What are the underlying reasons behind socioeconomic differences in doctor‐patient communication in head and neck oncology review clinics?

Abstract Objective To explore socioeconomic status (SES) differences in patterns of doctor‐patient communication within head and neck cancer clinics and why such differences exist. Methods Thirty‐six head and neck cancer review appointments with five Physicians were observed and audio‐taped, along with follow‐up interviews involving 32 patients. Data were analysed using Thematic Analysis, and compared by patient SES (education, occupation and Indices of Multiple Deprivation). Results Three main themes were identified: (a) Physicians used more humour and small talk in their consultations with high SES patients; (b) Low SES patients were more passive in their participation, engaged in less agenda setting and information‐seeking, and framed their clinical experience differently; (c) Low SES patients had different preferences for involvement, defining involvement differently to high SES patients and were seen to take a more stoical approach. Conclusion Low SES patients take a more passive role in medical consultations, engage in less relational talk and are less likely to raise concerns, but were satisfied with this. Physicians may adapt their communication behaviour in response to low SES patients’ expectations and preferences. Practice Implications A question prompt list may help low SES patients to raise concerns during their consultations. This may reduce inequalities in communication and health.

pate in consultations by asking questions and expressing opinions and concerns; and doctors involve patients by encouraging them to ask questions and express opinions or preferences. 7 Such input from patients is necessary in order for shared decision-making to occur. 7 Studies show that there are socioeconomic status (SES) differences in doctor-patient communication. Patients from low SES backgrounds tend to participate less actively in their consultations by asking fewer questions and are less likely to volunteer information unprompted or express their opinions or emotions. 8,9 Observations also show that clinicians give less information, spend less time building rapport, and listen less attentively to these patients. [8][9][10][11] Thus, a patient's socioeconomic position affects both the way in which they communicate with doctors, as well as how doctors communicate with them. This is likely to contribute to socioeconomic inequalities in health.
There are a number of theories which have been proposed to explain these socioeconomic differences in doctor-patient communication. Status characteristics theory posits that patients' and doctors' perceived differences in social status influence how much they participate in the consultation, and the value placed on their contributions. Status is determined by characteristics such as expertise, qualifications, SES, ethnicity and gender. The higher an individual's status, the more opportunities they are given to participate and the more their participation is valued. 12,13 Thus, patients who believe their status to be low may deny themselves participation, and doctors perhaps unconsciously deny participation to patients of lower status. Lay theories of social class provide a slightly different perspective. Low SES individuals who believe their social status is biologically determined and unchangeable may experience more shame, anxiety and negative affect than those who do not because of fatalism, thus reducing participation in consultations. 14 However, there has been a lack of research to explain these socioeconomic differences. Furthermore, the impact of SES on the effectiveness of the doctor-patient relationship and communication in head and neck cancer has not been well studied. 15 Malignancies of the head and neck are the 8th most common cancer in the United Kingdom; approximately 4000 die annually from the disease. 16 Previous work has demonstrated an association between low SES and unfavourable health-related quality of life outcomes after a diagnosis of head and neck cancer, 17,18 although there has been little attention in the doctor-patient communication literature. To fill this knowledge gap, the objective of this study was to explore differences in doctor-patient communication practices based on patient SES and why any differences exist using a qualitative design.

