Patient satisfaction with availability of general practice: an international comparison

Abstract

Introduction

Availability of the primary medical care team 24 hours a day is a core public demand, a demand that can only increase if the responsibility for patient care shifts from secondary to primary care. Delivery of out of hours services has received a great deal of attention in the past years, especially in the health care systems where family physicians/general practitioners (GPs) are no longer able to provide these services on a personal basis []. Personal continuity of care is highly valued by patients [2–4], but it requires that the care provider is available for patient care most of the time. It has been suggested that the ideal of personal continuity in general practice should be replaced by that of organizational continuity [5]. The percentage of GPs working in solo practices varies between countries (for example, 16% in THE UK and 69% in Belgium), but it is decreasing in most countries [6]. Personal continuity of care may be difficult to maintain, given the increasing number of GPs in larger health care organizations. In this paper we examine how patients in different countries evaluate the availability of general practice care.

Previous research has shown that patients are more satisfied with general practice if they have a personal GP and if they experience short waiting times [5,7,8]. Different types of out of hours services have proved to be satisfactory for providers, but less satisfactory for patients [,6,9]. Free flow of patients and administrative barriers might be damaging for the quality of care [10]. We expected to find that patients were more satisfied with the availability of general practice care if their GP worked more hours per week, if their GP had more years experience in general practice, and if their GP worked in a practice with few or no other GPs or care providers. We did not expect to find a relationship between patient satisfaction and the GPs’ age or sex or the setting of the practice (village or city). We also hypothesized that the existence or non-existence of these associations was not influenced by the national health care systems, because the factors studied vary within countries and patients’ priorities in general practice are consistent across countries [4].

A previous paper provided a global descriptive overview of patients’ evaluations of general practice care [11]. The current study aims to determine: (a) which characteristics of the GP and the general practice setting were associated with patient satisfaction with availability of general practice, and (b) whether the same associations were found in different European countries.

Methods

Design

In 1998, an international study of patients’ evaluations of general practice was carried out in 16 European countries using written surveys completed by patients.

Setting

The patient surveys were administered in general practice care settings and included approximately 24 000 responding patients. For this paper we selected the nine countries that also provided data on GP characteristics and a unique GP identifier in the data on patients’ evaluations of care: Denmark, Germany, The Netherlands, Norway, UK, Belgium (Flanders and Wallonia), Switzerland, Slovenia, and Spain.

Study participants

In each country, a stratified sample of approximately 36 practices was sought; in Belgium separate studies were carried out in Flanders and Wallonia (as if these were separate countries). The sample of practices was stratified according to practice size and urbanization in each country to reflect the national situation as closely as possible. In Denmark, Belgium, Switzerland, Slovenia, and Spain only one GP from each practice could participate to reduce statistical clustering of data. Where feasible, this GP was randomly selected. Short written questionnaires were mailed or given to each participating GP.

The patient population comprised adult individuals who had recently visited the GP. We aimed at 1080 patients per country (30 per practice) to allow a reliable comparison between countries (alpha = 0.01, P = 0.90, icc = 0.05, standard deviation = 0.8, and minimal relevant difference = 0.3 on a five-point scale; figures based on pilot studies with the instrument). The number of patients approached varied between 45 and 80 per practice (a consistent number per country), depending on the response rates that were expected by the national researchers. Patients were included if they were 18 years or older, and were able to understand the national language. The GP handed out a written questionnaire to all eligible patients consecutively visiting their practice after a chosen starting point. The patient was asked to complete the questionnaire at home and send it in a prepaid envelope to the research unit. Where feasible, reminders were sent three weeks after handing out the questionnaire.

Measures

GP questionnaire. Each GP filled out a questionnaire that asked questions on their age in years, sex (male = 1, female = 2), the number of years working in the practice, the number of hours worked per week, and the number of GPs (categories: 1, 2, 3–4, ≥5) and care providers (including GPs) in the practice organization (categories: 2–4, 5–10, >10). Note that this information was collected independently from the patient survey. Finally, the researchers recorded the urbanization level of the area where the practice was located (village, <15 000 inhabitants = 1, city/town, >15 000 inhabitants = 2).

Patient questionnaire. We used the Europep instrument, a multidimensional instrument comprising 23 questions on evaluations of specific aspects of general practice care and using a five-point answer scale with the extremes labelled as ‘poor’ and ‘excellent’ [12]. For this study we focused on patients’ evaluations of five aspects of the availability of the general practitioner: (1) getting an appointment to suit you, (2) getting through on the phone, (3) being able to speak to the general practitioner on the telephone, (4) waiting time in the waiting room, and (5) providing quick services for urgent health problems. In this paper we report the percentage of patients using the two most positive answer categories, because we felt that this measure had a consistent meaning across different countries. The questionnaire focused on the GP who was usually seen by the patient, which was often (but not always) the GP who handed out the questionnaire.

