General practice and ethnicity: an experimental study of doctoring

Inequalities in health status for ethnic minority groups are well documented, with ethnic minorities consistently faring worse in terms of morbidity and mortality [1‐3]. Moreover, in recent years, the gap in mortality rates has widened critically for many illnesses [4].

Many factors have been suggested to explain these health inequalities. Shavers listed the factors most commonly assessed in research: access to health care; socio-economic status; lack of knowledge of the health care system; cultural beliefs or preferences, and race/ethnicity-based discrimination [5].

Smedley concluded that ethnic discrimination existed at all levels of health care, from the policy level to the level of the individual provider [6]. Smedley’s research showed that for a wide range of medical procedures, ranging from highly technological interventions to basic diagnostic and treatment procedures, minority patients are less likely to receive adequate medical procedures and more likely to experience a poorer quality of medical care than non-minority patients. This pattern of differences remained true even when studies adjusted for differences in patients’ health insurance, socio-economic status, the stage and severity of the disease, and comorbidity.

Pager suggested three major mechanisms as underlying contemporary forms of ethnic discrimination: the individual level, the organizational level, and the structural level [7]. In this paper, we focus on the individual level of ethnic discrimination in health care. According to van Ryn, negative stereotypes that doctors have about ethnic minorities are likely to play a role in discrimination during medical encounters [8]. Explicit and implicit biases on behalf of providers can lead to discrimination in health care for ethnic minorities and, ultimately, to disparities in health [9‐12]. These stereotypes negatively influence the course and the outcomes of clinical encounters and lead to distrust within inter-ethnic medical interactions [11, 13].

Tajfel described “in-group bias”, and showed, through a set of pioneering experiments, that humans tend to favor in-group members over out-group members in the distribution of rewards, even when these groups are arbitrarily categorized [14]. The idea is that we like, and are more motivated to help, people that we think are like us [9]. Even though doctors are bound by professional and ethical standards to help and care for vulnerable patients, they may unknowingly stereotype members of ethnic minorities, particularly when working under conditions that diminish cognitive capacity, such as time pressure, fatigue, and information overload [15].

There is extensive evidence to indicate that patient ethnicity modifies conscious and unconscious beliefs and practices on the part of doctors [16‐19]. Doctors may be unaware of the fact that they hold implicit attitudes and stereotypes about ethnicity, but these implicit social biases can significantly influence quality of care [19]. According to van Ryn, these biases affect doctors when it comes to decision-making, interactions, and inter-personal behaviors [8]. For example, there is evidence that doctors contribute to ethnicity inequities in access to kidney transplants, cardiac procedures, psychiatric care, and pain control [8]. Ethnic minority patients are also more likely than non-minority patients to be recommended for unnecessary surgery [20]. Another major dimension of doctoring that is affected in inter-ethnic consultations is communication and the doctor-patient relationship [21]. For example, Roter revealed that when dealing with black patients, American physicians were more likely to adopt a “narrowly biomedical” communication pattern, characterized by low levels of patient control over communication, low levels of psychosocial talk, and high levels of biomedical information-giving [22]. Johnson found that doctors were less patient-centered, more verbally dominant, and expressed less positive affect with African-American patients compared to white patients [23]. Kaplan and Cooper found that minority patients rated their physician as having a less participatory decision-making style than did non-minority patients [24, 25].

There has been little research focusing on ethnic discrimination in general practice. Indeed, most studies on this topic have been conducted within the field of Specialized Medicine. However, GPs are often the primary access to health care for ethnic minority groups, and play a ‘gatekeeper’ role within the health care system [26]. An investigation into the issue of ethnic discrimination with respect to general practitioners is therefore relevant [27]. Moreover, most of the research and most of the publications on ethnic discrimination in health care have come from the U.S.A., a societal context that is characterized by a long history of migration and racial segregation. Our study was conducted in Belgium, which provides a different societal, historical, and cultural context, as well as a different organization of the health care system. It has been shown that different contexts may lead to different medical practices in terms of ethnic discrimination. For example, one study concluded that the Danish health care system was equitable for ethnic minority groups [28]. In 2011, the resident foreign population represented 10.2% of the entire population of Belgium [29, 30]. The principal ethnic minority groups living in Belgium are Moroccans, Turks, and Congolese [30].

