Medical Error and Technologies Research Workshop, London, 3 November 2004

Summary of presentations

Maisoon Ghaleb (School of Pharmacy, London University) described work on paediatric prescribing errors. The vast majority of medication error studies have been done in adults, with only 22 published reports found relating specifically to children. Taken together, these studies suggested that medication errors - particularly dosing errors - were more common in paediatrics than in adults. However, no drug chart-based review study had been done in the UK. The aim was to test a method for investigating the incidence and nature of paediatric prescribing errors in one large children's hospital.

Over a two-week period, senior pharmacists visiting intensive care unit, surgical and medical wards were asked to review drug charts using definitions of prescribing error developed by the research team. A researcher went with the pharmacists to record errors identified. A total of 1066 medication orders hand-written by doctors for 82 patients were assessed. The majority concerned the way the prescription was written, with 45% of orders lacking essential information and 8.8% classed as illegible. Use of electronic prescribing systems could reduce these types of error. Thirteen dosing errors were found, with rates ranging from 1.2 to 2.5 per 100 orders written. The rate on the surgical ward was more than twice than on intensive care unit.

Sylvia Birch (research pharmacist, Charing Cross Hospital) described the Serve-Rx computerised medication system being piloted on one surgical ward. Doctors prescribed on portable computers, there were no paper drug charts, and patient medication was dispensed into individual drawers in a computer-controlled drug trolley, which only opened after reading the barcode on a patient's ID bracelet.

Sylvia is a member of the Hospital Electronic Prescribing Pilot Assessment [HEPPA] team, which was investigating the impact of "paperless" IT systems such as this on medicine errors. She was using two retrospective methods: a detailed record review and a Trigger Tool in which the ordering of specific drugs and tests were taken to indicate possible adverse events. Notes for patients on the ward before and during the Serve Rx trial, were being reviewed. This had had taken longer than anticipated because of difficulty in finding records, and finding information in them. The review was not quite complete but very few errors had been detected using the Trigger Tool. The notes review suggested more administration errors with the old paper charts, although many these were medicines recorded as "not given" with no reason supplied. Overall prescribing error rates had probably not changed much, but the types of prescribing error were different with the electronic system.

Gerry Armitage (Bradford University) described the work he was doing on the factors contributing to drug error, and its reporting. Drug errors could be traumatic for all concerned. They also cost the NHS money. Yet current reporting systems only picked up the tip of the iceberg.

His study will examine the process of drug error reporting, and local actions taken as a result using relevant cultural and error theories. The ultimate aim is to produce a better way of reporting incidents.

The first stage of his project has involved quantitative and qualitative analysis of a random 50% of written error incident reports made in one NHS hospital trust between 1999 and 2003. There had been a clear increase over the 5 year period, with a huge jump in reporting rates the year when an allergic patient had died after being given IV penicillin. That incident had increased awareness and reporting of that type of error.

Free text comments made on these error reports will help to inform the next phase of the work, in which a snowball sample of up to 40 health professionals who have experienced drug errors will be interviewed. Recruiting these people and structuring the interview would both be difficult. There was also the question of how generalisable the data generated would be. His trust was a "zealot" and wanted to convert to a learning organisation; others were not all as supportive.

Garry Barton (Nottingham University) is starting work to explain the variation in the use of non-steroidal anti-inflammatory (NSAID) drugs. GPs write 15m prescriptions every year for this type of drug. It has been estimated that each year there are up to 12,000 adverse events (bleeds, ulcers, perforations) related to NSAID use, and 2,500 potentially avoidable deaths.

Official NICE guidance is that patients at high risk of gastrointestinal events should be prescribed the newer selective NSAIDS (COX-2 inhibitors). This would include people with a history of peptic ulcer. But what impact have these guidelines had on GPs?

Garry plans to extract data from GP computer systems to assess the impact of these guidelines on clinical practice, and to identify doctor, practice or patient characteristics which make NSAID prescribing more or less likely. As part of this he will be doing qualitative work on GP and patient reasons for using the drugs, and the trade-offs between pain relief and adverse effect risk.

It should be noted that on December 21st the UK Medicines & Healthcare Products Regulatory Agency announced it was reviewing the cardiovascular safety of COX-2 inhibitors (http://www.mhra.gov.uk)

Growing attention is being paid in Britain and the United States to the safety and efficacy of selective serotonin reuptake inhibitor (SSRI) antidepressants particularly in children and adolescents. Linsey McGoey (Biocentre, LSE) will focus on the controversy over use and regulation of the SSRI Seroxat, manufactured by GlaxoSmithKline.

