A critical analysis of medication errors in relation to management

Published Date: 23 Mar 2015

'The process of medication use is a continuum of activities involving multiple health care professionals and multiple steps (that is, prescribing, transcribing, dispensing, administering, and monitoring), thereby creating multiple opportunities for error' (Zhan et al., 2006, p353).

The principle aim of this essay will be a critical analysis of the management of medicines by nurses in the hospital environment in order to avoid medical errors and their possible adverse affects on patients. The prevalence, types, causes, sources and some consequences of medication errors will also be discussed. In order to provide a foundation for exploring these issues, definitions of medical error and related terms will be expressed. Types of medical errors originating in four sources, namely nurses, patients, doctors and pharmacists, will be discussed, as they may be severally, or individually, responsible for a medical error.

Reference will be made to five personal observations of potential medical errors of nurses whilst I was on recent clinical placements; these observations will form the central focus of the essay. Strategies will be outlined for reducing potential medical errors associated with these five observations, using evidence-based practice and relevant government policies. Medication errors associated with two high risk medicines, namely potassium chloride and insulin, will also be examined, together with strategies and protocols designed to reduce their potential danger. Finally, the importance of team work, reflective practice, good communication, and the need to overcome barriers in reporting medical errors by nurses will be emphasised.

Most medical errors do not harm patients (DoH, 2004). However, the National Patient Safety Agency (NPSA) (2009a) reported 3,426 medication incidents in Wales and that 1-2% of these incidents led to harm or death in the National Health Service (NHS) during the period April 2008 to March 2009; 32 medication incidents caused severe harm and 6 medication incidents caused death. In England, over the same period, the NPSA (2009b) reported that less than 1% of all incidents leading to harm or death in the NHS were due to medication errors; 155 medical incidents caused severe harm and 42 caused death.

In order to understand medication errors and discuss strategies for reducing medication errors, it is useful to understand the terms 'medication', 'medication error', 'adverse drug event', and also to classify types of error.

A medication is a compound which is taken or administered for one or more of the following reasons:

'to prevent a disease, to modify a physiological, biochemical, or anatomical function or abnormality; to replace a missing factor; to ameliorate a symptom, to treat a disease; to reduce anaesthesia' (Aronson, 2009, p601).

A medication error can be defined as 'a failure in the treatment process that leads to or has the potential to lead to, harm the patient' (Aronson, 2009, p599); the 'treatment process' is a complex process that includes prescribing, dispensing and administration of a drug. This definition does not indicate whether the error is due to a doctor, pharmacist, nurse, patient, or other person.

Medication errors should not be confused with adverse drug reactions (ADRs), although the two terms may overlap (Britten, 2009). An ADR has been defined by Bowman et al. (1996) as 'any adverse experience associated with the use of the drug including any side effect, injury, toxicity, sensitivity or reaction' (p10). The term 'adverse drug event' (ADE) is sometimes used for a medication error; an ADE includes human factors such as delay in administration, accidental overdose, or incorrect medication (Hendrie et al., 2007). ADEs may be linked by their intrinsic toxicity (inherent adverse effects) or by their extrinsic toxicity, that is, the way the drug is used; examples of extrinsic toxicity include interactions between two or more drugs or drug-food interactions (Guchelaar et al., 2005).

According to psychologists, errors may be classified as 'mistakes' (knowledge-based or rule-based errors), 'slips' (observable action-based errors such as the slip of pen or technical error) or 'lapses' (non-observable memory-based errors); a slip or lapse occurs when the action conducted is not what was intended (Reason,1990, p54), for example, an intention to write a prescription for 100mg of a drug, but writing 300mg instead (DoH, 2004). The disadvantage of this type of classification for medication errors is that it focuses on human rather than systems sources of errors (Aronson, 2009).

The causes of medication administration errors by nurses may be viewed as 'person' or 'system' based (Tang, et al., 2007). The 'person' approach to medication errors is the prevailing approach in medicine and is directed to 'aberrant mental processes such as forgetfulness, inattention, poor motivation, carelessness, negligence, and recklessness' (Reason, 2000, p768). Attaching blame to an individual for unsafe behaviour is easier than targeting an organisation (Reason, 2000).

The basic principle of the 'system' approach is that humans are prone to err, and errors are due to 'error traps' and 'organisational processes' in the workplace. Errors committed by individuals are transmitted through 'holes' in the layers of 'defences, barriers, and safeguards' of an organisation; these layers with their holes are analogous to the holes in slices of Swiss cheese (Reason, 2000). The presence of holes in a single layer does not normally lead to an error, but if the holes are lined up for a brief moment, there is an opportunity for an error to follow a trajectory through a faulty system and cause damage to a victim. Reason (2000) contended that adverse events usually involve a combination of 'active failures' by individuals and 'latent' factors in the working environment. 'Active failures' include forgetting or ignoring an established policy or procedure, whilst 'latent' factors include staffing shortages, inadequate equipment and lighting (Brown, 2000).

