What sort of evidence do we need in primary care?

In this week's BMJ (p 635), Mant and colleagues raise again the question of whether large scale randomised controlled trials provide evidence relevant to primary care.¹ In a cross sectional study they question whether the UK national clinical guidelines for stroke are applicable to primary care patients. These guidelines, largely based on the PROGRESS trial,¹ recommend a target blood pressure of 140/85 mm Hg, with further lowering beyond this target desirable through use of a thiazide diuretic and an angiotensin converting enzyme inhibitor.²

Mant and colleagues critiqued the applicability of these guidelines to primary care patients by comparing the characteristics of patients in English general practice who had confirmed stroke with participants in the PROGRESS trial. They concluded that these populations were not sufficiently similar to warrant widespread use of the guidelines. Patients with stroke in English primary care were generally older, equally comprised men and women, and had had their cerebral events less recently than the trial participants. They may not necessarily benefit from the relatively aggressive treatment recommended in the national guidelines. Given that most chronic stroke patients are managed by general practitioners, Mant and colleagues are calling for more appropriate research to provide evidence applicable to primary care.²

Primary care has specific needs for research. This setting is contextually complex, with patients who present with a broad range of signs and symptoms. Therefore, practitioners need research evidence for the whole range of diagnostic, prognostic, interventional, and phenomenological questions that emanate from this contextual complexity.³ Well documented gaps in the evidence limit the quality of patients' care. Some evidence is simply lacking—for example, on the aetiology of upper respiratory tract infections, and on the reasons for variation in health seeking behaviours between people who have the same symptoms. Even when there is relevant evidence, it is not always implemented.⁴

For evidence to be implemented in primary care it must be accessible as well as relevant.⁵ The sheer volume of published studies documenting medical and clinical research means it is impossible for busy clinicians to read, let alone digest, sufficient material to keep up to date. Clinical guidelines can provide clinicians with concise, evidence based recommendations to enable the delivery of high quality and well researched care for their patients.⁶ However, the guidelines must be regularly updated,⁵ and they should evaluate the methodological rigour of included studies and be clear about both the application and the limitations of the recommendations.⁷

With good clinical guidelines, clinicians can confidently use research evidence as a basis for their individualised decision making with patients, while also considering patients' personal histories and preferences. Mant and colleagues indicate that these criteria are not met by the UK clinical guidelines for stroke, thus under-mining the applicability of the recommendations.²

The PROGRESS trial not only had a different patient profile than that seen in primary care but yielded results that are open to interpretation. In particular, the cause of the substantial reduction in stroke risk seen in the trial (relative risk reduction 43%, 95% confidence interval 30 to 54) is unclear: is it due to the use of a thiazide diuretic and angiotensin coverting enzyme inhibitor in combination or alone, or simply due to the lowered blood pressure?⁸ Really useful guidelines would help clinicians to interpret these findings, rather than just uncritically basing clinical recommendations on them.

Evidence from randomised controlled trials is only a portion of the real knowledge that is needed in primary care. Such trials provide vital information about the effectiveness of interventions in selected populations of patients. However, general practitioners require evidence about how an individual patient will respond to an intervention. Active involvement of patients in decisions about their health care can improve health outcomes. For instance, patients with breast cancer have less anxiety and depression if treated by doctors who involve them in the decision making, and patients with diabetes achieve better blood sugar control if involved in discussions about their management.⁹

Mant and colleagues call for further research in the primary care population to determine the efficacy and appropriateness for individual patients of intensive lowering of blood pressure.² Their call is well justified and should be heeded.