Assessing the acceptability to general practitioners of the French College of General Medicine’s recommendations on considering patients’ social situations: a Delphi study Authors: Bastien Richard, Sohela Moussaoui, Kim Bonello, Mariela Skendi, Théo Duguet, Hugo Figoni, Gladys Ibanez Published: 2024 ABSTRACT Abstract AbstractBackgroundSocial determinants of health (SDH) impact the health status of individuals around the world. General practitioners (GPs) can take into account the social situation of patients in their care practice. To this end, the College of General Medicine (CGM) issued recommendations in 2022 to propose 100 methods of action.ObjectiveTo assess the acceptability to GPs of the recommendations set out by the CGM to improve the consideration of the social situation of patients in consultations.DesignQuantitative, cross-sectional, descriptive study, using the Delphi method.SettingThe data were collected through the administration of a questionnaire to GPs practising in France.ParticipantsParticipants were recruited by email from the academic network of Sorbonne University. The only criteria for inclusion were to be a GP from Sorbonne University and to complete the full questionnaire. 25 participants were included.InterventionsProposals were grouped into 24 themes. Participants had to rate the acceptability of these themes by rating their degree of relevance and degree of applicability on a Likert scale.Primary outcomeThe primary outcome was the acceptability by GPs of proposals to take into account the SDH. Acceptability was defined as relevance with a median greater than or equal to 7 and applicability with a median greater than or equal to 7, in the absence of disagreements.ResultsAfter 2 rounds, 12 themes were accepted: 5 addressed interventions at the individual level (eg, ‘understanding the context of patients’ lives and identifying social difficulties’), 4 addressed interventions at the organisational level (eg, ‘communication actions aimed at vulnerable patient populations’) and 3 addressed interventions at territorial level (municipality and national). Relevance was very good for all of them, with median responses ranging from 8 to 9 and with no disagreement. Applicability was more mixed, with 12 themes deemed applicable. Justifications were provided through participants’ comments. Proposals were made to improve the applicability.ConclusionThis study explored how SDH could be taken into account through the perspective of GPs in the context of the CGM’s recommendations. While all proposals were deemed relevant, some were not applicable. The findings emphasise the need for adaptations in the organisation of the practice, of care pathways and more generally, in the organisation of the health system. Those actions require the commitment of professionals and political actors.Keywords: GENERAL MEDICINE (see Internal Medicine), STATISTICS & RESEARCH METHODS, Protocols & guidelines, SOCIAL MEDICINE, Psychosocial Intervention STRENGTHS AND LIMITATIONS OF THIS STUDY The Delphi method has enabled a reasoned consensus to be generated, which can be used to support the proposed recommendations. The Delphi method has enabled a reasoned consensus to be generated, which can be used to support the proposed recommendations. The anonymisation of responses and the absence of face-to-face meetings avoided the ‘opinion leader effect’ and limited conflicts of interest. The anonymisation of responses and the absence of face-to-face meetings avoided the ‘opinion leader effect’ and limited conflicts of interest. The use of the Likert scale provided precise quantified answers and made it easy to analyse the results. The use of the Likert scale provided precise quantified answers and made it easy to analyse the results. The results of this Delphi procedure are specific to the panel of experts who were general practitioners practising in Île-de-France and mostly women, working in a multiprofessional group. The results of this Delphi procedure are specific to the panel of experts who were general practitioners practising in Île-de-France and mostly women, working in a multiprofessional group. The anonymisation of the responses did not allow participants to exchange views with each other. The anonymisation of the responses did not allow participants to exchange views with each other. Introduction The social determinants of health The social determinants of health (SDH) are the conditions in which individuals are born, grow, live, work and age as well as the set of forces and systems that shape their daily living conditions.1 These factors include economic systems, and social and development policies. They include income, education, employment, housing, ethnicity, working conditions and social ties, and they are shaped by the distribution of money, power and resources. These factors create a ‘social gradient’ that impacts the health status of populations across countries around the world. They account for nearly 50% of the modifiable factors that determine the health of populations.2 In the UK, for example, according to the Office for National Statistics in 2023, age-standardised mortality rates from 2021 to 2023 are much higher among the unemployed than among people with a high status or level of employment, and this is true for almost all causes combined (stroke, chronic ischaemic heart disease, diabetes, dementia, cancers, etc).3 In the USA, the most socially vulnerable populations are more likely to die early from cardiovascular disease and several types of cancer.4 5 These social inequalities in health have also been widely documented in the health context of the COVID-19 pandemic.6 7 In France, the level of social inequalities in health is one of the highest in Western Europe.8 A strong social gradient exists in