Socio‐economic differences in patient participation behaviours in doctor–patient interactions—A systematic mapping review of the literature

Abstract Background The degree to which patients participate in their care can have a positive impact on health outcomes. This review aimed to map the current literature on patient participation behaviours in interactions with physicians and the extent to which differences in these behaviours can be explained by socio‐economic status (SES). Search strategy Four electronic databases were searched from 1980 onwards using key words related to socio‐economic status and patient participation behaviours. Study selection Titles, abstracts and full texts were screened by two reviewers, with the second reviewer screening 20% of all entries. Data extraction Data on year of publication, country, patient population, setting, patient participation behaviour studied, and SES measure used were extracted. Main results Forty‐nine studies were included in the review. Most studies were conducted in the United States, and the most commonly studied patient participation behaviour was involvement in decision making. Most studies measured SES using education as an indicator, with very few studies using occupation as a measure. Many studies did not report on participants’ medical condition or study setting. Patient participation in their health‐care appointment increased with increasing SES in 24 studies, although in 27 studies no significant association was found. Discussion and conclusions Current literature was found to be mainly US‐centric. Many studies did not specify participants’ medical condition or in what setting the study was undertaken. More studies are needed on less commonly studied patient participation behaviours. It would be helpful for further studies to also include a wider range of SES indicators.


| INTRODUC TI ON
Patient-centred care has been associated with beneficial outcomes such as a greater adherence to treatment, satisfaction and improved quality of life. [1][2][3][4] The Institute of Medicine defines patient-centred care as providing care that is respectful of and responsive to individual patient preferences, needs and values, ensuring that patient values guide all clinical decisions. 5  Although the review found that patients with low socio-economic status (SES) tended to participate less actively in their care, the study and its search strategy were insufficiently sensitive to allow identification as to whether certain patient participation behaviours were more researched or more important than others, since the focus of the study was doctor-patient communication in general.
The Verlinde et al 8 review also limited identification of literature exploring the social gradient in doctor-patient communication and social gradient, to studies reporting the 'social class related concepts of' educational level, income or occupation. Confusingly, three of the studies included in this review measured SES using 'social class', although the authors did not specify exactly how this was defined. However, there are several other indicators of SES which may also be associated with patient participation behaviours such as the patients' health insurance status or receipt of benefits, and also area-level measures of deprivation related to the patients' home address (Indices of Multiple Deprivation), which may not have been captured previously, and may still be relevant. 9 Bearing in mind the potential importance of this area and its likely relationship to beneficial health outcomes, we undertook a systematic mapping review to identify what research had been done which specifically examined how patient participation behaviours in doctor-patient interactions are related to differences in a wide range of possible measures of socio-economic status.
We chose to conduct a systematic mapping review, as such reviews are useful for detecting patterns in a large body of literature in order to identify areas for future research. As such, details of the included studies are summarized without quality assessment or presenting statistical analyses. 10

| DATA SOURCE S
An electronic search was undertaken of the following databases: Medline, CINAHL, PsychINFO and Web of Science. Literature was searched from 1980 to 2018; since prior to 1980, there was much less electronic indexing. A pilot search was conducted to identify potentially eligible papers, assess the amount of relevant literature in the field and identify suitable search terms. At this stage, we found that including screening appointments and emergency admissions made the scope of the review far too broad and unmanageable; therefore, we decided to introduce limits in the electronic search terms regarding ongoing doctor-patient relationships. The electronic search contained free text and subject headings including patient-centred care, question asking, raising concerns, involvement in decision making, building rapport, expression of preferences, emotions or opinions, educational status, income, occupational status, employment, social class and socio-economic factors. This was modified as necessary for each database and can be found in Appendix S1.
Inclusion criteria for the review were as follows: • Studies involving patient perspectives on actual and desired question asking, raising concerns, involvement in decision making, rapport building, or expression of opinions, preferences and emotions.
• SES gradient measured in the form of education, income, occupation or 'other measures' which included patients' health insurance status, income indicators of state benefits and area-based measures relating to the patients' home address.
• Studies involving adult patients.
• Only studies which focused on doctor-patient interactions.
• Written in English language only. Studies were excluded if: • They included only health-care professional perspectives on patient participation.
• Patients under 18 or parents of patients only were recruited.
• Adult patient perspectives of childhood experiences were collected.
• The study was conducted in a country on the OECDs Development Assistance Committee list of Official Development Assistance recipients. 12 This was in order to limit literature to higher income countries where the health-care systems were likely to be similar.
• The appointment involved emergency attendances or screening.
• The interactions were with health-care professionals who were not medical doctors.
• They were opinion articles.
• They were systematic reviews.

