Throughout the world, places in medical school are highly-prized with a large number of applicants applying for a limited number of places. Although attrition from medical programmes is low compared to other university courses 1 , every year students leave medicine, either by choice or by necessity. This has important consequences for the university, society, the profession and most importantly, the students themselves.

Having a low attrition rate is important, not least because of the difficulty in securing a place in medical school and the high cost of tax-funded medical training. For the medical profession, dropout results in a loss of useful contribution and impacts on medical workforce planning. A high attrition rate can affect the academic reputation of a medical school and staff morale and may have financial consequences with subsequent impact on research and teaching. Most important of all, however, is the effect of leaving medicine on the individual student - dropout has considerable financial, social and emotional consequences and can cause great distress, low morale and poor self-esteem.

Medical school attrition is therefore an important area in medical education. Research in this area can be used to inform medical school selection methods, teaching and curriculum, and student support services. The time-frame of this study covered a period of considerable change in our institution, including change in admission policies 2 .

A MEDLINE search identified 36 relevant studies (mainly US and UK, but also the Netherlands, Australia and South Africa). The studies varied in size, scope, design, variables, and methods of calculating dropout. A recent meta-analysis found an average attrition rate of 11.1% (range: 2.4–26.2%) 3 while a UK ten-year retrospective study found a dropout rate of 14% 4 . A retrospective cohort study at Nottingham University found an attrition rate of 6% 5 .

Variables commonly studied in medical school attrition studies include age, gender, nationality, academic preparedness (prior academic performance, grade point averages) and medical school selection methods 3 4 5 6 7 8 9 10 11 12 13 14 15 16 . Besides ‘observable’ student variables such as gender, nationality and age, dropout may also be associated with ‘unobserved’ student characteristics such as commitment, resilience and motivation to study medicine and also with medical school (institution) factors (entry requirements, teaching, assessment procedures, curriculum design and delivery) 12 . There are a small number of studies (mainly Netherlands) on the impact of curriculum content and delivery on dropout 17 and a few studies (US, UK and Israel) 4 18 19 on the relationship between academic struggle at medical school and dropout. There are very few studies on physical/psychological morbidity 3 and little is known about the role of financial problems, relationship problems, homesickness, engagement with student welfare services, absenteeism and the effects of educational initiatives on dropout. Similarly, there is a lack of research on the effect of actual change in curriculum on dropout and on the personal effects of dropout on individual students.

We know, however, that there is no distinct ‘dropout profile’ and students leave medicine for a variety of reasons. Some leave because they have no choice (persistent academic failure) while others leave for non-academic reasons including change of mind about a medical career. Common reasons for dropout are persistent academic failure, wrong career choice, personal reasons, and physical and psychological illness 4 . However, academic preparedness has been the only consistent variable shown to be associated with dropout (US, UK, Australia, Netherlands) 3 6 8 9 11 12 14 20 21 22 23 24 25 .

Our study was a retrospective descriptive study carried out at University College Cork looking at medical school attrition over a Ten-Year period (2001–2011). In line with international trends in medical education, a new Graduate-Entry programme was introduced during this period. This was also a decade that saw the introduction of problem-based learning medical curricula in many medical schools.

The term ‘drop-out’ was used to describe students who failed to continue their medical studies including students who transferred to other medical schools (‘transfers’) during the period September 2001-August 2011. Dropout students were identified from archived medical school dropout files.

Student files, examination records and other available information were analyzed with regard to the variables of interest. Coding of a variable as positive was conditional on the actual documentation of the presence of that variable in the student file i.e. presence of financial problems was coded as positive only if there was a file entry stating that financial problems existed. Similarly, absenteeism was only coded positive if there were file entries documenting student absences from clinics, lectures etc. Health problems and psychological problems often required an element of judgement but were only included if they were considered to have had a probable impact on the student’s progression (i.e. a student feeling stressed a few days before examinations would not be categorized as having a psychological/psychiatric disorder). In some of these cases, staff interviews helped to clarify the situation.

