High profile initiatives such as the incentive programme introduced by the Centre for Medicare and Medicaid Services in the US¹ and the new general practitioner contract in the UK² highlight the enthusiasm of policy makers for using financial incentives as a way of improving the quality of care. This enthusiasm is understandable, given the burden of healthcare costs experienced by most countries. It makes sense to ensure that resources are targeted on buying desirable behaviours from health professionals and producing beneficial outcomes for patients.

But is the fascination with financial incentives based on sound empirical evidence? At a general level the answer is a guarded “yes”. We know from observational studies that the way in which doctors are paid is associated with particular patterns of clinical behaviour. For example, doctors paid under fee-for-service schemes undertake more visits and conduct more investigations than those paid under capitation schemes.³ In contrast, it is less easy to find a convincing causal link between targeted incentives and the behaviour of individual doctors, and little attention seems to have been paid to what might be termed “spillover” effects—that is, the impact of incentives on behaviours other than those incentivised. In part, this lack of evidence results from the methodological challenges associated with linking interventions to complex behavioural change. Even taking this into account, the evidence still leaves us with the impression that incentives do not induce the rational and predictable response that some observers would have us believe.

There are several examples to illustrate the problem. Firstly, the size of an incentive does not have a linear relationship with its impact. Indeed, there is some evidence that doctors may have a target income—perhaps a fixed sense of financial worth—above which they are no longer motivated to respond.⁴ Secondly, it also appears that the economic component of what appears to be a financially based incentive scheme is not what motivates professionals. In a local improvement project in the UK, much vaunted as a “successful” example of incentivising quality improvements, the costs to some of the participating general practices of implementing more effective systems of chronic disease management were greater than the resulting financial rewards.⁵ This did not seem to dampen the enthusiasm of those involved. Similarly, in a study conducted in Ireland, incentives to change prescribing behaviour were just as effective in dispensing practices (where there is a countervailing incentive to dispense expensive drugs) as in non-dispensing practices.⁶ These examples indicate that something more than personal financial gain is driving professional behaviour.

In attempting to explain this anomalous evidence, attention has focused on confounding variables such as the age and sex of physicians, their previous experience of incentives and payment methods, the type and severity of the condition being incentivised, the volume of activity, and the location and type of practice.⁷ All of these factors seem to be important, but together they fail to account for the unpredictable and variable impact that has been observed. A more convincing explanation perhaps lies in the relationship between external incentives such as material rewards and the internal “moral” motivation of health professionals.

Frank defines moral motivation as a force which encourages people to behave in ways which have no obvious advantages to the individual and may even prove contrary to their interests.⁸ Every day we see examples of this kind of behaviour—customers who leave tips in restaurants to which they will never return, people who make anonymous charitable donations, and health professionals who “go the extra mile” with their patients with no thought of financial reward. One study even showed that insurance salesmen—a group not popularly recognised for their moral drivers—have been shown to be guided by a strong, almost religious, moral code.⁹

If policy makers and managers have convinced themselves that they can buy “desirable” behaviours, why should they be concerned with the internal motivation of their workforce? The answer lies in the potential of externally imposed incentives to impact on internal motivation, even where such activities are recognised as the right thing to do. Evidence of the effects of disregarding moral motivation can be found in the literatures of economics, social psychology, and organisational sociology.

From an economic perspective, Frey describes this as “crowding out”.¹⁰ The psychological processes underlying the phenomenon have been explained in two ways. Firstly, external incentives may impair self-determination, resulting in a shift in the locus of control and the resulting loss of professional autonomy. Secondly, external drivers may damage self-esteem, resulting in the perception that professionalism is no longer valued. Crowding out appears to be more marked when external incentives are linked to perceived regulatory activity and managed in a bureaucratic fashion by people unknown to the recipients of the incentives. In contrast, if people feel that they “own” the incentives, then they can have the effect of enhancing internal motivation (the “crowding in” effect). It appears that more mechanical tasks are less likely to be crowded out than creative ones. This might explain the support for incentives from those who are inclined to focus on the technical aspects of delivering care, and the antipathy of others who focus on the “art” of clinical practice.

From a social psychological perspective there is ample evidence, including a meta-analysis of 128 experimental studies,¹¹ that “crowding out” is a real phenomenon. The literature of organisational sociology has recognised for more than half a century that incentivisation of rule governed behaviour is likely to lead to “goal displacement” in which rule following becomes a means to an end other than that intended by the designers of the system.¹² This observation has been drawn upon in more recent sociological writing about the displacement of trust and moral motivation brought about by the current emphasis on “managing” the performance of health and social care professionals.¹³

Financial incentives will no doubt continue to play an important role in the armoury of tools available to improve the quality of health care. Their contribution will, however, only be maximised if we understand the impact of financial incentives on the internal drivers of health professionals. In this respect there are some quick wins for those designing incentive schemes. It seems likely that financial incentives will be more effective if they are owned by their target audience and aligned to the professional values of this audience. It is also likely that overly bureaucratic schemes are more likely to damage professional motivation and that the incentives should be targeted more on the technical aspects and less on the indeterminate aspects of professional practice. It would therefore be inappropriate to attempt to link financial rewards to complex diagnostic processes or to the psychosocial aspects of care provision.

Beyond this, there is much that we do not know about how best to use incentives to change the behaviour of health professionals. In particular, it is essential that we develop a deeper understanding of the relationship between incentivised and non-incentivised professional work. The new UK GP contract provides a case in point. On the one hand, the incentivisation (agreed with the profession itself) of indicators about the routine treatment of single chronic conditions does make sense in the context of the evidence. On the other hand, it seems possible that, as an increasing proportion of total GP work is incentivised, the risks of crowding out of motivation to perform the non-incentivised more complex or simply caring tasks is increased. This is an area of policy that really does need to be underpinned by high quality evidence.