Clinical Trials Chapter 6

Question 1

What is meta-analysis?

Answer 1

A statistical method to combine results from multiple studies that answer similar research questions, these studies are often identified by a literature search and have to meet the inclusion criteria

Question 2

What is the fixed effect model?

Answer 2

Suppose that there are n independent studies in the analysis. There is a common outcome reported for each individual. The parameter representing the overall treatment effect is denoted by theta.
The observed treatment effect in a study is given by
theta_i hat = theta + error_i , I=1,…,n

where error_i is the difference between the common true effect and the observed effect for study i.
error_i ~N(0,s_i^2)
with s_i^2 as the within-study error variance

Question 3

Assumptions of the fixed effect model

Answer 3

The fixed affect model assumes that all studies in the meta-analysis share a common true treatment

Put another way, all factors which could influence the treatment effect are the same in all the study populations

It follows that the observed treatment effect varies from one study to the next because of the sampling error inherent in each study

Question 4

Fixed effect model

Estimation for the overall effect

Answer 4

theta_hat = sum for i=1 to n theta_i hat w_i/ sum from i=1 to n w_i

w_i = 1/s_i^2
confidence interval
theta_hat +/- 1.96 x sqrt(1/sum from i=1 to n w_i)

Question 5

Fixed effect model

U statistic

Answer 5

U = (sum for i=1 to n theta_i hat w_i)^2/ sum from i=1 to n w_i

under H_0 the U statistic follows a Chi-square distribution with one degree of freedom