Bivariate random-effects meta-analysis and the estimation of between-study correlation

Table 3 URMA and BRMA results for the telomerase and CD4 datasets

Dataset	Model	Pooled value endpoint 1 $\hat{β}$ ₁ (s.e.)	Between-study variance endpoint 2 ${\hat{τ}}_{1}^{2}$	Pooled value endpoint 2 $\hat{β}$ ₂ (s.e.)	Between-study variance endpoint 2 ${\hat{τ}}_{2}^{2}$	Between-study correlation $\hat{ρ}$ _B
Telomerase	Normal URMA	1.155 (0.186)	0.186	1.964 (0.541)	2.386	NA
	Normal BRMA	1.166 (0.186)	0.202	2.058 (0.554)	2.584	-1.0
	Generalised URMA	1.182 (0.176)	0.155	2.215 (0.578)	2.680	NA
CD4	Normal URMA	-0.049 (0.0695)	0.025	17.300 (5.561)	379.73	NA
	Normal BRMA	-0.109 (0.0748)	0.048	18.314 (5.740)	412.96	-1.0

s.e. = standard error; NA = not applicable; 95% confidence interval calculated using t-distribution with 9 degrees of freedom. Restricted maximum likelihood estimation was used for the normal models, whereas maximum likelihood estimation was used for the generalised model.

ISSN: 1471-2288