Impact of tiered restrictions on human activities and the epidemiology of the second wave of COVID-19 in Italy

Abstract

To counter the second COVID-19 wave in autumn 2020, the Italian government introduced a system of physical distancing measures organized in progressively restrictive tiers (coded as yellow, orange, and red) imposed on a regional basis according to real-time epidemiological risk assessments. We leverage the data from the Italian COVID-19 integrated surveillance system and publicly available mobility data to evaluate the impact of the three-tiered regional restriction system on human activities, SARS-CoV-2 transmissibility and hospitalization burden in Italy. The individuals' attendance to locations outside the residential settings was progressively reduced with tiers, but less than during the national lockdown against the first COVID-19 wave in the spring. The reproduction number R(t) decreased below the epidemic threshold in 85 out of 107 provinces after the introduction of the tier system, reaching average values of about 0.95-1.02 in the yellow tier, 0.80-0.93 in the orange tier and 0.74-0.83 in the red tier. We estimate that the reduced transmissibility resulted in averting about 36% of the hospitalizations between November 6 and November 25, 2020. These results are instrumental to inform public health efforts aimed at preventing future resurgence of cases.

Fig. 1. Changes in the time spent in different locations relative to pre-pandemic values over time⁸, September 25–November 25.

Colors represent three distinct intervention periods, separated by vertical darker lines: September 25–October 13 (gray, no interventions), October 14–November 5 (teal, national interventions), November 6–November 25 (yellow, orange and red, aggregating provinces by the corresponding tier). Thin lines represent values for individual provinces. Thick lines are regression lines computed over provinces with the same restrictions and are reported to help the visual identification of trends in different periods.

Fig. 2. Temporal dynamics of the net reproduction numbers R(t) and of the assigned tiers between October 30 and November 25.

Each line shows the mean R(t) for an Italian province (black) or region (blue). Provinces are grouped by region as tiers were assigned on a regional basis. Colored rectangles refer to the timeframe when the different tiers were adopted: teal = national interventions, yellow, orange and red as the corresponding tier (see Table 1 for restrictions associated with the different tiers).

Fig. 3. Estimates from LMMs.

A Net change of the reproduction number between the week October 30–November 5, 2020, and the week November 19–25, 2020, by tier group. n = 107 observations (provinces) observed over 2 time periods for model A, B,C, and E, n = 21 observations (regions) observed over 2 time periods for model D. B Relative change of the reproduction number between the week October 30–November 5, 2020, and the week November 19–25, 2020, by tier group, n = 107 observations (provinces) observed over 2 time periods for model A, B, C, and E, n = 21 observations (regions) observed over two time periods for model D. C Mean reproduction number in the week October 30–November 5, 2020, and in the week November 19–25, 2020, by tier group, n = 107 observations (provinces) in each time period for model A,B,C, and E, n = 21 observations (regions) in each time period for model D. Dots (center of the error bars) represent the mean values, vertical lines represent 95% CI. Colors (yellow, orange and red) are associated with the corresponding tier group. See Table 3 for a description of models.

Fig. 4. Incidence of hospital admissions by tier level over the period October 14–November 25, 2020.

Regions are grouped by the maximum tier assigned over the study period. Black lines represent observed hospital admissions. Teal lines represent mean projected values (shaded area: 95% projection interval) under the assumption that national restrictions were maintained after November 6, i.e., that the reproduction number was constant over the projection period. Shaded areas between the two curves highlight averted hospitalizations, the colors of each shaded area are associated with the corresponding tier.