EDITOR—Rice et al present what they describe as a “needs based” formula for allocating prescribing budgets.1 This seems an improvement on the ASTRO-PU (age, sex, and temporary resident originated prescribing unit), which it replaces, but in view of its crucial impact on the resources available to general practitioners and their patients its limitations should be stated more clearly.

Despite its title, the formula does not assess need directly but relies on data from the national census to generate proxy measures. The association between these measures and prescribing costs “explains” observed variation in these costs only in the narrow statistical sense of the word. Bains and Parry,2 and Majeed3 point out further important limitations.

These criticisms are of more than academic importance. The formula is “needs based” only in a vague and general sense, but despite its manifold weaknesses there is a danger that NHS organisations will use it as if it were an adequate basis for budget setting and monitoring prescribing performance. The patients whose access to treatment will be thus determined and rationed will not be proxies.

Primary care groups have the difficult task of salvaging something from this minefield. If we wish to explain variations in prescribing in the full sense of the word and if we are serious about the pursuit of equity and quality, we have a great deal of work to do. Several actions spring to mind.

Firstly, computerisation gives us the means to collect detailed morbidity data at practice level. This will allow us to test the formula for allocating prescribing budgets against real measures of need and, if necessary, make allowances in practice based budgets.

Secondly, by combining prescribing and morbidity data, and auditing standards of care, we can ensure that measures of quality are built into our incentive schemes. We will also gain considerably in our understanding of the many causes of variations in prescribing costs.

Thirdly, we might consider foregoing the right to keep savings from our practice prescribing budgets. In a cash starved NHS, large handouts for cheap prescribing should be seen as the occasion for red faces rather an opportunity for red carpets.

EDITOR—Equity is a central concern of primary care groups, which are responsible for allocating prescribing budgets to practices. Unfortunately, existing weighted capitation formulas can produce anomalies at practice level. Primary care groups face having to make subjective adjustments that are neither transparent nor acceptable. The promise of a new, intuitively plausible formula,1 especially one that is at last based on registered practice populations, will therefore, as Majeed warns,2 be attractive to primary care groups.

However, primary care groups should not use Rice et al's formula, for at least three reasons. Firstly, it uses the fatally flawed method of trying to predict the needs of practice populations from attributed data derived geographically: the ecological fallacy.

Secondly, applying existing formulas at practice level has long been controversial.3 Rice et al do not propose their formula for calculating practice prescribing budgets.

Thirdly, the inherent weaknesses of using existing census data are readily admitted by health economists.4 Why then continue to conjure formulas from poor data of doubtful relevance?

Using registered practice populations instead of attributed census counts is a breakthrough, but it is not sufficient on its own. Clinical research into measures of healthcare need should be funded and promoted by primary care groups. Indeed, information on the back of prescription forms has already been used by Lloyd et al for a low income index of deprivation.5 This is immeasurably more plausible than using old census data on the percentage of dependants in no carer households.

Years of health economics and statistics have produced practice budgets that are like rainbows. They have shape and colour but do not touch the ground. What we know about best, on the ground in general practice, is prescribing for individual patients. I prefer a bottom up approach to budget setting, driven by collaboration among all the prescribers in the primary care group. As clinical prescribing data are increasingly computerised, audit can become more extensive and the quality of care be assessed in greater detail, including cost effectiveness. We should aim to set prescribing budgets for our practices on summated data about individual patients and their care. Meanwhile, cost growth comparisons among practices will alert the primary care group to unequal use of the budget.

Perhaps the doctors and nurses charged with promoting equity in the “New NHS” are best placed to assess whether practice prescribing budget calculations are “intuitively plausible,” at least until rainbows touch the ground.