What is evidence-based medicine?
I heard the phrase evidence-based medicine by chance when a doctor friend of mine shared his impressions about the students to whom he lectures on the relevant issues. If my interlocutor was surprised by such ignorance, then he kept these assessments to himself, approached my computer and ... Then it was my turn to be amazed…
I asked what several of my friends and acquaintances have heard about evidence-based medicine, who are either professionals in one of the fields of medicine, or read a lot and extensively, moreover in different languages. They didn't hear anything. And then it seemed to me useful to fill this gap as much as possible within the framework of the format adopted at this site.
I am asking Doctor of Medical Sciences Evgeny Yakovlevich Parnes.
But "convincing medicine" simply does not sound in Russian, hence, I think, the term "evidence-based medicine". However, if I correctly understood the central idea of the method, any studies of living organisms can be proven precisely and only in the same sense – as fundamentally and unconditionally evidence–based.
So, let's not discuss the term itself. But what is curious is to find out when the Russian term evidence–based medicine became common among our doctors? After all, clinical trials under the "double-blind control" scheme (that's what it was called in the late 60s, when I began to get acquainted with psychiatry) are not new in themselves.
E.P. Of course, the methods of double (and even triple) blind control are not new in themselves. Any "natural scientist" will tell you that. The scale of modern randomized clinical trials can be considered radically new: not only are they amazing in terms of patient coverage, but the attitude to the results of such studies that we expect from a truly modern practitioner, and not just from a researcher, is very important.
Here is a vivid example for you. Doctors are constantly faced with the fact that many medications, surgical and other methods of treatment, which theoretically should be effective, in practice do not have sufficient effect, and sometimes, on the contrary, contribute to death. So, it has long been noticed that patients with heart disease and rhythm disturbance (for example, extrasystole) have an increased risk of sudden death from rhythm disturbance.
Naturally, many drugs were soon developed that were able to prevent the appearance of extrasystoles very effectively. It would seem great: here is the initial daily ECG monitoring – extrasystole is full, repeated ECG after a month of treatment – there are almost none. Is it good or bad?
Until evidence-based medicine appeared, they thought it was good, and when they conducted a study in accordance with the principles of evidence-based medicine, it turned out that it was bad, since most of the most effective drugs shortened the life of patients, increasing the likelihood of sudden death.
R.F. In accordance with the principles of evidence–based medicine is?..
H.P. Have you heard anything about the Framingham study?
R.F. No.
E.P. In medical journals, you can often find a link in this form: "as the Framingham study showed..." It really showed a lot of things – and it will show more.
Framingham Heart Study (Framingham Study) began in 1948 in the typical American city of Framingham, Massachusetts, with the aim of studying cardiovascular health. This research was conducted under the auspices of the National Heart Institute (later it was renamed the National Heart, Lung and Blood Institute – National Heart, Lung and Blood Institute, abbreviated as NHLBI). Initially, the study covered 5,209 men and women. In 1971, it included 5124 representatives of the second generation of participants – “offspring". Now the researchers are planning to start a survey of 3,500 grandchildren of those who entered the study more than 50 years ago – the “third generation". The study has no equal in duration and cohort size; its importance for modern medicine – and, first of all, cardiology – is difficult to overestimate.
Over the years of careful observation of the study participants, the main risk factors leading to diseases of the cardiovascular system were identified: high blood pressure, elevated cholesterol levels in the blood, smoking, obesity, diabetes, etc. (A similar approach belongs to the section of medicine called clinical epidemiology). Since the beginning of the study, about 1,200 articles have been published based on its results in the world's major medical journals.
We have all heard conversations in the spirit of "my grandfather smoked until he was 90 and nothing, but his neighbor never smoked and died of lung cancer at the age of 50." In fact, it happens! But…
R.F. ... a friend always suggested to me to recall the memoirs of Academician Krylov, who drank a glass of vodka in the morning all his life – and lived to old age...
E.P. So, only when all cases of lung cancer in smokers and non-smokers were calculated, it turned out that the risk of getting lung cancer is significantly higher in smokers. The Framingham study answered many questions related to factors influencing the occurrence of diseases. But the most remarkable thing is that when the revealed patterns began to be used in the prevention of diseases, it led to a sharp decrease in morbidity and mortality. First of all, this concerns cardiovascular diseases. Mortality from coronary heart disease in the USA has decreased by 2.5 times since the 50s of XX to date, and from stroke - by 4 times. It is interesting to note that the mortality curve went down only 20 years after the start of the study.
And another plot related to the Framingham study.
