Letters

Reducing error, improving safety

Defensive culture of British medicine needs to change

Log of errors is needed

Relation between reported mishaps and safety is unclear

No fault compensation protects patients in Nordic countries

Doctors could certainly take lessons from aviation

"Do all things practicable to reduce risk" should apply in health system

Health professionals should take responsibility for gross carelessness

Blaming individuals is more emotionally satisfying than targeting institutions

Medical errors must be discussed during medical education

Safety of systems can often be improved

Systems approach to intrapartum risk management is important

Crew resource management training should be mandatory in anaesthesia

Anaesthesia is different from anaesthesiology

Technology cannot replace healthcare workers

Defensive culture of British medicine needs to change

EDITORIt was brave to devote a whole issue to medical error¹how to recognise, how to investigate, how to analyse, and how to change systems to improve patient safety.¹ However, we regret that the edition was dominated by American studies, ignoring the British contribution of confidential inquiries and analyses of closed claims, which have significantly improved safety in some well defined areas of medical practice.

In the United States the insurance industry provided the impetus for the study of adverse events,² and in Australia the government funded a similar study³ because it was considering "no fault" compensation.³ In the United Kingdom, for 25 years the Department of Health has financed all successful claims against NHS hospitals and their staff. As a result the need to take a British study beyond the pilot phase may not be supported.⁴

Be that as it may, an important issue was not addressed in the BMJ. Behind each adverse event there is a patient, a doctor, and a doctor-patient relationship. A patient must be told when things have gone wrong. Every effort must be made to minimise the after effects, including financial compensation where necessary. Most patients wish to know in detail what happened and what is being done to reduce the possibility of a recurrence. And members of healthcare teams need mechanisms to come to terms with their fallibility. It is to be hoped that clinical governance will make a difference.

Meanwhile a change in the ethos of medical practice is required, and it is to this end that Action for Victims of Medical Accidents has set up a group for doctors. Action for Victims of Medical Accidents is often regarded as dealing solely with compensation and litigation, but its raison d'être has always been to improve patient care.

In February this year the doctors' group met informally to discuss how best to translate into practice the General Medical Council's requirements of "good medical practice when things go wrong."⁵ We are determined to take our discussion forward and would welcome input from others who see the need to change the defensive and exclusive culture of British medicine. Doctors who would like to be involved should contact Dr Anne Savage, who is acting as secretary to the group.

Victor Barley, consultant oncologist. 
Bristol Oncology Centre, Bristol BS2 8ED

Graham Neale, retired physician. 
London SW17 7BB

Christopher Burns-Cox, consultant physician. 
Wotton-under-Edge, Gloucestershire GL12 7PB

Paul Savage, retired surgeon. 
London NW3 5RA

Sam Machin, professor of haematology. 
Department of Haematology, University College Hospital, London WC1E 6AU

Adel El-Sobky, consultant psychiatrist. 
Hepscott, Northumberland NE61 6LB

Anne Savage, retired general practitioner. 
c/o Action for Victims of Medical Accidents, 44 High Street, Croydon CR0 1YB

Log of errors is needed

EDITORWe welcome the BMJ 's contribution to discussions concerning the study of medical error.¹ Of concern, however, is that although most clinical encounters in the health service take place in primary care, almost all deliberations on error to date have focused on delineation of error in the hospital sector. Error in primary care is neither well characterised nor well understood. As far as we are aware, there are few, if any, initiatives designed to document its occurrence or determinants in general practice.

A first step to studying error in this setting would be the creation of a log of errors.² If based on a voluntary, confidential, self reporting scheme, akin to logs used by the Federal Aviation Authority, this would enable systematic study of medical error without fear of reprisal. ^{3 4} Funded and administered at the level of the primary care group, in the context of clinical governance initiatives, such logs would enable patterns of error and latent deficiencies in service organisation and delivery of health care to be identified, including those that put patients at risk of avoidable harm. Although doubtless subject to underreporting, such a move would help to bring error out of the shadows of secrecy and blame and into the light of systematic description and study.

