Christine Clark BSc, MSc, PhD,
The purpose of the seminar was to provide a platform for the sharing of information about the best ways for hospitals to improve patient safety. Introducing the seminar, Laurence Goldberg (independent pharmaceutical consultant) said that the patient safety agenda should not be compromised in the present economic climate and that one aim of the day was to highlight ways in which patient safety could be improved without placing too great a burden on overstretched budgets.
Major incident in Mainz, Germany
The deaths of three newborn babies in a university hospital in Germany sparked an investigation into the pharmacy aseptic compounding services which concluded that pharmacy procedures were robust but a contaminated commercial product had been used to prepare parenteral nutrition. Professor Irene Krämer (director of pharmacy, University medical centre, Mainz, Germany) described how the event had unfolded.
On a Friday in August 2010 parenteral nutrition was prepared for 11 children. Reference samples were kept and additional samples were sent for routine microbiological testing, in accordance with the departmental protocol. The next morning Professor Krämer was contacted because bacterial growth had been found in two culture bottles. Further samples were taken from the reference material for repeat microbiological testing. By this time, three children were becoming ill and all intravenous feeds were withdrawn and sampled for potential microbiological contamination. On the Sunday there were several more positive cultures and the possibility of the feeds being contaminated was raised. The original part-used containers of the bulk solutions were retrieved from the waste bin and samples of the contents were taken. Enterobacter cloacae and Escherichia hermanii were found in all 11 IV feeds and in the original amino acid bulk solution. A thorough investigation of the pharmacy was performed. Neither of the specific strains of the two organisms, Enterobacter cloacae and Escherichia hermanii was found in the clean room or in stool samples from the staff. Simulation tests in which 10% amino acid solutions were inoculated with Enterobacter cloacae and Escherichia hermanii showed that there was a long lag phase after which high levels of bacteria were reached and persisted for a long period. These findings suggested that the contamination occurred long before the IV feeds were compounded.
Further tests showed that there were 30,000 (3 x 104) bacteria per ml in the reference samples. However, these solutions were clear; turbidity only occurs when the bacterial concentration exceeds 10 million (107) organisms per ml. The original bulk amino acid solution was found to contain high levels of endotoxin – more than 1000 IU per ml (the acceptable limit is less than 0.1 IU per ml). Such levels would have needed a bacterial count of 2–4 x 106 organisms per ml in the bulk solution whereas what was found was only 1.2 x 105 per ml. The head of microbiology concluded that this pattern – high levels of endotoxin combined with a low bacterial count – pointed to a bacterial population in the dying phase, said Professor Krämer.
The evidence suggested that the bulk amino acid solution had been contaminated weeks or even months before it was used. The product, packed in glass bottles, had been made in Italy and delivered to the hospital in Germany two months before it was used. “Glass might not be as good a primary packaging material as you might think”, said Professor Krämer. Investigations by experts from Schott, the specialist glass manufacturer, had shown that a fine crack in a glass bottle can allow bacteria to penetrate into the solution, although the bottle shows no signs of leakage. In this case, there had been a hairline crack in the bottle underneath the label that had not been seen during the preparation of the IV feeds.
A further twist in this case was that all 11 children had received infusions of endotoxin, although amino acid solutions were routinely administered through giving sets with filters with zeta potential, intended to remove endotoxin. Unfortunately, it was not widely known that such filters were not effective in the presence of electrolytes because the electrolytes neutralise the zeta potential and allow endotoxin to pass through. Professor Krämer recommended that filters with zeta potential should not be used in this situation because they give a false sense of security.
At the end of the investigation the pharmacy department was completely exonerated. This was largely due to the rigorous adherence to procedures and the keeping of meticulous records, concluded Professor Krämer.
There is a compelling rationale for dose-banding (dose-rounding) for chemotherapy although it is not universally accepted and there is a need for a consensus amongst doctors before further progress can be made, according to Burhan Zavery (pharmacy operational services manager, Mid-Cheshire hospitals foundation trust). Chemotherapy dosing can be based on flat dosing or adjusted according to body weight or body surface area (BSA). Flat dosing schemes give the same dose to everyone and are safer to use because no dosage calculations are required. BSA-based dosing was originally devised to allow translation of doses derived from animal studies to humans for phase 1 trials. It was intended to minimise interpatient variability and not to calculate the optimal dose, explained Mr Zavery. It was adopted by paediatricians and later by oncologists. However, the formula on which it is based was derived in 1916 using just nine subjects. Drug toxicity does not correlate with BSA, he added.
Random errors in the delivery of chemotherapy doses can arise in a number of ways. These include weight and BSA determination, variability in vial contents, syringe accuracy and residual volumes. Individual patients also handle the drugs differently. Often the errors cancel each other out but they can also be additive, said Mr Zavery.