terms of all-cause mortality in France for both men and women.9 According to the National Institute for Statistics and Economic Studies (Insee), among the wealthiest 5%, life expectancy at birth for men is 13 years higher than among the poorest 5%.10 For a woman, this gap would be 8 years. There is also a significant difference according to socioprofessional category or level of education. These social inequalities in health have also been widely described in chronic and emerging diseases, neurological and degenerative diseases, psychiatric disorders, preventive practices and also in children.11 12 Addressing the social inequalities In the spirit of social justice, the WHO established the Commission on Social Determinants of Health in 2005 to gather evidence on how to promote health equity. In WHO reports, the genesis of social inequalities in health is attributed to a set of determinants, from the most proximal to the most distal.13 14 To improve the health of populations, individual-directed modes of action (proximal determinants) must go hand in hand with more global strategies, involving a set of structural determinants.15 Well-organised healthcare systems, with a strong focus on primary care, have an impact on social and territorial inequalities in health.16 In the wake of the COVID-19 pandemic that has pushed an estimated 120 million new people into extreme poverty, WHO has again urged countries to invest in strong primary care.17 In several countries, recommendations for physicians and other professionals have been published, such as in the USA by the American College of Physicians, in Canada by the College of Canadian Family Physicians, in the UK by the Royal College of Physicians and in Japan by the Japan Federation of Primary Care Associations, as tools to help take into account the SDH in medical consultations.18,21 General practitioners (GPs) are among the first health professionals to be able to take SDH into account.22 In France, they carry out about 300 million consultations per year and on average see each patient 3–4 times a year.23 Because of their situation as primary care physicians, they have frequent and repeated contact throughout patients’ lives, which places them in a potential situation of identification, orientation and intervention. However, despite knowledge and appreciation of the social context, it is not easy in practice to address the SDH and relevant social needs. In France, reducing social inequalities in health is an objective of the current national health strategy and its previous health policies.24 In 2014 and again in 2022, the National College of General Medicine (CGM) published recommendations aimed at collecting and taking into account the social situation of patients in general practice consultations.25 26 This college proposed an assessment of the patients’ social situation through the study and understanding of four major areas of life: their socioemotional life, their professional life, their home and physical environment, their access and care pathway. Finally, the 2022 recommendations proposed a list of 100 methods of action grouped into 24 themes, as a new tool for GPs to take into account potential social difficulties identified during consultations. The recommendations of the CGM have so far not been evaluated in terms of their acceptability to French GPs. Our objective here was to carry out such an evaluation, considering issues of relevance and applicability for practice. The aim of this work was to improve the design and implementation of the proposed interventions and thus improve primary care physicians’ ability to address the SDH and related needs. Methods Study design Different approaches can be used to develop or assess the quality of tools such as medical recommendations and good practice guides. These include consensus methods, some of which can be used for quantitative assessment of qualitative data.27,29 The Delphi method consists of interviewing a group of individuals acting as experts on a given subject, whose opinions or judgements are of interest when information is missing on the subject in question, via systematised and iterative questionnaires over 2–4 rounds.28 30 This study was carried out after the recommendations had been developed by the members of the CGM, following exchanges of divergent positions in order to reach a consensus on content and editorial form. The study was not requested by the CGM. None of the participants in this study were involved in drawing up the recommendations. The recommendations of the CGM were developed using a modified Delphi approach.26 28 This study was carried out using a similar approach, using a Delphi method. The Sorbonne University Faculty of Medicine includes a network of GPs who are particularly aware of the themes of the SDH (n=381 GPs). They have the capacity to implement the recommendations expected from a theoretical point of view. These experts, university GPs, were asked to assess the acceptability of the proposals recommended by the CGM. Recruitment of participants Participants were recruited by email from the academic network of GPs at Sorbonne University. Those who agreed to participate received a summary of the 100 interventions proposed by the CGM and grouped into 24 themes, as well as a leaflet explaining the modalities of the study, the web link to the details of the protocol and the initial questionnaire. The recruitment was carried out in the Paris region (Île-de-France region) from June 2023 to December 2023. A minimum of 15 participants remaining at the end of the process was expected as usual.28 For each of the rounds of the Delphi method: 10 days were expected for the response time. A follow-up email was then sent to non-respondents. After 2 months, the questionnaire for each