Mechanism vs Demonstrated Benefit

What a mechanism is

The National Cancer Institute defines a drug's mechanism of action as how a drug or other substance produces an effect in the body, for example by affecting a specific target in a cell, such as an enzyme, or a cell function, such as cell growth. [1] Knowing it can be useful: NCI says it may provide information about safety and help identify the right dose and which patients are most likely to respond. [1] Mechanisms guide research. They do not replace its results.

Why a plausible mechanism is not proof

The reasoning "it affects pathway X, X matters in disease Y, so it should help Y" has a long record of failing when tested in people. Two examples:

Suppressing a warning sign made outcomes worse. After a heart attack, irregular extra heartbeats are a risk factor for sudden death. The Cardiac Arrhythmia Suppression Trial tested drugs that suppress them. In the preliminary report, 1,727 patients whose arrhythmia was suppressed were randomly assigned to active drug or placebo. Deaths from arrhythmia and nonfatal cardiac arrests were more common with encainide or flecainide (33 of 730 patients, 4.5%) than with placebo (9 of 725, 1.2%), and total mortality was higher (56 of 730, 7.7%, versus 22 of 725, 3.0%). [2] The drugs were effective at suppressing the arrhythmia, yet the trial found more deaths. [2]

Healthy-diet reasoning did not carry to a supplement. Observational evidence had linked carotenoid-rich foods and higher blood levels of vitamin E and beta carotene with lower lung cancer risk. A randomized, placebo-controlled trial of 29,133 male smokers in Finland then tested supplements. It found no reduction in lung cancer with alpha-tocopherol, and unexpectedly a higher incidence with beta carotene (18% higher, 95% confidence interval 3 to 36%). [3] These findings apply to that population and those products, not to every supplement or every person.

Both trials measured outcomes people care about, such as deaths and cancer cases. That is why they could overturn a mechanism-based expectation.

Lab and animal findings are mechanism-level evidence

Cell and animal studies are mainly about mechanisms and safety signals. A systematic review found that corticosteroids showed benefit in animal models of head injury but not in clinical trials, and tirilazad reduced infarct volume in animals but was associated with a worse outcome in patients with stroke. [4] We label such findings preclinical (cells/animals only). See cell studies vs animal studies vs human trials.

Markers are not outcomes either

A biomarker, such as a molecule in blood, is a sign of a normal or abnormal process. [5] A surrogate endpoint is an indicator used in place of a stronger one, such as longer survival, and NCI states surrogate endpoints are not always true indicators of how well a treatment works. [6] In oncology, 34 of 65 reported links between a surrogate and survival (52%) were of low strength. [7] See surrogate vs patient-important outcomes.

How we write about mechanisms

• We state the mechanism as proposed, with who or what it was studied in.
• We never write that a technology or compound "treats" a condition from a mechanism, a lab association or a marker change alone.
• We separate the field, the mechanism, the product and the claim, as set out in what counts as emerging health research.
• We use evidence labels to show which kind of study supports each statement.

Questions to ask

• Was benefit measured in people, and was the outcome something they would notice?
• Was there a comparison group, such as a placebo?
• Did it work for this exact product and population, or only in a related one?
• Were harms looked for and reported?