Data analysis

In this cross-sectional study with patients clustered within GPs, we used a multilevel logistic regression analysis (using SAS version 6.12) to examine the associations between GP/practice characteristics and patients’ evaluations of care [13]. The analyses were based on the logistic mixed effect model (GLIMMIX MACRO in SAS), which analyses fixed and random effects, because of the hierarchical structure of the study (patients nested within practices). We performed separate analyses for each aspect of care evaluated by the patients (response variable). We did not take potential clustering between GPs into account (particularly relevant in countries that provided more than one GP per practice), as the proportion of patient data clustered within GPs was small and we preferred to use the same two-level model in all countries. The regression models included patients’ age and sex as potential predictors of the variation of scores between patients (level one) and the six GP/practice characteristics mentioned above as potential predictors of the random variation of scores between GPs (level two). Patient age and sex were included in all the models.

Firstly, we calculated unconditional and conditional odds ratios for the total dataset from all countries (unconditional odds ratios were not controlled for the effect of other GP/practice characteristics, but were controlled for patients’ age and sex). For the calculation of conditional odds ratios, we entered all potential predictors in the model and reported on the results. We substituted the missing number of care providers per practice in The Netherlands on the basis of the data from Denmark, and the missing number of hours worked per week in Wallonia on the basis of the number from Flanders.

Next, countries were entered into this overall model to check for differences between countries and for interaction effects of countries with GP/practice characteristics.

Finally, we analysed each country separately, using a backward selection of significant indicators. GP/practice characteristics that were non-significant predictors (P > 0.05) were removed to determine the final model. We report the significant associations of GP/practice characteristics as the conditional odds ratios in the final logistic regression model. We performed 350 statistical tests (ten countries, five outcome variables, and seven predictors), so the results should be interpreted cautiously as associations may be significant by chance.

Results

The nine countries that could be included in the analysis provided 15 996 patients, but in Wallonia and Spain the intended sample size was not reached (Table 1). The response rates varied from 47 to 89%, but were >70% except in Wallonia (47%) and Switzerland (69%). The mean age of patients was about 50 years, and approximately two-thirds of the patients were women. A total of 481 GPs was included in the analysis (Table 2). Their mean age was in the forties, except in Denmark and Germany where the mean age was 51 and 50 years, respectively. The majority were men, except in Slovenia where most GPs were women. GPs in Flanders worked the most hours per week. In the UK and Spain, they typically work with several other GPs in a practice. In Slovenia and Spain, GPs worked in health care organizations with many other care providers.

In all countries, a majority of patients had very positive evaluations of the accessibility of general practice (Table 3). There was, however, significant variation in the percentages of patients who rated aspects of care as ‘very good/excellent’ across different countries, different patients, and different GPs. The proportion of variation in patients’ evaluations of care that related to systematic differences between GPs varied between aspects of care and countries, but it was usually <10%.

The overall odds ratios, based on data from all countries combined, are reported in Table 4. Patients’ evaluations of each of the five aspects of availability were more positive in practices with smaller numbers of GPs and care providers. Patients’ evaluations were also more positive if the GP worked more hours per week, except for the patients’ evaluations of the provision of services for urgent health problems. In addition, there were some associations with GPs age, sex, and number of years in practice. Looking at the conditional odds ratios, the number of GPs in the practice proved to be the dominant factor; it remained significantly related to patients’ evaluations of all aspects of the availability of care. Furthermore, GPs in villages received more positive patient evaluations for getting through on the phone, and male GPs had more positive patient evaluations for waiting time in the waiting room.

This analysis of overall conditional odds ratios also showed significant differences between patients’ evaluations in different countries. A second regression model, which added the interactions of country and GP/practice characteristics, showed that the number of significant interaction effects varied between three and six (out of the seven interactions studied) for the different aspects of availability. Therefore, we considered it relevant to continue with exploratory analyses for the countries separately.

Patient satisfaction with getting an appointment was highest in Germany (94%) and lowest in the UK (62%) (Table 3). Table 4 shows that patients in Denmark, Germany, Norway, and the UK were more satisfied if fewer GPs worked in the practice. Patients in The Netherlands and Norway reported more positive evaluations if the GP worked more hours per week. Patients in Denmark were more satisfied if the number of care providers in the practice was lower. In Slovenia the number of care providers was related to patient satisfaction, but the direction was unclear. Patients in Germany were more satisfied if the practice was in a city, but patients in the UK were more satisfied if the practice was in a village. Patients in Denmark were more satisfied if their GP was male.

Patient satisfaction with getting through to the practice on the phone was highest in Germany (95% positive) and lowest in Denmark (53%) and Norway (56%). Patients in Denmark and The Netherlands were more positive about this aspect of care if the number of GPs in the practice was smaller (Table 4). In Denmark, this effect was caused by a less favourable patient evaluation of practices with two GPs. Patient satisfaction in Denmark and the UK was higher if fewer care providers worked in the practice, and in Switzerland, patients were more satisfied in practices with more care providers. In the UK and Slovenia, patients in villages were more satisfied about this aspect of care, but in Germany, patients in cities were more satisfied. In The Netherlands and Norway, patients had more positive evaluations if the GP worked more hours, but in Spain and the UK, patients were more satisfied if the GP worked fewer hours. In Spain, patients with younger GPs and with GPs who had more experience were more satisfied.

