Factors associated with GPs’ knowledge of their patients’ socio-economic circumstances: a multilevel analysis

Abstract

Background.

To determine appropriate management for individual patients, GPs are supposed to use their knowledge of the patient’s socio-economic circumstances.

Objective.

To analyse factors associated with GPs’ knowledge of these circumstances.

Methods.

Observational survey of GPs who were internship supervisors in the Paris metropolitan area. Each of 52 volunteer GPs completed a self-administered questionnaire about their own characteristics and randomly selected 70 patients from their patient list. Their knowledge was analysed as the agreement between the patients’ and GPs’ responses to questions about the patients’ socio-economic characteristics in questionnaires completed by both groups. The association between agreement and the GPs’ characteristics was analysed with a multilevel model adjusted for age, sex and the duration of the GP–patient relationship.

Results.

Agreement varied according to the socio-economic characteristics considered (from 51% to 90%) and between GPs. Globally, the GPs overestimated their patients’ socio-economic level. GP characteristics associated with better agreement were sex (female), long consultations, the use of paper records or an automatic reminder system and participation in continuing medical education and in meetings to discuss difficult cases.

Conclusion.

Knowledge of some patient characteristics, such as their complementary health insurance coverage or perceived financial situation, should be improved because their overestimation may lead to care that is too expensive and thus result in the patients’ abandonment of the treatment. Besides determining ways to help GPs to organize their work more effectively, it is important to study methods to help doctors identify their patients’ social-economic circumstances more accurately in daily practice.

Introduction

The ongoing medical care that GPs provide gives them access, more or less directly, to different material and human aspects of their patients’ circumstances. This knowledge of the patient’s environment is supposed to enable them to plan management in accordance with the biopsychosocial model, in taking into account factors that are not strictly medical, such as their type of health insurance, their financial situation and their level of support by family and friends ( 1 ). Knowledge of these social and economic aspects is especially important because they are also major determinants of individual health status ( 2 ).

Different publications have described GPs’ knowledge of their patients’ socio-economic circumstances ( 3 ). Although some aspects, such as household composition, appear well known, GPs may know less about other elements, nonetheless essential, such as occupation.

GPs’ knowledge of their patients’ socio-economic circumstances varies according to patient characteristics, especially, the duration of the doctor–patient relationship ( 4 ), as well as according to GP characteristics ( 1 ). Doctors most distant from the biopsychosocial model may pay less attention to socio-economic details mentioned spontaneously during consultation or may avoid asking about them ( 5 ). Women doctors appear especially likely to discuss socio-economic factors with their patients ( 6 ). Another aspect, sparsely covered in the literature, is the concrete organization of GPs’ practices. This organization is important because the context of the physician–patient encounter determines in part the quality and type of conversations that take place ( 7 ). In principle, long consultations provide more time for GPs to get to know their patients better. Similarly, the type of medical file used may influence GPs’ knowledge. This primary tool of doctors’ activity structures their work: well used, it allows GPs to be better able to perceive the socio-economic aspects mentioned during the consultation; and well organized, it allows GPs to record these points when they are mentioned. Paper files, probably easier to use, seem to be associated with better knowledge of occupation than electronic files ( 8 ). Beyond these organizational aspects, GPs’ knowledge might also vary as a function of their participation in peer-group meetings: integration of the patients’ socio-economic circumstances into personalized management is supposed to be one aspect of quality that these groups are supposed to improve ( 9 ).

The aim of our study was to analyse the association between GPs’ characteristics (in particular, their organization of their work and their participation in peer-group meetings) and their knowledge of their patients’ socio-economic circumstances.

Methods

This study is an ancillary analysis of data from an observational survey named Prev Quanti ( 10 ). This survey was initially designed to document social inequalities in preventive care (breast, cervical and colorectal cancer screenings, discussion of tobacco and alcohol consumption and cardiovascular risk) provided by French GPs (see Box 1 for a brief overview of the French health care system).

To have a sample size large enough to be able to study the cancer screening tests recommended for patients aged 50–74 years, while still being able to analyse young patients at low cardiovascular risk, we chose to include only patients aged 40–74 years. A power calculation determined that we would require 50 GPs and 70 patients per GP to be able to demonstrate social gradients for the types of preventive care studied.

GP recruitment

The Prev Quanti study was conducted in 2008–09 among GPs who supervised students training in general practice during an internship at their surgery (office). We used email and telephone to recruit from among the 215 GPs working with two medical school departments of general practice in the Paris metropolitan area (participating GPs were paid €300 for their time).