| ME THODS
The setting for the study was head and neck oncology review clinics. After obtaining ethics board approval, patients over the age of 18 years who had previously completed treatment for cancer of the head or neck during the previous 12-24 months were recruited for study participation from one of five specialty clinics at a tertiary care hospital in the North-West of England. Each of these clinics was run by male physicians, and 36 patients were recruited in total. Non-English-speaking patients, those unable to provide informed consent, and those whose SES could not be determined were excluded.
Patients were approached to take part in the study while waiting for their consultation. Informed consent about the study was taken.
A researcher observed and audio-taped the patient's head and neck review appointment with their Physician, sitting in the corner of the room. The researcher's field notes were kept to record phenomena not available to audio recording. In particular, notes were made on body language such as eye contact, whether the patient and Physician shook hands, facial expressions, etc Appointment length was timed from introduction to closing statements by either patient or surgeon. A semi-structured interview with the patient was arranged a few days later, which was undertaken by telephone or faceto-face. The consultation recordings were used to derive questions for each follow-up interview and personalize the topic guide, so that the interviewer could ask each patient about specific things which happened during their appointment (eg, You mentioned you were having difficulty swallowing, how did you feel about the physician's response to that?). The researcher engaged in specific training in qualitative observation and coding at formal workshops and though private study.
Thirty-two patients participated in post-consultation interviews.
Three patients could not be contacted by the researcher following the appointment, despite repeated attempts, and one patient declined to be interviewed following their consultation. There were consequently 68 data sources (36 appointments and 32 linked interviews). Recruitment was stopped when a spread of participants by SES characteristics was achieved, and when saturation of data was reached.
We collected multiple datapoints on each patient to describe SES, including indices of multiple deprivation (IMD) based on postcodes, patient occupation mapped to the Office for National Statistics standard occupational classifications (SOC), and patient-reported education data and employment status. IMD decile one represents the most deprived postcodes. 19 SOC is a commonly used classification of occupational data which is collected and routinely updated by the UK government, and has been used as an SES measure by a number of studies. [20][21][22] It consists of nine groups, with group 9 representing the lowest SES. 23,24 These data were aggregated for each patient, as available, to categorize study subjects into 'low' or 'high' SES. Details regarding patient SES categorization are available in Table 2.
Audio recordings of appointments and interviews were transcribed and, with fieldnotes, organized using the NVivo 10 software, analysed using thematic analysis. This analysis involved identifying common conversational characteristics that were combined into themes and sub-themes, which together addressed the research questions. Differences by SES background were identified using a process of constant comparison between low SES and high SES cases. This involved seeking out contradictions in themes and subthemes as cases were added to the analysis, and amending themes until adding new cases to the analysis did not result in any reframing of the emergent themes. A researcher (SA) undertook initial coding and generated an initial thematic framework which was refined following discussion with other authors (RH, SB). Analysis took place alongside data collection. Each patient was assigned a number to anonymize the data. These codes used are to denote participants in the results below.

| Sample characteristics
Although 36% of participants lived in IMD decile 1, there was also a spread of participants across the SES gradient. The majority of participants were men (n = 24, 66.7%), with the most common primary treatment being surgery alone (n = 13, 36.1%), closely followed by chemoradiotherapy alone (n = 10, 27.8%). Average age was 64.9 years (Table 1). Table 2 shows the distribution of patients across SES categories.

| Overview
We identified three main differences by SES in patterns of communication, which were sub-divided into seven sub-themes in all (Table 3). We noted that consultations differed by SES in firstly the extent of relational talk; secondly in the extent to which patients ap-

| The use of humour
We observed that the tone of consultations was warmer and more familiar when patients from higher SES groups were seen. Doctors were friendly and talkative towards patients in higher occupational classes, spending time to recognize and build rapport in the relationship between the two. The use of humour in these interactions was particularly characteristic, with banter usually initiated by the doctor. In Table 4

| Small talk
In contrast, doctors seemed much colder in their interactions with lower SES patients, providing very little space for patients to speak and not attempting to initiate or encourage any sort of rapport in the same way. Physicians did much of the talking, most of which was centred around the medical aspects of the consultation with very little small talk or attempts at humour. In Table 4  Medical discourse is known to involve a dialectic between institutional (eg medical) frame and socio-relational frame for talk. 25 Thus, the consultation can be divided into task (instrumental) talk and relational (small) talk. 25 We found that a lower proportion of the appointment was given to relational talk for low SES patients. This difference in proportion of time spent on small talk could in part be due to differences in total appointment time. Table 5 shows the mean appointment lengths for each IMD decile. Barring a few outliers, patients from less deprived areas appear to generally tend to have longer appointments, although because the sample size of this qualitative study was relatively small, it was not appropriate to use statistical tests to investigate this further.

| Active or passive participation
We observed that low SES patients were comparatively passive in consultations: Physicians generally took the lead, with low SES patients raising relatively few questions and new topics of discussion. This pattern appeared to be influenced by three factors: firstly by patients' with knowledge and experience gained in higher education or in their occupation consciously applying this to 'oil the wheels' of the interaction; secondly by higher SES patients coming to the consultation with their own agenda; and thirdly because lower SES saw it as the Physician's responsibility to give information-rather than it being the patient's responsibility in obtaining it.

