Abstract
Background Clinical empathy may enhance patient satisfaction and health outcomes. The interaction between ethnicity and clinical empathy is underexplored.
Aim To compare how people from different ethnicities perceive GPs' communication of clinical empathy.
Design & setting A 2 (consultation style) x4 (participant ethnicity) web-based experiment using film vignettes in the UK.
Method In total, 274 UK Black British, White, East Asian, and South Asian adults (50% female; mean age 39.7 years), recruited from an online participant pool, were randomly assigned to watch one of two films showing enacted GP consultations for osteoarthritis: enhanced consultation (high empathy) or standard consultation (limited empathy). Participants’ perceptions of clinical empathy were assessed quantitatively and qualitatively.
Results Across the whole sample and within all four ethnicities, enhanced consultations were rated as more empathic than standard consultations; there were no significant effects of participant ethnicity on ratings of empathy. Bayesian analysis confirmed an effect of consultation style and indicated there were no differences between ethnicities in ratings of clinical empathy. Qualitative comments talked about the doctor: (not) showing interest in the patient; responding with(out) respect; (not) conveying warmth, empathy, and hope; and (not) clearly explaining management options or clinical decisions. Participants of all four ethnicities commented on similar aspects of the enhanced and standard consultations.
Conclusion In this well-controlled experiment, the empathic communication skills modelled in the enhanced consultation were interpreted similarly positively by UK adults of Black British, White, East Asian, and South Asian ethnicities. Core elements of clinical empathy appear relevant and similarly valued across these ethnic groups.
How this fits in
Studies show there are differences in how people from different ethnicities experience clinician empathy. These differences are incompletely understood and could impact health disparities. This study shows that core elements of clinical empathy appear relevant and similarly valued by UK adults of Black British, White, East Asian, and South Asian ethnicities. If, as part of broader cultural competence, GPs adopted the skills modelled in this study then empathy could be communicated more effectively to UK patients.
Introduction
Clinical empathy involves healthcare practitioners putting themselves in their patient’s position, acknowledging their patient’s feelings, concerns, and expectations, and communicating that understanding through their behaviour.1,2 When healthcare practitioners communicate clinical empathy effectively their patients report higher satisfaction with health care3 and, potentially, improved clinical outcomes including reduced pain.4 However, much of the evidence for the effects of clinical empathy rests on data from White patients.5
Within the broader context of complex ethnic health inequalities in the UK6 and globally7 the interaction between ethnicity and clinical empathy is potentially very important but remains poorly understood. A systematic review of 40 US studies found that, compared with White patients, Black patients reported poorer communication in general and poorer empathy-related aspects of communication in particular, including positive talk, partnership building, and participatory decision making.8 A meta-analytic systematic review of 18 studies found that Black or African American, Asian, and Native Americans perceived their clinician to be less empathetic than White patients and these differences are increasing over time, although are not statistically significant.9 However, the primary studies in these reviews typically did not systematically sample patients from diverse ethnic groups, resulting in imbalanced and likely unrepresentative samples. In the UK, patient satisfaction in the GP Patient Survey is consistently lower among minoritised ethnicities; this may be partially attributable to (not) feeling treated with care and concern, a key component of empathy.10 In summary, there appear to be differences in how people from different ethnicities experience clinician empathy.8,9 Such differences could be explained by clinicians expressing empathy differently to patients of different ethnicities (for example, owing to implicit bias)11,12 and/or there might be fundamental differences in how people from different ethnicities perceive similar expressions of clinician empathy during medical consultations.
We aimed to compare how UK adults from different ethnicities perceive GPs’ communication of clinical empathy. The objectives were to test for differences in ratings of clinical empathy between people from four ethnicities: Black British, White, East Asian, and South Asian; examine how any ethnic differences in ratings of clinical empathy interact with the level of clinical empathy shown in a consultation; and explore how participants perceive high and low-empathy consultations.
Method
Design
A vignette-based approach tested how the same empathy behaviours performed by the same clinicians were perceived by people from different ethnicities.13
A 2×4 between-participants experiment was completed by UK adults. The independent variables were participant ethnicity (four levels: Black British, White, East Asian, South Asian) and consultation style (two levels: enhanced, with high levels of empathy and optimism; standard, with limited empathy and optimism). The dependent variable was participant perceptions of clinician empathy. Covariates were consultation gender (female patient and GP; male patient and GP) and participant gender (female; male).
The experiment was hosted on Qualtrics, a cloud-based survey platform, which enabled us to display participant information sheets and document informed consent; display films to participants; randomly allocate participants to different films; collect responses to closed and open-ended questions; and seamlessly recruit from an online participant pool.
Participants
Inclusion criteria were UK adults (aged ≥18 years) of Black British, White, East Asian, or South Asian ethnicity. Participants were recruited via Prolific (www.prolific.com) (accessed 12.01.26). Prolific is an online research participant pool providing quick access to a large pool of adults seeking online surveys for payment. Prolific verifies email address, phone number, and photo ID, helping enhance the trustworthiness of this participant pool. Prolific outperforms other crowdsourcing online participant pools in terms of data quality.14
We used Prolific’s pre-screening function to invite participant pool members likely to meet our inclusion criteria. Eligibility was confirmed via screening questions on Qualtrics, presented immediately after the participant information sheet and consent. No support was provided for participants with low English proficiency.
Without sufficiently similar work to inform a power calculation, we assumed a medium effect size of ethnicity on perceptions of clinician empathy. To detect a medium effect size (f = 0.25) in one-way analysis of variance (ANOVA) with four groups at alpha = 0.05 requires 45 participants per group.15
Materials
Measures
Participants reported age (in years), gender (male, female, I define myself in another way, I prefer not to say), ethnicity, highest level of education completed, global subjective health (excellent, very good, good, fair, poor), and whether they had osteoarthritis or cared for someone with osteoarthritis. These items were chosen to facilitate sample description and comparison with the UK population; additional measures of socioeconomic status were not collected to minimise participant burden.
The validated and reliable 10-item Consultation and Relational Empathy (CARE) scale measured participants’ ratings of the GP’s clinical empathy in the filmed consultation.16 Each item has six response options: ‘poor’, ‘fair’, ‘good’, ‘very good’, ‘excellent’, and ‘does not apply’; responses were converted to a numerical score and summed to produce an overall rating of clinical empathy ranging from 0 (none) to 50 (high). ‘Does not apply’ responses were replaced with the participant’s average response across completed items. The CARE’s internal consistency was very high in this sample (Cronbach’s alpha = 0.985).
To explore participants’ broader views, two open-ended questions invited participants to share what they liked and disliked about the consultation.
Consultation films
Four films were professionally acted and produced for the EMPathicO e-learning package on empathy and optimism for primary care practitioners.17 Each film lasted approximately 8 minutes and showed a patient consulting a GP for knee pain, diagnosed in the film as osteoarthritis (a painful and disabling condition commonly managed in primary care;18 (see Supplementary Material for transcripts). Four films represented all combinations of consultation style and GP–patient gender: enhanced consultation (high empathy and optimism), male GP, male patient; enhanced consultation, female GP, female patient; standard consultation (limited empathy and optimism), male GP, male patient; standard consultation, female GP, female patient. All actors were White .
Procedure
In June 2023–July 2023, an advert on Prolific directed potentially eligible individuals to Qualtrics where the participant information statement was displayed with a checkbox to indicate consent. Potential participants then completed screening questions. Eligible participants completed the demographic and health questions, then were randomly allocated to watch one of the four consultation films (stratified by participant ethnicity). After watching one film, participants completed the empathy items and reported any interruptions, technical or other difficulties. Participants were thanked and redirected to Prolific for payment.
Analysis methods
Self-reported ethnicity, gender, and age collected via Qualtrics were compared with responders’ prolific profiles; any with discrepancies were excluded. Responders were excluded if they reported stopping watching before the film ended (n = 1), were interrupted while watching (n = 11), or completed the study quicker than film duration (n = 5). A factorial ANOVA tested for (fixed) effects of consultation style and participant ethnicity on CARE scores. Including participant and actor gender as covariates did not alter the inferences. Inspection of Q-Q plot and histogram confirmed residuals approximately followed the Normal distribution. Effect sizes indicated by partial eta-squared are interpreted according to Cohen.19,20
In traditional hypothesis testing, a non-significant effect (for example, of factor x) on a measure (for example, measure y) means there is no evidence of an effect of factor x on measure y. This is not the same as evidence that there is no effect of factor x on measure y. Rather, it could mean either that the study was insufficiently powered to detect an effect or that there is no effect to be detected. Traditional hypothesis testing cannot distinguish between these explanations, but Bayesian statistics can supplement traditional approaches by generating evidence that there is no effect of factor x on measure y.21 Table 1 summarises this approach to null effects.22,23
Following a non-significant effect in the traditional ANOVA, we conducted a 2×4 Bayesian ANOVA using JASP (version 0.19.3). We used the default prior option of 0.5 for the fixed effects; this is an appropriate approach to modelling when there are no strong (empirical) grounds for choosing other priors. Sensitivity analyses run with priors of 1 and 0.2 did not change the conclusions.
