Table 4.

Sensitivity analysis results of the GLM model using genetic covariate selection algorithm for cumulative deaths, accounting for spatial correlation (N = 38)

	Estimate	Standard Error	P value
During two months before the second wave peak
Cumulative deaths during two months before the second wave peak
Intercept	5 · 1807	0 · 5837	<·0001
Average proportion of EU2 variant	0 · 8938	0 · 3197	0 · 0086
Average proportion of B.1.1.7 variant	1 · 4900	0 · 3368	<·0001
Percentage of population aged 65 or more	0 · 04691	0 · 03145	0 · 1453
GDP per capita [1 mln USD]	−13 · 9679	4 · 9446	0 · 0080
From the second wave start to 25 February 2021
Cumulative deaths from the second wave start to 25 February 2021
Intercept	6 · 2515	0 · 5010	<·0001
Percentage of population aged 65 or more	0 · 04539	0 · 02736	0 · 1060
GDP per capita [1 mln USD]	−12 · 0909	4 · 7775	0 · 0160

GLM multivariate models with normal distribution and logit link function was used to explore factors associated with COVID-19 cumulative deaths. The number of cumulative deaths and average variants proportions were calculated during each of considered periods. Each model was run using 38 observations. Models were selected based on the use of genetic selection algorithm.