Table 2.
Between-study heterogeneity for the Bayesian and frequentist random-effects models
| Exposure | Bayesian (values based on the posterior distribution of τ) | Frequentist random-effects | ||
|---|---|---|---|---|
| τ2 (95% CrI) | I2 (95% CrI) | τ2 | I2 | |
| Dietary intake | ||||
| Riboflavin | 0.001 (0–0.046) | 1.7 (0–51.0) | 0.000 | 0.000 |
| Vitamin B6 | 0.001 (0–0.023) | 1.2 (0–33.6) | 0.000 | 0.000 |
| Folate | 0.001 (0–0.039) | 2.1 (0–49.1) | 0.007 | 15.360 |
| Vitamin B12 | 0.001 (0–0.046) | 1.5 (0–48.0) | 0.000 | 0.000 |
| Methionine | 0.001 (0–0.1) | 1.9 (0–64.6) | 0.000 | 0.000 |
| Choline | 0.001 (0–0.084) | 1.7 (0–60.1) | 0.000 | 0.000 |
| Betaine | 0.002 (0–0.18) | 2.9 (0–76.8) | 0.076 | 58.357 |
| Biomarker status | ||||
| Riboflavin | 0.001 (0–0.077) | 1.4 (0–54.8) | 0.000 | 0.000 |
| Vitamin B6 | 0.047 (0–0.807) | 37.0 (0–90.9) | 0.312 | 79.384 |
| Folate | 0.001 (0–0.09) | 1.4 (0–56.4) | 0.000 | 0.000 |
| Vitamin B12 | 0.001 (0–0.097) | 1.4 (0–57.7) | 0.000 | 0.000 |
| Homocysteine | 0.001 (0–0.079) | 1.3 (0–53.9) | 0.000 | 0.000 |
The Bayesian method allows for uncertainty in the estimation of heterogeneity and therefore 95% CrIs are reported for the Bayesian model parameters