The evidence on which to base practice: different tools for different times

The tangible reward of clinical surgery is that with our own hands we change the course of disease and so improve and prolong our patients’ lives. That is what we find so compelling. From earliest times the goals of surgery were to staunch haemorrhage, to relieve obstruction and to drain pus. These simple tasks remain the most dramatically effective things we do. Every time we save a patient who is bleeding to death from a simple stab wound or a complex aneurysm; or when we relieve obstruction of a coronary artery, a heart valve, the oesophagus or trachea; and when we drain an empyema and see the improvement in a sick and toxic patient within hours, it is our knowledge, our skill and our craft that benefit the patient. These treatments are based on our knowledge of anatomy, physiology, pathology and the natural history of disease. We apply our knowledge in a rational and logical way; we see a direct cause and effect between what we do and the benefit we give. In terms of evidence this can be simplified to the parachute analogy—you would be very foolish to try to fall to earth without one [1]. No trial will now be done of the parachute. Much of surgery may be viewed in that way. No trials will be done to test the fundamental principles of hip replacement, cataract surgery or valve replacement for aortic stenosis. In each case there is a clear mechanical cause for the effects of the disease and in each case they can be rectified by surgery. In these examples the effect is large and the relationship clear so no trials are needed. When the effect is small, or the cause and effect relationship is less obvious, or it is obscured by other factors, we need better evidence. When challenged to justify what we do we have to convince those who pay. Whether it is an insurance scheme, a health maintenance organisation or a health service, we must be prepared to demonstrate that the intervention we offer is truly beneficial and cost effective or the health care resources should be used in another way. Remember that their vested interest is at least as great as ours. Health care concerns everyone. They are likely to be patients at some stage in their lives (Box 1) [2]. The challenge to justify what we do was spelt out by the three times Governor of Colorado, Richard Lamm [3]. It is a robust challenge. American infant mortality rates are the second worst in the developed world, in an economy that spends the highest proportion of any nation of its enormous wealth on health care. Can they justify chemotherapy in dying patients, repeated organ transplants and prolonged mechanical support, while other Americans do not have access to the basics of antenatal care? Lamm is not asking for less to be spent but better justification for how it is spent. John Yates [4] in his book Private Eye, Heart and Hip criticized British surgeons for spending too much time in private practice and neglecting the Health Service. He was not asking for more evidence but more delivery of highly effective operations. Lamm and Yates are examples of a well informed, rational and influential coterie who would control the health care purse strings. At the same time as conventional medicine is challenged to provide evidence that what we do is effective, there is an increase in alternative and complimentary medicine—a growing cult of anti-science. Take for example cupping, a therapy about two and a half thousand years old. Cups containing a little burning oil or alcohol are applied to the skin and generate suction and produce ecchymoses—like love bites! In modern America its devotees have the notion that cupping sucks out the badness of modern life. As practitioners of what is called modern medicine [5] we should remember that many accepted medical treatments will, in the future, be regarded as ineffective or actually harmful. There are different ways surgeons obtain evidence for what we do. Controlled trials are best and the first we know of was in 1747. Of nearly 2000 sailors on a 2-year expedition around the world, more than half died of scurvy. No one knew the cause. It was two centuries later in the 1930s that the Hungarian Szent-Györgyi identified Vitamin C. But James Lind, a naval surgeon, knew that scurvy was a disease acquired at sea and that sailors recovered when they returned home. He concluded that the diet on land must contain an antidote. Lind took 12 sick men and treated them in pairs with different additions to their aboard ship diet. The two to whom he fed oranges and lemons recovered in days. Lind is given credit for the first prospective controlled trial. The British navy from then on always carried citrus fruit—it was the limes that led to the British being known as ‘limeys’ [6]. ‘So long as all the physician had to offer the patient was his own time and advice and a few herbs and powders, both medicine and society could comfortably claim that the physician's duty of fidelity was owed solely to the individual patient. When physicians can, with the stroke of their pens, literally bankrupt the community, the community may no longer be able to tolerate that view of the physician's duty.’ Brody H, Miller FG. The internal morality of medicine: explication and application to managed care. J Med Philos 1998;23:384–410 (ref [2] in the main list). In the 19th century, the biggest problem blighting surgery was sepsis. The death rate from infection following compound fractures or amputation was up to 50% in London Hospitals [7]. The death rate for childhood fever reached 29% in the Vienna General Hospital in the 1840s. Nearly a third of women giving birth died—if they were admitted to the medical wards. Meanwhile on the ward run by midwives the death rate was a tenth of that—only 3% [6]. The obstetrician Ignaz Semmelweis performed an experiment. He had the nurses and the doctors swap wards—and the tenfold higher death rate followed the doctors. Then in 1847 his pathologist Jacob Kolletschka cut his finger and died of septicaemia. Semmelweis wrote: ‘Day and night the image of Kolletschka’s illness pursued me … as we found identical changes in his body and those of the childbed women, it can be concluded that Kolletschka died of the same disease’. He saw the medical students coming directly from autopsies with soiled hands and instruments. Was it not obvious he wrote that ‘puerperal fever is caused by conveyance to the pregnant woman of putrid particles derived from living organisms … carried on the hands of the very people who were there to ensure the safety of their babies and of themselves?’ Semmelweis ordered hand washing with chlorinated water and mortality fell dramatically. Neither James Lind nor Semmelweis had an accepted mechanism on which to base treatment. They relied on trials with control groups. These two trials were far from perfect: neither used blinding; neither had a system of unbiased allocation or randomisation that would meet present day standards; but they were good enough. The effects were large and both were controlled trials ahead of their time. A hierarchy of evidence used in development of guidelines is now familiar [8]. Systematic reviews, meta-analysis and individual Randomised controlled trials (RCTs) are best. Controlled trials as used by Lind and Semmelweiss follow and then collected case series, preferably prospective rather than retrospective (Table 1 ). Evidence and examples Empyema illustrates a logical application of our knowledge of the natural history of disease. Hippocrates is credited with draining empyema but the example I will use is the King of England George V. A bulletin on 21st November 1928 reported the King to have pneumonia. Antibiotics were unknown. Twelve days later he had fever and pleurisy pain. Another 10 days later the King was in a coma. His physician aspirated 450 ml of pus from his chest and summoned the Serjeant-Surgeon to the King, Sir Hugh Rigby who performed a rib resection and drained the empyema [9]. From then the King improved. Management of empyema relies on the logical application of basic knowledge gained over millenia. In England in the 1920s if they did it to the King it must have been the right treatment. That and the King’s survival were evidence enough. The lowest forms of evidence are dogma and doctrine. Under this heading I place practices that cannot stand up to logical scrutiny nor have been proven in an empirical trial. Here I would include cupping, bleeding and homeopathy. We should not dismiss treatments because we do know the mechanism but these fail both the test of a scientific rationale and empirical evidence. Homeopathic medicine is based on a doctrine promulgated by Samuel Hahnemann [6,10]. He was educated in Liepzig, Vienna and Erlangen and saw doctors poisoning their patients. His theory is based on infinitely small doses of the active material achieved by sequential dilutions until there is nothing but a memory of the substance in the remaining water. At least this ensured that patients could not be poisoned. Less benign were the emetics, purgatives and the regular and copious blood letting practised by our predecessors [6]. Bleeding persisted until the 19th century, 200 years after Harvey had demonstrated the circulation of the blood. These doctors were no less rational or intelligent than us. The human intellect has not evolved over the intervening few generations. They would have offered plausible justification based on their observations and experience, and the authority of their professional status. Furthermore patients expect something to done—a justification we hear today when unproven surgery is used on cancer (Table 2 ). Knowledge and evidence implementation The ways in which we apply evidence can also be categorised. The epitome of the good doctor patient relationship is to listen and understand the individual patient’s symptoms, fears, hopes and wishes and respond with a plan of treatment appropriate to that particular patient. Patients value highly this personal interaction with the doctor—even if scientifically the treatment itself is worthless. That is why people go to homeopaths and other non-scientific practitioners. At the other end of the spectrum of means of implementation are protocols and guidelines. These are the basis of public health measures such as vaccination programmes and health screening and for example prophylaxis against peri-operative DVT and pulmonary embolism. We cannot rely on the memories of individual surgeons to derive a policy to minimise the combined risk of thrombo-embolism and bleeding. The event rates are less than one in a hundred. We cannot calculate the relative cost effectiveness from personal experience. We rely on evidence from large trials, systematic reviews and meta-analysis. We derive guidelines and implement them in the protocols.1,2 But as we adopt ever more evidence based medicine (EBM) and implement it in protocols, we appear less personal in our dealings with our patients. The practitioners of alternative and complimentary medicine are only too happy to fill this gap. Most human beings need more in their lives than rational science so we should not begrudge the time spent in listening, comforting and explaining if we are not to lose our patients to ‘quacks and mountebanks’ [11] (Fig. 1 ). The graph allows us to explore the evidence level and implementation strategies in Tables 1 and 2. Works in my hands (WIMH) to evidence based medicine (EBM) is on the vertical axis and the spectrum of implementation strategies, from the most personalised to those entirely protocol based, is on the horizontal axis. We can depict these two different scales of evidence and its implementation graphically. On the horizontal axis I have abbreviated the concepts in Table 2 using the polarities personal and protocol. Evidence ranges from the familiar EBM to WIMH—‘works in my hands’ (Table 1). Draining an empyema, and knowing when and how to, are matters where judicious application of craft and experience are accepted and the same is true of most emergency and trauma surgery. They are in the bottom left corner. After hip and cataract surgery the lame walk and the blind see. Case series have born that out and they are offered according to agreed criteria within guidelines. They are centre stage in clinical practice and occupy the centre ground in my graphical analysis. Semmelweis and Lind are high on the evidence scale even though they did their experiments 150 and 250 years ago. Lind did not intend his naval surgeons to form a meaningful therapeutic relationship with individual sailors. He did not want the captain of the ship to use his judgement to pick and choose which sailors would get the oranges and lemons. They were all to be given citrus fruits. Semmelweis intended that all staff