My weekly chart showing the positivity rate for Washington State is a tad late. The Washington Department of Health (DOH) has been busy fixing and upgrading its web site. The new site is a quite an improvement with informative charts, graphs and tables.

Looking at their data, I can appreciate the depth of their collection and reporting problems. Their old database was a “peacetime” tool, collecting positive results from a few of Washington’s many counties. In the fight against COVID-19, all counties are reporting at once and epidemiologists and decision makers want and need additional data about the number of tests, sex, age, etc. Of course, doing all of this work in the middle of a crisis must raise the IT stress level!

The state DOH site now reports detailed testing information in the form of a bar chart. The bar chart shows the number of positive and negative tests each day based on specimen date. This is a meaningful time base as it allows a snapshot of the positivity rate on a given day. It’s much harder to organize test data in this way than just lumping test results willy-nilly as they come in. (See “stress level” above.)

Because all test results for a given specimen day do not arrive at once, the DOH doesn’t guarantee complete results for the days just past. In keeping with their approach, my graph shows the Washington State positivity rate through April 8, 2020. [Click to enlarge graph.]

The true peak in the rate occurred on March 28-29, 2020 when it hit 14.49%. The positivity rate has been slowly declining since then. The peak rate was more than double the 5% to 6% rate established around March 10. The new chart agrees reasonably well with raw statistics from the University of Washington Virology Laboratory. UW results, of course, are large portion of the State’s results. A back-of-the-envelope estimate shows that the rate has improved after April 8. Stay tuned.

So, what about those high rates in early March?

The positivity rate is a rather poor indicator which is subject to bias. COVID-19 tests are only administered under certain specific conditions because testing is rationed. Criteria include COVID-19 symptoms (fever, cough, tiredness, difficulty breathing) or contact with a known infected person. Properly, first responders may be given preference since they are most likely to come into contact with infectious people who may or may not be symptomatic. Further, COVID-19 hits nursing homes with vulnerable residents early and hard.

The general public has not been tested in a systematic epidemiological study, i.e., a random sampling of the population at large. Thus, the positivity rate as measured here is imperfect and may not accurately represent what the disease itself is doing in the overall community.

Contact is an important biasing factor and it is why I do not consider results in early March to be baseline. Public health authorities were actively pursuing contacts in early March after the first case of community spread was detected. The number of daily tests ranged from 74 to 500 — small numbers which are easily skewed.

As I’ve mentioned in earlier posts, today’s statistics tell us what the disease was doing 10 to 14 days ago due to the time required for incubation, development of symptoms, clinical presentation, and testing. Washington’s first community mitigation measures went into effect on March 11, with home-stay recommended on March 16 and home-stay ordered on March 16. I think we can safely say “Community mitigation works.”

Now we need the patience to finish the job and win the fight.

Stay distant and stay healthy — P.J. Drongowski

Another week has past and it’s time for new charts. I’ve been tracking the daily positivity rate for Washington State. The “positivity rate” is the percentage of positive test results for each 24 hour period. Dividing by the number of tests each day adjusts for variations in daily testing throughput, i.e., the number of tests performed each day is different.

First, here is the chart using test results from the Washington State Department of Health (DOH). [Click to enlarge.]

The gaps in the data are due to database issues at DOH. DOH receives and integrates test reports from around the state. Their database has been stressed and has not been tallying negative test results from which one can compute the number of tests per day (i.e., the sum of positive and negative results). Thus, I wasn’t able to compute a daily positivity rate for certain days.

Fortunately, I have the University of Washington Virology Laboratory as back-up. The UW Virology Lab is performing a large portion of tests in Washington. The Lab maintains a dashboard showing the number of tests they performed, the positives, negatives and inconclusives. Inconclusive tests are presumed positive. So, the daily positivity rate is the sum of the positive and inconclusive tests divided by the number of tests performed that day.

Both graphs have a similar shape which shouldn’t be surprising as the UW Virology Lab is a major component in the DOH results. The positivity rate has more than doubled since early March. As of today (April 6), the positivity rate appears to be at a plateau and may show the beginning of a downturn. That’s good news and we need to remain committed to social distancing and other community mitigation measures.

Washington State recommended home-stay on March 16 and closed some businesses (e.g., hair salons, gyms, churches, etc.) I regard these measures as “social distancing lite.” On March 24, the State imposed a stay-at-home order. Given the time lag (10 to 14 days), the turn in the positivity rate is most likely due to the full stay-at-home order.

Fortunately, Governor Inslee has not declared victory. Instead, social distancing measures have been extended to May 4. Further, school has been cancelled through the end of the Spring term.

From my own observations, people in our part of Snohomish County (near Bothell and Everett) are treating COVID-19 seriously and are complying with the stay-at-home order. If a state, county or city is waiting for signs of an outbreak, don’t wait to issue a stay-at-home order. Response now will keep the size of your problem manageable. An ounce of prevention is worth a pound of cure.

Cleveland, OH

I keep an eye on the old hometown for no other reason than the Cavs and Browns. Cleveland.com published a few useful looking charts and graphs.

As we all go forward, we need to know when our communities have turned the corner and, importantly, when we can restart some business activities. The Ohio Department Health tracks the percentage increase in confirmed cases day-over-day.