round was closed and the response data were used. The research team was not part of the group of participants. Primary outcome The main outcome measure was the acceptability to physicians of proposals to take into account the social situation of patients in general practice. The assessment of this acceptability was based on two criteria: the assessment of the relevance of these proposals and the assessment of their applicability in practice. Acceptability was defined as relevance with a median greater than or equal to 7 and applicability with a median greater than or equal to 7, in the absence of disagreements. Disagreement was defined as a percentage of responses between 1 and 3 (ie, in the first third of the Likert scale) that was greater than or equal to 30%, with a percentage of responses between 7 and 9 that was greater than or equal to 30%.27 Data collection The entire procedure was conducted anonymously without any contact between the participants. Responses were collected through the administration of questionnaires in each round, carried out by the research team. A minimum of two rounds and a maximum of four rounds have been set. The questionnaires were named according to the rounds (questionnaire 1 for round 1, questionnaire 2 for round 2, etc). In these questionnaires, each participant could express via a Likert scale of 1–9 the level of relevance and applicability of the proposals. The 100 proposals to be assessed were grouped into 24 themes, following the recommendations of the CGM, and then were classified by level of intervention. The purpose of this classification into themes and level of intervention was to structure the assessment process and avoid having a questionnaire that was too long. Online supplemental appendix 1 presents the 24 themes by levels of intervention (individual, organisational and territorial) and the corresponding proposals. For each of the 24 themes, the 100 proposals were recalled for assessment. All themes of the CGM recommendations were submitted for assessment. In questionnaire 1, participants were additionally asked for the following information: gender, age, department of practice (Île de France regional subdivision), environment of practice (urban, rural or semirural location), practice mode (alone or in group) and site of practice that refers to the facility where the doctor practices and receives patients. Finally, in each round, a box allowed each participant to comment on his or her assessment in an open way (called ‘verbatims’). The order of the different stages of data collection was as follows: (Step 1) Sending questionnaire 1 via an internet link by email, containing the study protocol and the 24 themes to be assessed in relevance and applicability, to all GPs of Sorbonne University; (Step 2) Synthesis of results; (Step 3) Sending questionnaire 2 to each participant included in round 1, containing both the remaining themes to be reassessed (who did not obtain a sufficient level of agreement in round 1), the reminder of their own responses to questionnaire 1 and the overall statistics of the responses for each theme that did not have a sufficient level of agreement, with a summary of the associated comments if necessary; (Step 4): Synthesis of results; (Step 5) Possible repetition of steps 3 and 4 in a third and fourth round if there are persistence of unacceptable themes with disagreements after round 2. Regulatory approaches The study has been the subject of a General Data Protection Regulation certificate, according to the reference methodology MR-004, registered under reference No Richard-2024-0204, to guarantee the conditions of confidentiality and security of personal data. All collected data were stored on an encrypted hard drive owned by the research team. Data from questionnaire 1 were also stored on Limesurvey. Personal data will be stored on the hard drive until 1 March 2024 and will then be destroyed. The research team made it clear in each email inviting participants to participate in the study that the data collected was anonymous. Participants could not identify other participants’ responses, nor could they identify the participants themselves. The research team had no direct access to the identity of the respondents. Statistical analysis We first described the characteristics of the responding physicians. Following this, we assessed the 24 proposed themes initially in terms of relevance and applicability, and then in terms of acceptability, which combined both relevance and applicability. A total of 48 responses were obtained from the assessment of the relevance and acceptability of the 24 themes. The participants’ different responses to each round via the Likert scale resulted in a median of 1–9 levels of agreement for each of these 48 elements. If at the end of a round, a given theme obtained a median score of 7 or higher with no disagreement among the experts, then the item was ‘validated’ and no longer submitted for assessment in the next round. In order for a theme to be defined as acceptable, it had to have been validated both in terms of relevance and in terms of applicability. Finally, we analysed the verbatim comments associated with each of the themes. No software was used for this study. Results Description of respondents The consensus exercise was conducted from June 2023 to December 2023. Figure 1 shows the physician selection flow chart for the study. Of the physicians surveyed, 32 agreed to complete the study, 25 responded completely to questionnaire 1 and 20 responded completely to both questionnaires. The proceedings were stopped after two rounds. Study flow chart (n=number of experts). CGM, College of General Medicine. Table 1 presents the characteristics of the responding physicians. The majority of them are women, aged