- Top of page
- Competing Interests
WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT
• 8.9% medication orders written by doctors contain prescribing errors, which are made by all grades of doctor from foundation year 1 to consultant.
• Recommendations to improve prescribing safety include continuing education from medical school to retirement, as well as improvement of NHS systems.
• Priority areas for education in prescribing with potential to have impact on prescribing errors have not previously been identified.
• This study has identified 100 drugs most commonly prescribed in primary and secondary care and shown that these drugs are those most frequently associated with prescribing errors.
• This core drug list has potential to direct prescribing training for both undergraduate and postgraduate prescribers as part of a national strategy to reduce prescribing errors.
AIM To develop a core list of 100 commonly prescribed drugs to support prescribing education.
METHODS A retrospective analysis of prescribing data from primary care in England (2006 and 2008) and from two London Teaching Hospitals (2007 and 2009) was performed. A survey of prescribing by foundation year 1 (FY1) doctors in 39 NHS Trusts across London was carried out.
RESULTS A core list of 100 commonly prescribed drugs comprising ≥0.1% prescriptions in primary and/or secondary care was developed in 2006/7. The core list remained stable over 2 years. FY1 doctors prescribed 65% drugs on the list at least monthly. Seventy-six% of FY1 doctors did not regularly prescribe any drugs not on the core list. There was a strong correlation between prescribing frequency (prescriptions for each drug class expressed as percentage of all prescriptions written) and error rate described in the EQUIP study (errors made when prescribing each drug class expressed as a percentage of all errors made), n= 39, r= 0.861, P= 0.000.
CONCLUSIONS Our core drug list identifies drugs that are commonly used and associated with error and is stable over at least 2 years. This list can now be used to develop learning resources and training programmes to improve prescribing of drugs in regular use. Complementary skills required for prescribing less familiar drugs must be developed in parallel. Ongoing research is required to monitor the effect of new training initiatives on prescribing error and patient safety.
- Top of page
- Competing Interests
Prescribing errors are common and have important implications for patient safety. In a recent study (EQUIP) commissioned by the General Medical Council, researchers in the North of England identified errors in 8.9% of medication orders written . Errors were made by all grades of doctor, although less commonly by consultants (5.9%) than by foundation year 1 (FY1) (8.4%) or FY2 (10.3%) doctors. In an associated systematic review of 16 separate studies, error rates were 7% (interquartile range 2–14) of medication orders .
Factors determining prescribing errors are extremely complex. Characteristics of the clinical environment, prescribing culture, workload, time and support all have an important impact and need to be addressed to improve patient safety [1, 3]. On an individual level, prescribing errors are attributable to lack of training in practical prescribing and failures to link theory with practice . Key recommendations to reduce prescribing errors include education in and summative assessment of practical prescribing, particularly with respect to commonly prescribed drugs .
Identification of ‘commonly-prescribed drugs’ to support prescribing training has proved controversial. The World Health Organization recommends that students themselves should identify drugs that they will use regularly in clinical practice and develop a core list of p (personal) drugs . In practice this is difficult for junior students who are presented with a vast array of different medicines for the first time. De Fries and colleagues found that use of a core drug list and associated prescribing information (formulary) improved prescribing skills in medical students . However, as long as students were studying in a problem-based curriculum, prescribing skills improved irrespective of whether the core drug list was tutor or student-led (personal). Previous ‘tutor-led’ core drug lists for medical students have been based on consensus of senior clinical pharmacologists  or created by a process of iteration between specialist consensus lists , local NHS formularies and local teachers . These approaches tend to produce longer, more comprehensive drug lists, which may be less useful in focussing learning towards clinical practice.
In this paper we describe the development of a core drug list based on common use in clinical practice. The core list was compiled from primary care prescribing data, secondary care prescribing information from two London hospitals and drugs prescribed in the emergency situation according to acute care guidelines. The list was validated by determining stability over 2 years and by determining whether drugs on the list were prescribed by FY1 doctors. The relationship of prescribing frequency to error rate was established by comparing the core list with error rate from the EQUIP study .
- Top of page
- Competing Interests
We identified a core list of 100 drugs or classes commonly used in primary and secondary care which changed little over 2 years, and demonstrated a strong relationship with drugs most frequently associated with prescribing errors. The stability of the list over time indicates that use of this list to support prescribing training for senior medical students would be relevant for their practice as newly qualified doctors. In support of this, FY1 doctors prescribed 65% drugs on the list at least monthly and 76% FY1 doctors did not regularly prescribe any drugs that were not on the list. The FY1 prescribing survey was conducted 2 years after the original list was developed, providing further evidence for the enduring relevance of this core list.