Like Washington, Ohio has not hesitated to issue a stay-at-home order and to implement other forms of community mitigation. Gov. Mike DeWine, like Gov. Inslee, may not be the most dynamic individual, but he, too, is leading his state effectively in this crisis. I’m pulling for the folks at home!

This is another Ohio DOH chart which caught my eye. It plots confirmed cases by the onset of symptoms. This chart better aligns case count with the actual time frame in which a patient was infected.

NYT: What’s next?

The New York Times published two good articles on community mitigation and the next phase in our battle against SAR-CoV-2 (COVID-19) — cycles of “suppress and lift” as restrictions are applied and relaxed, so-called waves.

• Lockdown Can’t Last Forever. Here’s How to Lift It.
• How Will We Know When It’s Time to Reopen the Nation?

These articles are flying by so fast and so thick that you’ll miss them if you blink.

Stay 2 meters apart and stay healthy — P.J. Drongowski

Time for what’s known in the journalism biz as a “round-up.”

I started blogging about COVID-19 numeracy (AKA “by the numbers”) as a way to get my head around all of the data thrown at us by media outlets. Blogging also gave me a way to have the illusion of control over that which I cannot control, namely, the spread of COVID-19.

I hope that my blog posts are a resource for math and science teachers. (I try to keep these posts free of politics, BTW.) If there is a silver lining in all of this, what an opportune time to teach critical thinking and numeracy! I tried to collect the best and most useful examples of data presentation and, I collected a few stinkers, too.

So, here they are, links to my blog posts about COVID-19 numeracy:

• Washington State, April 6, 2020 (4/7/2020)
• Seattle Scene (3/30/2020)
• Epidemiologic Testing (3/27/2020)
• NYT models and cases (3/26/2020)
• Geography including facts about Italy (3/25/2020)
• New York City positivity rate (3/24/2020)
• Washington State positivity rate (3/23/2020)
• Washington and Seattle useful charts (3/21/2020)
• Affect of time on analysis (3/20/2020)
• Testing and bias (3/19/2020)
• Comment on NYT curve flattening by country (3/19/2020)
• COVID-19 key performance indicator: Acute cases KPI (3/18/2020)

If you’re a regular reader, you know that I usually deep-dive music technology and electronics. Fear not, I have been plugging away on music projects, too: Blokas MidiBoy, Toontrack EZKeys, recording demos with Genos, and more. All the projects for which I didn’t have time. Guess I’ve got the time now, just like everyone else. 🙂

I hope to blog about music-related topics as I’m not sure how much value I can add to the on-going COVID-19 discussion. I will continue to track the positivity rate for Washington State since this is my home with family and friends nearby. Plus, there are always subjects (e.g., reverse transcription polymerase chain reaction/RT-PCR testing) that pique my scientific interest.

Washington Post: Bending the curve

Here’s a quick hit from the Washington Post. It’s a set of graphs illustrating the effect of social distancing for select U.S. cities.

With respect to Washington State (Seattle), it’s still early days, really. I think my original guidance holds — we won’t really see a bend from another 7 to 14 days out from this point (April 2). Washington State imposed its first social distancing on March 16 followed by a stay-at-home order on March 23. With the delay due to SARS-CoV-2 incubation, development of symptoms, clinical presentation and testing (10 to 14 days according to medical experts), we should just now be seeing a bend in the curve.

The Washington State Department of Health (WADOH) has had technical issues with its disease reporting database. The original database could not handle the incoming stream of reports including negative test reports. So, the WADOH database went dark for a few days recently and is just now coming back. Due to these technical glitches, we need a little extra time to see the effect of social distancing in Washington.

Stay apart and keep all of us healthy — P.J. Drongowski

A New York Times article (March 29) suggests that social distancing is beginning to pay off in the Seattle area. The NYT article refers to work performed at the Institute for Disease Modeling in Bellevue, WA. The authors are cautious saying that data is preliminary and unpublished.

Quoting the IDM web site:

IDM shapes global efforts to eradicate infectious diseases and to achieve permanent improvements in the health of those most in need. By developing, using, and freely sharing computational modeling tools, we advise policymakers, promote quantitative decision-making and advance scientific methodologies.

IDM is an institute within the Global Good Fund, a collaboration between Intellectual Ventures and Bill and Melinda Gates.

Currently, IDM is working on disease transmission dynamics for malaria, polio, tuberculosis, HIV, pneumonia, typhoid, and more.

Personal thanks to Bill and Melinda Gates for their philanthropic work.

IDM has a COVID-19 InfoHub. For the latest results, I recommend the IDM InfoHub Research and Reports page. This is the real deal.

IDM published a report (as of March 29) on the effects of social distancing and mobility reductions on COVID-19 in King County, WA. Quoting the report:

Still, based on this data for King County, the trend clearly shows R_e decreasing from about 2.7 in late February to roughly 1.4 on March 18th. These estimates come with high uncertainty, and while the trend is encouraging, it remains unlikely that COVID transmission in King County was below the 1.0 threshold on the 18th.

[We] cannot say with confidence if R_e is below or above the critical threshold of 1 today.

R_e is the effective reproductive number for COVID-19.

Thank goodness, Governor Inslee is not declaring victory either. Community mitigation measures remain in place and likely will be extended. (The federal guideline is currently April 30.) In his interview on CNN’s “State of the Union” on Sunday, the Governor cited shortfalls in testing supplies (swabs, vials, etc.) which are limiting the number of tests performed daily. As I’ve said before, we need community-wide epidemiological studies to get ahead of the disease. We cannot perform epidemiological tests (surveillance testing) when we can barely perform diagnostic testing for symptomatic patients.

