Skip to main content

Quality and costs of primary care in Europe

Final Report Summary - EUPRIMECARE (Quality and costs of primary care in Europe)

Executive summary:

Throughout three years, EUPRIMECARE project developed methods to describe primary care (PC) models existing in Europe, as well as to assess their quality in different dimensions and to determine their cost. The final step of this research project was to analyse the links between quality and costs and to provide recommendations in order to improve the value of PC.

The description of PC models was based on a framework including several domains scope of PC, financing, regulation, paying, organisation, and behaviour . The most relevant factors to classify PC models are:

- Financing: Social Security of National Health System.
- Ownership: solo or group practices, or integrated networks of health centres.
- Facilities: public or private.
- Access to specialist: direct access or gatekeeping through PC.

The review of the existing systems to measure macro-costs of PC identified a wide variation in accounting systems across countries with no comparable measurement of PC costs available. A questionnaire was developed to collect cost data based either on activities performed in PC or on costs for professionals and medical goods in PC. For the micro costs part, the consortium developed a series of clinical situations (vignettes) and using time-driven activity-based costing the different components of cost (time of professionals, drugs or consumables, additional tests) were identified.

Quality was assessed through a process which started with information generated from focus groups both with patients and with professionals. The output of those discussions were used to conduct a population and a professional survey. These surveys investigated equity, appropriateness, satisfaction and users access to health care services, including screening, counselling and prevention activities, from population and professional perspectives. Information from medical records related with PC role in managing diabetes and hypertension was also elaborated.

All this information permitted a valid comparison between the identified European PC models and an assessment in terms of agreed quality parameters and costs.

Project context and objectives:

While health expenditure continues to rise (up to 11 % of GDP in some European Union (EU) countries) the needs of patient groups such as the elderly and chronically ill are still not well met. Increased wealth and an ageing population as well as increased complexity in health systems are contributing to inadequate treatment and care.

EC white paper (2008-2013) highlights the need of EU health systems to address common public expectations across Europe, migration and mobility of patients and health professionals, as well as the enlargement of the EU. It identifies the need for a community framework to be developed by 2013 for safe, high quality and efficient health services. PC will play a pivotal role in reforming health services. PC is usually the point at which a patient enters the health care system, where most of patient’s health problems could be examined, and where decisions on how best to treat the patient could be made. To raise the standards of PC there is a need for transparency regarding quality and cost performance by health care providers at regional, national and EU level.

However, while literature reviews of health systems are plentiful, a common framework to describe PC models in the EU is not available. In addition, a consensus on how to define quality of care in health systems does not currently exist. Costs of PC are not well identified in national accounting systems. Research tends to focus more on secondary care due to higher budgets and innovative technology intensive procedures. Comparable databases and research methodologies are therefore required to analyse and assess the quality and cost of PC across EU Member States. The key strategy of this proposal is to establish, in line with EU health policies, a common research method to explain PC models in EU countries and best practice for identifying quality and measuring the associated cost. The outcome will be a set of research methods and tools for informed policy decisions to improve quality and efficiency of PC, to contribute to equity and social cohesion, sustainability and quality of health systems.

The overall objective of this project was to use research methods to describe different PC models in the seven EUPRIMECARE countries (Estonia, Finland, Germany, Hungary, Italy, Lithuania and Spain), evaluate their quality in different dimensions and determine their cost. In order to analyse variations of both quality and cost as they apply to specific organisational models in Europe, and to study the possible trade-offs between quality and costs in each model, development of specific .

In order to assess costs of PC in different countries or models, two different approaches were combined: the macro and the micro approach. While macro level data reflected aggregated costs of PC at regional or national level, the micro level approach focused on specific and relevant cases of providing specific PC services at the clinical level. As the macro cost data were supposed to suffer from a lack of precision, the micro approach was an essential module for accurate cost comparisons of different countries or models.

With regards of quality, after the literature review the consortium agreed four quality dimensions: access, equity, appropriateness, and patient and professional satisfaction. Moreover, each beneficiary carried out focus groups discussions (FGDs), with the aim to study the perception and priorities of professionals and patients regarding quality criteria in PC. These quality dimensions and the results of the FGDs supported the selection of indicators for quality assessment of PC. Once the quality indicators were identified, the project started the assessment of quality through a professional survey, a survey among PC users, and the collection of clinical information through medical records.

In the last stage, the project aimed to investigate the links between cost and quality. Using the sources of information previously built (micro-vignettes, surveys and clinical records) and the stated goals of PC in each model, we were able to prove the capacity of improvement, taking into account the stakeholders interests and cost of improvement.

The differences across countries and models in the following variables were investigated:

(1) health status and socio-demographic characteristics;
(2) perceived satisfaction;
(3) intensity of activity of PC;
(4) control and follow up of chronic diseases;
(5) prevention of disease and health promotion;
(6) utilisation of health care services;
(7) availability of electronic medical records and computerised information systems.

Project results:

Defining models of pc systems in Europe

The project started with the identification of the PC models existing in EU, specifically in EUPRIMECARE countries. The process followed combined both a literature review and a narrative description of types of PC, as well as a statistical analysis. This proposal, besides the traditional classification of health systems (Beveridge and Bismarck) included five domains (financing, regulation, payment, organisation and organisational behaviour).

