Features of Meta-Epidemiology, Meta-Meta- Epidemiology and Network Meta-Epidemiology in Emergency Medicine

Introduction

The effectiveness of treatments ideally comes from randomized clinical trials (RCTs) or systematic reviews of trials that assess final endpoints. Many aspects of the design and conduct of RCTs have been shown to lead to overestimation of treatment effect size. These include [1-7]:

• Inappropriate random sequence generation
• Inadequate allocation concealment
• Lack of blinding
• Single center status
• Use of composite outcomes
• Inadequate intention to treat analysis
• Inadequate double blinding/placebo control
• Meta-Confounders, such as genotype, study design, and the number of participants

The definition of meta-epidemiology was introduced with considering the methodological limitations of systematic review for intervention trials. Meta-epidemiology study aims to describe the distribution of research evidence for a specific issue, to examine the heterogeneity and associated risk factors, and also to control bias between studies and summarize evidence. Diverse methods, such as meta-regression, imputation, informative missing odds ratio, two statistical models, and others, were attempted, and the term metaepidemiology [8-15].

Meta-epidemiology is focused as a research paper not being a simple meta-analysis or narrative review we usually encounter in the literature; it is clearly though a sort of meta-review. In metaepidemiology, one restriction is that informative meta-analyses must include at least one trial with and one without the risk factor of interest, and a minimum number of trials per meta-analysis may be required, depending on how heterogeneity is modelled and multivariable analyses are undertaken [8-15].

The meta-epidemiological, the point of analysis are meta-analysis of randomized controlled trials; for meta-meta-epidemiology, the point are meta-epidemiologic studies, and for network epidemiology, the point are meta-analysis (MA) of randomized controlled trials published where data had been analyzed with a valid statistical method for indirect comparisons or network meta-analysis(NMA) [8-16].

The meta-epidemiology is based on the combination of two concepts: epidemiology and meta-analysis. To fit the purposes of these two concepts, meta-epidemiology strives to achieve the following [16]:

• To describe the distribution of research evidence for a specific question;
• To examine heterogeneity and associated risk factors; and
• To control bias across studies and summarize research evidence as appropriate.

More differences are shown in Table 1 [8-16].

  

    
Meta-epidemiology 
    
Meta-meta-epidemiology 
    
Network    meta-epidemiology 
  
  

    
Data    sources 
    
A    collection of MA of randomized trials 
    
A    collection of meta-epidemiologic studies, combined into a harmonized dataset    without overlap between MA 
    
Networks    of RCTs 
  
  

    
Restrictions 
    
Informative    MA must include at least one trial with and without the risk factor of    interest 
    
The    different meta-epidemiologic studies investigate various sets of risk    factors, potentially assessed with different methods 
    
Eligible    networks must include more trials than interventions 
  
  

    
Trial-level    risk factors 
    
Reassessment    from individual trial reports or reliance on assessment from each selected MA 
    
Assessment    from each meta-epidemiologic study 
    
Reassessment    from individual trial reports or reliance on assessment from each selected    NMA 
  
  

    
Regarding    direction of bias 
    
In    active–inactive comparisons, a risk factor is expected not to favor the    inactive comparator 
    
In    star-shaped networks, a risk factor is expected not to favor the common    comparator 
  
  

    
In    active comparisons, an assumption regarding direction of bias is needed 
    
In    networks with closed loops, an assumption regarding direction of bias is    needed 
  
  

    
Impact    of risk factors on intervention effect estimates 
    
Effect    estimates are compared between trials with and without the risk factor within    each meta-analysis; the mean impact of the risk factor is estimated across    all MA 
    
Effect    estimates are compared between trials with and without the risk factor within    each network; the mean impact of the risk factor is estimated across all    networks 
  
  

    
Impact    of risk factors on intervention effect estimates 
    
Between    trials within MA 
    
Between    trials within networks 
  
  

    
Between    MA 
    
Between    networks 
  

Table 1: