Understanding characteristics of internal medicine residents matching into pulmonary critical care medicine fellowships

Abstract

Background

Internal medicine (IM) residents face significant challenges when pursing subspecialty fellowships. This study determined the factors that differentiate IM residents entering pulmonary and critical care medicine (PCCM) fellowships from those pursuing other careers.

Methods

We completed a retrospective study of 12 classes of IM residents at a single institution completing residency between 2010 and 2021. Data included pre-residency characteristics, global residency performance, and PCCM-specific experiences. Logistic regression models examined associations between these variables and the primary outcome of matching into a PCCM fellowship within one year of completing IM residency.

Results

Among 522 residents, 10.3% matched into PCCM. Completing a pulmonary elective significantly increased the odds of matching into PCCM (OR 7.78, 99% CI 3.10–19.53, p < 0.0001). Residents who match into PCCM were more likely to have < 3 publications than 3 + (OR 3.51 (1.20–10.25), p = 0.003).” A stated intent to enter PCCM was positively associated with matching into PCCM in the univariable, but not the multivariable, model.

Conclusions

Matriculating into PCCM fellowship was significantly associated with completing a pulmonary elective during residency. PCCM-bound residents were less likely to achieve high numbers of publications, suggesting these residents’ preferences for clinical learning and practice over scholarship. This study provides insights into characteristics of residents who match into PCCM and guides mentors as they counsel residents considering PCCM.

Introduction

Approximately 6,000 internal medicine (IM) residents apply for a subspecialty fellowship yearly. With 5,000 fellowship positions available, one in six will not match.[1] Residents often develop career interests before residency training,[2–4] and early exposure to specialty rotations may further impact their career decisions.[5, 6] Alternatively, some residents’ career choices may evolve later in training.[7] IM residents who pursue fellowship training may have higher general medical knowledge than those who do not[8]; however, some competitive subspecialties may require early specialization leading to decreased general medical knowledge.[9] Scholastic accomplishments such as Alpha-Omega-Alpha (AOA) membership, scholarship, and class rank have been shown to predict future performance in training;[10–16] yet, resident publications may be a poor predictor of fellowship publications.[17] Recently, authors identified relationships between IM residents’ medical knowledge, early career intentions, rotation evaluations, and pre-residency characteristics and matriculation into cardiology fellowships.[9].

Pulmonary and Critical Care Medicine (PCCM) is the second largest IM fellowship in the US with 629 training positions. More than 80% of intensivists are IM-trained.[1, 18] This fellowship has become competitive with a 24.8% unmatched rate between 2004 and 2019.[19] As the US population ages, projections suggest a shortage of critical care providers in the US.[20–22] Trainees pursing PCCM must develop confidence leading multidisciplinary teams during stressful situations, demonstrate empathy with distressed patients and family members, and learn psychomotor skills for numerous bedside procedures.[23] Considering the growing importance of work-life balance on IM resident career selection,[24] residents may feel disinclined to pursue careers in PCCM.[25].

No studies have demonstrated how residents matching into PCCM fellowships differ from their peers. Furthermore, understanding unique characteristics of residents who match into PCCM should help mentor residents with interests in PCCM, assist with attracting residents into PCCM, and alert PCCM fellowship program directors regarding features of successful applicants. This study sought to compare IM residents entering PCCM fellowships with other IM residents based on standardized measures of performance, stated career interests on electronic residency application service (ERAS) personal statements, and exposures to PCCM rotations.

Methods

This study’s methods reflect a modification of a prior study from our group focused on internal medicine residents entering cardiology fellowships. [9] While methodological approaches of the current and prior studies are similar, we have revised the methods section to reflect changes relevant to the current study.

This study included residents who matched to the Rochester, Minnesota Mayo Clinic categorical IM residency from 2007 to 2018; thus completing residency between 2010 and 2021. Those who completed residency in under 3 years or left the program were excluded. The primary outcome was matching into PCCM fellowship within a year of completing residency to account for those who completed a chief medical resident year and those who delayed fellowship match for 1 year. Residents who matched into PCCM were compared to all other graduating residents.

