Professional Reference articles are designed for health professionals to use. They are written by UK doctors and based on research evidence, UK and European Guidelines. You may find one of our health articles more useful.
People often prefer certainty in their lives. Generally people avoid risk and uncertainty. There is an expectation that professional people, including doctors, will give clear and unequivocal advice. However as the scientist Benjamin Franklin wrote: "Nothing in life is certain except death and taxes".
For healthcare professionals coping with uncertainty presents a number of problems:
- The professionals own experience of uncertainty. For example confidence may be affected and this may influence patient management decisions.
- The patients uncertainty. This will, for example, influence delivery of care including explanations of uncertainty, the nature of risk and risk management. This in turn will shape patients choices and decisions about their care.
Balancing risks and benefits can be very difficult. Advising and informing patients in this context is also inevitably difficult. It is possible for the health professional when trying to present choices to appear uncertain, or worse, unhelpful and disinterested on the one hand or too persuasive and directive on the other. Ultimately a bad outcome may be taken as an indication of incompetence or malpractice, but also raises issues about the quality of advice and explanation given to the patient.
The development of evidence-based medicine has been a major advance in healthcare. It can help when making choices in healthcare. The gold standard of clinical research is the randomised controlled trial (RCT), preferably with double blind placebo control. Sometimes the rarity of a condition makes large trials impractical. What do they prove?
The best that they can hope to achieve is to demonstrate that there is a low probability that the null hypothesis is true and the intervention is ineffectual. That is not the same as absolute proof that it works. People sometimes say, "You can prove anything with statistics." Quite the opposite is true. It is possible to prove nothing with statistics, only to show that there is a very low likelihood that the observations were a random occurrence.
Very often proof of the efficacy of an intervention is based on a P value around 0.05. This means that even without poor methodology and selective publication, that there is a chance of around 1 in 20 that it is totally useless. Even P<0.001 is not absolute proof. A chance of 1 in 14 million may seem as good as impossible but at least one person usually wins the National Lottery every week.
Meta-analysis is a technique that is often highly rated by reviewers but it is fraught with danger. The summated trials must have similar methods for selecting participants and measuring results. Any incongruity makes the summation meaningless. It is assumed that all are of good methodology but this is impossible to assess simply by reading the paper. If there are errors, the errors may be multiplied.
A problem with meta-analysis is that of compounding the effect of selective publication. For this reason, meta-analysis should really be accompanied by funnel plotting and 'cut and fill'. One example is the meta-analysis of selected, 'good' papers about homoeopathy. These showed benefit. However a few years later, further analysis by funnel plotting with cut and fill showed that selective publication had occurred. When account was taken of other unpublished papers, benefit was not demonstrated. Negative results are just as important as positive ones.
When considering evidence affecting treatment choices we should ask:
- What evidence exists?
- How valid is the evidence?
- How applicable is this to the patient in front of me?
A lot of research is done using patients who have been referred for specialist care. This can pose problems for the primary care physician. For example:
- There is a great deal of high quality evidence about the management of heart failure. However for methodological reasons subjects have been selected. For example patients are of similar age (often around age 55) with ischaemic heart disease but with no other pathology. These selected patients are quite different to the typical or average hear failure patient in primary care (average age 85 with multiple pathologies and on lots of other medication). This needs to be considered when looking at evidence and is not a reason to reject what may be the best available evidence.
- Similarly, although most depression is treated in primary care, much of the research evidence has been gathered from patients referred for treatment to psychiatrists. The benefits of antidepressant treatment are likely to be different in these different groups of patients.
- When participating in screening patients seek clear results.
- Any screening procedure will have specificity and sensitivity:
- Specificity relates to the number of false positives (positive result when no disease).
- Sensitivity relates to false negatives (negative result in the presence of disease).
- No screening system is devoid of uncertainty. Criteria that are too stringent will result in too many false negatives whilst being too lax will give an unacceptable number of false positives.
