Studies carried out in hospitals in the USA suggest that prescribing errors cause harm to about 1% of all patients,(1) and there have been estimates that deaths due to medication errors in America exceed those due to road traffic accidents.(2)
Prescribing errors are inevitable. Prescribers are human and, as such, make mistakes. Historically, the reaction to such errors was always to apportion blame, whatever the circumstances. However, it is now recognised that a “no blame” culture will produce greater improvement in performance. Industrial psychologists have developed frameworks to analyse the causes of errors and to suggest solutions. Reason developed one such framework,(3) which has been applied to medical error – see Figure 1.
Studies based on this model have stated that prescribers must be accountable for their actions; however, errors usually arise as a result of other factors.(4) Therefore, prescribing errors could be reduced by training, adherence to systems of work and the introduction of new working practices.
The UK Audit Commission reports that medication errors are common because of major systemic weaknesses in prescribing arrangements.(5) Around 70% of prescribing decisions are made by preregistration house officers and senior house officers, even though they have limited knowledge of medicines. It is imperative therefore that prescribers are supported fully by the rest of the healthcare team to reduce the frequency and potential impact of these mistakes. An integral step in this process is the clinical screening of discharge prescriptions by pharmacists before dispensing.
Pharmacists in secondary care have access to a great deal of patient information. Prescriptions are written on discharge letters, which often contain diagnosis, follow-up plans and comments to the GP. With the focus moving towards technician-led dispensaries, it is imperative that clinical pharmacists have access to a patient’s details on a ward level. This will ensure that they can make informed decisions so that the prescriptions reaching the dispensary are safe, accurate and appropriate. All of this means that hospital pharmacists can gain a much fuller picture of the patient’s treatment, enabling them to make more informed interventions.
Much of the dispensary staff’s time is taken up with details that have been omitted from the prescription during writing, such as ward number, patient’s unit number, length of an antibiotic course, and so on. These represent more of an inefficiency than a risk to the patient; however, they are regarded as errors because they require correcting. Electronic prescribing can assist in reducing these types of errors, as prescribers are prompted to enter the relevant information and given specific choices to make. However, it potentially gives rise to a different type of error, such as picking the incorrect option from lists. One of the aims of this study was to identify whether handwritten or electronic prescriptions contained fewer errors.
Stobhill Hospital in Glasgow has recently introduced checking technicians into the dispensary to allow pharmacists to spend more time in the clinical setting. It is therefore essential that prescriptions have been screened accurately by the clinical pharmacists at ward level. If they are not seen by the clinical pharmacist, they must be screened by a pharmacist in the dispensary, who has less access to clinical information. The study looked at the incidence of errors not being identified by the first pharmacist and having to be corrected later on in the process.
To identify and quantify interventions made by the dispensary pharmacist in discharge prescribing and to identify any training issues that arise.
- To quantify how frequently pharmacist interventions are required.
- To identify the most common errors in discharge prescribing.
- To establish whether electronic or handwritten prescribing is more accurate.
- To establish whether errors are caused by a lack of training, pressures on the prescriber or active mistakes (slips, lapses, loss of concentration, etc).
- To identify training issues that arise for pharmacists and prescribers.
- To establish patterns in pharmacist interventions.
- To identify accuracy in pharmacist clinical screening.
Methodology and timescale
Data collection was carried out in the pharmacy dispensary at Stobhill Hospital over four weeks in June/July 2003. Data was collected by all pharmacists and, when relevant, technicians. A standardised data collection form was used to record:
- The date of the intervention.
- The ward involved.
- Whether the pharmacist screening the prescription, or the pharmacist checking after dispensing, picked up the error.
- The type of error, recorded in code form.
- Whether the prescription was electronically prescribed or handwritten.
The pharmacist at the point of intervention decided the causative factor behind the mistake. This was a subjective measure that depended on the individual pharmacist’s interpretation of the conversation with the prescriber and could only be used as an approximate indicator.
The errors were coded and recorded on the data collection sheet in tick box form, and errors or interventions considered potentially more serious, “crucial interventions” or those needing further clarification described in further detail.
Over the four-week period during which data collection took place, 1,206 prescriptions were screened and dispensed (60.3/day). In total 313 interventions were made, which equates to 15.65 interventions per day. Interventions were required in 25.95% of prescriptions, and 40% of all interventions were classified as “crucial”.
Two-thirds of prescriptions were electronically generated. Overall, the number of interventions required was split almost evenly between handwritten and electronic prescriptions (see Figure 2). However, the proportion of interventions required in electronic prescriptions (20%) was significantly less than that required in handwritten prescriptions (32%).
The vast majority (91.2%) of interventions were made at the screening stage, but a significant number (8%) were not picked up by screening and were made at a later stage. In 1.8% of cases, this information was not recorded.
According to Reason’s model of accident causation, errors can be classified as active, caused by performance factors or latent issues (see Figure 1). The results in Figure 3 show that two-thirds of the prescribing errors were active mistakes by the prescriber. A quarter could not be classified, and the rest were split evenly between latent issues and performance pressures.
Figure 4 shows what classes of intervention were made most often. Type 11 (dosage incorrect, ie, sub-therapeutic or potentially toxic) was the most common error (18.2%), with the second most common concerning course lengths for antibiotics or steroid reducing doses.
The most common errors included:
- Incorrect doses of thiamine and vitamin B compound strong.
- The eye for application of eye drops not being stated.
- Modified- or sustained-release preparations not being indicated.
- Proton pump inhibitor doses not being reduced after triple therapy, or triple therapy antibiotics being prescribed incorrectly.
- The timing on nitrate doses (ie, no nitrate-free period).