Variables were listed in the database as ‘evidence of physical ill-health, ‘evidence of social isolation’ etc. rather than ‘physical ill-health’ or ‘social isolation’, highlighting the importance of having documented proof and objective identification of the variable. The authors analyzing the files had no previous knowledge of these students, thence allowing assessment of the data without personal prejudice.

Examination failures, repeat years and leave of absence were all cross-checked with data from the Examinations Office. All data entries were reviewed by BM. Data omissions/ambiguities were clarified by reviewing student files and discussing with the research team. Decision was taken by consensus.

Overall attrition rate (completed cohorts 2002–2007) was 6.8% (53/779) with rates ranging from 4.38% (2002/03) to 9.15% (2007/08) (Table 2). Eight of these dropout students left to continue their medical studies at another medical school (usually the student’s home country). Dropout rate was 5.7% (45/779) when these 8 transfer students were excluded (rates ranged from 4.38% (2002/03) to 7.74% in 2007/08). Current dropout rate for the 2008/09 cohort is 5.5% (excludes dropouts which may yet occur in the remaining six months of the students’ final year).

Dropout rate in the ten-year period was 5.9% (81/1361) and 5.2% when transfer students were excluded.

Analysis of 5 successive completed cohorts (September 2002–September 2006 inclusive) showed that males (22) were more likely to drop out than females (18); however, the relative risk was not significantly increased (RR = 1.70 [95% CI: 0.93, 3.11]; p = 0.079) (Table 3). When transfers were excluded, the relative risk decreased in males (RR = 1.58 [95% CI: 0.82, 3.05]: p = 0.165) (Table 4). The Ten-Year study gave similar results (Tables 5 and 6).

UAE United Arab Emirates, DE Direct Entry.

UAE United Arab Emirates, DE Direct Entry.

Δ Total number of students who dropped out during the period September 2001-August 2011.

# Total number of students who enrolled September 2001 – September 2011.

Δ Number of students who dropped out excluding students who transferred during the period September 2001-August 2011.

# Total number of students who enrolled September 2001 – September 2011 excluding students who transferred.

Analysis of 5 successive completed cohorts found an increased rate of dropout in Kuwaiti and UAE students combined (RR = 5.70(2.65, 12.27); p < 0.0001). Irish and EU students demonstrated a low risk of dropout and were therefore used as the reference group (Table 3). There was increased risk of dropout in North American students (RR = 2.68 (1.09, 6.58); p = 0.027), but this was not significant when transfer students (Table 4) were excluded (RR = 1.32 (0.38, 4.62); p = 0.75). Similar results were found in the Ten-Year study (Tables 5 and 6).

The Graduate-Entry Medicine programme commenced in 2008, thus we could only compare programmes for the past four years and numbers are small. Graduate-Entry students were less likely to drop out than school-leaver students (RR = 0.84 (0.32, 2.21) p = 0.72), even moreso when students who transferred were excluded (RR = 0.51, (0.15, 1.69), p = 0.26).

The commonest age at entry to university of students who dropped out was 18, 20 and 21 (8 students in each category). The average entry age of all medical students during this period was 19 (SD 5.6 years).

Over 60% dropout students were in First Year, with Year 3 having the next highest rate (16%). Final Year had the lowest rate (5%) (Table 7). Wrong career choice and academic difficulty were the commonest reasons for dropout in the early years of the programme (Table 5). Late dropout was associated with persistent academic failure, and psychological/physical ill-health. In many cases, multiple factors were associated with dropout and were often interlinked i.e. some students with academic difficulty may have left unexpectedly just before exam time.

Data relating to 70 students (excluding transfers) who dropped out of the medical programme over the ten-year period 2001–2011 was analysed.

There was documented student file and examination results (repeating in autumn, repeating a year) evidence of academic difficulty in 55.7% of dropout students. Twenty dropout students (28%) were required to leave because of the Two-Year Rule reflecting consistent academic underperformance.

Thirty-four dropout students (34/70; 48.5%) had to repeat at least 1 year. Examination records for the entire medical student population during that period show a yearly year repeat rate of 1.2% to 3.9%.

