Gender Differences in Hospitalization After Emergency Room Visits for Depressive Symptoms

Abstract

Background:

Depressed women have greater than three times the odds of hospitalization as clinically comparable men. The objective of this study is to understand if these gender differences emerge in admissions decisions after depressed individuals' arrival at the emergency room (ER).

Methods:

We used multivariate logistic regression to examine gender differences in hospitalization after 6266 ER visits for depressive symptoms in the nationally representative 1998–2007 National Hospital Ambulatory Care Medical Survey.

Results:

ER visits by depressed women have only 0.82 the odds of hospitalization (95% confidence interval [CI] 0.70-0.96, p=0.02) in models adjusted for sociodemographic, clinical, and system covariates. Sensitivity analyses demonstrate gender differences in visits by patients with no injury but not in visits by patients with self-inflicted injury.

Conclusions:

These findings suggest that admission decisions after ER visits are not responsible for the increased risk of hospitalization previously reported in depressed women, as ER visits by women with depressive symptoms actually have lower odds of hospitalization than visits by men. We encourage further research to explore the causes and consequences of this practice pattern to move toward rational delivery systems committed to providing comparable treatment to clinically comparable individuals regardless of gender.

Introduction

National community studies demonstrate that women have 1.4 times the odds of recent major depressive disorder (MDD) compared to men.¹ Among people with recent depression, women have over three times the odds of any outpatient depression treatment (primarily delivered in general medical settings) compared to men.² Given greater outpatient depression treatment has the potential to prevent acute crises, it is surprising that depressed women have over three times the odds of hospitalization,³ a gender difference that has not been reported for hospitalization for other diagnoses.⁴ Because emergency room (ER) visits are an important but not exclusive pathway to hospitalization for many conditions, including depression, we expected depressed women to have higher rates of hospitalization after an ER visit for depression than their male counterparts. Evidence that admission after ER visit differs by gender is important because it provides evidence that gender may impede the delivery of clinically comparable care to clinically comparable patients. Whereas men are at greater risk for a subsequent completed suicide,^5,6 women are at greater risk for subsequent suicide attempt,⁵ morbidity that potentially could be prevented by a comparable probability of high-quality hospitalization and discharge planning.

Materials and Methods

To address this objective, we conducted a secondary analysis on ER visits sampled in the 1998–2007 National Hospital Ambulatory Care Survey (NHAMCS) published by the Ambulatory Statistics Branch of the Centers for Disease Control and Prevention (CDC) National Center for Health Statistics (NCHS). These visits represent an annual national probability sample of visits to providers delivering ER care in nonfederal hospitals during a randomly selected 4-week period. NHAMCS uses a four-stage probability sample design to obtain primary sampling units, hospitals within primary sampling units, ERs within these hospitals, and patient visits within ERs. During the 10-year study, NHAMCS sampled between 458 and 546 hospitals each year. Of the sampled hospitals, 391–462 hospitals were eligible to participate in the survey. Response rates ranged from 89.4% to 98.0%. ER providers completed a standardized questionnaire on each visit, collecting sociodemographic, clinical, and systems data. These data include up to three patient reasons for visit, up to three provider diagnoses based on International Classification of Diseases codes (Ninth Revision) (ICD-9), treatment provided during the ER visit, and visit disposition. NCHS entered all data using a two-way independent verification procedure for 10% of sample records, which established that coding errors for various NHAMCS items ranged from 0% to 0.7%. NCHS then weighted each visit to allow extrapolation to national estimates. NHAMCS data files were combined in accordance with NCHS direction for analysis.⁷

The sample for this analysis is all ER visits by people ≥18 years for depressive symptoms from 1998 to 2007. ER visits were eligible for inclusion in the study if the ER provider noted in up to three fields (1) reason for visit as depression or two plus depressive symptoms, including sleep disturbances, trouble concentrating, low self-esteem, restlessness, or suicidality, or (2) a diagnosis of depressive disorder (ICD-9 codes 296.2, 296.3, 300.4, 309.0, 309.1, 309.28, 311.0, and 648.4). The broad definition was used in our primary analyses to avoid biasing the sample to ER visits made to psychiatric hospitals, whose ER providers may be more likely to use Diagnostic and Statistical Manual of Mental Disorders (DSM) diagnoses.