GSK has faced, and continues to face, a number of legal actions surrounding the testing, disclosure of clinical results and marketing of Seroxat In the UK, the Medicines and Healthcare Regulation Agency is allegedly considering bringing charges against GSK for with-holding data.

Linsey's research will be on the politics of the controversy surrounding the regulation and use of Seroxat, investigating the perspectives of the varying parties involved. These include the drug company, government regulators, psychopharmacologists and patient groups. The current Seroxat controversy will be compared with earlier debates over the safety of psychopharmaceutical treatments, such as concerns surrounding the use of the benzodiazepine tranquilizers. The use and regulation of Seroxat will also be considered in light of literature from the field of neuroethics, which addresses some of the concerns raised by the use, regulation and potential of medicines of this type.

Reshma Gandecha (Brunel University) is focusing on the National Programme for IT, particularly the effective delivery of the integrated NHS Care records.The NHS vision is that every person in England would have their own life-long electronic health record (EHR). Used with a unique patient ID number, this would provide a single source of reliable up-to-date information for all health professionals, so reducing errors, delays and fraud and providing a rich data source to monitor health trends. Patients would have limited access to their data, but would be able to change basic information. These records would be managed by the NHS Care Records Service, but other powerful organisations (eg National Patient Safety Agency) also had a stake.

Socio-technical perspectives such as Actor Network Theory (ANT) used in this work offer the potential to study the way that complex healthcare information technology/systems like this are developed. ANT views organisations as actor networks who enrol other actors, (both human and non-human) thereby forming and transforming a network for some purpose identified by a 'focal actor'. Within this perspective, actors 'translate' an object such as the "patient record", forming a sort of hybrid which is distributed and accessed by other actors in the evolving network.

The direction and pace of this change to the actor/record will be dictated by which actors interacts with it at any given time. In turn those actors who form alliances with the record are also changed as a result. The idea of a health record becomes translated into that actor's particular language and the situation prevailing at the time of its use. Clinical records are seen within this view as 'mutable mobiles' capable of mutating to accommodate to the prevailing circumstances of use.

IT has the potential to introduce problems as well as to solve them. Penny Ross (Portsmouth University) is investigating the methods used to display biochemical and haematology test results on screen to clinicians, and the effect this has on clinical decision-making. She said that clinical information systems should be viewed as 'Critical Systems' as healthcare data was different from other data. It was more complex and more sensitive, and the environment and culture within which the system has to operate was also significantly different to that of other information systems.

Her research team was investigating the factors that affect a clinicians' ability to recognise critical data on the computer screen, especially when they were busy, or often interrupted, or just tired. They were designing a series of experiments in which test result data were presented in a number of formats.

Did the order in which results were presented make any difference? Should certain results be grouped together? How many test results should be shown at once before the display becomes confusing? Were charts, graphs, patterns, or trends helpful?

The question of which reference values to use was another topic to be studied. For example, for patients on chemotherapy, would using "normal for chemo" make critical abnormalities easier to spot, or would it be better to show deviation from the normal range?

The aim was to provide empirical evidence that carefully designed IT systems can improve patient safety and work in a routine setting. They did have something which "worked" at her study site, but it had taken a while to get all the oncologists to agree on something they would all use. Now the big question is - would this set-up work the same in another department or in a different hospital.

Sisse Olsen (Clinical Safety Research Unit, Imperial College London) described the early days of a collaborative project with colleagues Charles Vincent and Catherine Tighe. The aim of the project was to develop an IT information-sharing system, based on the hospital intranet, for junior doctors. At the moment, JHDs ending a shift handed over information about their patients on scraps of paper. There were lots of pitfalls.

The system they were developing was intended for junior doctors and nurses, and would be used at nights and weekends. The system requirements would be defined through observing current practice, and talking to staff to define the types of problems and identify possible solutions. The precise clinical information to be logged will be decided with clinical staff, but observations and tests would be entered at specified intervals, and clear alert signals given for nursing staff. Junior doctors would need to define specific warning signs, not just tell nurses to ring them if a patient "goes off".

The system would be trialled for a month, with researchers on site for the first two on-call periods in case of problems. Once the system is up and running, they would investigate the impact on communication and attitudes to safety.