The Fine Print

Let’s get nerdy. How do IDM define R₀ and R_e. Quoting the IDM glossary:

R₀ (R Naught, Basic reproductive number): The average number of secondary cases a typical infectious individual will generate without any interventions in place to reduce transmission. The number of secondary cases per source is expected to change over time and space with different interventions and as the proportion of the population with immunity grows (R_Eff).

R_Eff (R Effective) or R_e: The number of secondary cases each infectious individual will generate, on average, after interventions are in place and/or part of the population is immune. We would expect R_Eff to be less than R₀.

Washington Department of Health

The Washington State Department of Health upgraded COVID-19 reporting on their web site. They are publishing better graphs and data, especially data about testing. I’d like to see graphs for the posivity rate over time since the ratio of daily positive tests divided by the daily number of tests adjusts for day-to-day variations in testing. (“Daily” means for each 24 hour period.)

Positivity Rate

Governor Inslee referred to this ratio as the “Positivity Rate” and I like that term. It’s more concise than “Percentage of positive tests per day.”

Below is my plot of the Washington State positivity rate starting March 12 to March 29. Although the press is optimistic, The positivity rate has doubled over the time period. [Click image to enlarge.]

The gap in the graph is due to missing data from the Washington State Department of Health. They ran into database glitches March 24 and 25. The March 26 data point aggregates positive cases and tests across the gap.

There are several factors that affect the positivity rate. First off, this is not a random sample of the population at large. People are tested only if they satisfy a protocol, that is, satisfy a set of criteria. Criteria include presence of COVID-19 symptoms (fever, cough, shortness of breath), contact with a known infected COVID-19 patient, or in-coming travel from a known COVID-19 hot-spot (e.g., China, Italy, New York City metro, etc.) No doubt, some of the tests are due to contact tracing as officials try to identify possible carriers in order to isolate them. Bottom line, the positivity rate is far from a controlled epidemiological study (surveillance).

Please remember that any measurements we take today show what the virus (SARS-CoV-2) was doing 10 to 14 days ago. The delay is due to incubation, development of symptoms, clinical presentation and testing.

At this point in time (March 30), I’m glad that Washington State is staying the course and is looking at additional restrictions.

BTW, I’m still trying to absorb the changes to the Washington State Department of Health site. Their graphs don’t always jibe with my table of previously published data.

Cautionary tale

Here is a cautionary tale from The Seattle Times.

At the beginning of March, The Skagit Valley Chorale discussed weekly rehearsals and whether they should go ahead. Skagit County had not yet reported any cases of COVID-19. On March 6, they decided to go ahead with rehearsal on March 10. The chorale members used hand sanitizer and minimized contact during rehearsal.

None of those precautions, however, prevented an outbreak. Three weeks later, 45 members have been diagnosed with COVID-19 or have had symptoms. At least three members were hospitalized and two are dead.

Fortunately, our own church group passed on rehearsal during the same time frame.

This tale should be a warning to any community that feels safe because they have not yet experienced the disease. Please pay attention to our public health officials who predict eventual outbreak everywhere. Start social distancing and isolation now.

    Car crash fatalities (2019)    38,800 people 
    Vietnam War fatal casualties   58,220 people 
    Hiroshima fatal casualties     90,000 to 146,000 people 
    Nagasaki fatal casualties      39,000 to 80,000 people 
    COVID-19                       ? 

— P.J. Drongowski

Epidemiologic testing

Yesterday, CNN carried interviews with Dr. Anthony Fauci and Bill Gates. Both gentlemen called for community-wide epidemiological studies of COVID-19. We need to know the prevelance of the disease in our communities in order to assess the effectiveness of community mitigation (i.e., social distancing).

Broadly speaking, there are two kinds of COVID-19 testing:

1. Diagnostic testing
2. Epidemiologic testing

Diagnostic testing is patient-focused. It determines if the tested individual is infected with SARS-CoV-2 (the virus which causes disease COVID-19) in order to isolate and treat the patient. Due to limited test capacity, all testing to date is diagnostic in purpose.

There are variants of diagnostic testing on the horizon such as a test to determine if an individual has acquired immunity to COVID-19 or not (antibody testing).

Epidemiologic testing is like a political poll. People are selected at random from the community, statistically called the “population.” The selected people are tested for SARS-CoV-2 infection using the same kind of medical test employed in diagnostic testing. Personal information is also collected such as age, sex, presence of underlying medical conditions, previous COVID-19 diagnosis, and so forth. Statistics are tallied and summarized such as the percentage infected, susceptible (not yet infected) and immunity. Like a political poll, the statistics are broken out by age, sex and so forth.

We have not yet performed a single community-wide epidemiological study of COVID-19. This leaves us blind to one important segment in the community — people in the community who are infected and are asymptomatic. We know that asymptomatic people are a significant factor in the spread of SARS-CoV-2 and COVID-19.

Impatient people are already chomping at the bit to restart the American economy. Given that we are only 2 weeks (at best) into community mitigation, this is preposterous. [Some communities have not even started social distancing and they are late.] At some point, though, we must restart the economy. Epidemiological studies will help us to decide if it is safe to reduce community mitigation and allow the free flow of commerce again.

