Within a litigious society and a national movement to deliver quality care, the solution to reduce medication errors has become actively sought.

A medication error is any preventable event, which may lead to inappropriate medication use or patient harm, while the medication is in the control of the healthcare professional or patient.1 Medication errors can be classified as occurring during prescribing, transcribing, dispensing or administration of a therapeutic agent.2 While a medication error may or may not result in patient harm, an adverse drug event is one which results in a clinically significant outcome.3

OUTLINE OF PROBLEM

For adults, the reported incidence of medication errors approximates 5% of all orders written.3 It has been acknowledged that the paediatric population is at greater risk due to the need to tailor drug doses to body weight and recognise the changing pharmacokinetic parameters through childhood.4 In paediatric inpatients, medication errors occur as frequently as 1 in 4.2 drug orders,5 with up to 80% of these being prescribing errors.

Assuming a reported paediatric dosing error rate of 1 in 100, this extrapolates to 50 000 paediatric dosing errors per year in England alone.6 Medication errors made involving paediatric patients have three times the potential to produce harm,7 with preventable adverse drug events occurring in 0.9% of medication orders.8

The Department of Health has recently highlighted this risk by publishing recommendations for the safer use of medicines in children,9 suggesting that the action to reduce the risks in paediatrics should focus on:

• training/competence assessment in drug dose and infusion rate calculations;

• availability of aides-memoire where validated computer software is not available;

• including the child’s age, weight and intended dose (mg/kg) on all prescriptions.

Although computerised technology with prescribing decision support has been hailed as the solution to most potentially harmful prescribing errors,7 such systems are expensive and thus out of reach for many hospitals. Given that concerns have been raised regarding calculation errors made by junior staff especially following staff change over1011 it seemed sensible to target the interventions which would perceivably most support our new junior staff until the National UK programme for IT can be implemented.12

CONTEXT

The study reviewed data collected from Airedale General Hospital, a 600-bed District General Hospital located in West Yorkshire, one of the top hospitals named in the Good Hospital Guide 2003 and 2004 (the Dr Foster guide that compares services and standards in the NHS and private hospitals within the UK13). The children’s unit comprises 30 beds accommodating paediatric medical, surgical and orthopaedic admissions.

ASSESSMENT OF THE PROBLEM

Retrospective analysis of inpatient drug charts was undertaken, with the charts analysed by a single research fellow (AD) according to criteria identified as good prescribing practice2314 to identify prescribing errors.

Patient notes were examined for documentation of adverse drug events including adverse drug reactions. A historical control was used to reflect an identical period of time the year before when the ward was as busy, and the junior doctors were beginning their paediatric rotation.

Results were analysed using SPSS version 11.5.

Strategy for change

After discussions with the paediatric junior doctors, it was felt that more support could be offered to them in the form skills of training within a prescribing tutorial and a written resource that could be referred to “on the job.”

The paediatric junior doctor prescribing tutorial was held in February 2004, prior to the commencement of clinical duties upon the children’s unit. The tutorial was conducted by a pharmacist (AM) with experience in prescribing training and AD. Fictitious medication charts were used to illustrate good and bad prescribing habits—a method used successfully in training pharmacy students.15 Emphasis was placed on legible handwriting, use of decimal points only where necessary, avoiding ambiguous units of administration and indicating frequency of administration especially for “as required” prescriptions. Weight-based drug calculations were practised, and infusion rate calculations discussed.

The bedside guidelines were formulated to encompass the most frequently prescribed medications utilised on the children’s unit. Liaison with the surgical, orthopaedic and paediatric teams resulted in the production of a guideline containing 22 drugs (listed broadly as antibiotics, analgesics and antiemetics) dosed according to age wherever possible, or weight in mg/kg/dose (as consistent with RCPCH “Medicines for Children” book).

After local Drug and Therapeutic Committee approval, they were introduced into the children’s unit in June of 2004. They were placed in each folder where the medication charts are held at the bed end.

Effects of the change

The medication charts were analysed before and 1 week after the introduction of the two interventions. Four sample groups were identified:

1. 10/02/03 to 19/02/03 pretutorial: The doctors in this group had access to the pre-existing ward protocols for prescribing and the RCPCH “Medicines for Children” formulary. Medication charts for 90 paediatric medical admissions were considered.

2. 10/02/04 to 21/02/04 post-tutorial: In addition to the above resources, these paediatric junior doctors had a prescribing tutorial within the first week of their appointment. Medication charts for 100 paediatric medical admissions were considered.

3. 10/02/04 to 21/02/04 preguideline: The same sample as (2) above, but inclusive of surgical and orthopaedic admissions to produce the baseline sample for the bedside prescribing guideline review—of 109 admissions (orthopaedic and surgical doctors did not have the prescribing tutorial but did have access to the bedside guideline).

4. 10/06/04 to 24/06/04 postguideline: These doctors (surgical, orthopaedic and paediatric) had the additional resource of a prescribing guideline as a reference at each bedside. Medication charts from 105 admissions were considered.