There was a strong relationship between the frequency of drug prescribing in our study and error rate identified by the EQUIP study . This indicates that error rate is related to workload and that learning focused on commonly-prescribed drugs could have the potential to minimize prescribing errors. The nature of prescribing training delivered around the core drug list requires as much attention as the core list itself.
Consideration of the dual process model of clinical reasoning developed in the cognitive psychology literature has potential to identify a role for the core drug list in prescribing training. The dual process model describes two modes/systems of decision making and judgement that depend on pattern recognition . System 1 processes are intuitive, reflexive and fast and are triggered automatically when a pattern is recognized. These processes require little effort and clinicians spend most of their time in this mode. System 2 processes are triggered when a pattern is not recognized. In this situation analysis and deliberate, slow reasoning are required to make a judgement and considerable effort is required to work in this mode. Although separate, the two processes are linked. Learning and practice convert system 2 to system 1 processes. Feedback, reflection and assessment (system 2 processes) are required from time to time to review and improve system 1 processes. System 1 and system 2 processes require different skills and are associated with different types of error.
Neal Maskrey, Director of Evidence-based Therapeutics at the National Prescribing Centre, suggests that this model can be applied to prescribing. Prescription of familiar drugs uses rapid system 1 processes, e.g. reflex prescription of paracetamol for a patient with mild pain. Prescription of unfamiliar drugs requires more time-consuming system 2 processes such as reference to the BNF, local and national guidelines and consulting with colleagues and pharmacy. A potential role for our core drug list is to support doctors in developing robust system 1 processes for safe and effective prescribing of drugs that will make up the majority of their workload. This must be complemented by development of system 2 processes for dealing with prescribing of unfamiliar drugs.
We found that emergency drugs were not prescribed commonly and are likely to be unfamiliar to newly-qualified doctors. The dual process model indicates that system 2 processes should be used for unfamiliar drugs, but they are unlikely to be appropriate in the emergency situation. We have therefore included emergency drugs in the core list to encourage students and trainees to become familiar with these drugs as part of their learning. Other processes such as easy availability of emergency guidelines, simulation training and pharmacy support are essential to promote effective prescribing in the emergency situation.
Although the core list was essentially stable over 2 years, some drugs moved into and out of the ‘top 100’. Around one quarter of these mobile drugs were antibiotics. Antibiotic prescribing is likely to change more rapidly than prescribing of other drugs due to changing microbial resistance and to vary nationally due to diverse infection control strategies and local resistance patterns. Design of prescribing education needs to take this into account.
Our study does have some limitations. The core list was developed from 752 million prescriptions in primary care but only 7705 prescriptions in secondary care. This discrepancy in numbers reflects the easy availability of electronic prescribing information in primary care, whereas secondary care information was collected manually. However even though secondary care prescription numbers were low, the 100 drugs used most commonly in secondary care remained reasonably consistent over 2 years. Furthermore only 10 drugs on the list were derived from secondary care prescribing data alone.
Primary care prescribing information was collected from England, whereas secondary care prescribing information was collected from two London teaching hospitals. Despite this, the core list appears to have broad applicability. In our FY1 prescribing survey, doctors from 39 hospitals across London prescribed 65% drugs on the list at least monthly and seldom prescribed drugs that were not on the list. The rate of prescribing errors made by doctors in the North West of England  was closely related to frequency of prescribing of drugs on the core list. Although our core drug list is based on prescribing practice in England, the majority of drugs also appear on the World Health Organization List of essential medicines . It is therefore likely that our core drug list would be useful and applicable to prescribing training in medical schools outside England, and that English graduates who have based their learning around this list would be equipped to prescribe elsewhere in the world.
Our core list contains 100 drugs, in line with British Pharmacological Society recommendations . However it is likely that this is too many for students and newly-qualified doctors to develop robust system 1 prescribing processes. Figures 1 and 2 show that the lower a drug was ranked in order of prescribing frequency, the more unstable its position was in the rankings. The core list may therefore more appropriately contain 60 or even 40 of the most commonly prescribed drugs.
Implications and future work
We have identified 100 drugs most commonly used in clinical practice and developed an enduring core drug list to support prescribing training. We propose that prescribers should develop robust system 1 skills for prescribing drugs in this list, complemented by system 2 skills for dealing with less familiar drugs. Challenges for the future include developing training programmes and learning resources to support prescribing training in this model and evaluating the effect of such training on prescribing errors and patient safety.
Primary care prescribing information was obtained from the NHS Health and Social Care Information Centre and is included in this paper with their kind permission.‘Copyright © 2009, Re-used with the permission of The Health and Social Care Information Centre. All rights reserved’.
The authors thank Neal Maskrey, Director of Evidence-based Therapeutics at the National Prescribing Centre for introducing them to the dual process model of clinical reasoning.