California: Roadmap for a Balanced Recovery
[8/9/20 Update: The second wave is larger than anticipated due to mitigation fatigue. Minor narrative edits and the addition of Case 5]
California case counts are on the rise. There is a modest uptick in deaths. The demographic of infections has shifted significantly to the younger cohort and deaths continue to affect the older cohorts. Mobility data increase in connectivity in the month of June, beginning in late May. This coincides with mass gatherings and a general loosening of society, in particular among the younger cohorts. With the rising infection levels, mobility data shows that behavior slightly reset in July.
Using our COVID Decision Model, we examine incremental reopening at different rates. We examine the effect of children returning to school. We assume continued protection for the vulnerable and low connectivity for those over 70. We assume connectivity modulates in proportion to risk. Connectivity is the level of potentially infectious interactions with other people. Social distancing, improved hygiene, indoor masks, telecommuting and other measures will significantly reduce the the risk of infection transmission. We adjust the connectivity factor by age to simulate these infection reduction behavior effects. Note that any sharp upward shift in connectivity will result in a peaking of deaths approximately 8 weeks later.
- Case 1: Nominal recovery case. Age proportionate risk with a gradual stepped recovery.
- Case 2: Nominal Case 1 with children back in school with full participation in all activities.
- Case 3: Nominal Case 1 with children isolated from society.
- Case 4: Similar to Case 2 with a faster short term increase in connectivity.
- Case 5: Updated to match actual mobility trends. Rapid age based recovery. The most likely scenario.
- Mass gatherings and mixed messages from public officials in June contributed to a shift in attitudes resulting in increased connectivity causing infections to spike up in July.
- Connectivity reduced slight in July as infections increased, but data shows this did NOT significantly damp the 2nd wave.
- As of August, over 15% of California residents have been infected with the virus. Ultimately 25% will be infected.
- A gradual recovery with age-proportionate risk management will allow the virus to run its course without a vaccine.
- If we follow a balanced risk approach, total deaths per million will be less than 500, far less than NY at 1666.
- Due to heterogeneous distributions of connectivity and susceptibility, the virus will die out once 25% of the population is exposed if reasonable mitigation behaviors are in place.
- Schools should be reopened, but we should make adjustments to protect vulnerable teachers.
- Returning children to school will have little impact on the total number of deaths and spread of infection.
- Our model assumes children infect others at the same rate they are infected themselves, a conservative assumption.
- Continued protection for elderly and high risk individuals with comorbidity conditions will effectively minimize deaths.
Death Rate Curves
Note that there is little difference between Case 1 (nominal) and Case 2 (full school for under 20). Completely removing the under 20 cohort from the simulation in Case 3 results in a slightly less deaths. An accelerated recovery and with a moderate connectivity pattern as shown in Case 4 results in a nominal increase in deaths. Case 5 reflects the most likely outcome after adjusting for mitigation fatigue and latest death curve data.
Dynamic Input Parameters
Case 1 through 4 assume a July mobility reset to be comparable to May. Mobility data published on the CDC show that this has not been the case. Case 5 has been calibrated to match these mobility trends in July. All cases reflect a 40% reduction in overall mortality due to improvements in care and therapy methods. Various long term recovery vectors are assumed with increased contact over time. All cases assumed continued protection for the vulnerable.
Distribution of Infections and Deaths by Age
Blue is the normalized general population age distribution. Red is the cumulative normalized percentage of overall infections. Green is the cumulative normalized percentage of deaths from COVID-19.
Infection Rate, Infection Fatality Rate and Death Rate by Age
Below are the final population infection rates (percentage of population infected), the infection fatality rates (percentage of those infected who die) and the overall population death rates (the percentage of deaths relative to population) for each age bin.
Date of Death Calibration
Extraction of actual date of death is critical to calibration of the analysis. California has no reporting of actual date of death available. The graph below shows extraction of approximate date of death from the reported death date data sets. This approach is far more accurate than a rolling average of reported deaths. The most recent week represents an incomplete count.
Late July Infection Status Distribution
The distributions in this model by age closely match the actual data found here: California COVID Dashboard
Worst Case Scenario for Reference
Below is a reference run where California returns completely to pre-outbreak conditions. This implies no protection for the vulnerable and the elderly. Note the 2 month delay from the increased contact to the peak of the death curves.
Comparing CDM Analysis to CA State Models
The worst case scenario is unrealistic; society is well aware of the risks and recognizes the need to protect the vulnerable. Many social distancing measures have become habitual. Yet, this unrealistic scenario is presented as a strong possibility on the California State website which tracks long term outlooks: California COVID Assessment Tool Below we compare our CDM scenarios against other predictive models for CA.
Worst case predictions for the outcome of the pandemic in California assume a complete letdown of protection for vulnerable people and a full scale return to life before the outbreak. This is NOT a valid assumption. Those at risk will continue to take abundant precautions and they should be reminded to do so by public health officials. Those who live in care facilities should be protected with strict safety measures. Schools can reopen with little overall effect, however at risk teachers should be afforded extra accommodations to manage their exposure. Middle-aged adults should reengage but do so with reasonable precautions to prevent the spread to older and more vulnerable populations. This measured approach will allow us to safely recover our economy and get back to normal.