EUPRIMECARE developed a framework to define PC models in Europe. The framework to define those models was based on a series of specific characteristics of PC systems related with variables associated with five domains: financing, regulation, payment mechanisms, organisation and organisational behaviour. These five categories cover the whole range of mechanisms and alternatives that could be influenced by healthcare decision-makers to improve health systems performance. The logic is that actions on those five global domains could have effects on determining quality and costs of health systems. So they should be considered as targets for proposing reforms of health care systems:

(i) Financing refers to all mechanisms for collecting the money that pays for activities in the health sector. These mechanisms include taxes, insurance premiums, and direct payments by patients. The design of the institutions that collect the money (e.g. insurance companies, social insurance funds) is also part of this control knob, as the allocation of resources to different priorities.
(ii) Regulation refers to development of public policies to alter the behaviour of actors in the health system, including providers, insurance companies, and patients.
(iii) Payment refers to the methods for transferring money to health-care providers (doctors, hospitals, and public health workers). These methods in turn create incentives, which influence how providers behave. Money paid directly by patients is also included in this control knob.
(iv) Organisation refers to the mechanisms used to affect the mix of providers in health-care markets, their roles and functions, and how the providers operate internally. These mechanisms typically include measures affecting competition, decentralisation, and direct control of providers making government service delivery.
(v) Organisational behaviour refers to the principles methods and strategies in a health organisation aiming to achieve a high level of quality in the provision of services including the satisfaction in meeting the needs of clients, providers and other involved stakeholders such as payers and society.

These global domains were discussed by all partners and validated with selected stakeholders (policy makers, health care managers and other PC experts), in order to obtain a high level of agreement. Validity was assessed from different perspectives such as content validity, face validity and reliability. Based on those concepts and after a literature search which reviewed the existing published knowledge regarding PC models in Europe, we prepared a template and conducted a descriptive analysis of the information gathered in the templates. The process of work we established tried to combine assessments of the existing evidence through review of the literature; group discussions among experts and partners; and collection of information from primary and secondary sources.

Analysis of these principles together with collected facts, observations and material from the seven countries allowed to point out the following tentative typology of envisaged five basic types/ models of PC:

(i) PC based on choice of specialists’ services
Primary medical care is provided typically by various ambulatory care specialists, who work as private entrepreneurs. The medical doctor is the central figure in the service provision. Typically the doctors own their enterprises and run them based on business principles. Essentially the funding comes from the health insurance agency or agencies that can also be of the sick fund type. This model is dominant in many Central European countries, with the Bismarckian model.

(ii) PC based on individual generalists
In this model, the patients / families may choose a PC doctor who is a general practitioner or family practitioner (these terms are interchangeable in this document). In a variation of this model, the generalist for children could be a paediatrician. In this model, the doctors are individual entrepreneurs and they deal with payers and other authorities as individuals. Gate-keeping and operating on a list of patients are two central features of this model.

(iii) PC based on group practices
The patients usually choose a practice, although they may formally choose to be patients of one doctor. The practice is the enterprise with its own administration, economy and many facilities. It is typical in the UK from practical needs to rationalise daily activities, manage premises, technology and also the growing numbers of staff employed.

(iv) Primary health care through health centres
The term health centre can refer to a variety of service arrangements. Earlier it was stated that group practices tend to be in the process of developing into health centres. In some countries with variable private practice, health centres have been established as service providers to segments of population with special needs. Health centres have not (at least in their original format) been enterprises. They employ doctors, dentists, nurses and various other health professionals typically on a salaried basis. Salary is not necessarily the only way of paying the professionals, but still the most common.

(v) Primary health care through integrated services
It has been perhaps surprisingly common for PC to keep a certain distance (separate organisations, separate facilities, often separate funding) to specialised services, although the specialised services are much needed for consultations and special investigations. Similar integration is also found in some of the integrated health systems of the United States.

(ii) PC based on individual generalists
In this model, the patients/families may choose a PC doctor who is a general practitioner or family practitioner (these terms are interchangeable in this document). In a variation of this model, the generalist for children could be a paediatrician. In this model, the doctors are individual entrepreneurs and they deal with payers and other authorities as individuals. Gate-keeping and operating on a list of patients are two central features of this model.

(iii) PC based on group practices
The patients usually choose a practice, although they may formally choose to be patients of one doctor. The practice is the enterprise with its own administration, economy and many facilities. It is typical in the UK from practical needs to rationalise daily activities, manage premises, technology and also the growing numbers of staff employed.

(iv) Primary health care through health centres
The term health centre can refer to a variety of service arrangements. Earlier it was stated that group practices tend to be in the process of developing into health centres. In some countries with variable private practice, health centres have been established as service providers to segments of population with special needs. Health centres have not (at least in their original format) been enterprises. They employ doctors, dentists, nurses and various other health professionals typically on a salaried basis. Salary is not necessarily the only way of paying the professionals, but still the most common.

(v) Primary health care through integrated services
It has been perhaps surprisingly common for PC to keep a certain distance (separate organisations, separate facilities, often separate funding) to specialised services, although the specialised services are much needed for consultations and special investigations. Similar integration is also found in some of the integrated health systems of the United States.

