Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Easy Isn't
- Ned Hoke (17 November 2003)
-
RCTs are not everything
- Kath H Checkland (17 November 2003)
-
A plea for research enhanced medicine
- Rod A Lawson (18 November 2003)
-
Dont confuse validity with levels of evidence
- Ben d Ewald, 2300 Australia (1 December 2003)
|
|
|||
|
Ned Hoke, ecological medicine USA
Send response to journal:
|
Given the problematic consequences of x-ray exposure and vain positive assertion physicians use when they depend too substantially on test results this report is a voice of necessary corrective in this subject. Taking the time to effectively evaluate the subjective signals and open broadly the mind to various themes of causality and simple mystery is something both physicians and patients need to embrace to respond to this situation. The easy answers are few and far between when really confronting the 'facts of life' in the human condition. Medical short-hands make for convenience and cost-effectiveness in service but degrade the human potential within the medical encounter and often leave clinical results entirely to the force of the wind. Competing interests: None declared |
|||
|
|
|||
|
Kath H Checkland, GP and Research Training Fellow NPCRDC, University of Manchester, M13 9PL
Send response to journal:
|
Whilst agreeing with the conclusion of these authors that more research is needed into the use of diagnostic tests, I fear that their paper betrays a common failure to understand the limitations of the randomised controlled trial. RCTs tell us about averages - if a large number of people are given aspirin following an MI, on average, more will benefit than suffer as a result. However, this "average" overall benefit conceals within it individuals who will suffer and individuals who will derive no benefit at all as well as individuals who do better as a result of treatment. Understanding the specificity of a test such as an exercise stress test is a legitimate goal, but I would question the application of similar methodologies to such things as trials of medication and the use of peak flow charts. Such things are used by practising doctors not to further understanding of the condition in question, but to try and apply existing evidence to the individual in front of them. What would an RCT of such things tell us that was useful? That some people will respond to the placebo effect? That giving a trial of inhaled steroids to patients with chronic obstructive pulmonary disease does not, on average, predict therapeutic benefit? Even if this latter point is true, does it matter, given that there will be individuals within this group for whom it will be a useful guide to their response? My plea is simply this: reserve the undoubtedly useful and expensive technique of RCTs for those areas where it is appropriate. Stop saying, with a reflex nod to Sackett et al that "the RCT represents the highest level of evidence" in all cases, and think outside the box when it comes to working out how to apply evidence to individuals. In the instance discussed above, rather than organising an RCT of trials of inhaled steroids, how about looking at what factors underly the placebo response, the meanings that patients with COPD ascribe to their inhalers and ways of persuading people to stop using them if there is no good evidence that they are benefiting. Such investigations would be qualitative and would fail the "level of evidence 1a or 1b" test. I would submit, however, that they would be of more practical use. Competing interests: None declared |
|||
|
|
|||
|
Rod A Lawson, Consultant in Respiratory Medicine Royal Hallamshire Hospital, Sheffield S10 2JF
Send response to journal:
|
Borrill et al's article is certainly thought provoking, but it encourages me once again to add my weight to those who believe the term 'evidence based medicine' should be replaced by 'research enhanced medicine'. If we practiced medicine that was strictly evidence based, we would find large gaps in our practice where we were paralysed, and unable to act, as the article demonstrates. The article clarifies the point that research on diagnostic tests may help us use them with greater precision and understanding, and that this may challenge preconcieved notions. At the same time, my PhD supervisor Chris Haslett's dictum that the most powerful statistical test is the 'BO' (or 'bloody obvious') test is also illustrated. Whilst we may argue about the precise sensitivity of venography for DVT, to award this test a poor level of evidence doesn't so much challenge the test as the very concept of the DVT. Again, serial chest radiology to help exclude a lung cancer doesn't need a clinical trial. If there's a lump getting bigger, it may well be cancer. If it remains static over a period of time, it's considerably less likely to be so; cancer tends to grow. Checkland in her response has already made the point that we need to be aware of the potential pitfalls of generalising from the average of a clinical trial to the specifics of an individual patient. In the end we can refine our tests and give a probability that a patient has cancer but patients don't get 73% cancer; they have it or they don't. No matter how precise our evaluation of diagnostic methods are, we are in the end thrown back to the art of medicine, the experience of the clinician, and the individual interaction with the patient. In the end, the foundation of our practice is art. We can and should enhance and supplement our practice with research where possible, but it is a fundamental mistake to suggest that medicine is or ever can be completely evidence based, or that the science is valid and the art is not. Competing interests: None declared |
|||
|
|
|||
|
Ben d Ewald, PhD student and GP Newcastle NSW, 2300 Australia
Send response to journal:
|
The paper by Borrill points out the lack of research into diagnostic tests, but seems to make the mistake of confusing levels of evidence with strength of evidence. A test may have high sensitivity and specificity, well demonstrated in a single good study and be categorised 3b (good reference standard but non- consecutive cases) while another test may have abysmal validity and be well researched so categorised as top level evidence.
As a doctor or as a patient I would rather have the valid test than the highly researched one. Of course it would be best if the valid test had good evidence to back it up, but given the inherent problems of diagnostic research the evidence is slow to accumulate. Dr Ben Ewald, General Practitioner Competing interests: None declared |
|||