GP characteristics

Three categories of GP characteristics were collected by self-administered questionnaire:

• - demographic characteristics: age (<50 years, [50–55], [55–60], >60), sex and duration of practice (≤20 years, [20–30], >30);

• - surgery organization: group practice, average duration of consultation (<20 minutes, ≥20), computerization of medical files and the use of an automated reminder system;

• - participation in peer-group meetings: both continuing medical education and peer-review groups were considered.

Patient recruitment

For each GP, a random sample of 35 men and 35 women aged 40–74 years was drawn from their patient list (patients who had declared the GP to be their regular doctor), furnished by the national health insurance fund. There was no exclusion criterion.

Patient characteristics

The patients’ socio-economic characteristics were collected simultaneously from the patients and their GPs by matching questionnaires. The patients received a questionnaire mailed to them by the GP, who in turn was supposed to complete a form for each patient included, using information in the medical files or what he or she knew about the patient that was not included in the file.

Eight socio-economic characteristics, collected from GPs and their patients, were categorized in three groups describing different aspects of the patients’ socio-economic circumstances.

• - Their health insurance status:

  • - Publicly funded special health insurance (for all very low-income legal residents, providing them with free health care);

  • - Complementary health insurance;

  • - Special coverage for a chronic disease: care, including medication, related to chronic conditions such as diabetes, is free (100% reimbursement).

• - Their socio-economic status:

  • - Employment status: coded in six categories (working, unemployed but seeking employment, student, retired, other not in the labour market or disabled on a temporary or permanent basis);

  • - Occupational group: based on the patient’s last job, coded into five categories derived from the standard classification of occupations in France and ranked as follow: professionals and managers, intermediate white-collar workers, shopkeepers and craft workers, office, sales and service workers and skilled or unskilled manual workers. This occupational ranking, derived from work by the national institute of statistics and economic studies (Insee), was determined by a multidimensional scaling procedure that used homogamy (i.e. marriage to a spouse belonging to the same social group, like marrying like) between different occupations. This technique, based on the statistical analysis of a distance or dissimilarity matrix between individuals, is often used to construct status or prestige scales;

  • - Financial situation: this characteristic was obtained by using the same question about the patients’ financial situation but with different multiple-choice responses for patients (‘I’m not managing’, ‘It’s tight, I need to be careful’ or ‘It’s OK, we’re comfortable’) and doctors (‘difficult’, ‘intermediate’ or ‘comfortable’).

• - Their social support:

  • - Lives with a partner;

  • - Social network: this binary characteristic (good versus not good) was obtained differently from patients and physicians. Patients completed Berkman’s social network questionnaire, which explored all their social ties ( 11 ). Their responses were used to calculate a four-level social integration index, which was then dichotomized as high or low social support. Doctors were simply asked: ‘does the patient have good social support?’.

Assessment of GPs’ knowledge of their patients’ socio-economic circumstances

For each of these eight questions about each patient’s socio-economic circumstances, the doctors could either answer or say ‘don’t know’. For each of these characteristics, the GPs’ knowledge was estimated by a binary variable called agreement (versus disagreement). Agreement was coded 1 when the patient and doctor provided identical responses and 0 when their responses were discordant or when the doctor responded ‘don’t know’ (regardless of the patient’s response, as the patient could not answer ‘don’t know’). When the response were discordant, we determined whether the GP’s response was an overestimate compared with the patient’s response, taken here as the reference (for example, if the doctor reported that the patient was a manager, when the patient said he was an office worker).

Statistical analysis

Agreement for each of the eight characteristics was analysed as a function of the GPs’ characteristics, with mixed logistic models with random intercepts to take into account the hierarchical structure of the data (that is, that the patients are not independent but are grouped by GP). Each model included only one GP characteristic at a time but was adjusted for age (40–49 years, 50–59 and 60–74), sex, duration of doctor–patient relationship (binary variable with a cut-off at 2 years) and the number of patient consultations during the past year (binary variable with a cut-off of 3). We were thus able to estimate not only standard odds ratios but also the inter-GP variance, that is, the variability in agreement rates between GPs.

All analyses were performed with SAS statistical software. The study was approved by the National Data Protection Authority, which is responsible for ethical issues and protection against inappropriate collection of individual electronic data. All patients provided informed consent.

Results

The first 52 GPs who volunteered to participate were included in the study. Of the 3640 randomly selected patients, the return rate for physician questionnaires was 98.9% ( n = 3600), while the patient participation rate was 71.6% ( n = 2605). Finally, these socio-economic characteristics were available from both the patient and the doctor for 71.4% ( n = 2599) of patients.