Forty-six dropout students (65.7%) had to repeat exams in Autumn at least once. Recent examination records (2006–2011) show a yearly medical student Autumn repeat examination rate of 8.9%-13.4%.

There was student file evidence of absenteeism (from lectures, practicals, wards, or electives) as a cause for concern in 21 (30%) of dropout students. One student ‘disappeared without trace for weeks on end’. Absenteeism was associated with social isolation and academic failure in many cases.

Fourteen dropout students (20%) took one or more years leave of absence. Reasons were primarily health problems (including depression) and also family problems.

Twenty-four (34%) of students who dropped out had attended the Student Welfare service.

Social isolation was documented in 20% of dropout student files and homesickness in 20%. Only one student who dropped out lived at home. Some students described themselves as ‘shy and withdrawn’ and not able to socialise. Many said they were lonely and had no friends. Social isolation was more prevalent in overseas students.

Four of the seven dropout students (10%) with documented English language problems in our study were from Kuwait.

Seventeen (24%) of students who dropped out said they still wanted to study medicine and planned to apply elsewhere. Law was the most common alternate course (4). Other courses included Pharmacy, Mathematics, Engineering, Research, Nursing, other health courses, Science, Arts/Music and Teaching, Accountancy, Philosophy and the Navy. Reasons for withdrawal and subsequent career choices were not always evident in the student files.

•Non-EU students (Kuwait and United Arab Emirates) were at increased risk of dropout (p = 0.0001). North Americans were also at increased risk (p = 0.008) but this was not significant when transfers were excluded (p = 0.75). Malaysian students were not at increased risk of dropout.

•Males were more likely to drop out than females but this difference was not significant.

•There was a lower rate of dropout in Graduate-Entry students but this was not statistically significant.

•Change of curriculum was associated with a transient and small increase in dropout in the first year of the new curriculum.

•Dropout can have serious emotional, financial and personal consequences for medical students.

•There was a high prevalence of pyschological/psychiatric problems in dropout students. However, most dropout students did not attend Student Welfare or other supports.

•Social isolation was common.

•There was a high rate of documented absenteeism in students who dropped out.

•Sixty per cent of dropouts occurred in First Year, 16% in Third Year and 5% in Final Year.

While studies have shown that ethnic minority medical students are at increased risk of academic ‘struggling’ 5 7 13 16 29 30 and one study found increased risk of dropout 16 , our study found an increased risk of dropout in students from Kuwait and UAE but not Malaysia (this country-specific effect has not been noted before). Academic admission criteria for Kuwaiti and Malaysian students are comparable and both cohorts are government sponsored. The higher number of Malaysian students, however, may protect against social isolation. Language fluency problems may also be a contributory factor.

The low dropout rate of Malaysian students is of particular interest in view of the large numbers of Malaysian students who travel abroad to study Medicine.

North Americans were also at increased risk of dropout (p = 0.008) but this was not significant when transfer students were excluded (p = 0.75).

Similar to other studies 4 11 12 , our study found that while males were at increased risk of dropout, this was not significant.

Similar to other studies 17 31 34 , the Graduate-Entry programme had a lower dropout rate, moreso when students who transfered were excluded. However, our numbers are small and it will be interesting to see future analyses in this area. Graduate-Entry programmes are more likely to utilise Problem-Based Learning (PBL) which is associated with lower dropout than traditional curricula 4 17 31 .

While studies have found higher dropout in students studying a traditional curriculum than a PBL curriculum 4 17 31 , there has been little research on effects of curriculum change.

