What can PACT tell us about prescribing in general practice?
BMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7121.1515 (Published 06 December 1997) Cite this as: BMJ 1997;315:1515
- Azeem Majeed, senior lecturer in general practice (a.majeed{at}sghms.ac.uk)a,
- Norman Evans, pharmaceutical adviserb,
- Paula Head, head of pharmacyc
- a Division of General Practice and Primary Care, St George's Hospital Medical School, London SW17 0RE
- b Merton, Sutton and Wandsworth Health Authority, Wilson Hospital, London CR4 4TP
- c Kingston&Richmond Health Authority, 22 Hollyfield Road, Surbiton KT5 9AL
- Correspondence to: Dr Majeed
- Accepted 22 August 1997
Introduction
Prescribing is important to general practitioners. Most consultations with a general practitioner result in a prescription being issued, and in 1995 general practitioners in the United Kingdom issued about 550 million prescriptions. The total cost of these prescriptions was £4700m, 11% of all NHS spending in 1995. Unsurprisingly, the general practice prescribing budget is seen by many managers and health economists as an area where there is scope for considerable savings through more cost effective prescribing.1 2 Hence, general practitioners are likely to be increasingly called on to defend their prescribing patterns.
Summary points
PACT (prescribing analyses and cost) data are derived from prescriptions issued by general practitioners
PACT data have been used by health authorities as a managerial tool, by health services researchers, and, more recently, by general practitioners to audit and to help improve their prescribing
The recent availability of electronic PACT data on the prescribing of individual drugs has allowed health authorities to look at prescribing in more detail and to develop prescribing indicators
To make more effective use of PACT, general practitioners who wish to examine their prescribing in more detail should be given access to electronic PACT data and be trained in its analysis and interpretation
In England the main source of information on general practitioners' prescribing is PACT (prescribing analyses and cost), and similar schemes exist in Scotland and Wales. Although PACT was first introduced in 1988, many general practitioners have made little use of PACT data. Because of the importance and potential value of such data, general practitioners need to be aware of the information available via PACT and how this information is used by health authorities' medical and pharmaceutical advisers. Moreover, as more general practitioners become involved in commissioning health services, either as fundholders or through local commissioning, they will need to make greater use of PACT data to audit their prescribing and help ensure that their prescribing is economical.
What information is available
The Prescription Prescribing Authority (PPA) collects information on all prescriptions issued by general practitioners that are dispensed by community pharmacists, dispensing general practitioners, or appliance contractors. The information collected includes the name and cost of the drug and the number of items dispensed (an item is defined as each preparation on the prescription). The drugs dispensed are then used to calculate the cost of each item, and the information is entered onto computer by the PPA. Drugs are categorised by the section of the British National Formulary that they fall in. Hence, information is available for individual drugs (such as salbutamol), for categories of drugs (such as bronchodilators), or for therapeutic areas (such as respiratory drugs). This information is available at individual practice level, health authority level, and national level, allowing different analyses (box box).
