Professional Documents
Culture Documents
Original Research
Sex Differences in Prevalence of Emergency
Department Patient Substance Use
Robert D. Cannon, DO1; Gillian A. Beauchamp, MD1; Paige Roth, MSW1;
Jennifer Stephens, DO2; David B. Burmeister, DO1; David M. Richardson, MD1;
Alanna M. Balbi, BS1; Tennessee D. Park, BA1; Stephen W. Dusza, DrPH1; and
Marna Rayl Greenberg, DO, MPH1
1
Department of Emergency and Hospital Medicine, Lehigh Valley Hospital and Health Network/University
of South Florida Morsani College of Medicine Allentown, Pennsylvania; and 2Department of Internal
Medicine, Lehigh Valley Hospital and Health Network/University of South Florida Morsani College of
Medicine Allentown, Pennsylvania
recently, specific attention has been brought to the Level 1 trauma center with an annual census of
opioid crisis in the United States.6 Of the 21.5 million 90,000 visits per year (site B), and a suburban hospital
Americans 12 years or older that had a substance use with an annual census of 58,000 visits per year
disorder in 2014, 1.9 million involved prescription (site C). All patients with International Classification
pain relievers and 587,000 involved heroin.7 From of Diseases, Tenth Revision, ED final diagnosis code
1999 through 2008, overdose death rates, sales, and (s) for substance use F10.000 through F19.999 (ex-
substance use disorder treatment admissions related to cluding F17 codes for nicotine) were abstracted
prescription pain relievers increased in parallel—the electronically by an automated query established by
overdose death rate in 2008 was nearly 4 times the an information technologist using SAP BusinessOb-
rate in 1999, while sales of prescription pain relievers jects Web Intelligence (WebI; SAP.com) who was
in 2010 were 4 times those in 1999—and the blinded to the study protocol for ED visits at all 3
substance use disorder treatment admission rate in hospitals. Data included ED clinical enrollment site,
2009 was 6 times the 1999 rate.8 primary substance used, sex, date of ED visit, and
Sex differences have been implicated as an impor- disposition (including left without being seen, left
tant factor in the etiology, pathophysiology, sequelae, against medical advice, discharged, admitted, and
and treatment of substance use disorders.9 Specifically treated in rehabilitation) for 18 months (January 1,
in the ED population, explorations of sex differences 2016 through July 31, 2017).
in substance use among adult emergency patients Descriptive statistics and graphical methods were used
have illustrated that males have a higher prevalence to assess the distribution of study variables by patient
of substance use, in particular, lifetime use of sex. Univariate analyses included Student t tests and
nonprescription opioids, as well as methadone or Pearson χ2 to assess differences in the distribution of
buprenorphine, are reported to be twice as high for study variables by patient sex. Exact 95% binomial CIs
males as females.9 Other sex-specific differences in were calculated for the relative frequency estimates for
opioid use are more concerning for females. Females each drug. Random effects regression was used to
are more likely to take prescription opioids without a evaluate differences in age by substance (drug), control-
prescription to cope with pain, even when they are ling for disposition of patient from the ED. A random
reporting similar pain levels.10 Research also suggests effect was added to these analyses because patient data
that females are more likely to misuse prescription were aggregated from 3 separate hospitals within the
opioids to self-treat for other problems, such as same health network. In addition, multinomial logistic
anxiety or tension.11 regression was used to evaluate associations between ED
While the overall prevalence of substance use and disposition (coded as admitted, discharged, and other)
misuse in the context of the current substance use and patient sex by substance type (coded on three levels:
epidemic has been reported, sex-specific trends in alcohol, opioids, and other). These analyses included
substance-related ED visits have been less well docu- assessments of the interaction between patient sex and
mented.9 We present the sex-specific prevalence in ED substance type. Marginal predicted probabilities were
encounters during this current epidemic related to estimated from these models and plotted to help better
patient substance use. visualize the relationship between being admitted and
substance type by patient sex. All regression models
included patient age to control for potential confounding.
