'It may be efficient to let marginal patients die from avoidable infections'

Over the past decade there has been increasing concern about the safety of medical practice. Gross failures to use routine data for measuring and managing clinical practice have been well reported in the worst cases such as GP Harold Shipman.

Added to these disasters all preventable if performance had been managed with the use of data such as hospital episode statistics there is panic over MRSA and Clostridium difficile, which can both be better controlled by targeted interventions.

As a result of these failures there have been large investments in improving safety. Various reports from the chief medical officer for England and the Health Foundation charity have supported the US Institute for Health Improvement's evangelical work in NHS hospitals.

The common characteristics of these interventions are the failure to cost them and the fragility of the cost-effectiveness evidence base. They are prime examples of what Princeton University economist Uwe Reinhardt calls faith-based policies.

The intention of medical and other policy-makers cannot be faulted. Unfortunately, economic sanity has failed to be incorporated into their advocacy and decision-making. In particular there has been a failure to recognise that there is an efficient level of damage to patients and this may not be zero.

Eradicating all errors in clinical practice and infection control would be enormously expensive. Targeting resources to give the greatest health improvement per pound spent should be informed by evidence of the relative cost-effectiveness of competing safety interventions. But the evidence base is poor and Whitehall continues to demand more safety investment. We may be throwing good money after bad.

Decision-makers fail in their duty to recognise that all such decisions have an opportunity cost: spending more on safety deprives patients of care from which they could benefit. Why is such unethical and inefficient decision-making condoned by ministers, civil servants and managers? The latter two groups condone inefficiency in the use of NHS billions to appease their political masters and the media.

We all will die and it may be efficient to let marginal patients die from avoidable infections and poor clinical practice if it is not cost effective to invest in better control. Trying to reduce such events to zero may use resources that would give more benefit if NHS patients are treated in environments which cannot be risk free but should always be as safe as is efficient.

The challenges are obvious. For government it is necessary to indicate the optimal level of hospital infection they wish to reach, and to cite or obtain evidence to inform the targeting of safety resources cost-effectively. They should no longer use the 'blunderbuss' approach as it wastes resources and deprives patients of care. It is moral civil servants' role to educate their more hysterical political masters about the reality of such difficult choices.

For the IHI the challenge is again to show high-quality evidence of cost effectiveness. Mere advocacy should not be a basis for investment by the NHS. Given the paucity of such evidence arising from their US work, it is to be hoped that UK investments in IHI will be more rigorously evaluated.

There are two lessons for local NHS management. First, we have the data to monitor clinical performance. So do it. We have collected and ignored hospital episode statistics for 20 years. The second is to respond to Whitehall policy nonsense and use simple economic arguments involving the concept of opportunity cost to challenge government choices. The agencies of NHS management and regulation such as the NHS Confederation, Monitor and the Healthcare Commission could lead such efforts. Failure to do so may be evidence of their lack of cost effectiveness and the need for their reform.