The Deceptive Allure of Evidence Based Medicine-W.W.
I have been in quarantine for a while, and during this time I have been following the coronavirus pandemic quite closely. There has been a deluge of information regarding recent and older studies related to possible treatments (both symptomatic and curative) for coronavirus. One particular issue related to the use of ibuprofen in symptomatic treatment of coronavirus cases provoked an argument between myself and another doctor. The core of the argument concerned some word of mouth claims (later proven to be false) that ibuprofen was associated with worse outcomes in coronavirus positive patients. However, previous studies on respiratory illnesses and ibuprofen have not come to any consensus, and no new investigations supported it either. Admittedly, performing useful studies of any kind during a pandemic is difficult.
Based on the above information, a group of general recommendations were made: There is no evidence these word of mouth claims are true. Prescribe paracetamol as a first choice for alleviation of symptoms associated with coronavirus. However, if you have to treat a coronavirus positive patient who is already using ibuprofen, do not remove ibuprofen from the treatment regimen unless there is another compelling reason to do so’. The rationale behind this recommendation is self-evident. Let’s not change what has always been done based on a couple word of mouth claims. If you do that, medicine would simply turn into doctors practising like headless chickens. The doctor I was arguing with, for some reason could not see my point, and began citing a medical app which pointed to a long list of side effects associated with ibuprofen at less than 1% of the population. She had started a smear campaign on ibuprofen based on word of mouth reports.
After this argument ended, I began thinking about EBM as a whole. What exactly went wrong in the argument I had with the doctor? Why was it difficult to get her to see the nuance in the argument? The nuance in the evidence? I’ve always had quite a few qualms in relation to EBM, and this particular argument brought quite a few of them to the fore.
EBM, or evidence based medicine can be defined as ‘the conscientious and judicious use of current best evidence in clinical care research in the management of individual patients”. It is the use of studies comparing one treatment with another treatment (or with placebo), in order to discover which treatment is best for a patient or particular group of patients. This may seem like an intuitive thing to do, but it only began in full force in the 1970s. Prior to this change of focus, medicine was practised based on ‘plausibility medicine’. If a medication or treatment was shown to perform x function, this medication was used in all diseases where it would be logical to think that this function would ameliorate or cure the patient’s condition. In the era of EBM there have been several cases where medications that were used based on plausibility were shown to have little or no effect. Some treatments have even been shown to be harmful.
EBM was meant to improve quality of care for all patients by providing a kind of consensus for treatment, based on objective criteria-eliminating the use of treatments which might harm patients, or provide them little benefit.
For any particular medical issue, a simple google search will reveal a plethora of studies, all of varying quality, related to the comparison of one treatment versus another for a particular disease. Simply type ‘hypertension guideline’ into google and you will find several recommendations for treatment, all backed by EBM. To perform these studies, several factors are taken into consideration-prior health, economic status, gender, the number of people enrolled in the study, length of the study, other medications being used by the patient. Any of these factors could complicate the analysis. It’s a lot to consider, and it is easy to imagine how any of these factors might create confusion when analysing results. Did John Doe live 5 years longer because of the new hypertensive you gave him? Or was he likely to live longer anyway because he is from the middle class and has access to better healthcare? Did he live longer because he was using other medications that are known to be effective? Confounding factors like these appear in any medical study.
These studies can and are repeated many times over the course of a few years. This invariably leads to differences in results. One month one could read that aspirin is great for secondary prevention of stroke, and a few months later one could find that after analysis of several groups of studies, no benefits were found. This is not even a fictional example. The 2019 ACC/AHA recommendations for aspirin use are downright confusing. One recommendation says that the use of aspirin in patients older than 70 is not recommended due to a higher risk of possibly fatal hemorrhagic events in this population. The next recommendation says the following: “for adults <40 years of age, there is insufficient evidence to judge the risk-benefit ratio of routine aspirin for the primary prevention of ASCVD”.This statement, while vague, can be interpreted to mean that since the patient is under 40, the risk of bleeding is no longer an important consideration, but the studies have not found a benefit or harm in the use of aspirin for the prevention of cardiovascular disease. But here comes the kicker:
“There is also insufficient evidence to comment on whether there may be select circumstances in which physicians might discuss prophylactic aspirin with adults <40 years of age or >70 years of age in the context of other known ASCVD risk factors”. This would mean that despite not finding a benefit or harm in the use of aspirin, it is still possible for a doctor to discuss the use of aspirin in cases of family history of heart attacks, inability to lower blood lipid levels, poor blood pressure, or blood glucose control.
Upon reading the recommendations several times, the meaning is understood. However, the recommendations are often very vague, and are interspersed with information that is common medical knowledge (bleeding risk is high in older patients) and the end result is probably one concrete recommendation, and the rest of it basically says ‘use your medical judgement’. How on earth is that helpful? The results are often unclear and very difficult to routinely apply in clinical practice.
This is only one example. Doctors often end up being more confused by new studies than before the studies appeared.
Another important problem with EBM relates to studies which show ‘negative’ results. These are studies performed on a medication or an intervention which show that it has no effect on the condition being evaluated.
In the above video, an epidemiologist explains how few of these studies are published. It is not fashionable or ‘sexy’ to report in a medical journal that a large double blind study was performed on a drug, and absolutely nothing happened. These negative result studies should be more important when making clinical decisions.
Studies are often also published in a frenetic manner. Fashionable or astonishing results are found from one study or group of studies, and these are almost immediately splashed across all the pertinent medical journals. In many cases, it is not long before contradictory results are published. This is particularly common in the cases of studies related to diet. It is understood that contradictory results are a necessary part of reaching the truth. However, this kind of ‘dissociative personality disorder’ that the research community appears to have can make it very difficult to make day to day clinical decisions. Helping doctors make these kinds of decisions was the very purpose of EBM.
Evidence based medicine, coupled with the reach of modern technology, was meant to be a great aid to health practitioners. However, the appearance of oftentimes vague recommendations, the omission of negative results, and the presence of contradictory and at times sensationalist reports have made me very skeptical about its utility.