One important factor in the decision to reduce mitigration is the degree of immunity in the community, the so-called “herd immunity.” The community (the herd) becomes protected from a disease when a large portion of the population (typically, 60%) is immune to infection. Infected individuals often become immune when they successfully fight off and recover from the disease (acquired immunity). Better yet, individuals can be made immune through vaccination.

Acquired COVID-19 immunity has not been confirmed by scientific study at this time. Based on prior experience with other coronaviruses, medical experts believe that recovered individuals will be immune to SARS-CoV-2.

“Flattening the curve” has two benefits. First, it reduces the strain on the health care system (hospitals, doctors, nurses and other caregivers). Second, it buys us the necessary time to develop a safe vaccine and to get a vaccination program ramped up and put in place. In some sense, developing the vaccine is the easy part. The time-consuming part is the medical trials and studies needed to assume safety and efficacy.

The Seattle Flu Study

During the Bill Gates interview, he mentioned The Seattle Flu Study.

The Seattle Flu Study was initially started to detect, monitor and control influenza outbreaks in Seattle using rapid virus genetic sequencing and mapping chains of transmission. The study is led by the Brotman Baty Institute, in collaboration with UW Medicine, The Fred Hutchinson Cancer Research Center, and Seattle Children’s.

With the outbreak of COVID-19, the Seattle Flu Study was “repurposed” to COVID-19. Quoting:

We are partnering with Public Health — Seattle & King County to launch the greater Seattle Coronavirus Assessment Network — or SCAN for short.

To track the spread of COVID-19, SCAN will collect nasal swabs from a sample of people across Seattle and King County, working to mirror the area’s population as closely as possible. We’ll collect swabs from both those who are healthy and those who feel sick and will also test de-identified clinical residual samples (these are left-over samples from tests performed for other reasons at clinical laboratories). The results of these tests will help us understand the outbreak more completely and, along with other data sources, help inform public health decisions. SCAN is not able to offer testing to every individual but the whole community will benefit from the program.

I look forward to the results of the SCAN study since we live a stone’s throw away from King County, Washington. This is our ‘hood.

Logarithmic graphs

Logarithmic graphs are good tools when analyzing phenomena with exponential growth. At this point, we’ve all seen linear scale graphs showing the steep rise of COVID-19 cases. Each tick in a linear scale graph has the same statistical weight, e.g., 100, 200, 300, 400, etc.

A logarithmic graph has a logarithmic scale. Each tick increases by a power of 10, e.g., 100, 1,000, 10,000, 100,000, etc. An exponential trend appears as a straight line in a logarithmic graph. Here is a logarithmic graph (from The Guardian, March 27).

Sadly, the United States is on an exponential trajectory. China and South Korea, in particular, have asymptotic curves, showing successul COVID-19 community mitigation. Those two countries are “flattening the curve.”

New York State Hospitalizations

Gov. Andrew Cuomo of New York presented an important and useful graph during his briefing this morning.

This chart shows the number of hospitalizations per day. Please note that the number of hospitalizations is increasing by almost 1,200 new patients per day. No health care system can sustain this kind of growth. New York State expects the apex of hospitalizations in 21 days, adding a further 25,200 hospitalized patients, if the current rate continues.

As I mentioned earlier today, The New York Times staff has published some excellent maps, charts and interactive simulations.

Their latest interactive simulation illustrates the effect of community mitigation (i.e., social distancing) for varying periods of time. I took snapshots at weekly intervals (social distancing for 7 days, 14 days, 21 days, etc.) [Click to enlarge.]

The curve in the lightest shade is the number of infected people. The middle shade curve is the number of people hospitalized and the darkest shade curve is the number of deaths. Each tick in the vertical scale is 10 million. Yes, millions. I recommend reading the article itself for assumptions and caveats on the mathematical model.

From this time series, it’s easy to see that two weeks (14 days) of community mitigation is wholly inadequate for suppressing the disease, let alone stopping it. To my eye, the shortest reasonable duration is 42 days (6 weeks).

Here is the latest NYT map (March 26, 2020) depicting the spread of COVID-19 across the United States. [Click to enlarge.]

Dedicated to Edward Tufte, who is still kicking around, thank goodness.

Updated March 26, 2020.

State-by-state

Governor Cuomo, in his briefing today, asked why New York State is leading the United States in positive COVID-19 cases. Let’s compare the top four states:

                     Total   Total   Percent    
    State            Tests   Cases  Positive  Date 
    -------------   ------  ------  --------  ---- 
    New York        38,390  14,904     38.8%  3/24 
    Massachusetts   13,749   1,159      8.4%  3/24 
    California      27,650   2,102      7.6%  3/24 
    Washington      33,933   2,221      6.5%  3/23 

Indeed, New York State has the highest percentage of positive tests — substantially higher. There must be a factor in play. Could there be a difference in testing (i.e., difference in the test kits? different test protocol?) Did SARS-CoV-2 hit New York State sooner? Is COVID-19 more prevalent due to the population density in the New York City region, as the Governor speculated?

Only a good epidemiological study could really measure the prevalence of SAR-CoV-2 in the community. As to timing, I don’t think we will ever know when SARS-CoV-2 hit our shores. New York City is a major global transportation hub. Even so, Washington and California have strong business and family ties across the Pacific and Pacific rim and have major transportation hubs (Los Angeles, San Francisco, San Diego, Seattle).

