Hospital Pharmacist
Vol 10 No 5 p198
May 2003

EAHP: Strategies for safety

Recognising that there is a problem is the first step in preventing errors, said Professor David Cousins (head of medication safety, National Patient Safety Agency [NPSA]). A recent report has estimated that 10 per cent of patients admitted to hospital experienced an adverse event. Currently in the UK the NHS pays out £600 million each year in compensation costs, so there is clearly an issue.

Two key targets have now been set by the Department of Health — no deaths from incorrect intrathecal injections and a 40 per cent reduction in the number of serious errors with prescribed medicines by 2005.

Encouraging incident reporting is an approach that could be adopted to help meet those targets. That method has been used successfully in the air transport industry, where the number of incidents reported rose steadily between 1994 and 1999, but the proportion of high-risk incidents fell dramatically, said Professor Cousins.

It is important to encourage reporting of incidents in every institution as they provide important qualitative information. "If you do not have many reports then you do not have information to improve your systems," said Professor Cousins.

One or two incident reports per bed per year is a useful rule of thumb for an acute hospital. Typically, a large number of reports from a clinical area reflects a good reporting culture. Areas with low numbers of reports are more likely to be problematic. A statistic known as the Heinrich ratio lends weight to the argument because it shows that every fatal error is preceded by 329 minor incidents. If there is a good reporting culture then a comprehensive picture of the danger areas can be built up. It is important to check that incident reports are capturing all processes and are being completed by all disciplines, he warned.

A recent audit in the UK showed that, in some hospitals concentrated potassium chloride injection was still available on general wards, even after the alert. The NPSA had recently issued a safety alert requiring chief pharmacists to remove that product from wards. In spite of the numerous reports of errors involving concentrated potassium chloride injection, many pharmacists telephoned the NPSA to say that they had not had a problem. "What they mean is that they have not had a problem yet," warned Professor Cousins. Some also claimed that it was safer for nurses to dilute potassium chloride on the wards. But Professor Cousins thought it was unlikely that less risk was associated with clinical staff preparing injections in a busy area than with administering a ready-to-use infusion. The main problem, said Professor Cousins, is a failure to learn from past mistakes.

Packaging and labelling

Packaging and labelling play an important part in medication safety, Professor Cousins went on to say. The Medicines Control Agency (MCA) has recently drawn up a best practice guide on this topic. Key points were that the name and strength of the product should be on at least three sides of the pack and that large font sizes should be used to improve legibility. It is particularly important to show the total dose in an ampoule rather than only the concentration. For example, a product such as clindamycin labelled as 150mg/ml, packed in a 2ml ampoule, could easily be accidentally overdosed as it is easy to misread and give two ampoules instead of one for a 300mg dose.

Innovative pack design, incorporating the judicious use of colour, is to be encouraged to ensure the accurate identification of medicine. When considering the acceptability of a particular pack design, it will be necessary to think about the relative distinguishing features compared with other packs in a range (a range may mean all packs bearing a corporate livery or a group of packs carrying the same design theme). The primary aim of innovative design of packaging is to aid in the identification and selection of the medicine. Professor Cousins recommends that the MCA's best practice guidelines should be incorporated into NHS purchasing specifications. Often drugs and therapeutic committees make decisions about products on the basis of clinical trial evidence alone without ever seeing the products themselves. A full risk assessment of product labelling, packaging and information leaflets should be carried out before their introduction into a new practice site, said Professor Cousins.

Defining errors

Extracting useful information on the incidence of errors from studies can be difficult, said Mike Cross (Barts and the London NHS Trust, London, England). In order to make sense of such studies, it is necessary to understand how they were conducted. In particular there are issues associated with how errors are defined and recorded.

Early research showed that medical errors occurred in 10.8 per cent of admissions. But those sorts of studies vastly underestimated the scale of the problem, said Mr Cross. There were a number of key problems with published studies so far. They had focused only on measurable and preventable harm, and their assumptions tended to be conservative. It was not possible to find an error-free control group and data collection methods were problematic. Mr Cross pointed out that chart reviews routinely missed 35 per cent of errors, but computer-based reviews missed 50 per cent and voluntary reporting missed 95 per cent. Administration errors were almost completely missed by retrospective chart reviews and could only be identified in a meaningful way by direct observation, he added. A study by Barber and Dean estimated that 5.5 per cent of all oral administrations were made in error, and a study by Bruce and Wong estimated that mistakes were made in 25.2 per cent of IV administrations.

