Psychiatric patients have high rates of medical comorbidity (1–4). This may be related to disease characteristics or lifestyle choices, or may be secondary to medication. Medical care for psychiatric patients, particularly the severely and persistently mentally ill, may be suboptimal for a variety of reasons (5–7).

Psychiatry residency programs need to ensure that graduating trainees are able to identify medical conditions appropriately. This may be increasingly difficult as medical care continues to become more complex and medical rotation requirements remain at a 4-month minimum. An examination of residents’ medical sophistication is important. As future practicing psychiatrists, residents will need to be able to inform their patients of what to expect when they are medically evaluated. In this context, our hypothesis is that psychiatry residents’ assessment of what is medically required when problems are identified may be a marker of the residents’ medical knowledge.

The emergency psychiatric service (EPS) may be an ideal environment to examine this question. Patients presenting for evaluation in the EPS are unknown to the residents. They may have known or unknown, active or stable medical issues. Substance intoxication or withdrawal may further complicate the picture. Examining psychiatry residents’ medical decision-making in this type of difficult situation might yield important information.

The study was performed in an EPS of an urban teaching hospital that serves as the city’s major 24-hour psychiatric emergency department. Psychiatry residents staff the EPS around the clock and are supervised by on-site psychiatry attendings for 16 hours daily and by phone overnight. Patients who present to the emergency department are triaged by a nurse to go either to the EPS or the medical emergency room. Patients who are evaluated in the EPS and have a medical problem can be sent to the medical emergency department for evaluation or treatment.

Psychiatry residents were asked to complete a survey sheet for every patient referred for medical evaluation from the EPS. We collected information regarding demographics, the level of subjective concern, and the anticipated level of medical intervention. Subsequent record review was performed to determine the actual level of medical intervention performed. The psychiatry residents knew of the design and purpose of the study but emergency medical physicians were blind to both. The scales utilized for levels of concern and medical intervention are presented in Appendix 1.

We collected data for 6 weeks. Statistical analysis utilized simple regression and a test for proportional data (8).

During the 6-week period of the study, EPS referred 27 patients to the medical emergency department for medical needs. Twenty-three (87.2%) of the records were completed accurately. The average age of these patients was 45.0 (SD=13.4) years. There were 17 men and six women. Five patients suffered from a mood disturbance (three from bipolar disorder, two from depression), two had schizophrenia, eight had primary substance-related disorders, five had other disorders (psychosis not otherwise specified, dementia, delirium), and three never received a psychiatric diagnosis because they were admitted to a medical unit.

There was a weak correlation between the residents’ subjective level of concern and the intervention in the medical emergency department (r²=0.11, F=2.69, p=0.12). However, there was a significant relationship between the residents’ anticipated level of intervention and the intervention actually carried out in the medical emergency department (r²=0.24, F=6.67, p=0.017). The residents accurately predicted the level of intervention in 10 cases (43.5%). They overestimated the required intervention in six cases (26.1%), and underestimated the required intervention in seven cases (30.4%). Since underestimation of medical needs is potentially dangerous, this category was compared with the others. The difference between the rates of accurate prediction and underestimation of medical need was not significant (z=0.92 nsec). However, when overestimation and accurate anticipation of medical intervention are combined (16/23), the difference was significant (z=2.6, p<0.01). In four cases (17.4%), the residents were ≥2 points off the actual intervention (two overestimation, two underestimation).

The Emergency Medical Treatment and Active Labor Act (EMTALA) of 1985 requires that all patients who seek emergency treatment be given an adequate medical screening examination (9). When there is a separate EPS, the medical screening must be performed by the examining psychiatrist. Significant medical comorbidity is present in approximately one-third of patients being evaluated for psychiatric emergencies (10, 11).

Residents not infrequently identified medical problems in psychiatric referrals that were missed by the emergency department medical triage nurse (27 patients out of total approximately 600 patients, or 4.5%). In five cases (21.7%), the patients were ultimately admitted to a medical or intensive care unit. In the majority of situations (69.6%), psychiatry residents either accurately predicted or overestimated the level of medical intervention required. They underestimated the level of intervention in 30.4% of the cases. In other words, psychiatry residents underpredicted the magnitude of medical intervention required nearly one-third of the time.

There are clear limitations to this study. Since we did not follow all of the patients evaluated in the EPS, the study does not address whether psychiatry residents might miss medical problems in their acutely ill psychiatric patients. The study examined agreement with intervention carried out by emergency department physicians. The design implies that medical emergency physicians are the gold standard. This is not the case. Immediately prior to initiation of this study, a resident referred a patient to the emergency department for abdominal pain. After infusion of some intravenous fluids, the patient was sent back. Dissatisfied with the workup conducted in the emergency department, the resident undertook her own workup and discovered acute pancreatitis. The patient was admitted to a medical unit.

Psychiatrists and medical emergency physicians in general have similar medical screening procedures, although psychiatrists tend to order more laboratory tests (12). The converse situation, the level of agreement on psychiatric disposition, provides a potential comparative marker for the current study design. The level of agreement regarding psychiatric disposition is generally low, with psychiatrists and medical emergency physicians disagreeing nearly half the time (13). This is probably due to the fact that emergency physicians undergo very little psychiatry training (14).

This is the first study to examine medical decision-making in psychiatry residents. We found that psychiatry residents not infrequently are able to predict the level of medical intervention eventually performed by the emergency department physicians. However, in nearly one-third of patients, psychiatry residents underestimate the level of medical intervention required. Additional research is required to determine whether this is an indicator that additional medical training is required, or whether this is a consequence of the variability in how different physicians address medical problems.