between 30 and 49, working in groups and in urban settings. Three departments of Île-de-France are mainly represented which are Paris (28%), Seine-Saint-Denis (28%) and Val-de-Marne (24%). Three departments of Île-de-France are not represented by any participant, which are Yvelines, Hauts-de-Seine and Val-d'Oise. Table 1Characteristics of responding experts (n number of experts)n%Gender Woman1768 Man832Age 20–2914 30–591040 40–491144 50–5914 60–6928Department of practice Paris728 Seine-et-Marne312 Yvelines00 Essonne28 Hauts-de-Seine00 Seine-Saint-Denis728 Val-de-Marne624 Val-d'Oise00Environment of practice Urban2392 Semi-rural14 Rural14Practice mode Alone416 In a group2184Site of practice City office728 Multiprofessional health centre/health centre1768 Hospital14 Other00 Characteristics of responding experts (n number of experts) Relevance and applicability of the 24 themes assessed Figure 2 shows the cumulative response rates to the different themes in terms of relevance and applicability to the two rounds, at the individual organisational and territorial intervention levels. There may be variability in responses depending on the level of intervention considered. Cumulative response rates to the different themes in terms of relevance and applicability to the two rounds and to the level of individual, organisational and territorial intervention (n=25 respondents in round 1 (R1)and n=20 respondents in round 2 (R2)). Regarding relevance, it was judged to be very good in the first round for all the themes assessed and did not require a second round for reassessment. The three themes that were judged to be the most relevant were theme 1 on understanding the context of patients’ lives and identifying social difficulties (96% of responses between 7 and 9), theme 5 on strengthening prevention (96% of responses between 7 and 9), and theme 13 on communication actions aimed at vulnerable patient populations (96% of responses between 7 and 9 9). It should be noted that these themes correspond to an individual or organisational level of intervention. The three themes deemed the least relevant were theme 15 on understanding the state of health of people in their health area as well as the resources available in the care offer (68% of responses between 7 and 9), theme 17 on improving care pathways in one’s territory (72% of responses between 7 and 9) and theme 18 on participation in local actions to promote healthcare in their healthcare area and health and public health (76% responses between 7 and 9). It should be noted that these themes correspond to a level of territorial intervention. There was no disagreement in terms of relevance to all the themes assessed. Indeed, there is a response rate of between 1 and 3 to only 8% at most, for theme 14 on the evaluation of professional practices. Regarding applicability, the results are more mixed. 9 out of 24 themes were initially deemed applicable to round 1, and 15 required a second round for reassessment. The three themes that were judged to be the most applicable were theme 4 on adapting care to patients’ needs in a biopsychosocial approach (64% of responses between 7 and 9 in round 1), theme 9 on understanding the state of health of its patients (64% of responses between 7 and 9 in round 1), and theme 12 on communication actions in common areas oriented towards prevention and the care pathway (68% of responses between 7 and 9 in round 1). It should be noted that these themes correspond to an individual or organisational level of intervention. The themes that were judged to be the least applicable were theme 7 on helping to coordinate care with local stakeholders (24% of responses between 7 and 9 in round 1, and 15% in round 2), theme 14 on the evaluation of professional practices (24% of responses between 7 and 9 in round 1 and 25% around 2) and theme 18 on participation in local health promotion and public health actions (24% of responses between 7 and 9 in round 1, and 20% in round 2). It should be noted that these themes correspond respectively to individual, organisational and territorial levels of intervention. No disagreement was found in terms of applicability for all the themes assessed. Indeed, the response rate between 1 and 3 in round 1 is at most 28% for theme 14 on the evaluation of professional practices. In round 2, the response rate between 1 and 3 is a maximum of 25% for theme 6 on social rights advice, for theme 7 on assistance in coordinating care with local stakeholders, and for theme 14 on the evaluation of professional practices. More detailed results are summarised in online supplemental appendix 2. Acceptability of the 24 themes assessed Table 2 shows the acceptability of the themes of the recommendations proposed by the CGM, following the two rounds of Delphi proceedings. It represents the medians and other quartiles (Q0, Q1, Q3 and Q4) of the responses for each theme, in terms of relevance and applicability, as well as a reminder of the presence of possible disagreements. Table 2Acceptability of assessed themes (Q0; Q1; Q3; Q4=Quartile 0; quartile 1; quartile 3; quartile 4)Assessed themesRelevance median (Q0; Q1; Q3; Q4)Applicability Round 1 median (Q0; Q1; Q3; Q4)Applicability Round 2 median (Q0; Q1; Q3; Q4)Presence of one or more disagreementsAcceptability (relevant and applicable)Interventions at the individual levelTheme 1: Understanding the context of patients' lives and identifying social difficulties9(5; 9; 9; 9)7(1; 6; 8; 9)NoYesTheme 2: Adapting communication to be understandable by all9(4; 8; 9; 9)6(1; 5; 7; 9)6(1; 5; 7; 8)NoNoTheme 3: Proper use of the electronic medical record9(5; 7; 9; 9)6(1; 5; 8; 9)7(1; 5,75; 7; 9)NoYesTheme 4: Adaptation of care to patients' needs in a biopsychosocial approach9(4; 9; 9; 9)7(4; 6; 9; 9)NoYesTheme 5: Strengthening prevention9(6; 8; 9; 9)7(2; 6; 7; 9)NoYesTheme 6: Advice on social rights8(5; 7; 9; 9)6(1; 5; 7; 9)6(1; 3,75; 6,25; 9)NoNoTheme 7: Assistance in the coordination of care with local stakeholders9(5; 7; 9; 9)5(1; 4; 6; 9)5(1; 3,75; 6; 9)NoNoTheme 8: Physician education9(3; 7; 9; 9)7(3; 6; 8; 9)NoYesInterventions at the organisational level (structure of practice, office/health centre)Theme 9: Understanding the health status of one’s patients9(5; 8; 9; 9)7(2; 6; 8; 9)NoYesTheme 10: Improving access to care in one’s office9(4; 8; 9; 9)7(1; 4; 7; 9)NoYesTheme 11: Improving care pathways in your practice9(1; 7; 9; 9)5(1; 4; 8; 9)6(1; 4,75; 8,25; 9)NoNoTheme 12: Communication actions in common areas oriented towards prevention and the care pathway9(3; 8; 9; 9)8(1; 6; 9; 9)NoYesTheme 13: Communication actions aimed at the fragile populations of its patients9(6; 8; 9; 9)6(1; 5; 7; 9)7(1; 5,75; 7,25; 9)NoYesTheme 14: Evaluation of Professional Practices8(2; 7; 9; 9)4(1; 3; 6; 8)5(1; 3,75; 6,25; 9)NoNoInterventions at territorial level (municipality, national)Theme 15: Understanding the health status of people in their health territory, as well as the resources available in the care offer9(3; 6; 9; 9)5(2; 4; 7; 9)6(2; 4,75; 6,5; 9)NoNoTheme 16: Improving access to care in the territory9(3; 7; 9; 9)6(1; 5; 7; 9)6(1; 4,75; 7; 9)NoNoTheme 17: Improving care pathways in its territory9(3; 6; 9; 9)5(1; 5; 7; 9)6(1; 5; 7; 9)NoNoTheme 18: Participation in local health promotion and public health actions9(3; 7; 9; 9)5(1; 4; 6; 9)5,5(1; 4; 6; 9)NoNoTheme 19: Information for the population on the actions and resources of the territory9(5; 7; 9; 9)7(1; 5; 8; 9)NoYesTheme 20: Development of research on quality and equity of care in the territories9(5; 8; 9; 9)5(2; 3; 7; 9)5,5(2; 4,75; 7; 9)NoNoTheme 21: Promoting primary care research for quality and equity of care9(3; 8; 9; 9)6(1; 5; 7; 9)6(3; 5; 7; 9)NoNoTheme 22: Better structuring of the care pathway in France9(6; 8; 9; 9)6(1; 5; 8; 9)7(1; 5; 9; 9)NoYesTheme 23: Improvement of business software interfaces and shared medical records8(3; 7; 9; 9)7(1; 4; 8; 9)NoYesTheme 24: Policy advocacy-health alert9(3; 8; 9; 9)6(1; 5; 7; 9)5,5(2; 5; 7,25; 9)NoNoColor green: "yes", color red: "no" Acceptability of assessed themes (Q0; Q1; Q3; Q4=Quartile 0; quartile 1; quartile 3; quartile 4) Color green: "yes", color red: "no" In terms of relevance, the medians of responses for each theme were all very high from the first Delphi round. Their median values were all greater than 7, and more accurately, were all between 8 and 9. In detail, 21 out of 24 themes had a median of 9. The three themes with the lowest median of responses (median at 8) were theme 6 on social law advice, theme 14 on the evaluation of professional practices and theme 23 on the improvement of business software interfaces and shared medical records. It should be noted that these themes correspond respectively to individual, organisational and territorial levels of intervention. In terms of applicability, the medians of responses were lower overall, again with variability in responses depending on the level of intervention considered. The values of these medians were all between 4 and 8 at the end of the two rounds. They were greater than or equal to 7 for 12 out of 24 themes at the end of the two rounds (9 themes in round 1 and 3 themes in round 2). The theme with the highest median (median at 8 in round 1) was theme 12 on communication actions in common areas oriented towards prevention and the care pathway. It should be noted that this theme corresponds to a level of organisational intervention. The medians of the remaining 12 themes were strictly less than 7 at the end of the two rounds. The themes with the lowest medians were theme 7 on helping to coordinate care with local stakeholders (median at 5 in rounds 1 and 2), theme 14 on the evaluation of professional practices (median at 4 in round 1 and median at 5 in round 2), theme 18 on participation in local health promotion and public health actions (median at 5 in round 1 and median at 5.5 in round 2), and theme 20 on the development of research in quality and equity of care in the territories (median at 5 in round 1 and median at 5.5 in round 2). For the record, there was no disagreement in terms of relevance and applicability for all the themes assessed during the two rounds. Finally, in light of these results, 12 themes were considered acceptable and 12 were considered non-acceptable. Nine were deemed acceptable in round 1 and another three were deemed acceptable in round 2. Five of the eight themes corresponding to an individual level of intervention were accepted. Four of the six themes corresponding to an organisational level of intervention were accepted. Only 3 of the 10 themes corresponding to a territorial level of intervention were accepted. It should be noted that the themes deemed unacceptable were nevertheless considered relevant but were not considered sufficiently applicable. Verbatim comments on the 24 assessed themes Online supplemental appendix 3 presents the summary of the verbatim comments associated with each theme assessed during the Delphi procedure. It can be seen that many positive comments emerged in the themes assessed, consistent with the strong overall relevance observed previously. The themes concerned belonged to the three different levels of intervention (themes: 1, 2, 3, 6, 7, 10, 13, 16, 18, 20, 24). These were comments that considered the proposals to be interesting, important and even essential. The following themes were considered essential: understanding the context of life and identifying social difficulties; proper use of the electronic medical record; adapting communication to be understandable by all; advice on social