| S TUDY S ELEC TI ON
One reviewer (SA) screened all titles and abstracts identified through electronic searches, and 20% of the entries were double screened by a second reviewer (DH). All full-text articles were then screened by one reviewer (SA), and 20% of the full texts were double screened by a second reviewer (DH). If the two reviewers disagreed on any papers, this was resolved by discussion with two other independent reviewers (RH and SR).

| DATA E X TR AC TI ON
Data extraction was independently conducted by both reviewers and the following information was obtained: year published, country the study was conducted in, study method and design,

| RE SULTS
The title and abstracts of 4718 articles were imported into Endnote, and 368 duplicates were removed. This left 4350 entries, of which 3989 articles were excluded leaving 361 entries. After screening all 361 full-text articles, the two reviewers disagreed on 11 papers.
Following discussion, seven papers were excluded. After screening, 49 studies were included in the review. The PRISMA diagram can be found in Figure 1.
Details of the characteristics of the 49 included studies can be found in Table 1 studies. Unfortunately, 16 studies did not specify which setting their research referred to when collecting data from participants. The most commonly studied patient participation behaviour was involvement in decision making (46 studies), whereas five studies examined raising concerns, [14][15][16][17][18] and only one study looked at rapport building. 19 Question asking and expression of opinions, preferences or emotions was more commonly studied, featuring in 13 and 12 studies, respectively. The rapport building study recruited participants from the general population and so there was a lack of studies which focused on rapport building which involved participants in a health setting. Three of the raising concerns studies were with breast cancer patients ( Table 2).
The most commonly used measure to explore relationships be- It is also important to note that many studies examined more than one type of patient participation behaviour or used more than one measure of SES. Most studies (33) used more than one measure of SES, whereas only 18 studies examined more than one type of patient participation behaviour (Table 2). Table 1 shows the direction of associations reported in the included studies. Of the 49 included studies, 5 did not perform statistical analyses as they had employed qualitative methodologies.
Positive associations between SES and patient participation be- One study did not perform statistical analyses.

| D ISCUSS I ON
Although the goal of systematic review searches is to identify all relevant studies on a topic, it is necessary to balance comprehensively covering a topic (or sensitivity of a search) with how manageable it is within resources available. 11 On the other hand, a wider search may reduce precision (identifying non-relevant articles), which while more comprehensive, may be more difficult to summarize because types of studies may vary quite widely. Systematic mapping reviews help by a method to overview a larger area so that gaps to inform future research can be identified. 10,11 Our study shows that while an earlier systematic review explor- whereas other reviews have had a main focus on behaviours in the consultation. 8,29 We found that the most commonly used measure of SES in studies of this type was educational level, while measures of participants' occupation have been much less frequently used. Income and employment status were not as commonly measured as educational level, although they were still used in some studies. Occupation is a key indicator of SES and likely to have an important influence on the doctor-patient relationship, 30 and so it is surprising to find so few previous studies using this measure.
We found that the most frequently studied patient participation behaviour was involvement in decision making, whereas raising concerns and building rapport were comparatively relatively neglected.   showed an association with education and patient participation behaviours, as many as 17 studies found no statistically significant association between the two variables; and so the relationship is likely to be complex. On the other hand, few studies seem to have found a significant association between patient participation behaviours and employment or income. Larger and more sophisticated studies are needed, using a range of SES indicators and a more in-depth description of patient participation behaviours, and the setting involved.
While the most common condition studied was cancer and the most common setting was secondary or tertiary care, 36.7% of studies did not specify what condition (if any) their participants were diagnosed with or what health-care setting their questions regarding patient participation referred to. This is potentially important information which is missing from these studies, as setting and condition which the patient is consulting for can influence a patient's preferred and experienced level of participation in a consultation. 7,32,33 Most studies included in the review were conducted in the United States, making the current research in this area very US-centric. This may limit the generalizability of the results of these studies, as other countries have differently structured health-care systems which might influence patient participation behaviours. There is a need for more studies on patient participation behaviours outside of the United States.

| CON CLUS ION
In conclusion, our findings suggest that most patient participation research relies on education as an indicator of SES and mainly explores involvement in decision making as the patient participation behaviour of interest. Most previous studies have been undertaken in the United States, but many lack important information on the setting or the patients' condition. More studies on specific patient participation behaviours such as rapport building and raising concerns are needed, and other studies undertaken outside the United States. Use of a wider range of SES measures such as occupation, housing situation, receipt of benefits and household finances would be useful additional data.

ACK N OWLED G EM ENTS
The authors would like to acknowledge the work of Danielle Hewitt as second reviewer on this mapping review.

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
Data sharing is not applicable to this article as no new data were created or analysed in this study.