Overall design and hypotheses
In order to fully understand the underlying mechanisms of PC leading to health system outcomes, this study distinguishes three levels of care. The first level is the system level of PC, encompassing features such as financing, governance and resources. The second level is the provision level, characterised as the delivery of care process at GP practice level. GPs can be seen as the core providers of PC. The third level, are the users of PC services. The features at these three different levels are expected to, directly and indirectly, contribute to health, access/equity, costs, process quality of services and perceived quality of services. The (inter)relations between the different levels and their features and the outcomes are visualised in Figure .
Elements of the study and their inter-relations.
A number of hypotheses will be tested in this study, concentrating on different domains: quality of service provision, patients' perceived quality of care, costs, equity, avoidable hospitalisation and good practices. The main hypotheses that will be tested are:
1. The degree of organisation of PC practices (e.g. higher skill mix and better organisation of out-of-hours care) is positively associated with the process quality of their services; (system → service provision)
2. A strong PC orientation at structure level is positively associated with the degree of organisation of practices and the process quality of services; (system → service provision)
3. Process quality of PC services is positively associated with patient evaluations of PC quality; (service provision → perceived quality of services)
4. The degree of the organisation of PC practices in combination with quality of the PC process is negatively associated with the incidence of avoidable hospitalisations; (system & service provision → process quality)
5. The strength of PC systems (in terms of strong PC orientation at structural level, good organisation of PC practices and high quality of PC services) is negatively associated with total health care expenditures; (system & service provision → costs/efficiency)
6. A strong PC orientation at structure level is positively associated with access at the practice level and patient perceived equality in access by socio-economic status; (system & service provision & access→ equity)
7. The scale of PC organisational units is associated with lower costs and higher quality. (service provision → costs & process quality & perceived quality)
8. Process innovation (case and disease management, patient-centered care, integration of prevention) is associated with lower costs and higher quality; (service provision → costs & process quality & perceived quality)
9. Delegation and substitution of tasks within PC is associated with lower cost and higher quality. (service provision→ costs & process quality & perceived quality)
Multiple methods of data collection are used in this study. First, the study builds upon existing knowledge, by making use of international studies which have previously invested in collecting evidence on PC at the system (structure) level (see Figure ).
Existing data sources to be used in QUALICOPC.
Secondly, new data are collected, using a cross-sectional survey methodology in a multi-actor design. The multi-actor design makes it possible to directly connect information on PC practices to information provided by patients of these practices. The survey consists of:
- A survey among GPs as core providers of PC collecting data at the process level. The survey also collects information on involvement and relations with other PC providers; part of the survey will be modelled on essential elements of the 1993 study mentioned in Figure [47
- A survey among patients that were treated at these PC practices to gather data on the process and outcome level; the methodology for this design has been developed and tested in the context of earlier WHO projects (e.g. performed in Turkey and Russia) [55
]. The survey among patients consists of two questionnaires: one about patients' experiences and one about patients' values. Measuring what patients find important enables the weighing of their experiences [57
Setting and sampling
Data is collected in 31 European countries (including all 27 EU Member States: Austria, Belgium, Bulgaria, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, UK and two Candidate Member States: Turkey and Iceland and Norway and Switzerland) and in three non-European countries (Australia, Israel and New Zealand). In each country we aim to realize a response of 220 GPs. In Cyprus, Iceland, Luxembourg and Malta the desired response is lower (around 75). In each country we aim to draw a nationally representative sample of GPs. Initially, this means that a simple random sampling procedure is used, drawing a random sample from the national register of GPs (if available). To avoid the inclusion of multiple GPs which are subject to the same practice variables, only one GP per practice will be included. In countries where a national register is not available a multistage sampling procedure is used e.g. by combining registers from different regions or municipalities. Furthermore, in large countries with differences in health care systems across regions, we selected a number of nationally representative regions and subsequently randomly selected GPs within these regions.
The patient survey will include patients above the age of 18 visiting a GP who filled in the questionnaire. Hence, in this study there is a focus on patients in PC who actually visited the practice. This means that the outcomes of the survey will represent the views of users of PC, rather than the general population. The questionnaires for patients will be distributed through the PC physicians who participate in the GP survey. The 220 physicians will be asked if a fieldworker may visit the practice to distribute questionnaires to patients who have consulted them. In practice, on a set date the fieldworker will visit the practice and ask patients to fill in the questionnaire in the waiting room, until a response of 10 patients has been reached. Per country we aim for a response of 1800 patients for the experiences questionnaire and 220 for the values questionnaire (see section 'Questionnaire development' for explanation on the questionnaires). In Cyprus, Iceland, Luxembourg and Malta the desired response is respectively 720 and 80 patients. In some countries where additional funds are available, a larger response will be realised to make comparison between different regions possible. In each practice the fieldworker will ask the first 9 patients, who are willing to participate, to fill in the experiences questionnaire and the 10th patient to fill in the values questionnaire.
Based upon earlier research it is known that the total numbers per country are sufficient to relate country characteristics, PC practice variables and patient evaluations; moreover numbers are large enough to produce reliable country level estimates of patient evaluations of PC [58
]. Several measures will be taken to acquire sufficient response to the survey. Firstly, per country, a national expert will be commissioned as a national coordinator of the fieldwork. Secondly, the coordinator will be asked to organise acquisition of national support from professional organisations for the study. Thirdly, financial resources will be made available to serve as incentives for GPs. Finally, in each country we will have extra versions of the patients' questionnaires available in the languages of the largest ethnic minority groups.