Our curriculum changed to an integrated systems-based curriculum in 2005/2006. Dropout rate that year (2005/6) increased by 1% (5.97 to 6.92), but fell again the following year (2006/07) to 5.38%. There was a rise in dropout (7.74%) the following year (2007/08, the year of highest dropout) with most of these being First Year students. The increase in dropout in 2005/06 may be due to a small, but transient effect of curriculum change or may have been circumstantial, suggesting that faculty adjust quickly to delivering a new curriculum and that change in curriculum does not adversely affect the student experience. If increased dropout numbers were simply organizational or ‘teething problems’ of a new curriculum, one would also have expected higher dropout numbers in 06/07 which was not the case. One would expect that a new curriculum might impact during the first few years when students who have to repeat exams or perhaps repeat a year, find themselves ‘falling between two curriculae’. The majority of dropouts in the 2007/08 cohort (third year of new curriculum) were First Years (may be related to unidentified institutional factors or may be a chance occurrence).

We identified a number of variables associated with dropout including examination failure, psychological problems (40%), absenteeism (30%), homesickness (20%), English fluency problems (10%) and social isolation (20%).

Academic difficulty (the main reason) and wrong career choice were the commonest reasons for dropout in the early years of the programme (Table 5). Our finding of higher dropout in the pre-clinical years, particularly First Year (60% dropouts in First Year), is in keeping with other studies 3 . This high attrition in First Year is not surprising. Students aged nineteen may lack the maturity to make informed career choices and may have difficulty adapting to the self-directed learning environment of University and to living away from home. Many students struggle with loneliness and homesickness. Others are unprepared for the volume of work. Some students quickly realise that they have made the wrong career choice and make an early decision to change courses. Only one student in our study cited parental influences to study medicine (although students may be slow to reveal this). Physical and psychological problems occur in students at all stages of the programme, but earlier on in the programme may not be revealed or may have been coped with by taking time-out.

Late dropout was associated with persistent academic failure, and psychological/physical ill-health, in particular, depression. Health problems tended to have been present for some time but may only have come to attention in later years. Students suffering from mental illness including addictions may have been reluctant to seek help for fear of ‘black-marking’.

The dropout rate in Final Year is low (5%) but dropout at this part of the programme is particularly traumatic and may reflect a failure to identify, support and offer alternative career strategies to students at an earlier stage in the course. Exiting the course in Final Year is particularly devastating and every effort must be made to identify and support students at risk of dropout before they progress to Final Year.

Our findings of 16% of dropouts occuring in Year 3 is higher than other studies 3 . Dropout in this year of the programme may be related to student commencement of full-time clinical rotations. The realities of a medical career may become apparent; students may be uncomfortable in clinical settings and may feel isolated from their colleagues.

Our study allowed us an insight into the personal impact of dropout, and events leading up to dropout, areas which have not received much attention. Of particular concern was the degree of social isolation (20% of dropout students). We identified recurrent themes of loneliness, failure, and despair, Dropout can be very traumatic for individual students who, having made the decision to study medicine, find that they cannot cope or realise that medicine was not the right choice.

Only one third of dropout students attended Student Welfare Services which is a cause for concern, as these are the very students who need help. Students may be afraid to disclose sensitive personal or health-related information because of worries about confidentiality. These students are also high achievers and may be reluctant to accept that they are having academic problems and that they need help.

In contrast to other dropout studies where Science was the most common alternate career choice 4 , Law was the most frequent career choice in our study.

Our attrition rate is relatively low (6.8% (53/779) and 5.7% (45/779) when students who transferred were excluded) compared to other studies 2 4 6 7 8 11 12 14 15 16 19 21 22 26 28 31 33 37 38 39 . A recent meta-analysis found an average attrition rate of 11.1% (2.4–26.2%) 3 .

In contrast to other dropout students, transfer students had a good academic record. Family and personal reasons were the usual reasons (students wanted to live near family). Most transferred after First Year, and some after the preclinical programme. Over half (6/11) of transfers were North American.

The strength of this study lies in the study’s detailed analysis of student factors associated with dropout (commonly studied variables along with other variables such as financial problems, social isolation, absenteeism, leave of absence) on a large cohort of students using a combination of qualitative and quantitative data.

The high incidence of absenteeism and leave of absence in our study suggests that these should be considered red flag signals and should be more closely monitored in student populations.

Importantly, this study identified a much higher rate of psychological illness (40%) in dropout students than previously described and deserves attention in all medical schools.