Measuring costs of PC

Micro-level costs:
The primary objective of this analysis was to cost specific and relevant clinical cases common to PC systems in all partner countries. In order to achieve this, we aimed to:
(a) measure resources consumption in the delivery of certain PC activities to which monetary values can be attributed;
(b) assess differences in the management of specific PC cases across and within countries in terms of types of professionals involved, time spent in the provision of services and clinical behaviours which might be driving costs.

Costing PC services in different countries represents a challenging goal, mainly because of the dissimilarity in the concept and organisation of PC itself. It is a difficult task to develop a one-fits-all method. For this reason, we used the 'vignette methodology' which we considered a lens flexible enough to encompass the variability among PC models but, at the same time, informative enough to provide a common and defined framework. This method is particularly suitable to compare different health systems because it allows for the identification of country-specific threshold parameters (King et al., 2004).

Vignettes are used by researchers to study a wide range of social issues and problems. We found relevant literature on using the 'vignette methodology' to investigate meanings, beliefs, judgments and actions of people in a given situation (Barter et al., 2000). King et al. (2004) suggested to use anchoring vignettes in cross-country comparisons; after that, several scholars focused on different aspects of the vignette-method, as, for instance, the importance of question ordering and wording, or testing the assumption of 'response consistency' (Gupta et al., 2010). The World Health Organisation adopted the 'vignette methodology' to measure and compare across countries people's judgments of healthcare systems’ responsiveness. In a recent paper Bleich et al. (2009) presented the results of the World Health Survey 2003. The main objective of the survey was to explore what determines people’s satisfaction with the health-care system above and beyond their experience as patients. The vignette method was also used in a EU-funded FP6 project, HealthBasket, in which vignettes were used to cost seven inpatient and outpatient services provided in 9 different European countries (Denmark, England, France, Germany, Hungary, Italy, Poland, Spain, The Netherlands).

There is a lack of evidence on using the 'vignette methodology' to cost PC services, however the consortium found it a viable, innovative and appropriate method to conduct the study. Clinical vignettes applied to the present project:

(a) solved the issue of the interpretation of identical questions, through the detailed description of a real medical situation;
(b) were a common denominator in a context of extreme heterogeneity;
(c) allowed to determine how a certain clinical case is managed in PC and all the resources consumed in the delivery of the correspondent PC services.

The selection of the vignettes aimed to identify clinical cases which could fit different PC national contexts and, at the same time, allowed to measure costs and quality of certain services common to all the partners of the consortium. Thus, the consortium designed four vignettes based on the following PC activities common to all partners, representing four main areas in PC:

(1) providing immunisation services to the population - area: disease prevention services;
(2) providing assessment, diagnosis and care of common acute or otherwise new health problems, for example respiratory infections, headaches, gastro-intestinal problems, back and joint problems and similar – to children at the age of 0 - 6 years of age - area: Care of acute but common problems – both for adults and for children;
(3) assessment, diagnosis and initiation of care of long term (somatic) illnesses, for example diabetes, asthma, hypertension and similar – of adults – Area: Care of chronic conditions;
(4) individualised preventive services to adults, for example for losing weight, help with smoking cessation, dietary and other help with hyperlipidemia / organisation of group-based preventive / health promotion activities, for example for weight loss, smoking cessation or similar - area: health promotion services.

Vignettes were built mainly through a review of the literature and face-to-face interviews with selected professionals, who were mostly involved in the clinical cases described. Basically, vignettes were made of two parts: (i) the overall description of the scenario ('What was the situation faced by the professional'); and ii) a questionnaire to guide the respondents to ask in the most specific and homogeneous way. We specified at the beginning of each vignette the dimensions of quality of care which the vignette is most related to.

Vignette 1: An old man wants to be vaccinated against seasonal flu
'A 70-years-old man in good health comes to the practice asking to be vaccinated against the seasonal flu. The patient is known to you.'

Vignette 2: A sick little boy
'A 2-years-old boy goes comes to the practice with his mother. The day before the boy had developed cough with nasal discharge and had fever up to 38.2 degrees of Celsius. The parent has noted a rattling sound in the child's chest. The child has drank liquids normally but has poorer appetite than usual. In the doctor's office he is alert and responsive. He has mild expiratory dyspnea. His breathing rate is 36 times per minute. Diffuse wheezing and rhonchi are heard bilaterally in the lungs. Pharynx is erythematous. Tympanic membranes are otoscopically without signs of inflammation. The boy has had similar episodes of cough, fever and wheezing at the age of 1 year 5 months. He has atopic dermatitis but otherwise has been healthy.'

Vignette 3: Management of a diabetic patient - time period: 12 months
'There is a 65-year-old woman among your patients, who has been diagnosed with type 2 diabetes. She comes in for a follow-up visit: the tests from last week show that her HbA1c is 7 %. She has no complications. She has been taking metformin 500 mg x2. You are her main PC provider for the next 12 months.'