Performance, career intent, and PCCM exposure were compared with pre-residency variables, characteristics of PCCM experiences, and global performance markers during residency. Only information available by the time of fellowship applications was included. After 2011, the fellowship match moved from June of PGY2 year to December of PGY3 year. As such, residents entering between 2007–2009 had data through the first half of PGY2; residents starting after 2009 had data included through the end of PGY2.

Pre-residency variables included declaration of PCCM intent, United States Medical Licensing Examination (USMLE) Step 1 and Step 2 Clinical Knowledge scores, pre-residency PubMed® publications (noting first author), AOA membership, and US News & World Report (USNWR) medical school research ranking.[26] Study team members (MC, TB, DK) reviewed personal statements for declaration of a PCCM subspecialty intent.

Data on resident PCCM experiences included previously validated faculty assessments for required PCCM rotations.[27, 28] We differentiated the timing of a resident’s first PCCM experience into the first versus second half of the PGY1 year, and assessed the choice to complete an elective pulmonary rotation. Similar to other evaluations, PCCM evaluation scores were deemed “highly professional” if in the top 20th percentile of their class. [8, 20] The percentile score on PCCM-specific content areas of the ITE was included as a measure of PCCM knowledge. Our analysis incorporated data from any PCCM rotations completed by the time of fellowship application for each resident in the study.

Residency clinical performance was assessed with rotation evaluations and mini-clinical evaluation exercise (mini-CEX) scores completed prior to fellowship application.[29, 30] Clinical evaluations were considered “highly professional” if in the top 20th percentile of their class.[31] In-training examination (ITE) was included as a marker of medical knowledge.[32, 33] These scored were restricted to the PGY2 year in order to ensure variance inflation factor (VIF) < 3 for all covariates in the multiple logistic regression model. The total number of PubMed®-indexed publications during residency served as a reflection of academic performance. This is the only variable with data across a resident’s entire residency rather than up to the time of fellowship application. Publications were analyzed as discrete variables with odds ratios calculated for zero publications vs. one, two, or greater than or equal to three publications during residency.

REDCap (Research Electronic Data Capture) data tools, hosted at Mayo Clinic, was used to collect and manage study data abstracted from residency application materials. [34] This is a secure, web-based application designed to support research study data capture. One author (AJH), who has no evaluative role within the residency, merged data. After merging data by resident, the data was deidentified prior to analysis to protect confidentiality.

Independent variables distribution was reported as mean (standard deviation) for continuous variables and n (%) for categorical variables. Relationships between independent variables and the binary primary outcome variable were assessed using logistic regression models. We visually examined functional form for continuous-valued covariates using Loess plots and objectively by Hosmer & Lemeshow goodness-of-fit tests, with those deviating from the assumption of linearity in the logit categorized by logical breakpoints. Potential multicollinearity among covariates was assessed using the VIF, with the highest VIF-valued covariate being excluded and re-assessing until all VIF < 3. A multivariable logistic regression model for the primary outcome adjusted for all modifiable covariates simultaneously. The threshold for statistical significance was set at α = 0.01. All analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC). The Mayo Clinic Institutional Review Board (IRB) and Internal Medicine Residency Education Group (IMREG) approved this study. The IRB and IMREG groups served as the ethics committees that waived the need for informed consent.

Results

Over this study period, 550 residents matched into the categorical program. Twenty-eight of them did not complete the program or graduated after only 2 years, leaving 522 residents for inclusion. Much of this population has been previously described.[9] We identified no significant differences in demographic or pre-residency characteristics between those included and excluded from our study (Table 1).

Table 1.