PATIENTS AND METHODS Additionally, a time series analysis was completed to
This study was approved by Lehigh Valley Health evaluate the overall counts and relative proportions of
Network’s Institutional Review Board using the ex- substances (drugs) encountered in the ED between
pedited review procedure in accordance with regula- January 1, 2016 and July 31, 2017, out of all the
tory requirements. Data on adult ED visits from 3 patients evaluated for drug-related issues. These analyses
hospitals in northeastern Pennsylvania were ab- aggregated data per month and were stratified by patient
stracted using a tool that had been developed for sex. Linear tests for trend were performed to assess any
hospital quality improvement. The contributing hos- trends in the data over the course of the evaluation time
pitals were an inner-city hospital with an annual frame. All analyses were performed using STATA soft-
census of more than 30,000 visits per year (site A), a ware, version 14.2 (Stata Corp, College Station, Texas).
Table I. Characteristics of patient visits to the emergency department in the Lehigh Valley Health Network
between January 2016 and July 2017 for substance use-related visits.
Age, y, mean (SD) 43.6 (16.4) 42.4 (16.3) 44.3 (16.4) o0.001
Drug, n (%) Alcohol 5704 (54.3) 1802 (49.9) 3902 (56.6) o0.001
Cannabis 1509 (14.4) 542 (15.0) 967 (14.0) 0.18
Cocaine 431 (4.1) 136 (3.8) 295 (4.3) 0.21
Opioid 2014 (19.2) 829 (23.0) 1185 (17.2) o0.001
Other 853 (8.1) 302 (8.4) 551 (8.0) 0.42
ED disposition, n (%) Admit 4746 (45.2) 1593 (44.1) 3153 (45.7) 0.01
Discharge 3531 (33.6) 1305 (36.1) 2226 (32.3) o0.001
Transfer 249 (2.4) 80 (2.2) 169 (2.5) 0.38
Left AMA 94 (0.9) 23 (0.6) 71 (1.0) 0.04
Observe 222 (2.1) 66 (1.8) 156 (2.3) 0.11
Other 50 (0.5) 24 (0.7) 26 (0.4) 0.05
Missing 1619 (15.4) 550 (14.4) 1099 (15.9) o0.001
Facility, n (%) Site A 1226 (11.7) 386 (10.7) 840 (12.2) 0.04
Site B 5376 (51.2) 1841 (51.0) 3535 (51.2) −
Site C 3909 (37.2) 1384 (38.3) 2525 (36.6) −
Table II. Measures of central tendency of age by patient sex and primary substance related to emergency
department visit.
Alcohol 5704 48.8 (15.9) 1802 46.7 (16.2)* 3902 49.9 (15.9)
Cannabis 1509 33.5 (14.9) 542 32.7 (15.0) 967 33.9 (14.8)
Cocaine 431 42.4 (11.8) 136 39.9 (11.4)* 295 43.5 (11.7)
Opioid 3014 39.9 (14.7) 829 41.3 (15.5)* 1185 38.9 (14.0)
Other 853 35.9 (14.4) 302 37.7 (15.6)* 551 34.9 (13.6)
⁎
Significant differences between sex, based on random effects linear regression model.
Admit Male
Proportion
18-month study period can be found in Table III.
0.348
0.225
0.262
0.056
Overall, patients with diagnoses of alcohol-related
issues were significantly older than patients with
diagnoses of other substances (48.8 vs 37.4 years;
Proportion P o 0.001). Compared with patients diagnosed with
Female
Admit alcohol-related problems, patients diagnosed with
0.202
0.229
0.055
0.312
cannabis-related reasons were 15.2 years younger
(95% CI, 14.4–16.1), patients diagnosed with opioid
reasons were 8.8 years younger (95% CI, 8.1–9.6),
Proportion
0.329
0.055
0.244
0.212
years younger (95% CI, 4.9–7.9). When further
Table III. Total emergency department visits, by participating hospital and sex for the 18-month study period.
P o 0.001).