Extrapolation of data from the most comprehensive adverse drug event (ADE) prevention study to a large NHS hospital trust suggested that a total of 5853 ADEs could be expected annually, of which 697 would be life threatening and 58 fatal. Approximately 13,000 additional bed-days would be required as a result of these ADEs, at an additional cost of about £10.8m.

Administration errors would always exceed prescribing errors in secondary care because prescribing occurs only once but medicines may be administered on many occasions from a single prescription and each event presents an opportunity for error, explained Mr Cross.

Confusion also arises at interfaces between different sectors. For example, one audit showed that changes are made to drug treatment in 93 per cent of patients after a second medication history has been taken by a member of the pharmacy staff. That means that only 7 per cent of patients would otherwise have received optimal therapy. "This is a strong argument for a pharmacy medication history taking service," Mr Cross said. A study in London had shown that giving a list of medication changes made in hospital to the community pharmacist could avoid serious complications in one in 19 patients. Eight percent of patients discharged from hospital suffered an ADE within three weeks, he noted.

Error rates are often higher in areas where more drugs are used. There have been attempts to identify high-risk patients but that has not been fruitful. In general it is better to stratify risk by systems, including drug and error types. High-risk drugs or drug groups include anticoagulants, anti-epileptics, cytotoxics, opiates, antipsychotics, steroids, antibiotics, potassium, insulin, digoxin and amiodarone. High-risk error types include prescribing the wrong dose, incorrectly transcribing admission or discharge prescriptions, administering oral drugs via the IV route, confusing one drug with another, and missing drug interactions.

In conclusion, Mr Cross said that although the prevention of drug errors is understated and poorly understood, high-risk drugs and activities are now well-understood. That should aid the prioritisation of effective steps to prevent errors.

Children

Paediatric dosing information is still not available to the extent it should be, according to Dr Gregory Kearns (Professor of Paediatrics and Pharmacology, University of Missouri). Although the situation has improved since 1968, when there was little or no information about dosing in children for 80 per cent of prescription medicines in the USA, a lack of dosing information still stands between an effective drug molecule and a product that is safe. Scientific advances have often failed to translate into better drugs for children, said Professor Kearns.

A physician-driven study published in 2001 had shown that the rate of adverse drug events was three times higher in children than in adult patients. Many of these events were preventable because the majority (79 per cent) arose from incorrectly transcribed orders, said Professor Kearns. Orders were written by doctors but transcribed by medical secretaries and then sent to the pharmacy. Dosage errors accounted for 34 per cent of the recorded adverse drug events.

Professoer Kearns mentioned data from Medmarx, a system produced by the United States Pharmacopoeia for voluntary reporting of medication errors. This showed that 52 per cent of errors involving paediatric patients occurred at the administration stage, 23 per cent during dispensing from the nursing station and 16 per cent during documenting. Only 9 per cent of errors occurred during prescribing. Regarding error type, the largest category was "improper dose" at 29 per cent.

Surveys had repeatedly identified the fact that the correct dose of many medicines for paediatric patients is simply not known "If we knew the doses, then children would be healthier", said Professor Kearns. There are numerous rules for calculating paediatric doses but they do not always work. Comparisons of doses calculated using a surface area-based formula with pharmacokinetic measurements showed that, for example, for gabapentin, a dose of 40mg/kg was needed to achieve therapeutic blood levels, but the calculated dose would only deliver 13mg/kg. Had this dose been used the drug would have been dismissed as being ineffective in children, said Professor Kearns.

Turning to the question of research in paediatric pharmacology, Professor Kearns pointed out that there was now legal requirement in the USA for manufacturers to undertake research to define the correct doses in children. To date, more than 50,000 children have been involved in drug studies and there have been no deaths. Extended marketing authorisations have been granted for 38 drugs and the labelling has been revised for 20 drugs. In seven cases this has involved the addition of specific warnings.

Many of the studies have been pharmacokinetic studies. Cisapride is frequently used in neonates and a pharmacokinetic study by Professor Kearns' team had shown that the half-life in neonates of 28–42 weeks was 11.5 hours whereas in those aged 42–54 weeks it was 4.8 hours. Other studies concerned drug metabolism. For example, pantoprazole, a CYP2C19 substrate is poorly metabolised by 90 per cent of the population. As a result, poor metabolisers are exposed to drug levels 10 times higher than fast metabolisers.

Genetic differences in CYP2D6 activity also give rise to relative differences in drug exposure. People of Chinese or African American origin are often slow metabolisers, said Professor Kearns. In particular, valproate hepatotoxicity probably relates to racial or ethnic differences in drug metabolising capability, he said.

