Article type
Year
Abstract
Background: Exhaustive systematic reviews require substantial time and other resources to conduct. The time-to-results often exceed the time-to-decision and thus may delay and/or impede evidence-based policy. This impediment has resulted in dialogue and exploration of rapid reviews, albeit with minimal impact on evidence-based policy. Cumulative meta-analysis has been characterized as a means for reviewing evidence as it is created, e.g., the second, then the third study, etc. until a consistent effect has been revealed and then theoretically additional studies would no longer be necessary. It hasn’t caught on—enough said! Recently, statistics have been developed that use cumulative meta-analysis that provide measures of sufficiency (of evidence) and stability (of cumulative effect size). Cumulative meta-analysis incorporating sufficiency and stability statistics may provide a pathway for reducing time-to-results and therefore enhance evidence-based policy.
Objectives: Examine the utility of sufficiency and stability statistics in cumulative meta-analysis in reducing time-to-results when studies are added randomly without respect to date of publication.
Methods: Fifteen published meta-analyses were selected and the studies within were used to develop five random sorts. Cumulative meta-analyses were conducted with Stata on each of the random sorts, and were also performed on both forward and reverse sorting by date of publication. Sufficiency and stability statistics were then calculated for each wave of each cumulative meta-analysis.
Results: Sufficiency and stability of studies were obtained (for some but not all of the selected systematic reviews used in this study) with fewer n-of-studies that had been included in the published exhaustive review. Sufficiency and stability did not appear to be affected by sort.
Conclusions: Cumulative meta-analysis, when incorporating sufficiency and stability statistics may reduce time-to-results compared to exhaustive reviews. Additional statistical parameter research is suggested for this methodology.
Objectives: Examine the utility of sufficiency and stability statistics in cumulative meta-analysis in reducing time-to-results when studies are added randomly without respect to date of publication.
Methods: Fifteen published meta-analyses were selected and the studies within were used to develop five random sorts. Cumulative meta-analyses were conducted with Stata on each of the random sorts, and were also performed on both forward and reverse sorting by date of publication. Sufficiency and stability statistics were then calculated for each wave of each cumulative meta-analysis.
Results: Sufficiency and stability of studies were obtained (for some but not all of the selected systematic reviews used in this study) with fewer n-of-studies that had been included in the published exhaustive review. Sufficiency and stability did not appear to be affected by sort.
Conclusions: Cumulative meta-analysis, when incorporating sufficiency and stability statistics may reduce time-to-results compared to exhaustive reviews. Additional statistical parameter research is suggested for this methodology.