Study population and setting
This study modeled the relaxing of physical distancing measures in Austin, Texas, USA. They compared strategies which reduced community transmission only, strategies which reduced transmission to high-risk individuals only, and a combination of both. These comparisons were done for a range of assumed effectiveness of transmission reduction.
Summary of Main Findings
The authors found that by April 23, 2020, physical distancing measures had reduced transmission in Austin, Texas by 95% (95% CI: 70%-100%). If physical distancing measures were completely relaxed on May 1, 2020, hospital capacity would be exceeded in 29 days (95% CI: 19-44). If, instead, partial relaxation of physical distancing measures was able to maintain a 75% reduction in transmission, hospital surge capacity would be reached after 109 days (95% CI: 72-184). The authors then examined the effect of additional “cocooning measures” to protect those at high risk, including incentives for these individuals to stay home, increasing resources and staffing at care homes, and enabling homeless individuals to physically distance. They found that if these measures could reduce transmission to high-risk individuals by 95%, and they were combined with the previous measures reducing transmission by 75% for the general population, then hospital capacity would not be exceeded, and 64.6% of hospitalizations and 74.7% of deaths would be avoided.
The authors estimated the proportion of high-risk individuals by age, and used age-stratified contact data to model the reduction in transmission. Combining these data enables more accurate estimates of the impact of cocooning, compared to models which treat all high-risk individuals as having the same risk regardless of age.
The methods are not very well documented, making reproducibility and validation more difficult. Sources for some model parameters, such as age-specific contact rates, are not clearly given. The confidence intervals for the number of hospitalizations and deaths under each scenario only take into account two sources of uncertainty: uncertainty in the impact of physical distancing measures so far, and inherent randomness in the transmission and disease progression processes. They do not account for uncertainty in other model parameters, such as the relative risk of hospitalization, and death for high-risk versus low-risk individuals, which are not well-known. The study also assumed that high-risk conditions for COVID-19 are the same as those for influenza, and that all high-risk individuals in a given age group have the same risk regardless of underlying condition; as risk factors for COVID-19 become better understood, these results may have to be revised. Finally, the study looks at combinations of different percentage reductions in transmission for the general population and in high-risk populations, but does not comment on the relative feasibility and ethics of these combinations — for example, whether a 75% reduction for the general population and a 95% reduction in high-risk groups is more or less feasible than an 85% reduction for both, and what practical measures are needed to reach these reductions.
This study highlights the value of targeted measures to reduce transmission in high-risk populations, and thus reduce the number of hospitalized cases and deaths. It also shows that cocooning high-risk individuals alone will not avoid exceeding hospital capacity, and will lead to substantial hospitalizations and deaths.
This review was posted on: 17 June 2020