Aziz Sheikh, NHS research and development national primary care training fellow. 
Brian Hurwitz, professor. 
b.hurwitz{at}ic.ac.uk Department of Primary Health Care and General Practice, Imperial College of Science, Technology and Medicine, London W2 1PG

Relation between reported mishaps and safety is unclear

EDITORThe articles from the special issue of the BMJ on reducing error rightly point out that mistakes are inevitable and that the way to alleviate their effects is to have an effective reporting system.¹ This is invariably advantageous to the organisation in which reporting is confidential internally and to external assessors.

Publication requires great sensitivity because the good organisation will have fewer mishaps but a greater proportion of them will be reported, and the opposite will apply in the bad organisation. Almost certainly, the organisation with the most occurrences of errors will not be the worst either in safety or efficacy, and the one with the fewest occurrences will not be the best. Indeed, these opposing effects may mean that there is no relationship at all between reported mishaps and safety or efficacy. When several units can be compared over time, a benchmark might be established for the optimal level of reporting, which will never be the lowest. Until a benchmark is established, anybody publishing such information should explain prominently in the introduction that there is no standard yet, and the contrary effects on the apparent frequency of incidents of honest reporting and good practice mean that no league tables can be construed.

The problem was well illustrated by the first publication a year or two ago of the Aldermaston risk management reports.² These were interpreted by an unsympathetic press as indicating an unsafe organisation. In the absence of any standard by which to judge, the frequency of reported errors could have been seen, with equal justification, as indicating a safety conscious organisation taking every precaution to avoid mishaps.

The medical profession must get this message over if it is to cooperate with the publication of league tables of this nature.

C K Connolly, director of research. 
Darlington Memorial Hospital, Darlington DL3 6HX

No fault compensation protects patients in Nordic countries

EDITORWeingart et al reviewed the epidemiology of medical error.¹ We would like to point out additional ways of addressing patient and consumer safety, both analytically and practically.

It is important for healthcare consumers to avoid not only adverse events related to specific medical errors but also adverse outcomes where a causal relation to errors in practice cannot necessarily be established. Clinical decision making is complicated and often includes an element of "normal" risk taking. The level of risk regarded as acceptable depends on the clinical situation and may vary from doctor to doctor, from patient to patient, and between patients and their doctors.²

Information about the factual risks associated with clinical interventions is a prerequisite for rational decision making. The possibility of finding facts about the risks of adverse events depends on the features of the healthcare system generating the events. In Sweden a national patient insurance scheme gives financial compensation to patients who have incurred physical or mental injury as a consequence of medical treatment. Compensation is granted regardless of medical responsibility or malpractice. Other Nordic countries have established parallel insurance schemes. Since the Swedish insurance scheme began in the 1970s, more than 100 000 claims have been filed, and about 40% of these patients have been compensated financially. Data about these cases are available in an extensive database.

Although we do not know what proportion of all injuries are reported to the insurance scheme, the vast database offers possibilities to analyse, for example, injury profiles for different types of healthcare units and the severity of consequences of different types of errors. In contrast to studies of hospital data, the database also permits analyses of outpatient care, comparisons between different levels of care, and studies of trends over time.

Previous analyses have addressed a range of topicsfor example, specific types of medical error, injuries associated with the use of selected healthcare technologies, and gender disparities in the quality of care.^3-5 A more novel approach is to study reports from patients with selected chronic diseases, such as diabetes, and to use the results to inform and educate practitioners and patients.

As the risk of adverse medical outcomes will never fall to zero, the least we can do for consumers is to provide them with available information about the relative safety of their options in everyday health care.

Pia Maria Jonsson, research associate. 
pia.maria.jonsson{at}phs.ki.se

Göran Tomson, associate professor. 
Karolinska Institute, Department of Public Health Sciences, S-171 76 Stockholm, Sweden

Lars Räf, professor. 
Swedish Patient Insurance Fund, PSR, Box 17830, S-118 94 Stockholm, Sweden

Doctors could certainly take lessons from aviation

EDITORHelmreich's article discussing lessons to be learnt from aviation is useful.¹ The principles outlined are mentioned in several other articles in the same issue of the BMJ, but they mainly refer to commercial rather than general aviation (private, small business, aerial photography, medical services, police work, etc).