The NCHS Institutional Review Board approved the protocols for NHAMCS including a waiver of the requirement for informed consent of participating patients. The Florida State University Institutional Review Board approved the analysis presented here.

Dependent variable

Hospitalization

Hospitalization was defined as ER provider notation that visits made by patients in the sample ended in hospitalization or transfer to another hospital/facility. Transfer was included in the definition of hospitalization to capture all patients who received any inpatient care after ER presentation. The NHAMCS dataset does not provide information on whether the hospitalization occurred in a general hospital or psychiatric facility, both of which provide inpatient care to depressed patients.

Sociodemographic variables

Gender

Gender was defined by provider report of female or male.

Age

Age was defined as provider report of years old coded as a continuous variable after recoding a small number of age values over 100 as missing.

Race/ethnicity

Race/ethnicity was defined as provider report as non-Hispanic white, non-Hispanic black, Hispanic, or other, including unknown.

Insurance

Insurance was defined as provider report of insurance as private, Medicare, Medicaid, none, or other, including unknown.

Rurality

Rurality was defined as provider report that the ER was in a nonmetropolitan statistical area, a designation used by the Office of Management and Budget (OMB) designation to determine eligibility and reimbursement for more than 30 federal programs.⁸

Clinical covariates

Injury

It is important to control for intentional self-inflicted injury as an important predictor of hospitalization.⁹ Injury as reported by the provider was coded as a categorical variable into three groups: self-inflicted injury, injury by other (assault, unintentional accident, legal intervention, adverse medical effect) or unknown source; or no injury.

Psychiatric and medical comorbidity

Psychiatric comorbidity was defined as the number of times any of up three provider diagnoses were coded as ICD-9 290.0–319.0, excluding ICD-9 codes for depressive disorder (296.2, 296.3, 300.4, 309.0, 309.1, 309.28, 311.0, 648.4). Medical comorbidity was defined as the number of times any of up to three provider diagnoses were coded using any ICD-9 code excluding codes for psychiatric comorbidity and depressive disorder.

ER medication

ER medication was defined as provider report of medications ordered or provided during the ER visit.

System covariates

It is important to control for healthcare system covariates that potentially impact the probability of hospitalization after an ER visit.¹⁰

M.D. provider

M.D. provider was defined by provider report as: (1) M.D., resident, or intern vs. nurse practitioner, physician's assistant, nurse, emergency medical technician (EMT), or unknown.

Weekend

Weekend was defined as a categorical variable by provider report as weekday (Monday, Tuesday, Wednesday, or Thursday) or weekend (Friday, Saturday, or Sunday).

Arrival time

Arrival time was defined as a categorical variable by provider report of patient arrival time in military hours between 0000 and 0759, 0800 and 1559, 1600 and 2359, or unknown.

Month of visit

Month of visit was categorized by provider report into one of four quarters.

Region

Region was defined as a categorical variable by practice location in the northeast, midwest, south, or east.¹¹

Analyses

We used recommended weights to account for selection probability, adjustment for nonresponse, and other factors necessary to reflect the universe of ER visits in the United States during the decade, producing national estimates with standard errors (SEs) that account for the complex sampling design.⁷ In the initial analysis, we compared gender differences in covariates using chi-squares for categorical variables and t tests for continuous variables. In subsequent analyses, we used logistic regression to test the relationship of gender to hospitalization after ER presentation for depressive symptoms, followed by sensitivity analyses in visits with and without self-inflicted injury to determine if gender differences in hospitalization were observable in both groups.¹² Using logistic regression, we also tested the relationship of gender to hospitalization after ER presentation for diagnosed depressive disorder. All logistic regressions were adjusted for covariates that predicted the dependent variable at p<0.2 in univariate analyses. In addition, each model included year coded as a linear variable after preliminary investigation identified no quadratic trends in hospitalization over time.