Vignette 4: Health promotion: smoking cessation
'A young woman, aged 35, comes to the practice to get a certificate of 'good health' for practicing a sport. She is in good health, she does sports, she has a good and satisfying job, she does not drink, nor uses drugs. But, upon your question, she reveals that she has been smoking 20 cigarettes per day for the last 10 years.

After defining the vignettes, each partner translated and validated them into their country languages with support of national professionals acquainted with the scenario described by each vignette. Once translated, the vignettes were fulfilled by those professionals responsible for each specific clinical case in each country. Thus, the GP was in charge of three out of four vignettes (with the exception of the vaccination of an old man in Finland and in Estonia, where it was more often provided by a nurse), while paediatricians were mostly implicated in the case of the sick boy. The contact method with professionals consisted of:

- Personally, by interviewers from each country.
- To a group of professionals of the same kind (e.g. a group of GPs, a group of paediatricians, a group of nurses): the number of the members for each group was 20 - 30 and different vignettes have been submitted to the same group.
- Through a written questionnaire: professionals of each group have been requested to answer the questions related to each vignette in writing.

This method allowed the interviewers to introduce the vignettes and the meaning of the Project to the respondents, and, at the same time, to include more interviewees. It also allowed to ease and to speed up the entire process of data collection.

For each vignette partners assessed:
(1) cost of the medical material directly used in the provision of the service described in the vignette (e.g. vaccine);
(2) cost of the medical material consumed as a consequence of the service described in the vignette (e.g. pharmaceuticals);
(3) direct cost paid by patients for the provision of the service described in the vignette (e.g. tariffs, prices);
(4) cost of the professionals involved in the provision of the service described in the vignette (e.g. time spent to visit the patient);
(5) cost of administrative staff involved in the provision of the service described in the vignette (e.g. time spent to register the provision of the service).

Points 4 and 5 were assessed using Time-driven-activity-based costing (TDABC). We defined the time spent in the specific vignette by each professional and then, considering the cost per hour work of these professionals, we obtained the time-driven cost of managing each clinical case on the vignettes (Kaplan et al., 2004; Everaert et al., 2008a; Demeree et al., 2009; Szychta, 2010).

Data collected through vignettes by each partner were put together and synthesised in four different databases, specific per each questionnaire, by the Bocconi University team. The standardisation of data took into account the specific features of each country and the precise aims of the analysis, both in terms of costing and willing to describe the management of the services.

For the analyses we took into consideration the PC models defined in WP2. Specifically, we sought to define different methods to assess the value of 'time-professional' for:
(a) PC model based on individual generalists or group practice;
(b) PC model based on the provision through health centres;
(c) PC model based on choice of specialists’ services.

In total, more than 200 professionals were interviewed for each vignette. The results showed that variability in the monetary value of the time spent in each scenario depended mainly on two circumstances:

- Services were not always provided by the same type of PC professional in every country.
- There were important differences across countries in terms of values and mechanisms of payment for the professionals, even the same type of professionals (e.g. GPs).

Since we cannot change the wages' levels in each country, analysing the variability across countries of how each specific case was managed represented the most interesting and valuable part of this WP.

Overall, the first criteria considered for evaluation was the type of professional directly involved in the provision of the specific service. For each vignette two sets of activities (and related time) were derived through the questionnaires: the clinical activities, to which the time spent in the provision of the service per se was linked, and the recording and administrative activities, namely the time spent in recording the service and to accomplish some administrative activities related to the provision of care. In this last case, a secretary or a medical assistant might be involved. Since the GP and the nurse proved to be the PC professionals mostly involved in the three vignettes, despite differences between systems, data on the variability of the clinical time were analysed for these two types of professionals. Results concentrated on the different amount of time spent by professionals in providing the specific service and on the dissimilar professionals’ clinical behaviour addressing the same clinical case.

Disregarding the specificities of the different cases observed, we could assert that in EU PC settings general practitioners and nurses were the most referential professional figures. Time spent in the provision of services varied significantly, and this variability depends on a number of factors.

In the 'Vaccination elderly man' vignette it came clear how the specific type of professional in charge of the case affected the clinical time spent in the activity: in Finland we observed a clinical time of 6.6 minutes per nurse involved, which was lower than the lowest clinical time resulting for GPs (7.0 in Germany). This aspect depended largely on the particular PC model adopted in the country and on the specific responsibilities that the system attributes to professionals. In the same way, in the 'Little boy' vignette, different professionals were in charge of the case: in Finland, Lithuania and Estonia paediatricians did not play a major role in PC, and in those systems children were generally visited by GPs. Similarly, the number of professionals involved derived from the organisation of PC settings: as for this two vignettes, in Germany the secretary was often involved in the provision because German GPs work as self-employed or entrepreneur professionals in a non-gate-keeping system and use to have an administrative collaborator in their practices.

Despite differences due to PC systems, we observed a certain level of heterogeneity across countries when comparing the average clinical time spent by the main health professional involved, namely the GP or the paediatrician: what surprised the most is the high variability witnessed within countries. Additionally, we found differences in number and type of professionals in the provision of the service within the same system. It is fair to believe that this kind of variability could be contained and controlled: in this sense, guidelines at national, regional and local levels should be circulated among professionals and regional and local authorities should put in place efforts in order to homologate the organisation of services (settings), at least within the area directly controlled.

