Project:No original research

Articles report what sources say; they do not draw new conclusions.

No original research prohibits the publication of analysis, synthesis or interpretation not already present in a published source. In practice on this wiki it prohibits three specific things: inferring a human effect from an animal study, combining two sources to reach a conclusion neither states, and computing a result and presenting it as a finding.

The commonest violation in this subject area. A rodent study showing accelerated tendon healing does not support a sentence about human tendon healing, however plausible the mechanism. The correct treatment is to describe the study, its model, and the absence of human data — which is more informative than the inference would have been.

Source A reports that a compound is degraded by an enzyme. Source B reports that a supplier's material contained that compound. Writing that the supplier's material would therefore have been degraded is synthesis, and is not permitted unless a source draws the conclusion.

Unit conversion, percentage change, and simple arithmetic on published figures are permitted as routine calculation, provided the inputs are cited and the operation is stated. The worked examples in Reconstitution of lyophilised peptides and Number needed to treat fall on this side of the line: every input is sourced and every step is shown.

A calculation becomes original research when it requires a modelling assumption. Estimating a plasma concentration from a dose requires assumptions about volume of distribution and bioavailability, and is therefore not routine.

The four calculator articles present arithmetic, not conclusions. They must show their formulae, state their assumptions, and refuse to produce a clinical recommendation. See Project:Medical disclaimer.