Demographics of residents and pre-residency attributes

	Eligible (N = 550)	Excluded (N= 28)	Included (N = 522)	P Value
Gender				> .99
Male	336 (61.1%)	17 (60.7%)	319 (61.1%)
Female	214 (38.9%)	11 (39.3%)	203 (38.9%)
Medical School Type	*152 unique		*147 unique	0.11
U.S. Public	329 (59.8%)	15 (53.6%)	314 (60.2%)
U.S. Private	180 (32.7%)	8 (28.6%)	180 (32.7%)
International	32 (5.8%)	4 (14.3%)	28 (5.4%)
Canadian	5 (0.9%)	1 (0.0%)	5 (1.0%)
Osteopathic	4 (0.7%)	2 (3.6%)	3 (0.6%)
Age,years	27.3 (2.7)	28.6 (3.4)	27.2 (2.6)	0.03
PCCM Career Intent				0.62
No	532 (96.7%)	27 (96.4%)	505 (96.7%)
Yes	18 (3.3%)	1 (3.6%)	17 (3.3%)
USNWR Top 50				> .99
No	351 (63.8%)	18 (64.3%)	333 (63.8%)
Yes	199 (36.2%)	10 (35.7%)	189 (36.2%)
AOA member				0.66
No	402 (73.1%)	22 (78.6%)	380 (72.8%)
Yes	148 (26.9%)	6 (21.4%)	142 (27.2%)
Publication				> .99
No	353 (64.2%)	18 (64.3%)	335 (64.2%)
Yes	197 (35.8%)	10 (35.7%)	187 (35.8%)
First Author				0.60
No	461 (83.8%)	25 (89.3%)	436 (83.5%)
Yes	89 (16.2%)	3 (10.7%)	86 (16.5%)
USMLE
Step 1	238.8 (15.7)	236.1 (17.3)	239.0 (15.6)	0.39
Step 2 CK	249.9 (14.8)	245.2 (14.6)	250.1 (14.8)	0.09
PCCM Pulmonary and Critical Care Medicine, USNWR United States News and World Report, AOA, Alpha Omega Alpha,USMLE United States Medical Licensing Examination

Summaries of the study variables, for the entire population and the PCCM-matched cohort, are seen in Table 2. Of the 522 included residents, 54 (10.3%) matched into PCCM. 187 (35.8%) residents had a publication at the time of residency application including 86 (16.5%) with a first author publication. One-hundred and eighty-nine (36.2%) graduated from a top 50 medical school as ranked by USNWR, and 142 (27.2%) were AOA members. Seventeen (3.3%) declared an intent to pursue PCCM in their residency application personal statement. Mean (SD) USMLE scores for Step 1 were 238.7 (15.6) and for Step 2 CK were 250.1 (14.8). Eighty-six (16.5%) of the residents studied completed a pulmonary elective rotation prior to fellowship application. During their residency, 204 (39.1%) authored ≥ 3 publications. The mean percentile (SD) on the PGY-2 ITE was 79.7 (18.3) overall and 62.4 (24.6) for the PCCM portion of the exam.

Table 2.