Remarkably, in Vignette 2 almost all the physicians in every country would prescribe at least one pharmacological treatment to this young patient, while the way physicians prescribed diagnostic tests showed a deep variability across countries: in general, it could be noticed that in countries where paediatricians are in charge of the case diagnostic tests are less prescribed and specialists less involved.

When analysing Vignette 3 (management of a diabetic patient for 12 months), it became clear that variability in clinical behaviour depended on the specific setting organisation where the GP works and on the particular clinical guidelines followed by health professionals: choices on drugs prescribed, diagnostic tests used and specialist involved were dependent on the organisational model in which GP deliver these services. To conclude, all the studies on variability of clinical behaviour and patients management highlight how this variability is related to inappropriateness and / or inefficiency, but causes of dissimilar clinical behaviours within the same country or across countries are not always controllable. Results reached by costing clinical activities perfectly reflect this type of variability, with the additional complication of extremely different levels of wages and stipends for the same kind of professional (especially GPs and nurses).

Macro-level costs:

With regards the macro-costs analysis, is well known and proven for a long time the relationship between health care spending and PC structures and activities. Strengthened PC improves health outcomes and mitigates the growth of health care spending. Most cost savings are achieved through lower predicted use of outpatient specialist care and hospital care. Literature reviews are also dealing with the analysis of inverse correlation of global cost and development of PC. Results from a systematic review of the literature, covering the period from 2003 through 2008 , state that PC contributes to the overall health system performance. The economic conditions (funding system, expenditures, employment status, and payment of the PC workforce) are associated with access, continuity, comprehensiveness, quality, efficiency and population health. A greater supply of family physicians and coordination of care, in terms of team size and composition and specialist outreach in PC is associated with lower total costs of health services and better health. A recent American literature review published in February 2012 demonstrates that PC policy changes (improved PC supply, increased proved PC utilisation, and improved PC practice architecture, like multidisciplinary groups or centres) likely reduce global cost and population health outcome.

At the macro-cost data analysis of PC expenditure, data were assessed with respect to PC models. As a first step, we examined existing international databases (like OECD, Eurostat, WHO HFA) and we could state that comparable PC cost data are not available in them. We, consequently, decided to collect PC costs data from partner countries together with the cost component information. The activities performed by PC were compiled in a previous phase of the project.

The most important factors the determine PC costs in an international context are:

(a) cost contents (human resource expenditure, other direct, indirect and capital cost, pharmaceutical expenditures, Out-Of-Pocket payments (OOP), etc.);
(b) input costs, price level differences of countries;
(c) political emphasis (money available) on PC in the different countries;
(d) the activity / service spectrum of PC.

The latter is related to the basic organisational characteristics of PC (PC is provided in solo or group practices or in health centres), with which the features of practice ownership, employment type of GP, payment method, gatekeeping role broadly determine the type of model of PC.

In the analysis of PC macro-cost data the effects of some determinants were controlled, but others remained, posing further constraints to comparability. Cost components were more or less similar across the countries (mostly without OOP and fully without pharmaceutical expenditures). To moderate differences in price levels of countries we used purchasing power parities. We captured political emphasis by the share of total health expenditure (THE) in GDP and the share of PC expenditures in THE.

A relevant finding is the existence of a high variation in terms of the type of activities or services which are offered to the population in PC in different countries. This information was collected through templates by the partners. Out of the 40 activity groups captured, 13 proved to be common, that is, they can be found in every country, except for Germany. (Hungarian PC includes or partly includes 27, Italian 14, Estonian 28, Lithuanian 28, Finnish 40, Spanish 29 and German 8 activities.) However, because of the characteristics of cost registration systems of different countries, we could not distinguish cost data for common core activities.

Countries provided data on costs mostly for 2005, 2009 and 2010. We chose 2005 as a control, to have data prior to the economic crisis. According to our observation, the effect of the financial crisis was not evident in most of the examined countries.

Regarding PCE, we observed a definite separation of models. The rate of expenditures on PC as a percentage of total health expenditures (in PPS), ranging from the Italian 4.5 % to the German 14.5 % in 2010 fell within three separate groups among the countries examined. The highest was 8.7 - 14.5 % in Spain, Finland and Germany (2010), 6.6-7.6 % in Estonia and Lithuania (2010, 2009), and the lowest was 4.5 - 4.8 % in Italy and Hungary. Models with more complex services - group practices, health centres - showed higher share of total health expenditures, because these systems provide more definitive care on the PC level, and solve those kinds of problems that are the task of specialist health care in models of individual practices. Individual practices had the lowest shares of health expenditure. It was hard to find boundaries of PC both within and across countries, but the defined models offered a viable solution for it. Models captured some of the core differences in the organisation of PC, but probably mainly the differences in the scope of services provided. This was underpinned by the fact that cost differences among countries were largely correlated with the models. In future analysis of PC costs across countries, it is advised to compare countries only within the same model to enhance comparability.

Another lesson for future analysis is that during the evaluation of the effects of PC models on the quality of care and cost effectiveness, PC should be analysed in relation with other levels of care. Keeping the population in good health (with prevention, health promotion), early detection of diseases, prevention of avoidable hospital admissions result in lower health care costs per population in the whole health care system, though not necessarily in PC.