The DoH (2004) guidelines for reducing medical errors stress that the following checks should be performed prior to medication administration: 'right medication, in the right dose, to the right person, by the right route, at the right time' (p61). Crouch and Chapelhowe (2008) also concurred with adherence to these 'five rights' in medicine management. However, rule-based administration of medicines using the 'five rights' may lead nurses to act ritualistically, and give a false assurance that their practice is safe (Crouch and Chapelhowe, 2008).

Deviations from the 'five rights' of medicine administration were elaborated in an extensive literature review of 26 studies by Brady et al. (2009). In one study, violations of the 'five rights' by 72 nurses in a Taiwanese hospital were reported by Tang et al. (2007); these were wrong dose (36.1%), wrong drug (26.4%), wrong time (18.1%), wrong patient (11.1%) and wrong route (8.3%). The main factors contributing to medical errors by nurses were personal neglect (86.1%), and 'systems-based' factors such as heavy workload and new staff. The main limitations of this study were the relatively small sample size and the limited experience of the nurses (average of two years).

Administering medications is prone to error since it is 'more than a technical mechanical process' (Eisenhauer, 2007, p86). Concurring with this view, Crouch and Chapelhowe (2008) stated that safe medication is a 'psychomotor skill' and requires:

'cognitive skills such as observation, listening to patients, analysis, critical judgement, clinical judgement, decision making, teaching and interpersonal skills' (p 487).

Judgment is needed in dosage, timing, selection of specific medications, checking on a patient's laboratory data (for example, potassium or blood glucose levels), observing and responding to adverse drug events, and deciding when to stop medication if adverse effects occur (Eisenhauer, 2007).

Safe medication is dependent on nurses acquiring motor skills using advanced technology to deliver medicines by a variety of routes, (for example, hypodermal, intramuscular, intravenous and patient controlled analgesia) and using different types of pumps, tubes and valves; such complexities have increased the risks of inappropriate dosing and wrong route (Sheu, et al., 2009; Tang et al., 2007).

During recent clinical placements in the hospital environment I observed five potential sources of error in the administration of medicines by nurses: these were poorly prepared, untidy and unattended drug trolleys, interruptions during nurses' drug rounds, unsafe patient self-medication, and complex activities involved in intravenous medication and drug calculations. These five potential sources of error I observed will now be examined in turn.

The first example of a potential source of error I observed was poorly prepared, unattended and untidy drug trolleys; these have the potential to lead to errors in administering medicines accurately and safely to patients (DoH, 2004; Castledine, 2006; Palese et al., 2009). Medicine trolleys are commonly too small and poorly designed for holding an ordered arrangement of diverse medications and related medical equipment; furthermore, identification of medicines by nurses is subject to error, since all labels are white, and medicines often have similar sounding names, for example, penicillin (an antibiotic) and penicillamine (an anti-inflammatory drug) (Crouch & Chapelhowe, 2008)

Unattended drug trolleys pose the risk of interference by patients and visitors; it is unsafe to leave a drugs trolley unattended while returning to the nurses' station (Palese, 2009). Castledine (2006) cited the case of a senior staff nurse in a busy hospital who frequently left the drug trolley untidy and left blister packs containing tablets out of their boxes, or placed them in the wrong boxes. On one occasion she left the drug trolley unattended; this resulted in a patient taking a container from the trolley, ingesting an excess of a sedative and becoming very drowsy; fortunately the patient recovered, but the nurse's name was removed from the Nursing Register.

An alternative to centralised distribution of drugs using a drugs trolley is the decentralised method which involves placing each patient's drug requirements in a locked cupboard by the patient's bedside (Jones et al., 1996; Manias et al., 2004, Bennett, 2006). Bennett (2006) showed that decentralisation in a 24-bed general medical ward in a hospital led to a 64% reduction in the number of interruptions nurses experienced while administering medications; on average, each nurse saved 23 minutes per 12-hour shift. Decentralising the medication system may improve medication safety by reducing the number of interruptions and distractions during the preparation and administration of medicines (Bennett, 2006). However, the NPSA (2007) cautioned that a patient's own medicines left in a bedside locker at the time of discharge may sometimes result in a patient going home without their medication, and possibly lead to the next patient being inappropriately medicated.

According to the DoH (2004) guidelines for improving medication safety, storage of drugs at the bedside of each patient 'reduces the range of drugs from which selection can be made' and may therefore reduce the risk of medication error when compared with selecting a drug in an untidy, busy, clinical room.

