Number of studies (k) |
k = 7 |
k = 9 |
k = 6 |
Log odds ratios |
Range: −1.11 to −0.11 |
Range: −1.15 to −0.35 |
Range: −1.63 to 0 |
Majority of estimates |
Negative (100%) |
Negative (100%) |
Negative (in 4 out of 6 studies, 67%) |
Estimated average log odds ratio |
\hat{\μ} = −0.7 |
\hat{\μ} = −0.8 |
\hat{\μ} = −1.08 |
95% confidence interval |
−1.05 to −0.35 |
−1.11 to −0.49 |
−2.34 to 0.17 |
Significance of average outcome |
z-score = −3.95, p-value < 0.001 |
z-score = −5.1375, p-value < 0.001 |
z-score = −1.6953, p-value = 0.09 |
Maximum weightage |
Yin et al. [9] (31.6%) |
Yin et al. [9] (24.2%) |
Rossi et al. [8] (32.3%) |
Heterogeneity |
Q-test p-value = 0.81, tau2 = 0, I2 = 0% |
Q-test p-value = 0.95, tau2 = 0, I2 = 0% |
Q-test p-value = 0.98, tau2 = 0, I2 = 0% |
Outliers |
No outlier studies, Studentised residuals < ±2.69 |
No outlier studies, Studentised residuals < ±2.77 |
No outlier studies, Studentised residuals < ±2.64 |
Influential studies |
None, Cook’s distances < 1 |
None, Cook’s distances < 1 |
None, Cook’s distances < 1 |
Funnel plot asymmetry |
Rank correlation p-value = 1, regression test p-value = 0.73 |
Rank correlation p-value = 1, regression test p-value = 0.98 |
Rank correlation p-value = 0.14, regression test p-value = 0.49 |