Abstract
Abstract
Public interest in NHS safety has been fired by cataclysmic reports of thousands of patients being killed through medical errors. These terrifying but unreliable figures almost certainly overstate the size of the problem, but the matter is complicated since there are presently no reliable data and there is no common definition of what an avoidable medical mishap is. So a very serious matter is somewhat clouded by confusion.
Focusing on death and severe harm through unambiguous human error (a good place to start), the most obvious way to reduce such events is to educate frontline clinical staff about the science of safety – how this has been developed in other fields of human endeavour and how it translates to healthcare. This will take time – a professional generation, probably, as the young are more amenable to new practices than the old. There are other obstacles, including disincentives to report mishaps and a present lack of professional ownership of the safety initiative. But as awareness grows the pace of change will accelerate and better data will give a clearer picture of what is going on.
How big a problem is safety in the NHS?
‘At the lowest estimate, preventable medical errors in hospitals exceed attributable deaths to such feared threats as motor vehicle wrecks, breast cancer and AIDS.’ So said the US Institute of Medicine in its 1999 mongraph ‘To Err is Human’. 1 Picking up on the theme, six months later the English Department of Health declared ‘The best research estimates we have reveal enough to suggest that in NHS hospitals alone adverse events in which harm is caused to patients occur at a rate in excess of 850,000 a year’. 2
While completely agreeing that safety in healthcare is hugely important, I would like to counterbalance the stridency of these reports by quietly pointing out that the NHS is surprisingly resilient given the demands placed on it and the volume of traffic it has to cope with. I also believe that apocalyptic estimates of serious harm through errors, at least for the UK, dwindle somewhat under more careful examination. I say this not to encourage complacency but to suggest a less fevered approach to changing the attitudes and behaviour of doctors, nurses and those in allied professions. What we are trying to solve is not a recent or escalating problem, nor can it be fixed overnight.
How to begin? ‘If you can't measure it you can't manage it’ is a mantra beloved of health service managers. For patient safety, it is very difficult to establish a clear baseline recording how often things go wrong against which any improvements can be reliably assessed. It is difficult for two reasons. First, the frequency of mishaps depends, of course, on how you define a safety-related problem. At one end of a spectrum is someone obviously doing something wrong by error (bungles or blunders, as the tabloids like to put it) and at the other is a complication blighting some procedure or treatment that might reasonably be considered plain bad luck. Add to this accidents that happen to patients while in hospitals or clinics (they fall over, out of bed, into and out of baths and occasionally out of windows and down stairs). For every event there is then another spectrum relating to its consequences varying from none (most) through trivial (common) to severe or fatal (rare). So when the topic of patient safety is raised, not everyone is talking about the same thing. This is very important when trying to measure the problem. Those who include all accidents and everything on both spectra will have a figure several orders of magnitude greater than those who (like me), at least for the time being, focus entirely on blunders and bungles causing severe harm or death. And I cannot over-emphasize that this is not the same thing as ‘serious adverse events’ since these include unavoidable complications of treatment.
The second reason hampering measurement of the problem is the fact that historically, apart from nurses faithfully recording the slips, trips and falls of hospital patients, the systematic reporting of other safety events, at least in the UK, has been at best patchy and at worst non-existent. The danger here is that where no reliable data exist, unreliable data are used instead. This is well illustrated by recent estimates of annual error-related deaths in the UK. These have been based on crude extrapolation of arguably unsuitable data, giving rise to headline grabbing figures including 25,000 (in the Bristol Inquiry Report) 3 and 34,000 (in a National Audit Office report). 4 The highest estimate was a widely quoted 40,000, which first appeared in the Sunday Times in 1999 (source unclear). The figure was subsequently reported to a conference in Australia, and that presentation was cited in the British Medical Journal in 2004. 5 Despite their shaky foundations, these figures, being more than 10 times the number killed in car accidents, have not surprisingly provoked indignation in newspaper columnists and scared the general public.
