Explaining variations in general practice prescribing costs per ASTRO-PU (age, sex, and temporary resident originated prescribing unit)BMJ 1996; 312 doi: https://doi.org/10.1136/bmj.312.7029.488 (Published 24 February 1996) Cite this as: BMJ 1996;312:488
- David K Whynes, reader in health economicsa,
- Darrin L Baines, research studenta,
- Keith H Tolley, lecturer in health economicsb
- a Department of Economics, University of Nottingham, Nottingham NG7 2RD
- b Nottingham School of Public Health, Medical School, University of Nottingham, Nottingham NG7 2UH
- Correspondence to: Dr Whynes.
- Accepted 3 August 1995
In the light of evidence which suggests that the use of medicines is in some degree predictable from the demographic structure of practice populations,1 a move has been made towards the use of a weighted capitation formula in determining general practitioners' prescribing budgets. One recently devised weighting structure, the age, sex, and temporary resident originated prescribing unit (ASTRO-PU), allocates individual weightings to nine age groups for each sex, plus one for temporary residents, in the practice list.2 Family health services authorities were advised that they might use weightings based on ASTRO-PUs to adjust the prescribing allocations, otherwise based on historic costs, for the financial year 1994-5.3
Over and above a budgetary allocation so determined, scope will always exist for discretionary or “soft” factors, reflecting particular patient need in individual practices.4 Evidence on what should be considered as genuine soft factors, however, is not readily available. We report a study of prescribing variations in a single English health authority, with a view to identifying factors influential in determining prescribing costs per ASTRO-PU.
Methods and results
Data on a wide range of general practice characteristics were available for 99 out of the 108 practices in Lincolnshire for 1993. Multiple regression analysis of the data was performed with prescribing cost per ASTRO-PU as the dependent variable. The mean cost per ASTRO-PU for the sample was pounds sterling18.898 (SD 3.193). Descriptive statistics of the independent variables and regression results are shown in table 1. The model was estimated in both a full and a final form, the latter entailing the addition or removal of variables from the full model in order to obtain significant coefficients and the best possible degree of goodness of fit. In the final model the coefficients indicate that, evaluated at the mean, a 1% increase in night visits raises costs per ASTRO-PU by about 4p, a 1% increase in payment exemptions raises these costs by about 5p, and a 1% increase in the rate of generic prescribing lowers costs by about 3p.
On the basis of the models we conclude, firstly, that prescribing costs per ASTRO-PU are associated with the prevalence in the list of payment exemption certificates issued by the family health services authority. These exemptions are offered to patients with a range of disabling medical conditions and to those holding prepayment vouchers precisely because they are likely to be extensive users of prescribed medication. Alternatively, general practitioners may be conscious that such patients bear no financial costs and thus feel less obligation to economise on their prescriptions.
Secondly, prescribing costs are related to night visits. This can be explained either by presuming that night visits indicate morbidity related to deprivation5 or by accepting that a night visit often leads to the issue of an emergency prescription, which is followed up with a regular prescription at a subsequent arranged consultation.
Thirdly, an increased rate of generic prescribing lowers prescribing costs. This result is to be expected, given the comparative costs of generic and branded medicines, and has been suggested in previous studies.4 Finally, both fundholding and the use of practice formularies seem to exert a cost reducing effect. Of equal interest, however, are the insignificant variables in the equation. There is no evidence that, for example, hospital referrals, geographical location, unemployment, or the staffing structure of the practice influences prescribing costs per ASTRO-PU.
Our final model succeeds in explaining about 42% of the variation in prescribing costs. The architects of the ASTRO-PU believed that their weightings would account for around 25% of prescribing cost variation,2 and our analysis has exposed additional factors which contribute to the overall determination of prescribing cost for each practice.
We gratefully acknowledge the financial and research support of Lincolnshire Health. The opinions expressed are solely those of the authors and do not necessarily reflect the views of Lincolnshire Health.
Funding Lincolnshire Health.
Conflict of interest None.