If I get a chance today, I’m going to look into our neighbor to the north, Vancouver. Vancouver is also quite connected with Asia and the Pacific rim.

The Washington Department of Health has had difficulties with its COVID-19 database. Although they posted a new number of positive cases, 2,469 as of March 24, the number of negative cases was not updated, showing the same total negative test count as of March 23.

The national picture

The New York Times has published many informative graphs and simulations during the COVID-19 crisis. The chart below shows the spread of SARS-CoV-2 (the virus) and COVID-19 (the disease caused by SARS-CoV-2) across the United States as of March 19, 2020.

The chart illustrates how population density affects the number of COVID-19 cases. [Click to enlarge.] The biggest circles coincide with the major urban centers in the United States. On the west coast are Seattle-Tacoma-Washington, the San Francisco Bay area, and Los Angeles, for example.

The chart demonstrates why it isn’t appropriate to think numerically about the country as a whole. Sure, Americans want to know how the U.S. is doing overall. However, when it comes to actual COVID-19 containment and mitigation, it’s a local, regional problem. That’s why our county and municipal health departments are so important and local statistics are the most meaningful and practical.

All is not local, however. The urban centers are connected by highways and air routes, giving SARS-CoV-2 passage. As the disease plays out, the coastal urban centers with the greatest population and density will likely succumb first, followed by so-called second tier cities. When the disease subsides in major urban centers, the second tier will maintain active reservoirs of SARS-CoV-2. If major urban centers drop community mitigation (e.g., social distancing) too soon, the active pools in the second tier will cause COVID-19 flare-ups in any major urban center that drops its mitigration too early.

Major urban centers are more tightly connected to global business and commerce than the second tier. SARS-CoV-2 likely arrived much earlier than the second tier and SARS-CoV-2 has probably been circulating in the first tier population for a longer time.

Italy

Media outlets have drawn many comparisons between the Unitet States and Italy. Let’s consider a few facts about Italy from the CIA World Factbook.

Italy’s land area is 294,140 sq km (square kilometers), roughly the size of Arizona (294,207 sq km). Italy’s population is estimated to be 62,402,659 people (July 2020). California is the most populous state with 39,512,223 residents (July 2019). Overall, Italy has as many people as California and New York State combined in a smaller land mass (Arizona).

Italy’s Lombardy region is, arguably, the Italian region most severely affected by COVID-19. Lombardy is located in northern Italy with the major urban center Milan as its capital.

Lombardy has 10,078,012 people (August 2019). It is the most populated region in Italy. It’s population is roughly the size of Michigan (or North Carolina, or Georgia). Lombardy, by population, would be the tenth largest state in the U.S. Lombardy is 23,844 sq km, roughly the size of West Virginia. Lombardy has a population density of 420 people per sq km.

The city of Milan has 1.369 million people. (The World Fact Book claims 3.140 million people, but it depends on how one draws boundaries.) Milan has a population density of 7,572 people per sq km (2019). That’s roughly the same density as Huntington Park (Los Angeles metro) or Somerville (Boston metro). New York City has a population density of 10,431 people per sq km. [I taught in Somerville; it’s dense.]

Lombardy accounts for 68% of COVID-19 fatalities in Italy. 23% of Italians are 65 or older, making it the second oldest population in the world. (Japan is first.) Age and underlying health conditions are known COVID-19 vulnerabilities. High population density and more elderly people (with existing health issues) are a fatal combination of factors.

Temperature

One final observation because it involves maps. The Kinsa US Health Weather Map has received media attention including MSNBC’s The Rachel Maddow Show. (Hi, Rachel.)

The Health Weather Map is a visualization of seasonal illness linked to fever as measured by Kinsa Smart Thermometers. The Kinsa Thermometer connects to your device (and the Kinsa App) via Bluetooth. The Kinsa Thermometer costs about $20-$25 USD retail and requires the Kinsa App for Web communication.

Here is today’s map (March 26). [Click to enlarge.]

I look at the Health Weather Map with interest, but take it with a grain of salt. There are two factors to consider.

• Although the Kinsa Thermometer is inexpensive, it needs an expensive smart device in order to communicate results via the Web.
• It’s not clear how people use the thermometer, i.e., do they take their temperature every day? When they feel sick?

Given the need for a smart device and Internet connection, I’m not sure if Kinsa users represent the population at large. Use of the device may bias results, too. If users take their temperature only when they are sick, then the map will show more high temperatures. Ah, well, I need to dig into their study. [Thank you, Kinsa, for posting your methodology.]

Stay apart, mate, and we’ll both be healthy — P.J. Drongowski

Yesterday, Governor Jay Inslee issued a full, statewide “stay at home” order for Washington State. Up to this point, home-stay was recommended and did not have the force of law. Now it does.

Thank goodness. When I first glanced at the percentage of new positive cases per day (Washington State), I felt optimistic. After reflection, I decided. “Yes, it’s good that the percentage is not increasing rapidly, but even no change is still bad.” Even a constant flow of new patients into the health care system will eventually overwhelm it. A linear curve, as opposed to an exponential curve, merely postpones the day when we exceed capacity.