Qualitative data were read repeatedly before being classified as positive or negative about the consultation, and then further categorised into inductive, data-driven categories. Most comments were very brief and many mentioned multiple aspects of the consultation (for example, body language and advice); categorisation was not mutually exclusive (that is, one comment could be placed in multiple categories). Categories were then mapped to participant ethnicity and comments were compared qualitatively. NVivo (version 12) was used to facilitate categorisation and to compare the nature of comments made by people from different ethnicities. The qualitative analysis was led by RD and FB, who discussed the categorisations in depth. Quotations for this report were selected to clearly illustrate the categories and are attributed to individuals using participant numbers.
Results
Participants
There were 274 participants, aged between 18 years and 77 years (mean [M] = 39.7, standard deviation [SD] = 14.2). A slightly higher proportion of participants reported being White (32%) than Black British (23%), East Asian (20%), or South Asian (25%) (Table 2). The gender split overall was balanced between males and females; this was also the case for three ethnicities, but there were more males than females among South Asian participants (60% versus 40%). Overall, most participants across all ethnic groups reported having obtained a university degree. Most participants reported having good (36%) or very good (34%) general health; self-reported health was broadly similar across ethnic groups. Fewer than 10% of the whole sample had osteoarthritis themselves or cared for someone with osteoarthritis.
Effects of ethnicity on ratings of empathy
Across the whole sample and within all ethnic groups, enhanced consultations were rated as more empathic than standard consultations (Table 3). The traditional factorial ANOVA confirmed a large and statistically significant main effect of consultation style on ratings of clinical empathy, F(1,266) = 814.81, P<0.001, partial η2 = 0 .75. There was no main effect of participant ethnicity on ratings of clinical empathy, F(3,266) = 2.039, P = 0.109, partial η2 = 0.02. While there appeared to be greater variation between ethnic groups in ratings of the standard consultation compared with ratings of the enhanced consultation (Table 3), the interaction between consultation style and participant ethnicity was small and not statistically significant, F(3,266) = 2.087, P = 0.102, partial η2 = 0.02.
In the Bayesian analysis, all possible models were assessed and compared with the best-performing model (Table 4). The comparison between the model that included only consultation style and the null model (with no predictors) overwhelmingly favoured the consultation-style only model, BF01 1.784×10+81; this confirms the results of the traditional ANOVA that there was an effect of consultation style on perceptions of empathy. The comparisons between the consultation style-only model and models that included consultation style and participant ethnicity also favoured the consultation style-only model. The data were 8.53 times more likely to be observed under the consultation style-only model than under the model with main effects of consultation style and ethnicity and their interaction. Furthermore, the data were 4.05 times more likely to be observed under the consultation style-only model than under the model with main effects of consultation style and ethnicity. These comparisons constitute moderate evidence in favour of the consultation style-only model compared with the models also including participant ethnicity,23 that is, there is moderate evidence that there is no effect of participant ethnicity on perceptions of clinical empathy.
To further illustrate this, an analysis of effects was used. This gives the Bayes factor for whether consultation style, participant ethnicity, and the interaction, should each be included or excluded from the final model. There was overwhelming evidence that consultation style should be included in the model (BFincl >100), ‘anecdotal’ evidence that the interaction should not be included in the model (BFincl = 0.38), and moderate evidence that participant ethnicity should not be included in the model (BFincl = 0.24). Overall, the findings of the Bayesian analysis support the traditional ANOVA and add that the evidence strongly favours this interpretation: there was only an effect of consultation style, and that there were no differences in ratings of clinical empathy owing to participant ethnicity.
Participants’ qualitative comments about the consultations
Two hundred and nine participants (76%) commented on either positive (89, 32% of all participants), negative (54, 20%), or both positive and negative (66, 24%) aspects of the consultation they watched. Enhanced consultations received more positive than negative comments (96 versus 20), standard consultations received more negative than positive comments (100 versus 59); this pattern was consistent across participants of Black British, White, East Asian, and South Asian ethnicities.
Comments were categorised into talk about the doctor: (not) showing interest in the patient; responding with(out) respect; (not) conveying warmth, empathy, and hope; and (not) clearly explaining management options or clinical decisions (see Table 5). In addition to commenting on aspects of the consultations that they liked and disliked, some participants related the enacted consultations to real-life experiences (see Table 6). All categories included comments about similar issues by participants from all four ethnicities. In other words, there were no major differences in how participants from different ethnicities talked about the consultations.