Records from one hospital showed that 189 different doses of cisplatin were prepared ranging from 20 -222mg. Similarly, 99 different doses of paclitaxel, ranging from 6-552 mg were prepared. In this latter case there were two clear peaks, one around 135mg and one around 175mg.
How does dose-banding work?
Dose-banding is defined as a scheme agreed with clinicians in which doses are rounded up or down within specified limits, usually within 5% of prescribed doses. A range of standardised, prefilled syringes can be prepared to deliver these doses. Drugs commonly prepared in this way include epirubicin, cyclophosphamide, 5-fluorouracil, doxorubicin, gemcitabine, oxaliplatin and methotrexate. About 60% of hospitals in the UK now use up to six cytotoxic drugs in a dose-banded form.
The problem with this type of dose-banding is that it is based on a simple arithmetic progression, with fixed intervals (e.g. 200 mg) between bands. This works satisfactorily when the first dose is calculated – the appropriate dose band is selected and the mid-point, ready-to-administer dose is given. However, difficulties can arise when a dose is adjusted, for example, because of toxicity. If a dose is decreased by 20%, the size of the difference depends on the size of the original dose – thus a 20% dose reduction for a dose of 2000mg is 400mg but a similar reduction for a dose of 1500mg is 300mg. The variance is larger with smaller doses and this could have important implications for small patients. One way to tackle this is to use a logarithmic scale of dose bands in which the midpoint of each band increases by a fixed percentage (see figure 1). Mr Zavery explained how a single scale which shows the percentage of the original calculated dose that is required can assist with this. The scale is constructed such that two bands below will always give a 20% reduction and two bands above will always give a 25% increase (see figure 2).
The beauty of this is that it provides a consistent relationship between doses and the maximum error is 6%. Furthermore, it means that the number of dose sizes needed decreases dramatically – in the case of cisplatin the number of different doses decreased from 189 to 20. Such an approach is well-suited to electronic prescribing where it can sit in the background, invisible to clinicians. However, for paper prescribing a look-up table is required.
In summary, Mr Zavery said that dose banding could reduce the potential for medication errors and protocol miscalculations. In addition, preparation of batches of standardised doses could reduce drug wastage from partly used vials and paved the way for ready-to-use and ready-to-administer products that had gone through formal validation.
Spending a week building a drug library for smart intravenous pumps would be time well spent because thereafter every intravenous dose would be checked using criteria defined by the pharmacy. So argued David Upton (clinical director, Sheffield Children’s NHS Trust, UK) as he described how hospitals are still failing to reap the full benefits of smart pumps.
The highest risk of medication errors occurs at the point of administration but at best only 2% are likely to be intercepted and therefore ‘a technological second check’ at this stage would be helpful.
The benefits of smart pumps are still not widely appreciated by health care staff. These pumps have three key functions – they calculate the infusion rate, hold an event log and contain built-in error-reduction software. The user need only input the dose (in mg/kg/hour) and the pump calculates the infusion rate required (based on the use of standardised concentrations). This discourages the risky practice of prescribing of doses in mls/hour, said Professor Upton.
Dose error-reduction software relies on having a ‘drug library’ or list of drugs and standard concentrations programmed into the pump together with upper and lower infusion rate limits for each drug. Building such a library is a valuable multidisciplinary exercise because it prompts reviews of current practice, explained Professor Upton.
The event log records each time a dose alert is generated and provides valuable information about patterns of use and training needs.
In most cases this type of information has never been available before because manual systems could not record or reject erroneous calculations.
Smart pumps have been available in Europe since about 2003 but currently uptake is estimated to be less than 10% of hospitals compared with 60-80% in the USA. Furthermore, many pumps are sold with ‘smart’ technology but it is not activated until the drug library is installed and this does not always happen.
Professor Upton suggested that the reasons for relatively poor uptake of smart pumps in Europe include lack of standardisation in hospitals, poorly targeted investment in new technologies, lack of evidence of effectiveness, under-promotion by manufacturers and a failure by pharmacists to embrace the technology as an integral aspect of medicines management.
Safe use of intravenous antibiotics
There are many opportunities for error associated with intravenous antibiotics, according to Hayley Wickens (microbiology pharmacist, Imperial College Healthcare NHS Trust, London). These include preparation errors such as mixing with incompatible solutions or drugs and calculation errors at the time of prescribing or administration. ‘Wrong route’ and ‘wrong rate’ errors also feature, she added.
In Dr Wickens’ hospital one third of patients are receiving antibiotics at any one time and 40% of these will be given intravenously.
One key measure that can improve safety is to provide adequate technical information about injectable antibiotics at the point of use, as recommended by the National Patient Safety Agency (NPSA) Alert, Promoting safer use of injectable medicines. The Medusa injectable medicines guide (http://medusa.wales.nhs.uk), which can be customised for local use, is a good way to do this, said Dr Wickens.
Therapeutic drug monitoring, intravenous-to-oral switching and prescribing for patients who are allergic to penicillin are all key measures in improving safe use of antibiotics. As a result of the high frequency of errors with neonatal gentamicin treatment the NPSA recently recommended that hospitals should follow a ‘care bundle’ of evidence-based interventions that included double-checking of prescribing and administration and administration of doses within one hour of the prescribed time.
Regarding IV-to-oral switching, Dr Wickens noted that intravenous antibiotics are rarely needed for longer than 48 hours outside intensive care units.
Prescribing for penicillin-allergic patients follows a traffic light scheme at Imperial College Healthcare NHS Trust. Penicillins are coded ‘red’ and are contra-indicated. Cephalosporins are coded ‘amber’ – for cautious use – because one in ten people with penicillin allergy are also allergic to cephalosporins, but if, for example cefotaxime were needed for meningitis it should be given because it is more effective than the alternative, vancomycin, explained Dr Wickens.
All other antibiotics are considered safe and are coded ‘green’.
Antibiotic stewardship describes a scheme that includes appropriate selection, dosing, route, and duration of antimicrobial therapy and aims to optimise clinical outcomes while minimising unintended consequences of antimicrobial use. Hospitals in England are now required to provide regular education about antibiotics to clinical staff, to have formal antibiotic policies and to undertake regular audits of compliance with feedback to prescribers. New guidance on antobiotic stewardship in the UK has recently been published to coincide with European antibiotic awareness day, which falls on November 18th.
Safety in homecare
An estimated £1.3 billion per annum is currently spent on medicines used in homecare in the UK but the true figure could be higher because there is no easy way to capture the information at present, Philip Aubrey (medicines procurement specialist, London and East of England) told the audience. About 100,000 patients receive homecare, including 15,000 who receive enteral products, and no fewer 13 homecare companies are now involved in providing the services.
The system is financially beneficial to the NHS because, unlike hospitals, homecare companies do not have to pay VAT on medicines. The types of drugs involved include intravenous chemotherapy, enzyme replacement treatments, erythropoiesis stimulating agents and antiretroviral agents. More than 50% of patients taking antiretrovirals receive them through the homecare route, said Mr Aubrey. Most homecare companies provide specialist nurses to administer medicines in patients’ homes. In general, homecare services are convenient and well-liked by patients. They also liberate time and resources in hospitals for complex in-patient care.
The rapid growth of homecare raises issues of clinical and financial governance for the NHS. From the financial point of view, NHS trusts need to have transparent service contracts and receive a clear breakdown of costs – something that has not always happened in the past, noted Mr Aubrey. They also need data on key performance indicators to ensure that value for money is obtained. From the clinical governance viewpoint numerous aspects of the service need to be clearly defined. These include dispensing arrangements and responsibilities, nursing support and standards of care and responsibility for serious untoward incidents such as anaphylaxis, adverse reactions or infections.
The current lack of data in homecare makes it difficult for the NHS to plan effectively. Mr Aubrey recommended that, ideally, all homecare should be processed through pharmacies so that all costs could be captured. In addition, there should be a standardised approach to managing homecare across health economies and homecare contracts should be based on the guidance and specifications developed by the National Homecare Medicines Committee (NHMC). The supply of medicines through homecare should always be in the best clinical interests of the patients and not merely a cost-saving measure, he added.
Pharmacists should ensure that measures for safe neuraxial administration are implemented in their hospitals, said David Lovett (principal pharmacist for aseptic services and clinical trials, University Hospitals of Leicester NHS Trust and member of the NPSA Safer Neuraxial Devices External Reference Group). Interest in this area had been driven by reports of accidental spinal administration vinca alkaloids and a number of ‘wrong route’ errors involving mix-ups with epidural and intravenous injections. He explained the key provisions of the NPSA Patient Safety Alert, Safer spinal (intrathecal), epidural and regional devices (2009/PSA004A & B). The alert is in two parts, the first of which calls for all spinal (intrathecal) bolus doses and lumbar puncture samples to be performed using syringes, needles and devices with connectors that cannot connect with intravenous Luer connectors by 1st April 2012. The second part calls for all epidural, spinal (intrathecal) and regional infusions and bolus doses to be administered using devices with connectors that cannot connect with intravenous Luer connectors or infusion spikes by 1st April 2013.