Secondary Logo

Share this article on:

National Trends and Factors Associated With Inpatient Mortality in Adult Patients With Opioid Overdose

Burton, Brittany N., MHS, MAS*; Lin, Timothy C., MAS*; Said, Engy T., MD; Gabriel, Rodney A., MD, MAS†,‡

doi: 10.1213/ANE.0000000000003755
Chronic Pain Medicine: Original Clinical Research Report
Continuing Medical Education

BACKGROUND: The prevalence of opioid misuse and opioid-related mortality has increased dramatically over the past decade. There is limited evidence on factors associated with mortality from opioid overdose in the inpatient setting. The primary objective was to report national trends in opioid overdose and mortality. The secondary objectives were to explore factors associated with inpatient mortality and report differences in prescription opioid overdose (POD) versus illicit opioid overdose (IOD) cohorts.

METHODS: Using the 2010–2014 Nationwide Inpatient Sample, we performed a cross-sectional analysis and identified a weighted estimate of 570,987 adult patients with an International Classification of Disease, Ninth Revision, or External Cause of Injury code of POD or IOD. We performed multivariable logistic regression to identify predictors of inpatient mortality. The odds ratio (OR) and their associated 95% confidence interval (CI) are reported.

RESULTS: Of the 570,987 patients with opioid overdose, 13.8% had an admissions diagnosis of IOD, and the remaining had POD. Among all opioid overdose admissions, the adjusted odds of IOD admissions increased by 31% per year (OR, 1.31; 95% CI, 1.29–1.31; P < .001); however, the adjusted odds POD admissions decreased by 24% per year (OR, 0.76; 95% CI, 0.75–0.77; P < .001). The mortality was 4.7% and 2.3% among IOD and POD admissions, respectively. The odds of inpatient mortality increased by 8% per year among IOD admissions (OR, 1.08; 95% CI, 1.02–1.14; P < .007). The odds of inpatient mortality increased by 6% per year among all POD admissions (OR, 1.06; 95% CI, 1.03–1.09; P < .001). Those with IOD compared to POD had higher odds of mortality (OR, 2.03; 95% CI, 1.79–2.29; P < .001). Patients with age ≥80 years of age and those with a diagnosis of a solid tumor malignancy had higher odds of mortality. Odds of inpatient mortality were decreased in African American versus Caucasian patients and in patients undergoing alcohol rehabilitation therapy.

CONCLUSIONS: The increase in mortality provides a strong basis for further risk reduction strategies and intervention program implementation. Medical management of not only the opioid overdose but also the comorbidities calls for a multidisciplinary approach that involves policy makers and health care teams.

From the *School of Medicine

Division of Regional Anesthesia and Acute Pain, Department of Anesthesiology

Division of Biomedical Informatics, Department of Medicine, University of California, San Diego, La Jolla, California.

Accepted for publication July 26, 2018.

Published ahead of print 26 July 2018.

Funding: The project described was partially supported by the National Institute of Health (NIH), Grant TL1TR00098 of Clinical and Translational Science Award (CTSA) funding before August 13, 2015 and Grant TL1TR001443 of CTSA funding beginning August 13, 2015 and beyond. The content is solely the responsibility of the authors and does not necessarily represent the official views of NIH.

The authors declare no conflicts of interest.

Reprints will not be available from the authors.

Address correspondence to Rodney A. Gabriel, MD, MAS, Department of Anesthesiology, University of California, San Diego, 9500 Gilman Dr, MC 0881, La Jolla, CA 92093. Address e-mail to ragabriel@ucsd.edu.

KEY POINTS

  • Question: What are the factors associated with inpatient mortality in patients admitted to the hospital for opioid overdose?
  • Findings: Illicit opioid overdose, older age, and solid tumor malignancy were associated with increased odds of inpatient mortality, whereas African American race and drug/alcohol rehabilitation were associated with decreased odds of inpatient mortality.
  • Meaning: Targeted program implementation and risk reduction strategies are urgently needed.