Someone might say, “Oh, linear isn’t bad.” Yes, it is and here’s why. It’s not like an acute care patient arrives, is given some magic cure and is sent home. Mean duration of hospitalization in Chine was 12.8 days. [Please see “Clinical Characteristics of Coronavirus Disease 2019 in China” in the New England Journal of Medicine.] Even with a so-called linear curve, acute care patients are admitted and stay. Most hospitals are “right sized” during normal times by administrators interested in efficency. So, it doesn’t take long before capacity and staff are exhausted.

China had hospitalization rates ranging from 15% to 20%. In the Chinese epidemic, 41.3% of patients received oxygen therapy and 6.1% received mechanical ventilation.

There is only one advantage to a linear curve vs. exponential — it gives hospitals and other facilities an opportunity to ramp up for the influx of acute care patients. The availability of respirators and ventilators is a critical concern as well as temporary beds and infrastructure for convalescence and recovery.

We need to push the number of new COVID-19 cases downward through social distancing and good hygiene. I’m glad that Governor Inslee has issued a full stay-at-home order. Now, we all need to listen to it and act.

New York City

Governor Cuomo of New York was visibly more alarmed today. I pulled together a quick table of new case and new testing statistics for New York City (NYC).

I’m alarmed, too.

New York City (NYC) 
             Cum       New     Cum     New          Percent 
Date    Positive  Positive   Tests   Tests  Ratio  Hospital 
------  --------  --------  ------  ------  -----  -------- 
Mar 20      4408      1939   14386    4852  40.0%       18% 
Mar 21      6211      1803   19463    5077  35.5%       15% 
Mar 22      9045      2832   26389    6927  40.1%       13% 
Mar 23     12305      3260   33003    6614  49.3%       13% 
Mar 24     14904      2599   38390    5387  48.2%       -- 
Mar 25     17856      2952   44076    5686  51.9%       12%
Mar 26     21393      3537   51031    6955  50.9%       14%
Mar 27     25398      4005   57954    6923  57.9%       15%
Mar 28     29766      4368   65902    7948  55.0%       --
Mar 29     33768      4002   73104    7202  55.7%       14%
Mar 30     37453      3685      --      --     --       14%

24% of hospitalizations are ICU patients (March 23) 
23% of hospitalizations are ICU patients (March 24) 
24% of hospitalizations are ICU patients (March 26)
24% of hospitalizations are ICU patients (March 27)
24% of hospitalizations are ICU patients (March 29)
25% of hospitalizations are ICU patients (March 30)

The percentage of new positive cases per day (which takes the ever-varying number of tests into account) for Washington State is around 6.5%. The ratio for New York City is 6 times higher than Washington. (Update: Washington State calls this ratio “Positivity Rate.”)

I fear that NYC is already in the weeds. Folks, NYC is the largest financial center in the United States. So, even if all you care about is money, this is the wrong time to lift restrictions and to restart the American economy. Plain and simple, that’s crazy talk. And, immoral.

Stay distant and stay healthy — P.J. Drongowski

Update quoting CNBC News:

New York and New Jersey are seeing coronavirus attack rates at least five times higher than other parts of the country, a U.S. official in charge of the White House’s pandemic response efforts said Monday.

“The New York metro area of New Jersey, New York City, and parts of Long Island have an attack rate close to one in 1,000,” Dr. Deborah Birx, a physician and the White House coronavirus response coordinator, said at a press briefing Monday evening. The attack rate is the percentage of a population that gets the disease.

She said roughly 28% of the specimens submitted in that region have tested positive for COVID-19, while less than 8% have tested positive for the disease in the rest of the country.

New York is currently the hardest-hit state in the country, ahead of New Jersey, California and Washington state. New York City, alone, accounts for 12,305 of the 20,875 confirmed infections in the state as of Monday morning.

% Positive New Cases Per Day

It’s time to show some numbers of my own!

Thanks to The Seattle Times and the Washington State Department of Health (DOH), I have built a table that summarizes COVID-19 case statistics since late February. Data before March 10 is somewhat spotty and lacks the number of tests per day, etc. The raw numbers are shown in the table below.

