opioid toxicity, 70.0% were classified as unintentional opioid poisonings. The 4 Annals of Emergency Medicine
triage severity score was high for most patients: 18.7% were classified as resuscitation and 62.4% were classified as emergency. More than a third of patients were hospitalized (39.5%), whereas 54.8% were discharged from the ED to home or their place of residence, and 5.8% left the ED before being discharged by a physician. In comparison of health service use before and after the index event, there were meaningful decreases in hospital admissions (42.2% versus 26.2%; standardized difference –0.34) (Table 2). In the 12-month follow-up, 118 subjects (1.9%) died of opioid-related causes, whereas 327 (5.3%) died from any cause (see Figure E1, available online at http://www. annemergmed.com, for histogram of all-cause mortality; a histogram is not available for opioid-related mortality because of small cells). Adjusting for other variables, contact with the health care system within 7 days after the index date was not associated with opioid-related mortality (HR 0.70; 95% CI 0.47 to 1.06) (Table 3; see Table E3, available online at http://www.annemergmed.com, for univariate analysis) or all-cause mortality (HR 0.98; 95% CI 0.78 to 1.24) (Table 4; see Table E4, available online at http://www.annemergmed.com, for univariate analysis). The standardized mortality ratio for all-cause mortality was
6904
6787
6548
6223
6140
Figure. Inclusion and exclusion criteria applied to populationbased analysis. Index date refers to the discharge date of the health care episode involved in the ED visit for opioid poisoning. Volume
-,
no.
-
:
-
2019
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
Table 1. Characteristics of individuals with nonfatal opioid poisoning at the index ED visit. Cohort (N[6,140)
Characteristic Median age (IQR), y
38 (27–53)
Age, No. (%), y
Table 1. Continued. Cohort (N[6,140)
Characteristic Mental health or substance use diagnosis (including poisoning with any intent)
4,507 (73.4)
10–24
1,122 (18.3)
Alcohol-related
595 (9.7)
25–54
3,700 (60.3)
Substance use
2,484 (40.5)
1,318 (21.5)
Mental health
3,571 (58.2)
3,239 (52.8)
Opioid poisoning
55–95 Men, No. (%) Rurality, No. (%) Rural
726 (11.8)
Urban
5,407 (88.1)
Missing
7 (0.1)
Income quintile, No. (%)*
Accidental
64 (1.0)
Intentional
28 (0.5)
Pain Osteoarthritis
613 (10.0)
Rheumatoid arthritis
103 (1.7)
1,942 (31.6)
Fibromyalgia
2
1,364 (22.2)
Low back pain
3
1,056 (17.2)
Posttraumatic pain
1 (lowest)
4
911 (14.8)
5 (highest)
806 (13.1)
Missing Accidental
4,297 (70.0)
Intentional
1,843 (30.0)
Triage level, No. (%) Resuscitation
1,147 (18.7)
Emergency
3,833 (62.4)
Urgent
1,015 (16.5)
Less urgent/semiurgent Nonurgent/unknown
131 (2.1) 14 (0.2)
Disposition, No. (%) Discharged from ED to home or place of residence Registered but left ED
3,363 (54.8) 355 (5.8)
Hospital admission
1,140 (18.6)
Admitted to special care unit
919 (15.0)
Mental health admission
363 (5.9)
Weighted ADG quintile, No. (%)* 1 (lowest)
1,228 (20.0)
2
1,237 (20.2)
3
1,339 (21.8)
4
1,122 (18.3)
5 (highest)
1,214 (19.8)
Diagnoses, No. (%)* Chronic lung disease
127 (2.1)
Chronic liver disease
477 (7.8)
Chronic kidney disease
221 (3.6)
Heart disease (coronary artery disease or arrhythmia)
839 (13.7)
-,
39 (0.6) 992 (16.2) 35 (0.6)
IQR, Interquartile range; ADG, Aggregated Diagnosis Group. *In the 12 months before the index date.
61 (1.0)
Reason
Volume
92 (1.5)
no.
-
:
-
2019
10.0 (95% CI 9.0 to 11.1), and the standardized mortality ratio for opioid-related mortality was 238.7 (95% CI 195.6 to 281.7). All interactions were nonsignificant, so we did not include interaction terms in the final models for allcause or opioid-related mortality. Examination of other model covariates revealed that predictors of opioid-related mortality were male sex (HR 1.98; 95% CI 1.32 to 2.97) and using opioid agonist therapy (HR 1.79; 95% CI 1.15 to 2.80) or a benzodiazepine (HR 1.54; 95% CI 1.02 to 2.31) in the 12 months preceding the index event (Table 3; see Table E3, available online at http://www.annemergmed.com, for univariate analysis). All-cause mortality was associated with age groups 25 to 54 and 55 to 95 years (reference age 10 to 24 years) (HR 1.81, 95% CI 1.13 to 2.91; and HR 4.13, 95% CI 2.51 to 6.79, respectively), male sex (HR 1.62; 95% CI 1.28 to 2.03), highest comorbidity (Aggregated Diagnosis Group quintile 4 or 5) (HR 1.92; 95% CI 1.50 to 2.46), benzodiazepine use (HR 1.44; 95% CI 1.13 to 1.84), and hospital or ED admission (HR 1.69; 95% CI 1.10 to 2.61) (Table 4; see Table E4, available online at http://www.annemergmed.com, for univariate analysis). For all-cause mortality, an addiction or mental health diagnosis was associated with a lower risk (HR 0.69; 95% CI 0.53 to 0.89). Furthermore, 1 or more visits to a family physician in the previous 12 months was a protective factor for opioid-related and all-cause mortality (HR 0.58, 95% CI 0.39 to 0.86; and HR 0.63, 95% CI 0.49 to 0.82,
Annals of Emergency Medicine 5
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning Table 2. Health service use and controlled medication use among individuals with nonfatal opioid poisoning. Health Service and Medication Use Health service use (‡1 visit), No. (%)
Past Subsequent Standardized 12 Months* 12 Months* Difference N¼6,140
N¼6,140 4,432 (72.2)
–0.36
ED
4,563 (74.3)
4,069 (66.3)
–0.18
Hospital admission
2,592 (42.2)
1,609 (26.2)
–0.34
836 (13.6)
772 (12.6)
–0.03
4,620 (75.24) 4,436 (72.2)
–0.07
Family physician
HR,* Adjusted Model (95% CI)
Characteristic (Cohort N[6,140) Hypothesis
ED or hospital admission 5,309 (86.5)
Mental health admission
Table 3. Characteristics associated with opioid-related mortality in the year after nonfatal opioid poisoning.
0.70 (0.47–1.06)
Any health service use in the week after index date Exploratory analyses Age, y
Monitored medication use, No. (%) Opioid
3,116 (50.8)
2,625 (42.8)
–0.16
OAT
1,431 (23.3)
1,657 (27.0)
0.08
Benzodiazepine
2,523 (41.1)
2,413 (39.3)
–0.04
OAT, Opioid agonist therapy. *Relative to the index date.
respectively). In accordance with this finding, we estimated that 20 opioid-related deaths (118 observed, 98 expected) and 12 all-cause deaths (327 observed, 315 expected) could have been prevented if all members of the cohort had been assessed by a family physician in the previous 12 months. Tables E5 and E6 (available online at http://www. annemergmed.com) present results stratified by whether individuals were discharged directly from the ED or from related hospital admission. Results were generally consistent, except that use of opioid agonist therapy was associated with risk of opioid-related death among patients admitted to the hospital but not for those discharged from the ED. LIMITATIONS Our study has several limitations. First, because we were using administrative data, there were no clinical details available that may have allowed us to be more specific in our description of patient characteristics or health care at the nonfatal opioid poisoning. Second, we ended our follow-up in 2017, and the current risk factors among patients presenting to the ED for nonfatal opioid poisoning may have shifted as the epidemic evolved. We have framed our hypothesis as care within 1 week, although immediate access is preferable for patients with opioid use disorder who need treatments such as opioid agonist therapy. There is risk of bias in our study through censoring; for example, if individuals died after discharge and before a health care visit. In the first week, there were 18 deaths, and 13 of 6 Annals of Emergency Medicine
10–24
1.0 [Reference]
25–54
1.43 (0.80–2.56)
55–95
1.21 (0.59–2.51)
Men
1.98 (1.32–2.97)
Weighted ADG quintile† 1–3 (lowest)
1.0 [Reference]
4–5 (highest)
1.43 (0.96–2.13) 0.83 (0.51–1.36)
Mental health or substance use diagnosis (including poisoning with any intent)† Monitored medication use†,‡ Opioid
0.96 (0.64–1.46)
OAT
1.79 (1.15–2.80)
Benzodiazepine
1.54 (1.02–2.31)
Health service use† ED or hospital admission
1.34 (0.75–2.41)
Family physician
0.58 (0.39–0.86)
*Adjusted for potential risk factors in the past 12 months, including baseline characteristics (age and sex), comorbidities (weighted ADG and mental health or substance-use-related diagnoses, including opioid poisoning), monitored medication use (in the previous 12 months) (opioid, methadone or buprenorphine, and benzodiazepine), and health service use (1 or more ED, hospital, or mental health admissions, or family physician visits). † In the 12 months before the index date. ‡ Monitored medications refer to opioids, opioid agonist therapies, and benzodiazepines.
these individuals did not have a health care visit before their death. This could affect the test of our hypothesis because death occurred before the opportunity to visit a health care professional. Additionally, we were unable to detect nonfatal opioid poisoning events treated in the community (eg, emergency medical services, community-based naloxone use without transfer to the hospital). Although some ethnic and cultural groups may be at higher risk for opioid poisoning, this aspect was out of scope for our initial exploration of these data. Databases included in our study capture the majority of physician services in Volume
-,
no.
-
:
-
2019
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
Table 4. Characteristics associated with all-cause mortality in the year after nonfatal opioid poisoning. HR* Adjusted Model (95% CI)
Characteristic (Cohort N[6,140) Hypothesis Any health service use in the week after index date
0.98 (0.78–1.24)
Exploratory analyses Age, y 10–24
1.0 [Reference]
25–54
1.81 (1.13–2.91)
55–95
4.13 (2.51–6.79)
Men
1.62 (1.29–2.03)
Weighted ADG quintile† 1–3 (lowest)
1.0 [Reference]
4–5 (highest)
1.92 (1.50–2.46)
Mental health or addictions diagnosis (including poisoning with any intent)†
0.69 (0.53–0.89)
Monitored medication use†,‡ Opioid
1.20 (0.92–1.57)
OAT
1.09 (0.80–1.49)
Benzodiazepine
1.44 (1.13–1.84)
Health service use† ED or hospital admission
1.69 (1.10–2.61)
Family physician
0.63 (0.49–0.82)
*Adjusted for potential risk factors in the past 12 months, including baseline characteristics (age and sex), comorbidities (weighted ADG and mental health or substance-use-related diagnoses, including opioid poisoning), monitored medication use (in the previous 12 months) (opioid, methadone or buprenorphine, and benzodiazepine), and health service use (1 or more ED, hospital, or mental health admissions, or family physician visits). † In the 12 months before the index date. ‡ Monitored medications refer to opioids, opioid agonist therapies, and benzodiazepines.
Ontario, although there are a few hundred family physicians who work in community health centers in Ontario, where the funding model does not require submission of individual service claims. We were unable to adjust our analyses for movement in and out of the province, although this likely represents a very small proportion. Patients treated outside of the ED may have important differences compared with those treated in the ED. The generalizability of our results to the Ontario population is limited to people who visit the ED in for opioid toxicity and survive to discharge, but our findings are not generalizable to those who may experience poisoning and never come to the hospital. Our study Volume
-,
no.
-
:
-
2019
results may not be generalizable outside of Ontario, where the dynamics of the opioid epidemic and health care systems differ.
DISCUSSION In this population-based study, we found that among people discharged from the hospital after opioid poisoning, 1 in 52 died of opioid-related causes and 1 in 19 died from any cause within 1 year. The opioid-related and all-cause mortality were higher among this group compared with the general population, particularly for opioid-related mortality. We also found that male sex and use of opioid agonist therapy or benzodiazepines in the previous 12 months were associated with an increased risk of opioidrelated death among patients in this cohort. Furthermore, risk factors for all-cause mortality included age, male sex, comorbidity, benzodiazepine use, and hospital or ED admissions in the previous 12 months. Visits to a family physician in the year before the ED visit for nonfatal poisoning were protective for opioid-related and all-cause mortality. Our study advances the literature by, to our knowledge, providing the largest available population-based analysis of individuals treated in the ED for opioid poisoning, and characterizing for the first time the risk factors for opioid-related deaths. The incidence of all-cause mortality at 1 year in our study (5.3%) was lower than in 2 previous studies conducted within a US health care plan,16,17 and similar to that in a recent statewide study in Massachusetts.23 The rate of opioid-related mortality is higher in the United States than in Canada, which may contribute to differences compared with our study findings.4,24 However, the variation between studies in the United States may be due to different methods in assessing the outcome. The study in Massachusetts could have produced a lower estimate because of challenges with data linkage.23 As in the previous US health care plan study, we found an increased risk for all-cause mortality among patients with benzodiazepine use.16 We found discordant results, with an association between mental health and substance-use-related diagnoses and lower all-cause mortality, but higher opioid-related mortality for individuals prescribed opioid agonist therapy. In previous literature, substance use disorders were a significant risk factor for all-cause mortality after opioid poisoning.16 Our results may reflect the decision to combine outcomes for mental health and substance use diagnoses, and the use of opioid agonist therapy in the 12 months before the index event may be an indicator for opioid use disorder as a risk factor. Our finding in regard to mental Annals of Emergency Medicine 7
Leece et al
One-Year Mortality After Nonfatal Opioid Poisoning
health and substance use is similar to that in a US study that found a protective association with mental health disorders and all-cause mortality after opioid poisoning.16 The finding that opioid agonist therapy in the previous 12 months was associated with opioidrelated mortality for patients discharged from the hospital but not for those discharged from the ED could indicate that these patients had additional complexities or were at risk because of potential interruptions in treatment. Likewise, our finding that family physician visits were protective could reflect that patients with additional disease complexity may not consult their physician as often. In summary, we found several predictors of mortality that may facilitate targeted interventions to reduce risk of subsequent death among people who attended an ED for nonfatal opioid poisoning. Particularly, our findings suggest that deaths may be averted by improving access to primary care for individuals at risk of ED visits because of opioid poisoning. Further research is needed to assess the effectiveness of interventions to prevent mortality after nonfatal opioid poisoning. Supervising editor: Donald M. Yealy, MD. Specific detailed information about possible conflict of interest for individual editors is available at https://www.annemergmed.com/editors. Author affiliations: From Health Promotion, Chronic Disease, and Injury Prevention, Public Health Ontario, Toronto, Ontario, Canada (Leece, Chen, Manson, Schwartz); the Department of Family and Community Medicine (Leece, Orkin, Schwartz), Dalla Lana School of Public Health (Leece, Manson, Orkin, Schwartz), Institute of Health Policy, Management, and Evaluation (Juurlink, Gomes), and Leslie Dan Faculty of Pharmacy (Gomes), University of