Until recently, it was believed that high blood pressure is an age–related phenomenon, but now, based on the Framingham study, arterial hypertension is called an increase in blood pressure above 140 and 90 mm Hg at any age, since it has been proven that an increase in blood pressure above 115 and 70 mm Hg is already beginning to be accompanied by an increased risk of cardiovascular diseases. Consequently, all patients with blood pressure above 130-140 and 80-90 should be treated, especially in the presence of risk factors.
Unfortunately, in Russia during this time, the mortality rate from coronary heart disease and strokes has increased almost 2 times – here we have reached the first place in the world. First of all, this is due to the neglect of the prevention of these diseases. And the scientific basis for prevention is precisely clinical epidemiology – one of the aspects of evidence-based medicine. The Framingham study perfectly illustrates the effectiveness of the corresponding methods.
R.F. For various reasons, both personal and scientific, I constantly read the sections etiology and pathogenesis present in the descriptions of diseases: etiology is about the causes, pathogenesis is about the mechanisms that determine the development of the disease. It seems that the treatment since about the end of the XIX century is based on the knowledge of both.
E.P. Evidence–based medicine does not reject these principles of treatment - they are called etiopathogenetic. For example, having revealed that the cause of most acquired heart defects is acute rheumatic fever, which is caused by streptococcus, doctors have been actively using penicillin and its analogues for the treatment of pharyngitis and tonsillitis (angina) since the 50s, which has led to a sharp reduction in patients with heart defects at present.
But evidence-based medicine postulates that a drug, a method of treatment, a diagnostic method can be used only when it has proved its feasibility. If it concerns a drug, then it should not only reduce the symptoms of the disease, but also improve the quality of life, prevent the development of complications of the disease and at the same time should not have severe side effects or side effects that occur frequently and force you to abandon the drug.
These requirements especially apply to new medicines that are compared with pre-existing ones. That's why reliable procedures and comparison methods are the main concern of evidence–based medicine.
R.F. If I understood you correctly, then the main obstacle and source of systematic errors both in diagnosis and in evaluating the effectiveness of treatment is a bias characteristic of a person…
E.P. You know that if a scientist wants to prove that his method, including the diagnostic method, is good, then he will definitely find confirmation of this in his experiments. A doctor who is convinced that one medicine is better than another will always look for confirmation that it is so.
Very often, the patient, receiving medicine from a doctor, begins to feel better. But why? Because the disease has a circadian rhythm, and spontaneous remission has occurred? Or does the patient wholeheartedly believe the doctor, and faith works miracles? Or did the "pill really help"? As a rule, the doctor attributes the effectiveness of treatment to the chosen medicine, but this is not always the case.
Bias leads to a systematic error, that is, to a non-random distortion of facts. Agree – it's in human nature, even a doctor sees what he wants to see. By and large, he is not to blame, unless he is engaged in banal forgery.
R.F. How is this bias blocked when studying the effectiveness of drugs?
E.P. One of the main principles of evidence–based medicine is the so-called blinding of research. In short, this is a scheme in which no one knows what the patient receives: neither he himself, nor the doctor who communicates with the patient, nor the statistician who calculates the result.
Of course, in such a study – it is conducted in a clinic – the ethical principle of informed consent of the patient is strictly observed. The patient knows which drug is being tested, with which it is being compared, and why the new drug may be better. But no one knows exactly what kind of drug this particular patient will receive.
If two drugs are compared, the patient will be given two packages. In one there will be a drug, in the other – a placebo, that is, a "dummy" – in shape, color and taste corresponding to the drug, but not containing an active principle. And every day the patient will take one of the two studied drugs from one package, and a placebo from the other. That is, the patient will not remain without treatment.
At the same time, one part of the patients will take a well-proven drug currently used, and the other part will receive a new, experimental drug, which presumably should be more successful, otherwise pharmacological firms would not invest crazy money in it.
A slightly different scheme corresponds to the case when a completely new drug has been created, which has no analogues, and it must be proved that it has a certain value. Then you will have to compare the new drug with a placebo. In this case, the patient is given one package, while some patients will receive the drug, and the other part will receive a placebo. And this is ethical, because in such situations there is simply no alternative.
R.F. And what is the function of a doctor when conducting a blind study?
E.P. The doctor must select the "right" patients who meet the conditions for inclusion in the study and who do not have exclusion criteria. (These conditions were developed by the authors of the "research protocol", and they are pre-determined). The doctor monitors the effectiveness of treatment by increasing or decreasing the dose, achieving optimal results (although he does not know which drug he uses at the same time), monitors side effects, carefully registering any deviations in the treatment process, including studying the patient's analyzes, which, as a rule, are sent to one central laboratory for all researchers, most often – in Belgium or the Netherlands.