With reference to table 1

A total of 76 prescribing errors were detected pretutorial. The most common error recorded was the absence of medication strengths, such as paracetamol, ibuprofen and chloral hydrate, where different formulations are available, and the dose could have been misinterpreted. Appropriate doses were administered to the patients concerned (as indicated by the nursing record of administration).

The second most common error noted was the absence of intended frequency of administration for “PRN” medication. The majority of the orders were for medication where clinical consequence would be unlikely from administering them more frequently than intended (such as Salbutamol 100 µg metered dose inhaler). Of concern was that 33% of these orders were for paracetamol; however, the doses had been spaced appropriately (4–6-hourly) by the nursing staff administering the medication.

The dosing errors identified were mostly subtle deviations from the RCPCH “Medicines for Children” formulary, captured because of strict 10% thresholds enforced for the purpose of collecting the data. Included was a paracetamol overdose of 7000 mg as required—clearly unlikely to have been administered given the volume of syrup it would have required to deliver the dose! Similarly, a budesonide metered dose inhaler was prescribed as 100 mg instead of 100 microgram.

In total, the pretutorial sample of medication charts contained 76 prescribing errors equating to 30.5% of all orders. The post-tutorial sample contained only 44 errors or 16.5% of all medication orders. This equates to a reduction in prescribing errors from 1 in 3.3 orders to 1 in 6.1 orders (p = 0.023, Mann–Whitney U test).

With reference to table 2

The preguideline sample of charts had 59 prescribing errors recorded (18.4%). The most common error recorded was absent medication strengths as before, followed by underdosages of medication (four out of 11 being underdoses of antibiotics that could have potentially had a clinical effect). Absent dosage frequency for “as required” orders was another common error, with a single order for oral morphine sulphate not indicating the frequency of administration intended. No morphine sulphate was administered.

A single adverse drug event occurred in the preguideline sample group before the introduction of the bedside prescribing guidelines. This involved the prescribing (and administration) of intravenous morphine sulphate at four times the recommended dose, prior to a surgical procedure, causing excessive sedation and resulting in intensive care admission. A calculation error with a deficit in core mathematical skills was central to this adverse event: support was provided for the surgical junior doctor involved.

The postguideline sample of charts yielded 56 prescribing errors (17%)—essentially similar to the preguideline group. The absolute error rate was reduced from 1 in 5.4 orders to 1 in 5.9 orders (p = 0.733, Mann–Whitney U test), which did not meet statistical significance. There appeared to be an improvement in the prescribing errors noted for the surgical and orthopaedic admissions to the children’s unit after the introduction of the bedside guideline (1 in 3.5 drug orders to 1 in 5.4 drug orders). Given small numbers (54 drug orders versus 27 drug orders), we cannot draw conclusions from this but could speculate that a bedside prescribing guideline may be useful to doctors who have not attended a prescribing tutorial.

Lessons learnt

Medication errors are an important cause of iatrogenic illness, although only 0.9% of reported medication errors are likely to produce harm.8 It has been acknowledged that the recognition of latent errors (errors waiting to happen, or systems encouraging medication errors) could help prevent more serious adverse drug events.2 Within this study, most of the medication errors were of little clinical significance on their own but served to illustrate prescribing practices that could potentially lead to a preventable adverse drug event. Indeed, some of the errors detected (such as those involving paracetamol or morphine sulphate) could have potentially led to an adverse event if they had been administered without insight. This highlights the essential role that each team member plays in preventing adverse drug events: in particular, vigilant nursing staff who are placed within the system to be the last person who has the opportunity to check the medication is appropriately prescribed and dispensed before it is administered. Given that there was only one adverse drug event recorded in this study, it could not be used as an outcome measure. This reflects a limitation of the study—the numbers are small, suggesting that a larger study is required from which to draw conclusions. The results of the junior doctor prescribing tutorial were encouraging enough to modify our local practice. Larger studies may help to clarify if these recommendations should be prioritised at other hospitals.

In this particular study, the promising result of the senior house officer prescribing tutorial and the nature of the adverse drug event suggest that training is in part the problem and the solution. This has been acknowledged by multiple authors715 and the Department of Health9 in the UK, with a call to strengthen undergraduate pharmaceutical prescribing training.

The current prescribing workforce will require a different solution to the problem. The challenge lies in how to address the issues within a highly educated profession, when the insinuation of lack of knowledge is not well received. Induction tutorials for new staff could partially address any knowledge deficits, as is currently planned by Airedale General Hospital in a move to promote safe prescribing practice.

In a recent systematic review of paediatric dosage errors, Wong et al recommended that professional regulatory bodies review postregistration education to ensure sufficient mathematical training and assessment for healthcare professionals prescribing in paediatrics.6 The General Medical Council revalidation process may be the appropriate means to ensure that prescribing doctors remain competent.