Table 3.

Results of the non-linear Poisson mixed model of the number of daily COVID-19 deaths with interactions between variants proportions and vaccination effectiveness.

Variable	Estimate	Standard error	P value
Intercept	0.409	0.204	0.054
Logarithm of the number of daily COVID-19 deaths 7 days beforea	0.926	0.007	<0.001
Logarithm of the number of COVID-19 deaths 7 days before/14 days before	0.158	0.015	<0.001
Proportion of 20A (EU2) variantb	0.012	0.036	0.747
Proportion of 20E (EU1) variantb	−0.034	0.047	0.467
Proportion of 20I (B.1.1.7) variantb	0.059	0.033	0.085
Max daily temperatureb	−0.005	0.001	<0.001
Mean daily wind speedb	−0.002	0.002	0.346
Arrivals at two biggest airports [in thousands]b	0.703	0.085	<0.001
Mobility change from prepandemicb,d	0.753	0.064	<0.001
Proportion of inhabitants aged ≥65 yearsc	−0.805	0.925	0.391
Diabetes prevalencec	0.124	1.285	0.924
Cardiovascular death ratec	0.025	0.023	0.278
Vaccination effectiveness against 20I (B.1.1.7)b	0.697	0.201	0.002
Vaccination effectiveness against variants other than 20I (B.1.1.7)b	0.778	0.379	0.049
Variance for RE on intercept	0.013	0.006	0.025
Variance for RE on mobility	0.035	0.023	0.132

AIC = 30,186.

Abbreviations: AIC = Akaike's information criterion; mln = million; RE = random effects.

Non-linear Poisson mixed model of the daily number of COVID-19 deaths per 1 mln inhabitants, with country-specific random effects on intercept and mobility, including interactions between variants proportions and vaccination effectiveness. Daily data for 32 countries were included, from 29 January 2020 to 15 April 2021. Assumed time to symptoms onset is 7 days; assumed time between symptoms onset and death is 14 days.

^aTime-varying variable with a 7-days lag.

^bTime-varying variable with a 21-days lag.

^cFixed variable.

^dAverage mobility change from prepandemic level calculated across retail/recreation, transit stations, and groceries/pharmacies [0.01].