Meta Analyses and Systematic Reviews

SYSTEMATIC REVIEWS AND META-ANALYSES

Discussion: Systematic Reviews and Meta-Analyses

According to Corcoran and Pillai (2008) systematic reviews come in handy in not only the identification, but also the synthesis of the evidence considered to be of relevance to the research query. With this in mind, systematic reviews could be considered an enhanced form of the traditional literature review approach. This is more so the case given that systematic reviews makes use of, or deploy, systematic methods in the combination as well as assessment of available evidence (Corcoran and Pillai, 2008). In a far as validity of systematic reviews is concerned, Patole (2021) observes that this has got to do with “how trustworthy the review's conclusions are for a reader” (193). It is important to note that in epidemiology, systematic reviews could be used in the formulation of strategies meant to impact the health and wellbeing of populations. For this reason, the relevance of relying on conclusions that could be deemed trustworthy cannot be overstated. The said trustworthiness can be assessed in terms of validity. On this front, two concepts come in handy, i.e. internal validity and external validity. Internal validity, as Patole (2021) points out, relates to the strength of the study methodology. Towards this end, as the author further indicates, the issue of greatest relevance happens to be bias extent. Thus, in the context of epidemiology, what one could look out for to ascertain validity is, amongst other things; performance bias and selection bias. On the other hand, external validity could be thought of in terms of the generalizability of systematic review results. Thus, when it comes to epidemiology, the all-important question on this front happens to be whether the results of the systematic review can be generalized to populations not captured therein. To ascertain external validity, one would ideally be interested in factors such as the characteristics of participants, sample size etc.

When it comes to meta-analysis Uman (2011) makes an observation to the effect that “often, systematic reviews include a meta-analysis component which involves using statistical techniques to synthesize the data from several studies into a single quantitative estimate or summary effect size” (Uman, 2011, p. 203). The author indicates that for a meta-analysis to be deemed valid, it ought to be representative of the research literature review. There are a number of factors that could be considered in the assessment of the extent to which a meta-analysis is valid. Some of the said factors are inclusive of; the inclusion criteria deployed, whether or not there was an appraisal of the included studies’ validity, homogeneity (i.e. the similarity of results across studies), data abstraction, etc.

In the final analysis, it would be prudent to note that one strength of both approaches happens to be a high level of evidence. This, as Corcoran and Pillai (2008) point out, is well captured in the evidence pyramid. As a result, the authors come to the conclusion that “a well-conducted systematic review and meta-analyses is considered a feasible solution in keeping health clinicians ahead regarding contemporary evidence-based medicine” (Corcoran and Pillai, 2008, p. 211). However, perspectives in systematic reviews could in some instances be biased by failure to include gray literature. Examples of gray literature on this front could be inclusive of, but is not limited to policy documents and government reports. Secondly, Corcoran and Pillai (2008) indicate that the absence of universal standards on various fronts (i.e. with regard to evidence evaluation) could result in poor quality or uncertain systematic reviews. It should also be noted that the problem of poor execution of meta-analyses could be a key limitation as it could make the same invalid. This is more so the case when appropriate studies are carelessly summarized or abstracted. Also, effects estimates could in some cases be meaningless – especially in those instances where different causal factors are grouped.