- Dose of aciclovir incorrect for different indications (eg, herpes simplex, varicella zoster).
Serious errors included:
- Steroid doses for ulcerative colitis orally and rectally. Oral dose double what consultant recommended for half the duration, with no reducing dose. Rectal dose a quarter of consultant’s instructions.
- Megestrol acetate prescribed as 40mg, but intended to be 160mg.
- Doses of co-carbidopa out by a factor of 100 (ie, 625mg instead of 62.5mg).
- Patient admitted and insulin commenced, but omitted from final discharge prescription.
- Comments section of prescription stated patient should be on indomethacin but prescriber was unsure of dose, therefore had not prescribed it.
- Patient newly started on warfarin, but missed off discharge prescription.
- Allopurinol dose three times higher than it should have been and ciclosporin dose incorrect. Patient also on warfarin, so potential interactions.
- Patient prescribed thiotepa (cytotoxic) instead of thiamine.
- Patient prescribed amiodarone instead of aminophylline.
- Patient with severe congestive cardiac failure, which was reason for admission, had frusemide omitted from discharge prescription.
- Mirtazapine prescribed instead of nitrazepam.
- Phenytoin dose halved because previous seizures not recorded in notes (at GRI) and doctor could not interpret lab results of phenytoin level, as was unsure of units the level was reported in.
This audit shows a worryingly high incidence of prescribing errors. Over a quarter of all prescriptions received by the pharmacy required some form of intervention by a pharmacist, and 40% of the interventions made related to a serious prescribing error. All of these errors, serious or not, take time to resolve and have an effect on the overall service provision of the department.
The wards that required most interventions were those where the pharmacists also have a commitment to the aseptic lab or dispensary and as such do not have as much time to spend on the ward, and in most cases cannot attend the ward at the most appropriate times. As a result they cannot screen every prescription. This results in prescriptions usually being authorised by another pharmacist, leading to more pharmacy time being spent on resolving problems at a later stage, and with less knowledge of the patient.
The electronic prescribing system (incremental discharge letter system: IDLS) produced proportionally less errors than handwritten prescriptions. This appears to be because the system prompts prescribers to input data where necessary, gives them a list of options in some cases, and eliminates problems due to poor handwriting. There were, however, incidents where the formulations, and even the drug, were selected incorrectly from a pull-down menu. Pharmacists only picked up these errors because the doses prescribed were incorrect for the drug that had been selected. Errors like these potentially have extremely serious consequences. This has to be established as a training issue for junior doctors.
The majority of interventions (91%) were made by the pharmacist screening the prescription. However, a further 8% were made during checking of the prescription after dispensing. With the advent of checking technicians, it is vital that prescriptions are effectively screened, as there will be no second clinical check. From these results it seems there is work to be done. Perhaps the 8% of errors not identified during screening would have been picked up had the prescription been screened by the ward-based pharmacist, who has a better knowledge of the patient.
In a quarter of patients, it was not possible to establish what had been the cause of the mistake. However, when it was possible, 67% were classified as “active” mistakes. Active failures include lapses of memory and attention, incorrect choices of objectives, or an incorrect path to achieve the objective. This, again, must be identified as a training issue for prescribers. Less than 5% of errors could be attributed to performance pressures, such as covering unfamiliar wards and patients, or latent issues, where organisational processes can be responsible.
The most frequent type of error related to the dosage of drugs on prescriptions. This is a potentially serious error, and prescribers must be encouraged to use the British National Formulary or clinical pharmacists for advice. Other common errors related to the omission of information from the prescription. Work pressures on PRHOs could result in them having insufficient time to write prescriptions carefully. This is another area where clinical pharmacists, as part of a multidisciplinary approach, could ease the burden, by being on the ward at appropriate times and offering advice.
Are these errors occurring because junior doctors have too much pressure on their time, and ward staff are encouraging them to write prescriptions in a hurry to fit in with other arrangements, such as transport or patients’ relatives? Or has the change in training of medical students produced doctors who are less well prepared for prescribing? Discharge prescribing is a huge part of a junior doctor’s role, so is there enough pharmacy input into their training? Doctors who have been involved with discussions regarding this audit appear to believe that medical students do need more training in prescribing, especially during the final year.
The results show that clinical pharmacists make a valuable contribution to patient care with regards to discharge prescriptions. Much of the poor prescribing that causes problems could be prevented by increased pharmacy involvement in the training of medical students. Also, an increased clinical role for the pharmacist on the ward, for example on ward rounds and advising on discharge prescribing, would improve the quality of care and reduce the number of adverse incidents relating to medication.
Type of intervention
- Directions, dose or drug not legible
- Directions incomplete
- Dose missing
- Drug prescribed generically when brand name required, eg, lithium, diltiazem
- CD prescription writing requirements not met
- Course length incomplete or inappropriate (eg, antibiotic course, steroid- reducing course)
- Ward incomplete
- Form incorrect/not stated
- Patients own medicines not indicated
- Dosage incorrect (ie, subtherapeutic or potential overdose – please state
- Potentially hazardous interaction – please state
- Inappropriate prescribing other (eg, laxatives omitted, 2 ¥ opioids, potassium etc) – please state
- Wrong drug prescribed – please state
- Other – please state
- Bates DW, et al. Incidence of adverse drug events and potential adverse drug events. JAMA 1995;274:29-34.
- Cox A, Marriott J. Dealing with dispensing errors. Pharm J 2000;264:724-5.
- Reason J. Human error. Cambridge: University of Cambridge; 1990.
- Dean B, et al. Causes of prescribing errors in hospital patients: a prospective study. Lancet 2002;359:1373-8.
- The Audit Commission. A spoonful of sugar: medicines management in NHS hospitals. London: AC;2001.