As mental health care systems shift emphasis from inpatient treatment to outpatient care, emergency room staff increasingly encounter patients needing acute psychiatric care. The Dallas Veterans Affairs (VA) Medical Center emergency room averages more than 22,000 visits annually; approximately 3,000 of these visits are for psychiatric disorders. In 1996, the psychiatry service began reorganizing into outpatient-focused continuity-of-care teams. With reductions in the number of beds available, the efficient use of inpatient resources became critical. It was determined that a disproportionate percentage of psychiatric admissions (40%) occurred in the evening through the general emergency room. This study was initiated to identify characteristics of patients seen in the emergency room for evening psychiatric consultation and to assess diagnostic correlates of subsequent psychiatric hospitalization and mortality.
Method
Data were collected on patients receiving psychiatric consultation in the emergency room between 5:00 p.m. and 12:00 midnight, December 1, 1996, to December 31, 1997. Only the first visit of repeat visitors was counted, creating a study group of unique patients. A data collection form and consultation note were available for each encounter. Information from the encounter form, progress notes, and discharge summaries included the first three DSM-IV axis I diagnoses, reason for consultation, basic demographic information, and disposition from the emergency room.
In March 2002, subjects’ medical records were reviewed for subsequent psychiatric hospitalizations at the medical center. Admissions from the initial emergency room contact or to partial hospitalization programs were not counted. Attempts to determine cause of death included referring 10 cases suspicious for suicide to the medical center risk manager, who verified the cause of death with the county medical examiner.
Results
Of 504 unique patients identified, 54 (11%) received two or more psychiatric consultations to the emergency room during the study period. The mean age of the 504 patients was 47.1 (SD=11.2). Ninety-four percent (N=473) were male, and 28% (N=140) were married. Ethnic distribution was 63% (N=315) Caucasian, 34% (N=169) African American, 4% (N=19) Hispanic, and 0.2% (N=1) Native American. Immediately following initial consultation, 31% (N=157) were admitted to the inpatient psychiatry unit, and 81% (N=408) received outpatient psychiatric follow-up care at the medical center.
The most frequent psychiatric diagnoses were addiction disorders, schizophrenia, major depressive disorder, posttraumatic stress disorder (PTSD), and cyclic mood disorders (schizoaffective and bipolar disorders). These constituted 90% of primary diagnoses. Two hundred eighteen (48%) of the 454 patients with these primary diagnoses had another active axis I diagnosis, and 73 (16%) had two or more additional diagnoses.
Seventy-eight patients (16%) died during the follow-up period. Nonsurvivors (mean age=54.7, SD=13.4) were older than survivors (mean=45.8, SD=10.2) (z=54, p<0.001, Mann-Whitney U test). Nonsurvivors were not statistically more likely to be male (N=76 [97%]) or unmarried (N=53 [68%]). Cause of death could not be determined for 15 (19%). Causes of death included myocardial infarction (N=15), alcohol complications (including gastrointestinal bleed and cirrhosis) (N=12), cancer (lung, esophagus, or bladder) (N=11), dementia of the Alzheimer type (N=8), and HIV infection (N=6). Other causes included renal failure (N=4), cerebrovascular accident (N=4), and automobile accidents (N=3). The medical examiner confirmed suicide in only two of 10 suspected cases. A significant effect of primary psychiatric diagnosis (χ2=29.5, df=15, N=504, p<0.05) was found: death rates ranged from 57% (N=8) for the 14 patients with dementia of the Alzheimer type to 7% (N=5) for the 69 patients with cocaine addiction. This finding appears strongly influenced by the age of the patients studied: those with dementia of the Alzheimer type had a mean age of 75.8 (SD=6.4), compared with a mean age of 41.4 (SD=6.5) for those with cocaine addiction.
The influence of psychiatric diagnosis on subsequent psychiatric hospitalization was evaluated. The four most prevalent diagnoses not related to addiction were schizophrenia (N=98 [19%]), major depressive disorder (N=111 [22%]), cyclic mood disorder (N=62 [12%]), and PTSD (N=74 [15%]). No other nonaddiction diagnosis accounted for more than 5%. Significant effects of diagnosis were found on hospitalization for schizophrenia and cyclic mood disorder. Seventy-nine percent (N=77) of the patients with schizophrenia were hospitalized at least once during the follow-up period (χ2=32.0, df=1, N=98, p<0.001), compared with 66% (N=41) of the patients with cyclic mood disorder (χ2=6.45, df=1, N=62, p<0.05). Diagnosis of major depressive disorder or PTSD did not increase the likelihood of hospitalization: 46% (N=51) of the patients with major depressive disorder (χ2=0.730, df=1, N=111, p>0.3) and 58% (N=43) of the patients with PTSD (χ2=1.94, df=1, N=74, p>0.1) were hospitalized during the study period.
Table 1 summarizes psychiatric hospitalization rates during the 5-year follow-up period. Significantly higher hospitalization rates for each diagnosis were found when comorbid diagnoses existed. Substantially different psychiatric hospitalization rates were found between patients with comorbid diagnoses and those with single primary diagnoses.
The effect of addiction disorders was examined by subdividing the subjects into five tiered groups: 1) no addiction, 2) addiction with no comorbid axis I diagnosis, 3) addiction as primary diagnosis with another axis I diagnosis, 4) addiction as secondary diagnosis, and 5) polysubstance dependence (three or more active addiction diagnoses). Addiction diagnoses (primary or secondary) were found in 295 (59%) of the patients; 200 (68%) of these patients were hospitalized during the follow-up period (χ
2=37.37, df=1, p<0.001). The 23 patients with polysubstance dependence had the highest rate of psychiatric hospitalization: 19 (83%) of these patients were hospitalized at some point during the 5-year study period (χ
2=9.78, df=1, p<0.005). Level of substance dependence had a significant effect on the number of hospitalizations (
Table 1).
Discussion
These results demonstrate the impact of psychiatric comorbidity on rates of psychiatric hospitalization. Rates of psychiatric hospitalization were higher for patients with comorbid psychiatric disorders, especially for patients with addiction problems comorbid with another condition.
Comorbidity between psychiatric and addiction disorders has been reported in 33% to 60% of psychiatric patients
(1,
2). Patients with severe mental illnesses and comorbid addiction disorders use significantly more hospital resources than those without comorbid addiction disorders
(3,
4). A VA study evaluating 16,066 patients
(5) demonstrated that psychiatric rehospitalization rates increased in direct relationship to the degree of addiction comorbidity. Studies from different settings
(6–
8) have found that up to half of psychiatric patients seen in emergency rooms have comorbid addiction disorders.
Limiting data collection to evening hours possibly influenced the severity of the patients’ disorders. Friends or family may be unable to bring seriously ill patients to the emergency room until after work. Fewer social support services exist in the evening to avert emergency room visits. Although some patients with less serious issues may avoid the emergency room after hours, diagnostic and demographic characteristics of this study group closely resemble those of patients hospitalized on the acute psychiatry unit around the same time
(2). Other limitations of this study include the lack of standardized diagnostic interviews, verification of interrater reliability of diagnoses, or data regarding outpatient follow-up appointment frequency, intensity, or adherence.
The 15% death rate is likely an underestimate because VA records were the only source used to identify deaths. Patients dying in other counties, out of state, or whose families did not notify the VA were likely missed. Excess mortality in this relatively young group of patients may have been the result of serious medical complications of alcohol, drug, and tobacco dependence. A study of deaths in VA psychiatric outpatients over an identical 5-year assessment period
(9) found similar causes of death, except for a higher percentage of suicides (6.9%), compared with the rate of less than 1% in this study. Although suicides were likely undercounted, changes made to the treatment system during the follow-up period, including full implementation of the continuity-of-care team model, expanded intensive case management, and improved access to dual-diagnoses addiction treatment, may have lowered the suicide rate (10). Additional measures appear indicated to improve preventive medical care and discourage health risk behaviors.