The main aim of this study was to investigate ethnic discrimination in general practice. Our general research hypothesis was that GPs would be more favorable towards non-ethnic minority patients (in-group members) and less favorable towards ethnic minority patients (out-group members). We used the van Ryn model to analyze the effect of patient ethnicity on GPs. This model suggests that there are two major pathways that lead doctors to give different treatment during medical consultations. The first pathway is the cognitive or decisional dimension, the second one is the interpersonal dimension [8]. In this paper, we assess whether patient ethnicity implies systematic differences in the medical care patients receive. In order to test our research hypothesis, we considered the outcome relative to GPs’ decision-making (assessment, recommendation, and prescription) and the outcome relative to the relational aspect of care (time devoted by GPs to examining patients’ medical case history), as suggested by van Ryn. Although the relational dimension of doctoring can be evaluated using many factors, in this study we will measure only one of these: time devoted to patient. It has been argued that consultation duration can be considered to be a generic enabling resource and a reliable indicator of the quality of the relationship and quality of care in general [31‐37]. We also looked at whether GPs’ working conditions led to differential medical care for ethnic minority patients.

This is one of the first experimental studies to investigate ethnic discrimination in general practice. Since GPs are ‘gatekeepers’ to the health care system, this issue is particularly relevant. Based on van Ryn’s model, we hypothesized that patient ethnicity would affect the quality of medical care through (a) biased medical decisions by GPs and (b) biased relational involvement on the part of GPs.

Before beginning a decision-making process, GPs evaluate a patient’s cardiovascular risk. We observed that cardiovascular risk assessment was not related to patient ethnicity. The next step is the medical decision and recommendation. Again, ethnicity was not related to ECG procedure recommendation. However, we detected that drug prescription was significantly more frequent for ethnic minority patients than for non-minority patients. These results are inconsistent with our first hypothesis, since ethnicity did not lead to an obvious decrease in the quality of medical decisions or recommendations. On the contrary, ethnic minority patients received more pharmacological treatment than non-minority patients. This result is surprising because previous studies have usually described fewer drug prescriptions for ethnic minority patients suffering from cardiovascular diseases than for non-minority patients [40]. So why did GPs in our study prescribe more drugs for ethnic minority patients, when the cardiovascular risk was equivalent in both groups? One possible explanation is that GPs may consider ethnic minority patients to belong to an at-risk population, due to their ethnicity, and therefore prescribe medication as a protective measure. GPs may have considered that most cardiovascular risk factors are more prevalent in ethnic minority populations than in the rest of the population [41]. For example, in Belgium, ethnic minorities are more likely to be living in a poorer socio-economic environment than the general population [42, 43]. The general health status of ethnic minorities in Belgium is worse than that of the general population [44]. Ethnic minorities are more exposed to cardiovascular risk factors and therefore incur higher cardiovascular morbidity and mortality [45, 46]. Research indicates that doctors over-apply population statistics to individual patients [47]. Population statistics may have a similar function to stereotypes in decision-making by doctors. Our results may therefore suggest that decisions made by GPs for ethnic minority patients are based on over-generalizations rather than on the actual individual assessment of the patient. This phenomenon suggests that GPs are more likely to perceive ethnic minority patients (out-group members) in terms of their group stereotype and pay less attention to their individual characteristics. An alternative explanation could be that GPs may prescribe drugs in order to avoid becoming involved with lifestyle modification, which is a more demanding clinical activity requiring more inter-personal communication. For example, studies show that doctors with a longer mean consultation time explore the psychosocial dimension of illness in greater depth and prescribe fewer medications than doctors with shorter consultation times. The same authors also showed that, interestingly, the doctors who were devoting more time to patients were those who were experiencing higher stress levels associated with “running behind schedule” [33, 34]. We speculate that, when facing ethnic minority patients, GPs may devote less time to them and prescribe them more medication as a coping strategy to avoid engaging in doctor-patient relations.