| Education and occupation as a cultural frame of reference
Eleven patients referred to their education and occupation during interactions. This was spread evenly across SES. However, SES differences were seen in the ways in which they referenced education and occupation. More educated patients and those with insights into the workings of health services by virtue of their occupation appeared to consciously apply this knowledge to help navigate the healthcare system and to reduce the doctor-patient difference in status. The example given in Table 4  This is in contrast with a patient with lower SES ( This was in contrast to low SES patients, who took a more passive approach as illustrated in sub-theme 2.2 in Table 3 (Patient 7): waiting for the Physician to set the agenda which was then focused on medical, or task aspect of the appointment. Low SES patients also appeared to be relatively more reticent in raising emotional concerns as an appointment agenda item.

| Responsibility for obtaining information
When asking whether they received sufficient information, we found that higher SES patients actively sought information from the Physician, even outside of the consultation if they had forgotten to ask something during their appointment-seeing this as their responsibility: if someone has not received sufficient information then this was seen as their fault (see Patient 3, Theme 2, sub-theme 2.3, Table 3). This contrasted with lower SES patients who saw it as the Physician's role to provide them with information which the Physician deemed to be important (see Patient 6,

| Defining involvement in decision-making
When patients were asked how involved they were in deciding what cancer treatment to undergo and whether they were happy with that level of involvement, we found that there were SES differences in how patients defined involvement. Lower SES patients interpreted 'involvement' as being fully informed of any decisions the Physician made regarding their healthcare (see Table 3, Patient 6, Theme 3, sub-theme 3.1). They seemed uninterested in being involved in making decisions, which was also reflected our observational data: Physicians made generally decisions for them without consulting them, and they seemed happy with this.
In contrast, patients with higher education levels were more interested in being a part of the decision-making process. Physicians also involved them in this process, talking through the various options available to them and the pros and cons of each so that these Erm…I have always been explained to me why they are doing things and initially when the treatment was. When I was diagnosed and they said to me, "Well there is two courses of treatment," he said "There is the tried and tested one or there is a new erm…one that they are trialling"… But obviously its not as, they do not know the results, so he said: "What do you want to do?", he said "It's got to be your decision" patients could make an informed decision. This is illustrated in data from Patient 3, Table 3, Theme 3, sub-theme 3.1 where the patient talked about the conversation which took place with her Physician when she was diagnosed where both discussed two alternative treatments, one of which was undergoing a clinical trial.

| Stoicism
Stoicism is the endurance of pain or hardship without the display of feelings and without complaint. 26 In terms of complaint, all patients brought concerns to physicians' attention and the degree of stoicism did not appear to differ across SES. Few complained about pain or discomfort in their interviews or appointments. However, while no higher SES patients referred to their own stoicism, several low SES patients explicitly referred to stoicism as a positive and self-defining value. This in illustrated by data from Patient 13 (Table 3, Theme 3, sub-theme 3.2) who had an outlook on life that 'You just have to get on with it'.
This contrasted with patients with higher education levels who were keen to talk through any issues with Physicians in order to help them deal with these problems. They not only wanted to make the Physician aware of their problems but also to understand them for themselves (Patient 12, Table 3 Patient: 'And my mother was quite a strong woman. You know she, we were never mollycoddled as children and we were expected… They were loving parents but we were expected to err…to get on with it. As they had,