When exploring the disposition from the ED and
with being admitted as the referent group, males were
14% less likely to be discharged than females (odds
ratio ¼ 0.86; 95% CI, 0.79–0.95). However, these
Population
Proportion
Female
0.544
0.556
0.527
0.572
8759
22,866
Admit
34,206
9860
Female
Admit
66,985
2671
18,619
Admit
Total
Total
A
1
* *
.6
.5
.4
.3
.2
50
.1
0
0
Ap 16
N 016
Ap 17
M 016
M 017
J u 16
ec 6
J u 17
Se 16
J a 16
Au 16
7
M 16
M 17
Ju 6
Ju 7
O 16
16
F e 17
1
01
1
1
20
20
20
20
20
20
20
0
20
20
20
20
20
20
20
2
r2
r2
l2
l2
Female Male
ar
ct
ar
ay
ov
ay
g
b
b
n
n
p
n
n
Fe
Ja
D
Alcohol Cannabis Opioid
Figure 1. Bar chart of the marginal predicted
probabilities of being admitted after B
1
emergency department visit for drug-
.9
Proportion of drug-related ED visits
related reasons by substance and
.8
patient sex. *P o 0.05 for differences
.7
.6
in predicted probabilities different by
.5
patient sex.
.4
.3
.2
.1
N 016
Ap 17
6
M 017
16
ec 6
Ju 17
Se 16
Ja 16
6
7
16
M 17
16
Ju 7
O 16
16
Fe 17
01
1
01
01
1
20
20
20
20
20
20
20
20
20
20
20
20
20
20
2
r2
r2
l2
l2
ar
ct
ar
ay
ov
ay
g
b
b
n
n
period. Much of this difference can be attributed to
p
n
n
Ap
Ju
Au
Fe
Ju
Ja
D
alcohol- and cannabis-related visits. Alcohol Cannabis Opioid
A C
.25
1
Cannabis
.9
.2
.8
Proportion of Cases
.7
.15
Proportion
.6
.5
.1
.4
.3
.05
.2
.1
0
0
16
16
16
16
16
16
16
16
16
16
17
17
17
17
17
16
16
16
16
16
16
ct 6
ov 6
ec 6
Ja 16
Fe 17
ar 7
Ap 17
ay 7
J u 17
17
7
01
01
01
01
01
01
01
01
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
r2
l2
r2
l2
r2
l2
r2
l2
n
ar
ay
ct
ov
ec
ar
ay
ar
ay
n
Ju
Ju
Ju
Ju
Ap
Ap
Ap
Ja
Fe
Ju
Au
Se
Ja
Fe
Ju
Ja
Fe
Ju
Au
Se
O
O
M
M
M
M
Male Female Proportion - Female Proportion - Male Overall
B D
.3
.7
Alcohol Opioids
.6
.25
.5
.2
Proportion
.4
.15
.3
.1
.2
.05
.1
0
0
16
16
16
16
16
16
16
16
16
16
17
17
17
17
17
16
16
16
16
16
16
16
16
16
16
17
17
17
17
17
7
01
01
01
01
01
01
01
01
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
r2
l2
r2
l2
r2
l2
r2
l2
n
ar
ay
ct
ov
ec
ar
ay
ar
ay
ct
ov
ec
ar
ay
n
Ju
Ju
Ju
Ju
Ap
Ap
Ap
Ap
Ja
Fe
Ju
Au
Se
Ja
Fe
Ju
Ja
Fe
Ju
Au
Se
Ja
Fe
Ju
O
O
M
M
M
M
Proportion - Female Proportion - Male Overall Proportion - Female Proportion - Male Overall
Figure 3. (A) Time series plot of the proportion of drug-related emergency department (ED) cases from January
2016 until July 2017 by patient sex. (B) Time series plot of the proportion of alcohol-related ED cases
from January 2016 until July 2017 by patient sex. (C) Time series plot of the proportion of cannabis-
related ED cases from January 2016 until July 2017 by patient sex. (D) Time series plot of the proportion
of opioid-related ED cases from January 2016 until July 2017 by patient sex.
Our study time frame for trend analysis was small, coding and the impact on study findings of the inherent
making it difficult to generalize patterns of use for many limitations of coding errors is not known. Due to the
of the substances over such a short period of time. Our method of abstraction, our study did not set out to
sex-specific trends found neither remarkable escalations determine whether this diagnosis of substance use was a
nor diminishing patterns in males or females presenting to new or established diagnosis or if there were other
the ED with opioid diagnoses in this time frame. secondary medical problems that complicated the dispo-
sition (admission). While our findings certainly imply
Limitations association of final diagnosis of substance use−related
Although the study was performed involving both diagnosis to admission, due to limitations of electronic
urban and suburban settings, these findings in northeast- data abstraction they do not clearly illustrate causation in
ern Pennsylvania may not be geographically generalizable all the cases that were analyzed. Out of design conven-
to other ED populations. Our results indicate statistical ience prevalence of tobacco use was not studied. We did
significance in age parameters (eg females tended to be 2 not capture ethnicity, insurance status, or other potential
years younger than males) that may not be clinically as confounders for our results. The impact on the study
significant. Electronic abstraction was based on diagnosis results due to the rate of missing data for disposition is
not known. Electronic abstraction of the records did not Rockville MD: (SAMHSA) Office of Applied Studies;
allow for determining whether females and males were 2006.