Participating GPs were mostly men ( Table 1 ), had a mean age of 54.3 years (SD = 6.3) and had been in practice for a mean of 25.4 years (SD = 6.3). The mean duration of their consultations was 21.7 minutes (SD = 4.8). The patients’ mean age was 53.9 years (SD = 9.5) ( Table 2 ).

GPs’ knowledge of their patients’ socio-economic circumstances

GPs were well aware of their patients’ health insurance status ( Table 3 ). The highest rates of agreement among all those studied were for publicly funded special insurance coverage and special coverage for a chronic conditions (both >85%). In contrast, the characteristics of occupation and financial situation had the highest disagreement rates (>45%). The characteristics related to patients’ social support had an intermediate position between the other two groups: the GPs could not always answer (13% to 20% of ‘don’t know’ responses), but their responses, when they were not ‘don’t know’, were only slightly discordant from those of the patients (~10%). The GPs’ discordant responses for all socio-economic characteristics were mainly overestimates.

Analysis of GP characteristics

The physician’s sex was the only demographic characteristic associated with better knowledge of the patient’s socio-economic circumstances ( Table 4 ). Agreement for the patients’ employment status and occupation was higher for women doctors than for men.

For surgery organization, physicians with consultations lasting longer than 20 minutes had higher agreement rates for employment and occupational group. Use of computerized files was associated with less frequent agreement for two patient characteristics: social network and financial situation. The use of automated reminders was associated with better agreement about ‘lives with a partner’.

The characteristics related to GPs’ participation in peer-group meetings were associated with all three groups of patient socio-economic characteristics. Participation in continuing medical education was the GP characteristic most strongly associated with good knowledge of these circumstances. It was the only GP characteristic associated with greater agreement about complementary health insurance. Participation in peer-review groups was associated with better agreement about chronic disease coverage and employment status.

Discussion

Main findings

The GP characteristics associated with better agreement were sex (female), longer consultations, use of paper records, use of an automated reminder system for those with computerized records and participation in continuing medical education and in peer-review groups.

Strengths and limitations

The principal strength of our study is its patient participation rate, which is better than that usually observed with this type of design. Furthermore, the definition of our agreement variable includes the ‘don’t know’ responses, which accounted for up to nearly 30% of the GPs’ answers. Analysis of agreement with a kappa coefficient or in terms of sensitivity and specificity would have required limiting analyses to patients for whom GPs had given responses other than ‘don’t know’. We considered that the absence of knowledge (‘don’t know’) and discordant answers both constituted erroneous perceptions of the patients’ socio-economic circumstances.

Even though our sample of GPs is similar to the entire population of French GPs in terms of age, sex, duration of consultation and group practice ( 12 ), our participants were all volunteers who were directing the internships of student GPs. This selection is the principal limitation of our study. It raises problems especially for the description of GPs’ knowledge of their patients, but less for the association between GPs’ characteristics and this knowledge. Women doctors, for example, seem to raise their patients’ socio-economic situation during appointments more often than their male counterparts. It is probable that this association is similar in ordinary run-of-the-mill doctors and in those in our sample.

Moreover, all the physicians of our sample have remained involved with their local medical school. This homogeneous recruitment of GPs might also have limited the variability within our sample and therefore the power of the tests we performed. Nonetheless, the low variability in inter-GP agreement for some patient characteristics may also be explained by the fact that they are frequently taken into account during the consultation: in France, very low-income patients with publicly funded medical coverage do not pay the doctor at the end of the consultation, unlike those with ordinary health insurance. The existence of a chronic condition is also involved in payment; moreover, the application to the health insurance fund for designation of this chronic condition must be completed by the GP, and all prescriptions must mention it. Partnership status is understood through various reasons for consultation (contraception, pregnancy, sexually transmitted diseases and appointments for children), just as employment status and occupation are indicated by the need for sick-leave certificates and for treatment of occupational diseases and workplace accidents.

Another limitation of this work is related to the different ways that financial situation and social network characteristics were collected from patients and physicians. For parts of the patients’ questionnaires, we used standard tools that have been validated in social epidemiology ( 11 ). Because these complex measurements could not be used for physicians (each of whom had to complete questionnaires for 70 patients), we decide to ask them simple questions for which the answers should have been available to them in their daily work. However, these differences might have induced a measurement bias.

Finally, our results are only slightly statistically significant. Hierarchical models have an advantage over standard models in that they provide unbiased estimates of the association between the agreement rate and GP characteristics. At the same time, however, they have the disadvantage of having less statistical power than the standard models. Confirmation of our results by studies including more GPs would thus be desirable.

