Table 3.
Random-effects meta-regression of crude case fatality risk over the first wave up to September 20, 2020.
| MODEL 1: a priori model | MODEL 2: bootstrap variable selection | |||||||
|---|---|---|---|---|---|---|---|---|
| Crude case fatality risk | Crude case fatality risk | |||||||
| Covariates | β | SE | P-value | OR | β | SE | P-value | OR |
| Intercept | −9.5118 | 1.3360 | – | – | −10.6256 | 1.2843 | – | – |
| Sociodemographic factors | ||||||||
| GDP per capita ($1,000 USD, 2019) | −0.0005 | 0.0084 | 0.950 | 1.00 | – | – | – | – |
| Urban population (%) | – | – | – | – | −0.0297 | 0.0079 | <0.0001 | 0.97 |
| Elderly dependency ratio (% of adults) | – | – | – | – | 0.1417 | 0.1004 | 0.164 | 1.15 |
| Proportion over 65 years (%) | 0.0256 | 0.0268 | 0.343 | 1.03 | −0.2221 | 0.1629 | 0.178 | 0.80 |
| Proportion overweight (%) | – | – | – | – | 0.0326 | 0.0108 | 0.004 | 1.03 |
| Proportion smoker (%) | – | – | – | – | −0.0201 | 0.0121 | 0.103 | 0.98 |
| Pandemic–related factors | ||||||||
| Time since 1st case (days) | – | – | – | – | – | – | – | – |
| Time since 100 cases (days) | – | – | – | – | – | – | – | – |
| Time since 1st death (days) | 0.0246 | 0.0066 | <0.0001 | 1.02 | 0.0314 | 0.0060 | <0.0001 | 1.03 |
| Testing coverage (n. tests per 10,000 pop) | −0.0153 | 0.0070 | 0.033 | 0.98 | −0.0112 | 0.0051 | 0.033 | 0.99 |
| Health system strength | ||||||||
| Healthcare workers (n. per 1,000 pop) | 0.0044 | 0.0330 | 0.895 | 1.00 | – | – | – | – |
| Hospital beds (n. per 1,000 pop) | −0.1352 | 0.0514 | 0.011 | 0.87 | −0.1055 | 0.0595 | 0.082 | 0.90 |
| Health expenditure (% of GDP) | – | – | – | – | – | – | – | – |
| Cultural characteristics | ||||||||
| Individualism vs. collectivism | 0.0147 | 0.0059 | 0.015 | 1.01 | 0.0123 | 0.0060 | 0.047 | 1.01 |
| Uncertainty avoidance | 0.0124 | 0.0052 | 0.019 | 1.01 | 0.0120 | 0.0056 | 0.037 | 1.01 |
| Indulgence vs. restraint | – | – | – | – | 0.0138 | 0.0055 | 0.015 | 1.01 |
| Long-term vs. short-term orientation | – | – | – | – | 0.0192 | 0.0063 | 0.004 | 1.02 |
| Power distance | – | – | – | – | – | – | – | – |
| Masculinity vs. femininity | – | – | – | – | −0.0085 | 0.0044 | 0.057 | 0.99 |
| Political characteristics | ||||||||
| Polity (democracy vs. authoritarianism) | 0.0023 | 0.0226 | 0.920 | 1.00 | – | – | – | – |
| pseudo–R2: 29% | pseudo R2: 47% | |||||||
| AIC:161.9 BIC:185.5 | AIC:143.8 BIC: 175.1 | |||||||
Random-effects meta-regression analysis of the crude case fatality risk at the last follow-up date in the main analysis (September 20, 2020) for 73 countries. Dependent variables were logit transformation to stabilize the variance of proportions. Random-effects meta-regression was used to explore the impact of cultural characteristics on fatalities while adjusting for important predefined covariates. The odds ratios (OR) represents the odds of a fatal outcome upon exposure to a risk factor relative to no exposure. For example, an OR of 1.03 indicates that a one unit increase the proportion of the population overweight, we expect to see a 3% increase in the odds of fatal outcome among infected individuals. Pseudo-R-squared value represent the proportion of heterogeneity explained by predictors included in the model. AIC, Akaike information criterion; BIC, Bayesian information criterion. Bold font indicates a statistically significant association with outcome at p < 0.05.