Our study sampled local and foreign students from local and overseas medical schools undergoing a specific clinical rotation in a public hospital. This is the first-of-its-kind medical education study applied to a heterogeneous population from different medical schools within the same hospital setting. This distinguishes our study from its predecessors where students sampled were from the same medical school and studies were conducted internally within universities or hospitals affiliated to universities to examine the learning approaches of their own students.
Our study found that the majority of students used deep and strategic learning approaches, with the predominant approach to learning being the strategic approach. In a study conducted by Shankar et al.  on undergraduate medical students in a medical school in Aruba, the majority used deep and strategic approaches to learning, with the median scores for deep, strategic and surface approach being 60, 73 and 52 respectively. The study population consisted of students mainly from the United States and Canada admitted to the undergraduate medical (MD) program. Another study conducted by Samarakoon et al.  in Sri Lanka also found that the strategic learning was the predominant learning approach in all three groups of pre-clinical, clinical and postgraduate students. In this study, the learning styles and approaches to learning in cohorts of undergraduate students in first (preclinical) year and final (clinical) year in the University of Colombo as well as postgraduate trainees of the Postgraduate Institute of Medicine, University of Colombo, Sri Lanka, were analysed. Both these studies similarly found the preference for deep and strategic learning amongst the students and also the predominance of strategic learners, which concurs with our study on medical students in our hospital. The predominant deep and strategic learning approaches amongst these student populations may be due to the global trend towards encouraging deeper learning from medical students through a shift in medical education paradigm from didactic lectures to integrated problem-based learning . Problem-based learning is a student-centered activity known to promote deep learning, unlike conventional teaching methods [8, 10]. In addition, the predominant deep and strategic learning approach may also relate to the inherent internal motivation and interest of medical students in the medical field .
Despite the differences in sample populations in other studies [6, 9] with ours, all three-study populations consistently showed an overall preference for the strategic approach to learning. This suggests that despite individual schools’ efforts to promote deep learning, majority of students studying medicine irrespective of culture or school, appear to prefer assessment-oriented strategic learning rather than deep learning. We posit that it could be potentially due to the need to cope with the heavy content, higher workload and tight course schedules in medical education that would require one to have good time management and organisation skills.
However, in Shah et al.’s  study of the learning approach among health sciences students which included medical students in a medical college in Nepal, he found that majority of the medical students adopted the deep learning as their predominant learning approach. A possible explanation for this difference may be that the students’ learning approaches have already been established prior to university entry . Learning approaches of medical students could have already been shaped by the quality of their teaching-learning environment and assessment procedures  in their pre-university years. As pre-university education in Singapore is largely didactic and lecture based, pre-university students are encouraged to use strategic learning. Many pre-university students in Singapore also attend supplementary private tuition classes . Attending private tuition classes reinforces the strategic and surface learning approach of students, as private tutors primarily promote exam-oriented learning to yield better student performance statistics and increase their own tutor credibility . Thus, Singapore’s pre-university education may be more competitive then Nepal’s as greater emphasis is placed on achieving good grades, a pre-requisite to entering medical school. In addition, the predominant deep learning approach in medical students may also be a reflection of the nature of medical school education in Nepal. Hence, it would be beneficial to compare how the local medical curricular differ from others to better modify our curriculum to promote deep learning.
Our study results showed that difference in age was not significantly related to the predominant learning approaches adopted by the students. This aligns with the findings from the study of pre-clinical undergraduate Nepalese students  which found no significant difference between age and learning approaches. As the ages of the students were either 19 or 20 years, the study population’s age range is significantly narrower compared to our study. It is unlikely that the relatively small difference in age will have an impact on the learning approaches on the students.
However, another study conducted by Wickramasinghe et al. on a mixture of pre-clinical, clinical and postgraduate students from the University of Colombo, Sri Lanka, concluded that scores for the strategic approach was significantly affected by age among the preclinical students (ρ = 0.206, p = 0.002), but no other significant relationship was seen in other groups for any of the approaches . This is in keeping with the results of our study on clinical medical students where age did not significantly impact the learning approaches adopted by students.
Since a significant association between age and learning approach was only noted amongst preclinical students in the other studies and not amongst the students in their clinical years, one explanation for this might be the differences in the format of teaching. During the first two pre-clinical years in medical school, didactic lectures, which promote surface and strategic learning, are the main method of delivery to impart large amounts of medical knowledge, and students’ approaches towards learning are constrained to some degree by the nature of environment and assessment strategy. Whereas in the later years, learning takes place in a clinical environment and more emphasis is placed on problem-based learning (PBL), which is able to increase deep learning .
