How has fundholding in Northern Ireland affected prescribing patterns? A longitudinal study
BMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7101.166 (Published 19 July 1997) Cite this as: BMJ 1997;315:166
- a Drug Utilisation Research Unit, Department of Therapeutics and Pharmacology, Queen's University of Belfast, Belfast BT9 7BL
- Correspondence to: Miss Rafferty
- Accepted 28 May 1997
Abstract
Objective: To compare prescribing patterns in general practices before and after the introduction of fundholding in April 1993 to determine whether fundholding changed prescribing patterns among practices that joined the scheme.
Design: Analysis of prescribing data from the Drug Utilisation Research Unit's database for all practices in Northern Ireland during April 1989 to March 1996.
Setting: Northern Ireland.
Subjects: 23 first wave fundholders, 34 second wave fundholders, 9 third wave fundholders, and 268 non-fundholders.
Main outcome measures: Prescribing costs per 1000 patients, prescription items per 1000 patients, average cost per item, and rate of generic prescribing.
Results: Prescribing costs and frequency increased in all groups throughout the study. Among the fundholders the rate of increase in costs after fundholding was significantly lower than among non-fundholders. The rate of increase in cost per item fell, coinciding with a significant increase in the rate of generic prescribing. However, with regard to first wave fundholders, their yearly increase in costs in their third year as fundholders (1995-6) was similar to that of the non-fundholders. The earlier practices that joined the scheme seemed to differ in some important respects from those that joined later.
Conclusions: After fundholders joined the fundholding scheme their patterns of prescribing changed compared with those of non-fundholders: the rate of increase in costs fell and there was a significant rise in the rate of generic prescribing.
Key messages
The effects of fundholding on prescribing costs have been disputed
Fundholders in Northern Ireland contained the rate of increase in prescribing costs more effectively than non-fundholders
Fundholders increased their rate of generic prescribing by an average of 13% in the first year of fundholding
The incentive to make further savings may have diminished after two years of fundholding
Introduction
Containing escalating health care budgets is an international problem. General practitioners' prescribing makes a big contribution to the rising health care budget and various measures have been used to influence their prescribing behaviour. These measures usually provide doctors with information relating to drugs and prescribing patterns. Some provide general information about a particular drug or group of drugs, encouraging restrained use.1 Others provide prescribers with detailed analysis of their prescribing behaviour with specific instructions for change.2 This type of information is sometimes presented to the prescriber by a prescribing adviser, a technique which Soumerai et al found to be the most effective.3 Some countries penalise doctors who fail to comply with guidelines or to prescribe within certain boundaries. In the Netherlands overprescribing can result in non-reimbursement; in France doctors can be fined if they blatantly ignore guidelines. Other countries reward doctors for good practice. In parts of the United States doctors are rewarded with discretionary salary increments for using resources efficiently, and French doctors receive increased fees for complying with guidelines.4
In 1991 Britain introduced the fundholding scheme to contain costs. The scheme is voluntary and allows practices to hold budgets for several services including prescribing. Money can be moved between the various elements of the budget. The incentive of the scheme is that the practices are allowed to redirect any money left in the budget at the end of the year back into the practice to be used to benefit patients.5 The prescribing bill has the greatest potential for savings, and several studies have shown the effects of this scheme on prescribing costs among general practitioners in Scotland and England. Most show that fundholders can contain costs more effectively than non-fundholders6 7 and that costs are contained by increasing generic prescribing7 8 and decreasing the average cost per item rather than by giving fewer items.9 However, these initial achievements may not always be maintained.10
Fundholding was introduced in Northern Ireland in 1993. First wave fundholders started in April 1993, and further practices joined the scheme in April 1994 and April 1995. This study looks at the effects of fundholding on the prescribing patterns of practices that joined the scheme.
Methods
Our unit's prescribing database records the cost and number of prescriptions written for each preparation by each practice in every quarter. This database is built from data downloaded from the Central Services Agency, whose system was designed for reimbursing pharmacists. We compared the performance of first, second, and third wave fundholders with that of non-fundholders during April 1989 to March 1996 using the following prescribing measures: prescribing cost (£) per 1000 patients, prescription items per 1000 patients, average cost (£) per prescription item, and rate of generic prescribing.
We included all practices in Northern Ireland except for those that did not exist for the entire study period. Dispensing dentists and hospices were also excluded as no list size is recorded for them. The term “non-fundholder” refers to practices which were not fundholders at any point in the study.
