Evidence synthesis is central to modern clinical and epidemiological decision making. Yet many published meta analyses suffer from conceptual misunderstandings, inappropriate model choices, and limited clinical interpretability. This one day intensive course provides epidemiologists and medical doctors with a clear and practical introduction to meta analysis using R.The course focuses on how to perform a meta analysis correctly and how to interpret results in a clinically meaningful way.

Participants will learn how to compute effect sizes, fit fixed effects and random effects models, quantify heterogeneity, evaluate bias, and translate statistical results into clinical conclusions. A major emphasis is placed on deciding when random effects meta analysis is appropriate. Instead of relying on mechanical rules or numerical thresholds, the course explains how model choice must follow the scientific question. Random effects models are justified when studies represent a distribution of true effects across different populations, interventions, or clinical settings. Participants will learn how to interpret tau squared, how to use prediction intervals to express clinical uncertainty, and why random effects models often provide more realistic and clinically relevant conclusions.

The course also provides a detailed critique of the I squared statistic, which is widely used but frequently misunderstood. I squared does not measure the magnitude of heterogeneity, is strongly influenced by sample size, and can be misleading in both small and large meta analyses. Participants will learn why they must exercise caution when using I squared, and how to rely instead on tau squared, prediction intervals, and conceptual reasoning grounded in clinical and epidemiological judgment.

By the end of the day, epidemiologists and medical doctors will be able to conduct a complete meta analysis in R, evaluate published evidence with greater confidence, and apply best practices that avoid common pitfalls. The course is ideal for clinicians, epidemiologists, public health researchers, and medical academics who want a modern and principled approach to evidence synthesis.

09:00 to 09:20: Course Overview and Motivation

• • Why meta analysis matters for epidemiology and clinical research

  • Strengths and limitations of evidence synthesis

  • The Borenstein perspective: conceptual clarity first

09:20 to 10:00: Effect Sizes in Clinical and Epidemiological Research

• • Mean difference and standardized mean difference

  • Risk ratio and odds ratio

  • Conversions and interpretation

  • Clinical versus epidemiological meaning

10:00 to 10:40: Fixed Effects versus Random Effects: Conceptual Foundations

• What question each model answers

• Why model choice must follow the scientific question

• Why I squared should not determine model choice

• When clinicians and epidemiologists should prefer random effects

10:40 to 11:00: Coffee Break

11:00 to 12:00: Deciding When to Use Random Effects Meta Analysis

• Real world heterogeneity in clinical and epidemiological practice

• Variation in populations, interventions, and settings

• Tau squared as the true measure of heterogeneity

• Prediction intervals for clinical interpretation

• Why model choice is conceptual rather than mechanical

12:00 to 12:30: Critique of the I Squared Statistic

• • 
    Why I squared does not measure heterogeneity magnitude

  • Dependence on sample size

  • Misleading values in small meta analyses

  • Misleading values in large meta analyses

  • Better alternatives: tau squared, prediction intervals, conceptual reasoning

12:30 to 13:30: Lunch Break

13:30 to 14:30: Hands On Session: Meta Analysis in R

• • Computing effect sizes

  • Running fixed effects and random effects models

  • Extracting tau squared, I squared, and prediction intervals

  • Forest plots and interpretation

14:30 to 15:00: Bias, Small Study Effects, and Sensitivity Analysis

• • Funnel plots

  • Egger test

  • Trim and fill

  • Influence diagnostics

15:00 to 15:30: Clinical and Epidemiological Interpretation and Reporting

• Translating statistical results into clinical meaning

• Communicating uncertainty

• Avoiding common misinterpretations

• PRISMA and reporting standards

15:30 to 16:00: Closing Discussion and Questions

• Integrating statistical rigor with clinical judgment

• Evaluating published meta analyses

• Final conceptual takeaways