Toronto, Ontario, Canada; The Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada (Chen, Juurlink, Gomes); the School of Public Health and Health Systems, University of Waterloo, Waterloo, Ontario, Canada (Manson); Schwartz/Reisman Emergency Medicine Institute, Sinai Health System, Toronto, Ontario, Canada (Orkin); the Department of Medicine, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada (Juurlink); The Sunnybrook Research Institute, Toronto, Ontario, Canada (Juurlink); and the Li Ka Shing Knowledge Institute, St. Michael’s Hospital, Toronto, Ontario, Canada (Gomes). Author contributions: PL, HM, AMO, BS, DNJ, and TG conceptualized the study. HM acquired funding. PL undertook project administration. TG supervised the conduct of investigation. All authors contributed to the investigation and methodology. CC conducted the analysis, and PL, CC, AMO, and TG validated the results. PL and CC completed the visualization and drafted the article. All authors contributed to revising and approving the final article. PL takes responsibility for the paper as a whole. All authors attest to meeting the four ICMJE.org authorship criteria: (1) Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the
8 Annals of Emergency Medicine
work; AND (2) Drafting the work or revising it critically for important intellectual content; AND (3) Final approval of the version to be published; AND (4) Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Funding and support: By Annals policy, all authors are required to disclose any and all commercial, financial, and other relationships in any way related to the subject of this article as per ICMJE conflict of interest guidelines (see www.icmje.org). This project was supported by funds from Public Health Ontario (PHO), which is funded by the Ontario Ministry of Health and Long-Term Care (MOHLTC). Dr. Orkin receives salary support from the Schwartz/ Reisman Emergency Medicine Institute, the University of Toronto Department of Family and Community Medicine, and the Canadian Institutes of Health Research Postdoctoral Fellowship program. Drs. Leece, Manson, and Schwartz and Ms. Chen report receiving employment income from PHO during the conduct of the study. Drs. Leece, Manson, and Schwartz report receiving grants from the Canadian Institutes of Health Research outside the submitted work. Dr. Gomes reports receiving grants from the Ontario MOHLTC outside the submitted work. Drs. Leece and Orkin report receiving nonfinancial support from Adapt Pharma for a study on related subject matter, outside the submitted work. Publication dates: Received for publication December 21, 2018. Revisions received May 30, 2019, and July 10, 2019. Accepted for publication July 15, 2019. Presented at the Canadian Centre on Substance Use and Addiction Issues of Substance meeting, November 2017, Calgary, Alberta, Canada; the Ontario Public Health Convention, March 2018, Toronto, Ontario, Canada; and the American Society of Addiction Medicine, April 2018, San Diego, CA. The sponsor had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the article. The opinions, results, and conclusions reported in this article are those of the authors and are independent from the funding source. No endorsement by PHO or the Ontario MOHLTC is intended or should be inferred.
REFERENCES 1. Degenhardt L, Charlson F, Mathers B, et al. The global epidemiology and burden of opioid dependence: results from the Global Burden of Disease 2010 study. Addiction. 2014;109:1320-1333. 2. Gomes T, Mamdani MM, Dhalla IA, et al. The burden of premature opioid-related mortality. Addiction. 2014;109:1482-1488. 3. Ahmad FB, Escobedo LA, Rossen LM, et al; National Center for Health Statistics. Provisional drug overdose death counts. Available at: https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm. Accessed May 28, 2019. 4. Special Advisory Committee on the Epidemic of Opioid Overdoses. National report: apparent opioid-related deaths in Canada (January 2016 to September 2018). Available at: https://health-infobase. canada.ca/datalab/national-surveillance-opioid-mortality.html#AORD. Updated April 10, 2019. Accessed May 28, 2019. 5. Weiss AJ, Elixhauser A, Barrett ML, et al. Opioid-related inpatient stays and emergency department visits by state, 2009-2014. HCUP statistical brief 219. Available at: https://www.hcup-us.ahrq.gov/
Volume
-,
no.