The second example of potential sources of medical errors I observed was the occurrence of interruptions of nurses on their drug rounds. Nurses are often interrupted by other healthcare workers when they are standing still by the drugs trolley (Pape et al., 2000). The drug round is a critical time and is 'vulnerable to a multitude of interruptions and distractions that affect the working memory and ability to focus' (Pape, et al., 2005, p108). Critical steps in the process of medication may be interrupted by noisy conversations with other staff members and visitors in a crowded hospital and also by multi-tasking during medication of patients (Bennett, 2006; Pape et al., 2005). Interruptions may affect clinical judgment in dosage, timing and selection of specific medications in a fast paced environment; 'constant vigilance' is needed to ensure appropriate medication for each patient (Eisenhauer, 2007).

Distractions may arise when nurses try to solve other problems while administering drugs. 'Solving other problems' was the leading cause of error in a list of 34 conditions that resulted in a medication error in the administration of medicines by the 72 female nurses in Tang et al.'s (2007) study in a Taiwanese hospital.

A quantitative and observational study by Kreckler et al. (2008) of interruptions of nurses' drug rounds in a 37-bed acute surgical ward at a U.K. teaching hospital indicated that nurses are interrupted on average 11% of each drug round. Interruptions were due mainly to doctors (21%), other nurses (17%), patients (11%) telephone calls (8%), and visiting relatives (3%); attempts to gain access to controlled drugs was also a serious source of interruption, since only one set of keys was available for several drug rounds which overlapped. More than 50% of the interruptions lasted more than 30 seconds (Kreckler et al., 2008). The main limitation of this study was that data were obtained from a small sample in one surgical unit.

Various solutions were proposed by Kreckler et al. (2008) to reduce interruptions. These solutions included: wearing some item of 'identifying clothing' which informs other healthcare staff they are not to be interrupted; and discouraging visiting or telephone calls from relatives during drug rounds. Research by Pape et al. (2005) demonstrated that placing warning signs with the words 'Stop - 'do not disturb' during medication administration reduced the number of interruptions and medical errors. Warning signs bearing the words 'Please do not disturb' above drug trolleys were displayed in a local NHS trust on a recent clinical placement, in order to minimise medical errors when drugs were being administered. O'Shea (1999) emphasised the need for provision of a quiet area where a nurse can prepare the medications for the drug trolley prior to the drug round in order to avoid interruptions.

The third potential source of error I observed related to patient self-medication. Drug tablets were sometimes left on top of patients' bedside lockers for self-medication by some patients; it was conjectured that the risk factors in this situation might be failure of patients to take the right medication at the right time due to lack of supervision by a nurse, unauthorised disposal or concealment of tablets, and removal of tablets by other patients or visitors. Jones et al. (1996) and Curry et al. (2005) acceded to the possible and frequent misappropriation of drugs by self-medicating patients in hospitals.

However, a study of 220 patients who had been screened for their ability to self medicate showed that patients educated about self-medication resulted in benefits for nurses and patients in an Australian hospital (Grantham et al., 2005). The study showed that 29% of patients were able to self-medicate safely and independently while in hospital; a further 26% of patients were able to self-medicate when directly supervised by a nurse. Although nurses perceived an increase in their workload, the main benefit was that no patient initiated medication errors occurred during the study period. Studies by Manias et al. (2004) and Jones et al. (1996) have demonstrated other benefits of self-medication for patients; these included the sense of autonomy and better preparation for drugs at discharge.

The need for caution by nurses in self-administration of medicines by patients is recognised in three standards presented in the NMC (2008a) document 'Standards of Medicine Management'; 'Standard 5' makes clear that self-medicating patients should have a secure, lockable bedside cabinet for safe storage of their medicines. 'Standard 9' states nurses are:

'responsible for the initial and continued assessment of patients who are self-administering and have continuing responsibility for recognising and acting upon changes in a patient's condition with regards to safety of the patient and others'.

'Standard 16' states that nurses: 'must assess the patients' suitability and understanding of how to use an appropriate compliance safely' (NMC, 2008a).

Adherence to Standard 9 of 'Standards of Proficiency for Nurses and Midwife Prescribers' (NMC, 2006) should also help to reduce medication errors; it states that the following information should be provided before commencing self- administration:

'the name of the medicine, why they are taking it, dose, frequency, common side effects and what to do if they occur, any special instructions, duration of course and how to obtain further supplies'

The fourth potential source of error I observed concerned the preparation and administration of intravenous medications by nurses. My observations of intravenous medications on a clinical placement in an intensive care unit revealed that the prescribed drug, number of doses, and corresponding patient were not always double checked by another nurse, mainly due to time limitations in a busy ward; in one case, a nurse administered a drug which was not the one prescribed.

An intravenous medication error has been defined as 'a dose prepared and/or administered by nursing personnel different than that prescribed by the physician and on the patient's records' (Anselmi et al., 2007, p1839). Intravenous doses are injected as a bolus, or given as a continuous infusion directly into the bloodstream via a vein (Crouch & Chapelhowe, 2008). Preparation for intravenous therapy may involve 'dissolving a powder, dilution or transfer of injection fluid from the original vial or ampoule into a container (a syringe or an infusion bag); these processes present multiple opportunities for errors' (Wirtz et al., 2003, p104).

