It is estimated that 16 to 20 per cent of primary care prescribing is initiated in hospital and a further 40 per cent could also be strongly influenced by hospital prescribing.
However, the evidence base for this estimate is unclear. In view of the fact that the NHS spends in excess of £10bn on medicines each year and 80 per cent of this is in primary care, it is important that the influence each sector has on the others’ prescribing patterns is explored.
In 2006 the NHS Institute for Innovation and Improvement in collaboration with the Department of Health, developed a set of “Better care, better value” indicators. These indicators showed trust-by-trust performance across a number of key efficiency and productivity measures.
The indicators were aimed mainly at commissioners in primary care trusts and acute hospital providers. Information relating to these indicators has been updated every quarter since October 2006 and published on the NHS Institute for Innovation and Improvement website.
However, the prescribing indicator “Percentage of low cost statin prescribing” is only measured and published in terms of PCT performance. Since statins are prescribed in both primary and secondary care — as recommended by the National Institute for Health and Clinical Excellence — for the prevention of cardiovascular events in patients at increased risk of developing cardiovascular disease, or for those with established cardiovascular disease,
this group of medicines may be a useful marker to establish if there is a link between secondary care and primary care prescribing. Furthermore, investigating the possible influence of secondary care on primary care prescribing of statins, would also support the work of the DoH to build on the “Better care, better value” statin prescribing indicator, as recommended by the National Audit Office report “Prescribing costs in primary care”.
Similarly, two NICE clinical guidelines (CG5 and CG34) recommend that patients requiring a medicine affecting the renin angiotensin system should be treated with an angiotensin-converting enzyme (ACE) inhibitor first, and that an angiotensin-2 receptor antagonist (ARB) should be reserved for patients intolerant of ACEs.,
Therefore, this group of medicines may also be a useful marker to establish if there is a link between secondary care and primary care prescribing. Furthermore, in 2005/06 NHS West Midlands developed a range of proposals for change in PCT prescribing consistent with NICE CG5, and PCT pharmaceutical advisers were tasked with delivering this change. Subsequently in 2006, this NHS West Midlands proposal also had a positive impact on implementation of NICE CG34 in the West Midlands. Therefore, investigating the possible influence of secondary care on primary care prescribing of ACEs and ARBs would provide an indication of any influence of hospitals on the implementation of relevant NICE clinical guidelines for use of these drugs.
The purpose of this study was to examine how data on hospital drug usage from IMS Health can be used to examine the influence, if any, of hospital prescribing activity on primary care prescribing activity for statins and ACEs/ARBs.
Two models were used for our investigation: (i) the NHS Institute for Innovation and Improvement “Better care, better value” prescribing indicator for statin prescribing and (ii) the NHS West Midlands local indicator for the percentage of prescription items for ACE inhibitors in relation to the total number of prescription items for drugs affecting the renin-angiotensin system. Data on use of statins, ACE inhibitors and ARBs were extracted from the IMS Health dataset for secondary care and ePACT data set for primary care for a three-month period in 2007. The raw data were extracted as the number of tablets of each product. The number of tablets for low-cost statins compared with all other statins expressed as a percentage was then examined for their respective use in primary and secondary care. A similar analysis comparing the use of ACE inhibitors and ARBs in primary and secondary care was also undertaken.
The combined use of simvastatin and pravastatin (low cost statins) expressed as a percentage of all statin use across 17 PCTs in NHS West Midlands for the period from June to August 2007 was 74.2 per cent (95 per cent CI 71.9–76.5). Similarly, the combined use of simvastatin and pravastatin expressed as a percentage of all statin use across 17 acute hospitals in NHS West Midlands for the same period was 75.6 per cent (95 per cent CI 73.0–78.2). Although the mean proportion of combined simvastatin and pravastatin use in hospitals is higher than in primary care, the difference is not statistically significant as the 95 per cent confidence intervals overlap. This was confirmed by performing a t-test on the difference between the means of the two groups which showed no significant difference (P=0.385).
The combined ACE inhibitor use expressed as a percentage of all drugs affecting the renin-angiotensin system (ACEs+ARBs) across 17 PCTs in NHS West Midlands during the period from June to August 2007 was 71.7 per cent (95 per cent CI 70.5–72.8). The combined use of ACE inhibitors expressed as a percentage of all drugs affecting the renin-angiotensin system across 17 acute hospitals in NHSWest Midlands over the same period, was much higher at 79.0 per cent (95 per cent CI 77.3–80.7). This difference was statistically significant as the 95 per cent confidence intervals do not overlap. This was confirmed by t-test, which showed the means were statistically different (P<0.001).
Where there was statistically significant difference between primary and secondary care a further analysis was undertaken. In the six instances where it was known that more than 80 per cent of a hospital’s contract activity was with its host PCT, the hospital use of angiotensin inhibitors expressed as a percent-age of ACE inhibitors against all drugs affecting the renin-angiotensin system was paired with its host PCT use. Linear regression analysis of these paired data for these six health economies showed no correlation between the hospital and PCT pattern of use of angiotensin inhibitors (R2=0.128).
