Table 2.
Association between weather variables and Re.
| +Variables | Contrast for which effect size is presenteda | Effect size 95% CI | P value | Difference in the likelihood ratio RLR2 statistic |
|---|---|---|---|---|
| Mean temperature (°C) | 10.2 vs 20 | 0.087 (0.025; 0.148) | 0.014 | +2.5 |
| Absolute humidity (g/m3) | 6.6 vs 11 | 0.061 (0.011; 0.112) | 0.036 | +2.0 |
| Relative humidity (%) | 60 vs 75 | 0.043 (−0.001; 0.087) | 0.058 | +1.5 |
| Surface solar radiation downwards (J/m2) | 248 vs 124 | −0.053 (−0.117; 0.011) | 0.208 | +0.6 |
| Wind speed (m/s) | 1.1 vs 3.0 | −0.038 (−0.090; 0.014) | 0.152 | +0.7 |
| Total precipitation (m) | 0.1 vs 6 | −0.031 (−0.075; 0.014) | 0.175 | +0.4 |
| OxCGRT (0–100) | 21 vs 66 | 0.285 (0.223; 0.347) | <0.0001 | +13.8 |
Effect size and variation explained by including, in turn, mean temperature (°C), absolute humidity (g/m3), relative humidity (%), surface solar radiation downwards (J/m2), wind speed (m/s), total precipitation (m) and OxCGRT (0–100) in the model of Re. P values were obtained from a two-sided Wald test in the multivariable meta-regression multilevel models adjusted by population (log scale), population density (log scale), GDP (log scale), % population >65 years, PM2.5 (μg/m3, log scale) and the OxCGRT Government Response Index, with cities nested within countries.
aThe exposure contrast for which effect size is presented is that between the values predicting minimum and maximum Re, where clear minima and maxima are observed (mean temperature, absolute humidity and relative humidity), otherwise the 5th to 95th percentiles.