Abstract
Objectives
To investigate whether demographic factors are associated with self-reported experience amongst medical trainees in the UK.
Design
Retrospective analysis of survey data.
Setting
General Medical Council (UK) National Training Survey data for 2014 and 2015.
Participants
A total of 105,549 responses were provided from 68,551 participants when no data were removed. After removing data to preserve participant anonymity, there were 64,278 participants providing 99,076 responses.
Main outcome measures
Considered trainee factors were gender, ethnicity, country of primary medical qualification, grade, post specialty and deanery. Self-reported outcome measures were ‘overall satisfaction’, ‘adequate experience’, ‘workload’, ‘clinical supervision’, ‘educational supervision’, and ‘access to educational resources’.
Results
The experience of medical trainees across various indicators is differentially related to gender, ethnicity, country of primary medical qualification, grade, post specialty and deanery.
Conclusions
It is demonstrated here that trainee factors are associated with subjective experience across different indicators. Further work is required to explore the reasons behind this, and how this relates to trainee quality of life, work performance and career progression.
Keywords
Introduction
The medical workforce in the UK is demographically diverse.1–3 Both undergraduate and postgraduate performance have been shown to vary across different groups, with associated factors including gender,4,5 ethnicity,5–9 and country of primary medical qualification.3,10,11
Ongoing progress is being made to explore the mechanism relating such demographic factors to performance.4,6,9,12–16 It is likely that subjective experience also plays a role.11,17 This may differ depending on grade,18,19 specialty, 20 and deanery of training.21–23
The General Medical Council National Trainee Survey monitors the quality of training and education of all doctors in the UK annually. 24 In 2014 and 2015, it collected survey responses to quantify the subjective experience of trainees to include the indicators ‘overall satisfaction’, ‘adequate experience’, ‘workload’, ‘clinical supervision’, ‘educational supervision’, and ‘access to educational resources’. 24 Trainee gender, ethnicity, country of primary medical qualification, grade, post (rather than programme of training) specialty and deanery were also included. 24 General Medical Council National Trainee Survey data for 2014 and 2015 were analysed to investigate whether trainee demographic factors are associated with self-reported experience.
Methods
Summary of responses for each of the trainee factors considered.
NTS: National Trainee Survey; IQR: Interquartile range; BME: black and minority ethnic; EEA: European Economic Area.
Though reported as numerical outcomes within a 100 point scale, measures of subjective experience were not continuous, and were thus modelled as ordered categorical variables. The considered measures were ‘overall satisfaction’, ‘adequate experience’, ‘workload’, ‘clinical supervision’, ‘educational supervision’, and ‘access to educational resources’.
Survey respondents must answer at least one fewer than the maximum number of questions that contribute to any given indicator to have a valid score; for indicators that have only two questions, respondents need a valid response for both questions. Due to differences in the nature of training in particular specialties and deaneries, no clinical supervision and workload scores were obtained for the 201 trainees in the Pharmaceutical Medicine Virtual Deanery. Missing data in the results released by the General Medical Council as a result of these issues were excluded from analysis: ‘workload’ (201, 0.2%), ‘clinical supervision’ (429 responses, 0.43%), ‘educational supervision’ (86 responses, 0.09%), and ‘access to educational resources’ (3161 responses; 3.19%).
Statistical analysis was undertaken using Stata 14 (StataCorp LP). Medians and interquartile ranges are used to display the (non-parametrically distributed) responses by participant category. The survey design function of Stata (svyset) was used to create a one-stage clustered design with stratification model, thus accounting for any reduced variance in reported experience within deaneries (clusters) and specialties (strata). Ordered logistic regression models were used to explore the adjusted association between trainee factors and the different subjective experience measures (dependent ordered categorical variables), with the largest category within each trainee factor variable used as the base level (highlighted in bold below). Coefficients and confidence intervals (CIs) are used as indicators of effect size; p values with a cut off of <0.001 are used as indicators of statistical significance (a Bonferroni correction is applied for multiple testing).
Trainee factors modelled as independent categorical variables were the year of survey (2014 or
Results
Results of the regression model for overall satisfaction.
NTS: National Trainee Survey; EEA: European Economic Area; BME: black and minority ethnic.
Results of the regression model for adequate experience.
NTS: National Trainee Survey; EEA: European Economic Area; BME: black and minority ethnic.
Results of the regression model for workload.
NTS: National Trainee Survey; EEA: European Economic Area; BME: black and minority ethnic.
Results of the regression model for clinical supervision.
NTS: National Trainee Survey; EEA: European Economic Area; BME: black and minority ethnic.
From 2014 to 2015, there was a statistically significant improvement in all considered outcome measures except educational supervision, which showed a non-statistically significant decline. Men reported superior subjective experience than women, reaching statistical significance for all six outcome measures.
Unspecified ethnicities reported worse experience in all measures compared to whites. Ethnic minorities reported worse experience in all measures compared to whites, except educational supervision and workload, where they reported better experience, though only reaching statistical significance for the former.
