Ray Fitzpatrick, BSc (Pharm), PhD, FRPharmS
Clinical Director of Pharmacy,
Royal Wolverhampton Hospitals,
and Professor of Pharmacy,
In 2009, the NHS in England spent £12.3 billion on medicines. Although the majority of this was in primary care, a significant proportion (30.9%) was spent in hospitals.1 Furthermore, hospital medicines expenditure grew at a higher rate (13.2%) than in primary care (2.6%).1 Detailed information on medicines use has been available within most hospitals for a number of years and has been demonstrated to help promote safe rational and cost-effective prescribing within hospitals.2 However, hitherto, it has not been possible to compare individual hospital trust’s medicines usage. This is in marked contrast to primary care, where Practice Detailed Prescribing Information (PDPI) data, previously termed PACT data, has been available since the mid-1990s.
The reason for this difference is because there is no readily accessible NHS national database of hospital medicines use, whereas, in England, all prescriptions dispensed in community pharmacies are processed for payment centrally through the NHS business service authority. This has enabled the development of a national database of general practitioner prescribing. The Healthcare Commission in 2007 in its national report on medicines management in acute and specialist trusts recognised the need to benchmark hospital prescribing and called for the development of hospital prescribing indicators.3 Over a decade ago in the UK, the National Prescribing Centre led a project to collect medicines’ use data electronically, but encountered difficulties due to variations in the pharmacy computer systems used and in the way different hospital pharmacies counted prescribed items. In the end, only 16 trusts participated, but the project demonstrated that it was possible to obtain useful comparative data on medicines’ use in hospitals.
However, it also concluded that there was no ‘quick-fix’ solution to the development of a comprehensive national NHS database of prescribing in secondary care.4 There is now a pressing need to be able to compare hospitals’ medicines use for a number of reasons including finance, patient safety and clinical effectiveness.
As stated earlier, hospital expenditure on medicines now represents over 30% all NHS medicines expenditure and is believed to influence a significant proportion of prescribing in primary care. As a result of the downturn in the global economy and subsequently the recession in the UK, the NHS has to save £20 billion over the next three years.
As medicines represent 15% of hospitals non-pay budget, this element of hospital costs is already under close scrutiny by NHS managers. Indeed, one of the key strategies to deliver the necessary savings is the NHS Quality, Innovation, Productivity and Prevention agenda (QIPP), and a number of therapeutic areas have been identified nationally as part of this agenda.5
Ensuring patient safety is one of the central roles of a pharmacist. A recent study sponsored by the UK’s General Medical Council showed that up to 9% of hospital prescriptions contain errors, but the majority of these are prevented from reaching the patient by the intervention of a pharmacist.6 Medication-related incidents are the third most frequently reported clinical incident. The National Patient Safety Agency (NPSA) has produced almost 70 alerts highlighting safety issues and, not surprisingly, a third of these are medicines-related. Clearly, having some way of comparing medicines’ use between hospitals would help hospitals better understand their use of high-risk medicines.
There are a plethora of guidelines on the use of medicines for specific indications. In the UK, the National Institute of Health and Clinical Effectiveness (NICE) was established in 1999 to provide definitive guidance on the use of particular medicines – usually high-cost medicines – in particular diseases. These guidance documents fall into two categories, clinical guidelines and technology assessment guidelines (TAGs).
It is mandatory on health economies in England to implement the latter. However, it is very difficult for healthcare managers to know how well NICE guidance is being implemented other than by asking. Having readily accessible data on the use of NICE-approved medicines would help hospitals understand how well this guidance is being implemented.
Developing suitable indicators
In the UK, two datasets exist which could be harnessed to allow comparison of hospitals medicines use. The first is information collected by the NHS pharmaceutical and supplies authority (PASA) on medicines usage in hospitals, which informs the NHS medicines contracting process. The second is a commercial dataset of medicines’ use collected by IMS Health, which covers approximately 97% of English hospitals.
This latter data is now starting to be used to compare hospitals medicines’ use. However, even when data is available, comparing hospitals of different size, activity and case mix is a challenge,7 and a number of different indicators have had to be developed to compensate for the wide differences in hospitals.
Proportionality of use within a class
This is probably the easiest way of compensating for hospital size and activity. It involves expressing the use of a particular medicine within a class of medicines as a proportion of the total use of medicines within the class.
The units can be numbers of tablets or capsules used, provided the number in a daily dose is the same. If these are different, then units used should be the defined daily dose (DDD) for the particular medicines, which can be obtained from the World Health Organization ATC index at www.whocc. no/atc-ddd-index.
Proportionality of use has been used to compare hospitals use of different formulations of the proton pump inhibitor lansoprazole in order to support change.8 In this paper, the authors compared the use of lansoprazole tablets and capsules between all acute hospitals in one NHS region in England. The results showed that four trusts were using significantly more of the tablet formulation than capsules (see Figure 1). This was because a special hospital price had been negotiated in those particular hospitals.
