Why Doctors should not follow Guidelines

There are large variations between different doctors in terms of prescribing, use of tests, and referrals to specialists. Some of this difference is clearly explained by doctor’s ignorance, and their failure to apply the lessons of medical research. A proportion of patients who suffer for example internal bleeding from stomach ulcers do so partly because doctors prescribe anti inflammatory drugs to high risk patients, such as those taking antidepressants, or patients with poor kidney function etc.

In order to prevent these harms, various institutions including NICE (the National Institute for Clinical Excellence), the Royal Colleges of various specialties, and various special interest societies such as the British Heart Foundation have systematically examined the research on many topics, and produced guidelines on the management of many conditions, which are widely disseminated and taught to doctors. Statistics are collected comparing the performance of doctors in complying with this guidance, and doctors who perform badly may be penalised in various ways. Reports from pathology labs, for example when they grow bugs in swabs, often only report whether the bug is sensitive to 'approved' antibiotics even though they are also tested against other drugs. The results supplied to the requesting doctors are restricted in order to nudge the doctors to use the treatments advocated by the microbiologist.

Surely this is a good idea, and likely to benefit patients?

I am not so sure for several reasons. 

One major problem is that the evidence on which they are based is insufficient or even biased, and is usually applied to groups of patients (such as the elderly), who were not included in the original research. Almost all research trials exclude patients over a certain age, or on other medication. Trials are conducted by the drug industry. They usually control all the data, and numerous cases of manipulation and fraud have been revealed. However, this possibility tends to be ignored by the relevant committees. For example, Guidelines advising the use of clot busting in stroke were issued in 2007 and hospital services have been reorganised to facilitate the use of this treatment. However, in a poll of emergency doctors, most said they did not believe the treatment to be beneficial, and in 2014 the MHRA launched a review of the treatment after reanalysis of the trial data showed glaring inconsistencies and a lack of benefit to patients in the majority of the hospitals taking part.

Similarly, after heart attacks patients are now treated with a new expensive clot preventer. The trial on which this advice is based showed a remarkable lack of significant benefit in some countries in the trial, with very large benefits in others. This could be a statistical blip, but it does seem suspicious, and surely the issue needs further research before it becomes official guidance.

Research is also biased in the sense that very little research is done into drug harms, so that guidelines are biased towards recommending medication. NICE guidance on using Statins in all over 80's is an example. Some of the results of Statin trials are probably misrepresented, there has been very little research in this age group, and there is a lot of uncertainty about harms, with huge numbers of patient reports of muscle pain, falls, memory problems, diabetes, cataracts, and nerve damage. Experts (some of whom are likely to have a conflict of interest) are sceptical, and say that the harms are exaggerated, and these problems are ignored by NICE.

Similarly, NICE advises medication for patients with brittle bones, although there are fears of these drugs causing cancer of the oesophagus. The Quality and Outcomes Framework target that makes up a large amount of GP pay promotes long term use, although the pharmacology of the drugs suggests that it is pointless using the treatment for more than 5 years. GPs are also paid for hitting targets for controlling blood sugar in patients with diabetes, although the evidence that this prevents deaths and complications is lacking. GPs are thereby encouraged to use new drugs that have no proven long term benefit, and are likely to have harms. This problem has lead to some embarrassing u turns by NICE, for example after the so called Cox2 selective anti-inflammatory drugs came out they advised their exclusive use in over 65's. However soon afterwards it became clear that these drugs caused heart attacks, and that they were particularly dangerous in the very patient group that NICE had recommended them for.

This is at least possible to remedy, but other difficulties seem insoluble. One big problem is that inferring actions from evidence needs value judgements about the pros and cons of the harms and benefits of treatment. NICE have attempted to deal with the cost effectiveness issue by adopting a cut off of £30k per year of quality life gained (This is of course extremely tricky to calculate in practice, and often the assumptions used are heroic), but there are lots of other harms and costs that defy mathematical solutions. The risk of stroke for patients with an irregular heart rhythm, so called AF, is roughly 1 in 20 per year and can be halved by blood thinning treatment. But the patient will have put up with various harms, including large increases in risks of surgery and accidents, not being able to have many other medications that upset the blood  thinning treatment, the need for frequent blood tests, which can make travel very difficult, a large increase in the risk of dangerous internal bleeding etc. The GP QOF system based on NICE Guidelines presently penalises doctors when patients are not being prescribed this treatment, and it is common to meet patients who have been prescribed blood thinners for this reason in hospitals without a detailed discussion of pros and cons. But balancing the reduced risk of stroke against for example the increased difficulty of overseas travel is surely a value judgement that only the individual patient can make. Even taking a daily pill is a cost that will vary between patients., some see it as depressing evidence of poor health, while others are not bothered.

