Meta-analyses

Meta-analyses

1. Introduction

   • used to accumulate evidence and reduce to a manageable size
   • collects data from quantitative studies

2. Aims

   • Refinement and reduction
   • power and precision
   • efficiency
   • generalisability

3. Statistical approach

   • Dichotomous
     • compared using relative risk of odds ratios
       • 1.0 is unity
   • Numerical
     • compared using mean difference
       • 0 is unity

4. Weighting

   • When combining the results of studies those with greater sample sizes and smaller 95% CI have more weight

5. Homogeneity and Heterogeneity

   • If estimates of an outcome between studies are similar (homogeneity)
   • Heterogeneity = p<0.05
     • significance applies to heterogeneity
     • looking for a difference
     • if heterogeneity exists a subgroup analysis may help
   • model type
     • homogeneity
       • fixed-effects model
     • heterogeneity
       • random-effects model

6. Ads and Disads

   • greater precision and power
   • prone to some bias (reporting bias) refers to significant study results
     • publication
     • time lag
     • language
     • multiple publication
     • citation