An overview of the methodologies used in the QUALICOPC study is shown in Figure .
Overview of the QUALICOPC study protocol.
To collect new data required for this study, questionnaires are already developed. The questionnaires need to contribute to the provision of usable data for a variety of topics on PC in Europe. For the GP questionnaire these topics concern activities and tasks of PC providers, process quality of PC and accessibility of PC at the organisational level. The patient questionnaire is aimed at gaining insight into the evaluation of services from the point of view of patients/clients by measuring the perceived quality of care, perceived access to care and actual cost barriers to PC.
To come to well-founded questionnaires several steps were taken. First, a framework, including important aspects regarding the process and outcomes of care, was defined. For the GP questionnaire the framework of Kringos et al (2010) was used. For the patient questionnaire a framework was used based upon the Consumer Quality Index of GP care [59
]. Secondly, a search in scientific databases and on the Internet on existing questionnaires on the topics included in the frameworks was performed. Thirdly, the questions from the selected questionnaires were grouped according to the topics of the identified frameworks. Fourthly, gaps were identified by experts on the different research topics (such as equity and costs). It was evaluated for which topics appropriate questions were lacking. For these topics, new questions were formulated. Based upon the findings from the third and fourth step, three questionnaires were developed: one for GPs and two for patients, distinguishing patients' experiences and patients' values. The questionnaires were then piloted in three countries (Belgium, Slovenia and the Netherlands). Based on the findings adjustments were made and consensus on the final questionnaires was reached based on experts' opinions (see section 'pilot'). Specifications on the outcomes of the search strategy, questionnaire development and questionnaires will be published separately.
The survey among GPs includes self-reported involvement in curative and preventive tasks and questions on the type and organisation of the practice, integrated provision of services and aspects of workload and use of time. The patients experience survey contains questions about the patients' backgrounds, distance to the PC practice, choice of doctor, copayments for services, time for the patients, availability of health education, experiences with services of the practice or centre, experiences with their own doctor and aspects of care coordination. The patients value questionnaire contains questions about the patients' backgrounds and their values regarding GP care.
As the survey will be held among GPs and patients in 31 European and 3 other countries, the questionnaires will be translated from the English master version into the national languages. Also, to reach the largest groups of ethnic minorities within the countries, some extra versions in languages such as Arabic will be made available for patients. An independent 'forth and back translation' procedure will be used.
A pilot was held to test the process of completing the survey in the GP practice and to test the relevance and comprehensibility of both questionnaires. Questions regarding the process that were addressed are e.g.: How long does it take to fill in the questionnaires? Are GPs and patients easily willing to participate? By testing comprehensibility of the questionnaires we tried to answer questions like: Are the instructions on the questionnaire understood by all respondents? For the closed questions, are all reasonable alternatives included for the respondents? The pilot was held in three countries in Europe (Belgium, the Netherlands and Slovenia) among a small sample of GPs and patients. GPs and patients were surveyed in the GP practice setting.
All questionnaire data will be centrally processed in the Netherlands. The questionnaires will have a uniform design and a closed answering format to allow optical reading technology for data entry. The data will be analysed initially to construct new variables to be used in the analyses for the hypotheses. This guarantees unity in the way important variables, such as process quality of PC and patient evaluations of PC quality, are defined.
The data collected in this study will be integrated by using statistical models for hierarchically structured data in multilevel models [60
]. Multilevel models enable to partition the variation in (e.g.) patient perceived quality of care into three parts:
- a part related to the individual patients (related to personal background, health status etc.)
- a part related to the PC practices they visit (e.g. related to the range of services that these practices provide and the process quality of the practices)
- a part related to the health care system of the countries (e.g. related to structural aspects of PC and the strength of PC).
Apart from studying general patterns and trends we will identify good practices which in a statistical sense are the outliers in the statistical analysis. Analyses will be made on the combinations of traits and circumstances that explain their position.
Data analysis techniques will include: data reduction by scale construction using factor analysis, reliability analysis and ecometrics; single level regression and correlation to analyse the relations at country level between PC structure and quality, cost and equity variables; multilevel analysis to relate country, practice and patient levels.
The use of multilevel statistical analysis (MLA) is essential in this study, in particular where survey data (from GPs and patients) are integrated with aggregate data at health care system level. The MLA approach has specifically been developed for these situations, where units on which variables are measured are nested within larger (higher level) units; such as patients within GP practices, or GPs within a country's health care system. MLA allows analysing variables at the country level and at the GP (practice) level at the same time.
Personal data confidentiality
For the survey among GPs random samples will be drawn of GPs from available lists or registers. Patients will be approached in the practices. Procedures which apply for this use of registers in each country and for survey research among patients will be identified and carefully observed. When necessary, we will apply for ethical approval in the participating countries. The survey is anonymous; respondents (physicians and patients) do not need to fill in their name. To be able to link the data of GPs to the countries and their patients, we will use identification numbers. In no publication results will be reported that can be related in any way to individuals or locations.
The research team consists of experienced researchers, with backgrounds in health services research, economic analysis, patient evaluation surveys, survey design, statistical modelling and PC research. The team has extensive experience in research on international comparisons of health care systems. Each of the involved research institutes are leading on one of the main study topics (equity, costs or efficiency, process quality of services, perceived quality of services) which fits within their expertise.