If adverse events occur, the doctor may decide to discontinue treatment, although he does not know which drug is associated with this side effect. All undesirable phenomena that occur during the study are recorded, and they are reported to special centers for their study. And if side effects occur frequently or they are severe, this serves as a basis for early termination of the study. Which has happened repeatedly.
R.F. But the selection of the "right" patients is a non–trivial task in itself, isn't it?
E.P. Of course, the principle of randomization, which ensures this selection, is one of the main pillars of evidence–based medicine. It is necessary that the group of patients receiving the studied drug and the group of patients receiving the already known drug should not differ in any way by gender, age, race, social status, duration of illness, complications that have developed, or previous therapy. Only then can it be said that the differences in the groups that will be observed further during the study were due to the characteristics of the drug, and not the initial characteristics of the patients.
In modern research, computer programs use the method of a random number generator, which gives the doctor only the patient's registration number. At the same time, many groups can be formed. The doctor does not know which group the patient is in. But with the help of a computer, medicine will be delivered to this number in the future.
R.F. In order for all this to be valid, you need to recruit a lot of participants. E.P. Of course, without this, we will not achieve any genuine randomization.
There really should be a lot of patients: for phase II of the study – 200-400 people, for phase III – at least 1000, and often 20-40 thousand
R.F. Wait, what is phase I?
E.P. During phase I on healthy volunteers (people do it for money), the tolerability of the drug and possible doses have already been determined.
During phase II, the effectiveness of the drug is studied, the effective dose is worked out, and side effects are studied. This is the first trial of the drug on patients. As a rule, it is during phase II that it is determined whether or not the drug should be. A feature of phase II is the selection of a "refined" group of patients – of a certain age, without concomitant diseases, etc.
During phase III, additional data on the safety and efficacy of a new drug are studied under conditions as close as possible to normal medical practice, that is, in patients with concomitant diseases undergoing concomitant therapy (for other diseases), with various complications.
R.F. How can you recruit 20 thousand study participants?
E.P. Of course, you can't do this with the help of two or three clinics. The study should be multicenter. Usually, research centers are organized on all continents, in most countries. In Russia, the centers are most often in Moscow, St. Petersburg, Novosibirsk, Samara. At the same time, there are 10-20 centers in Moscow alone. This is done so that an average of 10-20 patients are recruited in one center.
A multicenter study makes it possible to neutralize the poor performance of any one center. If the information from any center contradicts the information coming from other centers, then this center will be very carefully checked. And if it is found that the errors are related to the work of doctors, the center will be closed.
Multicenter studies, and usually this is phase III, also allow us to establish the characteristics of the drug's action (efficacy and side effects) in patients of different races, with different heredity, with different social status, different diet and lifestyle.
Another important principle of evidence–based medicine is the analysis of pre–selected "points" that characterize the course of the disease - first of all, the so-called endpoints. Endpoints in randomized controlled trials are called phenomena that should be influenced by treatment. The main or "solid" endpoints are obvious outcomes of the disease (for example, death, acute non–fatal myocardial infarction, cases of hospitalization due to a sharp aggravation of the course of the disease).
There are also soft "surrogate" endpoints – for example, a significant decrease in extrasystoles, the achievement of acceptable blood glucose figures in the treatment of diabetes mellitus, an increase in the ejection fraction of the heart in the treatment of heart failure, a decrease in blood pressure, etc.
It is important that the purpose of the study is not to influence the "surrogate points", but to prevent the onset of solid! And the surrogate points should rather explain why the treatment somehow affects the hard points.
Solid endpoints are methodically important also because they are difficult to falsify.
R.F. What is a more or less permanent subject of clinical research of the type described above?
E.P. A new drug. A new dosage form: for example, there used to be pills, and now intravenous injections or candles, ointment. New prescription mode: previously, one tablet 1 time a day for 6 days, and now 3 days in a row, but 2 tablets each. A new dosage (fixed or titrated) for example, atorvastatin was previously used at a dose of 10 mg, now 80 mg can be used, since in 4 studies it was shown that this dosage is more effective, and the number of side effects increases slightly.
Finally, a new duration of treatment: cystitis was previously treated for 7-10 days, now it has been proven that the effect will be the same if treated for 1 day with modern uroseptics.
