Fair access to medicine? Retrospective analysis of UK medical schools application data 2009-2012 using three measures of socioeconomic status

Postcode-assigned IMD is the most commonly used measure of SES in this context in the UK, but may be very misleading, since the majority of applicants had at least one parent in NS-SEC 1 across the entire distribution of IMD [26]. It is also notable that the observed social gradients in terms of the application ratio are much larger for the individual NS-SEC measure than postcode-derived IMD and school type, indicating that the latter two are likely to overestimate how wide participation is. However, irrespective of the measure used, it is clear that school-leaver applicants to medical courses are predominately drawn from the more affluent, and this largely drives the distribution of applicants with accepted offers. There are marked differences between medical schools in terms of the mix of applicants they attract, with almost a fourfold variation in the proportion of applicants from NS-SEC 4/5. Selection processes do therefore matter, but there will not be large-scale widening participation in medicine in the UK unless applicants come from a broader range of the population. There are also differences in applicants and those with accepted offers between the UK countries, with England appearing to have a more equitable distribution than the other three countries (although differences in IMD and school systems between countries may at least partly account for this).

This analysis shows that the dominance of medicine by the more affluent [8, 11, 27] is persistent despite increasingly intense activity over the last decade intended to widen participation [2]. Direct comparisons over time are difficult because SES measures have changed, but new measures like IMD and NS-SEC have the same problems as older measures. Do and Parry’s examination of English medical school applicants between 1996 and 2003 using older measures of SES (Townsend Score and Registrar General Classification) demonstrated that 60 % of applicants were from the professional/managerial class, and that while there was a positive association between parental occupation and Townsend score this was far from perfect [28].

A strength of the study is the use of linked UCAS and UKCAT data, ensuring that multiple measures of SES were available for the majority of medical schools and a large pool of applicants. Limitations include that the analysis presented does not account for ethnicity or gender. The interaction between ethnicity, gender and SES is of interest but was not the focus of this study, reflecting that widening participation in the UK is largely framed in terms of SES. The findings are consistent with low participation being largely driven by low rates of application from the least affluent, although this will partly reflect variation in academic achievement by SES which was not examined [29]. The analysis also shows that the mix of applicants and selection outcomes vary between medical schools implying that widening participation is possible, and previous analyses have shown that medical schools that more actively use UKCAT test scores in selection admit more applicants from lower SES backgrounds [30]. Eight UK medical schools were not in the UKCAT Consortium during the period of analysis, although in practice a large proportion of applicants will apply to at least one UKCAT Consortium school and will therefore be included. Finally, the analysis is only of UK domiciled ‘secondary school leaver’ applicants. However, measuring SES in graduate entrants is fraught with difficulty, because from a widening participation perspective it is childhood or family SES rather than current SES which is of interest, but existing data collection systems are based on current, individual postcode or income. Additionally, Mathers et al in their analysis of successful applicants to medical school between 2002 and 2006 reported the socio-economic profile of medical students did not seem to have been affected by graduate entry programs [31].

There are two key implications of this research. First, modifying selection processes is unlikely to have major impact on widening participation because so few people from less affluent backgrounds apply in the first place. Contextualised selection processes may be worthwhile and the evidence that individuals from non-selective schools outperform their independent school counterparts with the same A-level or equivalent grades [32] provides some rationale for varying requirements for entry based on socioeconomic background. From a widening participation perspective, more attention therefore needs to be paid to supporting the process of applicants “getting ready” (considering a medical career and preparing to apply) as well as “getting in” (what happens during selection from those who apply) [8, 16]. The large variation between medical schools in terms of the applicants they attract, and the proportion of applicants from low SES backgrounds who accept offers may indicate that some medical schools have implemented effective strategies to widen participation and better understanding of why these differences exist could help define and disseminate best practice. Second, there is no ideal measure of SES available. Consistent with our results, the experience of the Kings College London Extended Medical Degree Program (EMDP) illustrates this. EMDP applicants are drawn from schools with low average educational attainment in inner London, and represent a significantly broader demographic of applicants than non-EMDP, but one third of enrolled EMDP students were still from professional, middle class backgrounds [33]. Postcode and school type measures are also not straightforward to apply consistently across the four UK countries because the indices of multiple deprivation (IMD) used in each country are not strictly comparable [19] and school systems vary considerably. One reason this matters is because medical schools disproportionately recruit from their local populations, making publicly available league table comparisons of medical schools potentially problematic if based on inconsistent measures [34]. Individual measures like NS-SEC or an assessment of household income avoid this problem, but are not universally collected and the validity of data about parental occupation collected from teenagers is uncertain [26]. Additionally, were such data to be used in selection, then it would inevitably be open to gaming, although postcode of residence is not immune to this either. However, measuring the success or otherwise of widening participation will be difficult in the absence of consistent and robust measurement of individual SES, which will require gathering potentially sensitive data from all applicants and will have resource implications [3].

At present there is no set quota which medical schools must achieve in terms of recruiting for lower SES. However the Selection for Excellence Group (SEEG) set up by the Medical Schools Council have for the first time set ten year targets for medical schools aimed at increasing the percentage of students from a lower SES. These targets are based on the POLAR3 measure [35], which like IMD is a geographic measure, based on small area rates of participation by young people in higher education. It is therefore likely to have similar problems to IMD in relation to medical school applicants not necessarily sharing the characteristics of most residents of the postcode they live in. Indeed the SEEG report highlights that there are challenges with using this measure, including the fact that it is unable to provide a ‘granular view’ of geographic location [35]. Additionally like IMD, POLAR3 may not be directly comparable across the four UK countries due to differing definitions of what constitutes further or higher education. The SEEG report recommends that work is required to ‘develop additional targets for widening participation that utilise different data sets’. They also recommend that currently medical schools should be using ‘more than one source and different type of contextualised data in their admissions process’ and that they should ‘triangulate data to ensure the individuals they identify are truly from a widening participation background’. In the short term, medical schools are likely therefore to use the POLAR3 measure and to triangulate data such as parental occupation, IMD and school type in their admissions processes. Based on the results of this study and recommendations in the SEEG report, research is urgently required to develop a valid and reliable measure which can be used to compare applicants to and students accepted to medical schools across the four UK countries.