As regards the relational dimension of doctoring, ethnicity had no effect on the amount of time devoted by GPs to examining the patient’s medical history. However, we found that GPs experiencing high volumes of activity devoted significantly less time to medical history examination for ethnic minority patients than did GPs with lower volumes of activity. This result is consistent with previous research that indicated that time pressure, cognitive load, necessity of prompt decisions, task complexity, busyness, distraction, fatigue, and anxiety increase the likelihood of stereotype usage [48‐52]. Our findings therefore support the hypothesis that ethnic discrimination is likely to be context-dependent. Baron and Pfeffer showed that organizational practices mediate the cognitive biases and stereotypes of actors [53]. In addition, Reskin states that “the proximal cause of most discrimination is in the personal practices of work organization that polarize the biasing effects of automatic cognitive processes” ([54], p.320). The underlying idea is that discrimination is expressed only if certain organizational conditions are met [55]. We observed that the amount of time devoted by GPs to examining socio-relational history was significantly lower for ethnic minority patients. These results are consistent with Roter and Johnson’s research, which revealed that doctors engaged in less psychosocial talk and more biomedical information-giving during inter-ethnic consultations [22, 23]. They are also consistent with Cooper, who found that race-discordant medical interactions were significantly shorter compared to race-concordant medical interactions [56]. When interacting with ethnic minority patients, GPs are thus focused on medical aspects of doctoring and put aside the socio-relational dimensions of doctoring. This is supported by previous findings. Doctors engaged in inter-ethnic consultations have been shown to be more reluctant to engage in partnership-building and emotional considerations [25]. The literature on doctor-patient communication supports the idea that white doctors are less comfortable when interacting with patients of other ethnic groups [9, 25].

There were at least three limitations to our study. Firstly, the participation rate was low (45%). The representativeness of our sample is therefore limited. Our GP sample was over-representative of younger, female, and group-practice GPs when compared with the overall population of GPs in Belgium. This may mean that we surveyed a more “migrant friendly” group of GPs, since younger and female GPs have been shown to be more patient-centered and more concerned about vulnerable patients, compared to older, male doctors [57]. Secondly, case vignettes are not real clinical situations and it is difficult to determine whether the relational engagement with patient and medical decisions by GPs in case vignettes represent their true clinical activity. Thirdly, the relational dimension of doctoring involves more than the time devoted to the patient, which is only one of many aspects. Further studies, using more ethnographic methods, could consider other factors such as participatory style, warmth, empathy, and information giving/seeking, which are also essential aspects of the relational dimension of doctoring.

Good doctor-patient relations are essential for the effective management of chronic illness in general practice. In this study, we found that GPs self-report devoting less time to ethnic minority patients than to non-ethnic minority patients. This may suggest that ethnic discrimination in general practice occurs primarily through the relational dimension of doctoring. However, as our study design was experimental, our results require confirmation through the use of observational data and routine health care databases. We also observed that GPs in a constraining environment (high rate of patient contact) devoted less time to examining medical history for ethnic minority patients. This last finding suggests that ethnic discrimination is associated with resource allocation priorities and is therefore a context-dependent phenomenon. Further research should, firstly, aim to corroborate these ethnic differences in medical decisions with routine databases that allow GPs’ prescriptions, patient clinical status, and patient ethnicity to be tracked. Secondly, further research should seek to understand why GPs engage less in the relational dimension of doctoring with ethnic minority patients. Third, could also attempt to identify key contextual variables that enhance the risk of ethnic discrimination. Further research is also necessary to replicate and confirm our findings. Research into the issue of ethnic discrimination in general practice potentially has important public health benefits. GPs are the gatekeepers to specialty care and play an important role in patient pathways. Discrimination may also magnify ethnic inequalities in health. Our study highlights the risk of inappropriate drug use in some ethnic-specific encounters.