offered admission with the same frequency or whether 3. McDonald AJ, Wang N, Camargo CA. US emergency
women declined the admission (but were not asked to department visits for alcohol-related disease and injuries
sign an against medical advice form). between 1992 and 2000. Arch Intern Med. 2004;164:
531–537.
4. D’Onofrio G, Becker B, Woolard R. The impact of alcohol,
CONCLUSIONS Tobacco and other drug use and abuse in the emergency
There are sex-specific differences in prevalence of department. Emerg Med Clin N Am. 2006;24:925–967.
those diagnosed with substance use in the ED setting. 5. Cherpitel CJ, Ye Y. Trend in alcohol- and drug-related
The prevalence of opioid use and the sex-specific emergency department and primary care vistis: data from
differences in disposition for those with opioid use four U.S. national surveys (1995-2010). J Stud Alcohol
in comparison with other substances are worthy of Drugs. 2012;73:454–458.
consideration when determining resources necessary 6. Center for Behavioral Health Statistics and Quality. Key
to combat the substance use and overdose epidemic. substance use and mental health indicators in the United
States: Results from the 2015 National Survey on Drug
Use and Health (HHS Publication No. SMA 16-4984,
ACKNOWLEDGMENTS NSDUH Series H-51); 2016.
The authors would like to acknowledge the research 7. Substance Abuse and Mental Health Services Adminis-
operations management of Anita Kurt, PhD, RN, and tration, Center for Behavioral Health Statistics and
the research coordinator assistance of Manuel F. Quality. Behavioral health trends in the United States:
Colón, BS and Micaela B. Wilson, BA. Results from the 2014 National Survey on Drug Use and
Health. Rockville, MD: Substance Abuse and Mental
Health Services Administration; 2015.
AUTHOR CONTRIBUTIONS 8. Centers for Disease Control and Prevention. Vital Signs:
All authors provided substantial contributions to con- Overdoses of Prescription Opioid Pain Relievers—US
ception and design, literature search, acquisition of data, 1999-2008. MMWR Morb Mortal Wkly Rep. 2011;60:
or analysis and interpretation of data. SD performed the 1487–1492.
analysis and created the figures. MRG and SD drafted the 9. Beaudoin FL, Baird J, Liu T, Merchant R. Sex differences in
article and all authors equally contributed to its revision substance use among adult emergency patients: preva-
for important intellectual content. All authors gave final lence, severity, and need for intervention. Acad Emerg Med.
approval of the version of the article to be published. 2015;22:1307–1315.
10. National Institute on Drug Abuse. Substance use in women—
sex and gender differences in substance use (pain relievers—
CONFLICTS OF INTEREST opioids). https://www.drugabuse.gov/publications/resear
The authors have indicated that they have no conflicts ch-reports/substance-use-in-women/sex-gender-differences-in-
of interest regarding the content of this article. substance-use. Accessed August 9, 2017.
11. McHugh RK, DeVito EE, Dodd D, et al. Gender differ-
ences in a clinical trial for prescription opioid dependence.
REFERENCES J Subst Abuse Treat. 2013;45:38–43.
1. Cherpitel C. Drinking patterns and problems: a compar- 12. Centers for Disease Control and Prevention. Vital signs:
ison of primary care with the emergency room. Subst prescription painkiller overdoses: a growing epidemic,
Abuse. 1999;20:85–95. especially among women. www.cdc.gov/vitalsigns/pre
2. Mallonee E, Calvin SL. Emergency Department Visits scriptionpainkilleroverdoses/index.html. Reviewed July 2,
Involving Underage Drinking. The New DAWN Report. 2013. Accessed August 9, 2017.