The fatal accident rate in general aviation perhaps reflects the problems, especially psychological, that affect pilots (often single handed) when they are not protected by the vast machinery of an international flying organisation and cockpit cross checks, etc. The three main causes of death in general aviation are loss of control (in either instrument or visual conditions), "controlled flight into terrain" (flying into a mountain), and fuel starvation.²

Many cases of loss of control have been due to failure to recognise lack of ability, being out of current practice, or overconfidence. Controlled flight into terrain occurs in instrument conditions and is usually due to pilots either being lost or failing to obey the rules for terrain clearance, or both. Most engine "failures" caused by running out of fuel defy belief. Yet although each of these groups of error is likely to result in a fatal outcome for the pilot and passengers, they still occur. Clearly the psychological factors involved are complex, but it is unlikely that any pilot set out with the intention of dying.

Airlines now have rigorous psychological assessment before appointing a pilot to training. In light of the accident rate in general aviation, a section on human factors and performance has been introduced to the private pilot syllabus. Whether this will help to reduce the human factors involved in deaths in general aviation remains to be seen, but some will inevitably still occur.³ Psychological assessment of doctors or medical students, or both, along with training in recognising personality types and error prone situations could be of benefit to both practitioners and patients and help prevent such scenarios as those given in Helmreich's article.

Ken McCune, research fellow and private pilot. 
Department of Surgery (Queen's University, Belfast), Institute of Clinical Science, Belfast BT12 6BA khmccune{at}emailmsn.com

"Do all things practicable to reduce risk" should apply in health system

EDITORFrankel et al state, "Even if [formal monitoring of mortality in general practitioners' practices] were restricted to deaths that occurred outside hospital, random variation would mask considerable illegitimate mortality."¹ What about deaths that occur inside the precinct of the practice, or within 24 hours of attending the practice? Would that be a more sensitive measure? After all, aren't deaths that occur within 24 hours of admission to a hospital referred to the coroner in many places?

Suspicion surrounding the high mortality associated with Harold Shipman's practice was raised. However, the health authority's inquiry did not follow up its request for a further five sets of case notes. Was this because doctors are perceived to be immune from the failings of humanity?

The health system must be required, by law, to do everything practicable to reduce the risk to others in the workplace. The airline industry provides a useful model.² Captains used to reign supreme; questioning their judgment handicapped career advancement. However, an accident where the co-pilot knew what was about to happen, but did not question the captain, resulted in captain management systemsCMSbecoming cockpit management systems. Then there was an accident where a flight attendant knew that a wing was iced over but said nothing because "who am I to question the judgment of the cockpit?" The "C" then referred to crew. Then came an accident caused by factors outside of the aircraft, so the "C" now stands for corporate. All incidents must now be reported and investigated to see how to further reduce the risks of flying (1 death per 8 million passenger flights).

Requiring employers to take "all practicable steps" to improve safety has reduced workplace deaths by about 30% over the past decade in New Zealand. Why hasn't this happened in medicine?

The "business" model has been imposed on the health system in many countries over the past 15 years, with neither the injection of capital nor the leadership needed to manage change effectively. This has resulted in a focus on economic efficiencies, but organisational objectives such as safety have been forgotten.

A non-punitive systems safety approach is proposed for our healthcare system. All accidents and near misses must, by law, be reported and investigated so that the system can learn what went wrong and change procedures to minimise repeat occurrences. Amputating the wrong limb or giving the wrong drug are unacceptable, and avoidable, errors and should no longer go unchallenged.

Ron Law, lecturer in management. 
Auckland University of Technology, Private Bag 92006, Auckland 1020, Auckland, New Zealand ron.law{at}aut.ac.nz

Dr Law is a member of the working group established by the New Zealand Ministry of Health to advise on the establishment of a nationwide mandatory medical error management system.

Health professionals should take responsibility for gross carelessness

EDITORSmith says: "The easy, understandable, and completely wrong answer to such an incident [removal of the wrong kidney] is to blame those who made the mistake."¹ However, a perfectly fair consumer perspective is: If you cannot tell left from right then are you fit to practise?

Although I understand all the valid reasons for avoiding a culture of wholesale blame, patients are entitled to require the people whom they trust with their lives to take responsibility and be held accountable for their actions. If the medical profession cannot cope with this reasonable demand, rebuilding public confidence in its trustworthiness will prove more of an uphill struggle than it need be.