            Cum      Cum    Cum 
Date   Positive Negative   Tests New Cases New Tests Ratio  Notes 
------ -------- -------- ------- --------- --------- -----  -------- 
Feb 28        1 
Feb 29        7 
Mar 2        18 
Mar 4        39                                            SnoEmerg 
Mar 5        70                                            Pence 
Mar 6        83 
Mar 7       102 
Mar 8       136 
Mar 9       162     1149    1311           
Mar 10      267                        105 
Mar 11      366     3037    3403        99                  Ban>250 
Mar 12      457     4350    4807        91     1404  6.48% 
Mar 13      568     6001    6569       111     1762  6.30% 
Mar 14      643     7122    7765        75     1196  6.27% 
Mar 15      769     9451   10220       126     2455  5.13% 
Mar 16      904    11582   12486       135     2266  5.96%  HomeRec 
Mar 17     1012    13117   14129       108     1643  6.57% 
Mar 18     1187    15918   17105       175     2976  5.88% 
Mar 19     1376    19336   20712       189     3607  5.24% 
Mar 20     1524    21719   23243       148     2531  5.85% 
Mar 21     1793    25328   27121       269     3878  6.94% 
Mar 22     1996    28879   30875       203     3754  5.41%  Everett
Mar 23     2221    31712   33933       225     3058  7.36%
Mar 24     2469                        248                  StayHome
Mar 25     2580                        111
Mar 26     3207    43173   46380       627    12447  7.92%
Mar 27     3700    49015   52715       493     6335  7.78%
Mar 28     4300    54896   59196       600     6481  9.26%
Mar 29     4896    60566   65462       596     6266  9.51%
Mar 30     5515                                             DOHIssues
Mar 31     5984    68814   74798      1088     9336 11.65%
Apr 1      6585    72833   79418       601     4620 13.01%
Apr 2      6966    75633   82599       381     3181 11.98%
Apr 3      7591    80327   87918       625     5319 11.75%
Apr 4      7984                        393                  DOHIssues
Apr 5      8384                        400
Apr 6      8682                        298
Apr 7      9097                        415
Apr 8      9609                        512
Apr 9      9887                        278
Apr 10     10224                       337
Apr 11     10411                       187
Apr 12     10538                       127
Apr 13     10694   112160  122854      166
Apr 14     10783   113500  124283       89    1429   6.23%
Apr 15     11152   117748  128900      369    4617   7.99%
Apr 16     11445   120182  131627      293    2727  10.74%
Apr 17     11802           135706      357    4079   8.75%
Apr 18     11790   126852  138642             2936          DOHCorrection
Apr 19     12085   128926  141011      295    2369  12.45%
Apr 20     12282   132749  145031      197    4020   4.90%
Apr 21     12494   135459  147953      212    2922   7.26%
Apr 22     12753   140623  153376      259    5423   4.78%
Apr 23     12977   147347  160324      224    6948   3.22%
Apr 24     13319   157275  170594      342   10270   3.33%
Apr 25     13521   161956  175477      202    4883   4.14%
Apr 26     13686   165993  179679      165    4202   3.93%
Apr 27     13842   168673  182515      156    2836   5.50%
Apr 28     14070   173730  187800      228    5285   4.31%
Apr 29     14327   179654  193981      257    6181   4.16%
Apr 30     14637   184087  198724      310    4743   6.54%
May 1      15003   192312  207315      366    8591   4.26%
May 2      15185   196820  212005      182    4690   3.88%
May 3      15462   200858  216320      277    4315   6.42%
May 4      15594   203859  219453      132    3133   4.21%
May 5      15905   208908  224813      311    5360   5.80%
May 6      16231   214449  230680      326    5867   5.56%
May 7      16388   219447  235835      157    5155   3.05%
May 8      16674   226315  242989      286    7154   4.00%
May 9      16891   231984  248875      307    5886   5.22%
May 9      16891   231984  248875      307    5886   5.22%
May 10     17122   234986  252108      231    3233   7.15%
May 11     17330   238991  256321      208    4213   4.94%
May 12     17512   243568  261080      182    4759   3.82%
May 13     17773   250158  267931      261    6851   3.81%
May 14     17951   255352  273303      178    5372   3.31%
May 15     18288   262705  280993      337    7690   4.38%
May 16     18433   266810  285243      145    4250   3.41%
May 17     18611   270524  289135      178    3892   4.57%
May 18     18811   274309  293120      200    3985   5.02%
May 19     18971   278971  297942      160    4822   3.32%
May 20     19117   283810  302927      146    4985   2.93%
May 21     19265   289093  308358      148    5431   2.73%
May 22     19585   296691  316276      320    7918   4.04%
May 23     19828   306765  326593      243   10317   2.36%
May 24     20065   310533  330598      237    4005   5.92%
May 25     20181   312610  332791      116    2193   5.29%  George Floyd
May 26     20406   315395  335801      225    3010   7.48%
May 27     20764   322327  343091      358    7290   4.91%
May 28     21071   327162  348233      307    5142   5.97%
May 29     21349   333005  354354      278    6121   4.54%  SEA protests
May 30     21702   339197  360899      353    6545   5.39%
May 31     21977   343295  365272      275    4373   6.29%
Jun 1      22157   346642  368799      180    3527   5.10%
Jun 2      22484   354843  377327      327    8528   3.83%
Jun 3      22729   360858  383587      245    6260   3.91%
Jun 4      22993   367870  390863      264    7276   3.63%
Jun 5      23442   377146  400588      449    9725   4.62%
Jun 6      23729   381327  405056      287    4468   6.42%
Jun 7      24041   386249  410290      312    5234   5.96%
Jun 8      24354   390700  415054      313    4764   6.57%
Jun 9      24642      ---     ---      288     ---    ---
Jun 10     24779   400433  425212      137   10158   4.18%
Jun 11     25171   414691  439862      392   14550   2.69%
Jun 12     25538   424608  450146      367   10284   3.57%
Jun 13     25834   436768  462602      296   12456   2.38%
Jun 14     26158   445107  471265      324    8663   3.74%
Jun 15     26531   453495  480026      373    8761   4.26%
Jun 16     26784   408232  435016      253     ---     ---  DOH no sero
Jun 17     27192   419707  446899      408   11883   3.43%
Jun 18     27601   428340  455941      409    9042   4.52%
Jun 19     28225   437844  466069      624   10128   6.16%
Jun 20     28680   446258  474938      455    8869   5.13%
Jun 21     28870   448334  477204      190    2266   8.38%
Jun 22     29386   457673  487059      516    9855   5.24%
Jun 23     29869   465629  495498      483    8439   5.72%
Jun 24     30367   475428  505795      498   10297   4.84%
Jun 25     30855   483573  514428      488    8633   5.65%
Jun 26     31404   494398  525802      549   11374   4.83%
Jun 27     31752   502691  534443      348    8641   4.03%
Jun 28     32253   515967  548220      501   13777   3.64%
Jun 29     32824   524451  557275      571    9055   6.31%
Jun 30     33435   538529  571964      611   14689   4.16%
Jul 1      34151   550838  584989      716   13025   5.50%
Jul 2      34778   565197  599975      627   14986   4.18%
Jul 3      35247   572029  607276      469    7301   6.42%
Jul 4      35898   576808  612706      651    5430  11.99%
Jul 5      36985   592271  629256     1087   16550   6.57%
Jul 6      37420   598104  635524      435    6268   6.94%
Jul 7      37941   607131  645072      521    9548   5.46%
Jul 8      38581   621749  660330      640   15258   4.19%
Jul 9      39218   629248  668466      637    8136   7.83%
Jul 10       ---
Jul 11     40656   645349  686005     1438   17539   8.20%
Jul 12     41757   666517  708274     1101   22269   4.94%
Jul 13     42304   675930  718234      547    9960   5.49%
Jul 14     43046   690840  733886      742   15652   4.74%
Jul 15     44313   708861  753174     1267   19288   6.57%  Yakima backlog
Jul 16     45067   722590  767657      754   14483   5.21%
Jul 17     46026   745760  791786      959   24129   3.97%
Jul 18     46946   762393  809339      920   17553   5.24%
Jul 19     47743   778611  826354      797   17015   4.68%
Jul 20     48575   792609  841184      832   14830   5.51%
Jul 21     49247   805905  855152      672   13968   4.81%
Jul 22     50009   820754  870763      762   15611   4.88%
Jul 23     50824   833158  883982      815   13219   6.17%
Jul 24     51849   851825  903674     1025   19692   5.21%
Jul 25     52635   866712  919347      786   15673   5.01%
Jul 26     53321   879983  933304      686   13957   4.92%
Jul 27     54205   891029  945234      884   11930   7.41%
Jul 28     54985   903322  958307      780   13073   5.97%
Jul 29     55803   917851  973654      818   15347   5.33%
Jul 30       ---                                             DOH outage
Jul 31     57541   943987 1001528     1738   27874   6.24%
Aug 1      58173   950107 1008280      632    6752   9.36%
Aug 2      58715      ---              542                   DOH issues
Aug 3      59379      ---              664
Aug 4      60084      ---              705
Aug 5      60917      ---              833