Death due to opioid overdose has increased dramatically over the past decade. The Centers for Disease Control estimates that 115 Americans die daily of an opioid overdose.1 Recent data from the Centers for Disease Control suggest that among those 18–44 years of age, unintentional injuries are the leading cause of death, specifically due to unintentional poisoning.2 The surge in opioid overdose–related deaths began in 1999, and by 2012, opioid overdose–related deaths more than tripled from prescription medications.3 In the United States, prescription opioid abuse is associated with a $9.5 billion public health burden.4 During 2015, 63.1% of drug overdose deaths were due to opioids, and more recently, research suggests that the opioid epidemic is driven by widespread illicit (ie, heroin and synthetic opioids) opioid use.1 Furthermore, there was a 62.5% increase in heroin use in from 2002 to 2013.5

Despite the rapid increase in opioid-related deaths, there are limited studies outlining the factors associated with mortality of patients with opioid overdose in the hospital setting. While many studies report findings of a national population–based sample, these studies only evaluate trends in either opioid overdose–related hospitalization rates or mortality. For example, in their evaluation of disparities in opioid overdose, Unick and Ciccarone6 found that there were racial/ethnic and geographical differences in rates of prescription and heroin-related hospitalizations.6 Similarly, Hsu et al7 also used a national database to evaluate rates of hospital admission and inpatient mortality in patients with opioid overdose. The primary objective was to report national trends in opioid overdose and mortality using a large national database. The secondary objectives were to report factors associated with inpatient mortality and explore unadjusted differences in medical history in patients admitted with prescription opioid overdose (POD) or illicit opioid overdose (IOD).

Back to Top | Article Outline

METHODS

Data Collection

Data were obtained from the publicly available dataset, the National (Nationwide) Inpatient Sample (NIS) database of the Healthcare Cost and Utilization Project (HCUP). HCUP was developed through a multiorganization partnership sponsored by the Agency for Healthcare Research and Quality. NIS is the largest multi-state all-payer inpatient health care database in the United States and approximates a 20% stratified sample of medical records from the US hospitals. NIS includes deidentified data of several variables (ie, demographic characteristics, clinical variables, payment source, and hospital factors). In 2012, NIS was redesigned to improve national estimates.8 The new design of 2012 affected the numbers and types of discharges for prior years.8 To address this, HCUP recommends users use “trend” discharge weights for NIS files before 2012 to minimize the effects of the redesign on estimated trends that cross 2012.9 For years before 2012, we used the trend weight data element “TRENDWT” in place of the original discharge weight data element “DISCWT” to create national estimates for trends analysis that are consistent across 2012.9 The NIS redesign deidentified hospitals and instead provided an NIS unique hospital identification number. As such NIS still retains hospital-level data elements for all hospitals. We used our analysis that was not affected as we used the “HOSP_NIS” and “HOSPID” data elements to identify hospitals. Finally, HCUP recommends users to use HOSP_NIS or HOSPID in mixed-effect models account for within-hospital correlation in regression models. NIS meets the criteria of the Health Insurance Portability and Accountability Act to protect personal information and therefore was exempt from the consent requirement by the University of California, San Diego Institutional Review Board. This retrospective analysis adheres to the STROBE checklist for cross-sectional studies.

