Meta-analyzability concerns in meta-analyses.

Publication date: Available online 25 January 2019

Source: Archives of Physical Medicine and Rehabilitation

Author(s): Chih-Yang Hsu, Yen-Nung Lin

Abstract

Many studies to date have conducted a meta-analysis on a mix of effectiveness and superiority studies. This methodological flaw will lead to difficulties in interpreting the results. We addressed this issue in this article, illustrated our point with a simulated experiment, and re-analyzed a recent meta-analysis study based on the effectiveness-superiority dichotomy to provide a real-world correlate of our point of view.

The number of meta-analyses has increased astonishingly over the past few decades. (Figure 1) In 2016 alone, approximately 13 thousand new meta-analysis studies were indexed in PubMed, more than 200 of which are rehabilitation-related. The sophistication and availability of statistical software may have partially contributed to the prosperity. Here we would like to address an important issue which has usually been neglected—the meta-analyzability.

Meta-analyzability refers to the rationale to pool a bundle of trials with heterogeneity in a meta-analysis. The heterogeneity can exist in the participants, interventions, comparisons, or outcomes (PICO). Among them, the heterogeneity in comparison seems to have received less, and perhaps inadequate attention. The majority of trials exploring the beneficial effects of an experimental intervention (EI) can be classified into two kinds of study: effectiveness studies or superiority studies. The former compare the outcomes between the group receiving an EI and the group not receiving an EI, while the latter compare the outcomes between the group receiving an EI and the group receiving the standard or competing intervention (CI). The comparison formulas can be expressed as below:

from Rehabilitation via xlomafota13 on Inoreader http://bit.ly/2CHuArI
via IFTTT