Staged social distancing: a COVID-19 mitigation strategy

Breaking: official Israeli policy is essentially staged social distancing Naftali Bennett, Minister of Defense.

My new medrxiv preprint on the math of pooled sampling based on Fisher Information Theory.

The COVID-19 pandemic is about to wreck both our health care system and our economy. The bottleneck for proper treatment is the number of ICU beds with respirators. Social distancing, to "flatten the curve" such that the number of critically ill always stays below a threshold, is unsustainable. Recently, Ferguson et al. 2020, proposed an adaptive triggering strategy, where a certain level of ICU occupancy triggers a strict social isolation period. According to their models, approximately 2/3 of the next two years would trigger the isolation policy to avoid overwhelming the British health care system. The economic cost of sustained isolation, even with an adaptive trigger, is inconceivable. Here we present an alternative approach, staged social distancing, that saves lives compared to the chaos of the "do nothing" approach that is sure to follow the present distancing policies when economic resources and people's patience run out.

A crucial observation is that the percentage of critical care cases rises steeply with age. We propose to divide society into two groups: a low-risk group of younger people without pre-existing conditions and the rest. The first group continues with their life and jobs, while the second goes into strict isolation. Accurate determination of the optimal split needs detailed and local public health models, but simple calculations based on documented COVID-19 cases and US population data and resources puts the optimal separation age around 50. Thus up to 80% of the workforce could keep the economy running without overwhelming our healthcare system. Once the epidemic ran its course in the first group, updated simulations designate another cohort to reactivate, and so forth, until retirement age.

If implemented as a public policy, staged social distancing could keep our economy running at reasonable levels and avoid the war-like healthcare collapse experienced by Italy.

The figure shows the cumulative ICU need at any time as a function of age. Horizontal lines represent availability for the UK, Hawaii, the US, the latter using 100k ventilators in the US. The two curves with shaded areas correspond to two ways of estimating with a range of reasonable parameters. The horizontal line at the age of 50 represents about 80% of the workforce.

The figure above shows what happens without social distancing. Traditional social distancing attempts to lower the whole curve until at does not intersect the horizontal line. The first stage of staged social distancing would remove the rightmost part of the curve that runs above with the horizontal line, i.e., overwhelms our ICU/ventilator resources.

I implemented the extended SIR ("SIRsir") model of Chikina and Pegden (2020). to predict the time line of the first stage of staged social distancing. A clear physical picture emerges: if we split the population into under and over 50, lock down the latter, the disease spreads exponentially in the first group until the susceptible population is exhausted. If there is some non-compliance, i.e., some interaction/infection between the groups, the internal spread in the second group will still be small. Most cases are due to interaction between the groups. Imagine two forests, a dense and a sparse one. Fire will burn through dense woods fast, and burns at the boundaries, but will not spread inside sparse areas. This lack of exponential growth in the second group means that a moderate level of non-compliance does not saturate the available resources. The caveat is that the accuracy of the prediction depends on how realistic the chosen parameters are.

I need to explore the parameter space, but the simulation above has a quite substantial leakage between the two population and still appears to stay below the ICU threshold. The parameters for this simulation are: 10 day recovery, 1/0.3 day infection rate within the young group, 1/0.11 for the old, and 1/0.2 for cross infection. The infection rate of the old is dominated by getting infected by the young group (for instance by non-compliance of policy). The split is at age 50, and I used standard US population distribution. More tests and a wider search of the parameter space is needed.

Disclaimer: The information herein is provided for informational purposes only,is just the author's opinion, and should not be misconstrued as medical or expert advice. There are many unknowns, and the author of this document makes no warranty, expressed or implied, as to the results obtained from the use of the information, and shall have no liability for the accuracy of the information and cannot be held liable for any third-party claims or losses of any damages. The information contained in this web site does not constitute a confirmed offer to sell or the solicitation of an offer to buy any product or service, and should not be relied upon in connection with any investment decision or action. Author may, at any time, revise the information on this web site without notice and makes no commitment to update this information. Any user shall have the right to view the information, reproduce it, and share it for non-commercial purposes.