A new indication for use: for example, aspirin used to be used to treat pain and to reduce fever, now it is the number 1 drug in the treatment of coronary heart disease in order to prevent myocardial infarction and stroke.
R.F. Can we name the areas of medicine where randomized clinical trials are practiced especially widely?
E.P., about several thousand randomized trials are conducted per year; there are especially many in cardiology, therapy, endocrinology, cancer chemotherapy and oncohematology. This is primarily due to the production of new drugs, the properties of which are easy enough to test in randomized controlled trials. It is clear that the new modification of surgical treatment is more difficult to verify, since such a study cannot be blind, and the result significantly depends on the skill of the surgeon.
R.F. I want to understand how the power and reliability of randomized clinical trials described by you and the results resulting from them are combined, which in practice are transformed (at least they should!) in recommendations that are brought to each practitioner in one form or another.
At least two points are visible here: (1) how can I (i.e., the doctor) find out if I am doing the right thing by choosing medicine A, and not B; (2) how can I be sure that two drugs with different names, produced in different countries and allegedly having the same composition, and in the are they identical?
E.P. Let's separate these problems. Question (2) is simpler. A branded ("native") drug is often compared with a so-called generic. Usually, such studies are ordered by generic companies to prove that the generic is just as good. And this is good, we need to compare, but not on 20 patients! On the other hand, one can only marvel at the practical recommendations arising from dissertations that recommend the use of certain methods of treatment, and there were 4-6 people in the compared groups.
Now about the validity of your choice in general. Randomized clinical trials, as well as evidence-based medicine in general, as a direction, of course, is not an end in itself. Based on the data obtained, practical recommendations are being developed for doctors – in particular, determining the treatment strategy and the choice in favor of a particular drug.
Such guidelines are designed for all doctors, while it is assumed that no matter what qualification the doctor would be, if he strictly adheres to the recommendations, in most cases he will achieve a good result. Manuals to a large extent began to displace textbooks, since they present the etiology, pathogenesis, clinic and treatment of the disease from the standpoint of evidence-based medicine, that is, what used to be the essence of the textbook. What is important is not the author's textbooks, in which contradictory provisions can often be read, but a reflection of the consensus of specialists in this field.
Moreover, the recommendations for this country are usually consistent with the recommendations for other countries. Therefore, these guides are very useful at least for general educational purposes. And more. All provisions on the expediency of diagnostic and treatment methods are arranged in the form of classes and levels of evidence.
Unfortunately, cardiologists and pulmonologists have somewhat different ideas about them, but in general, there are 3 classes that determine the ratio of benefits relative to risk.
So, the recommendation of class 1 means that the benefit relative to the risk is maximum, which means that the proposed (whether it is a research method or a medicine) must be fulfilled.Class 3 means that this method of research is meaningless, and the medicine is more likely to cause harm.
Class 2 occupies an intermediate position – rather useful, it is advisable to perform. The levels of evidence inform us about the quantity and quality of the studies from which the classes of recommendations follow.
At the same time, the level of evidence A means that similar results were obtained in several (more than 3), well-organized randomized controlled trials in accordance with GCP (good clinical practice). And it does not matter whether we are talking about the fact that a certain drug is useful or harmful. Level B means that the results are obtained in 1-2 RCTs. Level C is the results of non–randomized studies and/or conducted with a small number of patients (for example, within the framework of PhD or doctoral dissertations). Level D is an expert's opinion and may reflect his private views.
In cardiology recommendations, level C and D (sometimes B and C) are combined into one, and then in the first case it will be level C, and in the second case – level B. Level C and D means that this recommendation can be ignored in some situations. It is not always necessary to conduct a study if the result is well known. So, although no one will conduct a study on the antipyretic properties of aspirin, since it has long been known and proven in practice, but in the recommendations one could find such a designation of the validity of the use of aspirin for this purpose: class 1, level C. That is, the effect is indisputable, although no one has studied it specifically.
And now the most important thing: these recommendations are not a guide to action, that is, not a direct instruction, but a list of methods of examination or treatment with varying degrees of usefulness and validity that are offered to the doctor during the examination and treatment of the patient. It is already up to the doctor to choose one or another method in accordance with his experience, the peculiarities of the course of the disease in the patient, the presence of concomitant diseases.
It's like in a tool shop. You come to the seller and say: "I need to cut one thick wide board into two, what should I choose?" The seller can offer you a machine, but why do you need a machine if there is only one board? He can offer you a small circular saw, but you have no experience with it; besides, you don't want to be left without a hand. Then he can offer you a hacksaw: you will suffer for a long time, but there is only one board, and you will be left with your fingers! But if you like a beautiful penknife, then the seller will tell you that it does not fit for your purposes in any way.
