Intended for healthcare professionals


The research needs of primary care

BMJ 2000; 321 doi: (Published 01 July 2000) Cite this as: BMJ 2000;321:2

Trials must be relevant to patients

  1. Paul Thomas, director, West London Research Network (WeLReN) (p.r.thomas{at}
  1. Department of Primary Health Care and General Practice, Imperial College School of Medicine, St Mary's Campus, London W2 1PG

    General practice p 24

    A health service that is led by primary care must be able to inquire into the practice of primary care; let research in primary care blossom. In England, at least, this logic now seems to be backed by political will. The Mant report (from a subcommittee of the health service's Central Research and Development Committee in 1997) states that there is an urgent need for both research and researchers in primary care.1 The full committee later challenged universities to support this recommendation.2 It is now government policy to develop research capacity through primary care research networks.3

    Two papers, one in this week's BMJ, the other recently published in the journal, point the way towards conducting randomised controlled trials that are relevant to primary care. Both papers argue that it is difficult for researchers to gather a sample that is representative of the whole population. Wilson et al (p 24) point out that clinical trials must be conducted in primary care rather than secondary care or else the sample will include only those who have reached secondary care.4 Rogers et al argue that trials must be relevant to a wide range of practices if a variety of practices are to be encouraged to participate in research.5

    I experienced the difficulty of recruiting a representative sample of practices recently when coordinating recruitment for a large randomised controlled trial on the management of hypertension in primary care in west London. Altogether, 106 general practices—one fifth of the total practice pool—sent personalised letters to eligible patients inviting them to take part. However, as in Wilson et al's study, only one tenth of the potential sample was recruited. Nine out of 10 patients in west London were excluded, or excluded themselves, from this study. The reasons relate to understandable everyday factors. Firstly, patients and doctors often have medical reasons for preferring a particular way of managing a condition and do not want it changed—for example, a β blocker may help someone's panic as well as their hypertension. Secondly, people may be motivated to use some treatments and not others—they may trust one drug over another because a relative derived some benefit from it. Thirdly, there may be difficulties in following a research protocol: people forget to take tablets, don't want to attend for follow up, want to choose which arm of the study they are enrolled in, or their hypertension is compounded by heart failure, asthma, or fear.

    Comprehensive representation of the population matters only if we decide that randomised controlled trials are the last word in research. Randomised controlled trials are designed to address one or two issues and make sense of them in isolation. But these trials represent only one way of looking at things. General practitioners often have to manage multiple, interacting factors. 6 7 Occasionally patients do present with a simple symptom that requires a simple diagnosis and cure, but more often they have a multitude of interwoven issues that need unravelling.8

    The challenge is to develop “real world” research that can capture something of these complexities. Wilson et al suggest developing primary care research networks to increase the number of service practitioners (such as general practitioners, nurses, and community practitioners) participating in research and thereby increasing the diversity of the population that can be studied. Rogers et al's team show a way of increasing the depth of findings by using quantitative methods and qualitative methods together and then using a group of key informants, carefully chosen to understand different perspectives, to interpret the data.

    Researchers will have to develop ongoing partnerships with those involved in developing services and continuing professional development programmes to increase reflective, inquiring practice across the board. The organisation of primary care in England into clusters of practices (primary care groups) offers the potential to do this. New theories are emerging that may show how different professional groups in a primary care group can work together most efficiently. Different professional efforts can be seen as connected parts of a whole system, the make up of which depends on the specific issue but the location of which can be in a primary care group area. 9 10 If the participants of a whole system operate as a learning organisation, lateral thinking can be encouraged and professionals can find ways to make different initiatives relevant to each other. For example, those involved in developing research projects and those involved in quality assurance programmes or continuing professional development may have shared interests around quality practice. Shared planning and discussions can result in cross pollination of ideas and skills and in mutual support. 11 12 Alternatively, locally derived research projects could be woven into initiatives led by academics, thereby providing support for teaching and discovery of other insights into the topic being explored. More ordinary primary care professionals need to get involved in research. If these ordinary professionals are not involved, other professionals who do not have to live with the inherent complexities of primary care will come in and describe our experiences for us. Primary care clearly needs more research and research needs more primary care.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    View Abstract