Measuring quality of PC

Surveys:

The EUPRIMECARE consortium considered access, equity, appropriateness, patient- and professional satisfaction as the main dimensions to assess quality of PC. Continuity and comprehensiveness were reviewed as components of appropriateness.

A series of indicators capturing information from those domains were selected in order to measure quality of the different PC models. Those indicators were capable to measure structure, process and outcomes. The process started from the perspective that both patients and professionals have of quality. Therefore, a series of dimensions of quality were identified through focus groups conducted in all countries. Focus groups included both patients and professionals. Their discussions reflected their understanding and views about quality in the different partner countries. Thematic analysis of the transcriptions of these discussions set the basis for elaborating a list of quality criteria. A literature review provided indicators for each of those criteria. Partners prioritised those criteria by scoring according to their importance and measurability. This process resulted in a set of 20 - 20 indicators to be measured by the population and professional surveys. Sixteen indicators were already available through national databases. The indicators which obtained the highest scores to be measured through the population survey were those related to patient centeredness. Regarding the professional survey the highest scores were obtained by indicators of professional satisfaction.

Based on the previous work the project conducted two surveys. The first one was focused on the population and the second one on the professionals of PC. The population survey aimed to measure the perception of quality care received by the patients. Meanwhile, the PC professional’s survey intended to assess their perception of the elements that affect quality of care provided.

The population questionnaire consisted in 5 sections: socio-demographics, satisfaction, utilisation, management of chronic patients, and frequency of prevention activities. The survey was designed in English by consensus in the seven countries. The sample was a representative group of the population in each country. The interviewees were randomly selected from stratified groups based on gender, age, socio-economic status, and region, according to the characteristics of the overall population. A pooling company (QUOTA) translated the English questionnaire into the 7 languages. The company contacted individuals through computer-assisted telephone interview (CATI) methodology until 430 usable surveys were answered in each country. 3020 respondents completed the survey, with less than 5% of missing responses.

The country with the highest number of visits to the GP was Hungary, followed by Italy, Lithuania, Spain, Germany, Estonia and Finland. In contrast, nurse visits were more frequent in Spain, followed by Estonia, Finland, Lithuania, Hungary, and Italy.

More than 75 % of the respondents were satisfied with all the PC items included in the satisfaction section. Items related to waiting times scored the lowest satisfaction levels; whilst social skills of the PC professionals got the highest punctuation. Estonia was the country with the lowest proportion of dissatisfied respondents, followed by Spain, Italy, Hungary, Germany, Finland and Lithuania.

The averages of people reporting chronic conditions across the 7 countries were 35 %, 25 %, 9 %, 8 %, and 6 % for hypertension, hypercholesterolemia, diabetes, chronic obstructive lung diseases, and asthma, respectively. Estonia, Hungary, and Lithuania presented higher proportions than the average in at least 4 of the 5 conditions. Spain, Italy and Germany showed lower prevalences for all the pathologies.

The highest proportion of chronic patients diagnosed by PC doctors was found in Spain (92 %), followed by Germany (80 %), Finland (76 %), and Italy (76 %). Hungary was the country with the highest percentage of prescriptions for these conditions (89 %), followed by Italy (85 %), Lithuania (83 %), Spain (82 %), Finland (80 %), Germany (79 %), and Estonia (76 %). More than 85 % of the respondents in all the countries reported to have their blood pressure measured at least once in their life. More than 80% of subjects received blood sugar and cholesterol measurement at least once in their life in Spain, Finland, Estonia, and Germany compared to Lithuania, Hungary and Italy where we found lower proportions. Weight was measured in more than 80 % of respondents only in Finland and Spain.

More than 75 % of the subjects presenting diabetes, hypercholesterolemia, or hypertension reported to have at least one blood pressure measurement in the last year in all the countries. More than 75% of chronic patients reported blood sugar measurements in the last year in Germany, Estonia, Spain, and Lithuania. Whilst, cholesterol measurements in the last year were performed in the 75% of the patients only in Germany and Spain. Weight measurements in these subjects were less frequent, with percentages below 70 % in all the countries.

The highest percentage of mammograms in women aged 50 to 69 performed in the last two years (as recommended by the European Commission) took place in Italy (93 %), followed by Finland (77 %), Hungary (77 %), Germany (70 %), Lithuania (68 %), Spain (65 %), and Estonia (48 %). The proportion of mammograms at any frequency in younger or older women than recommended were 79 %, 54 %, 54 %, 53 %, 49 %, 47 %, 47 % in Italy, Hungary, Germany, Spain, Lithuania, Estonia, and Finland, respectively.

The proportion of Papanicolaou tests performed in the last five years in women 30 to 60 years old were 87 %, 82 %, 71 %, 71 %, 70 %, 68 %, and 44 % in Italy, Finland, Hungary, Spain, Lithuania, Germany, and Estonia, respectively.

Counselling regarding alcohol consumption, smoking habits, and physical activity was more common in Finland, Italy and Spain, with proportions over 50 %, than in the rest of the countries.

