Ned Notes (not blog): 08jan21 COVIData Sweep

In remembrance of five fellow citizens taken from us by the COVID epidemic this week. Ranging in age from twenty-one to one hundred-one; average age of seventy; median of eighty-six. Three women and two men; one African American and one Bolivian immigrant. Hailing from Illinois to Florida. 

B.L.U.F. (bottom-line, up-front): a week dominated by politics, yet the deadliest COVID 'weep' to date.

INTRODUCTION
Summary. The back-story is thin since my attention has been far too focussed on an attempted coup d’état mid-week as Trump sought to overturn the election.

Despite the dominance of politics in the news this week, 22,000+ people died during the week ending 08jan21, almost certainly the deadliest week to date of the American epidemic. The coronavirus returned to the states hit last Spring with a vengeance and it continued its blight of others in the in the South and Midwest. The two key questions are the magnitude of community spread in New Year’s week and the pace of national vaccination. For me? I am guessing I will see the needle in August for the first time.

REVIEW of DATA
Summary. Deaths per thousand returned to higher levels among the more densely urbanized states. Cross-border transmissions, almost certainly going both ways along the Southwest frontier with México is not helping.

Relative to population, the states with most deaths per 100,000 residents this week were New Jersey (222 per 100,000); Massachusetts (187); Connecticut (177); the Dakotas (176); Mississippi (172); Louisiana (168); Illinois (150); Michigan (139); Arizona (136); and, Pennsylvania (135).

Rhode Island, wedged as she is between Massachusetts and Connecticut, split the difference between her neighbors with 180 deaths per100,000 residents. Puerto Rico, D.C., and the other territories are managing well with 58 deaths per 100,000, well below the national average in the seventies. New York presents a case in guidance for other metro areas. Whereas the Empire State has 122 deaths per 100,000, the City has 303 and the tri-state metropolitan area scores 256.

The bellwether states are showing a renewed infection rate in the Southwestern states first listed last Summer (i.e., Arizona, California, Florida, Texas, and Oklahoma). The second group of Colorado, Illinois, Wisconsin, Idaho, and Missouri appears to be converging with the national range of growth rates in cases and deaths (i.e., 4-6% for fatalities). Of the original five bellwether states, only Florida and Texas appear to be managing the spike affecting the whole country, though the Lone Star State is teetering. 

California and Arizona and are basically losing control of the contagions spreading within their fronteras, while Oklahoma is skating close to the edge. The struggles faced by states bordering México with two epidemics deteriorating steadily may be exacerbated by a cross-border spread heading north. This directional assessment is less likely for Southern California, since the Tijuana-San Diego mega-region tends to have a lot of flow both ways. Aggravating this threat among the border states is the lack of testing capacity, except in New México. 

Though California’s testing diffusion moderately exceeds the national average, levels should be higher, given the Golden State’s vast resources. Texas and Arizona are only testing at two-thirds and one-half the national level, respectively. These negligent levels create two problems: detecting potential hot-spots in time to shut them down, as New York does, and letting passive carriers breathe freely. The latter seemingly benign neglect presents two challenges: mapping community spread and gauging the unknown long-term effects on asymptomatic cases.

NOTE on testing percentages. When I write percentages for 'population tested', please remember that the calculation implicitly assumes that anyone taking a test does so only one time. Many people are getting multiple tests. For example, for various reasons, I have had four tests. That would count as four people taking tests for the datum calculated for Maryland. My S.W.A.G. (scientifically wild-assed guess) is that a more accurate level of people actually tested is roughly half of the percentage cited in the ‘38+ table’ all the way down.

The following states and territory enjoy 'population' testing rates above 100% of their populations: Massachusetts (167% of the Commonwealth’s seven million people); New York (139%); D.C. (133%); Connecticut (129%); Vermont (119%); Illinois (110%); and Minnesota (104%). Of these states, Vermont is notable because the Green Mountain State has tripled her testing capacity in less than a month to respond to one of the largest spikes in the country. Nevertheless, Vermont still suffers the lowest death toll relative to her population in the nation, as per the ‘38+ table’ below.

The eleven states listed in the table above are of particular concern for the coming week should the next seven days prove to be as intense, or just little less lethal, than this week. Several states that have struggled mightily in recent months -- namely Wisconsin, Iowa, Louisiana, Idaho and the Dakotas -- are not included here. Their recent performances, albeit high cumulatively, have stabilized and there is ample intensive care capacity available. Though available data are scarce, the supply of ventillators appears to remain plentiful across the country.

¿WHAT's NEXT?
Summary. Strap yourselves in, me-lads and lasses. My base case death toll of 335,301 of 05apr20 – deemed stringent at the time – is likely to fall anywhere from a third to a half short of the ultimate number of deaths at 05mar21. 

My revised projections run well above those of the C.D.C. (i.e., the Centers for Disease Control) and of the University of Washington due largely to the sensitivity of forecasts to initial conditions. The S.W.A.G. (i.e., the composite average of the forward-looking scenarios) is 565,805 fatalities at 05mar21 inside a range 522,535 and 579,208. A level matching the 675,000 Americans lost in the three years of the 1918 Flu, while unlikely, remains conceivable after the first year of the epidemic.

