Last month's consultation document, The New NHS: Modern and Dependable: a national framework for assessing performance, presents a key aspect of the Department of Health's attempts to replace the internal market.

Ministers claim that the best way to achieve improvements in standards of performance and reduce variations across the NHS is by comparing performance and sharing best practice - not by financial competition between different parts of the service. 'Sharing' sounds nice of course; better than the spikiness of 'financial competition'. And it rhymes with caring.

But what is going to be measured, and is sharing it really going to work? The use of performance measures in the NHS in the past has not been an entirely happy experience. As the consultation document makes clear, many of the health service indicators related to inputs and process are likely to be scrapped. Also, that mother of all performance measures - the purchaser efficiency index - will be, in the face of its multiple perversities, scrapped in due course.

Three (at least) problems arise with the new framework. First, what should be chosen as relevant and meaningful indicators of performance?

Second, to what extent can variations or changes in indicators be interpreted unambiguously?

And crucially, what mechanisms (instead of 'financial competition') should be put in place to ensure that poor performance is improved?

Answering these what, why and how questions will be vital to the success of the new framework.

The framework suggests grouping indicators into six areas: health improvement, fair access, effective delivery of appropriate healthcare, efficiency, patient/carer experience and health outcomes of NHS care. By and large, it proposes performance indicators which are already collected as part of the routine statistical system.

So, for example, day case and vaccination rates, unit costs and conception rates for girls under 16 are all suggested as useful measures of performance. However, the greater emphasis on measures of outcome to which the framework aspires highlights the relative dearth of hard information in this area. Focusing on outcomes is clearly the way to go, as was indeed recognised by the previous government.

But the long-recognised problem is obtaining such information without incurring possibly prohibitive costs.

Even if we can be sure we are collecting the right measures - right, primarily, in the sense that they are the ones which accurately reflect the objectives of the NHS - possibly insurmountable problems can arise in our ability to understand what they mean and/or what are the factors producing variations. Why one health authority may perform badly compared with another is, as the framework admits, likely to be a combination of poor performance (and hence rectifiable) and factors outside the control of authorities (and hence unrectifiable), and for which there may, of course, be legitimate reasons.

There are ways to minimise the influence of confounding factors - standardising for factors (such as age and sex) known to produce (acceptable) variations. But can the lack of money, or the historical pattern of hospital buildings, or decisions by local authority social services, or the existence of environmental hazards, or factors such as unemployment and poor housing, be legitimately used to standardise performance indicators?

Finally, having chosen the right indicators and also managed to identify what (and how much) can be done by the health service to rectify poor or variable performance, how is performance to be changed?

The internal market had an answer (theoretically, at least): competition. In practice, competition did not turn out to be as effective as proponents of the market claimed. But poor performance by the market is no justification for poor performance by any substitute process.

Bringing about change in the absence of a market mechanism is going to be difficult, and the new framework is somewhat coy on this matter.

But it has been suggested that the new Commission for Health Improvement (and managerially performed services - CHIMP - as a York-based academic waggishly mused) will have the powers to 'assist locally in raising standards, and as a last resort, intervene'.

HAs, the new GP commissioning groups and providers should ask themselves what the rather ominous words 'assist' and 'intervene' will mean in practice.

The source for all graphs is The New NHS: Modern and Dependable: a national framework for assessing performance. Department of Health, January 1998.

The deadline for responses to the consultation document is 20 March. Please write to Sue Probert, NHS performance branch, Room 4W27, Quarry House, Quarry Hill, Leeds LS2 7UE.

John Appleby is senior lecturer in health economics at the school of health, University of East Anglia.