The ability to calculate risk, as with risk for coronary heart disease or stroke, can be useful when deciding which patients are most likely to benefit from treatment. However it is essential to explain the results to patients and understand the levels of risk. No one can predict with absolute certainty who will for example develop heart disease. We can adjust some risk factors to reduce risk.
A patient with cancer may have been advised of the average 5 year survival figures for their condition. Remember to weigh the balance of optimism and pessimism. A 20% survival means both 4 patients in 5 being dead within 5 years, and 1 person in 5 surviving. A 90% survival is very encouraging but 1 person in 10 will die. There is usually more value for patients in emphasising optimism.
Patients who understand risk can understand treatment decisions better. If possible use a range of techniques and try to have unbiased visual aids available.[4, 5] There is an increasing body of evidence supporting the design of effective evidence-based communication tools but there is variable access to such tools in practice.
What is an abnormal result? Many parameters have a Gaussian distribution and often laboratories take arbitrary cut off points such as 2 or 2.5 standard deviations from the mean. If 2 standard deviations is taken as the line then 5% of people will fall outside it with 2.5% above and 2.5% below the range. Hence 1 person in 20 will be abnormal for that parameter and if 20 different parameters are measured, most people will be abnormal for at least one. Nowadays laboratories tend to call it the reference range rather than the range of normal.
If a result is unexpectedly abnormal the test should be repeated. However, remember also that a normal result may give false reassurance. There is no absolute guarantee that a result, positive or negative, is correct. The interpretation of prostatic specific antigen (PSA) is especially fraught and is unreliable as a screening tool for prostatic carcinoma.
Making a diagnosis involves balancing evidence. Several differential diagnoses may have to be considered. Diagnosis involves selection of the most likely option. An obvious diagnosis may eventually prove to be wrong. In the words of Robert Burns, "There is no such uncertainty as a sure thing". It is essential to keep an open mind and to consider the possibilities. A degree of self-confidence inspires confidence but should not blind either doctor or patient to other possibilities.
Between 20 and 30% of appendices that are removed for acute appendicitis, are completely normal. The surgeon who removes more than 80% abnormal specimens may not be cautious enough and be waiting too long to intervene. Acute appendicitis and myocardial infarction are common diagnoses but not easy diagnoses. Admission to hospital with these and many other conditions is based on the premise that the risk of a serious diagnosis warrants admission. This is often the case when there is less objective evidence for a diagnosis. This is particularly the case in primary care and when the patient is a child. Making decisions when there is both uncertainty and risk is very challenging. Inevitably some patients are admitted, observed and discharged. There is again a balance required between failing to admit patients with serious illness and overburdening the emergency system with trivial illness. Careful history and examination can aid discrimination but it will not provide absolute certainty.
Is treatment safe?
There are many claims for herbal or "natural" remedies that are 100% effective and without side-effects. There is no therapy that is completely safe and entirely effective. Which was the largest island in the world before Australia was discovered? The answer, of course, is Australia. So too, adverse effects of treatments cannot be assumed to be absent because no one has looked for them.
When faced with an osteoarthritic elderly patient do you prescribe an NSAID, a COX-2 inhibitor, a codeine-based or related drug, plain paracetamol or nothing? How do you justify your decision? When faced with a patient at high risk of arterial disease the benefit of low dose aspirin usually outweighs the risk. For a patient at low risk of arterial disease this may not be so. The younger patient is at lower risk of adverse effects but has a very small chance of any benefit.
Assessing new drugs
We often hear of new "wonder drugs" that patients believe they must have to save their lives. No one can be sure that a particular therapy will alter their outcome unless it changes the result from 0% to 100%. Imagine that a new drug for cancer has been subjected to a vigorous trial in which 1,000 patients received the current therapy and 1,000 the new treatment. During the course of the trial 60 patients in the group receiving standard therapy died as did 40 in the trial group. Which of the following statements is true?