Logistic regression analyses

		(N = 522)	Matched to PCCM (N = 54)		Bivariate Logistic Regression		Multiple Logistic Regression
Pre-Residency Characteristics	N (%)	N (% of)	p value	OR (99% CI)	p value	OR (99% CI)	p value
Research publications, total	≥ 1	187 (35.8%)	17 (31.5%)	0.55	0.805 (0.364–1.1782)	0.48	1.134 (0.401–3.202)	0.76
	0	335 (64.2%)	37 (68.5%)		-		-
Research publications, 1st author	≥ 1	86 (16.5%)	4 (7.4%)	0.08	0.377 (0.095–1.489)	0.07	0.260 (0.048–1.403)	0.04
	0	436 (83.5%)	50 (92.6%)		-		-
PCCM stated intent	Yes	17 (3.3%)	6 (11.1%)	0.005	5.195 (1.328–20.330)	0.002	4.282 (0.851–21.551)	0.02
	No	505 (96.7%)	48 (88.9%)		-		-
USNWR Top 50 medical school	≤ 50	189 (36.2%)	21 (38.9%)	0.66	1.136 (0.531–2.431)	0.67	0.999 (0.386–2.587)	> .99
	> 50	333 (63.8%)	33 (61.1%)		-		-
AOA member	Yes	142 (27.2%)	11 (20.4%)	0.26	0.658 (0.265–1.635)	0.24	0.367 (0.119–1.136)	0.02
	No	380 (72.8%)	43 (79.6%)		-		-
		Mean (SD)	Mean PCCM vs. Not
USMLE Step 1 Score		238.7 (15.6)	239.1 vs. 239.0	0.96	1.000 (0.977–1.024)	0.96	0.998 (0.958–1.040)	0.89
USMLE Step 2 Clinical  Knowledge Score		250.1 (14.8)	251.1 vs. 250.0	0.57	1.005 (0.980–1.031)	0.60	1.008 (0.965–1.054)	0.63
PCCM Rotation Characteristics		N (%)	N (% of)
PCCM Elective	Yes	86 (16.5%)	26 (48.2%)	0001	6.314 (2.875–13.868)	< .0001	7.782 (3.102–19.525)	< .0001
	No	436 (83.5%)	28 (51.9%)		-		-
Timing of first PCCM experience	1st Half	254 (48.7%)	24 (44.4%)	0.57	0.828 (0.393–1.743)	0.51	0.675 (0.287–1.587)	0.24
	2nd Half	268 (51.3%)	30 (55.6%)		-		-
PCCM clinical evaluations top 20%	Yes	103 (19.7%)	12 (22.2%)	0.59	1.184 (0.484–2.897)	0.63	1.412 (0.464–4.292)	0.42
	No	419 (80.3%)	42 (77.8%)		-		-
		Mean (SD)	Mean PCCM vs. Not
PGY2 ITE PCCM percentile		62.4 (24.6)	67.8 vs. 61.8	0.06	1.010 (0.994–1.027)	0.09	1.012 (0.991–1.034)	0.14
Global Residency Performance		N (%)	N (% of)
Publications during residency	3 +	204 (39.1%)	9 (16.7%)	0.0007	0.257 (0.081–0.815)	0.003	0.236 (0.064–0.876)	0.003
	2	103 (19.7%)	18 (33.3%)		1.180 (0.433–3.216)		1.243 (0.388–3.983)
	1	123 (23.6%)	13 (24.1%)		0.658 (0.227–1.906)		0.509 (0.151–1.713)
	0	92 (17.6%)	14 (25.9%)		-		-
Mini-CEX evaluations top 20%	Yes	104 (19.9%)	7 (13.0%)	0.21	0.570 (0.193–1.684)	0.18	0.445 (0.121–1.629)	0.11
	No	418 (80.1%)	47 (87.0%)		-		-
Clinical evaluations top 20%	Yes	106 (20.3%)	12 (22.2%)	0.72	1.137 (0.465–2.779)	0.71	1.437 (0.450–4.591)	0.42
	No	416 (79.7%)	42 (77.8%)		-		-
		Mean (SD)	Mean PCCM vs. Not
PGY2 ITE overall percentile		79.7 (18.3)	80.1 vs. 79.7	0.85	1.001 (0.981–1.022)	0.86	0.996 (0.962–1.031)	0.75

PCCM Pulmonary and critical care medicine, USNWR United States News and World Report, AOA Alpha Omega Alpha, USMLE United States Medical Licensing Examination, PGY2 Post-graduate year 2, ITE In training exam CEX Clinical evaluation exercise

Bivariate logistic regression found a positive association with declaring PCCM in residency personal statements (p = 0.002); however, it was no longer significant in the multivariable model after accounting for all other modifiable independent variables (OR 4.28 (0.85–21.55); p = 0.02). Multivariable logistic regression found completion of a pulmonary elective rotation was significantly associated with matching into a PCCM fellowship (odds ratio (OR) 7.78, 99% confidence interval (CI) 3.10–19.53, p < 0.0001). PCCM matriculants were more likely to have < 3 publications than 3 + publications (OR 3.51 (1.20–10.25), p = 0.003).