However, in primary care – where the patients would be maintained on therapy after discharge – the basic NHS price of the tablets was much greater than the capsules, and as the tablet formulation was dissolvable in the mouth and pleasanttasting, it was likely that patients would request the same formulation from their primary care doctor. Once the relevant hospital chief pharmacists were shown the data, they took action to change. As a result, the usage pattern of the different formulations had changed significantly when re-examined six months later (see Figure 2). This is a good example of using an indicator to compare hospitals medicine use to support more costeffective prescribing. However, using proportionality as an indicator only works when comparing similar medicines within a therapeutic class or, as in the case above, different formulations of the same medicine.
Measures of hospital activity
Where there is only a single medicine within a class, it becomes more complex to compensate for hospitals’ differences in size, activity and case mix. In these circumstances, it is necessary to use some measure of hospital activity as a denominator to medicines’ use – for example, Finished Consultant Episode (FCE). Using DDD/FCE has been demonstrated to be a valid indicator when comparing hospitals use of antibiotics.9
In the UK, FCE is only one of a number of measures of hospital activity, which can be used; alternatives could include number of admissions or occupied bed days. Indeed, occupied bed day is the denominator used in European antibiotic surveillance initiatives. In the UK, individual hospital activity data can be accessed from the Hospital Episode Statistics website www.hesonline.nhs.uk.
While general activity data can compensate for hospital size and activity, it does not compensate for differences in hospitals case mix. This is less of a problem when using proportionality within a class to compare hospitals, as the therapeutic class of the medicine tends to be used in similar groups of patients. Where this is not the case, or where there is only one medicine in a class and case mix is potentially a significant variable, then some disease-specific activity indicator is needed as a denominator with medicines use. Oncology medicines are a one such group, since there is often only one in a class or the medicine is used for a number of different cancers. A good example of the former is trastuzumab for the treatment of breast cancer.
The author has undertaken some initial research to examine the possibility of comparing use of this medicine between different NHS regions within England. To do this, data on the use of trastuzumab was obtained from the Health and Social Care Information centre for the 28 strategic health authorities (SHA) in England in 2005- 06. As there is no DDD for trastuzumab, the raw usage data was converted into average weekly doses (AWD). Figure 3 shows the use by SHA. As can be seen, there is significant variation across all SHAs in England. However, if this usage data is divided by the number of FCEs for breast cancer by SHA for the same period, then this variation is reduced as a result of compensating for disease treatment activity (see Figure 4).
It can be noted that three SHAs still show very high usage of trastuzumab. However, all three host specific cancer hospitals where the use of trastuzumab would be expected to be high.
The average usage of trastuzumab across all SHAs in this analysis was 0.76 AWD/FCE of breast cancer and, as well as the SHAs with high usage, there are a few SHAs with apparently very low usage, which warrants further investigation. This sort of macro-level analysis could be used as an initial assessment of the implementation of NICE guidelines on this medicine.
There is a clear need to be able to compare hospitals medicines’ use to support cost-effective use of medicines. Using raw usage data alone is not appropriate, as hospitals vary in size, activity and case mix. However, indicators have been developed which can compensate for these variables, including using proportionality of a specific medicine within a class, DDD/FCE or other activity indicator, and DDD/ disease-specific activity indicator. The indicator used will depend on the specific medicine and its indications for use.
Such indicators will become increasingly important in the future as health systems come under increasing pressure to deliver better outcomes for the same or less resource. Having some way of promoting better use of resources, delivery of safer medicines’ use, and assessing the implementation of bestpractice guidelines such as those from NICE will prove invaluable.
1. NHS Information Centre. Hospital Prescribing, England: 2009. Health and Social Care Information Centre, Leeds: October 2010
2. Fitzpatrick RW et al. Pharmaceutical Journal 2001;266:588-89
3. The Best Medicine. The Management of Medicines in Acute and Specialist Trusts, ISBN 1-84562-124-7. Published by the Healthcare Commission, January 2007
4. Walker D. Prescribing information in secondary care – the value of a national database. Pharm 2000;264:262-5
5. National Prescribing Centre. Key Therapeutic Topics 2010/11 – Medicines Management Options for Local Implementation, National Prescribing Centre Liverpool UK: July 2010
6. Dornan T et al. The EQUIP Study. Final report to the GMCl, December 2009
7. Fitzpatrick RW & Pate RG. Health Services Journal 2008;13 November:28-29
8. Pate RG et al. Pharmaceutical Journal 2009;283:597-98
9. Fitzpatrick RW & Edwards CM. Pharm. World. Sci. 2008;30:73-78