Back to Top | Article Outline

Patient Population

Table 1

Table 1

Hospitalizations of patients ≥18 years of age from 2010 to 2014 with a diagnosis of POD or IOD were included. The primary objective was to explore national trends in opioid overdose (POD and IOD) from 2010 to 2014. The secondary objectives were to explore factors associated with inpatient mortality, defined as death during the inpatient stay, and to present an exploratory analysis of unadjusted rates among opioid overdose cohorts. We used the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) and External Cause of Injury codes to identify POD and IOD, as well as patient comorbidities and inpatient interventions. Here, POD was defined with the following ICD-9-CM codes: 9650.00 (poisoning with opium), 965.02 (poisoning by methadone), 965.09 (poisoning by other opiates and related narcotics), E850.1 (accidental poisoning by methadone), and E850.2 (accidental poisoning by other opiates and related narcotics). IOD was defined as 965.01 (poisoning by heroin) and E850.0 (accidental poisoning by heroin). We extracted medical history from patients with any diagnosis of IOD or POD; such extraction has been shown to be the most sensitive definition of opioid overdose.10 Sociodemographic and hospital variables supplied in the NIS include race, sex, age, median household income, insurance status, hospital ownership, hospital location and teaching status, hospital region, and weekend hospital admission. Table 1 lists the ICD-9-CM codes used to identify medical history for each case.

Back to Top | Article Outline

Statistical Analysis

R, a software environment for statistical computing (R version 3.3.2; R Foundation for Statistical Computing, Vienna, Austria), was used to perform all statistical analysis. We weighted all patient records to produce national trends; we used data elements “TRENDT” before 2012 and “DISCWT” for 2012 and onward.9 , 11 TRENDT and DISCWT were used to weight each patient record, and these weights are stored in each patient record. When the discharge weights are applied to the unweighted NIS data, the result is an estimate of the number of admissions/discharges for all inpatient discharges from community hospitals in the United States.9 , 11 The stratification and weighting was used to provide national estimates for Tables 2 and 3. We specifically used the “survey” package in R version 3.3.2.

For bivariate analyses, the P value for the comparisons between 2 cohorts (POD versus IOD) was derived from the Pearson χ2 and Wilcoxon rank sum test for categorical (with continuity correction) and non-normally distributed continuous variables, respectively. Time trend analysis was performed using logistic regression for binary outcomes with year of hospital admission as the independent continuous variable. We also adjusted our time trend analysis for race, gender, age, median household income, and insurance status. The binary outcomes in the time trend analysis included the following: (1) IOD admission among all patients in NIS; (2) POD admission among all patients in NIS; (3) mortality among all patients admitted for IOD; and (4) mortality among all patients admitted for POD. To assess the association of potential risk factors with inpatient mortality, we first performed a mixed-effects univariable logistic regression followed by a mixed-effects multivariable logistic regression. All cases that had missing values for any of the variables were removed from the final analysis. Mixed-effect logistic regression analysis was done on complete cases. The random effect was “hospital identification number” (a unique value assigned in the NIS database for a specific institution). The mixed-effect logistic regression analysis was not weighted. Using this data element allowed us to account for clustered observations within hospitals.

Table 2

Table 2

Table 3

Table 3

In the initial mixed-effect multivariable logistic regression model, we included all covariates with P < .2 from the univariable analysis. We included the following 32 covariates in the univariate analysis: sex, race, age, median household income, source of payment, hospital location and teaching status, weekend hospital admission, type of opioid overdose (POD versus IOD), opioid abuse history (yes/no), respiratory complications (mechanical ventilation or tracheostomy), cardiopulmonary resuscitation, cardiac evaluation, hemodialysis, packed red blood cell transfusion, central venous catheter placement, any naloxone use, alcohol rehabilitation and therapy, drug rehabilitation and therapy, alcohol abuse, pneumonitis, metabolic acidosis, metabolic alkalosis, rhabdomyolysis, acute kidney injury, sepsis, pneumonia, hypotension, solid tumor malignancy, chronic pain, and cocaine, amphetamine, and hallucinogen poisoning. Backward elimination was then performed by stepwise removal of covariates with the largest P value until all covariates in the model were P < .05. The odds ratios (ORs), 95% confidence intervals (CIs), and Wald test P value were reported for each independent variable. We used a Bonferroni-corrected P value for 21 covariates (final model). Two-sided P < .002 was considered statistically significant. We assessed multicollinearity with variation inflation factor statistic, in which a value <5 was deemed adequate with no collinearity. We evaluated model discrimination with area under the receiver operating characteristic curve.