Results

Sample

The NHAMCS dataset contained 6266 eligible ER visits weighted to represent 15,180,590 visits/year between 1998 and 2007. In this sample, 3133 ER visits by women were weighted to represent 8,064,216 visits/year, and 3133 ER visits by men were weighted to represent 7,116,374 visits/year. Visits by women differed from visits by men by patient age, insurance, injury, medical comorbidity, psychiatric comorbidity, and arrival time (Table 1).

Table 1.

Visit Sample

Variable	Total n=6266 (weighted n=15,180,590)	Women n=3133 (weighted n=8,064,216)	Men n=3133 (weighted n=7,116,374)	p value
Sociodemographic characteristics
Mean age (SD)	40.83 (0.30)	41.88 (0.46)	39.65 (0.40)	<0.0001
Race/ethnicity
% Non-Hispanic white	69.91	71.59	68.05	0.06
% Non-Hispanic black	17.77	16.45	19.24
% Hispanic	9.75	9.13	10.44
% Other	2.56	2.83	2.27
% Rural	13.42	14.61	12.07	0.07
Insurance
% Private	24.67	27.99	20.92	<0.0001
% Medicare	16.01	16.36	15.61
% Medicaid	24.95	28.27	21.18
% None	23.40	17.27	30.35
% Other	10.97	10.12	11.94
Clinical characteristics
Injury
% No injury	57.93	69.06	49.85	<0.0001
% Self-injury	16.71	14.85	18.82
% Other	25.36	20.09	31.33
Mean medical comorbidities (SD)	0.54 (0.01)	0.62 (0.02)	0.45 (0.02)	<0.0001
Mean psychiatric comorbidities (SD)	1.01 (0.02)	0.79 (0.03)	1.25 (0.03)	<0.0001
Mean ER medications (SD)	1.50 (0.05)	1.54 (0.06)	1.46 (0.06)	0.10
System characteristics
% M.D. provider	96.22	96.03	96.43	0.59
% Weekend	39.46	39.11	39.85	0.69
Patient arrival time (MT)
% 1200–0759	16.42	14.41	18.70	0.01
% 0800–1559	39.90	40.43	39.30
% 1600–2359	42.33	43.77	40.70
% Unknown	1.35	1.40	1.31
Month
% January–March	24.22	24.49	23.91	0.64
% April–June	25.56	26.06	25.00
% July–September	26.07	26.27	25.84
% October–December	24.15	23.18	25.25
Region
% Northeast	24.45	24.48	24.42	0.99
% Midwest	23.63	23.42	23.86
% South	30.63	30.67	30.57
% West	21.30	21.43	21.14

ER, emergency room; MT, military time; SD, standard deviation.

Hospitalization

In ER visits with depressive symptoms, 42.2% of ER visits resulted in hospitalization, 39.4% of ER visits by depressed women compared to 45.3 % of ER visits by depressed men (p=0.001). ER visits by depressed women had lower odds of hospitalization (odds ratio [OR] 0.82, 95% confidence interval [CI] 0.70-0.96, p=0.02), controlling for sociodemographic, clinical, and system covariates (Table 2). Other predictors included older age (OR 1.01, CI 1.00-1.01, p=0.0005), Medicare (OR 1.39, CI 1.06-1.81, p=0.02), Medicaid (OR 1.39, CI 1.11-1.75, p=0.004), nonrurality (OR 0.70, CI 0.51-0.95, p=0.02), self-injury (OR 2.30, CI 1.83-2.90, p<0.0001), greater psychiatric comorbidity (OR 1.12, CI 1.01-1.24, p=0.03), less medical comorbidity (OR 0.81, CI 0.72-0.92, p=0.001), ER treatment by M.D. (OR 1.85, CI 1.27-2.69, p=0.001), nonweekend visit (OR 0.81, CI 0.69-0.95, p=0.0005), and earlier year (OR 0.94, CI 0.91-0.97, p=0.01). In visits with self-inflicted injury, depressed women and men had statistically comparable odds of hospitalization (OR 1.09, CI 0.71-1.66, p=0.70), controlling for sociodemographic, clinical, and system covariates. In visits with no injury, depressed women had lower odds of hospitalization (OR 0.76, CI 0.63-0.90, p<0.002) than men, controlling for sociodemographic, clinical, and system covariates. In ER visits with diagnosed depressive disorder, ER visits by depressed women continued to have lower odds of hospitalization compared to men (OR 0.81, CI 0.66-0.99, p=04).

