Attracting and retaining GPs: a stakeholder survey of priorities

Setting

Belgium has compulsory health insurance covering the entire population and a very broad benefits package covering 77% of all healthcare expenditures. It has a high medical density (four physicians per 1000 inhabitants) and high per capital healthcare expenditure (10.6% of the GDP). Most GPs work in solo practice with a predominantly, fee-for-service payment and no auxiliary workers. Compared with the UK, the Netherlands and countries in Scandinavia, Belgium has no gatekeeping function, more GPs per inhabitant, less-developed IT systems, and strongly competing professional bodies. The health system resembles many parts of the French system in these respects.

The sample

In 2008 a stakeholder survey was carried out in Belgium. This was a tool for generating knowledge from actors and understanding their intentions, interests, and interrelations.20 According to the literature, stakeholders are classified as policymakers, professional groups, academics, media, and GP leaders.21

A panel of researchers and leaders in general practice drew up an initial list of 155 names of the most influential persons in each group, such as senior administration officers, ministerial advisers, union leaders, faculty deans, senior sickness-fund administrators, editors, journalists, and students' representatives — that is, all those with major responsibilities in, or influence on, the Belgian healthcare system or the training of GPs. The extent of each individual's influence was rated by seven external leaders from different professional and public health organisations. Those who received at least three votes out of seven were retained for the survey.

In order not to leave out an important group, several validation meetings took place among the researchers. The sample was complemented by snowball sampling: each interviewee was asked to name up to three important persons in each stakeholder group. The most-cited persons were then also contacted.

In total, 116 stakeholders were contacted and 102 (88%) agreed to participate. A breakdown of the sample is given in Table 1.

Design and measurements

For the questionnaire22 the study used the multicriteria analysis approach (MCA), which is designed to assess decision making between different policies, based on several criteria. MCA has recently been used in the health sector for assessing obesity prevention23,24 and health technologies in primary care.25 It involves four steps:

• identifying various policy options to solve a problem;

• identifying criteria to appraise those policies;

• allocating a score to each policy option on each criterion; and

• weighting each criterion to reflect its relative contribution to overall decision making.

In the research, an initial list of possible policies was collected from an extended literature review, as well as from recent policy initiatives in Europe.1,26 The list was discussed with the research team, adapted, then piloted with experts in the medical field. No policy was excluded because of being supposedly controversial.

The study ended up with 23 policies, classified into five groups: training, financing, work–life balance, governance, and practice organisation (Table 2). In addition, each policy was classified as either conservative or innovative by the panel of researchers. The policy was considered to be conservative if it was related to already-existing solutions (such as increasing fees) and innovative if it required either a change in existing regulations (such as targeted payment) or the setting up of a new executive body. A Delphi process was carried out, with the panel of researchers classifying each policy into these two categories.

The criteria were defined according to the literature on healthcare prioritisation and were then discussed and adapted by the panel. Four criteria were retained:

• effectiveness in improving the attraction and retention of GPs;

• cost to society;

• acceptance by other health professionals; and

• accessibility of care and patients' freedom of choice.27,28

Data collection

Stakeholders were interviewed in a face-to-face computer-assisted interview, conducted by a trained interviewer. During the interview the stakeholders were asked to score each policy according to each criterion on a seven-point Likert scale ranging from very negative (-3) to very positive (+3). Each criterion score was weighted to reflect its relative importance. For each interviewee, the weights were obtained during the interview, using the swing-weight method. Weights were then normalised to compute an overall score for each policy.23 For each interviewee, the overall score of policy P equals:

Each stakeholder could also make open-ended suggestions or comments on the policies; this helped to clarify their choices. After the interview, the interviewer rated the interview process. In total, 87% (n = 89) of the stakeholders had a good level of interest in the subject and 83% (n = 85) understood the questions well. Stakeholders also were requested to choose their first, second, and third most-favoured policies for each policy group.

Analysis

The difficulty of reforming primary care may be due to the complexity of the issue itself, the political context that makes new policies more or less likely to be adopted, and the political divide between powerful groups.18 Accordingly, the study analysed the policy score in three steps.

First, the study compared the scores according to two policy characteristics: policy group of the interviewee (training, finance, work–life balance, practice organisation, and governance) and whether the policy was innovative or conservative in the Belgian context. As each interviewee rated 23 policies, the study used a regression model including a random component to take account of clustering at the stakeholder level. Second, to understand why some policies received a higher score than others, the study broke down the policy score by the contribution of each criterion (standardised weight_i * score criterion i).

Finally, the study compared each policy scores separately between stakeholder groups.