I have tried to make this table as accurate as possible. Please forgive any transcription errors. The Washington Department of Health has suffered technical issues, so there are gaps in the data. Their database has been overwhelmed and they have had trouble tallying negative test results and the number of tests.

Don’t forget about the University of Washington Virology Laboratory COVID-19 dashboard as another reliable data source.

The State DOH reports the cumulative number of positive cases (tests) negative test results daily — the second and third columns in the table. The fourth columb, cumulative number of tests, is the sum of the cumulative positive and negative results each day.

From these numbers, we can compute the number of new positive cases and the number of new tests for each 24 hour period:

    Tests.Today = Positive.Today + Negative.Today 
    New cases = Positive.Today - Positive.Yesterday 
    New tests = Tests.Today - Tests.Yesterday 
    Ratio = (New cases / New Tests) * 100.0 

The seventh column, ratio, is the percentage of positive new cases per day. I believe that this percentage is an appropriate measure for tracking COVID-19 over time. The ratio normalizes the number of positive cases for the number of tests performed. Thus, it makes daily results comparable even as our test capacity and number of daily tests grow.

Update: Washington State calls this ratio/percentage “Positivity Rate.”

The bar chart below is a quick and dirty plot of the daily percentage of positive cases over the last ten days. Be sure to check the vertical scale when interpreting the chart!

I hope this trend declines. Please remember that any measurements taken today show what COVID-19 was doing 10 to 14 days ago. The delay is due to the time required for COVID-19 incubation, worsening of symptoms, clinical presentation and testing.

The last column in the table, Notes, are my mental notes marking significant events along the COVID-19 timeline in our area. The events are:

• Snohomish County (my home) declares a state of emergency.
• Vice President Pence visits the State of Washington.
• Governor Inslee bans groups of more than 250 people.
• The State advises people to stay at home, especially the elderly.
• The City of Everett (a stone’s throw away from home) imposes a stay at home order.

Up to today, Governor Inslee has not imposed a blanket “stay at home” order. Suggestions made by the state are similar to a stay at home order, but do not have the force of law.

Don’t do this

Here is the egregious chart of the day by way of Cleveland.com. I still check in with my original hometown. Hope the Cavs can hit the court again.

To be fair, the author did put a few caveats about number of tests, etc. in the article. The caveats are lost in the TL:DR age. Please don’t publish meaningless charts and comparisons.

If I were teaching a math or science class today, I think I would use this chart in class about critical thinking and interpetation of the graphs thrown at us by the media.

No one (CNN, etc.) should compare state-by-state number at this point in time. There isn’t a standard for reporting and test capacity is, literally, all over the map.

Stay distant and stay healthy — pj