The intravenous route for the administration of medicines by nurses has a high potential for patient harm (DoH, 2004; NPSA, 2007; Wirtz, 2003; Simonsen et al., 2006). Medicines administered intravenously are very difficult to 'retrieve' when compared with the oral route, since it is difficult to reverse their pharmacological effects once they have been administered; however, antidotes are sometimes available in the case of overdoses (Simonsen et al., 2006).

A six-day observational study conducted by Wirtz et al. (2003) concerning intravenous medication errors executed by 61 nurses in three hospitals (one British hospital and two German hospitals), indicated a total of 26% preparation errors and 34% administration errors. The most common errors for all three hospitals were wrong dose, omissions, and wrong administration rates. The most common type of administration error on all wards was the wrong rate; seventy three (88%) out of eighty three injections were faster than the recommended rate (3-5 minutes for an intravenous push).

As a result of this study, Wirtz et al. (2003) recommended the following strategies for reducing intravenous medication errors: improved communication between doctors, pharmacists and nurses; educating nurses about preparation of intravenous medicines and the clinical consequences of medication errors; provision of a centralised intravenous medication service; regular checking of drug charts by nurses; and standardised ready-to-use medicines. A disadvantage of using standardised ready-to-use medicines is that nurses may lose skills or have reduced efficiency in preparing intravenous drugs; consequently, serious medication errors may arise in emergency situations (Taxis and Barber, 2003).

In order to reduce medical errors arising from intravenous infusions, the DoH (2004) guidelines state that staff should receive training in the correct use of intravenous devices, have access to the suppliers' user manuals, and report any actual or suspected damage of these devices. 'Standard 20' of the NMC (2008a) guidelines on medicines management advises that 'wherever possible', two nurses should 'check medication to be administered intravenously' and that one of them should be the nurse who administers the intravenous medication.

The advent of computerised intravenous infusion 'smart pump' technology has enabled nurses to reduce some of the medical errors associated with intravenous infusions, especially errors relating to inappropriate dose rates and overdosing (Bowcutt et al., 2008). 'Smart pump' technology employs a programmable, computerised system which can control up to 4 attached infusion devices (Nuckols et al., 2008); it delivers intravenous medications, supported by computerised 'drug libraries, volume and rate calculations, dose limits, soft and hard alerts, and bolus dose limits' (Bowcutt et al., 2008). Safety software in the 'smart pump' displays alerts if doses exceed safe limits (Nuckols et al., 2008).

A survey of 1056 nurses in an American hospital led Bowcutt et al. (2008) to conclude that infusion pump technology is 'perceived by nurses to enhance safe nursing practice'. However, Nuckols et al. (2008) observed no significant difference in the number of adverse drug events associated with 4604 critically ill adult patients in two American hospitals, before, and after the introduction of 'smart pump' infusion systems.

Husch et al. (2005) pointed out that 'smart pumps' do not alert clinicians concerning wrong dose errors, wrong medication, or changes in a patient's condition. Husch et al. (2005) conjectured that significant improvements in patient safety will only be possible if new technology facilitates the interfacing of 'smart pumps' with other systems, such as computerised prescriber order entry, administration of bar coded medication, and pharmacy information.

Intravenous medications delivered by a pump system pose a serious risk of overdosing and harm, since they have an immediate systemic effect; also, they work at lower doses than oral medicines since they bypass the gastrointestinal tract (Crouch & Chapelhowe, 2008). Overdosing of high risk drugs by nurses using intravenous injection is an important and critical type of medical error since high risk drugs work in a narrow therapeutic range (DoH, 2004); furthermore, overdosing may lead to serious harm or fatality (NPSA, 2007). Some of the consequences of overdosing of two high risk drugs, namely potassium chloride and insulin will now be briefly explored, and also strategies and protocols for their safe administration by nurses.

Potassium chloride is commonly 'administered intravenously in diluted solutions to treat low potassium levels (hypokalaemia)' (NPSA, 2002). Inappropriate administration of concentrated potassium chloride solutions, and also ampoules of sodium chloride mistaken for potassium chloride, have led to accidental overdosing, cardiac arrhythmia, and fatality due to cardiac arrest in the U.K. and the USA (NPSA, 2002; Hadaway, 2001).

According to Mather (2008) a lethal dose of concentrated potassium chloride was given by intravenous injection in the neck of a 91 year old lady by a student nurse in a British hospital. The student had tried to deliver the drug in one dose rather than over the course of a few hours; as a result the patient suffered a cardiac arrest and died the next day. The coroner criticised the training of this student and said that the patient's death was preventable. On reflection, the student nurse should have worked within the limit of her competence (NMC, 2008b), and not agreed to administer potassium chloride.

In Australia, concentrated potassium chloride ampoules were removed from the clinical areas due to their link with patient deaths (Sladdin & Lee, 2006). Sladdin & Lee (2006) recommended a protocol involving central distribution of concentrated potassium chloride.

Protocol implementation for the preparation and administration of intravenous drugs has been shown to significantly reduce the incidence of medication errors (Tromp et al., 2009). The DoH, (2004) recommended the preparation of dilutions of concentrated potassium chloride in the pharmacy, rather than on the wards. In order to reduce the risk of accidental overdosing of potassium chloride, a local NHS trust published a 'Policy for intravenous potassium supplementation' in line with a patient safety alert issued by the NPSA (2002). The policy provides guidelines for the safe administration of potassium chloride by nurses; these include:

'the withdrawal of concentrated potassium chloride injections from all critical areas except critical care units; nursing staff must use ready prepared fluids when intravenous potassium replacement is required as far as possible; procedures for checking the use of potassium chloride injection must be in place in clinical areas; induction programs should include training about this policy and the risks of potassium chloride injection if used inappropriately; a second practitioner must check for the correct product, dosage dilution, mixing and labelling when a 20mmol/10ml potassium chloride injection has to be used, and sign the record accordingly' (p2).

A local NHS Trust made three other recommendations in their policy for managing safe dosing of potassium chloride; firstly, the dose rate using a peripheral line should not exceed 10mmol/hour in adults, otherwise there is an increased risk of thrombophlebitis; secondly, if potassium chloride is administered through a central line, dose rates should not exceed 100mmol/ 100ml over at least 5 hours, or 40mmol/100ml over at least 2 hours; thirdly, if the dosage is high, the nurse should contact the prescriber for clarification.

Insulin is another example of a high risk drug associated with intravenous medication errors. Insulin was not always administered on time in a local NHS trust. The timing of insulin administration by nurses to hospital inpatients is a common source of error and may be unsafe (Gangopadhyah et al., 2008). Since the onset of insulin action varies from minutes to 8 hours, inappropriate timing of insulin with respect to meal times can lead to hypoglycaemia (Cohen, et al., 2003). The timing of insulin administration depends on the type of insulin being used. Analogue insulin, for example, insulin lispro produced by recombinant DNA technology (Bates, 2002) is considered effective if administered five minutes before or after a meal, while non-analogue insulin is considered effective if administered 10-30 minutes after a meal (Gangopadhyah et al., 2008).

Research conducted by Gangopadhyah et al. (2008) compared the effect of timing of insulin administration on two groups of hospitalised patients in an English NHS Trust (group 1: insulin administered by nurses to 25 patients over 175 meals; and group 2 (insulin self-administered by 10 patients over 75 meals). There was no significant difference between the two groups with respect to age, duration of diabetes, or type of diabetes. Patients recorded their meal time on a timing sheet. In group 1, nurses administered insulin in only 19% of cases at the correct time, while in group 2, insulin was self-administered on time in 78% of cases. These results led Gangopadhyah et al. (2008) to recommend a trust-wide plan for patient self-administration of insulin in hospital for carefully selected patients, as it is safer than administration of insulin by nurses. Education of nurses and doctors with respect to the timing of insulin administration was also a feature of the plan.

No clear advice is given in the NICE (2009) guidelines about the timing of insulin in relation to mealtimes for patients with type 1 diabetes; the guidelines advise patients:

'to seek advice from professionals knowledgeable about the range of available meal-times and basal insulins, and about optimal combinations thereof, and their optimal use'.

The administration of insulin by a diabetic specialist nurse prescriber has been shown to reduce the number of medical errors associated with insulin and also oral hypoglycaemic agents (OHA) in a study by Carey et al. (2008); the results of this study indicated a significant reduction (about 80%) in errors in the intervention group when compared with the pre-intervention group. The delivery of insulin by a diabetes specialist nurse prescriber in the intervention group reduced the median length of stay by 3 days (Carey et al., 2008).

The fifth potential source of medical error I observed was the performance of drug calculations by some nurses. A frequent cause of medication errors in nursing practice is poor drug calculation skills; these have become a national concern (DoH, 2004; Wilson, 2006; Lee, 2008).

According to Kapborg (1994), poor mathematical skills may be related to poor performance in drug calculations. Kapborg (1994) studied the drug calculation skills of 545 experienced nurses and 197 nursing students in Sweden by giving them a test consisting of 14 items (percentages, dose calculation of liquid and solid drugs, transformation of units and calculation of time for giving intravenous fluids and transfusions); approximately 90% of nurses and student nurses who performed the test failed to solve all items. No significant difference was found in the error rate between the nurses and student nurses; both groups solved 9 out of 14 items correctly. If this standard was reflected by nurses in their daily work then inappropriate drug dosages would be administered to patients with possible consequences of harm and fatality.

