This study found that a significant percentage of internal medicine residents were interested in pursuing sub-specialty fellowship. A typical internal medicine resident with fellowship interest at the institution surveyed was more likely to be male, international medical graduate, aged 29 or less with no prior formal clinical training prior to residency and with interests in Cardiology, Pulmonology or Hematology/Oncology.
The authors hypothesized that internal medicine residents would devote much study and research time to their subspecialty of interest in order to remain competitive for fellowship applications. This observation is supported by two findings in this study. Firstly, the finding that the need to be competitive was a major and significant factor that influenced the study patterns of IM residents with fellowship interest compared to their counterparts without fellowship interest; Secondly, the finding that a majority (75%) of IM residents with fellowship interest had spent at least 2-3 months rotating through their sub-specialty interest areas.
Data relating to the associations between fellowship interest and medical knowledge of internal medicine residents is lacking. Our working hypothesis prior to the conduct of this study was that based on the need to be competitive in their chosen fields, leading to proportionately less study of general internal medicine concepts, residents with fellowship interest would have lesser knowledge of general internal medicine.
A number of factors could potentially confound this hypothesis, including 1) Time-length since graduating from medical school and 2) Training in internal medicine (or other specialty) prior to current residency. A resident with fellowship interest could have lower IM-ITE scores, not because of his interest in fellowship training but because he had lost touch with academic medicine due to a long interval between graduation and commencement of residency training. Likewise, a resident without fellowship interest could have higher scores due to knowledge acquired from prior training in internal medicine.
Even though there was no significant difference in the groups with respect to these potential confounders, in multivariable analysis, they (in addition to gender and CMS) significantly confounded the relationship between fellowship interest and objective general medical knowledge. From our analysis of potential confounders, males had higher scores than females; International Medical Graduates (IMGs) had higher scores than graduates of US medical schools. Also residents with prior formal clinical training and higher number of elapsed years since graduating from medical school had higher scores than their counterparts without prior training and with lesser time-lengths since graduating. While the confounding effects of prior clinical training and number of years since graduation have real world plausibility, the confounding effect of gender is not easily explicable. With regards to Country of Medical School, IMGs are known to outperform U.S. graduates from assessments of IM-ITE scores up to the year 2000 [17
]. IM-ITE data from more recent years however is yet unpublished.
None of the confounding demographic variables had a direct relationship with the primary independent variable (fellowship interest). After adjusting for plausible confounders, Internal Medicine residents with fellowship interest had statistically significantly higher overall percentile scores.
The intriguing finding of greater scores among residents with fellowship interest, even after multivariable adjustment, does not support our original hypothesis. Are residents with fellowship interest more knowledgeable than their counterparts at baseline? Or does the competitive drive to excel make them better as they progress through residency?
Our study is not without limitations. The interpretation and generalizability of our results is hindered by sample size and other limitations. Among the survey respondents, eight (8) had no interest in pursuing subspecialty fellowship training and quite a few subgroups derived from this control group include only one or two persons. This is especially important as it relates to certain confounding variables that were adjusted for. Our sample size may not have allowed for an effective evaluation of the differential effects of the identified confounding factors on the observed results. Our study was also limited to one cohort of each residential year of training in a single institution in which only 13% of trainees were graduates of US medical schools.
In trying to study factors that affect a resident's knowledge within a limited time frame defined by a 3-yr residency training format, further limited by our independent variable only being definable in the second year of residency training, our study was designed as a cross-sectional study. This is not without its own limitations. If possible, a prospective study with repetitive measurements of objective examination knowledge would provide more valid results. This is important as it is possible that residents who have less general knowledge and are not planning on a fellowship do not yet do so because they still lack the necessary prerequisite general knowledge
In view of these limitations, the demonstrated associations should not be interpreted in a causal context as having established a causative relationship between career intentions and level of knowledge (or vice versa). Despite the single study site and sample size limitations, the consistency of the findings and the lack of statistically significant differences in the groups with respect to factors that may have biased general medical knowledge suggest that this early finding may be indicative of a more general phenomenon in IM training.
There is an ongoing debate over whether IM residency education needs to be redesigned to respond to changes in medical practice and health care delivery. Some recommend that IM training be lengthened to 4 years so that residents can fully grasp the depth and width of current knowledge. Others suggest shortening the training to 2 years so that residents can quickly advance to subspecialty fellowship training. The debate needs to be better informed by empirical data, and our study represents an initial attempt to provide such data.