E.2 Using meta Package
R에서는 meta package가 meta-analysis에 많이 사용된다.
다음과 같이 예제의 자료를 분석할 수 있다.
require(meta)
= c(49, 44, 27, 102, 85, 246) # number of death in the treatment group
e.t = c(615, 758, 317, 832, 810, 2267) # total subjects of the treatment group
n.t = c(67, 64, 32, 126, 52, 219) # number of detath in the control group
e.c = c(624, 771, 309, 850, 406, 2257) # total subjects of the control group
n.c = metabin(e.t, n.t, e.c, n.c)
r6 summary(r6)
RR 95%-CI %W(common) %W(random)
1 0.7420 [0.5223; 1.0543] 11.6 14.3
2 0.6993 [0.4828; 1.0129] 11.0 13.4
3 0.8225 [0.5051; 1.3393] 5.6 9.2
4 0.8270 [0.6487; 1.0545] 21.6 20.9
5 0.8193 [0.5927; 1.1326] 12.0 15.7
6 1.1183 [0.9411; 1.3289] 38.1 26.5
Number of studies combined: k = 6
Number of observations: o = 10816
Number of events: e = 1113
RR 95%-CI z p-value
Common effect model 0.9130 [0.8166; 1.0208] -1.60 0.1099
Random effects model 0.8607 [0.7257; 1.0209] -1.72 0.0851
Quantifying heterogeneity:
tau^2 = 0.0209 [0.0000; 0.1419]; tau = 0.1446 [0.0000; 0.3767]
I^2 = 49.4% [0.0%; 79.9%]; H = 1.41 [1.00; 2.23]
Test of heterogeneity:
Q d.f. p-value
9.88 5 0.0786
Details on meta-analytical method:
- Mantel-Haenszel method
- Restricted maximum-likelihood estimator for tau^2
- Q-Profile method for confidence interval of tau^2 and tau