Throughout the study period, practices in Northern Ireland received regular visits from a prescribing adviser. Practices also had equal access to prescribing reports similar to those issued by the Prescription Pricing Authority in England and to the COMPASS report (computerised online monthly prescribing analysed for science and stewardship), which gives specific indications where each practice can make savings.2 When the fundholding scheme was introduced in 1993 the indicative prescribing scheme was already in place, but there were no positive or negative incentives for practices to prescribe within their budgets. No other incentives schemes to improve prescribing existed after the introduction of fundholding.
We analysed the data using the SPSS for Windows package. Differences were compared with the multiple comparison t test procedure with the Bonferroni correction.
Results
Table 1 gives the characteristics of the practices included in the study. List sizes ranged from 449 to 13 243 (mean 4479) among non-fundholders and from 1366 to 15 993 (mean 6887) among fundholders.
Characteristics of fundholding anh non-fundholding practices. Values are numbers (percentages) of practices unless stateh otherwise
Prescribing costs per 1000 patients rose in all groups throughout the study (fig 1). The fundholders had slightly lower costs than the non-fundholders until they joined the fundholding scheme, although the differences were not significant (table 2). After this their costs were significantly lower than those of non-fundholders. The percentage yearly increases in cost per 1000 patients were similar for fundholders and non-fundholders before fundholding, but once practices became fundholding their costs were significantly lower than those for non-fundholders. Although the costs of first wave fundholders remained significantly lower than those of non-fundholders, in the third year of fundholding these practices had a yearly percentage increase similar to non-fundholders (table 2).
Mean cost (£)/1000 patients, items/1000 patients, cost (£)/item, and generic percentage for fundholders and non-fundholders for each quarter from April/June 1989 to January/March 1996
Mean cost (£) (mean % increase on previous year) per 1000 patients for fundholders anh non-fundholders
Prescribing items per 1000 patients rose in all groups throughout the study (fig 1). First and third wave fundholders prescribed significantly fewer prescription items per 1000 patients than non-fundholders throughout the study (table 3). The second wave fundholders prescribed significantly fewer items per 1000 patients than non-fundholders after becoming fundholders and for three of the five years previously. Fundholders and non-fundholders had similar yearly percentage increases in items per 1000 patients for most of the study, although in their first year of fundholding the first wave's yearly rate of increase was half that of the non-fundholders (table 3). In their second and third year of fundholding the rate of increase in items per 1000 patients among first wave fundholders was the same as for non-fundholders.
Mean number of prescription items (mean % increase on previous year) per 1000 patients for fundholders anh non-fundholders
Cost per item rose in all groups (fig 1). The yearly figures show that before becoming fundholders the first wave had a significantly higher cost per item than non-fundholders, but on joining the scheme their cost per item did not differ significantly from that of the non-fundholders (table 4). Second and third wave fundholders were similar in cost per item to the non-fundholders from 1989 to 1993, but on becoming fundholders their cost per item fell below that of the non-fundholders and the figure was significantly lower for the second wave in their second year as fundholders. Fundholders had generally similar percentage yearly increases to those of non-fundholders until they joined the scheme, when their yearly increase was notably less than that of the non-fundholders.
Mean cost (£) (mean % increase on previous year) per item for fundholders and non-fundholders
Generic prescribing—The percentage of generic prescribing for non-fundholders remained around 25% throughout the study (table 5). The fundholders remained at a steady, though somewhat higher, level until they joined the scheme. On becoming fundholders their rate of generic prescribing significantly increased compared with non-fundholders. Before fundholding was introduced only small increases (≤2%) in generic prescribing occurred among all groups. In the year they became fundholders, however, first wave fundholders had a yearly increase of 14%, the second wave had an increase of 11%, and the third wave an increase of 15%.
Mean generic prescribing rate (%) (mean % increase on previous year) for fundholders and non-fundholders
Discussion
This is one of few studies of prescribing behaviour that examines an almost complete population and includes behaviour for all practices before fundholding began. Our results confirm that fundholding has a significant effect on prescribing patterns. Fundholders contain costs more effectively than non-fundholders, and this change coincides with practices joining the fundholding scheme. The same pattern can be seen in all three waves of fundholding. This cost containment seems to have been achieved partly through an increase in generic prescribing, which is reflected in the cost per item. However, our data cannot show if there were other changes in prescribing patterns such as substitution of less expensive alternative drugs, or if prescriptions were written for shorter periods or for lower doses. A study looking at specific treatment groups using the defined daily dose method would give some indication of what changes occurred.
Differences between fundholders and non-fundholders
Before joining the fundholding scheme first wave fundholders had slightly lower costs per 1000 patients, fewer items per 1000 patients, and a higher cost per item compared with non-fundholders. Second and third wave fundholders had slightly lower costs per 1000 patients, fewer items per 1000 patients, and a similar cost per item compared with non-fundholders. Work in other regions suggests that the characteristics of the first wave fundholders are typical of practices in more affluent areas, which are larger and more organised than those in poorer areas.7 11 First wave fundholders were mainly larger practices, with almost 50% having five or more partners per practice (table 1), partly as a result of the entry requirements (minimum list size of 7000 patients and no single handed practices).
Data on the number of deprived patients in each practice, which is based on a modification of the Jarman index,12 suggests that first wave fundholders had the fewest deprived patients per partner. The fact that first wave fundholders are larger practices and have fewer deprived patients does not necessarily prove that they are in more affluent areas, considering the criticisms of the Jarman index.13 14 However, it does show that first wave fundholders are not typical practices, and this may have affected their response to the fundholding scheme. Comparing each fundholding group with the non-fundholding group for each quarter we found that the cost per 1000 patients of first wave fundholders started to differ significantly from that of the non-fundholders in the quarter before joining the fundholding scheme (P=0.04). Later fundholders differed significantly from the non-fundholders in the quarter they became fundholders, but not before.
It has been argued that the way in which budgets were set may have encouraged prospective fundholders to inflate their costs in order to obtain more money.15 Several studies, however, found no evidence of this.7 16 In Northern Ireland prospective fundholders had to carry out a prescribing study during their preparatory year to identify how they might prescribe more effectively once they became fundholders. Only very high cost practices were expected to show some improvement during their preparatory year. Any inflation of costs, therefore, would have occurred in the year before their preparatory year. Our results show no evidence of generalised wilful cost inflation, but five practices may have inflated their costs in the year before their preparatory year, one of which reduced its costs in the first year of fundholding and two in their second year of fundholding.
Possibilities for savings
Bradlow and Coulter found that none of the practices that they studied managed to reduce absolute costs on becoming fundholders.11 In our study seven practices managed to reduce absolute prescribing costs in their first year of fundholding: 42 practices decreased their yearly increase on the previous year and 10 increased their yearly increase on the previous year. Nevertheless, in the third year of fundholding, first wave practices had a similar yearly percentage increase in prescribing costs to non-fundholders (table 2). Whether second and third waves will follow this pattern remains to be seen, but the finding may suggest that the incentive for making further savings has diminished after the first two years of fundholding. In addition, in their second and third years of fundholding, the rate of increase in items per 1000 patients among first wave fundholders was the same as among non-fundholders, giving further evidence that the incentive for reducing prescribing frequency cannot be maintained.
Generic prescribing rates in Northern Ireland are lower than in England. Figures for 199517 show that 55% of drugs were prescribed generically in England compared with 27% in Northern Ireland for the same year. However, 11% of the generically prescribed drugs in England were dispensed as proprietary drugs. This is because certain drugs are patented under a specific brand name and cannot be produced by another company until the patent expires. Drugs prescribed generically while still under patent are not recorded as generic prescribing in Northern Ireland. However, even after such drugs are excluded from the English figures, generic prescribing in Northern Ireland on average lags behind England by 17%. First wave fundholders had a generic prescribing rate of 31% in 1995, but this is still 13% lower than the rate in England. There is obviously room for improvement in generic prescribing in Northern Ireland.
Conclusions
We found that fundholding does affect prescribing in that the rate of increase in costs is reduced and the rate of generic prescribing is considerably increased. Each year's budget is reset to take account of changing factors in a practice from year to year such as, for example, pay and price increases. However, the amount left in the previous year's budget is deducted from this reset budget. This makes it increasingly difficult to make further changes each year and may also reduce the incentive to do so.
The savings that the practices can keep do not indicate what effect fundholding had on the overall regional drugs bill. Had fundholding not been introduced and costs continued to rise at the same rate as before, costs would have been higher than they actually were. So what savings did fundholders actually make? We calculated savings made in the first year of fundholding for each group by assuming that groups experienced the same percentage increase in costs in their first fundholding year as the previous year and by subtracting from this estimated cost the actual first year's costs. We estimate that the 23 first wave fundholders made savings of £1 267 748, the 34 second wave fundholders made savings of £1 476 394, and the nine third wave fundholders made savings of £451 903. This is a crude estimate and does not allow for increases in the cost of drugs (and so is probably an underestimate). However, it shows that general practitioners in Northern Ireland responded to the fundholding incentive and is clear evidence that fundholding alters the costs of prescribing.
Acknowledgments
Funding: The Drug Utilisation Research Unit is funded by the Department of Health and Social Services in Northern Ireland.
Conflict of interest: None.