Written paper tests used to predict nurses' calculation skills have been criticised since they do not reflect numeracy skills in nurses in the clinical area (Wright, 2007). According to Wright (2007), a written assessment is not a valid method of testing the numeracy skills of student nurses and ensuring their fitness for practice, since the context is divorced from the reality of the hospital environment. Nurse education needs to develop numeracy skills of student nurses in the context of their clinical placements (Wright, 2007).

In order to improve performance in drug calculations the DoH (2004) recommended that nurses who lack confidence in drug calculations should seek confirmation of accuracy from those who have demonstrated these skills in practice. Nurses should not rely on calculators but employ arithmetical skills (NMC, 2006). In the case of complex dose calculations, a second nurse should 'check the calculation independently to minimise the risk of error' (NMC, 2008a). Annual numeracy tests for qualified nurses, set by the NMC or local NHS trusts, might improve confidence and reduce medical errors in the administration of drugs. Empirical research using questionnaires by Fry & Dacey (2007) indicated that 92% (126/137) of nurse respondents felt that 'regular updates on calculations would be of benefit' (p679).

High standards of proficiency have been set by the NMC (2006) in order to reduce the incidence of calculation errors by nurse prescribers. The portfolios of nurses studying to be nurse prescribers should include evidence of 'numeracy skills, writing prescriptions, and prescribing in a range of scenarios' and 'numerical assessment within the context of prescribing practice'; furthermore, 'students must achieve a 100% pass' (NMC, 2006).

Although nurses make errors in drug calculations and in the administration of drugs, some of the medical errors for which nurses are blamed have their origin in inappropriate prescriptions by medical prescribers and dispensing by pharmacists (DoH, 2004). However, nurses are accountable for the safety of their patients (NMC 2008b) and are responsible for correct administration of drugs (O' Shea, 1999).

In order to ensure correct drug administration, it is advisable that nurses check prescribed drug doses with the most recent British National Formulary (BNF) handbook or online version (Crouch & Chapelhow, 2008; DoH, 2004). However, Cantrill (2000) questions the 'gold standard' reputation of the BNF, since some doctors knowingly prescribe lower doses than those recommended in the BNF for those patients who appear asymptomatic, or who are suffering unpleasant side effects, or who are responding to lower doses which are therapeutic. Aronson (2004) a clinical pharmacologist, criticised the BNF because in his view 'the lists of interactions in the BNF are difficult to read and understand' (p371).

A systematic review by Ross et al. (2009) of 24 studies focussing on prescribing errors of junior doctors, mainly in hospitals, indicated that medical errors are made more frequently by junior doctors than senior doctors; however, this review did not make a firm conclusion about the error rate of junior doctors since their were differences in the studies concerning error definitions and variations in the settings and grades of junior doctors (Ross et al., 2009). Inappropriate prescribing by junior doctors may be due to stress and deficits in knowledge and training (Velo & Minuz, 2009). Prescription errors may arise because doctors lack immediate access to drug information relating to its indications, contraindications, interactions with other drugs, allergies, therapeutic doses and side effects, and the latest laboratory results for patients (Ridley et al., 2004).

A five-year study detailing 1115 significant prescribing errors of doctors by Lesar, et al. (2002) showed that patients were frequently at risk due to confusion concerning inconsistent prescribing of brand names and generic names, and failure to specify controlled release formulations on medication dosage forms. Prescribing errors may also arise due to inaccurate prescriptions, omissions of important details (such as the total volume of solvent, duration of infusion), ambiguous abbreviations or symbols, and handwriting errors (Velo & Minuz, 2009).

Illegible prescriptions make it difficult for nurses to interpret doses and may lead to 10-fold overdoses and serious patient harm (DoH, 2004). Poor handwriting led to an intended prescription of vasolastine (a drug administered intramuscularly for treqting angiopathy) being interpreted by a pharmacist as vinblastine, a cytoxic drug (Bardzo et al., 2009). This error led to the death of an 83-year old patient, and two nurses and a pharmacist receiving fines and suspended prison sentences. In order to avoid this serious medical error, the pharmacist should not have interpreted the handwritten prescription but contacted the doctor for clarification and the nurses should have consulted the manufacturer's instructions (Bardzo et al., 2009).

According to the DoH (2006), prescriptions 'must be written legibly and in ink' and the names of drugs and preparation must be written clearly'. The DoH (2004) made the following recommendations to prescribers for reducing prescription errors: prescribers should receive guidance from the British National Formulary (BNF) when writing prescriptions, adopt recommended international non-proprietary names, pay particular attention to the dosage form and dosage route, and also be careful in the use of abbreviations, decimal points and zeros.