Based on the work by Peduzzi et al,12 we estimated the minimum sample size needed for our study. Per Peduzzi et al,11 p is the smallest of the proportions of negative or positive cases in the population and k is the number of covariates (the number of independent variables), then the minimum number of cases to include is N = 10 k/p. Given that the proportion of patients with inpatient mortality was 0.026 and we have 32 covariates in the final model, the minimum number of cases required is 12,308.

Back to Top | Article Outline

RESULTS

There were 584,373 patients with a diagnosis of either POD or IOD from 2010 to 2014. After excluding patients with missing data (2.29%), our final sample included 570,987 patients, of which 13.8% and 86.2% had a diagnosis of IOD and POD, respectively. Table 2 outlines the distribution of patient and hospital characteristics. Patients with IOD were younger than patients with POD (31 vs 52 years old; P < .001). The proportion of patients with POD versus IOD was higher in Caucasian (81.5% vs 74.1%) and Native American (0.9% vs 0.4%) patients (P < .001). Female (56.6% vs 29.1%) patients were more likely to have POD, while male (70.9% vs 43.4%) patients tended to have IOD (P < .001). Patients with IOD were more likely to live in zip codes with the lowest quartile of median household income (P < .001). Medicare (41.1%) was more common among those with POD; however, Medicaid (33.9%) was more common among those with IOD (P < .001). Uninsured patients had higher rates of IOD versus POD (P < .001). Rural (13.9%) and urban (41.3%) nonteaching hospitals had higher unadjusted rates of POD inpatient diagnoses (P < .001).

Back to Top | Article Outline

National Trends

Figure 1

Figure 1

Figure 2

Figure 2

In our unadjusted logistic regression model, IOD admissions increased by 24% per year (OR, 1.24; 95% CI, 1.22–1.25; P < .001); however, POD admissions decreased by 19% per year (OR, 0.81; 95% CI, 0.80–0.82; P < .001; Figure 1A). In our sociodemographic-adjusted logistic regression model, IOD admissions increased by 31% per year (OR, 1.31; 95% CI, 1.29–1.31; P < .001); however, POD admissions decreased by 24% per year (OR, 0.76; 95% CI, 0.75–0.77; P < .001). The overall mortality rate was 2.6%. Furthermore, we estimated a 4.7% and 2.3% mortality rate among IOD and POD admissions, respectively (Table 3). In the unadjusted logistic regression analysis, the odds of inpatient mortality increased by 8% per year among IOD admissions (OR, 1.08; 95% CI, 1.02–1.14; P < .007) and by 6% per year among POD admissions (OR, 1.06; 95% CI, 1.03–1.09; P < .001; Figure 1B). In the sociodemographic-adjusted logistic regression analysis, the odds of inpatient mortality increased by 9% per year among IOD admissions (OR, 1.09; 95% CI, 1.03–1.15; P = .003) and by 6% per year among POD admissions (OR, 1.06; 95% CI, 1.03–1.09; P < .001). Figure 2 shows the prevalence of IOD and POD admissions among all inpatient admissions in the NIS database during 2010–2014. IOD is more prevalent in the Northeast (69 per 100,000 hospital admissions) and Midwest (60 per 100,000 hospital admissions) geographical regions, whereas POD is leading along the Western region (311 per 100,000 hospital admissions).

