Dynamics of career intentions in a medical student cohort: a four-year longitudinal study | BMC Medical Education

Study design

We analyzed longitudinal data from an undergraduate medical student cohort followed over four years. The present report is one of several from this cohort study, which started in 2011. It was designed to study the impact of student and institutional factors on academic performance and career choices over the course of the entire undergraduate curriculum. The analysis presented in this paper is based on a selection of variables considered relevant for answering the research objectives. We briefly present the study design and context; more details and other findings related to this cohort may be found in previous publications [21,22,23,24].

Setting

The study took place at the Faculty of Medicine in Geneva, Switzerland. The six-year undergraduate curriculum consists of a pre-selection year (year 1), two pre-clinical years (years 2 and 3), two years of compulsory clinical clerkships (years 4 and 5), and one elective year (year 6). Medical graduates pursue postgraduate training in the field of their choice; change of career orientation is possible during postgraduate training.

Participants

All students from two consecutive classes, starting medical school in 2011 and 2012, were recruited into a cohort study during their first academic year and invited to complete a yearly paper-and-pencil survey. From a total of 306 students in these two classes, 290 (95%) enrolled into the study. Participants signed a consent form after being informed about the study. They provided a unique student identification number at each data collection to allow longitudinal matching of questionnaires. Researchers involved in the analysis could not identify students through their identification numbers to guarantee confidentiality. For the present study, we considered data collected from academic year 3 (i.e., end of pre-clinical training) to year 6 (i.e., before final certification exams). We included individuals who had provided career intentions in at least three out of these four years. Our sample thus included 262 students (i.e., 85.6% of all students in these two classes).

Student-related variables

Table 1 presents the variables chosen from the cohort dataset according to their relevancy for the objective of the present study. Career intentions were assessed by two single-choice questions: (1) “What type of practice do you plan to exercise in the future?” (options: private practice, hospital practice, or teaching/research), and (2) “If you are considering a specialization, which one?” (options: anesthesiology, general internal medicine, internal medicine subspecialty, emergency medicine, obstetrics-gynecology, pathology, pediatrics, psychiatry, radiology, surgery, academic activity, undecided, and other with the option of specifying a hand-written specialty). The list of specialties was derived from postgraduate specializations available in Switzerland. For the statistical analysis, specialties were grouped into the following six categories: General medicine (general internal medicine, pediatrics), medical specialties, surgical specialties (surgery, gynecology/obstetrics, ophthalmology), acute care (anesthesiology and intensive care, emergency medicine), technical specialties (radiology and medical informatics, pathology and forensic medicine), and other (psychiatry, academic activity, other specialties).

Students were asked whether they had identified a positive role model during their studies (answer options: yes or no); this variable was included in our analysis as we hypothesized that having a role model could be associated with more stable career plans [25].

Motives for becoming a physician were included because of their previously observed association with certain career choices [20]. In our previous study, we also observed that motives remained mostly stable; therefore, we only included measures from study year 3 in the present analysis. A list of motives for becoming a physician was presented to the students: academic interest, prestige, reward, private practice, saving lives, caring for patients, cure diseases, vocation, mission, and altruism. They were asked to rate the importance of each motive on a 6-point scale (i.e., “Describe how important each of these keywords is for your choice of medicine” from 1 = not important at all to 6 = very important). This set of motives was developed based on a literature review aiming at achieving a wide enough description of different typologies of motives [26,27,28,29]. Validity analyses of the set of motives and their correlation with empathy and learning approaches has been reported previously [30]. Students’ overall motivation to become a physician was also included as a variable, as we hypothesized that a stronger motivation could lead to a stronger commitment to a specific career choice. Students were asked to rate their overall motivation on a 6-point scale (1 = not motivated at all to 6 = very motivated).

Empathy was included because it is considered a key correlate of medical students’ career choices [31, 32], emotional intelligence [33, 34], and psychological distress [35, 36], thus potentially contributing to changing career preferences. Empathy was measured using the student’s version of the Jefferson Scale of Empathy (JSE-S), consisting of a 20-item questionnaire assessing students’ perception of the importance of empathy in the doctor-patient relationship (Cronbach’s Alpha α = 0.83) [37]. Answers to questions are structured on a 7-point scale; a total score is calculated by summing up all answers. We used a French version of the JSE-S whose validity and cross-national generalizability has been confirmed [38].

Personality traits were included because of their associations with career indecision [39]. Also, personality has been hypothesized to play a role in changing career preferences [15]. The validated French version of the NEO Five Factor Inventory (NEO-FFI) was used [40]. It is the short version of the classic revised NEO Personality Inventory, which is the assessment instrument most often used to measure personality with the “Big Five” model, assuming five underlying personality dimensions. NEO-FFI includes 12 items for each dimension scored on a 5-point scale: neuroticism (α = 0.87), extraversion (α = 0.74), openness (α = 0.69), conscientiousness (α = 0.86), and agreeableness (α = 0.73).

Coping strategies were included because they have been described as important influences on career decision-making in medical students; notably, maladaptive coping strategies were associated with career indecision [18]. Career decision-making may be regarded as a potential stressor and approached differently depending on students’ strategies coping with this stress. Coping styles were measured by the Coping Inventory for Stressful Situations (CISS), which assesses coping strategies that individuals might use when facing stressful situations [41, 42]. Three dimensions of 16 items each are rated on a 5-point scale: emotional coping (α = 0.87), task coping (α = 0.89), and avoidant coping (α = 0.80). This questionnaire is widely used in the domain of psychology and has been validated in a number of languages, including French.

Score of career intention instability

We developed a score quantifying the instability of career intentions. It was calculated for each student and composed of three elements, detailed in Table 2: (1) change of practice type intentions over the four years; (2) change of specialty intentions from one year to another; and (3) the number of different specialties indicated over the four years. We also accounted for certain specialties being closer to one another by applying a corrective factor. Career intention changes occurring later were attributed more weight as we considered these to potentially reflect a more pronounced career indecision. The score ranged from 0 (= maximum stability, i.e., career intentions never changed) to 10 (= maximum instability of career intentions). Seven students indicated being undecided in year 6 and in at least two of the three previous years; we manually attributed them the maximum score of 10 as we considered them to be highly undecided.

Analyses

We examined data for accuracy and missingness and calculated descriptive statistics for age (mean and standard deviation) and gender (N and % of females). For an overview of our cohorts’ characteristics, we cross-sectionally described career intentions (N and % per academic year) and overall number of changes in career intentions and the respective specialties. The career intention instability score was analyzed descriptively (mean, standard deviation, median, quartiles).

We used a regression analysis to search for associations between the score (the dependent variable) and the other variables, selected because of their potential impact on career-related decision-making (see above). Specialties were grouped into six categories to limit the number of variables. Given the score’s distribution, we applied a beta regression considering the scaled instability score as the dependent variable. Moreover, considering the large number of covariates, we selected the model based on a stepwise procedure using the Generalized Akaike Information Criterion [43]. This stepwise regression method is a standard approach to select relevant covariates and to obtain an interpretable model in the presence of a large number of covariates compared to the number of observations [44, 45]. The adequacy of the model with the data was assessed with a residual analysis based on the randomized quantile residuals [46]. The model’s estimated coefficients were interpreted in terms of their magnitude (absolute value), sign (positive or negative), and p-value.

All analyses were carried out with R statistical software [47] and the package GAMLSS [48]. P-values smaller than 0.05 were considered statistically significant.