So, if a doctor can choose from classes 1 and 2, then it is not allowed to use the recommendations of class 3.
Unfortunately, standards are a logical continuation of practical recommendations. Standards are developed by medical officials or officials together with individual doctors at the hospital, city or country level. And this is bad, since the standards take into account the interests of officials and sharply limit the capabilities of the doctor, clearly prescribing which drug to take and in which situations. Maybe this is good for very bad doctors, or rather, it is good for their patients, but it will definitely interfere with good doctors, since the patient rarely corresponds to "school" cases and everyone needs an individual approach. And who are the judges…
Modern randomized clinical trials (RCTs), the results of which can be found on the Internet, allow the doctor to solve complex issues in relation to the treatment of a particular patient. For example, a woman at 5 weeks of pregnancy has severe pneumonia, most likely caused by mycoplasma, a pathogen that is not sensitive to conventional antibiotics. Such pneumonia should be treated with modern antibiotics – macrolides (these include, for example, clarithromycin and sumamed). However, the instructions say that they are not recommended during pregnancy, because they negatively affect the development of the fetus.
What to do – "in a simple way", i.e. prescribe kefzol, which is allowed during pregnancy, although it does not affect mycoplasma, or?..
Two RCTs – for 2006 (from Canada) and for 2008 (from Israel) – indicate that the appointment of modern macrolides practically does not lead to teratogenic – i.e. negatively affecting fetal development – effects in the 1st trimester of pregnancy. Therefore, such pneumonia should be treated with macrolide.
R.F. Finally, tell us about some memorable incident.
E.P. I remember treating a patient in 1993 who made a great impression on me. He could not sleep lying down because of suffocation and could only sit, greedily grabbing air; the circumference of his shins was, as I remember now, 61 and 56 cm on the left and right, temperature below 40 degrees, blood pressure 200 and 110 mmHg, and even atrial fibrillation with a heart rate of 130 per minute. He had deep vein thrombophlebitis with recurrent thromboembolism of small branches of the pulmonary artery, despite the fact that he had a so–called cavafilter - a special trap for blood clots in the inferior vena cava. Pneumonia was heard, the liver was enlarged, there was fluid in the abdominal cavity – ascites. And at the same time there was a high concentration of uric acid in the blood. He had already been treated before me, but it didn't help much.
Consequently, it was necessary to break all the vicious circles that he had formed. I had to prescribe drugs that reduce blood clotting (they had not been given before), kapoten at a maximum dose of 150 mg per day – it was then the only available angiotensin converting enzyme inhibitor; digoxin (although by that time it had already been persecuted, and in the USA it was practically abandoned), but in small doses (already there was an article on this topic) it practically did not cause side effects; furosemide mixed with urate, veroshpiron in large doses, allopurinol to reduce uric acid levels and prevent kidney death against the background of massive diuretic therapy, and, of course, antibiotics to cope with pneumonia.
I have consulted many people about this patient. And many now high-ranking therapists have told me that I am crazy. Firstly, I am engaged in polypragmasia (we counted seven medications, and it is supposed to give no more than three at the same time). Secondly, digoxin cannot be prescribed at all – it is dangerous. And, finally, it is impossible to prescribe ("no one treats like that") veroshpiron to kapoten, since both delay potassium, and this will kill the patient. But I saw signs of secondary aldosteronism in the patient, i.e. a secondary (disease-caused) increase in the hormone aldosterone, one of the manifestations of which is potassium deficiency, and even gave a high dose of furosemide, so veroshpiron oh, how I needed it.
Therefore, despite the criticism of colleagues, I did not change the treatment.
Gradually, the patient got rid of shortness of breath, edema, his legs became slim, his blood pressure normalized, and despite the persistent atrial fibrillation, the heart rate was no higher than 80 per minute. Six months later, he boasted to me that he could run up the stairs to the 4th floor, and then left for America. And, as I heard, recently published a book.
How I rejoiced when, 5 years after that, I got acquainted with the results of the RALES 4 study (first published in December 1995), in which it was shown that veroshpiron must be added to ACE inhibitors (and kapoten belongs here), and this was included in the guidelines for the treatment of heart failure. By the way, digoxin in patients with heart failure and atrial fibrillation is now a Class 1A recommendation.
...Let's imagine for a second that if there were standards then, I could easily have been deprived of my diploma, despite the results of treatment.
Portal "Eternal youth" www.vechnayamolodost.ru17.04.2009