In the previous post, the levels of evidence were classified using two systems. The first of those was proposed by Sackett. However, the second one is currently in favour. Since the latter is more comprehensive, I will discuss that system here.

Let us understand what the various levels really mean:

1a: Evidence obtained from Meta Analysis of Randomized Trials

Meta Analysis: A statistical procedure in which the results of many (small and large) studies are pooled. The analysis gives different weights to the studies based upon their sizes. Larger studies contribute more towards the final (pooled) results, and vice versa. The pooled results are considered to be much more accurate than those reported by the individual studies included in the analysis.

Randomized Trial: An experimental study in which participants are allocated to the various study arms based upon the result of randomization. It is the only study design that can test a hypothesis, hence held in high esteem.

Randomization: A statistical procedure that distributes study participants in such a way that the various arms of the trial are similar with respect to their characteristics (age, sex, etc.). Randomization (or the lack of it) significantly influences the findings of a trial. Trials that employ randomization are therefore considered superior to those that are non-randomized.

A well conducted Randomized Controlled Trial is considered to generate the best quality of evidence. Therefore, a meta analysis of such studies is expected to provide top-class evidence.

1b: Evidence obtained from at least one Randomized Trial

From the explanations provided above, one can see why the next level of evidence insists on evidence from at least one randomized trial. It is assumed that such evidence will be superior to that obtained from other study designs.