The proportion of subject with at least one chronic condition or older than 64 who were recommended to have seasonal influenza vaccination in the last year were 59 %, 56 %, 54 %, 54 %, 50 %, 24 %, and 12% in Italy, Germany, Hungary, Spain, Finland, Lithuania, and Estonia, respectively.

The professional survey aimed to cover 4 main aspects of the quality of PC systems: accessibility-equity, appropriateness, patient-centred, and professional satisfaction. It consisted in 6 sections: socio-demographics, organisational and financial aspects, activity performed, satisfaction, patient access, and burn-out. The survey was designed in English by consensus by the partners. A pooling company (QUOTA) was in charge of the translation into the 7 languages and of the interviews with the GPs.

Previous to the survey a pilot of 10 interviews was conducted in each country. The final questionnaire included the most relevant and common suggestions. The method for collecting information was different for each country. In Spain interviews took place on face to face basis. In Germany, Estonia, Finland, and Hungary the interview consisted in a web survey. In Italy and Lithuania CATI system was used. For all the countries GPs were selected through random sampling considering geographical representativeness. GPs were contacted until a minimum of 176 interviews were completed in each country. A total a 1 331 surveys were completed.

The novelty of the EUPRIMECARE indicators was that we measured criteria which contain not only the perceptions both of patients and professionals, and referred to the overall characteristics of the PC model as well. Existing tools usually measured quality of a practice (for ex. EPA_European Practice Assessment) and quality of the management (non-clinical) of the whole PC system.

Clinical records:

The collection of information about quality was complemented with data from clinical records, in countries in which this extraction was possible. the diversity across countries both in the modalities of medical records and in the regulation protecting access this type of data for research purposes, implied that this process varied a lot in each country. In fact, Hungary, Germany and Italy were not able to collect information at individual patient level. As we can see in the table above, the results the clinical indicators available for analysis showed significant variations. Concerning hypertension, the proportion of hypertensive patients with blood pressure level at or below goal (= 140 / 90 mmHg) ranged between 33 - 56 % in different countries, being lowest in Lithuania and highest in Estonia. Similarly, the proportion of hypertensive patients whose lipid profile was measured ranged between 24 - 63 %, being lowest in Lithuania and highest in Estonia.

The proportion of patients with diabetes having average blood pressure at or below goal (=130/85 mmHg) ranged between 8 - 62 %, being lowest in Spain and highest in Lithuania. Lipid profile was measured in 23 - 76 % of patients with diabetes type 2. The proportion of patients with diabetes with total cholesterol at or above 4.5 mmol / l and statin treatment ranged between 15 - 42 % and the proportion of patients with diabetes type 2 with total cholesterol < 4.5 mmol / l was 28 - 53 %. HbA1c was screened in 57 - 90 % of the patients with diabetes type 2 (lowest in Hungary and highest in Lithuania). The proportion of patients with diabetes type 2 with HbA1c < 7 % was between 49 - 71 %. Similarly, the proportion of patients with diabetes with HbA1c = 7 - 8.5 % and insulin treatment provided was 49-72%. The proportion of patients with type 2 diabetes having an eye examination ranged between 1 - 46 %, being the lowest in Spain and the highest in Lithuania.

As conclusion in terms of quality, we found that there were significant differences in data collection across countries and thus, the high variability that was identified in the results of clinical indicators.

The relationship between quality and costs of PC

EUPRIMECARE analysis contributed to produce a methodological and analytical framework and data to this debate concerning PC in Europe, including for that purpose, possible trade-offs between different quality dimensions in each model in order to achieve a better balance, possible win-win changes in each model, possible measures to increase coherence between financing options, regulation mechanisms, payment and other policy initiatives, in order to achieve a set of quality standards, or organisational measures, management structures and quality improvement initiatives associated with better quality of clinical services.

Strong PC has been associated with better health outcomes. Nevertheless, the precise definition of what constitutes 'strong PC' has not yet been full defined. Differences in the provision of PC could have also different costs. To what extent these differences in costs translate into better health outcomes has not either been yet clarified. Although this project has advanced the existing knowledge in this area, there are important limitations on the available information in the participants countries that represent a significant barrier to clarify some of the existing question.

Nevertheless, the data we have been able to gather and analysed could provide answers to policy question. The following tables offer the main results of this study.

Potential impact:

Although there is broad policy consensus that both cost containment and quality improvement are critical, the association between health care costs and quality is one of the more controversial topics in health policy. One possibility is that improvements in quality will require increases in cost (or conversely, cost reductions could reduce quality). On the other hand, improvements in quality could lower costs by reducing complications, further diagnostics tests or hospital readmissions. In reality, the association between cost and quality probably falls between these 2 extremes, so that some types of health care costs are associated with high quality and others with poor quality. The effect depends on where the money is spent.