It is clear from the results of this study that primary and secondary care prescribing of statins is similar. However, the volume of use of low cost statins (simvastatin and pravastatin) compared with all other statins is consistently higher in secondary care (West Midlands mean ratio 75.6 per cent) compared with primary care (West Midlands mean ratio 74.2 per cent). Although this is not statistically different, it is reasonable to conclude that secondary care is not adversely impacting on primary care reaching its “Better care, better value” indicator target for appropriate statin prescribing. Indeed it appears that secondary care prescribing of statins is more consistent with NICE guidance
However, analysis of prescribing of medicines affecting the renin-angiotensin system demonstrates an even wider diversity in prescribing patterns between primary and secondary care. The mean ratio of use of ACE inhibitors as a percentage of all drugs affecting the renin-angiotensin system is statistically higher in secondary care (West Midlands mean ratio 79.0 per cent ACEs to 21.0 per cent ARBs) than it is in primary care (West Midlands mean ratio 71.7 per cent ACEs to 28.3 per cent ARBs). Given the different population of patients and that secondary care tends to deal with more complex patients it would be expected that secondary care might actually have a higher use of ARBs to ACEs. This is not what we found and since this difference is statistically significant (P<0.001) it would suggest that secondary care prescribing of ACEs and ARBs is more consistent with NICE guidance in relation to use of drugs affecting the renin-angiotensin system.,
Furthermore, the sub-group analysis of six health economies confirms there is no relationship between primary and secondary care prescribing for this group of medicines.
Therefore, the results of this study in these two groups of medicines do not support the suggestion in the National Audit Office report that secondary care significantly influences primary care prescribing.
This is not surprising since no literature evidence is quoted in the report. Furthermore, the groups of medicines investigated in this study are not specialist medicines and therefore would normally be expected to be routinely initiated in primary care rather than in hospital. In addition, previously the NHS West Midlands secondary care pharmaceutical adviser, in line with the NHS West Midlands primary care strategy, had encouraged hospitals to initiate ACE inhibitors in preference to ARBs where clinically appropriate.
Hospital prescribing data from IMS Health have limitations in that they do not include stock provided via clinical trials, FP10(HP) prescriptions, sample stock, or patients’ own medicines. Although this may have some minimal effect on the raw volumes of medicines used, by expressing the data as a proportion rather than as an absolute value these confounding factors are removed. In addition, given the nature of these medicines and the associated volume of prescribing, this limitation, if it occurs at all, is likely to be insignificant. Another confounding factor may be a hospital’s size and its level of activity.
However, by using proportional use of specific medicines within a class rather than raw usage data, it is possible to compensate for this.
Although we accept the limitations of using hospital prescribing activity data from IMS Health, the use of percentages as a comparison of drug use within a class such as statins is the model chosen in the “Better care, better value” indicators. It is not unreasonable, therefore, to use percentages as a comparator when comparing use of ACEs and ARBs.
The results of this study on two groups of medicines widely prescribed in primary care and hospitals demonstrate clearly that while prescribing patterns are similar in both sectors, hospital prescribing is not adversely influencing primary care prescribing in the context of “Better care, better value” indicators, NHS West Midlands primary care prescribing proposals, and implementation of NICE CG5 and CG34. Indeed for initiation of ACEs in preference to ARBs for new patients, as proposed in NHS West Midlands primary care prescribing targets and NICE CG5 and CG34, hospital prescribing is better than primary care prescribing in NHS West Midlands. This will prove helpful to NHS West Midlands given that increasing the use of low cost drugs affecting the renin-angiotensin system as a proportion of all such drugs is now an indicator of prescribing performance in the “Better care, better value” indicators.
About the authors
R. G. Pate, MCPP, FRPharmS, is secondary care pharmaceutical adviser at Keele University
R. W. Fitzpatrick, PhD, FRPharmS, is clinical director of pharmacy at Royal Wolverhampton Hospitals and professor of pharmacy at Wolverhampton University
P. Woodvine, BSc, is data analyst in the department of medicines management at Keele University.
In collaboration with West Midlands hospital chief pharmacists.
Correspondence to: Ron Pate, Department of Medicines Management, Keele University, Keele, Staffordshire ST5 5BG
 National Audit Office. Prescribing costs in primary care. London: The Stationery Office, 2007.
 NHS Institute for Innovation and Improvement. Better care, better value indicators. Available at: www.productivity.nhs.uk/default.aspx (accessed 6 April 2009).
 National Institute for Health and Clinical Excellence. Statins for the prevention of cardiovascular events in patients at increased risk of developing cardiovascular disease or those with established cardiovascular disease (TA 94). London: NICE, 2006.
 National Institute for Health and Clinical Excellence. Clinical Guideline 5. Management of chronic heart failure in adults in primary and secondary care. London: NICE, 2003.
 National Institute for Health and Clinical Excellence. Clinical Guideline 34. Hypertension: management of hypertension in adults in primary care. London: NICE, 2006.
 Pate RG, Fitzpatrick RW, Woodvine P. Using comparators in secondary care prescribing activity to change drug use. The Pharmaceutical Journal 2009;283:597–8.