- The death rate fell from 6% to 4%
- The death rate fell by 2 percentage points
- The new treatment can prevent a third of deaths
- It is necessary to treat 50 patients with the new treatment to save 1 life
- The benefits may not be sustained with longer follow up
All these statements are true but describing a one third drop in death rate sounds impressive. Spin is not limited to cricket and politics. As Benjamin Disraeli once put it, "There are lies, damned lies and statistics."
Decisions are based not just on the level of risk, but on the potential outcome. Would you be prepared to walk along a beam 10cm wide and 1 metre above the ground? Would you walk along that same beam if it were 3 metres above the ground or 30 metres above the ground? Would you walk along that beam 30 metres above the ground if it were the only way to escape a fire? The risk of falling is the same each time. It is the outcome of falling and the benefit of crossing the beam that varies. Thus a 10% risk of a rash whilst taking a life saving drug is acceptable but a 10% risk of a serious adverse effect when treating a minor illness is not.
Getting the message across
The fact that 20% of parents nationwide and 30% of parents in London did not let their children have the MMR vaccine is testament to an abysmal failure to communicate. The BBC and ITV still insist on referring to it as the controversial MMR vaccine and pretend that expert opinion is evenly divided. Experts often mar their testament by expressing a degree of uncertainty that is appropriate for the scientific community but which undermines the message to a lay audience. To say, "All the evidence points to the vaccine being safe and there is absolutely nothing to link it with autism or bowel disease," is correct and decisive. Autism often becomes apparent about the age of vaccination but association and causation are not the same. Retrospective analysis has often showed that there was some evidence of regression before the injection.
It is worth considering the following facts:
- The original paper linking the MMR vaccine with autism and inflammatory bowel disease has been retracted by all its authors except one and the editor of The Lancet says that he would never have published it if he had known that it was such bad science.
- This flawed paper was based on 12 children. A study from Denmark, based on over 537,000 children found no association between the vaccine and autism.
- Many others, both in the UK and throughout the world have looked to see if there is an association between the MMR vaccine and autism and none has been found.
- The incidence of autism started to rise around 1980 but the MMR vaccine was not introduced until 1987. Within 18 months the uptake was over 90% but this had no effect on the rate of rise in autism.
- No country is the world has been sufficiently concerned as to withdraw the vaccine because of safety.
- Japan had problems with production of MMR vaccine and so had to revert to separate injections for a number of years since when they have re-introduced it. None of this has affected the incidence of autism in Japan.
- A large body of independent people including the editors of journals like the BMJ and The Lancet, the Royal Colleges and many academics throughout the world consider the vaccine safe.
In conclusion, the risks of being unprotected against the diseases far outweigh any risks associated with the vaccine. Still parents may ask for separate vaccines. A paper from Bandolier is recommended.
Clinical governance and risk management
An essential part of clinical governance is the management of risk. Good results have been achieved with protocol driven care, often by nurses. Doctors are not perhaps as ready to follow protocols which are seen as painting by numbers. However much of what doctors do, for example when conducting history and examination follows a protocol. Protocols represent best practice rather than variable practice. If things go wrong and if a doctor is called to account for his actions it may be easier to defend actions consistent with protocol. Protocols do not prevent clinical freedom but deviation from them must be justified.
Critical event monitoring was introduced to the RAF in the Second World War by a psychologist called Flanagan. It has been taken up by the National Patient Safety Agency.
- An adverse event is when things go wrong.
- A critical event is when things may have gone wrong.
It is important that such events are discussed openly. They offer an opportunity to learn. Critical events outnumber adverse events many-fold. It is important to ask if education needs to be offered or care pathways changed to prevent repetition and possibly disaster. Particularly with the changing face of clinical responsibility it is essential to assure that knowledge and skills match demands. Skills for Health are able to offer support and advice.
No matter how carefully you practice, adverse events will occur. The risk can be minimised but not eliminated. This is why everyone needs personal indemnity insurance. It is sobering to think that despite advances in risk management, the annual subscription to the MPS or MDU in 1974 was £25. We have had a great deal of inflation since then but not enough to account for a 100 fold increase.