-
:
-
2019
Leece et al
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
One-Year Mortality After Nonfatal Opioid Poisoning
reports/statbriefs/sb219-Opioid-Hospital-Stays-ED-Visits-by-State.jsp. Accessed May 28, 2019. Canadian Institute for Health Information (CIHI). Opioid-related harms in Canada. Available at: https://secure.cihi.ca/free_products/opioidharms-chart-book-en.pdf. Published September 14, 2017. Accessed January 2, 2019. Leece P; Ontario Agency for Health Protection and Promotion (Public Health Ontario). Evidence brief: mortality following non-fatal opioid overdose. Available at: https://www.publichealthontario.ca/en/ eRepository/EB_Mortality_following_non-fatal_opioid_overdose.pdf. Published September 15, 2017. Accessed December 3, 2018. Lund C, Bjornaas MA, Sandvik L, et al. Five-year mortality after acute poisoning treated in ambulances, an emergency outpatient clinic and hospitals in Oslo. Scand J Trauma Resusc Emerg Med. 2013;21:65. Wichmann S, Nielsen S, Siersma VD, et al. Risk factors for 48-hours mortality after prehospital treatment of opioid overdose. Emerg Med J. 2013;30:223-225. Boyd JJ, Kuisma MJ, Alaspää AO, et al. Recurrent opioid toxicity after pre-hospital care of presumed heroin overdose patients. Acta Anaesthesiol Scand. 2006;50:1266-1270. Rudolph SS, Jehu G, Nielsen SL, et al. Prehospital treatment of opioid overdose in Copenhagen—is it safe to discharge on-scene? Resuscitation. 2011;82:1414-1418. Wampler DA, Molina DK, McManus J, et al. No deaths associated with patient refusal of transport after naloxone-reversed opioid overdose. Prehosp Emerg Care. 2011;15:320-324. Stoové MA, Dietze PM, Jolley D. Overdose deaths following previous non-fatal heroin overdose: record linkage of ambulance attendance and death registry data. Drug Alcohol Rev. 2009;28:347-352. Nielsen K, Nielsen S, Siersma V, et al. Treatment of opioid overdose in a physician-based prehospital EMS: frequency and long-term prognosis. Resuscitation. 2011;82:1410-1413. O’Brien BP, Murphy D, Conrick-Martin I, et al. The functional outcome and recovery of patients admitted to an intensive care unit following
Volume
-,
no.
-
:
-
2019
16.
17.
18.
19.
20.
21.
22.
23.
24.
drug overdose: a follow-up study. Anaesth Intensive Care. 2009;37:802-806. Boscarino JA, Kirchner HL, Pitcavage JM, et al. Factors associated with opioid overdose: a 10-year retrospective study of patients in a large integrated health care system. Subst Abuse Rehabil. 2016;7:131-141. Maeng DD, Han JJ, Fitzpatrick MH, et al. Patterns of health care utilization and cost before and after opioid overdose: findings from 10year longitudinal health plan claims data. Subst Abuse Rehabil. 2017;8:57-67. Health Quality Ontario. Effect of early follow-up after hospital discharge on outcomes in patients with heart failure or chronic obstructive pulmonary disease: a systematic review. Ont Health Technol Assess Ser. 2017;17:1-37. Benchimol EI, Smeeth L, Guttmann A, et al. The Reporting of Studies Conducted Using Observational Routinely-Collected Health Data (RECORD) statement. PLoS Med. 2015;12:e1001885. Dhalla IA, Mamdani MM, Sivilotti MLA, et al. Prescribing of opioid analgesics and related mortality before and after the introduction of long-acting oxycodone. CMAJ. 2009;181:891-896. Gomes T, Greaves S, Martins D, et al; Ontario Drug Policy Research Network. Latest trends in opioid-related deaths in Ontario: 1991 to 2015. Available at: https://odprn.ca/wp-content/uploads/2017/04/ ODPRN-Report_Latest-trends-in-opioid-related-deaths.pdf. Published April 20, 2017. Accessed May 28, 2019. Ontario Agency for Health Protection and Promotion (Public Health Ontario). Interactive opioid tool: opioid-related morbidity and mortality in Ontario. Available at: https://www.publichealthontario.ca/en/dataand-analysis/commonly-used-products/interactive-opioid-tool#/drug. Updated March 20, 2019. Accessed May 28, 2019. Weiner SG, Baker O, Bernson D, et al. One-year mortality of patients after emergency department treatment for nonfatal opioid overdose. Ann Emerg Med. 2019; In press. Centers for Disease Control and Prevention. Drug overdose deaths. Available at: https://www.cdc.gov/drugoverdose/data/statedeaths. html. Updated December 19, 2018. Accessed May 28, 2019.
Annals of Emergency Medicine 9