The initial lockdown in Hawaii was severe and occurred before COVID-19 gained any significant steam. This lockdown was effective but certainly inflicted significant economic and psychological pain. As mobility increased and activity increased through the summer, infections and death rates are now increasing. The majority of the cases are within the island of Oahu, largely within the greater Honolulu area. We limit our analysis to this island and scale Covid Tracking Project data by the ratio of deaths between HI and Oahu (~80%).
Since the disease penetration in Oahu, Hawaii was so low going into the summer (~1%), the summer surge is essentially a first run outbreak in an uninfected population already educated and predisposed to manage risk by age and vulnerability. Advances in treatment have also greatly lowered death rates. The overall death for HI will be well below the average for most states, likely in the range of 250 deaths per million.
Given the low initial R0 of 1.33 for Hawaii as calculated by the rt.live HI, a homogeneous herd immunity threshold (HIT) of 25% would be predicted using standard equations. Allowing for heterogeneous characteristic distributions, the true HIT threshold for HI will likely range between 10% to 20%, depending on behavior and risk management of vulnerable populations. This is consistent with the case study scenarios presented below. For more information see our post: Are We Closer to Herd Immunity than Most Experts Say?
As a practical matter, it is very difficult to completely lock down an entire state indefinitely. Mitigation fatigue sets in and eventually younger cohorts start to return to normal activities. Elderly and vulnerable people will be more cautious as they protect their self interests, risk appropriate behavior modification occurs. We see this in Hawaii. If health and civil authorities balance economic and psychological stresses with common sense strategies to protect the vulnerable and elderly from COVID-19, Hawaii can get back to normal in a balanced manner consistent with recommendations in our article: Rational Policy Strategies for the COVID-19 Pandemic.
Hawaii Surge Case Studies
We examine four surge models which bound the range of outcomes for the island of Oahu. Case 1 is an unlikely lower bound. Case 2 and 3 are likely outcomes. Case 4 is an upper bound worst case.
- Case 1: Active June, slower July, September return to May activity levels.
- Case 2: Active June, slower July, September return to June activity levels. .
- Case 3: Active June, slower July, September return to June levels for elderly. Schools open, active 20-50 group.
- Case 4: Active June extended flat into the future.
- Oahu, Hawaii is going through a first significant COVID-19 surge, but will fare better than most states.
- Increased mobility in Hawaii generally tracks trends in the mainland.
- Our analysis assumes no external introduction of the virus after the lockdown.
- There has been a slight reduction in mobility in July/August, but a return to April lockdown levels is unlikely.
- Over 5% of Oahu residents have likely been infected by mid August.
- Infection rates are higher among individuals in the 20-40 age range, ~10% in late August.
- Total deaths are ultimately be ~250/million (compare to NYC ~2,800/million, Sweden ~750/million)
- Death rates continue are strongly biased toward elderly populations.
- The infection fatality rate (IFR) is relatively low, in the range of 0.15% and will rise if elderly become active too early.
- Returning children to school and extracurricular activities will not cause a significant increase in deaths or infections.
- Risk proportional mobility lowers herd immunity threshold; reached when 25 to 35% of 20-40 year olds are infected for HI.
- Vulnerable and elderly populations are at risk and should continue to take precautions and avoid high risk situations.
- The summer surge will die out through the fall, we will not see a returning infection peak if elderly and vulnerable are protected.
- Improvements in therapeutic methods of treatment have significantly reduced the mortality rate, we assume a 40% reduction since March.
Case 1 reflects the fewest number of deaths but is an unlikely scenario given lockdown fatigue effects. Case 2 and 3 are more realistic scenarios and have slight variations in mobility in younger and middle age adults. Case 4 is a worst-case extension of peak activity in July and not likely. Until the infection has damped down, elderly and vulnerable people should continue to avoid risky situations and practice social distancing. In all cases the infection dies out by winter. Our COVID Decision Model (CDM) results are compared to the current IHME predictions for reference (only through November); IHME predicts less deaths in the short term, but a potential exponential increase later in the year.
Trends from COVID Tracking Project Data processed by JHU show a late August surge in the percentage of positive tests. Note that tests and collection of results reflect a delay of 1 to 2 weeks from the actual date of infection. This is consistent with our results.
Hawaii Testing Trends: John Hopkins
Our analysis shows infections peaking in late August and into early September for Case 1 and 2. Case 3 and 4 peak later representing more worst case scenarios. The percentage of the population currently infected shows a peak a couple weeks after the peak of new infections. With the delay from infection to test, and from test to results, we would expect to see the peak of positive tests to be in late September into early October. The cumulative percent of population infected ultimately will be in the range of 10% to 20%. As the average age of infection shifts to the lower mortality rate cohorts, we see the infection fatality rate in the range 0.1 to 0.2%. The reproductive factor peaked at 1.4 and currently is in the range of 1.2 to 1.0 and will soon be under 1.0.
Date of Death Extraction
Determination of actual date of death is critical to calibration of the analysis. The graph below shows the derived date of death the reported death date data sets found at the COVID Tracking Project. Our approach is far more accurate than a rolling average of reported deaths and is detailed here: Reported versus Actual Date of Death. The most recent 2 weeks represent incomplete counts.
Mobility trend data for Hawaii from the IHME indicates a minor slowdown in July following a June peak. Mobility data does not include the effects of enhanced personal hygiene or indoor social distancing, but does indicate a general trend. We stratify these trends by age group in our analysis.
A simulation population of 1 million discrete agents is scaled by a factor of 0.953 to represent the 0.953 million population of Oahu Hawaii. We assume 50% of those infected are asymptomatic.