Statistical Methods for Meta-Analysis with Life Science Applications

This is a two-day course on statistical methods for meta-analysis using the package STATA. The first day gives an overview on traditional techniques used in meta-analysis. The second day present more recent state-of-the-art modelling including mixed Poisson and binomial regression. The teaching style of the course is a mix of lectures and practical work. 

The course covers: 

• Basic elements: Effect measures of interest; Mantel-Haenszel, case studies, continuous outcomes; graphical techniques, precalculated effect measures
• Heterogeneity: why is there heterogeneity, study as random effect, sub-group analysis, meta-regression
• Meta-analysis of rare events: what are the issues? Mantel-Haenszel estimation 
• Modelling approaches for rare meta-analysis: Poisson and binomial with fixed and random effects
• Meta-analysis for rare events: single- and double-zero studies and their effect
• Conditional logistic regression

By the end of the course participants will:

•  Be able to do a simple statistical meta-analysis using traditional techniques
•  Be able to do a more advanced model-based meta-analysis 
•  Know in which meta-analysis setting to apply the most appropriate statistical approach 

Preparatory Reading

Introduction to Meta-Analysis, second edition (authors: Michael Borenstein, Larry V Hedges, Julian P T Higgins and Hannah R Rothstein.

Knowledge in STATA is helpful but not an essential pre-requisite.

This course will take place at the University of Southampton from 9am to 5pm both days (please note refreshments will be provided but lunch will not).

Course Description

Programme

DAY 1

9.00 -9.30    Registrations and coffees

9.30-11.00    L1: Basic elements: Effect measures of interest; Mantel-Haenszel, case studies, continuous outcomes; graphical techniques, precalculated effect measures

11.00-11.30    Morning break – Coffees and biscuits 

11.30-12.30    P1: Basic elements

12.30-14.00    Lunch

14.30-15.30    L2: Heterogeneity: why is there heterogeneity, study as random effect, sub-group analysis, meta-regression

15.30-16.00    Afternoon break – Tea and biscuits

16.00-17.00     P2: Heterogeneity

DAY 2

9.00-10.00    L3: Meta-analysis of rare events: what are the issues? Mantel-Haenszel estimation 

10.00-10.30    Morning break – Coffees and biscuits

10.30-11:30    L4: Modelling approaches for rare meta-analysis: Poisson and binomial with fixed and random effects

11.30-12.30    P4: Modelling approaches for rare meta-analysis

12.30-14.00    Lunch

14.00-14.30    L5: Meta-analysis for rare events: single- and double-zero studies and their effect

14.30-15.00    L6: Conditional logistic regression

15.30-16.00    Afternoon break – Tea and biscuits

16.00-16.30    P6: Conditional logistic regression

16.30-17.00    Discussion

L = Lecture, P= Practical