Guidelines from different countries varies hugely because of this value issue, despite being based on the same evidence. For example, high blood pressure is the commonest reason for the use of long term medication and there have been countless research trials into its treatment, which often requires several different medications used in sequence. Despite this, comparisons of guidelines from the US, Canada, Denmark, and the UK shows that there is no agreement even on the most basic issue of which drug to start with. Most countries advise starting treatment with a diuretic (which makes the kidneys excrete more salt and water), but this treatment is only step 3 in the UK Guidance.

The AF guidance focuses on preventing strokes without paying a lot of attention to harms. Other guidelines focusing on a particular problem can have unintended harmful consequences in other areas. GPs are monitored on how much they prescribe certain antibiotics that increase the risk of the superbug Clostridium difficile, usually by looking at the percentage of prescriptions for 2 particular classes. One of the main types that is encouraged are so called macrolides, like clarithromycin. However, it is now clear that these drugs have a dangerous effect on the heart in some patients and significantly increase the risk of sudden death.

But the biggest problem with guidelines is that they completely fail to take account of the fact that most patients have more than one long term condition, and many have several. Does it matter? Yes of course it does. How well do the various drugs go together? We often do not know. If a patient has very poor kidney function, does that affect his diabetes treatment? Well obviously. If a patient has dementia, are we keen to start treatment aimed at preventing sudden death? Well, maybe not. There has been an attempt to assess how many guidelines may be relevant to the treatment of typical illnesses in typical patients. For a particular elderly patient with a broken hip it was calculated that 75 guidelines were relevant. Simply to read them all would have taken several weeks.

So how can this situation be ameliorated?

First of all, the advice that is usually buried somewhere in the small print, that the guidelines need to be interpreted in the context of the individual patient, and that it may be entirely correct to ignore them occasionally, needs to be printed in bold on every page. Along with this, most targets that incentivise treatments should be removed. Removing doctors conflicts of interest in this way would also improve patients trust in doctors.

A recent research study BMC Cancer 2015; 15;734  looked at patients treated for breast cancer in Germany comparing results for patients whose treatment adhered to guidelines, with those whose treatment diverged from the guidance.  The latter group showed a significant improvement in survival! The authors conclude in a somewhat Teutonic formulation “This is an indicator for the appropriate application of guidelines. A maximization of guideline-based decisions instead of the ubiquitous demand of guideline adherence maximization is advocated.” i.e. patients may benefit when doctors do not adhere to guidelines.

Secondly, NICE etc. need to be more rigorous with examining all the data and ignoring any trials with a whiff of fraud. They should cease to extrapolate to patients for whom there is no direct  evidence. They also need to commission research into the harms of treatments because no one else is doing it. Conflicts of interest among members of NICE and other guideline producers need to be rigorously policed.

Thirdly there needs to be a big effort to make guidance appropriate to real patients, with multiple conditions.

Last of all but most importantly, it needs to be emphasised that good medical care is a matter of well educated doctors evaluating evidence together with patients and making appropriate decisions together, or in the words of Glasgow GP Des Spence “A doctor’s role is not ….. to follow guidelines but to interpret, and often ignore, them—that is, to short circuit the mechanised medical machine for the benefit of patients. Otherwise what is the point of doctors: any idiot can follow a guideline"

Just how bad is Cancer care in the UK?

The UK has poor cancer care compared with our neighbours. This fact is used as a stick to beat GPs especially, and has led to a huge investment in hospital cancer care in the last five years. But can we trust the stats, and has the investment been a good idea?

First of all we need to examine whether we really are making fair comparisons, and what factors other than quality of care affect the results. We generally compare survival figures such as the percentage of patients alive 5 years after diagnosis of a particular disease. There is an obvious problem with this approach if the patients in one country differ at the outset from those in another. This happens in various ways.

First of all, cancer of any particular organ is not one disease but a collection of different mutated cell types that have changed in a way that allows them to evade the immune system and invade other tissues. Some cancers are much more deadly than others and some countries may suffer deadlier subtypes more than others. For example, one type of lung cancer, so called small cell, is never curable by surgery, and has a 5 year survival near zero. A high proportion of this type among lung cancers will adversely affect overall lung cancer survival. Cancer types do vary between European countries, so this a significant problem.

