How we label evidence strength
Division of Pulmonary Medicine, Taipei Veterans General Hospital Hsinchu Branch · Dr. Hwa-Yen Chiu
The commonest problem with medical information is not that it says something wrong, but that it states things we are sure of and things we are not with exactly the same confidence.
“Inhaled corticosteroids reduce deaths from asthma” and “intravenous laser improves chronic fatigue” both read like a piece of advice on the page, but the evidence behind them is nowhere near the same level — the first has decades of randomised controlled trials in tens of thousands of people, the second only small studies in a few dozen.
So on this site, every treatment and every test recommendation carries a label for how strong the evidence is. You do not need to become a doctor, but you have a right to know how certain a thing really is.
The four grades
Supported by several high-quality randomised controlled trials or meta-analyses, and listed as standard treatment in international guidelines.
Read it as: unless there is a particular reason not to, you should do it. Both the direction and the size of the effect are reasonably certain, and the results agree across studies.
For example: inhaled corticosteroids for asthma, stopping smoking in COPD, CPAP to improve symptoms in moderate-to-severe sleep apnoea, LDCT lung cancer screening in heavy smokers, the four drug classes for heart failure.
Supported by randomised controlled trials, but limited in size, in consistency or in how widely they apply; guidelines mostly list it as “may be considered”.
Read it as: it makes sense and is worth discussing, but it is not something you have to do. The effect may well be real; it is just smaller, or holds only in a particular group.
For example: an oral anti-snoring device for mild-to-moderate sleep apnoea, acupuncture for nausea caused by chemotherapy, rTMS for fibromyalgia, FeNO as an aid to interpretation in asthma.
Mostly small studies, observational studies or surrogate measures, with no standardisation and no long-term follow-up.
Read it as: you can try it, but treat it as a trial. A placebo effect is hard to rule out, the methods and the results vary widely between studies, and there is no way to be sure it will work for you.
For example: intravenous laser for chronic fatigue, hyperbaric oxygen for anti-ageing and for long COVID, rTMS for tinnitus.
No reliable studies in people, or the studies that do exist are neutral or even negative.
Read it as: there is at present no reason to do this. That does not mean “not yet studied” — some of these have been studied, and the results showed no benefit.
For example: chest X-ray for lung cancer screening (shown not to reduce mortality), intravenous laser as a cancer treatment, external counterpulsation in healthy people to “look after the circulation”.
What do we base the judgement on?
When we assign a grade, this is the order of priority:
- Treatment guidelines from international professional societies — GINA (asthma), GOLD (COPD), ATS/ERS (lung and sleep), IDSA (infection), ESC (cardiovascular), UHMS (hyperbaric oxygen), AASM (sleep medicine) and others.
- Systematic reviews and meta-analyses, particularly the kind that use individual patient data.
- The main results of large randomised controlled trials, and their limitations.
- Taiwanese research and epidemiological data — because lung cancer, tuberculosis and sleep apnoea do not look the same in Taiwan as they do in Europe and North America.
- Domestic regulation and reimbursement rules — indications approved by the Taiwan Food and Drug Administration, and the conditions and review rules for coverage by the National Health Insurance (NHI).
Evidence grade ≠ covered by the NHI
These are two independent axes, and they are constantly mixed up. So this site labels them separately:
Evidence strength: Strong evidence Moderate Limited Insufficient
Payment status: Covered by the NHI Conditions apply Self-paid
Some treatments have strong evidence but are not covered by the NHI (some newer drugs, for instance, and self-paid screening); others are covered, but the evidence is only moderate. Knowing that these are two different things makes a decision much clearer.
Three things that are often misunderstood
1. “Safe” does not mean “effective”
Many complementary therapies do have a good safety record, with few adverse events. But safety is only a threshold; it is not proof of effect. A thing can be both safe and of no help — and then what you pay is time, money, and possibly a delay to treatment that does work.
2. “Insufficient evidence” does not mean “it certainly does not work”
Here is where science is honest: no evidence of benefit, and proof of no benefit, are two different things. Some treatments simply have not been studied properly yet. But until the evidence arrives, it should not be recommended as though it were already settled, and still less should it replace treatment that already works.
3. Grades change
Medical evidence moves. Today’s “limited evidence” may be upgraded in three years because of one large trial, or downgraded because the trial came out neutral. The labels on this site reflect the position at the time they were written, not a permanent verdict. If you see a label on a page that does not match the latest research, do tell us on LINE.
How to discuss each grade with your doctor
| Grade | A sensible way to ask |
|---|---|
| Strong evidence | “Is there any reason this would not be suitable for me?” — the focus is on contraindications and side effects, not on whether to do it at all. |
| Moderate | “In my situation, how much improvement would you expect from this? Is there an option with stronger evidence I could try first?” |
| Limited | “If I try it, how many sessions, what measure tells us whether it is working, and how long without improvement before we stop?” — agree the stopping rule first, so it does not become an open-ended course of treatment. |
| Insufficient | “Is there an alternative with stronger evidence?” — if your doctor still recommends it, ask exactly why. |
One principle that runs through the whole site
Treatments with strong evidence come first, and treatments with weak evidence do not replace treatments with strong evidence.
A complementary therapy can be added on top of existing treatment, but it should not take its place. Heart medication, medication for depression, diabetes control, asthma inhalers — none of these becomes something you can stop because you have started a course of some therapy.
Where these labels are used
- Symptoms: every test suggested carries an evidence grade, with a note on what question it answers.
- Wellness: for each of the five non-drug treatments, every indication is labelled with both the strength of the evidence and the payment status.
- Clinical pathways （中文）: every recommendation the tools produce carries the guideline class of recommendation (Class I / IIa / IIb / III), shown in the same colour scheme.
- Health Topics: where a treatment is described, the evidence behind it and its known limitations are stated.
Questions, or found a mistake?
Medical knowledge is updated, and this write-up may have gaps. If you find a label that does not match the latest research or guidance, please tell us — it helps every reader.
This page explains how evidence is labelled on this site, so that readers can judge how certain the content is. All labels are health information only; they cannot replace a doctor’s professional judgement about your individual situation.