Discussion

Pharmacist prescribing at the three hospital sites in this study is common practice, but little was known of the prevalence of prescribing by pharmacists and the prevalence of errors. This study has demonstrated hospital pharmacists prescribed for 39.8% of hospital inpatients, prescribing 12.9% of all medication orders on a single census day.

A mean prescribing error rate of 0.3% (medication orders) by pharmacist prescribers was demonstrated, which compares favourably with the EQUIP and PROTECT studies.21 ,22 However, it is recognised that this was a pragmatic study, and has a number of limitations. First, there may be bias due to selection of patients; the patients prescribed for were those present on the study weeks, so may not be representative of a standard group of patients, and it is not known whether other prescribers were initiating more complex prescriptions; the level of prescribing errors made by other prescribers was not evaluated, and may not have been as high as the EQUIP study. However, the results from the PROTECT study21 suggest that prescribing errors by doctors is a problem that is not just isolated to one part of the UK. It is also not known whether the results are generalisable to other hospitals. Nevertheless, there are few data on the prevalence and error rate of pharmacist prescribing, and this study does provide an indication of these parameters, which warrant further study in other institutions.

Results from parts 1 and 2 of this study show that the main area of prescribing was regular medicines not being prescribed following admission. This is where medicines taken by the patient prior to admission are omitted in error.29 This can potentially lead to patient harm from not taking medicines (eg, worsening of symptoms) or receiving interacting medicines, as clinicians are unaware of what the patient was taking. The EQUIP study reported that one in three prescriptions for regular medicines was omitted.22 In this aspect, this study demonstrates that prescribing pharmacists perform a key function in improving medicines reconciliation on admission to hospital and thus reducing the occurrence of unintentional omitted doses. This practice is not replicated in all hospitals, where pharmacists are still dependent upon medical staff—often junior—in order to complete full medicines reconciliation following admission.

It is also noteworthy that 13% of the medication orders prescribed by pharmacists were for new therapy, suggesting that pharmacists are not just using their prescribing rights to correct doctors’ errors but are actively managing patients. Further research is needed to identify differences, if any, between the type of medicines prescribed by pharmacists and doctors. However, this study does show that pharmacists are not focusing on a limited formulary of medicines but are prescribing from all but three sections of the BNF. This underlines one of the potential strengths of pharmacists as prescribers, in that their underpinning knowledge gives them a good understanding of a wide range of medicines in many clinical settings. This allows pharmacists to use their prescribing expertise in varied environments such as wards, where they may be expected to prescribe a range of treatments for a range of conditions. This differs from many models of nurse prescribing where the majority of prescribing is specialist, focusing on a narrower range of medicines or a clinic-based model of pharmacist prescribing often seen in primary care.2 In both parts of the study, a small proportion of prescribing activity was to stop medicines; stopping medicines (deprescribing) is an essential part of the prescribing pathway, preventing problematic polypharmacy, and can only be performed by prescribers.

The EQUIP study undertook regression analysis to identify errors between the type of prescription and stage of the hospital stay.22 Although these data were collected, statistical analysis was not appropriate as only four errors were detected. Of the four errors reported, three were potentially significant according to the EQUIP criteria and one had no impact on patient safety as the prescribing was not signed for. Two errors were for interacting drugs which had the potential to cause patient harm; increased risk of rhabdomyolysis. The final error was the substitution of oxycodone with morphine sulfate for breakthrough pain relief. This could have potentially led to insufficient pain relief. In all three cases, however, the patients came to no actual harm as errors were identified and corrected before they reached the patient.

A further potential limitation of part one of this study was that the data were collected over 2 months with each ward being visited once, rather than all data from all wards being collected at a single point in time. Our approach allowed the same researcher (OC) to undertake the data collection ensuring consistency of data collection. Finally, the prescribing pharmacists knew that the study was being undertaken and, thus, their prescribing accuracy may have potentially been subject to the Hawthorne effect.30

This study suggests that prescribing pharmacists can provide a valuable role in safely prescribing for a broad range of inpatients in UK general hospitals, especially at the point of admission where medicines are known to be frequently omitted by admitting medical staff.