Back to Top | Article Outline

Comorbidities, Interventions, and Outcomes Characteristics

Table 3 lists the distribution of comorbidities, inpatient interventions, and outcomes of patients admitted with opioid overdose. Active smoking (37%), hypertension (31.4%), depression disorder (25.9%), and benzodiazepine poisoning (25.8%) were the most common comorbidities among the overall study population. The median number (interquartile range) of diagnoses per inpatient was 12 (9–16). Compared to POD, patients with IOD tended to have a diagnosis of cocaine abuse and poisoning, central nervous system stimulant poisoning, amphetamine poisoning, alcohol abuse, opioid abuse, cannabis abuse, active smoker, hallucinogen poisoning, suicidal ideation, pneumonitis, metabolic acidosis, cardiac dysrhythmias, rhabdomyolysis, leukocytosis, or septicemia (all P < .05). The most common inpatient interventions included mechanical ventilation (21.1%) followed by noninvasive mechanical ventilation (4.5%) and cardiac evaluation (3.4% [eg, electrocardiogram, diagnostic ultrasound, and cardiac catheterization]) (P < .001). The median hospital length of stay was longer for POD versus IOD (3 vs 2 days; P < .001). Hospital charge was higher for POD (P < .001). Unadjusted mortality was significantly more common among patients with IOD (P < .001).

Back to Top | Article Outline

Factors Associated With Mortality

Figure 3

Figure 3

Figure 3 lists the results of inpatient mortality of the mixed-effects multivariable logistic regression analysis. We included an unweighted number of 116,407 patients in the regression analysis. Those with IOD compared to POD had higher odds of mortality (OR, 1.73; 95% CI, 1.56–1.94; P < .001). The odds of inpatient mortality were decreased in African American versus Caucasian patients (OR, 0.67; 95% CI, 0.57–0.79; P < .001), in patients undergoing alcohol rehabilitation therapy (OR, 0.19; 95% CI, 0.09–0.41; P < .001), and in patients with chronic pain (OR, 0.55; 95% CI, 0.49–0.61; P < .001). The odds of inpatient mortality were increased in patients 80+ years of age versus 18–49 years old (OR, 2.36; 95% CI, 1.91–2.91; P < .001) and in patients with solid tumor malignancy (OR, 2.44; 95% CI, 2.13–2.80; P < .001). For inpatient mortality, model discrimination demonstrated an area under the receiver operating characteristic curve (95% CI) of 0.9141 (0.9093–0.9189).

Back to Top | Article Outline

DISCUSSION

In this retrospective analysis of NIS data, we examined trends and the impact of perioperative factors on inpatient mortality (ie, primary outcome) and explored unadjusted differences among opioid overdose cohorts (ie, secondary outcomes). We estimated a mortality rate of 2.6% in patients admitted for opioid overdose from 2010 to 2014. Small cohort studies have published findings related to opioid overdose; however, most are descriptive and only evaluate trends in outcomes.13–16 To our knowledge, there are limited studies using a large national database to identify factors associated with inpatient mortality. Given the public health and economical burden of opioid misuse, recognizing the differences in vulnerable populations and risk factors of opioid overdose has important implications for targeted and effective intervention and prevention.

The sales of prescription opioids quadrupled from 1999 to 2014, and the National Institute of Drug Abuse estimated a 167% increase in opioid-related death from 2002 to 2015.17 Of patients prescribed opioids, 8%–9% develop an opioid misuse disorder and 4%–6% progressed to heroin abuse.17 From this, further understanding demographic factors and comorbidities associated with fatal opioid overdose serves to direct primary, secondary, and tertiary prevention strategies and proactive intervention. Knowledge of these factors associated with mortality will help physicians to not only target the vulnerable patients but also educate and counsel them on substance abuse in the setting of their comorbidities. Furthermore, identifying vulnerable populations will allow health care providers and public health officials to develop screening tools aimed at detecting opioid misuse, which may help to reduce the incidence of opioid misuse. Appropriately allocating resources to manage underlying risk factors may help to decrease the morbidity and mortality, and such allocation will serve to reduce the negative impact of the opioid epidemic.

We first sought to describe the demographics of patients admitted for either IOD or POD. IOD was more commonly found in younger, male, Caucasian patients, and those with Medicaid, whereas POD patients were more likely to be older, Caucasian, female, and have Medicare. In a previous 12-year study, POD patients were more likely to be female, Caucasian, and have Medicare when compared to heroin opioid overdose patients.7 We found that patients admitted for overdose had a median number of 12 diagnoses with smoking, hypertension, depression, and benzodiazepine poisoning being most common. These demographics and comorbidities may play a role in risk stratification; thus, early recognition is important for preventative strategies or interventions during hospitalization.

Here, we showed that IOD admissions increased by 23% per year, whereas POD admissions decreased by 19% per year. Unick and Ciccarone6 similarly demonstrated that during 2012–2014, the rate of POD hospitalization declined, whereas the rate of heroin overdose hospitalizations steadily increased. We also found that IOD admissions are more prevalent in the Northeast and Midwest regions of the United States, while POD admissions are most common in the West followed by the South. These geographical patterns have been similarly described.6 The drop in the opioid prescribing rate due to stringent prescription drug monitoring programs coupled with an increasing demand of more potent opioids may have contributed to the increase in IOD admissions.18–22

While we only observed an increase in IOD admissions from 2010 to 2014, inpatient mortality increased among both cohorts. IOD admissions had a 2 times increased odds for inpatient death compared to POD. This may be explained partly by either an increasing prevalence of illicit opioid use or widespread use of more potent formulations of illicit opioids. The development and widespread misuse of illicit fentanyl made it 30–50 times more potent than heroin and led to its significant contribution in opioid overdose–related deaths.23 The National Forensic Laboratory Information System suggests that direct use or heroin “laced” with nonpharmaceutical fentanyl contributes to the rising prevalence of IOD.24

Identifying factors associated with mortality is an important next step for risk-stratifying opioid users. We found that several baseline characteristics were associated with inpatient mortality, including age, ethnicity (African Americans were protective of mortality), and presence of solid tumor malignancy. Recent studies show a surge in opioid-related death rates among African Americans.16 , 25 Further work is needed to evaluate the impact of the opioid-related morbidity in racial/ethnic minorities. The explanation of why solid tumor malignancy is associated with increased death in this patient population likely related to the underlying malignancy, comorbidity burden, or previous opioid abuse/misuse history. Other medical factors such as sepsis, hypotension, and inpatient interventions were also associated with mortality. These comorbidities either represent the severity of opioid overdose (ie, metabolic acidosis or acute kidney injury) or a direct result of the disease process itself (ie, sepsis).

Once patients are admitted for an opioid overdose, strategic interventions should be in place at respective institutions to prevent further morbidity and mortality. Active enrollment in drug and alcohol rehabilitation and therapy conferred a significant protective effect against inpatient mortality. This supports a previous finding that opioid replacement therapy was associated with a reduction in heroin overdose deaths.26 While treatment may not fully prevent overdose events, it plays a significant role in decreasing mortality. These findings underscore the importance of assessing a patient’s readiness to engage in rehabilitation and arranging the necessary support that patients may require on discharge. In a previous study, inpatient consultation by an addiction consult team reduced addiction severity as well as increased number of days of self-reported abstinence after discharge in adults at high risk for alcohol or drug use disorder.27 Likewise, a patient engaged in rehabilitation or therapy may have greater social support or access to other resources, which may positively impact prognosis as well.28 , 29

There are important limitations in our analysis. By analyzing only hospitalizations, we exclude cases of opioid overdose that result in death before admission or cases of overdose that were managed in the emergency department without need for admission; this introduces significant bias in our analysis. Because the NIS is an administrative inpatient database, we are unable to determine whether IOD occurred after prescription opioid misuse. Moreover, other important clinical data are not available, such as the severity and degree of chronic pain, cause of death, dose, type, and route of administration of opioid, length of opioid misuse/abuse, time to event, or events that transpire after discharge. NIS does not capture cause-specific mortality; therefore, it is unclear if mortality is secondary to opioid overdose or other inpatient comorbidities. NIS does not allow us to determine the time of diagnosis of comorbidities. There is debate on the optimal model building strategy. Here, we have used backward model selection with P values. This method, however, has several limitations which include the following: (1) multiple testing problem (ie, the probability of observing significance by chance alone) that leads to an elevated type 1 error rate, (2) CIs falsely narrow, and (3) using P values to exclude potential confounding covariates.30 , 31 Strengths of our study include a large sample size, allowing us to examine a relatively rare outcome (ie, mortality), using data over a span of years, and the use of a nationally representative sample.