The debate over the cost–quality association has been largely framed by several seminal studies that compared use and outcomes of health care services across areas and populations. These studies documented large variations in cost across areas, with no evidence that higher-cost areas had better quality or health outcomes. Both the methods and the interpretation of these studies have been heavily debated. Other studies of the cost–quality association have compared units other than geographic areas (such as hospitals) using various methods and have come to different conclusions. In fact, there is yet no systematic literature review of evidence on the cost–quality association in health care. Among studies on the association between health care costs and quality, several design characteristics may be critical. First, level of analysis is important because area-level studies may yield different results than provider- or patient-level studies. Second, there are many ways to measure quality, each of which may have different associations with cost. For example, a structural measure of quality, such as nurse staffing per patient, will probably have different cost implications than higher performance on an outcome measure, such as patient functional status. Third, 'cost' can be measured in many ways, such as reimbursement from a health plan or the amount of resources used by a provider. Fourth, studies may use different statistical methods, particularly in adjusting for the effects of health status on quality and costs.

In health policy, we could refer to the iron triangle of health care. The components of the triangle are access, cost, and quality. This concept was developed by William Kissick to illustrate the inherent trade-offs in health care systems. His point was that at any time, you can improve one or perhaps even two of these things, but it had to come at the expense of the third.

Health care systems could have lower costs, but that can happen with reductions in access in some way or in quality. Health care systems, could improve quality, but that will either result in increased costs or reduced access. And of course, access could increase, but that will either cost or result in reduced quality. Based on this approach, the health care system could be more universal, improve quality, or reduce costs, but not the three of them at the same time. The lesson of the iron triangle is that there are inherent trade-offs in health policy. In order to conduct the debates on health care honestly, we would acknowledge these and allow the public to decide what they really want - and what they are willing to sacrifice to get it. The impact of this project lies in the results of the studies carried out and therefore, in the conclusions that we deduced. It is clear that these conclusions will have a different impact depending of the healthcare systems existing in that country.

(1) EUPRIMECARE has provided new tools to evaluate PC quality across Europe through the definition of a set of quality indicators, which have been assessed through:
(a) focus groups of patients and PC professionals;
(b) population survey and PC professional’s survey;
(c) clinical information extracted from medical records.

(2) As we mention before, to strengthen PC improves health outcomes and mitigate the growth of health care spending. EUPRIMECARE offers a method to measure quality of management of different PC systems and allows the comparison of different models independently they are based on health centres, polyclinics, group- or solo practices. Thus, the evidence-based evaluation of quality contributes to improve decision makers and optimise healthcare investment.

(3) EUPRIMECARE has allowed seeing the interaction of PC with other levels of care and the necessity to analyse the quality of PC and cost effectiveness in relation with other levels of care.

The main dissemination activities carried out so far are the following:

(1) International and local events:
UEMO General Assembly (June 2011, Hungary) (ISCIII),
- WONCA, Word Organisation of National Colleges and Academies Europe (OALI),
- 31st Congress of SEMFYC, the Spanish Society of Family and Community Medicine (June 2011, Zaragoza) (ISCIII),
- EHMA European Health Management Association (June 2011, Portugal) (ISCIII),
- HTAi, Health Technology Assessment (June 2011, Brazil) (ISCIII),
- XIth Conference on Quality Management in Health Care DEMIN (June 2011, Hungary) (OALI),
- EHTEL/ELO (European Health Telematic Association) Meeting at eHealth Week 2011 (May 2011, Budapest),
- Workshop at Laín Entralgo Agency (April, 2011, Madrid) (ISCIII),
- Workshop at AETS, (February, 2011, Madrid) (ISCIII),
- National conference of the Hungarian Scientific Association of General Practitioners (October 2010, Hungary) (OALI),
- Workshop at National School of Healthcare (October 2010, Madrid) (ISCIII),
- EFPH, European Forum of Primary Health Care, (August 2010, Italy) (ISCIII),
- Group of Research in PC in Aragón (August, 2011) (ISCIII),
- Workshop at European Forum of PC, Graz (Austria), Integrating Public and Personal Health Care in a World on the Move: the European PC perspective, (September 2011),
- Workshop at Universidad Nova de Lisboa, Inovação e Desafios Organisacionais nos Cuidados de Saúde Primários: Perspectivas de España & Portugal, (November 2011),
- External Expert Panel at Euro REACH Web Platform, Brussels (March 2012),
- 2012 EHMA Congress, European Health Management Association: 'Public Healthcare: who pays, who provides?'. Berna, Switzerland (June 2012),
- 9th HTAi meeting, 'HTA in Integrated Care for a Patient Centered System.' Bilbao, Spain. (June 2012),
- XVI Workshop of Spanish Network of PC. Madrid (June 2012),
- 4th European Forum of PC Biannual Conference, 'Crossing borders in PC'. Gotheburg, Sweden (September 2012),
- 30th Meeting of Spanish Society of Epidemiology. Santander, Spain (October 2012),
- ISPOR 15th Annual European Congress, 'Challenging Times for Health Care Decisions in Europe.' Berlin (November 2012),
- Workshop at National School of Healthcare, 'PC in Europe: does it value what it costs?' Madrid (November 2012).

(2) Publications
- Public Health Research and Policy Journal of the Ministry of National Resources State Secretariat for Health and the Hungarian Association of Public Health Training and Research Institutions, volume 88. No 4. pages 282-283 GYEMSZI,
- ESKI Institute's reports and yearbook.

List of websites: http://www.euprimecare.eu