Box 1: Definitions of key terms
PACT—Prescribing analyses and cost. Information on general practitioners' prescribing obtained from prescriptions dispensed by community pharmacists, dispensing general practitioners, and appliance contractors
PPA—Prescription Prescribing Authority. Located in Newcastle, the PPA is the source of PACT data
PUs—Prescribing units. A measure of patients' needs for prescribed drugs weighted for age; patients aged under 65 years count as one unit, patients aged 65 and over count as three units
ASTRO-PUs—Age, sex, and temporary resident originated prescribing units. A more sophisticated measure of patients' needs for prescribed drugs than prescribing units
STAR-PUs—Specific therapeutic group age-sex related prescribing units. Measures of patients' needs for prescribed drugs in specific therapeutic areas. These are not yet in widespread use
Defined daily doses—A measure of the amount of drug prescribed based on standard therapeutic units. The number of defined daily doses is calculated by dividing the total amount of a drug prescribed by the defined daily dose for the drug. For example, the defined daily dose of ranitidine is 300 mg. If a practice prescribed a total of 186 000 mg of ranitidine, this would equal 620 defined daily doses (186 000/300)
PACTLINE—An electronic link between health authorities and the PPA which allows health authorities to download computerised prescribing data about the prescribing of their general practices
HEAPACT—Health authority electronic PACT. An enhanced version of PACTLINE which allows health authorities to obtain computerised information about the prescribing of individual drugs
EPACT—Electronic PACT. A unified system that will replace PACTLINE and HEAPACT
Box 2: Units of analysis and levels of PACT data
Unit of analysis
England
Health authority
General practice
Level of analysis
Total prescribing, prescribing by British National Formulary chapters
Prescribing by subsections of British National Formulary chapters
Prescribing of individual drugs
PACT includes information on prescribing costs, the number of items prescribed, and generic prescribing. This information can be looked at using different units of analysis or different levels of prescribing data. General practitioners will be mainly interested in their own practice's PACT data and how their prescribing compares with other practices in their health authority. Health authorities will want to know how they compare with other health authorities and about the prescribing of their general practices. The NHS Executive will want to know about prescribing at national and health authority levels.
After analysis, the data are fed back to both health authorities and general practitioners. General practitioners generally only receive a printed version of the analysis. This information (the standard PACT report) contains information on the practice's rates and costs of prescribing along with comparative information. General practitioners can ask for more detailed information but this more detailed report is very unwieldy. By contrast, health authorities receive both printed reports on general practices' prescribing and also have access to computerised PACT data (PACTLINE). Recently, the PPA has also developed HEAPACT (health authority electronic PACT). This allows health authorities' pharmaceutical advisers to obtain information on the prescribing of specific drugs directly from the PPA via a modem link. With HEAPACT, health authorities can look at prescribing in great detail. For example, the prescribing of new classes of drugs such as selective serotonin reuptake inhibitors can be examined to determine which general practices are the first to make major use of them.
How is PACT used
Traditionally, PACT has been mainly used as a financial tool to help health authorities set and monitor general practice prescribing budgets. However, PACT is now increasingly used for other purposes, including audit and research, improved methods of funding high cost drugs, and the development of practice formularies. Because prescribing is heavily influenced by general practices' demography, weightings for age and sex have been developed so that rates and costs of prescribing in different practices or health authorities can be compared.3 4
Budget setting
Health authority budgets
The NHS Executive has recently begun to use PACT data to develop a formula for allocating prescribing budgets to health authorities.5 Variables were used to develop a statistical model that could be used to explain variations in prescribing costs between health authorities. Variables that were found to be associated with prescribing costs included the age and sex weighted population (measured with ASTRO-PUs), cross boundary flows (patients living in one health authority but registered with a general practice in another health authority), and the percentage of people in a health authority with permanent sickness (obtained from the 1991 census). Health authorities that were more than 2% below their predicted spending were given a larger increase in their prescribing budget than other health authorities. The NHS Executive is now encouraging health authorities to consider using similar methods when they in turn allocate budgets to practices.
General practice budgets
PACT data are used by health authorities to help set and monitor general practices' prescribing budgets. For many years, these budgets were based on prescribing units, but most health authorities now use ASTRO-PUs as these provide better weightings for age and sex than do prescribing units. Because there is no valid capitation based formula for allocating prescribing budgets to practices, health authorities have to use their judgment when allocating budgets. In effect, this means that practices with a good grasp of their past and current PACT data are likely to be able to argue a better case than other practices and hence obtain a larger prescribing budget.
Health services research
PACT data have been widely used by researchers to investigate variations and trends in prescribing costs.6 For example, studies that have examined differences in prescribing costs between fundholding and non-fundholding practices have generally used PACT data.7 8 Studies of the management of conditions such as asthma, comparing admission rates with prescribing rates, have also generally used PACT data.9
Prescribing indicators
A relatively new development is to use PACT data to develop prescribing indicators. Many health authorities now do this, often in consultation with general practitioners, and a large number of indicators have been developed (box box).10 The assumption is that these indicators can be used to measure the quality or prescribing in general practice. Some indicators, such as the ratio of inhaled corticosteroids to inhaled bronchodilators, are considered to measure good prescribing while other indicators such as the rate of prescribing of appetite suppressants are considered to measure bad prescribing.
