Increases in neuropsychiatric knowledge have dramatically impacted the practice of clinical psychiatry (+1—+4), leading educators to question whether psychiatric residency training programs adequately prepare their trainees to deal with these changes (+5,+6). The increased emphasis on neurosciences is reminiscent of a time when neurology and psychiatry were highly related disciplines. During the late 1920s and early 1930s, when the nascent national examining bodies were forming, the majority of physicians who passed the first examination were board-certified in both specialties (+3,+7—+9). After World War II, both neurology and psychiatry witnessed a gradual separation, such that by 1970 the American Board of Psychiatry and Neurology (ABPN) temporarily dropped the internship year requirement for psychiatric residents, and by 1981 neurology was completely eliminated from Part II of the board exam. This legacy of separation between the two fields still pervades psychiatric training programs, perhaps negatively impacting on residents' abilities to diagnose and treat neurological and neuropsychiatric disorders (+10,+11).

At the University of Michigan Department of Psychiatry, we currently provide 2 months of a neurology rotation (both inpatient and outpatient) in the first year of training as recommended by the Residency Review Committee (RRC). We wanted to examine whether this type of training ensured both good performance on standardized testing, as well as a high degree of resident self-confidence in the diagnosis and treatment of neurological and neuropsychiatric disorders. Our concern is that although the practice of psychiatry has changed rapidly over the last several decades, residency programs still devote only 2 months during the first year to neurology training. This at a time when psychiatrists increasingly need a better command of neurology to assess and treat patients with neurological and neuropsychiatric disorders. Residents also need a solid understanding of neuroanatomy, neuroendocrinology, and the functioning of parts of the brain especially important to psychiatry, like the limbic system, to keep pace with contemporary psychiatric research. An adequate foundation in neurology is essential to the ability to practice contemporary psychiatry with its increasing neuropsychiatric emphasis (+12).

To assess both the changes in self-confidence and the ability to diagnose and treat neurological disorders, we compared ratings of clinical confidence and the Psychiatry Resident In-Training Examination (PRITE) scores of our residents throughout their training. We predicted that by the time residents were ready to graduate, their confidence in their neurologic skills would decline, since they would not be actively participating in the treatment of these disorders and would no longer have consistent exposure to this material in their curricula. We also predicted that working with psychiatric patients would not prevent this process. Likewise, we expected to see a decline in residents' performance in the neurological subsection of their PRITE scores over time.

We anonymously surveyed first- through fourth-year psychiatry residents at the University of Michigan during the 1996—1997 academic year by using a modified previously described questionnaire (+13) to examine levels of neurological, psychiatric, and neuropsychiatric proficiency for psychiatrists. An investigator brought surveys to resident meetings, where prospective respondents were informed that a study on neurology education was being conducted, and their participation was requested, but voluntary. We further explained that their decision to participate would not affect their education or evaluation and that their responses would be confidential and anonymous. If residents declined to participate, they were not reapproached. A total of 32 residents from the first through fourth year of residency were assessed, for an 84% response rate. Only two of the eight first-year residents had neurology training at the time of the survey. The other six began either a 6-month block of psychiatric training or other intern year requirements in medicine, pediatrics, or family practice. These two groups of first-year residents were analyzed separately. Residents in the second through fourth year of training all had their required 2-month rotation in neurology.

Respondents evaluated their own ability to diagnose and treat 24 conditions. We separated psychiatric, neuropsychiatric, and neurological proficiencies based on assumed levels of exposure throughout the residency. For instance, neurological disorders included the seizure disorders and Parkinson's disease. Neuropsychiatric disorders included Alzheimer's disease and depression attributable to central nervous system disease, whereas psychiatric disorders included schizophrenia, acute extrapyramidal symptoms (EPS), and tardive dyskinesia (TD). Traditionally, EPS and TD are designated to the neurological or neuropsychiatric realm; however, we categorized them as psychiatric proficiencies since residents routinely work with these diagnoses on general psychiatric rotations.

These data were grouped by year of training, then analyzed by one-factor (simple) analysis of variance (ANOVA), regression analysis, and two-way ANOVA, with the year of training and the type of self-confidence (psychiatric, neuropsychiatric, and neurological) as factors.

First-year residents with neurology training were included in analyses for neurological and neuropsychiatric proficiency, whereas first-year residents without neurology training were only included in the psychiatric and neuropsychiatric groups.

To assess standardized test performance, we anonymously examined percentile PRITE scores for 39 residents (from the academic years 1988 through 1995), as they tested consecutively in their second, third, and fourth years of our training program. Scores for first-year residents were not available during that period. We analyzed the neurology and psychiatry subsections of the PRITE and performed ANOVA, repeated measures, and regression analysis to test for statistically significant changes from year to year.

We found a significant difference (simple ANOVA F=5.2, df=3, P<0.01) in confidence levels for diagnosing and treating neurological disorders among psychiatric residents in different years of residency training, with first-year residents reporting the highest level, and the fourth-year class reporting the lowest level. Furthermore, there was a significant positive correlation between the proximity of neurology training and self-confidence. Residents who most recently completed their neurology clerkships felt the best able to diagnose and treat neurological disorders, followed by a significant decay in confidence for each successive year (R=0.60, P<0.003; see +Figure 1).

Self-confidence in the treatment of psychiatric disorders showed the opposite trend, with first-year residents scoring significantly lower than upper level residents (F=5.2, df=3, P<0.01). There was no decrease in confidence in later years of training; in fact, there was a weak (trend level) positive correlation between further residency training and psychiatric self-confidence (R=0.34, P<0.07). Self-confidence in the management of neuropsychiatric conditions revealed improvement with continued training at the trend level (F=2.4, df=4, P=0.07).

