National curriculum reform
In 2000, the joint committee of deans of medical schools in Taiwan passed a curriculum reform resolution to improve the medical education system. An expert panel was formed to design the new 6-year medical system [18]. An experimental 6-year curriculum was tested at the National Taiwan University College of Medicine with selected students, and their learning results did not differ much from those of students following the original 7-year curriculum [15]. In 2009, the joint committee announced that a new medical education curriculum would be carried out in 2013, and the first year of the new curriculum would yield graduates in 2019. Following decreased undergraduate rotation, the postgraduate year (PGY) rotation was increased from 1 year to 2 years. PGY training is a general practitioner rotational program undertaken before further specialist training (residency) [14, 22]. After the outbreak of SARS (severe acute respiratory syndrome) in 2003, the Taiwan Ministry of Health and Welfare set up this one-year PGY training to improve medical graduates’ core competencies and prevent them from becoming specialized too early [17]. In a way, the reform placed the graduation and board certification 1 year earlier than they were previously (Fig. 1).
Study setting
This was a prospective, longitudinal comparative cohort study conducted between 2017 and 2020. In July 2017, students of the last generation to receive the old curriculum (2012 cohort, 3 years of clinical rotation) and the first generation to receive the new curriculum (2013 cohort, 2 years of clinical rotation) started their last two years of clinical learning simultaneously. They spent their last two years in teaching hospitals completing their clinical rotation, graduated, took national board exams, and then started their postgraduate clinical learning at the same time (Fig. 1). A deidentified online survey was given to the participants at multiple prescheduled times. The study was approved by the Chang Gung Medical Foundation institutional review board (IRB No. 201601758B0, 201701981B0). All methods were carried out in accordance with relevant guidelines and regulations.
Ethical approval
The study was approved by the Chang Gung Medical Foundation institutional review board (IRB No. 201601758B0, 201701981B0).
Participants
Undergraduate students admitted to Chang-Gung University College of Medicine in 2012 (old curriculum) and 2013 (new curriculum) were eligible to participate in the study. We invited them via the Internet to join this three-year longitudinal survey study from two years before graduation to one year after graduation. All participants were given a one-hour orientation before providing written informed consent. To eliminate the effect of change itself on learning experiences, we enrolled another group of participants from the 2014 cohort (new curriculum) using the same method described above. This group of participants followed the new curriculum but were enrolled one year after the year of the change and thus served as the control group among the new curriculum students.
During the first year of the study (2017/06-2018/05), the participants of the 2012 cohort and 2013 cohort entered their clinical rotation at the same teaching hospital (site A), which, with a 3600-bed capacity, is the largest teaching hospital in Taiwan. During the second year of the study (2017/06-2018/05), the participants of the 2014 cohort entered their clinical rotation in the same teaching hospital (site A), and the participants from the 2012 and 2013 cohorts were distributed among the four teaching hospitals in Taiwan (sites A, B, C, D) according to their preference. These four teaching hospitals, which differ in geographical location and patient population backgrounds, consist of two medical centers and two regional hospitals. The participants from both the 2012 and 2013 cohorts graduated in 2018/06 and took the national board exams in 2018/07. During the third year of the study (2018/06-2019/05), most of the participants from both the 2012 and 2013 cohorts began their mandatory postgraduate rotation (PGY) training. The PGY training lasts for one year in the old curriculum and two years in the new curriculum (Fig. 1). A few of the participants who did not proceed to PGY training either entered military service, in the case of some male graduates; prepared for another round of board exams due to failing their first attempt; or were lost during follow-up. A detailed list of the distribution of participants is provided in Supplementary file 1.
Data collection and measurements
We collected basic personal demographic information including age, sex, and self-reported previous academic performance before entering clinical rotation. Learning-related variables included which study hospital they were assigned to. The participants from the 2012 and 2013 cohorts were asked to complete surveys regarding their preparedness and degree of burnout every 6 months after joining the study. These measures were labeled from 1 to 6 according to chronological order, with the 1st to 6th measurements being conducted at 18 months before graduation, 12 months before graduation, 6 months before graduation, 1 month before graduation, 5 months after graduation, and 11 months after graduation, respectively. The participants from the 2014 cohort, who served as a control group against the variable of year of change, completed only two surveys with a 1-year gap, which were spaced to match the timing of the 1st and 3rd measurements of the study group. A diagram of serial measurements and number of respondents in each measurement is presented in Fig. 2. The survey process was conducted online and anonymously, and respondents were instructed to answer the questions in reference to their rotations in the current and previous months.
Their preparedness for independent clinical practice was assessed using the Chinese version of the Preparedness for Hospital Practice Questionnaire (CPHPQ). This eight-domain measurement, consisting of 41 items evaluated on a six-point Likert scale, was previously developed and validated by Chaou et al. with good internal consistency (Cronbach's alpha=0.94) [23]. The eight subscale domains were interpersonal skills (IS), clinical confidence (CF), team collaboration (CL), patient management (MG), medical science (SC), disease prevention (PV), holistic care (HC), and self-directed learning (SDL).
We evaluated the degree of burnout using the Copenhagen Burnout Inventory (CBI), a measurement developed by Kristensen et al. to measure burnout in the populations of medical professionals [24]. The CBI measures three dimensions, including personal burnout, work-related burnout, and patient-related burnout, using a 5-point Likert scale. It has been shown to be more straightforward for measuring burnout among medical professionals than another frequently used questionnaire, the Maslach Burnout Inventory [25]. A Chinese version of the CBI has also been developed and validated in previous studies [26,27,28].
Statistical analysis
We determined the sample size before the research was started. For the descriptive results, the mean and standard deviation (SD) are used to describe the central tendency and spread of continuous variables and the count and percentage for categorical variables. We used independent t tests for the comparison of continuous results between groups and a chi-square test or Fisher's exact test for categorical results, where appropriate. We treated the sum of the Likert scales as continuous variables in accordance with Norman et al. [29].
For the analysis of curriculum change on the preparedness and burnout of medical students during the graduation transition, we used linear mixed models to cope with dependence within the subject or training hospital while exploring the effects of possible explanatory variables [30]. Serial measurements of total PHPQ scores and CBI scores were taken as dependent variables. Individual learners and training sites were added into the model as random intercepts. The models tested the effect of curriculum change (old vs. new) and the year of change while adjusting for personal demographic variables, including sex, age, and self-reported previous academic performance. The models were fit using the residual maximum likelihood (REML) approach, and the covariance structure of the two random effects was set as unstructured. All analyses were performed using SAS statistical software version 9.3 (SAS Institute Inc., Cary, NC). A reported p-value of less than 0.05 was considered statistically significant.