Discussion

We endeavored to understand how residents who match in PCCM fellowship differ from their peers. Completing a pulmonary elective was significantly associated with matching into PCCM. Such PCCM electives are essential for residents advancing their skills in managing PCCM patients, meeting faculty members, and obtaining letters of recommendation. Additionally, it is likely that, in many cases, these elective rotations alter residents’ pre-existing career plans.

Matriculants into PCCM in this study were more likely to have < 3 publications than ≥ 3 publications. Our institution trains many residents pursuing cardiology, gastroenterology, or hematology/oncology fellowship training and we believe that these highly competitive groups are more likely to have ≥ 3 publications. Supporting this, a recent similar study at our institution found that twice as many residents matched into cardiology than PCCM and that those pursuing cardiology fellowship had more publications during residency than their peers.[9] In the current study, the average number of publications in the PCCM-matched group was 1.56, indicating that residents choosing PCCM are more focused on clinical skill enhancement than research. Furthermore, PCCM fellowships may place less emphasis on publications when ranking applicants for selection.

Adjusted analysis did not reveal significant associations between stated career intent in personal statements with matching into PCCM. Many programs may place lesser emphasis on the personal statement vis-a-vis selection decisions; however, this same association was significant in previous research and deserves further study in the field of PCCM. [9, 35, 36].

Strengths of this study include large sample size over a long timeframe and inclusion of both subjective and objective variables. Limitations include the single institution as a large academic medical center, which may reduce generalizability. That said, the study cohort included diverse residents from over 140 different medical schools. This study incorporated residents who trained before the Covid-19 pandemic, which potentially constrains conclusions about fellowship selection since the pandemic. Finally, this was a purely quantitative study; therefore, future qualitative research should examine reasons why residents choose to enter PCCM and other specialties.

To our knowledge, this is the first study on characteristics of residents who match into PCCM fellowship training. This study identified the importance of PCCM elective rotations during residency training on entering PCCM fellowships. While scholarship is an important feature among all fellowship applicants, this study suggests that residents entering PCCM training may place greater emphasis on clinical skills enhancement, and that PCCM program directors may focus more on other qualities, such as clinical experience. These findings should assist in mentoring residents who select careers in PCCM. These results can offer valuable guidance to trainees as they prepare for fellowship applications and prioritize their tasks. By providing general benchmarks, trainees may also be motivated to engage more actively in research by recognizing its differentiating value. Future research should further explore the potential value of personal statements and expressions of career intent, potentially through qualitative inquiry, to better understand the motivation and characteristics of residents who match into PCCM.

Conclusions

We found that completion of a pulmonary elective in residency was significantly associated with matching into PCCM. Additionally, PCCM-bound residents were less likely to achieve high numbers of publications suggesting a preference for clinical exposure over scholarship. This data may help provide insight into residents who match in PCCM and aid in mentoring these residents.

Abbreviations

IM

Internal Medicine

PCCM

Pulmonary and critical Care Medicine

AOA

Alpha omega alpha

ERAS

Electronic residency application service

USMLE

United States medical licensing examination

USNWR

United States news and world report

Mini-CEX

Mini clinical evaluation exercise

ITE

In training exam

ACGME

Accreditation counsel for graduate medical education

MICU

Medical Intensive Care Unit

REDCap

Research electronic data capture

VIF

Variance inflation factor

IRB

Institutional Review Board

IMREG

Internal Medicine Residency Education Group

Authors’ contributions

BS aided in design, data acquisition, and drafted the work. TB, MC, and DK aided in design and data acquisition. AH aided in design, data analysis, interpretation of data, and prepared Tables 1-2. All authors approved this submitted version and agreed to be personally accountable for their contribution.

Funding

None.

Data availability

The datasets generated and/or analysed during the current study are not publicly available due to trainee privacy but are available from the corresponding author on reasonable request.

Declarations

The Mayo Clinic Institutional Review Board (IRB) and Internal Medicine Residency Education Group (IMREG) approved this study. The IRB and IMREG groups served as the ethics committees that waived the need for informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.