rights; improving access to care in their practice; communication action towards the fragile populations of its patients; improving access to healthcare in the territory; development of research on quality and equity of care in the territory; participation in local health promotion and public health actions; political advocacy, health alert. Negative comments were rarely concerned with the relevance of the recommendations made, except when they seemed to be at the limit of the role of the GP itself: advice on social rights and training of doctors (themes 6, 8). The other negative comments were primarily related to the difficulty of accessing available resources (by their lack, saturation, lack of knowledge or understanding on the part of the physician and their lack of knowledge or lack of understanding on the part of the patient). These could include human resources such as medical, paramedical, social and administrative staff (themes: 2, 3, 5, 6, 7, 8, 10, 11, 13, 14, 16, 17, 19, 20, 21, 23); technical resources such as interpretation services, computers, shared medical records, digital health records (themes 2, 3, 6, 23); or territorial, intervening in particular in the coordination of care, such as the Territorial Health Professional Communities (THPC), municipalities and local authorities (themes 7, 17, 18, 20, 24). Another major limitation mentioned was the physician’s lack of time to implement the proposed recommendations (themes 2, 6, 7, 8, 9, 11, 13, 14, 15, 16, 17, 18, 20). Other important limitations were related to the costs involved (in particular concerning the remuneration of time dedicated outside of pure medical activity, for example, the doctor’s training, meetings with other actors, etc). (themes 2, 4, 7, 11, 16, 18, 20), the language barrier (themes 1, 2, 5, 7, 13, 19) and the mode of practice of the physician alone (themes 2, 7, 10, 11, 15, 18). There were also other notable limitations such as the lack of public policy intervention (18, 20, 22, 24), the diversity, complexity and lack of standardisation of medical software (themes 3, 14, 23), and the ethical dimension posed by data sharing (themes 3, 23). A number of solutions have been proposed to overcome the limitations mentioned. Among them, the concept of multidisciplinarity has often been proposed, with the facilitated use of professionals trained in social issues, the transfer of competences and even the delegation of tasks in terms of social care (themes 2, 6, 7, 11, 13, 14, 21). The role of the THPC has also been strengthened on several occasions, particularly with regard to themes related to the improvement of coordination and care pathways (themes 2, 10, 14, 15, 17, 18). The other solutions proposed were the introduction of more appropriate IT tools (themes 3, 5, 14, 23), the strengthening of the training of doctors, particularly on social issues (themes 6, 7, 23), and the promotion or financing of tools such as professional interpreting services (themes 2, 7). Discussion Synthesis During this Delphi procedure, 25 participants assessed the relevance and applicability of 24 themes grouping together 100 proposals of the CGM, to improve the consideration of the social situation of patients in general practice. Of the 24 assessed themes, 12 were deemed acceptable by the GPs surveyed, that is, both relevant and applicable. The remaining 12 themes all achieved a sufficient level of agreement in terms of relevance but not in terms of applicability. Themes that were deemed not applicable frequently related to a level of territorial intervention. Several comments were made to illustrate these limitations. These include issues related to the lack of saturation of social resources, their lack of knowledge by the doctor and the patient, the doctor’s lack of time, the problems related to the financing of certain tools and measures, the difficulties related to the doctor’s mode of practice and the limits of the role of the GP. These results highlight the overall relevance assessed on all the elements proposed by these recommendations, and the importance for GPs to take into account the social situation of patients in general practice. However, the applicability of these recommendations requires adaptations in the organisation of work, the organisation of care pathways and, more generally, in the organisation of the health system. Some areas for improvement can be proposed. Comparison with international data Recommendations to improve the consideration of patients’ social situation have been developed by several scholarly organisations in the USA, Canada, England and Japan.18,21 The very existence of these recommendations reflects the universalism of the problematic but also the importance of designing recommendations specifically adapted to each country and healthcare system. The recommendations from both England and the USA are directed not only at physicians and healthcare professionals but also at various professionals involved in the social and healthcare sectors, health departments and public sector organisations. They advocate for a change in the healthcare and social systems and commit to changes in the physician’s training in order to better prepare them to take into account the social situation of patients in care. Other recommendations, such as those of the Canadian College of Family Physicians (CCFP), are of particular interest. In 2015, the CCFP published a series of recommendations in the form of a practical advice guide. In these, three levels of intervention have been proposed. First, an individual (micro) level corresponding to the immediate clinical setting of the physician. Second, a community and local level (meso) that includes the patient’s community, the medical community and the ‘civic community’, where health