Box 3: Prescribing indicators commonly used by health authorities
Markers of good prescribing
Rate of generic prescribing
Frusemide and bendrofluazide as a percentage of diuretic drugs
Ratio of bendrofluazide 2.5 mg to bendrofluazide 5 mg
Atenolol and propranolol as a percentage of ß blockers
Ratio of inhaled corticosteroids and cromoglycate to inhaled bronchodilators
Cimetidine as a percentage of H2 receptor antagonists
Generic antibiotics as a percentage of all antibiotics
Markers of bad prescribing*
Diuretic potassium combinations
Cerebral and peripheral vasodilators
Appetite suppressant
Topical non-steroidal anti-inflammatory drugs
Benzodiazepines
Cough mixtures and decongestants
Health authorities currently use many different indicators, but they are largely unvalidated and an important subject for further research will be to see if these indicators are valid measures of the quality of general practitioners' prescribing. Much of the research in this subject has tried to validate indicators by examining their association with other process measures. For example, several researchers have looked at the association between the ratio of inhaled corticosteroids to inhaled bronchodilators and admission rates for asthma.11 12 13 The next step in this research will be to validate the indicators against better quality measures.
Current NHS information systems make it straightforward for health authorities to produce a large number of prescribing indicators, and care is needed in deciding how indicators are selected and used. Ideally, prescribing indicators should be used to improve the quality and effectiveness of prescribing in general practice and not simply to reduce prescribing costs.14 Furthermore, like all indicators of performance, prescribing indicators can create perverse incentives, and it would be fairly straightforward for practices to improve their performance as measured by many of the indicators without improving their clinical management of patients.15 For example, a practice with a low ratio of inhaled corticosteroids to inhaled bronchodilators could improve its ratio by indiscriminately increasing its prescribing of inhaled corticosteroids without improving other aspects of patient management such as patient education or adequate record keeping.
Despite their limitations, however, prescribing indicators do offer potential benefits to general practitioners. For example, general practitioners can use them to set and monitor their practice's progress towards a target rate of prescribing generic drugs.
Identifying expensive drugs
Patients who need expensive drugs can substantially increase a practice's prescribing costs. Some general practitioners may therefore be concerned that having too many patients who need expensive drugs on their list will prevent their practice from remaining within its prescribing budget. With PACT data, however, it is possible to separate the costs of expensive drugs from other prescribing costs. This information can then be used to adjust general practices' prescribing budgets to take into account the cost of these drugs.
This approach has now been adopted by many health authorities. For example, Kingston and Richmond Health Authority has used PACT data in combination with notifications from general practices to identify patients taking any drug costing more than £1500 a year (table 1). Patients who need a drug that costs more than £2000 a year have the cost of the drug met from a special health authority budget for expensive drugs. One byproduct of this is that the health authority has been able to produce a database of patients who need expensive drugs including information on diagnosis. This information can then be fed back to local practices to help them manage their prescribing budgets and to plan the clinical management of these patients. The register also has other potential uses: for example, it can be used to identify patients with specific conditions and hence can be used to identify samples of patients for audit and research.
Characteristics of 27 general practitioners, analysed with respect to their average rates of recording of clinical questions during atient consultations (n=85)
At present, it is much easier to identify expensive drugs than to identify expensive patients, as high cost drugs can be identified through PACT data. Unfortunately, patients who require multiple low cost drugs and who therefore have high overall drug costs cannot be identified through the PACT system. This is because PACT can not yet provide patient based data.