Electronic prescribing systems linked to the patients' records enable access by all health carers and may reduce the risk of handwritten prescriptions (DoH, 2004). A comparison of hand written prescriptions with computerised physician orders in a review of 12 studies by Shamliyan et al. (2008) indicated a 43% reduction in the total for dosing errors, 37.5% reduction in the total for adverse drug events and a 66% reduction in total prescribing errors for adults. Computerised physician order entry (CPOE) makes sure that medication orders are legible and that they contain all necessary information concerning dose, dosage form and route for administration (Agrawal, 2009).

Computerised physician order entry (CPOE) with decision support has the advantages of reducing prescription errors by providing access to information concerning patient allergies, pharmacokinetic data (drug interactions) and pharmacodynamic data (data pertaining to the biochemical and physiological effects of drugs on the body) (Agrawal, 2009; Ridley et al., 2004). Furthermore, CPOE systems perform calculations for dosage adjustment in respect of renal impairment and body weight of patients, and platelet count in the case of patients receiving anticoagulation therapy (Agrawal, 2009). Another advantage is that CPOE may reduce pharmacists' processing times for medication; a study by (Wietholter et al., 2009) in a tertiary care hospital showed the processing time to be reduced by about 90% (from 31 minutes to 3 minutes).

Although several claims have been made for the potential benefits of CPOE, a study by Zhan et al. (2006) comparing voluntary medication error reports of 570 health care facilities in America indicated that medication errors may arise due to faulty computer interface design, typing errors, and deficits in knowledge and clinical decision support. This study did not demonstrate the expected benefits of CPOE due to variations in reporting diligence. A study by Bedouch et al. (2009) of drug related problems (DRP) in medical wards in a French hospital showed that DRP are common even with CPOE.

Cheung et al. (2009) advocated intensifying of prescription checks by pharmacists to reduce prescription errors. The importance of prescription checks by pharmacists in hospital wards was demonstrated in an 880-bed hospital in North West England by Tully & Buchan (2009). Prescription checks on ward rounds by nine pharmacists in medical teams revealed prescription errors in 10.5% of new medication orders on patients' drug charts; this lead to the patients' drug charts being amended (Tully & Buchan, 2009).

The effect of a pharmacist participating in daily rounds with doctors in the medical team monitoring patients in a hospital ward was investigated by Scarsi et al. (2002). Pharmacist participation in the medical rounds reduced medication errors by 51%. Scarsi et al. (2002) suggested that this reduction might be explained by the pharmacist obtaining 'a more thorough clinical picture' of each patient; the pharmacist could evaluate the prescription by knowing the patient's medical history, drug contraindications and the reason for the prescription.

The majority of research on medical errors is concerned with prescribing and administration errors and the harm caused by them. However medical errors also occur in dispensing of medicines by pharmacists and may result in significant harm to patients (Beso et al., 2005). A dispensing error is 'a discrepancy between a prescription and the medicine that the pharmacy delivers to the patient or distributes to the ward on the basis of this prescription' (Cheung et al., 2009, p676).

Beso et al. (2005) investigated the frequency of different types of dispensing errors and their potential causes in a U.K. hospital pharmacy by interviewing pharmacy staff. At the final check stage, 2.1% of 4849 dispensed items were associated with one or more dispensing errors. Dispensing errors included 'content' errors (incorrect drug, incorrect strength, and incorrect dosage form), 'labelling' errors (incorrect patient name, incorrect instructions) and 'documentation' errors (absent/incorrect controlled drug records). In a total of 106 error producing conditions, the most common causes were being busy (21%), being short-staffed (12%), time constraints (11%), fatigue (11%), interruptions during dispensing (9.4%) and look-alike/sound alike medicines (8.5%).

Strategies recommended by Beso et al. (2005) for reducing dispensing errors included firstly, better labelling and storage of containers, secondly, appointing a member of staff to manage queries, and thirdly, introducing automated dispensing systems to reduce 'content' errors. The use of robots for dispensing has been introduced in a few large UK hospitals (DoH, 2004)

An overall strategy for reducing medical errors requires teamwork between doctors, pharmacists and nurses; teamwork is able to integrate the prescribing, dispensing and administering stages of drug management, and ensure patient safety (DoH, 2004). Doctors should recognise their inherent fallibility to make medical errors and acknowledge their interdependence with team members (Cosby & Croskerry, 2004), especially nurse prescribers who relieve the work of junior doctors (Crouch and Chapelhowe, 2008) Nurse prescribers are advised to maintain their standards of proficiency by adhering to a mandatory requirement for continuing and professional development (CPD), and keeping up with the latest developments in prescribing and pharmacology (NMC, 2006; Green et al., 2006). Nurses may reduce errors by making checks, and being double checked by other nurses in adherence to the 'five rights' of medicine administration (DoH, 2004). Pharmacists can reduce errors by being present on the ward drug rounds and sharing their drug knowledge (DoH, 2004; NPSA, 2007; Horns et al., 2002).