Previous findings reported by Emilia et al.  in an Indonesian medical school and Wickramasinghe et al.  found that gender was not significantly associated with the predominant learning approach. In the Nepalese study conducted by Shankar et al., there was also no significant difference in scores by gender . While our results were similar, it was interesting to find that the male students in our study are deemed to be less likely than female students to exhibit strategic learning, with p value = 0.06 almost reaching statistical significance of 0.05. An important fact to also note is that learning styles may change over a shorter time frame than over the course of a medical degree. Learning styles indeed may change based on the context, environment and topic being learned and is likely a flexible changing trait rather than a fixed innate trait a student possess [23, 24]. Our study is a pilot study and serves as a starting point for student awareness of different learning styles and to start reflecting on adopting more appropriate learning styles in different situations.
Our findings found that age, gender, highest education qualification have no significant association with one’s predominant learning approach. Therefore, further longitudinal studies as well as further studies to monitor and identify key factors that cause students’ learning approach to change through the years would be beneficial for planning and modifying medical curriculum in the medical schools in Singapore. For instance, if factors such as the syllabus structure or learning environment in medical school have been identified to play a crucial role in changing the predominant learning approach of the individual, then efforts can be focused on these areas to encourage students to adopt the deep approach from the start of medical school.
Further studies can also be conducted to investigate the relation between learning approaches and academic performances amongst medical students in Singapore since other studies have suggested a positive correlation between learning approaches and academic performance, where students with deep approach achieved higher performances and vice versa . With the increase in the number of medical students and limited amount of resources and time, finding the solution to encourage independent deep learning amongst the students would be beneficial in the long run, ensuring that students develop the most advantageous learning approach from the start. Hence, motivating medical students towards deep learning would be beneficial in achieving expected long-term goals.
There are limitations to the ASSIST questionnaire despite it being valid and internally consistent. The ASSIST questionnaire is a self-reporting instrument and may not always reflect the true approach to learning of students, especially if they answered the questions in a way that they thought would have been the approved answers . Furthermore, since there were many different questions to check for consistency in the student’s response, this renders it difficult for students to provide certain approved answers, which may ensure that the results obtained are more objective.
We feel that a more critical approach to describing the nature and categorisation of learning approaches could have been taken. The student population may have experienced a range of medical curricula prior to taking part in the study. Out of the three cohorts we included, the students from Singapore Medical schools are likely to have experienced similar curriculums. The third cohort consisting of foreign students on their elective placements might have been exposed to very different medical educational approaches. Though they comprise of only 18.5% of total study population, it would have been worthwhile considering their previous educational experiences and cultural backgrounds.
We attempted to use other tests that allowed for the determination of a single approach such as using paired t-tests. When we tried to use paired t-test for analysis, there was a significant difference between deep (M = 0.40, SD = 0.49) and surface (M = 0.09, SD = 0.28) approaches; t (237) = 7.75, p < 0.01) and between strategic (M = 0.51, SD = 0.50) and surface (M = 0.09, SD = 0.28) approaches; t (237) = 9.98, p < 0.01). These results suggest that there is a statistically significant difference between the mean scores of deep versus surface, and strategic versus surface. There was also a borderline significance between deep (M = 0.40, SD = 0.49) and strategic (M = 0.51, SD = 0.50) approaches; t(237) = − 1.70, p = 0.09. There was a significant difference between students adopting surface approach and the other two approaches and this was what we expected to identify, since we wanted to determine the percentage of students adopting the surface approach so that we could focus on interventions to reduce this percentage in our future studies.
There are other limitations of this study which include the small sample size and examining a small range of demographic variables. These limitations also do not allow us to make more changes that are specific to the curriculum in order to encourage students to use higher order learning approaches, which are desirable in medical education. However, given these limitations, we felt that there is a need to still conduct and publish this study because it provides initial direction and approach to modifying the curriculum. This is a baseline pilot study with the intention of determining the course of future directions to more specifically tailor interventions in larger scale future studies. We firstly would need to explore whether these learning approaches are related to demographic variables for which we can then more appropriately plan future studies, especially designing interventions and tailoring where best to intervene along the students learning time course.