Table 2.

Hospitalization After Emergency Room Visit

Parameter ^a	Beta	SE	P value	Odds ratio	95% CI around ORs
Intercept	−1.23	0.22	<0.0001	-	-
Female	−0.19	0.08	0.02	0.82	0.70-0.96
Age	0.01	0.00	0.0005	1.01	1.00-1.01
Race/ethnicity
Non-Hispanic black vs. non-Hispanic white	0.01	0.10	0.88	1.04	0.83-1.23
Hispanic vs. non-Hispanic white	−0.24	0.13	0.10	0.78	0.60-1.01
Other vs. non-Hispanic white	0.32	0.25	0.19	1.38	0.85-2.24
Insurance
Medicare vs. private	0.33	0.14	0.02	1.39	1.06-1.81
Medicaid vs. private	0.33	0.12	0.004	1.39	1.11-1.75
None vs. private	−0.01	0.13	0.91	0.99	0.77-1.26
Other vs. private	.02	0.15	0.85	1.03	0.76-1.38
Region
Midwest vs. northeast	0.22	0.13	0.10	1.25	0.96-1.63
South vs. northeast	0.04	0.11	0.73	1.04	0.83-1.29
West vs. northeast	−0.24	0.16	0.12	0.79	0.58-1.07
Rural	0.36	0.16	0.02	0.70	0.51-0.95
Injury
Self vs. none	0.83	0.12	<0.0001	2.30	1.83-2.90
Other vs. none	−0.12	0.11	0.26	0.88	0.71-1.10
Psychiatric comorbidity	0.11	0.05	0.03	1.12	1.01-1.24
Medical comorbidity	−0.21	0.06	0.001	0.81	0.72-0.92
ER medications	−0.05	0.03	0.07	0.95	0.91-1.01
M.D. provider	0.62	0.19	0.001	1.85	1.27-2.69
Weekend	−0.21	0.08	0.01	0.81	0.69-0.95
Year	−0.04	0.01	0.01	0.94	0.91-0.97

Covariates from Table 1 that did not meet criteria for inclusion in the model: Month and patient arrival time.

CI, confidence interval; SE, standard error; ORs, odds ratios.

Discussion

ER visits by depressed women have only 82% the odds of hospitalization as visits by men, after adjusting for sociodemographic, clinical, and system covariates. These findings suggest that admissions after ER visit decisions are not responsible for the increased risk of hospitalization previously reported in depressed women. Further research on two alternative explanations may shed further light on gender disparities in hospitalization: (1) physicians may directly admit (bypassing the ER) depressed women more often than men or (2) depressed women may be more likely to visit an ER than men, resulting in greater overall rates of hospitalization despite lower rates of admission after an ER visit.

Rather than the higher risks we expected to observe, depressed women who come to the ER have a lower risk of hospitalization than men for a potentially life-threatening condition. Although depressed men and women are hospitalized at comparable rates when they deliberately injure themselves, depressed women who have not injured themselves are less likely than men to be hospitalized. Because depressed people who have not injured themselves are rarely hospitalized without significant suicidal ideation, it is possible that depressed men reported greater rates of suicidal ideation in the emergency room visits than did women. Whereas suicidal ideation in community residents is more prevalent in women,¹³ suicidal ideation in ER patients may be more prevalent in men, consistent with their delay in seeking outpatient mental health services.¹⁴ If the women in our sample have greater suicidal ideation, our results underestimate depressed women's reduced odds of hospitalization. If the men in our sample have greater suicidal ideation, our results overestimate depressed women's reduced odds of hospitalization. However, it is also possible that suicidal ideation was not comparably evaluated in depressed male and female patients,^15
–17 consistent with gender differences observed in quality of care for coronary artery disease (CAD),^{18