Back to Top | Article Outline

CONCLUSIONS

In this retrospective study, we reported national trends, evaluated factors associated with inpatient mortality, and explored various morbidity points in patients with opioid overdose. Despite recent efforts to curtail the opioid epidemic and its devastating impact, our results show that the epidemic is shifting from one driven by prescription use to one driven by illicit use. Medical management of not only the opioid overdose but also the comorbidity burden calls for a multidisciplinary approach that involves policy makers and health care teams.

Back to Top | Article Outline

DISCLOSURES

Name: Brittany N. Burton, MHS, MAS.

Contribution: This author helped search the literature, analyze the data, and prepare and review the manuscript.

Name: Timothy C. Lin, MAS.

Contribution: This author helped search the literature, analyze the data, and prepare and review the manuscript.

Name: Engy T. Said, MD.

Contribution: This author helped search the literature, analyze the data, and prepare and review the manuscript.

Name: Rodney A. Gabriel, MD, MAS.

Contribution: This author helped search the literature, collect the data, study the design, analyze the data, and prepare and review the manuscript.

This manuscript was handled by: Honorio T. Benzon, MD.

Back to Top | Article Outline

REFERENCES

1. Rudd RA, Seth P, David F, Scholl L. Increases in drug and opioid-involved overdose deaths: United States, 2010-2015. MMWR Morb Mortal Wkly Rep. 2016;65:1445–1452.
2. Ruhm CJ. Corrected US opioid-involved drug poisoning deaths and mortality rates, 1999–2015. Addiction. 2018;113:1339–1344.
3. Frenk SM, Porter KS, Paulozzi LJ. Prescription opioid analgesic use among adults: United States, 1999–2012. CDC NCHS Data Brief. 2015:1–8.
4. Birnbaum HG, White AG, Reynolds JL, et al. Estimated costs of prescription opioid analgesic abuse in the United States in 2001: a societal perspective. Clin J Pain. 2006;22:667–676.
5. Jones CM, Logan J, Gladden RM, Bohm MK. Vital signs: demographic and substance use trends among heroin users: United States, 2002-2013. MMWR Morb Mortal Wkly Rep. 2015;64:719–725.
6. Unick GJ, Ciccarone D. US regional and demographic differences in prescription opioid and heroin-related overdose hospitalizations. Int J Drug Policy. 2017;46:112–119.
7. Hsu DJ, McCarthy EP, Stevens JP, Mukamal KJ. Hospitalizations, costs and outcomes associated with heroin and prescription opioid overdoses in the United States 2001-12. Addiction. 2017;112:1558–1564.
8. Healthcare Cost and Utilization Project National Inpatient Sample (NIS) Database. Introduction to the HCUP National Inpatient Sample (NIS). February 2018. Available at: https://www.hcup-us.ahrq.gov/db/nation/nis/NIS_Introduction_2015.pdf. Accessed June 1, 2018.
9. Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS) Database. Trend weights for HCUP NIS Data. February 2018. Available at: https://www.hcup-us.ahrq.gov/db/nation/nis/trendwghts.jsp. Accessed June 1, 2018.
10. Slavova S, Bunn TL, Talbert J. Drug overdose surveillance using hospital discharge data. Public Health Rep. 2014;129:437–445.
11. Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS) Database. HCUP methods series. Using the HCUP National Inpatient Sample to Estimate Trends. February 2018. Available at: https://www.hcup-us.ahrq.gov/reports/methods/2006_05_NISTrendsReport_1988-2004.pdf. Accessed June 1, 2018.
12. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol. 1996;49:1373–1379.
13. 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.
14. Feng J, Iser JP, Yang W. Medical encounters for opioid-related intoxications in Southern Nevada: sociodemographic and clinical correlates. BMC Health Serv Res. 2016;16:438.
15. Pfister GJ, Burkes RM, Guinn B, et al. Opioid overdose leading to intensive care unit admission: epidemiology and outcomes. J Crit Care. 2016;35:29–32.
16. Visconti AJ, Santos GM, Lemos NP, Burke C, Coffin PO. Opioid overdose deaths in the city and county of San Francisco: prevalence, distribution, and disparities. J Urban Health. 2015;92:758–772.
17. National Institute of Health NIoDA. Available at: https://www.drugabuse.gov/drugs-abuse/opioids/opioid-overdose-crisis. Accessed June 15, 2018.
18. Levy B, Paulozzi L, Mack KA, Jones CM. Trends in opioid analgesic-prescribing rates by specialty, US, 2007-2012. Am J Prev Med. 2015;49:409–413.
19. Pezalla EJ, Rosen D, Erensen JG, Haddox JD, Mayne TJ. Secular trends in opioid prescribing in the USA. J Pain Res. 2017;10:383–387.
20. Guy GP Jr, Zhang K, Bohm MK, et al. Vital signs: changes in opioid prescribing in the United States, 2006-2015. MMWR Morb Mortal Wkly Rep. 2017;66:697–704.
21. McClellan C, Lambdin BH, Ali MM, et al. Opioid-overdose laws association with opioid use and overdose mortality. Addict Behav. 2018;86:90–95.
22. Chang HY, Murimi I, Faul M, Rutkow L, Alexander GC. Impact of Florida’s prescription drug monitoring program and pill mill law on high-risk patients: a comparative interrupted time series analysis. Pharmacoepidemiol Drug Saf. 2018;27:422–429.
23. Stanley TH. The history and development of the fentanyl series. J Pain Symptom Manage. 1992;7:S3–S7.
24. O’Donnell JK, Gladden RM, Seth P. Trends in deaths involving heroin and synthetic opioids excluding methadone, and law enforcement drug product reports, by census region: United States, 2006-2015. MMWR Morb Mortal Wkly Rep. 2017;66:897–903.
25. Kandel DB, Hu MC, Griesler P, Wall M. Increases from 2002 to 2015 in prescription opioid overdose deaths in combination with other substances. Drug Alcohol Depend. 2017;178:501–511.
26. Schwartz RP, Gryczynski J, O’Grady KE, et al. Opioid agonist treatments and heroin overdose deaths in Baltimore, Maryland, 1995-2009. Am J Public Health. 2013;103:917–922.
27. Wakeman SE, Metlay JP, Chang Y, Herman GE, Rigotti NA. Inpatient addiction consultation for hospitalized patients increases post-discharge abstinence and reduces addiction severity. J Gen Intern Med. 2017;32:909–916.
28. Cavaiola AA, Fulmer BA, Stout D. The impact of social support and attachment style on quality of life and readiness to change in a sample of individuals receiving medication-assisted treatment for opioid dependence. Subst Abus. 2015;36:183–191.
29. Damian AJ, Mendelson T, Agus D. Predictors of buprenorphine treatment success of opioid dependence in two Baltimore City grassroots recovery programs. Addict Behav. 2017;73:129–132.
30. Mundry R, Nunn CL. Stepwise model fitting and statistical inference: turning noise into signal pollution. Am Nat. 2009;173:119–123.
31. Altman DG, Andersen PK. Bootstrap investigation of the stability of a Cox regression model. Stat Med. 1989;8:771–783.
© 2019 International Anesthesia Research Society