Comparison with existing literature

Our finding that GPs have limited knowledge about their patients’ complementary health insurance, occupational group and financial situation [the most difficult information to obtain, after ethnic origin ( 13 )] is consistent with the literature ( 14 ). The fact that women GPs have a better knowledge of their patients’ socio-economic circumstances is also consistent with the results of meta-analyses ( 15 ) showing that patients provide significantly more psychosocial information to women physicians. We looked for but did not find any significant interaction between the patients’ sex and that of the GPs (results not shown).

It is not obvious that the association between better knowledge of patients’ socio-economic circumstances and longer consultations can be generalized to other countries where the durations of consultation are different. But the fact that the same type of association has been observed in England for consultations averaging 9 versus 7 minutes ( 16 ) and that a psychosocial motive for consultation is associated with longer consultations in diverse countries ( 17 ) strengthens our results.

Clinical implications

Patients’ socio-economic circumstances are not only an important element in good medical management but also a major determinant of their health status. GPs may be the health care professionals best positioned to combat social inequalities in health. But carrying out this mission presupposes an accurate assessment of each patient’s circumstances ( 18 ). Our results show that improvements in this assessment are needed. The GPs in our sample had only an imperfect knowledge of their patients’ occupational group, financial situation and investment in complementary health insurance. All these elements, however, play an important role in the patients’ ability to use burdensome or expensive management. Most characteristics related to patients’ health insurance status are well known by GPs because they play a role in the payment process, at the end of the visit. Coverage by complementary health insurance, on the contrary, has the highest proportion of ‘don’t know’ responses (nearly 30%) and should be mentioned more often in medical files. Moreover, the global tendency of GPs to overestimate their patients’ socio-economic circumstances may lead them to suggest management that is too expensive or too frequent for the patient’s work schedule and therefore result in some patients at the bottom of the social hierarchy abandoning treatment ( 19 ).

Our findings do not allow us to conclude that there is a causal association between GPs’ characteristics and their knowledge. The causality might well be circular: long consultations enable better knowledge that, in return, makes the consultations longer to enable the doctor to integrate socio-economic factors into medical decision making. Nonetheless, the GP characteristics associated with better knowledge of their patients’ socio-economic circumstances can be interpreted as the reflection of the doctors’ commitment to their patients, simultaneously quantitatively, by giving them time during consultations, and qualitatively, in working to improve their practices. As such, these organizational factors can be promoted as useful to practitioners.

Further research

According to our results, physicians using paper files or computerized files equipped with an automated reminder system know their patients’ socio-economic circumstances better than those with computerized files and no reminder system. This result should be examined in more detail in other studies but suggests several hypotheses about the more or less ergonomic relation that GPs maintain with the tool they use for patient records. We can suppose that both keeping paper files and using an automated reminder system are associated with greater ease in handling files and that this allows GPs to add more information to their files (such as those related to socio-economic circumstances) or to be more available during the consultation to talk about their patients’ lives.

Research should also be conducted into how to integrate precise identification of patients’ socio-economic circumstances into the standard consultation. A first path for this research should attempt to understand how GPs assess their patients’ social position so that their misunderstandings can be corrected [does this result from selective sociability or erroneous deductions ( 20 ) or stereotypes?]. Another path would be to teach GPs sociological approaches to situating people socially indirectly (that is, without explicitly asking excessively intrusive questions) and to determine if this training improves their knowledge of their patients’ socio-economic circumstances.

Conclusion

The GPs of our sample have fairly good knowledge of their patients’ socio-economic circumstances. Nonetheless, their systematic overestimation of their patients’ social position could lead to increasing social gradients if the management devised by the GPs causes some less well-off patients to abandon it. Having adequate time to spend with each patient, mastery of patient files and participating in peer meetings are organizational aspects to be promoted among GPs, to enable them to practice in accordance with the biopsychosocial model and to optimize the social equality of the care they provide.

Declaration

Funding: Union Régionale des Professionnels de Santé (URPS) Médecins libéraux d’Île de France; Institut national de prévention et d’éducation pour la santé (Inpes).

Ethical approval: the study was approved by the National Data Protection Authority (CNIL, Commission nationale de l’informatique et des libertés), which is responsible for ethical issues and protection against inappropriate individual electronic data.

Conflict of interest: none.

Acknowledgements

The authors are extremely grateful to all the GPs and patients who took part in this study, as well as to the Société de Formation Thérapeutique du Généraliste (SFTG) for its logistic support and to the general practice department of Paris Descartes and Pierre et Marie Curie Universities for its active collaboration in the performance of this study.

References