Summary

Overall, the stakeholders were keen on moving the GPs towards group practice, improving their role in the multidisciplinary team, helping them to offload administrative tasks, having compulsory clerkship and sharing a common infrastructure. They were also keen on improving the early and integrated exposure of all medical students to general practice. However, three elements show that reforming general practice is problematic:

• Two groups of policies (financing and work–life balance) were not strongly supported by the stakeholders, although these are currently supported by the research and among the main claims of the GPs' unions;

• Some policies failed to win acceptance, not because of their lack of effectiveness, but because of their poor performance regarding cost to society, acceptance, and accessibility to care.

• Innovative policies scored poorly when compared with conservative polices.

In several European countries, GPs have long been calling for improved financial conditions as well as an improved work–life balance, and research evidence strongly supports changing the payment mechanisms applied to general practice.29 However, these two groups of policies received low scores. Stakeholders were not keen on them and were more willing to use training and organisational policies. For some policies, therefore, there is a mismatch between what GPs are looking for and the support of stakeholders. According to the study's findings, this mismatch appears to occur because stakeholders take a multidimensional perspective on policy making. In particular, their reluctance to use financing policies as leverage to improve GPs' attraction and retention is due to their concern to not alienate other actors, such as other health professionals. As priority-setting research has made clear, criteria such as acceptability and equity, as well as other institutional constraints, are also taken into account in decision making.30

Innovative policies received lower scores than their conservative counterparts. This is consistent with the theory of path dependency, according to which, choices that have been made in the past systematically constrain the choices that are available in the future — a mechanism clearly at work in cost-containment policies.31 Path dependency is partly due to the current institutional arrangements for decision making in health care in Belgium; decisions are made in bodies organised along professional lines (medical doctors, hospitals, home care, nurses, and physiotherapists) which impedes the emergence of a strong primary care political community in which GPs, nurses, and other primary care providers can draw up and advocate a strong primary care agenda.32

Finally, both groups of GPs (the leaders and those from the medical unions) were as divided as the other groups on all policies, including innovative ones, such as the creation of a nurse assistant master's degree, gatekeeping, and clinical delegation. These professional groups could not come up with a clear-cut programme on which they agreed. This lack of mobilisation also explains the difficulty in coming up with bold reforms of primary health care in pro-coordination polices or cost-containment policies.33 The emergence of a strong policy network related to primary care is less likely if the profession is divided. The lack of a strong corporate interest weakens mobilisation and means there is no strong political partner to negotiate and implement reforms; as such, a weak GP profession also weakens public health stewardship.34

Strengths and limitations

To the authors' knowledge, this study is the first stakeholder survey to use a multidimensional evaluation of a set of policies aimed at attracting and retaining practitioners into the profession, and it helps to distinguish three obstacles to making general practice more attractive. However, two possible risks of this study's approach are that the criteria do not correspond to the responders' own values or that the responders prefer a policy for which they have a preference independent of the criteria. In spite of this, such biases were not observed; indeed, only a minority of the responders (6%) gave a zero weighting to any criterion (acceptance by other health professionals). Moreover, average inter-criteria correlation was rather low (0.34) and only four stakeholders displayed a clear tendency to rate all criteria in the same way. It was also found that in most cases (96%), the policies with the highest score were also likely to be selected as first-, second-, or third-best policy, suggesting a strong agreement between scoring and choosing.

The study was carried out in a specific continental European setting and, as such, the priorities may be more relevant to a conservative social insurance health system and to similar primary care systems, such as those of France and Germany than systems with a national health service.35 Nevertheless, the study helps to understand why reforming primary care has been a difficult task until now in many countries.

Comparison with existing literature

In Europe, studies have shown that reforms of the primary care sector have occurred for one of the following three reasons:

• to increase the power of primary care as a purchaser of services, coordinator, and gatekeeper;

• to broaden the service portfolio of primary care; and

• to provide supportive conditions in order to promote a stronger role for primary care.35

This work suggests that stakeholders support the third reason — the least-radical reform — while the degree of support for the second is unclear: on one hand the multidisciplinary team received a high score but, on the other, the policies related to the delegation of clinical tasks or nurse-assistant practitioners did not score well. This ambiguity is not unique and parallels the difficulties countries like France and Germany have faced in enhancing the role of primary care as coordinator of care.36

Implications for practice and research

Although the ‘disappearing GP' may jeopardise the healthcare system, this stakeholders’ survey showed that the policies favoured by the stakeholders do not indicate an emerging consensus on radical change of the present situation. Changing the profession internally is on the agenda. As practices will serve more patients in the future, among the actions that can be performed within the profession are practice rearrangements like working in larger teams, adjusting GPs' task profiles, and introducing transfer of tasks to the most appropriate echelons of care.

Countries with strong general practice systems have already introduced large numbers of auxiliary personnel in primary health care. In the laissez-faire approach observed in the key stakeholders of health care, a stronger political primary care community, with prominent GP leaders showing the efficiency of general practice, would help to promote more radical alternatives.