As an NHS hospital consultant for over 30 years and latterly serving three years as medical director of the National Patient Safety Agency (NPSA) I have to say that these figures have never felt right to me. If, in England, around 150 patients died each year in every NHS acute Trust due to a bungle or blunder (the sort of number necessary for 25,000 to perish this way), the staff would be pretty well aware of it and the coroners would conduct rather more than the 7000 inquests arising from hospital deaths that is presently the case. I have no idea what the true figure is, but the best data on deaths through evident error that the NPSA could assemble in its first five years would suggest that it is much, much smaller than 40,000 – probably less than one-tenth of that figure, even allowing for a good proportion going unreported. (I should qualify that statement by saying that I include only deaths obviously due to error and not, for example, non-fatal errors that occur in the terminally ill immediately prior to their natural death – again, we have to be careful to understand what we are talking about.)
So, focusing on severe harm or death through human error, I would summarize by saying that the problem is difficult to quantify reliably, but has probably been overstated. The relative risk, of course, also has to take into account the opportunities for error as a denominator and once that is done the chances of being killed by an NHS blunder become very small indeed.
By setting the scene in this way I am not in any way trying to trivialize such tragedies, quite the reverse. By cutting them down to size I see them as a more clear-cut and manageable target than the mountain of what I consider to be less pressing incidents recorded in NHS risk management databases associated with little or no harm. I should also make clear that I am not including, for instance, the problems with resistant cross-infection which are a challenge more for microbiologists than students of the psychology of human error.
So how do we tackle the most pressing safety concerns?
My interest in safety was kindled by being involved in two enquiries into deaths of children with leukaemia. Both patients had responded well to treatment and had a good outlook but both were killed abruptly by the mistaken injection of vincristine intrathecally rather than intravenously. I was first aware of this happening in 1975, and last happening in 2001. In between there were at least another 12 such episodes in the UK. 2 There are approximately 3000 opportunities each year for this error to arise in children, so the chance of disaster is around 1:6500 or less. Long odds, perhaps, but after the first event it should have become zero.
This type of disaster is not unique. As I write the news is full of the sad story of a lady who died immediately following the birth of her son. Her death was the result of an epidural infusion of bupivacaine being switched with an intravenous infusion of saline and causing fatal cardiotoxicity. Like intrathecal vincristine, this is a rare but well known error that has been responsible for similar deaths previously.
There are two main reasons why these catastrophes occur repeatedly; first, the errors involved can obviously be made by many different individuals and so must be easy to commit, and secondly the lessons to be learnt from each event do not travel far outside the institution concerned (all the vincristine calamities happened in different hospitals).
Most serious errors involve one or more of three main areas: communication, medication and identification. Wrong drugs, wrong doses, wrong patients and wrong routes of administration jostle with wrong site surgery, incompatible blood transfusions and vitally important lab and X-ray reports getting lost or misfiled before being seen. So how do we stop these things happening repeatedly?
The answer is in one way simple. It requires all NHS professional and managerial staff to know and understand why and how people do wrong things and to design systems, tasks and processes in a way that makes it harder or impossible for them to do so. There is much to learn from safety experts outside medicine. This includes techniques such as root cause analysis of adverse events and breaking down errors into subcategories like absent-minded slips and lapses, mistakes through ignorance and violations of standard procedures. Most catastrophic errors arise because of multiple systematic failures. The final act may be carried out by an individual but that person is set up to fail by a series of pre-conditions that set the stage. For example, one intrathecal vincristine disaster was a mistake by a single individual but was preceded by two violations, a lapse and a mistake by four others beforehand, added to a breakdown in communication between doctors and the patient not being admitted to the usual ward because of over-crowding.
If professional staff better understand these matters and other aspects of what one might call the ‘science’ of safety they will be receptive to learning about the redesign of working practices, adapting these as appropriate to local needs and generating their own ideas to share with others. If, on the other hand, they do not and think safety is just about being careful, they are likely to be dismissive, defensive or unreceptive to the need for change.
How do we get the message across?
In a word, the answer has to be education. And we should focus our efforts on the young professionals in training – doctors, nurses, midwives, scientists, pharmacists, radiographers, physiotherapists and other allied professions. It is they who will effect real change. Safety should be an obligatory part of their early postgraduate training, and cross-disciplinary. The core knowledge about human error and the systems approach to controlling it should be assessed by examination. Later in training they should have more discipline specific instruction on the hazards of their particular craft group. By the time they reach career grade they should be well able to understand the need for safety to be built in to everything they do.
