Electronic prescribing – key to medication safety

Closed loop medication administration is the best way to ensure medication safety, and it starts with the medication history, Uwe Buddrus (Senior Consultant, Health Information Management Systems Society (HIMSS) Europe) told the audience. Research in the US has shown that the two most error-prone steps in medication use are prescribing and administration. Electronic medication records could help to avoid 50–60% of such risks, he said. However, the implementation of a complete electronic medication record system relies on a complex infrastructure. A seven-stage electronic medication record adoption model (EMRAM) shows how the relevant components work together.

The closed-loop medication administration process starts with an accurate medication history at admission. Electronic prescribing is the next step and, ideally, this should incorporate a stratified warning system so that users are not overwhelmed with trivial warnings. Many interactions and contraindications can be checked electronically, he noted. Validation of the prescription by a clinical pharmacist is essential for patient safety. In one hospital in Barcelona, every prescription is validated by a clinical pharmacist, 24 hours a day, seven days a week, but in many hospitals such validation is performed from 08.00 hours to 17.00 hours, Monday to Friday. Automated dispensing further helps to avoid errors. At the bedside, the ideal system is one in which electronic medicine carts are used to deliver bar-coded, unit doses. The patient, the medication and the prescription are then electronically matched before administration and an electronic record is made.

There is at present considerable variation in progress with electronic medication records (EMR). For example, the penetration of electronic prescribing is 28% in Germany compared with 71% in Spain and 94% in The Netherlands. Although a number of hospitals in Europe have reached stages 6 and 7 the EMR adoption model, no single hospital has the complete system, said Mr Buddrus.

A study in the US in 1995 had shown that reducing the number of steps in the medication use process from 17 to five reduced the rate of harm from medication incidents from 3.5 to 0.5 per 1000 patients. More recently, in a hospital in Barcelona, a 16% reduction in the cost of medication per patient and a 30% reduction in stockholding had been achieved through the introduction of electronic systems. Similarly, a hospital in Navarra reported a 9% reduction in adverse side-effects and a 29% reduction in adverse interactions after the introduction of computerised prescribing and electronic administration records.

In summary, Mr Buddrus said that what is needed for successful implementation of electronic medication records is an understanding of the problem, a range of solutions, as there is no ‘one size fits all’ system, and adequate finance. The other prerequisites are process optimisation and culture change – especially in relation to the role of pharmacists in performing the second validation.

In order for the closed loop medication administration (CLMA) to work most effectively, there has to be a unit dose dispensing system in place, argued Michael Baehr (Director of Pharmacy, University Medical Centre Hamburg-Eppendorf, Germany). Pharmacists have little influence over the safety of the medication use process when traditional ward medicine cupboards are used. The use of automated cupboards and barcoded medication increases the safety of the process but can be cumbersome for nurses. The provision of patient-specific, barcoded unit-doses shifts the workload and the responsibility for safety to the pharmacy and significantly reduces the nurse’s workload, said Dr Baehr.

In 2002 the University Medical Centre Hamburg-Eppendorf had a traditional drug distribution system and it was decided to move to a CLMA system. Over the past few years pharmacists have taken a central role in medication management – all paper systems have been eliminated and the overall process has been greatly simplified. Now, prescriptions are written electronically at the bedside and validated on the ward by clinical pharmacists before unit doses are dispensed in the pharmacy for the next 24 hours. Pharmacists visit the ward daily to validate prescriptions and discuss treatment with doctors and nurses. Validation could be performed remotely from the pharmacy but visibility is important, said Dr Baehr. The presence of pharmacists on wards has allowed them to make a number of important interventions, for example, advising on the duration of antibiotic treatment and dosage adjustment in renal insufficiency. A study published in 2013 showed that 93% of pharmacists’ recommendations were accepted.

At the University Medical Centre Hamburg-Eppendorf, unit doses are prepared in the pharmacy and provided in a ready-to-use form. This supports a paperless administration system, which is much easier than the previous paper system, said Dr Baehr. Closed loop medication administration is now in operation in ten intensive care units, 59 wards and the Accident and Emergency department. Doctors like the system because medicines are delivered promptly and the collegial relationship with pharmacists is welcomed.

Nurses like the system because they have been able to relinquish some of the error-prone steps in medication administration. The hospital management board likes the reduction in the risks of drug therapy and the transparency of costs. Best of all, the board likes the fact that University Medical Centre Hamburg-Eppendorf is the first in Europe to reach HIMSS stage 7. Finally, patients like the personalised doses and believe that their treatment is safer as a result; they also value the opportunity to speak to a pharmacist, he said.

A recent study has shown that when electronic prescribing and unit-dose drug distribution operate together the error rate is very sharply reduced – from 21% to 0.7% – compared with manual issue of doses at ward level.

In conclusion, Dr Baehr said that electronic patient records make CLMA possible. He advised pharmacists not to wait for the industry to pack unit doses but to undertake unit dose packaging in-house.

Unit dose distribution systems are recommended by the International Pharmaceutical Federation de Pharmacie (FIP) and Alliance for Patient Safety (APS – Aktionsbündnis Patientensicherheit) in Germany, said Jochen Schnurrer (University Hospital, Essen, Germany). In fact, it is essential to operate a unit dose distribution system with continuous quality control in order to reach level three of the APS medication safety checklist, he added.

Unit dose distribution systems are the safest for patients because each dose can be identified clearly and fully documented up to the point of administration. It is possible to match the patient, the prescription and the individual dose if all are bar-coded and linked to electronic prescribing and administration systems. As yet, most medicines are not supplied in suitable unit-dose packs and Dr Schnurrer called upon the pharmaceutical industry to introduce unit-dose packs as soon as possible. Currently, more than 20 German hospitals have implemented unit dose systems and the majority of these are packing their own unit-doses using blister packaging machines.

