Why we need systematic reviews and metaanalyses

It is generally accepted that the best primary evidence on the effectiveness of healthcare interventions comes from the results of well-designed and well-conducted RCTs. As in any scientific experiment, the observed effect in a clinical trial is composed of the true

Fig. 11.2 Schematic representation of the effect of random and systematic error in a research project. Reproduced with permission from [8].

underlying effect plus the effect of both random and systematic error (Fig. 11.2) (see also Section 3.4.1).

Random error can, of course, never be eliminated, but it can be minimized by evaluating the data from large numbers of patients (see Section 3.4.1 and Chapter 5). Systematic error or bias can arise through poor design or conduct of a study, and its effects can easily be as large or larger than the size of clinically important treatment effects [9]. This is likely to be most problematic if the underlying true effect is moderate. The aim of any systematic review and meta-analysis should be to minimize both types of errors. The systematic component addresses the issue of bias and the meta-analysis element addresses the issue of numbers and random error. The observed effect should then be as close as possible to the true underlying effect, thereby providing a reliable and realistic estimate on which to base treatment policy and future research.

0 0

Post a comment