Hawai’i COVID-19 - Preliminary Analysis III

Hawai’i COVID-19 Stats - Preliminary Analysis III Statistical analysis shows Hawai‘i has one of the lowest rates of positive COVID-19 test results compared with other US territories and states. Rates of testing in Hawai’i per capita are above the US average. Since March 28th, the frequency that positive test results have been returned by Hawai‘i Department of Health remains largely unchanged (less than 1% change). Testing breakdowns on the individual islands reveals Oahu and Maui have a higher rate of positive tests results than Hawai’i Island and Kaua’i. Social distancing metrics produced by Google COVID-19 Community Mobility reports a significant change in the movement of the population on Hawai’i that coincides with a very large drop in passenger arrivals to Hawai‘i. Reports from Hilo Medical Center indicate the Emergency Room is at roughly 33% of normal occupancy, and no hospitalizations related to COVID-19. Comparing Hawaii’s Testing Both the expanse of testing in the population (per capita testing) and the number of tests per confirmed case (hit-rate) allows for some understanding as to true spread of the virus from available data between different regions. Using these two metrics it is possible to compare Hawai‘i’s testing and spread of the virus across states and even some countries. [1] There is a stark contrast in COVID-19 testing in Hawai’i statistically when compared with the other States and US territories, especially those now in the midst of an outbreak. The cumulative testing data indicates that Hawai‘i’s hit-rate is currently at 2.5% after 20,525 tests as of April 15th. For comparison, the United States as a whole has a hit-rate of 24%, with 795 people on average being screened for COVID-19 per 100,000 people. In Hawai‘i there have been 1,367 tested per 100,000 people, i.e. almost twice as many. What An Outbreak Looks Like New York State has a well documented outbreak underway, and the data shows that. New York has the second highest rate of testing per capita in the nation, at 5,788 per 100,000 people. It also has the second highest hit-rate for positive test results in the states, at just over 40%. These tests can be assumed to be only a subset of a larger unknown infection. New York State has seen the hit-rate on testing rise +2.9% in 8 days, going from a hit-rate of 37.9% on April 3rd, to 40.9% on April 11th. At the same time Hawai‘i’s hit-rate remains mostly unchanged, up roughly .1%. Testing Over Time In the first Preliminary Analysis I posted on March 28th it was suggested there was a sufficient sample size in test results to begin to use the data set on COVID-19 test data statistically. At the time, the hit-rate on positive tests was ~2% with 5,800 samples. Since then, the sample size has more than tripled in Hawai‘i, and the hit-rate has changed less than one percent (.6%). See ‘Image 1’ for testing data since April 3rd, and Preliminary Analysis II posted on Hawaii Tracker. As can be seen in image 1, the states with the highest testing rate and with the the highest hit rate are not the same but in states that are rather dissimilar. After reviewing the statistics, the hit-rates on tests from Hawai‘i are among the lowest in the nation. Both the rate of testing per capita, and the rate of expansion of the testing program Hawai‘i is near average compared with 57 states and US territories. The testing rates for Hawai‘i is not accelerating on pace with other US States per capita. Full spreadsheet: https://docs.google.com/spreadsheets/d/1KL1d75JZUscUSd1bk62ECpEGWesZKiSFKMSrZYyRcd8/edit?usp=sharing ------------------------------------------------ Testing By Island Breakdown A recent breakdown on testing per island was provided by the Hawaii Department of Health shows that the amount of testing and hit-rates on each island. (See Image 5) Oahu: 3% hit-rate, 11, 794 tests total Maui: 2.6% hit-rate, 3,577 tests Hawaii: 1.6% hit-rate, 2,698 tests Kauai: 1.8%, 1,034 tests Hawai‘i's hospitalization rate. The hospitalization rate is determined by the amount of positive cases in a state, and the amount of cases that require hospitalization. This provides an additional metric to contrast states based upon the frequency that people have severe symptoms of COVID-19. Not all states that have a high hit-rate in test results have high hospitalization, and not all states provide statistics on hospitalizations. ------------------------------------------------ Social Distancing Metrics As would be obvious to many, there is now strong evidence that social distancing results in decreased COVID-19 transmission rates, as multiple simulations that used SIR models suggested [2][3]. Google setup metrics to track mobility of people in each region worldwide. The nationwide average for the United States has the following breakdown, as defined by Google COVID-19 Community Mobility Report [4]. Retail & recreation -56%, compared with baseline. Grocery & pharmacy -36% Parks -65% Transit stations -72% Workplaces -45% Residential +16% Starting March 15th, Hawai‘i began to see relatively large changes in mobility when compared with other US states. Retail & recreation -56%, compared with baseline. Grocery & pharmacy -36% Parks -65% Transit stations -72% Workplaces -45% Residential +16% Updated data from Google shows Hawai‘i has had only a small rise Grocery & pharmacy usage (+7%) between March 28th and April 11th. Retail & recreation -55% Grocery & pharmacy -29% Parks -66% Transit stations -74% Workplaces -47% Residential +18% Oahu’s significant population heavy skews the data for the state, neighboring islands have seen variances from the statewide trends. (See Image 2.) One thing to consider when dealing with this data set from Google Mobility; the impact from Hawai‘i’s changes in tourism on the metric. Over the course of March passenger arrivals to Hawai‘i have dropped by ~99%, with most of the changes in travel happening in mid-March at roughly the same time frame. In the first half of April, Hawai‘i State has seen an average of 409 visitors a day. [5] (See Image 3) Findings Mobility and traffic volume metrics are likely skewed by large decreases in travel which happened roughly over the course of the same time span, and cannot be easily separated. Things are looking pretty good here in Hawai’i all things considered. There are many states that have more cases of COVID-19, and reasons to suspect a significantly larger outbreak than the officially documented. Hawai’i still has active case clusters that are within the transmission period of COVID-19. The five day trend on cases over in the last month shows a declining trendline since early April. (See Image 4) Commentary It can be expected that the true expanse of SARS-CoV-2 infection is more widespread than testing suggests, in some areas the outbreak could be over a magnitude larger than test positive cases of COVID-19. Widespread testing is still not available in Hawai’i or the United States as a whole, based upon data compiled by The COVID Tracking Project. [6] Mahalo to Maren Purves, Philip Ong, and the rest of the Hawaii Tracker team for assistance on this preliminary analysis in research and editing. Limitations While Hawaii Tracker will work with the numbers available, there are inherent flaws in data collection and distribution related with an emerging outbreak. The arguments presented are the best I can make with the data at hand, and should be looked at through a generalized lens due to some anomalies in reporting between states. Secondly, the Google Mobility Reports are based upon devices that have Location History enabled, which by default it is not. This reduces the sample size of devices polled for information. Third, delays in test results means there is some lag on results. Sources [1] Our World In Data, Joe Hasell, Testing data provides us with two indicators of the quality of data on COVID-19, https://ourworldindata.org/covid-testing [2] 3Blue1Brown, Simulating an Epidemic, https://youtu.be/gxAaO2rsdIs [3] MedRxiv, Oscar Patterson-Lomba, Optimal timing for social distancing during an epidemic, medrxiv.org/content/10.1101/2020.03.30.20048132v2 [4] Google, COVID-19 Community Mobility Reports, https://www.gstatic.com/covid19/mobility/2020-04-11_US_Hawaii_Mobility_Report_en.pdf [5] Department of Business, Economic Development & Tourism, PASSENGER COUNT UPDATE, April 16th. http://dbedt.hawaii.gov/visitor/daily-passenger-counts/ [6] The COVID Tracker Project, https://covidtracking.com/data