A number of factors contribute to successful implementation of unit dose systems, including electronic prescribing, understanding of the medication use process, good cooperation between doctors, nurses and pharmacists and stable products.

In practice, the majority of preventable severe adverse drug reactions are caused by a limited number of specific medication errors and it is important to have a prescribing system that can alert prescribers and prevent such errors, said Stefan Russman (Professor of Epidemiology, University of Zürich, Switzerland). The growing speciality of interventional pharmacoepidemiology has helped pharmacologists to identify critical errors or interactions and develop prescribing systems that will prevent them, he explained.

The combination of computerised prescribing with a clinical decision support system (CDSS) can prevent many adverse drug events by electronically detecting contraindications, drug interactions, the need for dosage adjustment and therapeutic duplications. However, analysis of the performance of decision support systems shows that only 6–8% of alerts about interactions are clinically relevant and prescribers quickly learn to ignore most alerts. If the most important errors, based on real-life data, could be identified, then highly-specific, real-time alerts could be built into prescribing systems.

At Zürich University Hospital, a pharmacoepidemiological database with medication error target identification has been developed. Over a two-year period just under 235,000 patient-days were logged and nearly seven million alerts were generated. Clearly, there were not seven million errors, commented Dr Russman, but 157,000 absolute contraindications and 42,000 major contraindications were identified. A case in point is co-prescribing of spironolactone with an angiotensin converting enzyme inhibitor – a combination that “trips every CDSS”, he said. All that is required is an inbuilt routine to check the patient’s potassium level. If it is satisfactory then no alert is required.

One example of a medication error of importance was paracetamol overdosage. A study in the US had shown that 6.6% of patients were exposed to doses of more than 4.0g per day. In Zürich, a rate of 6.1% had been found over a two-year period, including about 100 patients who received more than 6g paracetamol in 24 hours. The main cause appeared to be the use of fixed-combination products together with on-demand doses. Since January 2014, an automated surveillance programme has been implemented to detect and prevent accidental paracetamol overdosage. Another example was automated identification of patients with estimated glomerular filtration rates of less than 60ml/min who are prescribed metformin. An automatic alert is sent to the clinical pharmacologists who evaluate the patient and provide a dosing recommendation in the patient’s electronic record.

Manual preparation of cytostatic injections is physically tiring, mentally stressful and carries the risk of repetitive strain injury, according to Mieke Mertens (Pharmacist, Heidelberg Hospital, Germany). She explained how the hospital in Heidelberg prepares 75,000 doses annually and 60% of patients are treated as outpatients or day-case patients. The need to protect staff and to ensure safe preparation of injections were driving factors in the decision to install a robot to prepare cytostatic injections, she said.

The PharmaHelp (Fresenius) semi-automated system was selected. The robot works in a laminar air flow cabinet; it can prepare doses from concentrated liquids and reconstitute dry powders. Up to 16 doses, comprising up to four different products, can be prepared in a single run. The robot uses RFID technology to identify the locations of products. Intravenous bags can be prepared but not, as yet, elastomeric devices. Intravenous bags and syringes from all manufacturers can be used provided that they fit into the PharmaHelp clamps that hold them in position during the preparation process.

The PharmaHelp software creates a batch list from the electronic prescriptions but information can also be entered manually if required. The materials are assembled and manually loaded into the cabinet. Drug vials are identified using a camera and then fixed in the pre-determined position in the robot. Once all the materials are loaded, the robot runs and doses are prepared. A special vented needle is used to minimise the release of cytotoxic aerosols into the working environment. The final products are then manually unloaded for a final weight check and labelling. The loading and unloading processes take two to three hours per day and, in practice, two people are required to work with PharmaHelp.

In the first instance the robot at the Heidelberg hospital has been used for high volume products to reduce the workload for pharmacy staff. Use of the robot has reduced the contamination risk markedly, although efficiency could be improved, particularly in relation to the cumbersome loading process, said Ms Mertens. The accuracy of filling shows less than 3% deviation for large volumes although there can be problems with small volumes, she added. Part of the problem is that volumes are checked gravimetrically during preparation and it is difficult for a sensitive balance to operate in a laminar flow cabinet.

Kirsten Spindeldreier (Pharmacist, University Hospital, Mainz, Germany) described how the Apoteca (Loccioni) robot is used at the University Hospital of Mainz to prepare cytotoxic doses. The robot prepares intravenous doses in a laminar airflow cabinet. It can prepare up to eight doses in a single run and 30 doses per day.

The Apoteca software compiles a batch list and then the loading process is driven by on-screen instructions. The starting materials are identified by barcode before loading. The compounded products are compared with standard images. The robot is able to prepare injections both from concentrated solutions and dry powders, although some powders take time to dissolve, said Ms Spindeldreier. It can pack final products in syringes, bags or elastomeric devices. There is in-process checking of the weight of the product during preparation and final product labels are generated automatically. The robot has to be cleaned manually and this is followed by UV irradiation for a period of four hours. Initially the daily cleaning procedure took 45 minutes but, with practice, the time was reduced to 20 minutes. In addition there is a big weekly cleaning procedure. The only difficulty so far has been the removal of platinum-containing drugs resulting from minor spillages.

The staff at the University Hospital of Mainz have concluded that the use of the compounding robot has increased the quality of the cytotoxic compounding service considerably. They estimate that 35,000 doses per year would be the break even point.

During the discussion, it was noted that use of these robots rarely led to savings in staffing. Users argued that the real gains were in flexibility and safety, especially when handling hazardous drugs. In addition, the capital costs of a robot need to be amortised over several years. One study in the US showed that a robot was good value if the workload was above 38,000 doses per year.