Association of tiered restrictions and a second lockdown with COVID-19 deaths and hospital admissions in England: a modelling study

Abstract

Background: A second wave of COVID-19 cases in autumn, 2020, in England led to localised, tiered restrictions (so-called alert levels) and, subsequently, a second national lockdown. We examined the impact of these tiered restrictions, and alternatives for lockdown stringency, timing, and duration, on severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) transmission and hospital admissions and deaths from COVID-19.

Methods: We fit an age-structured mathematical model of SARS-CoV-2 transmission to data on hospital admissions and hospital bed occupancy (ISARIC4C/COVID-19 Clinical Information Network, National Health Service [NHS] England), seroprevalence (Office for National Statistics, UK Biobank, REACT-2 study), virology (REACT-1 study), and deaths (Public Health England) across the seven NHS England regions from March 1, to Oct 13, 2020. We analysed mobility (Google Community Mobility) and social contact (CoMix study) data to estimate the effect of tiered restrictions implemented in England, and of lockdowns implemented in Northern Ireland and Wales, in October, 2020, and projected epidemiological scenarios for England up to March 31, 2021.

Findings: We estimated a reduction in the effective reproduction number (R_t) of 2% (95% credible interval [CrI] 0-4) for tier 2, 10% (6-14) for tier 3, 35% (30-41) for a Northern Ireland-stringency lockdown with schools closed, and 44% (37-49) for a Wales-stringency lockdown with schools closed. From Oct 1, 2020, to March 31, 2021, a projected COVID-19 epidemic without tiered restrictions or lockdown results in 280 000 (95% projection interval 274 000-287 000) hospital admissions and 58 500 (55 800-61 100) deaths. Tiered restrictions would reduce hospital admissions to 238 000 (231 000-245 000) and deaths to 48 600 (46 400-50 700). From Nov 5, 2020, a 4-week Wales-type lockdown with schools remaining open-similar to the lockdown measures announced in England in November, 2020-was projected to further reduce hospital admissions to 186 000 (179 000-193 000) and deaths to 36 800 (34 900-38 800). Closing schools was projected to further reduce hospital admissions to 157 000 (152 000-163 000) and deaths to 30 300 (29 000-31 900). A projected lockdown of greater than 4 weeks would reduce deaths but would bring diminishing returns in reducing peak pressure on hospital services. An earlier lockdown would have reduced deaths and hospitalisations in the short term, but would lead to a faster resurgence in cases after January, 2021. In a post-hoc analysis, we estimated that the second lockdown in England (Nov 5-Dec 2) reduced R_t by 22% (95% CrI 15-29), rather than the 32% (25-39) reduction estimated for a Wales-stringency lockdown with schools open.

Interpretation: Lockdown measures outperform less stringent restrictions in reducing cumulative deaths. We projected that the lockdown policy announced to commence in England on Nov 5, with a similar stringency to the lockdown adopted in Wales, would reduce pressure on the health service and would be well timed to suppress deaths over the winter period, while allowing schools to remain open. Following completion of the analysis, we analysed new data from November, 2020, and found that despite similarities in policy, the second lockdown in England had a smaller impact on behaviour than did the second lockdown in Wales, resulting in more deaths and hospitalisations than we originally projected when focusing on a Wales-stringency scenario for the lockdown.

Funding: Horizon 2020, UK Medical Research Council, and the National Institute for Health Research.

Figure 1

Model fit to NHS England region-specific data from March 1, to Oct 14, 2020 NHS England region-specific data on the number of COVID-19 deaths, hospital admissions, and occupied hospital and ICU beds; proportion of residents PCR positive and seropositive; and the regional attack rate (proportion ever infected). Black lines show reported data. Coloured lines and shaded areas show medians and 95% credible intervals from the fitted model. The crosses on the PCR positivity and seropositivity graphs show the time period over which the data were collected (horizontal lines) and the 95% CIs associated with the data (vertical lines). ICU=intensive care unit. NHS=National Health Service.

Figure 2

Projected impact of a Wales-type lockdown in NHS England regions R_t, the daily number of deaths and hospital admissions, and the daily number of occupied hospital and ICU beds are compared across seven NHS England regions for three different scenarios: tiered restrictions only and a Wales-type lockdown with and without schools open. Lockdowns extend from Nov 5, to Dec 2, 2020 (indicated by grey shading on the graphs). Lines and green, red, or blue shaded ribbons indicate median and 95% projection intervals, respectively. Step changes in R_t show the introduction or relaxation of tiered restrictions and lockdown measures. ICU=intensive care unit. NHS=National Health Service. R_t=effective reproduction number.

Figure 3

Projected impact of altering the type of intervention or duration or timing of lockdown on cumulative deaths, pressure on hospitals, and time spent living under restrictions (A) Impact of altering the type of intervention. Baseline refers to a counterfactual scenario with no tiered restrictions and no lockdown. Note that the lines for NI with and without schools closed overlap in the top graph. (B) Impact of altering the duration of lockdown. Graphs show the effects of introducing different lengths of a Wales-type lockdown (with schools open) in England on Nov 5, assuming tiered restrictions are already in place. (C) Impact of altering the timing of lockdown. Graphs show the effects of varying the time a 4-week, Wales-type lockdown (with schools open) is introduced in England, starting from up to 4 weeks before to 2 weeks after Nov 5. Hospital pressure was defined as the population-weighted mean number of weeks that an National Health Service region's hospital or ICU bed occupancy exceeded 50% of the peak occupancy for that region during the first wave of COVID-19 in England. All graphs show medians, with shaded regions defining 95% projection intervals. ICU=intensive care unit. NI=Northern Ireland.