It may be hard in so far as scarcely any doctors deliberately damage their patients, but the public expects privileged professionals to accept their obligations, including penalties for inexcusable carelessness. Perhaps readers can explain why health professionals should not suffer the consequences of gross carelessness like employees in every other trade and calling.

Roger M Goss, director. 
Patient Concern, PO Box 23732, London SW5 9FY rogerconcern{at}hotmail.com

Blaming individuals is more emotionally satisfying than targeting institutions

EDITORThe 18 March issue of the BMJ, on reducing error, raised some pertinent issues for me as a "second victim,"¹ and it was good to know that I am not alone.

The retraining period that immediately followed my medical accident showed me how blaming individuals is more emotionally satisfying than targeting the institution. The very existence of error seemed to damage my colleagues' professional self image such that they needed to correct and purge the source of the error.

The only way I felt able to protect myself was to maintain professional dignity while my character as well as my competence was being scrutinised. I still break out in a cold sweat when approached by someone saying "Can I have a quick word?"will it be criticism, judgment, or rebuke?

I do not know how someone without a caring partner could cope. I vividly remember mine (non-medical) spending four hours trying to write an essay set by my retraining supervisor on the causes of medical error. They say that the road to hell is paved with good intentions. That was four hours of helltears, grief, helplessness, and the torment of "What if?" and "Why?"

I can identify with the possible need for confession, restitution, and absolution, or at least resolution. Unfortunately, restitution for me was delayed for 18 months and came in the form of the coroner stating that "responsible" did not mean "negligent" in this case. It was too late in some respects, as this sensitive and reflective person, now deeply wounded, was burnt out.

Of course, this is all yesterday's news for some. I have learnt to live with itthe error, the fallout, the burn-out. As Reason stated, "It is often the best people who make the worst mistakes."² This is comforting. Someone else said, "Father forgive them."³ This is challenging. If I can forgive myself, and then those who reacted as though they were beyond fallibility, I suppose this would be resolution?

Medical errors must be discussed during medical education

EDITORThe profession has an opportunity to improve medical education, using the momentum generated by the Institute of Medicine's report on medical errors¹ and by general raising of awareness as in the BMJ of 18 March.² We used personal examples to increase awareness of the significance of errors.

We showed a videotape on errors in medicine to a graduating class of medical students. After they had viewed the tape we asked them to close their eyes (both to maintain anonymity and to increase response) and to raise their hands if they had been exposed to medical errors. We asked three questions.

• In your experience, have you seen a medical error that resulted in anything, from no harm to death? All 67 had.
• How many of those resulted in major harm or death? Thirty (45%) of the 67 indicated that it had done (95% confidence interval 33% to 57%).
• How many of those have you been personally involved with or have first hand information about? The response was 6/67 (9%; 2% to 16%).

Students were then asked to describe an error that resulted in death or major harm. Sixty two did so. We typed all comments and found that the errors fell into five categories: decision making, drugs, procedural, system, and others. The interrater agreement for the type of error was moderate (=0.55, P<0.001).³

Errors in decision making were noted in 12 responses (for example, wrong diagnosis; a pregnant patient sent home after abdominal trauma). Drug errors accounted for 18 responses (a switch of drugs with similar brand names (analgesic instead of antidepressant); a long acting drug crushed). Procedural errors accounted for seven responses (an error due to insufficient training; pneumothorax due to inadequate technique). System errors accounted for 15 responses (inability to obtain medical records; staffing shortage). Finally, other errors accounted for 10 responses (fear of correcting a superior; inadequate blood sampling). We further categorised the written comments as indicating errors that resulted in death (nine cases; moderate agreement, =0.57, P<0.001) and errors that we deemed preventable (43; fair agreement, =0.38, P=0.003).

How do we interpret the finding that 45% of graduating students are aware of an incident that has resulted in major harm or death, yet 9% had first hand information? If such estimates are accurate and representative they are astounding. We all face the challenge to change the culture of blame and to provide a safe forum for discussion among medical students.

