• EN    |    TH
  • +66 2-038-5115

Scientific evidence reliability is from the most reliable to unreliable level.

The principle of Level of evidence – LOE of each institute has the similar principle. The one explained here is from ILCOR. It is difficult to understand the LOE of Medical Science evidence; even some physicians do not totally understand this matter. However, though it is difficult, it is unavoidable to talk about it otherwise the information on the internet would rather destroy the reader’s health than to better it. To classify LOE based on ILCOR criteria [1], it can be divided into five level from the highest to the lowest as follows.

Level 1 Research (High reliability) which is the randomized controlled trial - RCT). It means the research that randomizes the patients into two groups and compares the treatment method, traditional and intervention treatment.

Randomization is the method to delete the exceeding confounding factors of each side. It is to delete the placebo effect to the disease. The example of level 1 research is the research on heart disease treatment with the change of lifestyle conducting by Dr. Dean Ornish [2,3] that aimed to answer the question whether the adjustment of lifestyle had the effect of heart disease treatment or not. He randomized 93 patients with Ischemic heart disease who already had cardiac catheterization & angiography and grouped them into two groups. The first was the control group who received the normal treatment and the second group was the intervention group who had to adjust their lifestyle completely in four aspects:

(1) ate only vegetarian food cooked without oil and non-polished,

(2) had regular exercise,

(3) had activity to release stress everyday and

(4) joined the group “friend helps friend” once a week.

Then, following up for 5 years to examine the difference between two groups by measuring the constriction of a coronary artery, the frequency of chest pain and the number to have treatment in hospital as the indicator.

After one year, made an appointment with all patients to have cardiac catheterization & angiography and assessed the indicator. It was found that the group that adjusted lifestyle to eat vegetarian food cooked without oil and non-polished (did not eat meat, milk, egg or fish), do exercise and have stress management had the vasodilator at 4.5% while the control group had the narrower vessel for 4.5%. The number was close but in a different way. Regarding chest pain, it was found that the group that adjusted lifestyle to eat low-fat vegetarian food, do exercise and have stress management had “less” chest pain for 91% whereas the control group had more chest pain for 165%.

After the completion of research, he followed up this two groups to complete 5 years and made cardiac catheterization & angiography and assessed the indicator again. The result showed the same difference but more obvious. That was to say when measuring the constriction, it was found that the group that adjusted the lifestyle had the “wider” constriction for 7.9% while the control group had the “narrower” constriction for 27.7%. When checking on the rate to admit in the hospital, it was found that the group that adjusted the lifestyle had to admit to the hospital 0.89 times at the average while the average of the control group was 2.25 times.

Thus, he concluded that to have patient with heart disease to adjust their lifestyle by eating vegetarian food cooked without oil and non-polished, do exercise and have stress management as well as join the helping group had the reverse result. The constrictive vessel could be expanded.

Level 2 Research (Moderate reliability) is the prospective cohort study. It is to register all patient, examine, record the indicators and keep examining at the interval to see what kind of risk factor or intervention they have and how it affect the disease development or the recuperation. This group of people obtained from the volunteer without random. To divide into a group based on the difference of the original of an individual.

The sample of level 2 research is the research of Harvard to find out that what kind of fat caused the illness and death from heart disease the most [4]. He began his research by interview 80,082 normal people about nutrition at the interval. In 14 years, there was 939 severe and dead patient from heart disease. Studying eating fat behavior compare to the chance to be severe ill and dead from heart disease, it was found that every 5% of calories the patient consumed comparing to the calories from carbohydrate indicated that the calorie from trans fat (vegetable oil with hydrogen such as non-dairy creamer, margarine, cake, cookies, and snacks) was “more” related to the severe illness and death than calorie from carbohydrate for 93%.

Calories from saturated fat (such as lard and palm oil) were “more” related to the severe illness and death than calories from carbohydrate for 17%.

Calories from monounsaturated fat (i.e. vegetable oil such as olive oil) were “less” related to the severe illness and death than calories from carbohydrate for 19%.

Calories from polyunsaturated fat (i.e. vegetable oil such as soy bean oil) were “less” related to the severe illness and death than calories from carbohydrate for 38%.

Therefore, the researcher team concluded that trans fat was the worst calorie source that was related to the severe illness and death the most.

