The themes that emerged on reviewing the 35 references, seven books, and nine policy documents are listed in Table 2 and will be explained below. There was considerable variation in the language found in the literature used to describe GPs, registrars, and educators. The list of definitions used in this review is listed in Table 3.

There were six medical educators, six GPs, and seven GP registrars who published the 18 surveys and cases publications. The main emphasis of non-practice based GP educators was on resources and educational influences for teaching research to GP registrars. In contrast, GP educators and their registrars emphasised their attitudes to research, or lack of time to carry out research within clinical practice.

In the early 1990s, the literature was largely dominated by the views of medical educators and GPs about the process of teaching research. It was only later that GP registrars had their views published. The first publication by an identified GP registrar occurred in 2004 17 and thereafter much of the GP registrar literature was published in the form of case studies. There was only one article describing the research environment for a GP registrar in an Australian general practice.

Five educational barriers for teaching of research were identified:

a) GPs struggled to identify research questions that were different from hospital practice and relevant to general practice 42.

b) GPs struggled with research methods. Community derived research questions needed complex research methods to answer them. These methods were usually a combination of qualitative and quantitative techniques rather than the more narrow focus of quantitative methods found in hospital specialist practice 25.

c) Overly ambitious curriculum guidelines produced by various colleges that resembled a PhD curriculum 536.

d) Lack of recognition that a single uniform approach to research training did not exist 31.

e) Among those residents who had been in a research training program, few reported that their training prepared them to do research 29.

There had been an increasing volume of published research from Australian general practice over the previous 25 years 1526. Table 4 lists the range of topics and suggests a number of gaps. For example, there were no research topics from rural general practice or research on health inequalities and the determinants of health in Australia. Less than half the published research focused on clinical topics 14. There was less research published from Australian primary care if compared to the production from internal medicine, surgery, or public health 12. This gap in publications was also found in United Kingdom 56 and The Netherlands 41 general practice.

The editor of The Lancet argued that the presence of these gaps indicated that "Primary care research is a lost cause" 57. This line has been vigorously denied by arguing that much of the clinical and preventive care that occurred in general practice must be underpinned by research produced from general practice 45. Furthermore, the long list of common diagnoses and their management found in general practice could not be informed by hospital-based research alone 4245.

It is a truism to state that the more common the condition in general practice, the less it is studied 58. This truism could be applied to the GP registrar who was commonly found working throughout Australian general practice, but the largest gap in publications was seen from GP registrars. Overall, there was very little literature describing the kind of topics in which GP registrars had done research. Grzydowski described an evaluation of research projects done by 271 Canadian GPs who had been residents involved in research between 1990 and 1999. The GP residents' contribution to research consisted largely of their collecting data for projects. 40% of their projects were cross-sectional studies. Only 7% of projects were published and half the remaining residents would have liked to publish their study. There had been only two Australian case studies describing what research GP registrars did. Chien described how a GP registrar and supervisor within one general practice could do research by using an action technique for research 17. Gartlan described her experience of being a novice researcher 19. Montgomery published five case studies of how GP registrars produced research proposals at a research workshop 22.

There were two cultural perspectives found in the review on the value of research for the GP registrar. First, although it was widely recognised that research was part of primary care 41, many people perceived general practice as being a non-academic discipline that should concentrate mainly on the provision of health care 46. This cultural perspective placed research in the background of what the GP registrar should learn when being taught general practice. For example, few providers of GP registrar training in Australia had included research as a compulsory project within their training [Tasmania was one exception] 159. Research had also played a minor role in Academic Family Practice in the USA 36.

A consequence of the low value placed on research was that the GP registrar became a late author of research in the history of general practice research. The first publication by a GP registrar in this review was in 2004 17. The first Australian GP training programs started over two decades before that time.

The second perspective found in the review placed research into the foreground of general practice by arguing the benefits of research. These benefits included an increase in the status of general practice in academia, improved professional standards, and improved standardisation of terminology and diagnostic and therapeutic procedures 41. GPs could obtain personal benefit from research in that it improved morale, reduced dissatisfaction to working in general practice and improved intellectual stimulation 41. The benefits could also be extended to the GP registrar. In addition, the GP registrar could gain educational benefits from performing research tasks including asking questions, critical appraisal and creating new knowledge. Such benefits were at the centre of the educational cycle of learning 11.

Svab argued that a culture of research would exist if clinician-researchers were paid to do general practice and conduct research at the same time, and also worked with a critical mass of interdisciplinary researchers 46. Svab suggested that a cultural change was required to develop "practices of excellence' throughout a country where clinician-researchers worked within supportive organisational frameworks. A profound change in cultural mind-set was required. GPs needed to change from not only being the traditional devoted and caring doctor of individual patients, but also a GP who became research oriented. If a GP registrar were to stumble upon "a practice of excellence", he or she would find a research culture where all individuals within the practice had a way of perceiving and understanding the world that included research.

This review found a few successful stories of learning research in general practice: GPs were more likely to undertake research if they had prior research experience in their undergraduate, medical school, or residency period 33. Individual GPs told success stories of how they started research as a young GP and went on to built a life of research 14. There is an account of a registrar and a GP supervisor working together in a practice environment where a research culture is clearly evident 19.

Stories about a research culture have as yet to be articulated in general practice. The literature suggests such stories would contain five elements. The first would include the active involvement of research mentors 1621323335. Such mentors would be found through colleges, university departments or research networks. They would be medical educators, supervisors, faculty staff, or GPs, actively involved in research. The second element would include a supportive teaching infrastructure for research 21293133. This structure would provide administrative resources, organise dedicated time and funding. The third element would require GP registrars to create and publish small research projects 2136. The final element would include a description of people who have an inherent enjoyment of research 33.

The review identified six enabling factors that not only increased research participation by registrars but also changed the negative cultural perspective. First, there should be an increase in the number of GP registrars who undertake research. The more registrars completing a project during training, the more likely it was that they would do research in their career 35. Second, more GP registrars should observe clinician-researchers doing their work in general practice 34. Third, many more GP registrars should become research fellows 34. Fourth, a GP Training Program should be linked to a collaborative research network in order to develop a critical mass of researchers 3446. Fifth, a GP Training Program should provide all local GP educators and supervisors access to information resources and opportunities to participate in relevant research activities 16. Finally, GP educators should start talking about general practice research early in a medical student's career pathway. Medical students should asked questions relevant to general practice when they engage in their learning about general practice 3142.