White paper proposals to determine primary care funding according to a set formula will not solve

inequitable resource allocation.

The white paper offers much of what many NHS staff have been suggesting for some time: locality-based commissioning, reducing unnecessary transactions and bureaucracy, longer-term contracts, avoiding duplication and, not least, abolishing the plain daft efficiency index. Above all, though, the white paper emphasises that this change is to be gradual.

But the seemingly subtle changes in structure could have a much more radical impact on resource allocation. The substitution of practice-based with locality-based commissioning will mean resources must be shared out according to a new unit of analysis. Setting 'fair-share' budgets at practice level under the fundholding scheme was the source of much debate; setting locality budgets is likely to be even harder. In the 'new NHS' resources will flow from the centre to HAs as they do now. HAs will then share out resources to their localities ('primary care groups') as they have done for GP fundholding practices. The key difference is that, between them, primary care groups will represent the whole district population, and a much larger share of HAs' total allocation will be 'delegated' as they will only commission 'special services'.

The source of potentially radical change is the proposed mechanism for resource allocation. The white paper proposes that a national resource- allocation formula be rolled out to primary care. 'The government will put in place new mechanisms to distribute NHS cash more fairly... There will be a national formula to set fair shares for the new primary care groups, as there is now for HAs.'1

There is a considerable literature on resource allocation by formula to large populations.2,3,4

The current version of the formula, used since April 1995 to allocate resources to HAs in England and Wales, uses weighted capitation where crude population data are weighted by age, need and market forces factors. This is simply to reflect the fact that healthcare expenditure is correlated with these variables. The formula is based on a set of regression equations designed to explain observed differences in use and spending between large populations. The formula contains two separate adjustments for need, one for acute services and the other for psychiatric care. The acute model has five variables with co-efficient weightings that are best able to explain observed differences in acute service use: the all-cause standardised mortality ratios for under 75-year-olds, the number of elderly people who live alone, the number of single carers, the rate of unemployment, and the standardised long-term illness ratio for under 75-year-olds. The variables in the psychiatric model are the number of permanently sick, lone parents, elderly people living alone, residents from the new Commonwealth and those without a carer. The current version of the national formula applies the two needs-adjustment models according to the national proportion of health expenditure. This means that 64 per cent of total hospital and community health services is weighted according to the acute model, 12 per cent is weighted according to the psychiatric model, and the remaining 24 per cent (mostly community services) is allocated without any adjustment for need.

There has been much debate about this national formula, not least about the effect of the unadjusted element on the effects brought about by the needs- adjustment models. Academics do not agree on exactly how this should be handled. There is, however, general consensus that, to date, after considerable research, no regression equations have been found that can satisfactorily explain variations in use at the sub-district level.

Given the lack of alternative methods of resource allocation within districts, several HAs have already carried out some work applying the national formula at sub-district level. Newcastle and North Tyneside HA has had a long- running interest in locality commissioning. In 1996, a sub-district resource allocation exercise was carried out with the aid of data-manipulation software from the York Health Economics Consortium. The objective was to derive 'fair-share' indicative budgets for each locality and each GP practice by applying the national formula. Separate work applying programme budgeting had already produced estimates of actual resource use by locality and practice. A comparison of results from the two strands of work revealed the extent of possible re-allocation that might occur with the introduction of the allocation formula.

Figure 1 shows the percentage change to each locality budget based on the difference between estimated current resource use and the budget calculated by applying the national formula to these results. It is clear that if this mechanism of resource allocation were introduced, change would be far from gradual.

Figure 2 shows the amount each locality budget would change by if the global HA budget was pounds200m. North locality would receive an additional pounds2.2m, while East locality would receive over pounds0.8m less. It is clear there would be winners and losers. North, Whitley Bay and Tynemouth localities all appear to be made better off by a move to allocating resources according to the national formula, while the East and West (inner and outer) ends of Newcastle, Longbenton and Wallsend localities all appear to lose out.

The objective of resource allocation by formula is greater equity. However, it seems apparent that the set of results generated from this rigid application of the national formula at the sub-district level are far from appropriate and appear, in fact, to recommend what appears to be a more inequitable allocation. The allocation arrived at by the formula is also likely to be politically unacceptable.

Results may be explained intuitively. It is likely that North locality gets a larger share under formula allocation due, in part, to the university practice there, where health service use is much lower than at other practices, but the per capita-based formula payment fails to adjust for this sufficiently. Similarly, Whitley Bay and Tynemouth localities include seaside areas with larger proportions of elderly residents; thus the age adjustment in the formula increases their share. All three of these localities also include the most affluent areas of the district, where service use (and need) are generally lower than elsewhere. East and inner-west Newcastle localities are well-known deprived areas where health service use is high, but the national formula does not compensate for this. Thus the formula recommends shifting resources away from some of the most obviously deprived areas within the district, and possibly the UK, to enable additional resources to be put into other areas of the district.

The issue of whether these other areas need more resources is a separate issue. Is the allocation arrived at by formula better (fairer) than that which existed previously?

Perhaps the underlying reason that this exercise generated such results is the general inappropriateness of the formula at this level. These results should not be surprising given that the formula

was not designed to be applied to smaller populations below the district level. The 'best-fit' variables arrived at in the needs-adjustment models and the co-efficient weightings calculated were based on a different unit of analysis.

These variables are unlikely to explain differences in use between localities, so spurious results are inevitable.

It is worrying that the white paper is proposing to allocate resources to primary care groups by means of a formula when rigid application of the current national formula is capable of producing inappropriate results, and researchers have not found suitable alternatives. Part of the attraction of the resource-allocation formula to policy makers must be that it is presented as a technical, seemingly apolitical, tool which 'automatically' shares out resources equitably. This is, of course, a grave misconception. The design and implementation of formulae are inherently political.

Despite the apparent absence of 'good-fit' models at the small area level, some improvements to the national formula could be made. Most significantly, the current national formula makes no direct adjustment for deprivation, and indeed the white paper proposes that deprivation be built into a formula in some way: 'The healthcare needs of populations, including the impact of deprivation, will be the driving force in determining where the cash goes.' The architects of The New NHS may, however, need a 'reality check' on their good intentions. Yes, resource allocation by formula can be a useful guide to decision-makers, but caution is needed for several reasons. A suitable evidence base for any formula is essential if it is to be reliable and valid; no such evidence base currently exists at the sub-district level. Even with the best models, rigid, unthinking application of a formula, without adjustment for local factors, is likely to be dangerous. When formulae have the potential to 'recommend' significant re-allocation of resources, we need to be sure they are right before such sweeping changes are implemented.

A further concern is that far-from-gradual resource re-allocation, brought about by the introduction of a formula, could be well disguised in the 'new NHS'. Current locality budgets are not generally known since the locality is a new entity in many areas. Practice budgets under the fundholding scheme will be known, but the goalposts will be moved: primary care groups will be more like 'group fundholding' with several practice budgets combined, but also the range of services commissioned will be greater than under the fundholding scheme. Other proposals such as practice-based contracts, and merging budgets for general medical services, hospital and community health services and pharmaceuticals could potentially further mask large- scale shifting of resources within a district.

While the white paper has received general praise from both the health service and research worlds, we should not be blinded by its good intentions. Methodologies must be examined and tested. Resource allocation by formula is not the holy grail. At best it will provide decision-makers with guidance and rules of thumb; at worst it will mislead and worsen existing inequalities in health.

We should not have blind faith in the capabilities of econometric models to re-allocate resources in a genuinely equitable fashion. Formulae can not substitute for good judgement and pragmatic management.