The original idea behind payment by results was disarmingly simple: more work, more money. But the dilemma was always how to pay for it in a closed, cash-limited system. So what can we anticipate from the consultation paper on the future of payment by results due by the end of February?

Two big themes stand out. One is the future scope of payment by results. Just how serious is the Department of Health about extending tariff-based funding, and the commissioning model that accompanies it, into mental health, long-term illness, community services and primary care?

Such an extension was always the DoH's aim. But it remains ambitious, and last autumn a curiously laissez-faire approach to 'unbundling' was adopted for 2007-08. Not quite a retreat, but hardly a charge either. The focus of unbundling effectively narrowed to those elements of care pathways, such as diagnostics and rehabilitation, which alert commissioners might wish to strip out of acute trusts and transfer to primary care.

This refocusing was pragmatic. It recognised the paucity of good information systems outside the acute sector; it also circumvented potential conflict with the practice-based commissioning lobby. But what now?

If the non-acute sector remains outside payment by results, will it continue to bleed resources? What will be the fate of mental health, for instance, in a context of continuing acute hospital cost pressures, driven by the maximum 18-week wait target and reduced financial growth?

The second key issue is the depth of analysis to be used in acute care. As if coping with the more subtle level of analysis that the new healthcare resource group set will offer were not challenging enough, attention has quietly moved to the level of supporting costing detail that trust management should have at its fingertips.

Let's be clear here. We are talking about quite different things. In the NHS the HRG set determines the tariff, or price list, by which providers get paid. Costing systems, on the other hand, are a management tool for identifying the actual costs a provider incurs.

Imagine you are factory worker paid by the item you produce: 'piece work', as it's traditionally called. It's one thing to redefine the scale by which you get paid; another to analyse how you spend your wages. In the NHS there is a link - the tariff is based upon actual costs - but the link is getting more distant and there are strong hints the NHS may move to a tariff based on what a procedure should cost. So what's going on?

Monitor has been advocating what it calls 'service-line economics': breaking down the activities of a hospital into distinct trading units, and assessing their individual profitability. It's a logical enough extension of the PbR philosophy and it fits management structures built around clinical directorates, but there are two fundamental challenges.

It requires a system of internal recharging covering the costs of all departments that do not directly earn revenue, from pathology to physiotherapy to human resources; and a process for allocating the additional income that, in principle, flows from commissioners when trusts perform extra activity.

Service-level budgeting pilots are already being run at three foundation trusts in southern England. But although this is pioneering work, there is an emerging view that analysis at a trading unit level is simply not sophisticated enough.

What is the appropriate building block for measuring the relative profitability of parts of a hospital? The consultant, perhaps? Or do we really need costing analysis built up from individual patient data?

This appears to be the belief of Peter Donnelly, of the DoH payment by results team, who argues that without patient-level detail, cost data will simply not be persuasive enough and comparisons of clinician performance will be too fragile.

The impact could include radically different ways of setting budgets. Nursing budgets, for instance, could be based not on incremental growth but on an anticipated number of patient days, a ratio of nursing hours per patient day and a standard cost per nursing hour. And that's before in-year adjustments for actual levels of activity.

Now, there is no essential incompatibility between patient-level costing and service-level accounting but it doesn't come for free, and working up from patient data needs a different league of software capability. At which point, for old NHS hands, alarm bells may start to ring. There is a body of overseas experience in using patient-level costing software, and indeed all private hospitals use it in some form to justify their charges, but creating the genuine software capacity to support NHS volumes is another matter entirely.

What's really interesting about the push for service-line economics and patient-level accounting is the way it switches management attention back into the hospital, and especially towards comparisons between clinicians. Instead of growth projections built around better use of fixed assets, we return to the safe, miserable territory of hunting down and eliminating waste.

This is a major change of direction from the early years of payment by results, when the point was to offer clinicians financial incentives to do more work. It recognises the strategic reality of reduced growth funding under the 2007 comprehensive spending review, and the short-term context of a new model contract, published in late January, that limits the financial exposure of commissioners. But it carries within it the potential for divisiveness between clinicians, and between clinicians and management.

It was never really payment by results; it was always payment for activity. But the NHS responded to the financial incentive, as predicted. Whatever else our funding system does, it must also keep good incentives for clinicians - and the organisations that employ them - to increase their productivity. 

Noel Plumridge is an independent consultant and former NHS finance director.