professionals are both practitioners and citizens. This community level also includes education, training and continuing professional development. Third, a global (macro) level corresponding to themes of humanitarian and political action on a larger scale, with physicians having to be concerned about the well-being of the entire population. While they may not appear to be perfectly transposable to the levels of intervention of the CGM’s recommendations, the levels of intervention in the CCFP’s practice guide can be broadly similar. These levels of intervention include a total of 14 themes, each with different indications and advice, as well as links to other practical guides specifically related to the concerned proposal. Among other examples, at the individual level, one of the themes of the recommendations for the physicians is to regularly screen for poverty and intervene if necessary. The proposal explains the ins and outs of this recommendation and also provides a link to a specific poverty screening and intervention tool developed by the Ontario Family Physician Poverty Committee. At the community level, for example, one of the recommendations is to encourage home consultations, which sometimes make it possible to better understand the living environment of patients, and therefore, to better understand their social situation. Finally, at the global level, a recommendation theme is to encourage physicians to join or create a health promotion organisation. Links to tools are provided, such as a physician-deputy contact programme, which connects Canadian Medical Association members with their corresponding members of parliament. To the best of the knowledge of the research team, these recommendations have not been the subject of an acceptability study. Strengths of the study The Delphi method has made it possible to generate reasoned consensus that can be used to legitimise at least in part the proposed recommendations. There were no geographical limits (apart from that imposed by the natural distribution of GPs at Sorbonne University, limited to Île-de-France). Thanks to the anonymity of the respondents, the use of the Delphi method also made it possible to protect against the dominance effect of one or more participants, known as the ‘opinion leader effect’ due to the absence of face-to-face meetings between the participants and to limit conflicts of interest. The use of the Likert scale allowed a quantitative estimation of qualitative data (in this case, being more or less in agreement with a given theme), and therefore, made it possible to obtain precise, quantified answers and to have a very good simplicity of analysis of the results. The choice of an odd-choice scale made it possible to give way to the undecided who would not have been able to make up their minds about their degree of agreement. Indeed, proposing a midpoint in the Likert scale (in this case, the value 5) made it possible to avoid cultivating frustration and limiting questionnaire abandonment by not ‘forcing’ participants to position themselves and to choose and by leaving them the possibility of an ‘average’ level of agreement. The choice of doctors from Sorbonne University allowed the research team to have privileged access to a very large number of expert doctors. Since the questionnaires were relatively dense and required real involvement on the part of respondents, it was all the more important that the distribution be made to a very large number of people, with the aim of obtaining at least the minimum threshold of experts to be included. None of the proposed themes were deemed ‘irrelevant’. This is not very surprising given that these recommendations were developed by a group of expert GPs, members of the CGM. This clearly shows that GPs consider the subjects dealt with by these recommendations to be of great importance, and makes the research, development and production of tools to help take into account the social situation of patients in consultations fully legitimate. Limitations of the study Different limitations can be identified in this study. The results of this Delphi procedure are specific to the panel of experts who participated and may not be found in other contexts. The participants all worked in Île-de-France, the majority of them were women working in a multiprofessional group. Some points of the recommendations may seem more relevant or more applicable to doctors practising in these types of structures, as has been mentioned several times in the transcripts. These are indeed structures where doctors can potentially have easier access to specific resources, paramedical, social, IT, coordination networks, etc. There is also a lack of responses from certain departments of Île-de-France (Yvelines, Hauts-de-Seine, Val d'Oise). It would, therefore, be interesting to develop similar studies in other regions, rural areas, less populated areas and also areas where the density of physicians and healthcare facilities is lower. Interviewing a representative sample of the French GPs was not feasible at the time of the study for budgetary reasons and timing with the release of the recommendations. The use of the university network was a facilitating factor both in obtaining responses within a reasonable timeframe and in disseminating the research, as this network encompasses the whole of the Île-de-France region (university GP practices are spread throughout the whole of the Île-de-France region). However, as it is a qualitative study we were not looking for representativeness, but rather for diversity of opinion. To this end, the study sample was very diverse in terms of age, gender and place and type of practice. In addition, Ile de France is a region of great size, that concentrates a substantial proportion of the GPs and the