would wash their hands before all examinations of mothers to be (Fig. 2 ). We can use the graph to position the examples of evidence used in the text and seen in Table 1. But how does evidence change practice? How are colleagues persuaded and converted? Lind converted the navy to give all the sailors fruit but Semmelweiss did not get support. He was resisted, driven out, and they went back to their old ways. The difference between the navy and the medical profession is that a fighting ship can only work on the basis that all the sailors understand the chain of command and obey orders but doctors are fiercely resistant to orders and value their autonomy and their entitlement to clinical freedom. That brings us into conflict with a health service and hospital administration. Many believe we are right to resist being regimented but do we allow every junior doctor to take an individual view on whether to scrub before an On the of is a plausible mechanism there was that was a good for there was a on in the 19th about what caused in regarded as to and in had the between the and childbed He washing in water. He was by colleagues who that fever was not and not caused by American the [6]. of infection was not until the time of many years later (Box the same time the London his book of the with case in a a the in the third right was admitted to Hospital with and on the left book to be based on individual It is more than in the American was in the left of the chest a little the The American was over when was only 10 years old. He was to case of the this is a case of which is more based than of his own experience. It is based on the experience of in this it is to our and A case is a it is not the evidence for what we for The is so very small cannot out at But many people see a His that A of His with Of and On each of which a of His of a these have been But who to us that they must be so us is James is most for one of the worst in the history of surgery. of the heart has reached the by for all no and no can the natural that a wound of the A more was in the in by a physician in private that with the of surgery some of the of will be was based on his of of in the at a physician at Hospital and out the problem in is not only one of the most forms of but in its forms it all treatment by On at the in a case of this disease one is by the of ever a which will the to the blood in a the small and the that one could the as as one can after was by the It was years later in that and on a up the to cut the In performed the that had the He his finger the valve The was and for there was a clear cause and of and first patient also but a patient died within days of surgery and he his He the work in the a on the surgery for The British of medicine which out in the 1930s at first in but there were no they to to relieve stenosis. Sir in his of of the Heart wrote: to relieve of … by the valve have so far to give I they will to not only because the is too but because the is based … an … the valve is the of the the of clinical the surgeons that surgery is both and much for what I have credited as the application of basic The was but the was surgeons this a in an American Hospital in England to and in and around the heart and great on There were no evidence years after and that it was to work within the heart used the by at some In he a and those interest he to and the medical (Fig. ). On two patients with to his the first patient was on the and two more in the following He had in his first two American patients he at In the and each had a of in and years after was The natural history was very well and it was as in large case series The effect was and such good that no one the evidence. From the of what to be called the in Hospital in As we explore my graphical of of evidence of implementation it a one from to EBM because two of the are Can something be into a based on evidence the bottom of the evidence Most of cancer surgery and all the practice of from cancer are based on series The most and used protocol in is experience On the other we have when we have evidence but it is only to in some patients, so it The trial which no benefit from in empyema is an example The evidence is against that intervention but it many to be on a in my hands’ basis (Fig. ). The graph can be with examples from Table 1 and in the Another controlled trial which a to a was the trial. is and the patients are The of disease was in the and was when this surgery was as surgery and the As a all over America patients were in care on and the Health care were not prepared to the and on a trial. rather in this the parachute The trial was by the of the and and the for Health and I believe this is a for between and trials are the of the small too many are by who will to trials in their There are many in surgery where the mechanical benefit is very and trials were needed. and aortic are clear There are far too many to each one to an We have to practice by clear in many There is also a by patients and their surgeons that must do some The is that trials half the patients a of these surgeons to know that the of the trial will their treatment. There are with of cancer treatments in each trial there is a It the or it the best How many the How many the The it that there is an even around no We can is that are because many no no difference is a that knowledge we can choose the less or the less Another is that the trial and did their these as a the claim to was There are many for which the benefit that is but there are many things we do in cancer that be The history of cancer surgery has many and better were the to it would have been achieved by rates in series appear better because of ever more rather than by more is an surgery in the form of has been offered to patients by a of surgeons the with surgery were so and then therapy were these patients do better than those the That is the we to in the trial are too as too in surgery. They are but not some there is no other to When effects are small, time scales are long and there is a to trials are The evidence for chemotherapy after resection of cancer relies on meta-analysis of patients in trials There is no disease to We have no of other than survival There is no that any individual can on the basis of personal clinical experience, if there is more benefit than The more I in the right of this graph rather than the bottom the more I than I more good than