Comparing acute trusts' prescribing activity is a valuable if difficult exercise, say Ray Fitzpatrick and Ron Pate
In 2005, NHS England spent more than£10bn on medicines. Per head of population their use grows by 5 per cent every year. While hospital prescribing makes up just a fifth of the total NHS medicines bill, it is still a financial risk. With four of every five inpatients on four or more medicines it also represents a significant clinical risk. Much primary care prescribing also appears to be strongly influenced by what gets written up in hospital.
But there is no national database of hospital medicines use for comparing individual trusts' performance. Attempts in 1998 by the National Prescribing Centre to electronically capture medicines usage data encountered difficulties, but with the participation of 16 trusts the project did demonstrate that it was possible to obtain some useful figures. It also concluded that there was no quick way to develop a comprehensive national database of prescribing in secondary care.
In 2006 the Healthcare Commission reviewed the medicines management systems of 173 acute trusts and was able to provide a benchmark of prescribing costs to patient. Using costs alone, however, is a crude indicator of prescribing.
Although the standard tariff allows hospitals to compare their overall costs for particular procedures with national averages, there is no way of comparing medicines use within those treatments. With the rise of medicines expenditure in hospitals exceeding that in primary care in recent years, hospital managers and commissioners need assurance that patients receive the most cost-effective treatment. Primary care trusts will also want to know that treatments in hospital are not adversely influencing prescribing in primary care, and that National Institute for Health and Clinical Excellence guidance is implemented.
Recognising the need to benchmark hospital prescribing, the Healthcare Commission recently invited proposals for the development of hospital prescribing indicators in a letter to strategic health authority chief executives.
But what should the appropriate indicators be?
Cost: used as a method recently by the Healthcare Commission, simply looking at the money spent on a particular medicine over one year is a crude means of comparing hospital prescribing. The main drawback is the lack of standard prices for medicines in hospital. The prescription price regulation scheme sets the basic cost of NHS medicines, but as a result of regional or local contracts with suppliers and wholesaler discounts, hospitals pay less than this for many generic and branded medicines.
Usage data: use of a medicine over a period of time could be compared. The main problem is determining a common unit of use. There are a variety of hospital pharmacy computer systems counting stock in different ways (full packs, outers of packs, etc) and applying conversion factors when tracking stock issued. Different dosage regimens only confound things further.
Defined daily doses: using the standard dose of a medicine for one patient for one day as a measure of medicines use overcomes problems of variable pack sizes and doses. By taking a hospital's use of a medicine expressed by weight (usually grams) and dividing this by the medicine's defined daily dose, also expressed in weight, it is possible to calculate the number of defined daily doses of a particular medicine used over a defined period. Within this standardised approach there are fewer confounding variables but, unless the hospitals being compared are of a similar size and case-mix, it is extremely difficult to draw any real conclusions using this indicator alone.
Dividing defined daily dose by bed days: an alternative that normalises for size and activity is to divide the number of defined daily doses of a medicine by bed days. However, this approach does not take into account case-mix, since some hospitals may be regional or national tertiary referral centres for particular diseases - one reason why drug expenditure in large teaching hospitals is greater than in district general hospitals. Another drawback is that hospital activity data has to be accessed.
Where there are a number of similar medicines in a particular class or therapeutic group (for example statins), then measuring the proportion of one medicine's use against the rest of the group is an alternative to defined daily dose/bed days. This approach reduces the delay associated with accessing hospital activity data. But this proportionality approach only works where there are a number of choices of medicines within a therapeutic class or group with different degrees of cost-effectiveness. A disadvantage to this proportionality indicator is that it cannot take into account case-mix differences between hospitals. Furthermore it cannot be used where there are no other medicines in the class.
Defined daily dose divided by finished consultant episode: the approach of dividing the number of defined daily doses of a particular medicine by the number of finished consultant episodes over the same period is similar to the defined daily dose/bed day as an indicator of prescribing. The finished consultant episode is an accepted way of measuring hospital activity in the UK.
Although using it has no real advantages over defined daily dose/bed day, as neither takes into account the case-mix of patients, this method offers the possibility of normalising for case mix by using disease-specific finished consultant episodes.
The main problem with all these approaches is that data has to be collected directly from individual hospitals. If any of these are to be used routinely to compare hospital prescribing, we need a single database of hospital medicines use.
International Medication Systems - an independent commercial company supplying information to a variety of customers, including the pharmaceutical industry - is probably the largest worldwide repository of medicines usage data. The NHS Information Centre is developing agreements with International Medication Systems to access this data and we are actively engaged with them in developing comparison tools.
Any measure of benchmarking different hospitals' prescribing, however, must be presented in a supportive, rather than performance management way. Such information can be a powerful tool for change within a trust, but must have buy-in from hospital chief pharmacists to deliver change.