Two-way ANOVA showed a significant main effect for both factors: the year of training (F=9.3, df=2,59, P=0.0003) and the type of the self-confidence (neurological, neuropsychiatric, or psychiatric: F=55.0, df=2,59, P<0.0001). Post hoc comparison attributed the significance in the year of training factor only to the change in self-confidence to diagnose and treat neurological conditions between the second- and fourth-year residents (Sheffe, P<0.02). As expected, the significant effect for the type of self-confidence factor was due to higher scores in the management of psychiatric disorders, followed by the neuropsychiatric disorders. The lowest self-confidence was seen for the neurological disorders. Neuropsychiatric and psychiatric self-confidence levels did not change significantly from year to year (see +Figure 2). There was no interaction effect (F=1.4, df=4,59, P=0.25).

Analysis of PRITE scores by repeated measures ANOVA (using year of training as a repeated measure) revealed that the neurology subsection scores significantly improved with training (F=3.8, df=3, P<0.03). We found a weak positive correlation between neurology subsection PRITE scores and the year of training (R=0.16, P<0.09). There were significant increases in test scores in the psychiatry subsection of PRITE as residents progressed from Year 2 to Year 4 (F=52.8, df=2, P=0.0001), with strong positive correlation between years of training and PRITE scores (R=0.45, P<0.0001) (See +Figure 3).

Assessment of self-confidence in the treatment of neurological disorders revealed that residents who were closest to their clinical rotation in neurology rated themselves the most competent. There was a gradual decay in this level of confidence, with the residents closest to graduation scoring themselves at the lowest levels for the entire group. This is consistent with our prediction that decreased participation in the treatment of neurological disorders over the course of psychiatric training would lead to an erosion of self-confidence in this area. As expected, residents' self-assessed ability to treat psychiatric disorders increased significantly during training. However, self-confidence in treating neuropsychiatric disorders improved only at the trend level with further training.

Interestingly, a different pattern was observed in the PRITE scores throughout training. Psychiatry subsection scores showed the expected improvement with experience; however, PRITE neurology scores also significantly improved. This finding suggests that while residents might be accumulating more neurological didactic knowledge, they feel less confident applying these skills in an actual clinical setting. This effect may be due to the length of time since those diagnostic and treatment skills were last used in the clinical setting (+14). In contrast, the improvement in PRITE scores may be due to increased study by fourth-year residents as they are closer to graduation and Part I of their ABPN exam.

The disparity in PRITE scores, compared with residents' survey responses, raises several interesting issues. One possible explanation is that PRITE may be a less accurate instrument in assessing residents' ability to diagnose and treat neurological problems. This assumption is supported by an earlier study by Webb et al., in which they demonstrated only a low correlation between neurology scores on the PRITE and scores on the ABPN Part I exam (+15). Therefore, using PRITE as the sole monitor of neurological ability may be inadequate. We may need other outcomes measures to ensure we are educating our residents thoroughly in this area (+16). Alternatively, some of the difference between our PRITE data and resident self-assessment scores may be accounted for by greater proficiency taking standardized tests with each passing year. This finding may explain in part why residents' scores increased in both psychiatry and neurology on the PRITE, whereas their actual confidence level in clinical neurology decreased. Finally, fourth-year residents on the verge of graduation may have a greater perspective on the entire field of psychiatry. This factor, in combination with anxiety around their upcoming transition into a practice setting with less supervision, might result in a more conservative estimate of their abilities, augmenting the disparity between neurology self-assessment scores and PRITE scores in the neurology subsection.

Certain limitations should be considered while attempting to generalize from this set of data. For example, we examined different groups of residents in our analysis of the self-confidence data and PRITE scores. While we believe that these are two samples representing the same population, this heterogeneity could also contribute to the obtained results. Furthermore, there was only partial correspondence between the definitions of neurology used in our survey and in the PRITE exams, which could have contributed to the disparity between PRITE scores and self-confidence ratings. Since our data does not address the validity of these definitions, this issue requires further study. Our study also examined residents within one department; replicating these findings in other departments would strengthen our conclusions. Finally, 16% of residents did not respond to the questionnaire, and while this subpopulation is in good standing academically, and are not different from the rest of their colleagues on any meaningful measure, one cannot completely exclude primary selection bias.

Given the results of this study, we question whether additional training, especially in the form of a clinical neurology rotation, should be required as a "refresher" sometime prior to graduation. Current RRC requirements stipulate "the training in neurology should provide sufficient didactic and clinical experience to develop expertise in the diagnosis of those neurological conditions that might reasonably be expected to be encountered in psychiatric practice and that must be considered in the differential diagnosis of psychiatric conditions [emphasis ours]" (+17). Given the depth and complexity of information in both fields, we may need to reevaluate the timing or duration of the clinical experience in neurology (+3). Conversely, while the aforementioned requirements listed are well-intentioned, they may be too ambitious for a 4-year residency, given the exponential increase in information in both psychiatry and neurology. In addition, more time spent in neurology training may further erode time needed for adequate psychotherapy training. Other educators have voiced similar concerns (+18,+19). Residency directors would be better guided in improving their training programs if the aforementioned criteria are better specified, and the scope of expected "expertise" clarified.

The authors thank James Meador-Woodruff, M.D., for his assistance with this paper.