Older established professionals offer more of a challenge, and some will find it difficult to see the need for change, particularly if towards the end of their careers they have been lucky enough to avoid the trauma of seriously harming a patient by accident. Arguably the best way to sway these individuals is through their up and coming junior colleagues in whose eyes they will be most anxious not to appear out of date.
What are the obstacles to progress?
I see three chief obstacles to achieving measurable improvement. The first is collecting reliable data on serious mishaps to see how common they truly are and if they are falling in number over time. This is not as easy as it sounds. The NPSA's early attempts to sort through the material dumped in NHS Trusts' risk management databases showed how true the ‘garbage in, garbage out’ principle of IT systems is. A great deal of work by many dedicated individuals has continued to refine and sift the data, but so far its reliability in terms of offering a true snapshot of the overall picture remains highly questionable. These databases also do not contain all that they should due to under-reporting of important events. Efforts are in progress to triangulate information from different sources such as National Confidential Enquiries, the Royal Colleges and the NHS Litigation Authority in a ‘patient safety observatory’, but the challenge is considerable.
The second obstacle is the reluctance of senior professional staff, particularly doctors, to admit and report errors. This may be simply because they are busy people and do not see the point – particularly if no harm has been done. It may also be for fear of ridicule, reprisal or litigation. Litigation in particular does little to help foster an open and fair working environment where error reporting is second nature. Apart from the growing blizzard of civil claims for compensation, the number of criminal prosecutions for gross negligence manslaughter has also risen significantly in the last two decades, and the Crown Prosecution Service seems to have lowered the bar on this charge under pressure from angry relatives who seek revenge. 6 To counterbalance this there is a need for more hospitals to look hard for systematic failures rather than an individual to chastise when things go wrong – though 60 years of a blame culture in the NHS will not disappear overnight.
The third obstacle, perhaps the least appreciated and most important, is lack of professional ownership of the safety agenda. In 2001 the initiative came from the English Department of Health via the NPSA through Trust risk management teams in England and Wales, and so arrived at the frontline as another imperative from the NHS executive. It produced a broadly predictable response from the weary professional foot-soldiers already reeling from an avalanche of guidelines, targets, waiting list initiatives and other Things From Above. Despite this there were sufficient numbers who took the matter to heart and there is evidence of progress, but super-tankers like the NHS take a long time for the whole ship to change course. Meanwhile the English Department of Health, impatient for change to happen more quickly, has set up a new Patient Safety Forum to increase momentum. This has 29 members representing mostly NHS national regulation, guidance and management and is chaired jointly by the Chief Medical Officer and the Chief Executive of the NHS. There is evidently a strong will to put safety as a higher priority in all NHS institutions, and also a clear ambition to develop regional teams to drive it forward, but I still worry that many of those in the trenches who look to their professional organizations and senior colleagues for a steer on how to do their jobs will still feel that safety is being handed to them from NHS management and does not belong to them. If so, I believe this is the wrong way round. Clinical delivery teams will be harder to engage and slower to respond while they are merely stakeholders rather than shareholders or directors.
So what happens next?
The pace of change will not be as fast as the Department of Health would like. The aviation industry, often held up as a shining example of how safety can become embedded in a single field of human endeavour, took well over a decade to win the hearts and minds of senior air crew. By comparison, the NHS is much bigger, more complex, many more professionals are involved and the numbers of therapeutic procedures involving single patients are several orders of magnitude greater than flights involving many passengers. The procedures are also much more varied in terms of risks involved and their control.
Even where design improvements are agreed by everyone they can take many years to be adopted. Accidental intrathecal vincristine or intravenous bupivacaine would become impossible if the standard Luer connector was not used for both routes, but developing a different connector for spinal needles (the blindingly obvious thing to do) has got tangled in the barbed wire of international standards and commercial profitability and has been inching forward in EU committees for at least six years.
But we will get there. And while we pursue the educational imperative for young professionals there is still much to do. Better ways of identifying patients (bar codes or radio frequency tags), better ways of packaging medication to avoid confusion in emergencies, and better ways of making vital clinical information available to the right person at the right time are all being developed. And we should share information about what goes wrong and why. The NPSA must be allowed freedom to publish the data it collects on medical mishaps (it was stifled for the first five years of its existence by political nervousness at the Department of Health). Let some fresh air into the problem and we will probably find that while there is much to do, things overall are perhaps not as bad as we initially feared.