A further complication is that drugs work differently at receptor sites in children. For example, a much smaller dose of cyclosporin is needed in children aged less than one month compared with those aged more than one month to achieve inhibition of specific cytokines.

In future, it should be possible to improve treatment for children by linking knowledge of the patterns of age and disease in drug disposition with knowledge of the genotype for drug metabolising enzyme transporters and receptors, knowledge of the genetics of disease expression and the flexibility to tailor drug treatment, said Professor Kearns.

"Children participate in studies – we do not experiment on children", he said. Even if we feel uncomfortable about children in trials, we know there is no substitute for doing such trials and doing them well, he added. Understanding how to translate discovery into therapy and understanding the important role of pharmacy in this process will mean that the next generation will have a better life, he concluded.

Automation advances

Advances in automation were discussed by a team from Chaim Sheba Medical Centre, Tel Hashomer, Israel.

Implementing a closed-loop medication management system (ServeRx) has brought real advantages for doctors, according to Dr Eli Mizrahi (head of the geriatric rehabilitation department). "I get information in real-time about patients' treatment without leaving my room. If I see anything I don't like I can stop it or change it or call the junior doctor to discuss it", he said. Another benefit is that once he has entered a treatment plan into the computerised system, no-one can override it, he added.

Prescriptions are entered using a hand-held device that is light and easy to use. Commenting on the software design Dr Mizrahi said, "Three-quarters of our juniors are new immigrants — they think in their mother tongue, speak in Hebrew and write prescriptions in English — and they have had no difficulties with this software." He predicted that this system would become a tool that they could not work without. Nurses have found the system simple to operate, according to Ms Hanna Sherman (head nurse, geriatric rehabilitation department) and patients have also responded favourably.

The ServeRx system uses bar-code verification to match prescriptions, medicines and patients. It has now been in operation, in a pilot scheme, for five months at Sheba Medical Centre. The ServeRx system, which can be free-standing or integrated with existing hospital systems, controls and logs the processes of prescribing, dispensing and administration. It comprises a custom built, computer-controlled medicines storage unit (that includes a fridge and a secure cupboard for controlled drugs), a medicines trolley with an integral computer and a hand-held device for prescribing. Medicines are dispensed from the storage unit and placed in the trolley individual patient drawers, marked with an LCD label. Reading the patient bar code releases the appropriate drawer and medicines administration is recorded on the computer screen. When the trolley is docked with the storage system after use the central computer is updated with the administration record.

The hospital's strategic vision emphasises patient safety, supported by electronic health records and electronic prescribing with inbuilt decision support systems. It has 1200 acute beds and 880 chronic beds and a staff of 30 pharmacists and technicians. "It is quite a struggle to maintain safety", said Dr Amitai Ziv (medical director). Dr Ziv feels sure that, as in most hospitals, there is under-reporting of errors. In 2002, 170 drug-related errors had been reported. None had been fatal but two had been serious and resulted in admission to the intensive care unit. The ServeRx closed-loop system meets the hospital's requirements for risk management, he concluded.

The impact of the system is being assessed by means of a prospective study comparing two geriatric rehabilitation units with similar case mixes, one using the ServeRx system and the other using the usual system, explained Dr Ahuva Weiss-Meilik (head of performance and outcome improvement unit). Although a full analysis of the results is not yet complete, some key preliminary findings have emerged. Analysis of the workflow has shown a marked reduction in the complexity of the process required to prescribe, dispense and administer medicines. The number of discrete tasks involved has fallen from 46 to 33 (a 28 per cent reduction). Similarly, the number of decisions that have to be made during the process has fallen from 13 to four (a 69 per cent reduction) and the number of places where patient details are recorded has been reduced to one. Future analysis will show whether these process changes translate into improved outcomes.

Ms Miriam Adam (chief pharmacist) commented that the system should allow better use of pharmacy technical staff. Ms Ann Jacklin, chief pharmacist at the Hammersmith hospital, where the UK pilot of ServeRx will take place, pointed out that, unlike paper systems, the electronic prescribing system will identify prescription changes thereby allowing pharmacists to target their efforts more effectively.

Experiences with automated systems in The Netherlands have thrown up some new problems. Mr Bertil Lenderink (director of pharmacy Midden-Brabant, Tilburg) said that problems were associated with bar-coded bracelets. Some patients had taken them off, some bracelets had been difficult to read, either because they had been damaged when the patient took a shower or because the scanner could not read the barcode on the curved surface, and some of the barcodes had been inaccurate.