French population and with high proportions of social inequalities in health in France. Thus, the doctors who practice in this region are the ones who will be most likely to consult and apply the recommendations. The anonymity of the responses meant that there was no debate among the participants. It would have been possible to integrate a face-to-face meeting to allow for a debate, but this would have moved away from the initial Delphi technique to be more assimilated to a RAND-UCLA method.29 It could also have lost the benefits of anonymity for participants. However, after this work, the results were presented at an academic meeting to which the participating physicians were invited. Choosing an odd-choice Likert scale with a neutral option (a value of 5) carries the risk that it will be misinterpreted. Depending on the respondent, the neutral option could mean, “’ don’t know’ or ‘I don’t want to answer’ or ‘I don’t agree or disagree’. Some people may also choose this value to avoid involvement in the decision, or to respond to chance, as a ‘safe haven’. To compensate for this, the questionnaires were materialised in such a way that it was concretely written what the extreme values of this scale corresponded to at the time of each vote (the value 5, therefore, corresponds to an average degree of agreement). The number of rounds, not predefined initially, was limited to 2 in the end because no disagreement (which could have justified additional rounds in order to seek consensus) appeared at the end of the second round. However, additional rounds might have been able to refine the results. In the first round of Delphi, several people did not complete the questionnaire (n=7/32), probably because it was long and dense to complete. To simplify the questionnaire, the 100 proposals were presented by theme, with their full description in the question statement. In the second round of Delphi, 20% of the participants (n=5/25) did not answer the questionnaire, probably again because of the length of the questionnaire, but in this round none of the respondents provided incomplete answers. The assessment of the themes, and not of each proposal in them, implied that participants who voted for a given theme might not have the same level of agreement for all proposals in the same theme. Perspectives To be implemented, these proposals need to bring together a range of stakeholders other than GPs, with a shared willingness to work together. The themes considered unacceptable were often those acting at the territorial level. Participants pointed out that these themes were often unacceptable due to a lack of resources or time. Innovative actions need to be designed to put these proposals into action. In this regard, social prescriptions could be of great value. Social prescription consists of a set of interventions that can be initiated by GPs, in answer to a patient’s social need.22 Those prescriptions are based on public policies, community-based social services or referral programmes and allow GPs to connect patients with supports or social actors. Social prescription seems to be useful for completing fields of action that are difficult for GPs to access.31 The implementation of these recommendations could in time be evaluated by means of interventional studies. Conclusion The objective of this work was to assess the acceptability of the recommendations of the CGM to improve the consideration of the social situation of patients in general practice. A Delphi study was conducted among GPs at the Sorbonne University Faculty of Health. It highlighted that while all the proposals were considered relevant, not all of them were found to be applicable. Their main limitations have been illustrated, and several suggestions for adaptation have been suggested both to improve the practice of physicians and the organisation of the health system in the territory. These suggestions will be discussed to be integrated into an improved version of the recommendations of the CGM. This study should be replicated to identify actions acceptable to GPs in other regions and different healthcare contexts. These results, in addition to those of this study, will enable future recommendations to be adapted. supplementary material Footnotes Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2024-084837). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: Approval from an ethics committee was not required for this study, as it was a study on professional practice and on the acceptability of professional recommendations. The GPs were under no obligation to participate and received no compensation (financial or otherwise). The study did not raise ethical issues. Data availability free text: Data are available on reasonable request and with the agreement of the College of General Medicine. Patient and public involvement: In our study, patients and the public were not involved in any way, nor in the design or the implementation or analysis or interpretation of the results.The recommendations were drawn up on the basis of several studies, some of which involved the patient's point of view. Author note: BR and SM are joint first authors. HF and GI are joint last authors. Contributor Information Bastien Richard, Email: rbast94@gmail.com. Sohela Moussaoui, Email: sohela.moussaoui@sorbonne-universite.fr. Kim Bonello, Email: kim.bonello@sorbonne-universite.fr. Mariela Skendi, Email: mariela.skendi@sorbonne-universite.fr. Théo Duguet, Email: theo.duguet@sorbonne-universite.fr. Hugo Figoni, Email: hugo.figoni@sorbonne-universite.fr. Gladys Ibanez, Email: gladys.ibanez@sorbonne-university.fr. Data availability statement Data are available on reasonable request. References Supplementary Materials Data Availability Statement Data are available on reasonable request.