There is a discrepancy between societal need for general internal medicine specialists and internal medicine residents’ choice in the R4 subspecialty match [1–7]. There has been a decline in internal medicine residents’ interest in general internal medicine in the United States with 54% planning to practice general internal medicine in 1998 [8] compared with 22% from survey data collected from 2008 to 2011 [9]. There are similar concerns in South Asia [4] and in several European countries [5–7]. In Canada, internal medicine residency training begins with three years of core internal medicine, and then all residents apply and match (known as the R4 match) to a subspecialty residency with general internal medicine as one of those options. Between 2008 and 2010, 21-24% of internal medicine residents pursued general internal medicine based on Canadian Post M.D. Education Registry (CAPER) data [10]. Concurrently, Canadian physician resource studies predicted 400–500 general internists would retire between 2010 and 2013 with the number of new graduates only meeting half of this demand, let alone the current deficit in many areas of the country and the expected population growth [11].

There have been several studies published examining demographic and non-demographic aspects associated with internal medicine residents’ choice of subspecialty [8, 12–21]. However, few have looked at the relative importance of personal goals and their impact on the internal medicine R4 subspecialty match [8, 14, 16]. These studies examined personal goals based on career path with participants being surveyed with either dichotomous (yes or no) items [8] or five-point likert scales [14, 16]. Although these approaches can answer the question, “Does this group value lifestyle?”, they may not be the ideal methods to answer the question “How high of a priority is lifestyle for this group as compared to others?” as there is nothing that forces respondents to prioritize the items they answer.

The purpose of this study was to assess if internal medicine residents can be grouped based on their personal goals and how each group prioritizes these goals compared to each other. A secondary objective was to explore whether we could predict a resident’s desired subspecialty choice based on their constellation of personal goals.

In 2009, 43 residents participated with an additional 16 in 2010 (total 59) out of a possible 114 residents. Thirteen Q sorts were excluded because of either missing or duplicate statements resulting in 46 Q sorts suitable for analysis. Residents were primarily 26 to 31 years old (63% of participants), with 16% younger than 26 years old and 21% older than 31 years old. The majority (63%) were female, and 42% were in PGY-1, 28% in PGY-2 and 30% in PGY-3. Nine percent of participants did not have a premedical degree, 63% had a bachelor’s degree and 28% had a master’s degree. Participants desired choice of subspecialties is shown in Figure 2.

The key statements that define and/or distinguish between these factors are breadth of practice, lifestyle, desire to do procedures, length of training, and future income potential. Three statements with the most agreement between factors are as follows: “I’d like to do a combination of inpatient and outpatient” (moderate agreement from the four factors); “My decision was predominantly based on experiences as a medical student” (mild disagreement from the four factors); and “Prestige and respect from colleagues matters to me” (scored as neutral by all four factors). The three statements with the most disagreement between factors were: “I am willing to do as many years as it takes to get the job and subspecialty I want” (moderate agreement by factor 3 but disagreement from all other factors); “I like a little bit of everything and variety” (strong agreement from factor 1, mild agreement or neutral from other factors); and “A two-year GIM fellowship is a deterrent” (mild agreement from factor 2 but disagreement from other factors).

The purpose of this study was to evaluate if we could group residents based on how they prioritized their personal goals for their career, and we were able to group 35 of 46 residents (76%) to one of four groups. Using these personal goals groups, we had moderate agreement (kappa = 0.57) between our predicted subspecialty group and residents’ self-reported first desired subspecialty, something that we have not seen in our review of the literature.

The main advantage of Q methodology is that it forces a quasi-normal distribution so respondents cannot select the same response for every item, as they can with traditional survey methods. Horn and colleagues’ [14] Canadian survey study grouped their respondents based on subspecialties (procedural, non-procedural, and non-procedural with declining interest) and found that all three groups valued diversity of clinical spectrum. We found that only one group valued variety compared to other groups. Similarly, Horn and colleagues [14] found their three groups all valued satisfaction among staff physicians whereas only one of our groups valued positive interactions with attending physicians. Although we are not comparing the exact same statements, the differences between these two studies are likely attributable to the methodology, with Q methodology allowing better differentiation between groups.

Examining the four groups of residents (see Tables 3, 4, 5 and 6), there are some key findings. It has been established in medical students that interest in a specialty correlates strongly with future income [28]. In internal medicine, a common finding across studies is that subspecialties such as cardiology, gastroenterology, hospitalists, nephrology, and respirology value a higher remuneration more than other subspecialties. We found that the group of residents interested in gastroenterology, nephrology and respirology valued remuneration more than others, but this was not the driver for those interested in cardiology or general internal medicine (which in Canada is somewhat similar to a hospitalist in the United States). The primary driver for those interested in cardiology and critical care was willingness to extend training, which is consistent with our anecdotal experience of these residents often doing seven, eight or sometimes more years of residency before ending training.

In our previous qualitative study [19], we often heard residents referring to general internal medicine as a dumping ground for patients other subspecialties did not want. In this study, it would appear that this sentiment is only a strong motivator for those interested in cardiology and critical care and was not a major concern for the other groups. In Canada, we have recently moved from a one to two-year general internal medicine fellowship after three core years of internal medicine. Our previous study indicated this increased length of training may be a deterrent to pursuing general internal medicine, and we have now found that this applies mainly to those interested in gastroenterology, nephrology and respirology.

In the current study, it was hard to tease out the non-procedural lower income subspecialties from each other, which may be due to a smaller sample size. However, this may also be due to a similar set of goals with these residents: they do not want to do procedures, are not as concerned with income, and want a focused area of expertise. What may distinguish them from each other is the specific patients whom they are interested seeing and the positive and negative interactions the residents have had with attending physicians.

Because this study was done with Canadian residents, it is worth noting the similarities and differences in the practice of general internal medicine in Canada compared to other countries. Ghali and colleagues [29] reviewed the clinical profile of general internal medicine in Argentina, Australia, Canada, Japan, New Zealand, Switzerland, and the United States. General internal medicine in Canada is a consultative service to primary care practitioners and is often hospital based caring for complex, multisystem patients. This overall profile is similar to that of general internists in Australia and New Zealand, and somewhat similar to general internists in Switzerland. This differs from general internists in the United States, Argentina and Japan who often have a large outpatient primary care role, though many also have consultative and hospitalist roles.

That said, one of the most interesting findings of our study is why a resident desires a career in general internal medicine. It is not surprising this group enjoys variety but what is surprising is this is essentially the sole driver for choosing general internal medicine over other subspecialties as exhibited by the extremely high Z-score of 2.71. A past president of the Canadian Society of Internal Medicine may have put it best when he said, “To my mind, this clinical smorgasbord is the best advertisement for doing general internal medicine” [30]. Regardless of the differences in the clinical profile of general internal medicine globally, this thirst for variety as a driver for residents to pursue general internal medicine is likely to transcend geographical borders.

Although previous aspects identified in the literature are important, this study suggests that residents fall into four groups based on a constellation of personal goals when choosing an internal medicine subspecialty. The key goals that define and/or distinguish between these groups are breadth of practice, lifestyle, desire to do procedures, length of training, and future income potential. Using these personal goals groups, we were able to predict residents’ first subspecialty choice/group with moderate success.