Our study showed a nationality-specific increased risk of dropout, suggesting the importance of analysing dropout rates for specific nationalities, rather than simply grouping students as ‘Overseas students’.

Another strength of the study was that it looked at the individual stories behind dropout allowing insight into the personal and emotional impact of dropout, an area where there has been little research. Qualitative analysis and inclusion of previously unstudied variables combined to gave a ‘big picture’ approach to determining the complex reasons underlying dropout. Behind every dropout statistic lies a personal story, some more tragic than others.

Another strength of the study was the analysis of the impact of changing the curriculum on attrition, a previously unstudied area.

The study also allowed us to compare dropout rates between a school-leaver programme and a Graduate-Entry programme, both programmes having similar teaching facilities/faculty. Results showed that dropout was lower in the Graduate-Entry programme, although this was not statistically significant. To date, only one cohort has completed the Graduate-Entry programme and student numbers are small.

Our study findings are of interest to all medical schools, as being able to identify students at risk of dropout is a concern for medical faculty in all institutions. Many medical schools are introducing new curricula and new Graduate-Entry programmes and are also increasing their intake of overseas students. Awareness of the effects of dropout on medical students is important for all medical schools.

Limitations of the study relate to restricted statistical analysis (lack of comparative data for the general medical student population).

This study was conducted in one medical school and our data may not necessarily generalise to other medical schools with different selection methods and student profiles. However, a similar list of underlying reasons for dropout has been found in international dropout studies, suggesting that our results may be relevant for other medical schools and can be generalised 5 20 .

Causes of dropout may be multifactorial and reasons for withdrawal may be more complex than we were able to ascertain.

Recording of data was based on information in the student files which depended on what the student disclosed. It is very likely that the prevalence of some problems such as psychological/physical illness, family/personal problems etc. in the dropout population is higher than our findings. Especially in the case of psychiatric illness including addiction, eating disorders and depression, many cases may have been undiagnosed, unreported, or undocumented. We did, however, seek to supplement information by interviewing staff (this is subject to bias).

We were not able to analyse the effect of change in admission policy (more targeted selection processes are generally associated with lower dropout 22 32 33 ) as none of the ‘HPAT’ cohorts have graduated yet.

We were unable to control for confounders when we analysed data for gender/nationality associations.

Qualitative analysis may have been prone to observer bias.

Medical school attrition is important and there is a need for more rigorous studies and qualitative research on this subject 3 .

Identification of institution factors (student selection, curriculum, course delivery and assessment) is an important research area and one which has not received much attention. Future studies need to address this. Future studies also need to examine the relationship between specific nationalities and rates of dropout.

Exit interviews should be considered in all medical schools.

Student Welfare services should be actively promoted, in particular to at-risk students (First Years, students who fail exams, and overseas students). Only a small proportion of our dropout students availed of Student Welfare services, despite the high prevalence of psychological ill-health and social isolation in this population. This suggests the need to regularly promote these supports to students and to reinforce the acceptability of accessing help. As numbers of non-EU students continue to expand, Student Welfare services are particularly important in this group. Students need to be reassured regarding confidentiality and the non-disclosure of health and social problems.

Mental Health services need to receive more attention in medical schools. Stress management programmes and programmes to increase awareness of depression and anxiety disorders should be promoted. Practical considerations regarding ongoing student health issues may need to be addressed by Fitness to Practice procedures.

It is important to identify language fluency problems at an early stage and arrange appropriate intervention.

Absenteeism should be viewed as an important red flag signal and prompt early interview. Consideration should be given to more vigilant attendance documentation.

Educational interventions (including key strategic learning skills)/communications skills/procedural skills) may be of benefit to some students with academic difficulty but research is necessary to evaluate the effects of these interventions 34 . Students who have difficulty with procedural skills or communication skills should be identified early in the programme.

Dropping out of medical school can be a very traumatic time in a young person’s life. It is important that we continue to support these students after they have left the programme and assist them in exploring other career options.