Discussion
Summary
Participants from all four ethnicities included in the study perceived the enhanced consultations as significantly more empathic than the standard consultations (P<0.001). Bayesian analysis confirmed the effect of consultation style and further indicated that there was no effect of ethnicity on ratings of empathy. Qualitative comments were markedly more positive and less negative about the enhanced consultations.
Strengths and limitations
Participants rated enacted consultations; they were observers rather than personally experiencing a clinical interaction. Only a small proportion shared the health concerns of the patients in the films and this lack of personal relevance may have blunted their responses, although this is unlikely given the extreme scores on the CARE scale and strongly worded comments. Providing videos in this experimental design meant controlling for variables, including the patient’s clinical condition and individual GP variability, which would not be possible in a clinical setting. Thus, we traded ecological for internal validity. The actors in the filmed consultations were all White; this may have shaped participants’ views in unknown ways. We successfully recruited similar numbers of participants across four ethnicities and the gender split within ethnicities broadly reflects 2021 census data, except for South Asians where our sample contained more males than the census (60% versus 50%, respectively). However, our participants were more highly educated than the general population: only 0.4% of participants had no formal qualifications, compared with 18% of adults in the 2021 census24 and likely had high English language proficiency.
Future studies could consider the following: recruiting participants with no formal educational qualifications from more deprived populations with higher levels of morbidity, as deprivation and multimorbidity can impact perceptions of clinical empathy;25,26 examining the impact of English language proficiency on perceptions of clinical empathy across ethnicities, given documented impacts of English language proficiency on patient–practitioner communication;27 exploring whether and how intersectional positions (for example, different combinations of gender and ethnicity for patients and clinicians) might shape experiences of clinician empathy;28 including participants from other ethnic groups or using more precise ethnic categorisations (for example, Arab, Bangladeshi, Chinese, White Irish);29 and increasing the sample size to permit detection of small effects of ethnicity on ratings of clinician empathy and interactions between ethnicity and consultation style. A qualitative approach, for example, using the enacted consultations to stimulate focus group discussions, would potentially elicit more nuanced data.
Comparison with existing literature
Our findings suggest that relatively highly educated UK adults of Black, White, South and East Asian ethnicities with good self-reported health perceive similar levels of clinician empathy from the same clinician behaviours. This rules out one possible explanation for understanding the documented ethnic disparities in satisfaction with real-world communication and empathy in primary care.10,28 Specifically, our findings support the need for further research to explore whether some clinicians communicate differently to patients depending on patient ethnicity, potentially owing to implicit biases, and whether intersectional factors, such as educational level, multimorbidity and socioeconomic factors, are also important. Addressing implicit biases may help reduce communication disparities11,12 but should be complemented by relevant accessible practitioner training to enhance empathy skills and embed them in all consultations.
Implications for practice
It is widely acknowledged that not all clinical consultations have optimal communication. If as part of broader cultural competence, clinicians are enabled to adopt the skills modelled in the enhanced consultation film (and taught in the broader EMPathicO intervention),17 empathy could be communicated more effectively to UK patients of Black British, White, East Asian, and South Asian ethnicities.
Notes
Funding
No external funding was received for this project.
Ethical approval
Ethical approval received from Psychology Research Ethics Committee, University of Southampton (ERGO81788).
Provenance
Freely submitted; externally peer reviewed.
Data
The anonymised data that support the findings of this study are available in eprints http://eprints.soton.ac.uk at https://doi.org/10.5258/SOTON/D3611.
Acknowledgements
The authors acknowledge the contribution of the participants in this study.
Competing interests
The authors declare the following competing interests: FLB (research grant from NIHR School for Primary Care Research paid to institution; speaker’s honoraria from Stoneygate Centre for Empathic Healthcare and New Scientist). HAE (research grant from NIHR School for Primary Care Research paid to institution; is Deputy Academic Capacity Development Lead for the NIHR SPCR and sits on the NIHR SPCR Board and Exec; works clinically as a GP at New Horizons Medical Partnership in Southampton as part of Professor of Primary Care Research post at the University of Southampton; Co-Authors the Oxford Handbook of General Practice published by Oxford University Press).
- Received April 17, 2025.
- Revision received June 3, 2025.
- Accepted June 17, 2025.
- Copyright © 2026, The Authors
This article is Open Access: CC BY license (https://creativecommons.org/licenses/by/4.0/)
References
In this issue
LinkedInJump to section
More in this TOC Section
Related Articles
Cited By...