–26} diabetes,^27,28 HIV,²⁹ and cancer prevention.^30,31 ER physicians may be more likely to ask men about suicidal ideation or to interpret their responses as indicating suicidal risk because they have been taught that men are at greater risk for completed suicide.^5,6

Our findings about gender differences in hospitalization rates have strong internal validity because the NHAMCS dataset provides ER discharge data, as well as important sociodemographic, clinical, and systems covariates; however, we recognize the possibility that the gender relationships we report potentially result from unobserved differences in suicidal ideation and other severity indicators that our clinical covariates did not adequately capture. We encourage future investigators to investigate causal factors for the gender differences we observed, as well as to characterize the consequences of these differences for subsequent suicidal behavior. Specifically investigators should examine (1) if gender differences in hospitalization are explained by suicidal ideation, which was not measured in this dataset, and (2) if gender differences are observable in both psychiatric and medical hospitalizations in depressed patients. Even though NHAMCS provides the most generalizable data on ER visits for depressive symptoms, the external validity of our findings is limited. Because the NHAMCS unit of analysis is an ER visit, frequently visiting patients may be overrepresented in our sample. Given that each ER sampled visits for a 4-week period, however, this source of measurement error is not likely to introduce a substantial bias.

Conclusions

Admission decisions after ER visits are not responsible for the increased risk of hospitalization previously reported in depressed women. Rather, these findings provide initial evidence that depressed women are not being hospitalized at the same rate as men after ER visits, identifying potential disparities in the delivery of crisis care for a potentially life-threatening disorder. We encourage further research to confirm or refute the findings we present, with concomitant exploration of the causes and consequences of this practice pattern. Such research is needed to inform the field's efforts to move toward rational delivery systems committed to providing comparable treatment to clinically comparable individuals, including hospitalization when it is needed, regardless of gender.

Footnotes

Disclosure Statement

No competing financial interests exist.

References

Kessler

, Berglund

, Demler

et al.

The epidemiology of major depressive disorder: Results from the National Comorbidity Survey Replication (NCS-R)

JAMA, 2003; 289:3095–3105.

Rost

, Zhang

, Fortney

, Smith

Jr . Expenditures for the treatment of major depression. Am J Psychiatry, 1998; 155:883–888.

Rost

, Zhang

, Fortney

, Smith

Jr . Rural-urban differences in depression treatment and suicidality. Med Care, 1998; 36:1098–1107.

Bertakis

, Azari

, Helms

, Callahan

, John

, Robbins

. Gender differences in the utilization of health care services. J Fam Pract, 2000; 49:147–152.

Nock

, Borges

, Bromet

et al. Suicide and suicidal behavior. Epidemiol Rev, 2008; 30:133–154.

World Health Organization. www.who.int/mental_health/prevention/suicide/country_reports/en/index.html

Hing

, Gousen

, Shimizu

, Burt

. Guide to using masked design variables to estimate standard errors in public use files of the National Ambulatory Medical Care Survey and the National Hospital Ambulatory Medical Care Survey. Inquiry, 2004; 40:416–415.

Hart

, Larson

, Lishner

. Rural definitions for health policy and research. Am J Public Health, 2005; 95:1149–1155.

Doshi

, Boudreaux

, Wang

, Pelletier

, Camargo

Jr . National study of U.S. emergency department visits for attempted suicide and self-inflicted injury, 1997–2001. Ann Emerg Med, 2005; 46:369–375.

10.

Smith

, Larkin

, Southwick

. Trends in U.S. emergency department visits for anxiety-related mental health conditions, 1992–2001. J Clin Psychiatry, 2008; 69:286–294.

11.

ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/dataset_documentation/NHAMCS/doc08.pdf

12.

Dombrovski

, Szanto

, Duberstein

et al. Sex differences in correlates of suicide attempt lethality in late life. Am J Geriatr Psychiatry, 2008; 16:905–913.

13.

Borges

, Angst

, Nock

, Ruscio

, Walters

, Kessler

. A risk index for 12-month suicide attempts in the National Comorbidity Survey Replication (NCS-R) Psychol Med, 2006; 36:1747–1757.

14.

Wang

, Berglund

, Olfson

, Pincus

, Wells

, Kessler

. Failure and delay in initial treatment contact after first onset of mental disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry, 2005; 62:603–613.

15.

Harbinder

, Adams

, Singleton

, Clark-Carter

, Kidd

. The impact of gender dyads on doctor-patient communication: A systematic review. Patient Educ Couns, 2009; 76:348–355.

16.

Bertakis

. The influence of gender on the doctor-patient interaction. Patient Educ Couns, 2009; 76:356–360.

17.

Babitsch

, Braun

, Border

, David

. Doctor's perception of doctor-patient relationships in emergency departments: What roles do gender and ethnicity play? BMC Health Serv Res, 2008; 8:82.

18.

Weitzman

, Cooper

, Chambless

et al. Gender, racial and geographic differences in performance of cardiac diagnostic and therapeutic procedures for hospitalized acute myocardial infarction in four states. Am J Cardiol, 1997; 79:722–726.

19.

Chou

, Scholle

, Weisman

, Bierman

, Correa-de-Araujo

, Mosca

. Gender disparities in the quality of cardiovascular disease care in private managed care plans. Womens Health Issues, 2007; 17:120–130.

20.

Chou

, Wong

, Weisman

et al. Gender disparities in cardiovascular disease among commercial and Medicare managed care plans. Womens Health Issues, 2007; 17:139–149.

21.

Correa-de-Araujo

, Stevens

, Moy

, Nilasena

, Chesley

, McDermottt

. Gender differences across racial and ethnic groups in the quality of care for acute myocardial infarction and heart failure associated with comorbidities. Womens Health Issues, 2006; 6:44–55.

22.

Rieker

, Bird

. Rethinking gender differences in health: Why we need to integrate social and biological perspectives. J Gerontol B Psychol Sci Soc Sci, 2005; 60B,Special Issue II:40–47.

23.

Bird

, Fremont

, Wickstrom

, Bierman

, McGlynn

. Improving women's quality of care for cardiovascular disease and diabetes: The feasibility and desirability of stratified reporting of objective performance measures. Womens Health Issues, 2003; 13:150–157.

24.

Daly

, Clemens

, Lopez Sendon

et al. Gender differences in the management and clinical outcome of stable angina. Circulation, 2006; 113:490–498.

25.

Penco

, Fratine

, Romano

, Novo

. Gender differences in the outcome of noninvasive cardiovascular treatment. Ital Heart J, 2003; 4:514–517.

26.

Care of women in U.S. hospitals. 2000. HCUP Factbook No. 3. AHRQ Publication No. 02-0044. Rockville, MD: Agency for Healthcare Research and Quality, 2002. www.ahrq.gov.data/hcup/factbk3/. 2010 July 21.

27.

Correa-de-Araujo

, McDermott

, Moy

. Gender differences across racial and ethnic groups in the quality of care for diabetes. Womens Health Issues, 2006; 16:56–65.

28.

Ferrara

, Williamson

, Karter

, Thompson

, Kim

. Sex differences in quality of health care related to ischemic heart disease prevention in patients with diabetes: The Translating Research Into Action for Diabetes (TRIAD) study, 2000–2001. Diabetes Care, 2004; 27:974–972.

29.

Hirschhorn

, McInnes

, Landon

et al. Gender differences in quality of HIV care in Ryan White CARE Act-funded clinics. Womens Health Issues, 2006; 16:104–112.

30.

Kosiak

, Sangl

, Correa-de-Araujo

. Quality of health care for older women: What do we know? Womens Health Issues, 2006; 16:89–99.

31.

Sherman

, Fu

, Joseph

, Lanto

, Yano

. Gender differences in smoking cessation services received among veterans. Womens Health Issues, 2005; 15:126–133.