A lack of clarity around the effectiveness of out of hospital interventions is preventing their potential cost efficiencies from being realised. But, says Nuffield Trust director Jennifer Dixon, there are reasons to be cheerful.

Let’s not be too sidetracked by the drama of the Health and Social Care Bill’s passage with respect to GP consortia. The real drama is how to gain much better efficiency in provision.

Perhaps the biggest hope is that more “out of hospital” interventions will prevent avoidable downstream ill health and hospital costs, particularly for older people and those with chronic or terminal conditions. That is why there is such interest in the concept of integrated care. It seems reasonable to assume that prevention (primary and secondary) is possible, and that it will cost less than the treatment in hospital that would otherwise result.

But finding strong evidence that out of hospital interventions are cost effective is harder than you might think – in particular, interventions that might be called integrated care, as a forthcoming review by my colleague Vidhya Alakeson will show. Why?

First, the interventions being piloted are often not accurately defined or described. In particular, the precise aims are not made clear.

Second, the context in which the intervention is implemented is rarely described, even though it may have significant impact. For example, the financial incentives operating in the wider health system. This point is particularly relevant because much of the literature comes from outside the UK, often from entirely different health systems.

Third, the patients selected to have the out of hospital intervention may not be those in whom the intervention would have much impact.

For example, preventive care to patients who are currently experiencing multiple hospital admissions can be inefficient because, even without intervention, such patients will on average have fewer unplanned hospital admissions in the future.

This phenomenon is called “regression to the mean” and it implies that hospital-avoidance interventions are best offered according to a person’s future, not current, risk of hospitalisation. Few pilots have stratified the future risk of the patients and identified patients at high risk for intervention (case management).

Costly and weak

Fourth, the strength of implementation of any intervention or pilot is assumed to be 100 per cent from day one, whereas in fact it might rise from a homeopathic to sub-therapeutic dose over the course of many months because of unserious management.

Fifth, the intervention may be evaluated too soon – at least one year is needed.

Sixth, the evaluations frequently do not examine the costs of an intervention, or may not use the right methods to test “what would have happened anyway” or use an appropriate control group. Too many evaluations are costly, weak and take too long.

Thus some initiatives may have erroneously been shown to reduce emergency admissions due to faulty evaluation. Others (I suspect the Evercare programme) did not show impact, not because the intervention was weak, but because risk stratification was not available at the time to select the right patients for case management. In the US a key lesson from the Medicare health support programme (a chronic care pilot programme running from 2005-08) was to avoid flawed strategies for selecting populations to target.

Until we iron out some of these problems, we cannot know what type of intervention works, and their potential to result in the hoped for efficiencies. This is not, repeat not, to say integrated care or out of hospital initiatives do not work.

This is all inconvenient for a number of reasons. Not least because clinicians and managers with good ideas need to construct strong business cases for petitions for invest-to-save funding from the local primary care trust clusters (where funds are available) or other sources. And Monitor, if the Health Bill is largely unchanged, will have to form a view as to the extent of “vertical integration” it is prepared to accept. If the case for greater efficiency of initiatives is flimsy, they may be rejected.

But there are reasons to be cheerful. There seems to be an emerging consensus as to the ingredients of effective interventions, in particular integrated care: integrated information systems, good clinical leadership, aligned financial incentives, and high patient participation. Risk stratification – identifying patients’ future risk of hospitalisation – is now a common activity by commissioners.

There is widespread clinical and managerial interest in greater collaboration across primary and secondary care, and with social services to reduce avoidable ill health and costs. Initiatives are widespread across England – for example, the Department of Health-funded integrated care pilots and whole-system demonstrator programme.

And there are now much better tools for evaluation which are rigorous, allow good matching, with controls using innovative data linkage, are cost effective and provide results quickly.

To speed up progress, we need a better system to give seedcorn investment to trial good provider-focused ideas; initiatives with stronger (financial) incentives and management embedded; rigorous evaluation; and national surveillance as to what works. Otherwise 10 years from now we will be none the wiser.