The burden of general medical illness among individuals with serious mental illness is of growing concern (
1,
2,
3 ). Efforts to improve the quality of general medical care must examine the use of the emergency department for poorly controlled general medical illness (
4,
5,
6 ). Although data suggest that individuals with serious mental illness have elevated use of medical emergency services (
7,
8,
9 ), little is known about their patterns of use or associated factors. We aimed to identify correlates of increased medical emergency department use over a six-month period among individuals with serious mental illness discharged from an acute psychiatric inpatient stay.
Methods
Data were drawn from a study testing the effectiveness of a three-month brief critical time intervention (B-CTI) compared with usual care for veterans with serious mental illness. Details of the study procedures and intervention, aimed at improving continuity of psychiatric outpatient care after hospitalization, are described elsewhere (
10 ).
Participants were recruited from four inpatient psychiatric units within the Veterans Affairs (VA) Capital Health Care Integrated Service Network between February 2003 and April 2005. We omitted one site without an emergency department. Participants were 18–70 years old, living within 50 miles of the facility, and had a chart diagnosis of a schizophrenia spectrum or psychotic mood disorder. Participants were at risk for dropout from mental health treatment and met one of the following criteria: co-occurring substance use disorder; medication nonadherence over the past year (per patient report or medical chart); or psychiatric inpatient admission in the past two years, followed by a readmission, mental health emergency department visit, or no outpatient mental health visit within 30 days postdischarge. After receiving a complete description of the study, patients gave written informed consent.
Of 212 inpatients approached, 156 (74%) consented to participate. Twenty patients were withdrawn from the study; 15 became ineligible because of changes in discharge plan or changes in diagnosis, and five were discharged before assessments could be completed. Eighteen participants from the site without an emergency department were excluded from the study, leaving 118 participants. The study was approved by the institutional review board at the University of Maryland, Baltimore.
Participants completed a 90-minute baseline interview, which included demographic and clinical characteristics. Chart reviews provided service utilization data at VA facilities within Veterans Integrated Service Network 5 during the six months after discharge.
We identified three groups based on the number of medical emergency department visits during the six-month follow-up period: individuals with no visits, with one visit, and with two or more visits. We constructed variables from the service utilization data for the six-month follow-up period to assess timeliness, intensity, and continuity of outpatient visits. Timeliness was measured by the number of days until the first outpatient visit and whether an outpatient visit occurred during the first 30 or 180 days after hospital discharge. Variables used to assess intensity of care were the number of days after discharge on which there were outpatient visits during the first 30 days and the first six months. To assess continuity of care, we created a variable measuring the number of 60-day periods with two or more outpatient visits in the first six months after discharge (possible values: 0, 1, 2, or 3). These variables were constructed for both general medical visits and outpatient visits for mental health or substance abuse treatment.
The treating inpatient psychiatrists' diagnoses of mental disorders were obtained for descriptive purposes. Participants were assigned to the category of schizophrenia spectrum disorders or mood disorders.
Items modified from the Medical Expenditure Panel Survey assessed perceived barriers to receipt of medical care and satisfaction with care quality.
The Brief Psychiatric Rating Scale (BPRS) assessed psychopathology in the past week. Subscale (positive, negative, hostility, anxiety, and withdrawal) and total scores were used. Self-reported alcohol and drug use in the past month was assessed with the Schizophrenia Patient Outcomes Research Team survey, which asks participants how many days in the past month they used substances and how often they used substances in the past six months (
11 ).
Items derived from the National Health and Nutrition Examination Survey III (
12 ) assessed whether participants currently had any one of 14 medical conditions, including high blood pressure, heart problems, and diabetes.
The Quality of Life Interview (QOLI) (
13 ) assessed objective and subjective quality of life in eight domains: living situation, daily activities, family relations, social relations, finances, work and school, health, and legal and safety issues.
We first compared the three medical emergency department groups across demographic characteristics, general medical conditions, medical care barriers, psychiatric symptoms, substance abuse and dependence, quality of life, and outpatient medical visits and visits for mental health-substance abuse treatment. Pairwise comparisons of demographic variables across the groups were conducted with chi square, Fisher's exact, and t tests. The remaining pairwise comparisons were based on regression models to adjust for intervention group and site.
Logistic regression was used to compare proportions of the three groups that had an outpatient mental health or substance abuse treatment visit within 30 and 180 days of discharge, a general medical outpatient visit in the 30 and 180 days after discharge, a general medical illness or substance use disorder, and medical care barriers.
Linear regression was used to compare means of continuous variables among the groups, including the number of 60-day periods in which there were at least two outpatient mental health or substance abuse treatment visits, the number of 60-day periods with at least two general medical visits, the BPRS symptom scale scores, and the QOLI subscale scores.
We used Cox proportional hazards modeling to compare the groups' average time to first outpatient mental health-substance abuse treatment visit and general medical visit. Negative binomial regression was used to compare the groups on number of days with outpatient medical visits in the first 30-day period and during the 180-day period after discharge. We did parallel analyses with mental health and substance abuse treatment visits.
Results
Of 118 participants, most were men (N=107, 91%), were either African American (N=65, 55%) or white (N=49, 42%), and had 12 or more years of education (N=103, 87%). The mean±SD age was 47.7±7.9 years. Approximately one-third (N=42, 36%) of participants had schizophrenia, and the others (N=76, 64%) had a psychotic mood disorder. Forty-seven percent (N=56) of participants received B-CTI. Nearly all (108 of 114, 95%) had at least one current comorbid medical condition.
Nearly half of the participants (N=54, 46%) had one or more medical emergency department visits in the follow-up period. About one-fifth (N=24, 20%) had two or more visits, and 30 (25%) had one visit. Medical emergency department use did not vary between treatment condition (B-CTI versus the comparison group) and was not associated with psychiatric diagnosis, number of current general medical conditions, gender, race, education, marital status, or age. Only heart and skin problems correlated with medical emergency department use. Participants with two or more emergency visits were more likely to have heart problems than those with one visit (eight of 23, or 35%, versus one of 27, or 4%) ( χ 2 =5.54, df=1, p=.019). Participants with one emergency visit were more likely than those with no visits to have skin problems (11 of 29, or 38%, versus 11 of 60, or 18%) ( χ 2 =4.53, df=1, p=.033).
Overall, increased medical emergency department use was associated with greater use of general medical outpatient services in terms of timeliness, intensity, and continuity (
Table 1 ). In contrast, medical emergency department use was not consistently linked to use of mental health-substance abuse treatment services in the first 30 days after discharge (
Table 1 ). However, outpatient mental health visits and substance abuse treatment visits in the six months after discharge appeared to be associated with increased medical emergency department visits.
Satisfaction with medical care was not correlated with emergency department visits, and only one barrier to care differentiated visit groups. Those with one visit were more likely than individuals with no visits to report delaying medical care because their provider did not give them an appointment (N=7 of 29, or 24%, versus 2 of 60, or 3%; χ 2 =6.50, df=1, p=.011). Individuals with two or more visits had higher total BPRS scores (2.3±.6 versus 2.03±.6 out of a possible 7 points, with higher scores indicating increased severity; χ 2 =2.23, df=1, p=.028) and anxiety subscale scores (3.93±1.1 versus 3.1±1.3 out of a possible 7 points, with higher scores indicating increased severity; χ 2 =2.87, df=1, p=.005) than individuals with no visits. No group differences were observed in other BPRS subscale scores or substance use disorder diagnoses. The only difference on QOLI scales was that the individuals with two or more visits reported lower levels of satisfaction with their current living situation than individuals with one visit (3.6±1.6 versus 4.8±1.7 out of a possible 7 points, with higher scores indicating better quality of life; t=–3.17, df=108, p=.002) or no visits (3.6±1.6 versus 5.1±1.7; t=–3.09, df=108, p=.003).
Discussion and conclusions
We found that medical emergency department use was relatively common in this population of veterans with serious mental illness who were at risk of psychiatric treatment dropout.
These data did not support our expectation that increased rates of medical problems and substance abuse and dependence would drive medical emergency department use. Of note, we did not measure severity of medical illness, which might predict emergency department use. We relied on self-report data regarding medical illnesses, so it is possible that they were underreported. We did not find strong support for the notion that barriers to outpatient medical care were related to increased medical emergency department use.
Our most striking finding was the consistent association between increased use of outpatient medical care and emergency department services. This is somewhat contrary to the expectation that use of outpatient services would replace and prevent use of emergency services, although it is consistent with findings from other studies in the general population, which have found increased emergency department use to be associated with increased outpatient service use (
5,
6 ). It is possible that participants may have had general medical conditions that were not identified until the emergency department visit, which may have led to subsequent medical outpatient visits. Of note, we found that those who had more medical emergency department visits also used more outpatient mental health and substance abuse treatment services. The overall increased use of services could be driven by increased severity of both mental illness and medical disorders, although our data cannot tease apart those possibilities. However, the increased overall psychiatric symptoms and anxiety of the patients with two or more emergency visits suggest that psychiatric symptoms may play a role in use of medical services and underscores the need to address mental and medical health needs in an integrated fashion.
Individuals with more medical emergency department use did not vary from those with less use in their type of living situation but did report lower satisfaction with their living situation. Perhaps dissatisfaction with one's living situation lowers the decision-making threshold for using the emergency department for medical problems.
This study was limited by its relatively small sample and restriction to veterans with a recent psychiatric hospitalization, who were mostly male; had access to care within the VA health care system, including outpatient medical services; and differed from the general population in overall health status. These limitations were balanced by the examination of a carefully evaluated sample with limited financial barriers to service use. Our study supports the proposition that persons with serious mental illnesses require global and integrated attention to both general medical and mental health needs in order to use health care services more effectively and efficiently.