Skip navigation

"When in doubt, do it" or "When in doubt, stop it"

review of "Three battles to watch in the 1990s" by David Eddy

Across the developed world the problem for governments is to deliver health care from finite resources. Beveridge believed that provision of free health care would, after a catch up phase when the health care of the poor was lifted to an adequate standard, result in decreased costs. By the mid 60's Buchanan was arguing that this could never have been the case.

Governments could provide free funeral services without undue losses in efficiency because each person dies only once. A zero price does not produce a larger demand for funerals than a high price. For many services this zero price elasticity condition does not apply. make the price of plastic surgery low, or offer it free, and demand will rise - demand will be higher at zero or marginal cost prices.

Growing health care costs

The escalation in the cost of health care over the last twenty years has led to our current debates. Eddy's analysis suggests that reform of financing alone will not solve the cost problem, and, crucially, that "no attempt to control the excess increase in health care costs will be successful over the long term unless it addresses the decisions physicians make about treatments".

Eddy breaks down the growth in health care expenditure into two factors:-
  • Those which health care reform cannot affect are general price inflation (42% of the increase) and growth and ageing of the population (9%).
  • Factors which could be affected are medical price inflation in excess of general inflation (17%) and increases in volume and intensity of service above and beyond anything explicable by demographics (32%).
It is this volume and intensity figure at which Bandolier's bullets are aimed. The battle is about "what practitioners do and how they do it".

Evidence

Eddy tackles the evidence issue under three headings:
  • What evidence is needed to be "sufficient"?
  • Who has the burden of proof?
  • Old versus new treatments.

Sufficiency

Few procedures have proven health gain or loss, and the great bulk of traeatments are best described as looking promising or having uncertain effects. For these Eddy asks "if enough experts agree that a treatment is effective, is it necessary to have any empirical evidence at all?" Eddy also poses the questions as to whether studies should look at actual outcomes, such as survival or quality of life, rather than proxy intermediate outcomes such as disappearance of tumour on X-ray, and whether a huge pile of retrospective uncontrolled studies will 'do' in lieu of a small number of controlled studies.

Burden of Proof

The issue of burden of proof is well illustrated by our current dilemma about mammography for women under 50. For a common and serious disease should this screening be provided even if the evidence for effectiveness is less than overwhelming? Eddy encapsulates the problem neatly with two phrases - "when in doubt, do it" or "when in doubt, stop it".

These questions are easier to address with new treatments than with old. As Eddy says, if we applied modern standards of evidence to all our existing interventions, "medical practice would be in chaos". The Dutch are striving to ensure that any new intervention is assessed properly before widespread introduction, and the Health Technology Assessment initiative of the NHSME's R&D programme has a similar brief.



Eddy's views on evidence are very pertinent. If Bandolier is to disseminate bullets of evidence then some way of rating these bullets should be explicit. How do we rate a bullet based on 30 years of experience but with no RCT, compared with a bullet based on a small number of RCTs? This is not an adequate stance for the bulk of our diagnostic tests, devices, procedures or services, and the importance of a bullet must also take into account
  • whether the problem is common
  • whether the problem is serious
  • the cost of making the intervention and
  • the cost of not making the intervention.
We hope to produce constructive thoughts on rating bullets later this year. In the meantime, we rate the Eddy paper highly.
Henry McQuay, MD
Regional Research and Development Director

References:

  1. Eddy DM. Three battles to watch in the 1990s. Journal of the American Medical Association 1993 270: 520-526.
  2. Buchanan JM. The inconsistencies of the National Health Service. Institute of Economic Affairs 1965 Occasional paper 7.
next story in this issue