| Discussion
In summary, firstly we found SES differences in sociable talk, that is, relational communication between Physician and patient. Although relational talk is recognized to be an important activity in itself, since the cocreation of a 'relational climate' promotes patient cooperation with care 29 ; our study found Physicians appeared to spend more time engaged in this, where patients were closer to their own social position. We particularly identified differences in the amount of humour within consultations. Previous research has found that individuals use humour and teasing to create power and solidarity within an interaction, with teasing particularly used in interactions between individuals of the same gender, 30 although since our study involved only consultations with male Physicians we recognize that there may be gender differences, possibly in line with gender concordance between doctor and patient, and so further studies of this would be helpful.
Our finding that patients from less deprived areas had longer appointments is consistent with the findings from our previous study which also showed a significant positive correlation between appointment length and SES as measured by IMD decile. 31 This may reflect the lack of much relational talk which occurred between Physicians and low SES patients. Coupland et al 32 report that both patients and doctors work to sustain the relational portion of the consultation which can delay the instrumental portion. 32 However, in our study we found that the relational portion appeared relatively reduced in consultations involving low SES patients, and instead, there was a prioritizing of the instrumental aspects of the consultation. Because of the qualitative nature of the study, with a limited sample size, we did not use statistical methods to investigate differences in relational talk by socioeconomic gradient, and so further (quantitative) studies would be needed to investigate the pattern which emerged in the data.
The finding from the sub-theme 'stoicism' that patients with lower education levels seemed to prefer to deal with problems on their own in private as a way of coping with their condition is sup- breast cancer patients found that low SES patients raised significantly fewer concerns during consultations with their oncologist. 9 Some studies have found that individuals from low SES backgrounds tend to expect negative outcomes, which can lead to hopelessness and chronic stress. 33 They engage in more fatalism and avoidance, as opposed to instrumental coping behaviours such as talking through issues with their doctor, displaying lower perceived control over events. 34,35 This is in line with the SES differences in the patterns of doctor-patient communication we observed in our study too.
Within the 'Defining involvement in decision-making' theme, we also identified that patients with less education were not only less likely to be involved in making decisions about their care, but that they defined involvement as being fully informed of decisions made rather than making decisions themselves. to know everything about it before it actually happened…' (Table 3).
Although previous studies which have also found that people from low SES backgrounds are less likely to seek information or have a preference for active involvement in decision-making, 36 Nevertheless, increasing patients' control over their own healthcare is increasingly seen as a good thing, 43 and something which contributes to better treatment outcomes. 1-5 Therefore, even if not driven by the need to satisfy low SES patients' expectations, there may still be a need to ensure that information exchange during consultations is more equitable by socioeconomic status. That our previous study found no SES differences in the number of concerns raised in head and neck cancer clinics when using a question prompt list 31 suggests that a pro forma approach might be an effective intervention to reduce socioeconomic inequalities in this area. The Patient Concerns Inventory (PCI) is a question prompt list which allows patients to select what topics they wish to discuss with their Physician, prior to their appointment. This is then used to streamline the consultation. 44 Studies have found that it is feasible to use with both elderly patients and those with little education, 45,46 and currently, a trial is being conducted to examine whether its long-term use may improve patients' quality of life. 47 While patients' control over their own healthcare has increased over time, 43 researchers and clinicians should be mindful that not all patients are interested in taking a more active role in their care.
Our study has a number of limitations. Firstly, our observation of the consultations may have altered the way in which they were conducted. Both patients and Physicians may have acted differently than they usually would have, because they were aware of being observed and recorded. In some consultations, references were made by both patients and Physicians to the recording equipment being used, so this is a possibility, although one which many qualitative studies of this nature experience. We could have also used video-recording equipment to collect more data from the consultations themselves, however, that may have potentially influenced participants' behaviours even more.
It is also important to acknowledge the impact of our own expectations and experiences on the analysis. The researcher grew up in a lower/middle class household, but has since received training as a healthcare professional and researcher and objectively sits in a higher SES bracket. These experiences are likely to have led to unconscious biases within the analysis.

| CON CLUS ION
In conclusion, there are a number of socioeconomic differences in communication behaviours during head and neck oncology followup consultations. This is through the use of relational talk, patients' active or passive participation, and patients' preferences for involvement. Patients from low SES backgrounds seem to take a passive role within the consultation, which may cause the Physician to alter their communication style accordingly.

| Practice implications
Given that patients from lower socioeconomic backgrounds seem to be reluctant to raise concerns with Physicians, the PCI may be a useful way to facilitate raising of concerns for such patients. This may help to reduce SES differences in this aspect of doctor-patient communication, and potentially reduce health inequalities in this group.

CO N FLI C T S O F I NTE R E S T
None.

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared.