Gender and Prescription Opioid Misuse in the Emergency Department

Introduction

Over the past decade, prescription opioid misuse and abuse has demonstrated a marked increase. In 2004, it was estimated that the number of emergency department (ED) visits involving the nonmedical use of prescription opioids was 144,644; in 2008 this number increased to 305,885.¹ Nationally, there were an estimated 14,800 prescription opioid overdose deaths in 2008, representing an increase of greater than 370% from 1999.²

Various explanations have been proposed for the upswing, including changes in prescribing practices and lack of public awareness about the potential of opioids to cause addiction and death. The existing literature demonstrates great heterogeneity in all aspects of opioid use: motives for opioid use, medical needs of the user, the source of the drug, concurrent other drug use, and comorbidities.

Gender has emerged as a distinguishing factor in the epidemiology of prescription opioidabuse^3–7 For example, although men still make up the majority of non-medical users of prescription opioids, the rate of rise in fatal prescription opioids overdoses in women has been higher: deaths among women have increased by 400% since 1999, compared to 265% among men.⁸ This statistic – only partially understood – highlights the importance of further investigations into the phenomenon of prescription opioid use and how and why outcomes may be affected by gender.

Studies have demonstrated gender differences influencing initiation and ongoing nonmedical use of prescription opioids, including high-risk times of consumption, routes of administration, and particularly motives for using prescription opioids.⁹ Men have been described as using opioids more often for pleasurable aspects or to enhance amusement, similar to the positive expectancies noted in male alcohol misuse. In contrast, women may engage in nonmedical use of opioids more often to deal with negative emotions and address interpersonal problems.^3,4,6,10,11 Women are also more likely than men to use additional medications such as sedatives to enhance the therapeutic effect of prescription opioids¹²; co-ingestion is a known risk factor for mortality from prescription opioids. In addition, women are more likely than men to be prescribed prescription pain medications, are given higher doses, and use them for longer time periods than men.⁸

If gender is indeed a significant factor in opioid use and misuse, it may affect how and why women and men present to the ED, as well as the treatments and services they need once they engage with ED care providers. To date, gender patterns in ED presentation for nonmedical us of prescription opioids have not been described. Using the Substance Abuse and Mental Health Administration (SAMHSA)'s Drug Abuse Warning Network (DAWN), a nationally-representative sample of drug-related ED visits, we sought to examine gender differences in: prevalence of visits for nonmedical use of prescription opioids; drugs used in combination with opioids as a proxy for gender-based difference in use patterns; and outcomes of opioid-related ED visits, including disposition and mortality.

Methods

The Drug Abuse Warning Network, or DAWN, collects data from a nationally representative sample of hospitals throughout the United States, including Alaska and Hawaii.¹³ Non-Federal, short-stay, general surgical and medical hospitals with a 24-hour ED are eligible for inclusion. This analysis used the 2011 dataset, the last year DAWN was funded to collect ED visit data. For 2011, data were collected from 233 participating hospitals, and a total of 229,211 drug-related ED visits were identified; by applying post-stratified weights to the data received from the participating sampled hospitals, the submitted cases were extrapolated to an estimate of 5,067,374 drug-related ED visits out of an estimated 126 million total ED visits. Of these the drug-related visits, 2,462,948 were considered to involve drug misuse or abuse, with the balance involving adverse reactions and accidental ingestions. DAWN does not capture any other data on individual visits, including admission diagnoses, measures of illness severity or procedures needed, length of stay, or service of admission, except for psychiatric admissions.

For this analysis, we selected out visits involving adults (≥18 years of age) and nonmedical use of pharmaceuticals and excluded those related to adverse reactions, accidental ingestions, or only involving illicit and/or alcohol use. The 3,300 individual drugs captured by DAWN in 2011 were reviewed by the study team to identify those falling into categories of: 1) prescription opioids, 2) illicit drugs, 3) antidepressants or 4) anxiolytics. Both single and combination prescription opioids were included in the first category. All drug names were reviewed by two authors (EKC and CD), and any unrecognized drugs were confirmed against a toxicology database. Alcohol was a pre-defined variable in the DAWN dataset. We created individual variables to indicate if the prescription opioids involved in the visit were taken alone or in combination with other substances (primary outcomes). Other variables of interest extracted for this study included gender, age, race, and clinical disposition, including: hospital admission or transfer; intensive care unit (ICU) admission; referral to outpatient detoxification; admission for inpatient detoxification or psychiatric care; and death.

As this study used only existing, publically available, de-identified data, it was exempt from IRB review.

Data Analysis

We calculated proportions and 95% confidence intervals (CIs) for demographic and drug use variables and compared differences between women and men using univariate (chi square) analysis, defining as significant non-overlapping 95% CIs. We identified the top three most frequently used drugs for women and men in each drug / drug-combination category; however, in presenting this list, we did not include the non-specific category “Narcotic analgesics NOS,” which was at or near the top for all subgroups, and reported rankings of specifically named opiates only.

We then developed logistic regression models to examine the associations between gender and specific drug presentations and clinical outcomes, adjusting for age and race. These were selected a priori, rather than through sequential or stepwise processes, based on evidence in the literature. Model variables were examined for evidence of collinearity. Model fit was evaluated using Hosmer-Lemeshow goodness of fit testing for sample survey data. We also examined interactions between gender and race, however, these did not have significant effect in any model. Adjusted odds ratios for which the 95% CI did not cross the null vaue of one were considered statistically significant.

For all analyses, we used “svy” commands in Stata to account for weights and clustering and obtain accurate point estimates, standard errors, confidence intervals and tests of hypothesis.

Results

Out of an estimation sample of 1,096,741, DAWN captured 426,010 visits related to opioid misuse, indicating that 23.9% (95%CI 21.3-26.5%) of drug-involved ED visits were for nonmedical use of prescription drugs and 38.8% (95%CI 34.4-43.2%) involved opioids. There were no significant differences between women and men with opioid use in the proportion of patients represented across age or race categories (Table 1). Visits by women and by men were equally likely to involve illicit drug use, including subsets of cocaine and heroin use, and anxiolytics; however, women were more likely to present with antidepressant use, while men were more likely to present with an alcohol co-ingestion (Table 1).

Table 1. Characteristics of ED visits with prescription opioid use involvement, by gender (n=426,010).

	Female (51.1%) Proportion (95%CI)	Male (48.9%) Proportion (95%CI)
Age Category
18-29	25.4 (20.6-30.1)	30.4 (27.4-33.3)
30-44	32.8 (29.4-36.3)	32.0 (28.3-35.7)
45-54	22.8 (19.8-25.8)	22.2 (18.9-25.6)
55 or older	19.0 (15.9-22.1)	15.4 (11.9-18.9)
Race
White	82.2 (75.7-88.7)	82.8 (77.1-88.5)
Black/African-American	12.1 (6.9-17.3)	11.5 (6.9-16.1)
Other	5.6 (2.5-8.7)	5.7 (3.0-8.3)
Opiates Only	55.8 (51.0-60.6)	50.1(44.4-55.9)
Alcohol	9.6 (7.7-11.4)*	16.7 (13.8-19.6)*
Illicit Drugs	22.4 (17.6-27.2)	30.5 (25.1-35.9)
Cocaine	9.1 (6.5-11.8)	11.7 (8.5-15.0)
Heroin	2.3 (1.3-3.3)	2.0 (1.2-2.8)
Antidepressants	6.5 (5.0-8.0)*	2.3 (1.4-3.2)*
Anxiolytics	19.3 (16.0-22.6)	16.6 (14.5-18.7)
General Hospital Admission or Transfer	31.8 (27.8-35.8)	28.7 (24.6-32.9)
ICU Admission**	23.3 (14.7-31.9)	23.6 (15.3-32.0)
Death	0.06 (0.00-0.13)	0.2 (0.03-0.4)

^*Statistically significant difference between women and men

^**ICU visits are presented as proportion of inpatient visits

There were no significant overall differences between women and men in clinical outcomes examined, including proportion referred to outpatient detoxification or admitted for detoxification or psychiatric reasons, hospital admission or transfer, ICU admission, or death (Table 1). More than 30% of both women and men required hospital admission and of those patients and more than 20% of both women and men admitted to the hospital required an ICU setting. Death was a rare outcome for both genders.

While “narcotic analgesic NOS” was one of the most frequently coded prescription opioids, when the agent was identifiable, the most commonly listed individual opioids were acetaminophen/hydrocodone, known by trade name “Vicodin,” single entity oxycodone, and acetaminophen/oxycodone, known as “Percocet.” The three most commonly reported specific opioids, by drug/drug-combination category and gender, are shown in Table 2.

Table 2. Top three drugs involved in prescription opioid-implicated ED visits, in order of frequency, by gender^*.

Drug Combination	Female (51.1%)	Male (48.9%)
Overall	n= 10,370 Acetaminophen/hydrocodone Oxycodone single entity (SE) Acetaminophen/oxycodone	n= 11,321 Oxycodone SE Acetaminophen/hydrocodone Acetaminophen/oxycodone
Opioids Alone	n= 6,310 Acetaminophen/hydrocodone Oxycodone SE Acetaminophen/oxycodone	n=6,021 Oxycodone SE Acetaminophen/hydrocodone Acetaminophen/oxycodone
Opioids + Alcohol	n=1,220 Acetaminophen/hydrocodone Oxycodone SE Acetaminophen/oxycodone	n=2,184 Acetaminophen/hydrocodone Oxycodone SE Acetaminophen/oxycodone
Opioids + Illicit drugs	n=1,800 Oxycodone SE Acetaminophen/hydrocodone Acetaminophen/oxycodone	n=2,797 Oxycodone SE Acetaminophen/hydrocodone Acetaminophen/oxycodone
Opioids + Antidepressants	n=554 Acetaminophen/hydrocodone Oxycodone SE Acetaminophen/ oxycodone	n=325 Acetaminophen/hydrocodone Oxycodone SE Acetaminophen/oxycodone
Opioids + Anxiolytics	n=1,714 Acetaminophen/ hydrocodone Oxycodone SE Acetaminophen/ oxycodone	n=1,632 Acetaminophen/ hydrocodone Oxycodone SE Acetaminophen/ oxycodone
Hospital Admission	n=3,275 Acetaminophen/hydrocodone Oxycodone SE Aspirin/oxycodone	n=3,212 Oxycodone SE Acetaminophen/hydrocodone Aspirin/oxycodone
ICU Admission	n=640 Oxycodone SE Acetaminophen/hydrocodone Morphine	n=590 Oxycodone SE Aspirin/oxycodone Acetaminophen/hydrocodone
Death	n=15 Acetaminophen/hydrocodone Oxycodone SE, Hydromorphone (tied)	n=15 Acetaminophen/hydrocodone Oxycodone, SE, Aspirin/oxycodone (tied)

^*The category “Narcotic analgesics NOS” was excluded from this table.

In the gender-stratified multivariable analyses (Table 3), women who presented with prescription opioid misuse with either concurrent illicit drug use or antidepressant use were more likely to require general hospital admission. Among men, presentations for opioids with alcohol and with heroin increased the odds for general hospital admission. Opioids in combination with antidepressants were associated with ICU admission in both women and men, although for men, wide confidence intervals for this outcome reflect the small numbers of male patients with this combination. Opioids in combination with anxiolytics were also associated with ICU admission. No studied drug combination was associated with increased odds of death.

Table 3.

Logistic regression models, with adjusted odds ratios^a (aORs) for clinical outcomes with specific drug combinations, stratified by gender, among ED visit related to prescription opioids (n=426,010). Each row represents a separate model estimating the aOR for a given outcome with a specific drug combination.

	Female aOR (95%CI)	Male aOR (95%CI)
aOR for General Hospital Admission / Transfer
Opioids Alone	0.65 (0.54-0.77)*	0.62 (0.46-0.83)*
Opioids + Alcohol	1.49 (0.69-2.06)	1.86 (1.36-2.53)*
Opioids + Illicit	1.62 (1.19-2.21)*	0.98 (0.69-1.38)
Opioids + Cocaine	2.14 (1.22-3.75)*	1.09 (0.79-1.50)
Opioids + Heroin	1.32 (0.51-3.37)	1.89 (1.11-3.23)*
Opioids + Antidepressants	1.82 (1.09-3.03)*	1.54 (0.64-3.66)
Opioids + Anxiolytics	1.39 (0.97-2.00)	1.16 (0.79-1.70)
aOR for ICU Admission
Opioids Alone	0.99 (0.64-1.53)	0.75 (0.38-1.51)
Opioids + Alcohol	0.92 (0.49-1.75)	1.00 (0.52-1.58)
Opioids + Illicit	1.00 (0.57-1.75)	1.22 (0.72-2.06)
Opioids + Cocaine	0.52 (0.28-0.96)*	1.03 (0.50-2.13)
Opioids + Heroin	0.08 (0.02-0.31)*	2.18 (0.44-10.72)
Opioids + Antidepressants	2.08 (1.19-3.63)*	4.16 (1.54-11.21)*
Opioids + Anxiolytics	1.19 (0.74-1.93)	2.10 (1.01-4.34)*
aOR for Death
Opioids Alone	0.55 (0.10-3.08)	0.48 (0.06-3.58)d
Opioids + Alcohol	0.22 (0.02-2.14)	1.64 (0.21-12.7)d
Opioids + Illicit	0.10 (0.01-1.23)d	0.46 (0.05-4.08)
Opioids + Cocainec	0.43 (0.04-4.69)	---
Opioids + Heroinc	---	---
Opioids + Antidepressantsc	---	---
Opioids + Anxiolytics	0.11 (0.02-0.65)**	1.60 (0.20-12.99)d

^aModels also include race & age (except as indicated)

^bThis model includes only race, given colinearity between race and age

^cModels without estimates lack sufficient sample size for model stability

^dIndicates poor model fit

^*Indicates statistically significant, ie, the 95% confidence interval for the aOR estimate does not include the null (1.0)

In both women and men, using opioids alone, rather than in combination with alcohol or other drugs or medications, was associated with decreased odds of general hospital admission. Of note, several models evaluating association with ICU admission and death demonstrated poor fit, likely due to the small sample sizes for these outcomes (Table 3, Appendix A and B).

Discussion

Previous literature has demonstrated a tremendous heterogeneity among opioid users in terms of patterns of use and the interactions between types of use and subgroups of gender, race/ethnicity, familial substance abuse, routes of administration, concurrent drug use, and comorbid psychiatric and medical disorders.^10,14–17 With the premise that understanding the needs of specific subgroups of users may help develop more effective screening and treatment approaches, we examined a gender-stratified, nationally-representative population of opioid users seeking ED care.

Opioid use in the study population was high, and men and women were equally represented among opioids users. Although men and women used similar types of prescription opioids in combination with alcohol, illicit drugs, and antidepressants, there were differences between them in clinical outcomes within drug combination categories. This may be due to patterns or amount of drug taken, polysubstance use, or different thresholds for seeking healthcare. The difference may also be biological: observed sex differences include greater susceptibility to adverse effects of drugs, which may also contribute to gender differences in hospital admission and ICU care involving these drug combinations.

For both men and women, opioids taken alone posed similarly lower risk for need for hospitalization than when taken in combination with other substances, an intuitive finding that confirms the higher risk of coingestion observed in previous studies.^18,19 Further, the lower risk of single agent was similar between genders even though the specific agent involved differed between men and women who presented with opioids alone.

Among the drug combinations studied, ICU admission was associated with opioids and antidepressants for both women and men, and with opioids and anxiolytics only for men. Although this study did not examine the characteristics of the non-opioid drugs involved in the ED visit, it may be that the long-acting formulations available for these drug categories played a role in the need for ICU-level care. There may be additional factors underlying the gender difference observed for opioids and anxiolytics, such as the quantity or formulation of anxiolytic taken by men who presented to the ED with this drug combination.

Women in the DAWN dataset were more likely to have an opioid ingestion in combination with antidepressants and men more likely to use opioids with alcohol, consistent with previous literature that described gender-specific reasons for opioid use.²⁰ We also found that these particular combinations were clinically severe (ie, were associated with elevated odds of hospitalization) for each gender, respectively. Further study is needed to understand the explanation for this; it may be that social expectations or biases on the part of medical/behavioral healthcare providers make women more likely to have opioids and antidepressants prescribed together, prescribed in a way or used or metabolized by women in a way that creates higher risk for presentation to the ED and hospitalization. Similarly, alcohol and opioids may be a particularly dangerous combination in men, whether because of gender-determined differences in the way opioids and alcohol are taken together, or less caution in screening for alcohol problems or prescribing opioids to men with histories of high risk alcohol use.

With ED visits for prescription opioid misuse still on the rise, our study adds to the literature supporting an important role for the ED in examining and preventing medication safety errors, improving safer prescribing of opioids, and educating patients about combinations and drug-drug interactions.²¹ Our study underscores the importance of considering the patient's history of substance abuse and mental health conditions and current medication lists –such as referencing a prescription monitoring program –when making prescribing decisions, and of providing patients, particularly those treated with other medications, of the serious potential for addiction or death.²²

Prior studies have characterized gender differences in prescription opioid misuse based upon self-report. This study provides more objective data on opioid misuse among a subset of men and women who experienced clinically significant consequences of drug use and associated co-ingestions. While this data cannot elucidate the root cause of these differences, it provides impetus for a deeper, gender-specific understanding of the complex factors that may lead to serious morbidity from prescription opioids.

Limitations

DAWN data were collected through retrospective chart review, and thus the determination of eligibility for an individual case was dependent on the quality and accuracy of clinical information captured in the patient chart. It is possible that biases on the part of clinicians could lead to over-representation of drug misuse for one gender or the other. Clinical information captured in DAWN is limited. Although we could observe drugs and drug combinations implicated in visits, we are not able to correlate these with clinical diagnoses, pain-related complaints, co-occurring medical or psychiatric issues, or other specific information that would enable us to more thoroughly evaluate potential confounders of the relationship between opioid use and gender. The data were confined to individuals 18 years and older; thus findings may not generalize to younger populations presenting to the ED. Finally, in this exploratory analysis, we did not achieve good fit for some of the models examining drug combinations and the rarer outcomes of ICU admission and death. These associations merit further exploration in a larger dataset powered to examine these serious outcomes more closely.

Conclusions

ED visits related to prescription opioid misuse are numerous and often involve combinations with other substances. We found gender differences in these visits, especially in patterns of co-ingestions of other substances and associations between specific drug combinations and hospital admission. A better understanding of the gender factors involved in the initiation, misuse, treatment needs, and clinical outcomes may inform the development of gender–specific interventions and preventive measures.