Uncertainty in life
Around 1850 it was thought that science had advanced so far that before long everything that was to be known would be known. Yet in the 4th century BC Aristotle had written a philosophy that may be summarised as "The more you know, the more you know you don't know". Uncertainty is wisdom, not foolishness. Mark Twain wrote, "Education is the path from cocky ignorance to miserable uncertainty."
Uncertainty affects medical practice as it does the rest of life. For example, the rehabilitation of offenders balances the risk an individual poses to society against the benefits of making that individual an integrated member of society. A rail crash in which 10 people are killed raises a public outcry whilst, on average, nearly that number is killed on the roads every day. Forensic psychiatrists will always be criticised if they declare a patient to be safe when he is subsequently released and does harm. Who would criticise the forensic psychiatrist who incarcerates all his patients for ever? It is again important to balance risk and benefit, good outcomes and bad outcomes rather than looking selectively or in an unbalanced way.
Francis Bacon wrote, "If a man will begin with certainties, he will end in doubts; but if he will be content to begin with doubts he will end in certainties." All our world, not least the NHS, is beset by constant change and that means uncertainty. Uncertainty is wisdom. Dogma is weakness. We must learn to understand and embrace uncertainty rather than to merely cope with it.
The following have been confirmed from the Oxford Book of Quotations:
- Nothing is certain in life except death and taxes. Letter from Benjamin Franklin to Jean Baptiste, 13th November 1789, recorded in The works of Benjamin Franklin.
- Mark Twain attributed the quotation to Benjamin Disraeli in his autobiography of 1924. There are three kinds of lies: lies, damned lies and statistics.
- If a man begin with certainties, he will end in doubt; but if he will be content to begin with doubts, he will end in certainties. Francis Bacon. The advancement of learning 1605.
The following are from the Internet and hence of more dubious authenticity. The quotation is followed by the years in which the individual lived.
- Robert Burns: There is no such uncertainty as a sure thing. 1759 to 1796.
- Mark Twain: Education is the path from cocky ignorance to miserable uncertainty. 1835 to 1910.
Further reading and references
Skolbekken J-A; Communicating the risk reduction achieved by cholesterol reducing drugs. BMJ 1998316:1956-1958
Linde K, Clausius N, Ramirez G, et al; Are the clinical effects of homeopathy placebo effects? A meta-analysis of placebo-controlled trials. Lancet. 1997 Sep 20350(9081):834-43.
Sterne JAC, Egger M, Smith GD; Systematic reviews in health care: Investigating and dealing with publication and other biases in meta-analysis. BMJ, Jul 2001 323: 101 - 105.
Schwartz LM, Woloshin S, Welch HG; Risk communication in clinical practice: putting cancer in context. J Natl Cancer Inst Monogr. 1999(25):124-33.
Edwards A, Elwyn G, Gwyn R; General practice registrar responses to the use of different risk communication tools in simulated consultations: a focus group study. BMJ. 1999 Sep 18319(7212):749-52.
O'Connor AM, Legare F, Stacey D; Risk communication in practice: the contribution of decision aids. BMJ 2003327:736-740 [full text]
Trevena LJ, Davey HM, Barratt A, et al; A systematic review on communicating with patients about evidence. J Eval Clin Pract. 2006 Feb12(1):13-23.
Thompson IM, Bermejo C, Hernandez J, et al; Screening for prostate cancer: opportunities and challenges. Surg Oncol Clin N Am. 2005 Oct14(4):747-60.
Wakefield AJ, Murch SH, Anthony A, et al; Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children. Lancet. 1998 Feb 28351(9103):637-41.
Thompson IM, Bermejo C, Hernandez J, et al; A population-based study of measles, mumps, and rubella vaccination and autism. N Engl J Med. 2002 Nov 7347(19):1477-82.
Honda H, Shimizu Y, Rutter M; No effect of MMR withdrawal on the incidence of autism: a total population study. J Child Psychol Psychiatry. 2005 Jun46(6):572-9.