All the health professionals within a team may minimise medical errors by reflective practice, although there is a lack of empirical research to support its beneficial effects (Mamede et al., 2007). According to Walsh (2006), health professionals can reflect by one or more of the following methods: inviting feedback from colleagues about strengths and weaknesses; critical incident reviews to find out what went wrong (for example an allergic response by a patient to a drug), why it went wrong, and how to prevent repetition of an error; use of a diary for self assessment and identifying knowledge gaps.

Although health professionals aim to work as a team, nurses often experience barriers in reporting medical errors. A survey conducted by Uribe et al. (2002) investigated the reasons given by a group of health professionals in an American medical centre for not reporting medical errors. The top six barriers for the nurses in the group were:

'time involved in documenting an error, not being able to report anonymously, extra work in reporting, hesitancy regarding "telling" on somebody else, thinking that it is unnecessary to report the error because it had no negative outcome, fear of lawsuits' (p273)

A study of the perceived barriers to nurses reporting medical errors in a Taiwanese hospital indicated the chief reasons to be fear of physicians' reprimands, fear of patients' negative attitudes and fear of being recognised as incompetent (Chiang et Pepper, 2004). Other important reasons were the time to fill reports and the lack of positive feedback.

Medical error reporting should make every effort to promote anonymity, confidentiality and cooperation between members of a health care team in an institution and at the same time provide for accountability (Uribe et al., 2002). Anonymity encourages good rates of reporting (Anderson & Webster, 2001).

Uribe et al. (2002) made the following recommendations for improving reporting of medical errors: a centralised database for all health professionals within an organisation, a centralised taskforce for managing and analysing errors, education programs which give clear guidelines on reporting, and agreement on definitions of medical error and related terminology. Milch et al. (2006) recommended a secure, web-based portal on all hospital computers to promote voluntary reporting of medical errors.

According to 'standard 25' in the document 'Standards for medicines management', NMC (2008a), nurses are instructed to record adverse drug reactions (ADRs) in the patient's notes and 'notify via the yellow card scheme immediately'; furthermore nurses are advised 'to take action to remedy harm' caused to patients who experience an ADR.

In conclusion, it has been seen that 'medication management is a complex, multifaceted operation involving multiple people and numerous steps' (Agrawal, 2009, p681). Medical errors have the potential to harm patients at the prescribing, dispensing and administration stages of drug delivery (Aronsen, 2009) and may be person-based or system-based (Reason, 2000). Medical errors made by nurses in the administration of drugs may be a direct result of aberrations in the intrinsic thoughts and physical actions of nurses, or may indirectly result from errors made by medical prescribers (chiefly junior doctors) and pharmacists. A multidisciplinary approach to medication safety should therefore be adopted (DoH, 2004).

Five instances of potential harm personally observed on clinical placements in hospitals were evaluated using a literature search; these were the method of drug distribution, interruptions in administration of drugs, self-medication of patients, intravenous medication, and drug calculations. It was argued that medical errors by nurses could be reduced by: firstly, decentralised drug distribution rather than centralised drug distribution; secondly, interruptions could be reduced by nurses wearing 'identifying clothing' which informs other healthcare staff they are not to be interrupted, and using 'Please do not disturb' signs when administering drugs; thirdly, in the case of self-medication, it was recommended that drugs should be placed in lockable bedside cabinets, that nurses should spend time educating selected patients about self medication, and drugs, such as insulin, should be offered to selected patients in order to improve the time of dosing; fourthly, intravenous medication by nurses could be improved by education, a centralised intravenous medication service, 'smart' infusion pumps, standardised ready-to-use medicines, regular checking of drug charts, and improved communication between doctors, pharmacists and nurses; fifthly, nurses should double check complex dose calculations before administering drugs.

In order to reduce medical errors, nurses should adhere to the 'five rights' in drug administration', check dosing instructions with the most recent British National Formulary and follow the protocols for the administration of high risk drugs such as potassium chloride and insulin. Errors made by doctors could be reduced by computer order physician entry with clinical decision support and checking of prescriptions by pharmacists and nurses. Errors made by pharmacists could be reduced by improvements in drug labelling and storage, appointing a staff member to answer queries, and utilising an automated dispensing system.

Finally, it was emphasised that health professionals responsible for the prescribing, dispensing and administration of medicines should work together as team members in the ward environment, engage in reflective practice individually and corporately, and that nurses should be able to report medical errors anonymously to a central data base within an organisation. Another recommendation would be creating a hospital ethos which encourages reporting medical errors, and which blames the 'system' rather than the 'person' (Anderson & Webster, 2001). However, nurses should remain 'accountable for actions and omissions' in their practice and 'must always be able to justify' their decisions (NMC, 2008b).