Medical graduates from the European Economic Area did not report a statistically significant difference in any of the six outcomes measures when compared to British graduates. As compared to British graduates, though those receiving their degrees from other countries (outside the European Economic Area) reported worse overall satisfaction, they experienced the workload, clinical supervision and access to educational resources favourably; differences in the other outcome measures did not reach statistical significance.
As compared to core and lower grade trainees, foundation year doctors reported worse overall satisfaction, workload, clinical supervision, and access to educational resources. As compared to core and lower grade trainees, ST4+ trainees reported superior overall satisfaction, adequate experience, clinical supervision and access to educational resources; they reported worse educational supervision, however.
There was marked variation in the experience of trainees across the various deaneries and specialties through the different measures.
The described associations only explain a small component of the heterogeneity in reported experiences, with no coefficient within any of the models having magnitude greater than 3 points (Tables 2 to 5, Supplementary Tables 1 and 2).
Discussion
Principal findings
Despite the use of a stringent Bonferroni correction for multiple testing, this work suggests that self-reported experiences amongst UK medical trainees in the indicators ‘overall satisfaction’, ‘adequate experience’, ‘workload’, ‘clinical supervision’, ‘educational supervision’, and ‘access to educational resources’ are related to gender, ethnicity, country of primary medical qualification, grade, deanery and post specialty. Although there were various statistically significant associations, these only related to small changes in the observed responses, with no single trainee factor associated with a more than a 3 point change in any of the considered outcome measures (Tables 2 to 5, Supplementary Tables 1 and 2). Thus, while the considered aspects are indeed associated with self-reported experience, they explain only a small component of the overall variation.
Limitations
These data have been anonymised by the General Medical Council by replacing respondent identifiers and removing responses from doctors in training where there are fewer than 3 respondents sharing the same demographic characteristics. After removing data to preserve participant anonymity, there were 64,278 participants providing 99,076 responses. Thus, a significant proportion of trainees will have completed the General Medical Council National Trainee Survey in both 2014 and 2015, in turn potentially impacting on the independence of the responses seen. Furthermore, a total of 6473 out of 105,549 (6.1%) cases were removed (Table 1). Though the overall proportion of removed responses was low, there is marked variation in how much data were lost from the different categories (Table 1), with more than 20% of responses missing from the unspecified ethnicities (33%), European Economic Area graduates (46%), Occupational Medicine trainees (38%), Ophthalmology trainees (38%), Pathology trainees (34%), Public health trainees (46%), NHS Education for Scotland East Region trainees (24%) and NHS Education for Scotland North Region trainees (21%) categories. The removal of responses was not a random process, and had greatest effect on trainees in minority groups. However, the available sample was still sufficient to detect differences within many of the considered trainee factors and experience measures. Insufficient response to survey questions also meant that data were missing for three of the outcome measures considered, though this was smaller in magnitude and less likely to have affected the results: ‘clinical supervision’ (228 responses, 0.23%), ‘educational supervision’ (86 responses, 0.09%), and ‘access to educational resources’ (3,161 responses; 3.19%).
Relation to other studies
This is the largest published analysis of survey responses to compare the subjective experience of medical trainees to demographic factors. While these findings are consistent with previous work,1–8,10,11,14,17–22 the numerical scores used also allow for quantitative comparisons to be made between the different trainee factors and outcome measures. 24 Furthermore, the General Medical Council National Trainee Survey is an annual survey, allowing the associations and differences between groups to be monitored yearly, and related to interventions targeting disparity. Work analysing the responses of the trainee survey in 2006 clearly highlighted that there were differences in the satisfaction and supervision of trainees in different specialty groups, but was not able to compare particular subsets of groups, and did not attempt to investigate the acknowledged effects of sex, year of qualification, time in post, and grade. 23 This work differed in its methodological approach as compared to the current study; with a sample size of 23,267 it used in a multi-level model considering deaneries, training providers and specialty groups. 23 Though adjustments were made for specialty group, training grade, time in post, sex, year of qualification, and the route used to respond to the questionnaire, ethnicity and country of primary medical qualification were not considered. 23 The use of a multi-level model potentially offered superior statistical robustness, but may have also hindered in its complexity, with adjustment for the year of qualification not possible when considering supervision scores due to an inability to converge the model, and the repeated division of the cohort within clusters potentially limiting the ability to identify significant differences between groups. 23
Implications and further work
As is a limitation with all such observational studies, this work only identifies associations, but does not explore the underlying mechanisms. To this end, while it is demonstrated here that trainee factors are associated with subjective experience across different indicators, further work is required to explore the reasons behind this, and how this relates to trainee quality of life, work performance and career progression. Given the demographic diversity of the medical workforce in the UK,1–3 and the recognised variation in performance across trainees of different gender,4,5 ethnicity,5–9 and country of primary medical qualification,3,10,11 it may be particularly prudent to explore how discrepancies in the experiences of these groups might affect their outcomes.
Footnotes
Declarations
Acknowledgements
This research has been undertaken using data from the National Training Survey (NTS) data set available on application from the GMC. The GMC does not however hold any responsibility for subsequent analysis done from raw data provided as this is seen as creating new information. The GMC offered clarification on missing data and calculation of indicator scores. The final draft was also approved by the GMC.