Carlos A Estrada, associate medical director. 
estradac{at}mail.ecu.edu

James Carter, medical director. 
Clyde Brooks, associate medical director. 
Clinical Information and Support OfficeSupport Building, University Health Systems, Greenville, NC 27835-6028, USA

Ann C Jobe, senior associate dean. 
Brody School of Medicine at East Carolina University, Greenville, NC 27858, USA

Safety of systems can often be improved

EDITORWe agree with the findings of Espinosa and Nolan's study on reducing errors made by emergency physicians in reporting radiographs.¹ We work at a district general hospital's accident and emergency department that has operated an almost identical system for over 10 years, in accordance with the British Association of Accident and Emergency's guidelines.²

Key points in our department are the rapid return of all radiographs to the requesting physician; the reporting of the radiographs by consultant radiologists within 24 hours; the recall of any patients with errors made in interpreting radiographs by telephone; and the use of any such radiographs as a teaching exercise for all staff. Differences in the systems include reporting of plain radiographs within 24 hours in our institution rather than 12 hours, and an additional level of input in the marking of radiographs as abnormal by radiographers.

Using the experience of the radiographers adds another tier of safety to the system. The radiographer marks all abnormal radiographs with a red dot. This part of the system is audited regularly (last audit: sensitivity 93%, specificity 97%; audit period two weeks, 449 radiographs; true positive results 80, false positive results 6, false negative results 9, true negative results 354).

Having such a fail safe system has several effects: patient satisfaction is subjectively better, with the knowledge that all radiographs are reported; few complaints are made about misinterpretation; and a culture of learning and cooperation exists among junior staff.

Continuous audit data show a remarkably low rate of clinically important misinterpretation: 0.64% of plain radiographs per month (mean 6.84 events per month, mean 1069 radiographs per month; range 0% (0/1049) to 1.4% (16/1151) per month, data from 90 consecutive months). This compares with the rate of false negative errors of 0.3% (0.26% to 0.34%) in Espinosa and Nolan's study.

This is an excellent systematic approach to what is an error prone activity, reducing mistakes by accident and emergency staff (often junior), increasing patient satisfaction, and reducing long term patient morbidity and litigation. We think that this is the type of approach alluded to in another article in the same issue, by Barach and Small, applied in a medical context.³

Jonathan Aldridge, senior house officer in accident and emergency medicine. 
jon{at}aldridge007.freeserve.co.uk

Peter Freeland, consultant in accident and emergency medicine. 
St John's Hospital at Howden, Livingston, West Lothian EH54 6PP

Systems approach to intrapartum risk management is important

EDITORThe BMJ of 18 March highlighted the issue of medical error, the subject of our own programme of research. We fully endorse the strategy of viewing medical error as a system failure and the importance of seeking the full range of root causes underlying particular incidents. ^{1 2} This strategy supports prevention rather than the apportioning of blame to individuals.

Our current research in the labour ward has been stimulated by observational studies that reported higher rates of error and injury than might be expected.³ Our research entails analysing the system of care in labour wards in each of seven maternity units in the north west of England. Additionally, we will be studying five adverse incidents in each unit. Staff will be interviewed to ascertain the sequence of events. We will use the cognitive interview technique, which can elicit nearly 50% more information than traditional interviewing techniques.⁴ We will then analyse the findings using the prevention and recovery system for monitoring and analysis to try to establish the root causes.⁵ The results will be compared with the analysed system of care for the individual unit to provide evidence based risk management data.

In view of the serious hazards in this specialty, we believe that we should report our initial investigations. These suggest hypotheses about the sources of risk in current practice in the labour ward:

• Lack of formal training and updating on interpretation of cardiotocographs for midwifery and medical staff
• Inappropriate deployment of midwifery staff because of the team midwifery system, with the least experienced midwives being assigned to the highest risk patients
• Dilution of labour ward skills through use of rotational team midwives, who lack consolidation of skills and confidence
• Reduced familiarity with protocols, including emergency strategies, because midwives rotate to labour wards as teams
• Reliance being placed on bank midwives to maintain adequate staffing levels
• Increased rates of elective procedures, especially caesarean section, which can result in extra workload where there are tight staffing levels
• Inadequate or no supervision of junior medical staff during emergency procedures
• Transfer of asphyxiated infants before resuscitation can be started because resuscitation apparatus is sited centrally
• Increased time constraints through the duplication of written records on to computerised systems and correction of malfunctioning equipment.