Prospective cohort study suits with the small amount of participant. However, if there is the good research design with the monitoring of indicators, it may have wider benefits. For instance, the research of Dr. Caldwell B. Esselstyn from Cleveland Hospital[5]. He conducted the research with 24 patients with Ischemic heart disease in severe stage and had experience in cardiac catheterization & angiography. Eight years earlier, these people had the heart attack 49 times. He let them eat vegetarian food cooked without oil and non-polished. During the first year, 6 patients could not stand the vegetarian food and quit. After 5 years observation of 24 patients, 6 patients who quitted had 13 acute heart attack whereas those who continued eating vegetarian food cooked without oil and non-polished had no acute heart attack. After 10 years, the second group still had no acute heart attack. At 5 years, 11 patients who ate vegetarian food agreed to do cardiac catheterization and it was found that none of them had more severity from the past 5 years. 8 of them (73%) had a reverse disease.

Besides, it reported that the average cholesterol of the group decreased from 246 mg/dl to lower than 150 mg/dl. The result of cardiac catheterization indicated that from 25 constrictions, 11 were wider while 14 were stable after 10 years. Further, this research illustrated the result of cardiac catheterization to see the patient’s vessel before the research which was long-narrow and rough that became the wide vessel as usual after eating vegetarian food for 3 years.

From this research, it was clear that though those who had been intervened and monitored was in a small group without random the comparative groups, with the monitoring method and clear evidence, it was acceptable that was used as the reference widely.

In general, the reason that the research without random sampling had less reliability than the research with random sampling is that the research without random sampling sometimes has a confounding factor that causes the illness which the researcher does not aware of. Therefore, the research without random sampling or cohort research should look for and resolve confounding factor problem for the reliable results.

The sample of confounding factor result is the research on monitoring two groups of Japanese people in Hawaii, drinking and not drinking coffee by studying on those who drank/not drank coffee without random sampling. After 5 years, it was found that those who drank coffee were ill and died from Ischemic heart disease more than those who did not drink coffee. Thus, it led to the faulty conclusion that coffee caused illness and death. Medical field finally found out later that this assumption was wrong. This was because there was the confounding factor in the research, which was smoking. In this research, the rate of people who drank coffee was higher than those who did not drink coffee regularly. When grouping them into a group of drinking people and non-drinking people, the smokers were in the group of drinking more than in the non-drinking group. A cigarette was the actual cause of death from Ischemic heart disease. With this reason, the group of drinking coffee had higher death rate than the non-drinking group. After the new analysis by excluding the smokers from both groups, the conclusion revealed that coffee did not increase the rate of illness and death from Ischemic heart disease.

Level 3 Research (Lower-Moderate Reliability) which is the retrospective study. The sample is the research to find out whether drinking coffee relates to non-dementia or not [6]. The research consisted of 54 patients with dementia and 54 normal people who had similar age and characteristic without dementia. Then, the researcher examined the quantity of coffee drinking within the past 20 years of both groups. It was found that the group of patients with dementia drank coffee containing caffeine 73.9 mg/day at the average (equivalent to a half cup of coffee) while those with non-dementia drank coffee with caffeine 198.7 mg/day (equivalent to one and a half cup of coffee). Therefore, it was concluded that people who drank coffee had less relation to dementia than those who did not drink coffee.

If comparing the research without random sampling, the retrospective research had more chance of confounding factor than the prospective cohort research. Thus, retrospective research has the less reliability in the third rank following RCT and prospective cohort research. Moreover, it is frequent that retrospective research concludes the fault conclusion which normally found out after the second research that applying prospective cohort research or RCT and obtains the opposite result.

However, the research on results of eating and exercising behavior for the good health should be monitored at least 10 or 20 years. The level 1 research, which is the randomized controlled trial cannot be implemented because it cannot force people to do something such that long period. With this reason, the medical field has to use research result from prospective cohort or epidemiological research as a basement by considering the overall. It means that if conducting research with several communities or countries, it would have the same result and the reliability will increase though it is not the level 1 research.

Level 4 Research (Low Reliability) is the patient report that does not have research methodology and case series, only the narration that some treatment has been done with the patients and what is the result.

Level 5 Research (Unreliable and unable to implement to human) such as animal model, laboratory model, mechanical model or extrapolation. Medical and scientific area consider them as the evidence that is unable to implement to human. However, there is a claim for this kind of research to support health products and services selling on the internet.

Furthermore, the information publishing on the internet is another kind of information that is not one of scientific evidence levels mentioned above. It is called non-level evidence such as anecdote, testimonial, expert opinion which normally the fake expert or the expert with hidden intention to sell products. Most of the information is the prediction or the fault information to make people misunderstand scientific evidence. Importantly, the information dispersing on the internet and online media is considered to be in this group (approximately 80% of all information).