Analysing prescribing in therapeutic areas
The recent development of HEAPACT (health authority electronic PACT) allows health authority medical and pharmaceutical advisers to download from the PPA detailed information on the prescribing of general practices. Through HEAPACT, information is available on the prescribing of individual drugs. This offers both health authorities and general practitioners a number of opportunities. For example, it is possible to examine prescribing in specific therapeutic areas in order to develop practice formularies, rationalise prescribing, and cut prescribing costs (table 2). HEAPACT also allows the identification of practices that are prescribing new, expensive drugs that cost more than older drugs of similar efficacy or drugs that are known to be ineffective. Although some practices now have access to and are using data derived from HEAPACT, many practices are not aware of this development or of the benefits it could offer their practices (see box box).
Sources of information used by 27 general practitioners to pursue answers to clinical questions during and after patients'consultations, and sources stated as most often used (15 of the 85 questions raised were not pursued by doctors)
Box 4: Potential uses of level 3 PACT data to general practitioners
Provide baseline data on prescribing to help in preparation of a practice formulary
Determine what percentage of drugs prescribed are in the practice formulary
Monitor progress towards prescribing targets
Identify areas in which practice prescribing differs greatly from health authority average
Identify therapeutic areas in which greater use of generic drugs is possible
Identify areas where there is scope for therapeutic substitution—for example, using frusemide and bendrofluazide in place of more expensive diuretics
Identify prescribing of drugs of limited therapeutic value such as cough mixtures and decongestants
Help standardise the quantity of drugs prescribed—for example, for 28 days versus 30 days
Allow calculation of defined daily doses instead of the number of items as a measure of the amount of a drug prescribed
Limitations of PACT
Although PACT data have many uses, they do have some important limitations, the most important being that the data provide only a narrow range of information, mainly on what drugs were prescribed and how much the prescribed drugs cost. Secondly, the data cannot be linked to demographic or clinical data on patients. Hence, they cannot be used to calculate age and sex specific prescribing rates or to look at prescribing rates for specific conditions. This is why ASTRO-PUs and STAR-PUs were derived from the computerised prescribing records of general practices and not from PACT data. Thirdly, because they are based on dispensed NHS prescriptions, they do not include private prescriptions or prescriptions that a patient does not have dispensed.16 Fourthly, the number of items prescribed is not always an accurate measure of the amount of a drug actually prescribed.17 Defined daily doses (which can be calculated from PACT data) can be used to overcome this problem and provide a more accurate measure of the amount of a drug prescribed than the number of items. Finally, PACT tells us only about the prescribing carried out in general practice and does not contain any information on prescribing in hospitals.
Making PACT more accessible
Most general practitioners currently receive their PACT data from the PPA in the form of a booklet. Many health authorities are also producing prescribing indicators for feedback to their general practitioners, and in some cases these indicators are distributed in electronic format. In some areas general practitioners have also been given access to their entire PACT data in electronic format. The next step in making PACT data more accessible will be to extend this initiative and to provide PACT data in electronic format to all practices that want this. However, to make effective use of electronic PACT data, general practitioners will need training in how to analyse, interpret, and use PACT data appropriately.
Improving PACT
There are two developments that would substantially improve the usefulness of PACT data. The first would be to include a unique patient identifier (such as the NHS number) on the prescription. This would allow prescriptions to be linked to individual patients, leading to patient based PACT data and the calculation of age and sex specific prescribing rates. It would also allow general practices and health authorities to identify patients who have high drug costs and thus lead to more reliable setting of prescribing budgets. The use of a unique patient identifier would also allow PACT data to be linked to other data sets, such as data on hospital admissions.
The other major development that would improve PACT would be to include diagnostic data (perhaps in the form of a Read code) on the prescription. This would allow prescribing for specific clinical conditions to be analysed. Because many drugs, such as ß blockers and angiotensin converting enzyme inhibitors, have more than one indication, PACT data cannot currently be used for this.
Neither of these developments would be easy to implement and, realistically, would not be feasible until all general practices are fully computerised and issue virtually all prescriptions with their practice computers. Including clinical data on the prescription form would also raise issues about confidentiality, and this might prevent the implementation of this measure. However, as well as improving the usefulness of PACT data, these developments, especially the use of a unique patient identifier, could lead to a lower level of prescription fraud and thus yield some financial savings.