Patients in the UK and Norway were least satisfied with the ability to speak to the general practitioner on the phone (51 and 54% of the patients had most positive evaluations, respectively). In Germany, The Netherlands, Norway, and Switzerland, a lower number of GPs in the practice was related to higher patient satisfaction, but in Denmark this effect was reversed (Table 4). In The Netherlands and Norway, patients were more satisfied if the GPs worked more hours per week, but in Spain this relationship was reversed. A lower number of care providers in the practice was strongly related to lower patient satisfaction in Denmark and the UK.

In all countries, waiting times in the waiting room were less favourably evaluated compared with other aspects of care (Table 3). Table 4 shows that patients in Denmark and Germany were more satisfied if fewer GPs worked in the practice. In Denmark and Flanders, patients in cities were more positive about this aspect of care. In Flanders and Spain, patients were less satisfied if the GP worked fewer hours per week. In the UK, patients with younger GPs were more satisfied, but in Slovenia, patients with older GPs were most satisfied. In the UK, fewer care providers in the practice was related to higher patient satisfaction, but in Slovenia, more care providers was related to higher patient satisfaction. Finally, patients in Flanders were more satisfied with waiting times if the GP was male.

Patients had very positive evaluations of the GPs’ provision of services for urgent health problems. In Denmark, patient satisfaction was higher if the number of GPs in the practice was larger, but in Norway this relationship was reversed (Table 4). In The Netherlands, patients had more positive evaluations if the GP worked more hours per week and if the practice was in a village. In Norway, patients were more satisfied if the GP was older. In the UK, patients were more satisfied if the number of care providers in the practice was low. Finally, patients in Spain were more positive if their GP was female.

Discussion

Patients’ evaluations of the availability of general practice care in the nine European countries were positive, but there was some variation, which was associated with characteristics of the patients and the doctors. We found confirmation of our expectation that more favourable patient evaluations of availability were associated with working more hours per week as a GP, and with fewer GPs and other care providers in the practice. There was little evidence to support our hypothesis that more years of experience as a GP was related to higher patient satisfaction. There were some associations with GPs’ age, sex, or the geographic location of the practice (village or city), but we did not identify a clear pattern. There were some contradictory relationships, and none of the associations was consistently found in all countries, whereas in Wallonia, patient satisfaction was not statistically associated with any of the GP or practice characteristics.

Overall, this study suggests that there is an association between patients’ positive evaluations of the availability of the GP and GPs spending many hours per week on patient care in practices with few other GPs or care providers. We were not able to elucidate in this cross-sectional study whether this was an effect of a selection of special GPs to work single-handed, a selection of specific patients to smaller practices, or an effect of cultural factors or organizational conditions.

This study used a previously validated patient satisfaction instrument and large, reasonably representative samples of patients and doctors. We did not have data on GP characteristics for all 16 countries that participated in the Europep study, which may have led to a selection bias. Countries not included were: Austria, France, Finland, Iceland, Israel, Portugal, and Sweden. Wallonia had a relatively low response rate, which may be related to a weaker research tradition. In group practices in Germany, The Netherlands, Norway, and the UK there may have been some confusion concerning the GP to whom the questionnaire referred. The high patient satisfaction ratings may have caused a ceiling effect, so that it was difficult to identify differences between GPs. Collinearity (correlations between the GP/practice characteristics) may have influenced the results of the stepwise analyses per country.

The findings of this study in general practice are similar to those in a study in a hospital setting, which showed that patient satisfaction is higher in smaller health care organizations [14]. The authors claimed that patients perceive larger hospitals as impersonal and intimidating. An alternative interpretation would be that larger health care organizations attract a specific type of patient or doctor. The availability of a doctor’s list of patients may have a positive influence on patients’ evaluations of care, because it emphasizes the personal relationship between a patient and a doctor [15]. Personal lists are less often used in larger health care organizations. Patients in managed-care settings were less satisfied with continuity of care compared with patients receiving fee-for-service care [16]. Further research should reveal which factors determine patients’ negative evaluations of larger health care organizations.

This study showed that the findings on determinants of patient satisfaction cannot easily be generalized across countries. There is considerable variation in the organization of general practice care within each of the countries, so a detailed analysis of the determinants of patient satisfaction within a specific country should complement an international comparison of patient satisfaction with care [3]. A positive association of patient satisfaction and the number of working hours was found in The Netherlands and Norway, but ths association was reversed in Spain. Do GPs in The Netherlands and Norway work too few hours, as far as patients are concerned, and GPs in Spain too many hours?

This study raises the question of whether patients’ preferences can be combined with the development towards larger practices and GPs working fewer hours per week in many European countries. A better collaboration between GPs provides opportunities for professional development, quality development, and organizational flexibility, which may ultimately overcome the disadvantages of larger practices and fewer working hours per week. Maybe GPs should organize themselves in networks of small practices rather than large clinics with many health care professionals. The organizational development of general practice should not only be determined by the professional perspective, but also by patients’ needs and preferences.

Acknowledgements

We would like to thank Reinier Akkermans of the Centre for Quality of Care Research, Nijmegen, The Netherlands, who performed the multilevel regression analyses for this study.

References