I suppose the process of acceptance will pass through the usual four stages:

1. This is worthless nonsense,

2. This is an interesting, but perverse, point of view,

3. This is true, but quite unimportant,

4. I always said so.

JBS Haldane Journal of Genetics 1963; 58(3), 464.

Few would disagree that it is good idea to provide clinicians with incentives to improve performance. Others propose sanctions for clinicians who do not meet clinical standards. Our proposal sees both propositions as the matching halves of a feedback control system, with “patient harm” the variable to be controlled. In these three thoughtful commentaries, we see the beginning of a necessary dialogue to reshape safety research and policy. The debate comes from different traditions, including health services, economics, governance and systems control. Finding a common language and theoretical foundation will take time.

It is fascinating to see, in the Donaldson and Meltzer commentaries, two mutually exclusive stories about why a safety market will not work.12 Coming from the UK with its strong public health system, Donaldson (see page 87) sees market reforms as a failed experiment. Coming from the US with a strong private system, Meltzer (see page 86) sees markets working well and in little need of government intervention.

In reply to Donaldson, while healthcare may or may not be a public good, safe healthcare surely is. When washing your hands, you will not know which patient avoids hospital-acquired infection, nor which clinician’s life you make easier. No one in hospital is excluded from the safer environment created nor competes for access to it. Donaldson suggests we return to priority setting based upon multiattribute utility theory, but clinical practice is local. There will always be a mismatch between centrally set targets and what is needed and feasible locally, and a disconnect between the costs of meeting priorities—borne mainly by clinicians, and the benefits—typically accruing to administrators, politicians and ideally patients.

Meltzer suggests that putting a price on safety will result in more expensive care, but this is the status quo today. Practitioners pay for system harms through insurance, and transfer those costs to patients. A safety market should reduce such costs. Patients will prefer providers that meet harm-reduction targets and are cheaper, to providers that flout targets and pass the cost of poor care onto patients. Meltzer rightly suggests that in a poorly designed system, there could be less incentive to provide care to higher-risk patients. We proposed the use of case-mix methods that allow providers to risk-adjust the way organisations are measured to avoid this.

Meltzer believes that the incentives to avoid patient harm through market forces are already present. If that is the case, they are ineffective. Even in countries with vibrant private health sectors where patient choice is not restricted, a key market component—the harm signal—is not clearly present. Where are the comprehensive public databases listing clinical outcomes and error rates for patients to make rational choices? And when this information becomes available, would not a major problem be that those managing the sickest would be most penalised?

Coase’s theorem states that government intervention is not needed to manage “external” costs like pollution whenever private rights are clear, and bargaining costs non-existent. As it is often difficult to ascribe causality with patient harm, safety “property rights” can be hard to recognise. Using Meltzer’s analogy, clinicians may be unaware that their actions are polluting their own and other’s property, and so have no incentive to stop. That’s why safe behaviours create a public good. Further, the high transaction costs of dealing with error may be one major reason why market forces have so far been ineffective. Meltzer observes that a safety market will coexist with the legal system. It is unclear if this produces significant double dipping, as the number of court cases are few compared with the daily incidence of avoidable harms. It would be interesting to ask what mechanisms would allow patients and clinicians to bargain up front about the risks both are willing to bear—a more radical universe than the one we currently propose!

We agree with Chin and Wilkes (see page 88) that equity is core, and that the rich have more lobbying power, influence, receive disproportionate benefits and can thus game the system.3 The vulnerable in society need protection. This is a major consideration of our proposal, reflected for example, in the need for equitable issuing of permits.

Our challenge to the community is to no longer consider that business as usual is acceptable. We must explore more pervasive, system-wide solutions, as patient safety is a problem that is endemic, and profoundly systemic—the system produces the outcomes it has been designed to produce. Our healthcare systems, designed in one century, now operate in an entirely different context, and surely need reinvention.4