Another problem is so called incidentalomas.  These are true cancers under the microscope, but do not behave as cancers, and rarely pose a threat to life. These are not rare, with kidney cancer, thyroid cancer, skin cancer, and so called Ductal Carcinoma of the breast being well known examples. In countries where more formal or ad hoc screening scans are done, a proportion of 5 year survivors are not surviving because of good treatment, but because they would have been alive anyway because their ‘cancers’ were never a threat to life.

Then there is so called 'lead time bias'. If patients are diagnosed earlier, they are more likely to survive for 5 years, even if there is no effective treatment, simply because the diagnosis is made earlier in the natural course of the disease. This problem is well known but it is difficult for experts to allow for it.

A useful example is Prostate Cancer compared between the US and the UK. The death rate of this condition has remained similar and very stable in both countries for many years. But the survival figures have improved dramatically in the U.S. The general interpretation is that U.S. Patients are getting a diagnosis, which is of more harm than benefit, and that treatment is largely useless at least in terms of postponing death.

Then there is the issue of cancer registries and death certification, especially of elderly patients who are likely to have advanced cancer, but who are frail and generally sick. These patients and their families may choose not to have scans or operations and they may therefore never be included in the statistics. This tendency is likely to vary between countries. A large proportion of patients are certified as dying of 'old age', some of who surely in reality died of cancer. Some countries have much higher rates of post mortem examinations than others. In those countries the true cancerous diagnosis will be recorded, thus adversely affecting survival statistics.

Then there is patient behaviour, which varies significantly. For example, lung cancer survival is strongly affected by whether smokers stop the habit. Other variables, such as Vitamin D blood levels are also known to have a major effect (so if you get cancer, move somewhere sunny). Behaviour in terms of the way patients seek medical advice is also important. Anecdotally, there seem to be 2 main types of patients in the UK, the 'worriers'-those who listen attentively to the health info publicity, and seek advice for every skin blemish or headache because it might be cancer,- and a significant number of 'deniers' who ignore symptoms like weeing blood. If a country has a higher level of health literacy, and more sensible behaviour, then it is likely to result in earlier diagnosis and better survival.

Another relevant variable factor is the extent to which patients accept definite harms in the hope of an improved chance of long term survival. For example, patients with early localised breast cancer may be offered drugs and radiation with the aim of mopping up cells that may have spread, even though that may be relatively unlikely, so that only a small minority of patients are likely to benefit. Patients and their doctors in some countries will conclude that it is worth taking the small risk of not having the treatments,  in order to avoid the certainty of months of feeling ill, losing hair etc. This will worsen the statistics for that country but does that mean that the standard of care is worse?

Taking the survival figures at face value has led to various actions in the UK. Cancer patients are now treated by so called MDT's, where the surgeons, oncologists, X-ray specialists etc. discuss patients and decide on treatment. This has been done in order to improve survival. But this ignores one of the most important principles of medical ethics, which is patient autonomy, the freedom to make ones own choices with ones body. It also commonly leads to severe communication failures, for example patients attending for follow up appointments at which the doctor that sees them assumes incorrectly that the patient has already been given some bad news, for example that a cancer has spread. This seems unbelievable but I personally have encountered this several times in the last 3 years just among my own patients.  

Also, patients seem to be being pushed into treatment with drugs which reduce their quality of life, when the prospects of extending their lives seem very poor. it seems that the imperative to improve the stats is once again going against standard medical ethical principles.

One suggestion made by our Health Minister on this issue, has been to 'name and shame' individual GPs who fail to refer patents within the first couple of consultations, or if their patients are diagnosed after being admitted to hospital as an emergency. This is a naive idea. First of all measuring this statistic per individual GP would be tricky, as defining which consultations counted would not be obvious. Also each doctor will only refer a handful of patients every year who are found to have cancer, and there would be a lot of random variation. But the main problem is that once again cancer is a mixed bag and certain types (pancreatic cancer for example), are very difficult to diagnose, while others are easy.  Once again there is a big risk of comparing apples and pears.

There is however one very obvious factor that is associated with better survival rates, which is short waiting times for 'routine' scans and specialist appointments. Results for the UK and Denmark, which both rely on GPs as gatekeepers to assign priority to patients, are poor.

The reason is that medicine is a difficult art, and a large proportion of patients in all countries are diagnosed with cancer after routine investigation of vague complaints rather than so called red flag symptoms. Reducing waiting times for these patients, and loosening the tick box restrictions on urgent referrals would certainly improve performance, but at a huge cost and of course not within the tenure of any particular minister.  Blaming GPs is easier.

  

So should we ignore the figures? No of course we should not, but like all statistics they should be examined carefully and suspiciously and kept away from politicians and never used as soundbites.

 

---