That sensitivity to initial conditions makes the unfolding contagion for New Year’s week quite pivotal. The question about New Year’s Eve 2020 is whether it has reflected the fêting of past years. If so, it may end up being almost as catalytic as Christmas week. If more people are like me – even in my mis-spent youth – of not being too terribly fond of so contrived a holiday, people may have stayed home, thus suppressing the community spread a bit. Basically, a few ranges lend some perspective.

1. Mine from April 2020: 335,301 deaths with a fatality range from 253,788 to 502,446.
2. University of Washington from November 2020: 470,494 deaths with no range cited.
3. Washington from January 2021: 536,279 deaths with a range from 512,786 to 565,195.
4. Mine revised for this week. 641,704 deaths with a range from 586,748 to 675,974.
5. The C.D.C. by simple extrapolation: 519,432 deaths with a range from 497,917 to 538,489.

The projections have only grown gloomier over time, a depressing trend largely attributable to the quality of (and example set by) national leadership; a lack of preparedness across the country; as well as, autonomous resources available to states and municipalities. Should this limited number of data not be to your liking, you should be able to find other data easily enough on the inter-net. Four factors account for the wide range among the three projections above:

• my lack of expertise and comprehension of epidemiological forecasting;
• the timing of the dissemination of the vaccines;
• the possible onset of the Italian nightmare of rationing resources away from those most afflicted; as well as,
• the contagiousness realized during New Year’s week.

That said, the scenario to stick with here is most likely that of the University of Washington. The data available to me from the C.D.C. forecast ahead for two weeks only. The six weeks that follow are derived from applying the C.D.C. current two week predictions to mine and taking that factor forward for the out-weeks against my numbers. For the second half of January the C.D.C. projects a range of cumulative deaths from 405,000 fatalities to 438,000, for a level that I infer of 422,500 deaths (i.e., the mid-point of the two numbers).

My projection for total deaths at the end of January comes in at 472,473. To forecast February and early March for the C.D.C. I apply a simple ratio of the C.D.C. death toll to my assumed weekly growth rates for the next two weeks. That yields a calculation of 422,500 / 472,473, or just below 90%, to be applied to my projections for the remaining six weeks. For clarification, my revised projections largely replicate the mortality experience after Thanksgiving.

Data from the ‘38+ table' versus the bellwether states shape the prospective experience after Christmas. The differentiating assumption by me hypothesizes that the last week of 2020 will prove nearly as contagious as Thanksgiving and Christmas weeks. These heroic assumptions are adjusted for possible impacts of accelerated vaccine roll-outs or more of the same lags seen to date leading to contention across segments of the population and local choke-points.

Currently, some 200,000 people have received both vaccines, though few have received their prescribed second dose. Dr Anthony Fauci, Director of the National Institute of Allergies and Infectious Diseases praised this week’s milestone of two million vaccinations administered. That is a large difference with the 200,000 cited above, reconciled by the obervation only 8-11% of the people receiving their first dose received their second shot. Even that lower percentage appears higher to me from what I have read. So, that slippery variable around vaccine roll-outs is extremely important. 

The table above illustrates that, as dramatic as the number of inoculations may sound, the United States is just beginning. Vaccinations are expected first to focus on the elderly and their care-givers, followed by the most vulnerable among so many essential workers, a disproportionate percentage of whom are citizens of color and remain more vulnerable than Whites to the virus due to co-morbidities often associated with poverty. (For an intelligible discussion of the constraints and challenges facing vaccine roll-outs, please revert to the video embedded in this article linked above.)

CLUSTER FACTS
Summary. Most of the news involves restrictions being continued or elevated, with one odd exception: Pennsylvania. The Commonwealth relaxed restrictions on 04Jan21, thus earning her place on the watch-list presented above. 

Overview. Big dichotomy between deliveries, already slower than expected, and actual shots administered (i.e., roughly the administration to people of a third of vaccines received, except for Philadelphia and New York at 50% or so). Few people have received a second vaccine yet (i.e., something like 5%, or less, of those who have had their first shot).

Baltimore-Annapolis: 1% of Cluster receiving at least one vaccine; 15-30% intensive care capacity (I.C.U.) available. Republican Governor Hogan loosens vaccine roll-out under pressure; choke-point around administration of vaccines. Remembering 2020 COVID victims in Anne Arundel County.

Boston & the Cape: 2-3% of the cluster with one vaccine; 20% I.C.U. availability in Boston; 52% I.C.U. availability on the Cape. Boston extends COVID restrictions in place now for three weeks; Commonwealth likely to follow. The Cape remains roughly 53% more safe than Massachusetts. Problems with vaccine roll-out.

Chicago proper: <3% of the cluster with at least one vaccine; 31% I.C.U. availability. Illinois the fifth state to register more than a million cases. Illinois loosening vaccine priorities modestly despite lag in inoculations and resistance among certain health-care workers. Outreach started to Chicago’s essential workers; ¿who are they and how many?

Colorado state-wide: <3% of the State with at least on shot; Denver I.C.U. availability of 36%; Dolores County availability of 46%. Governor Polis aiming for three-quarters of Coloradans over seventy to receive their first vaccine by the end of February. State-wide guidance from University of Colorado. Dolores County deemed one of the healthiest communities in U.S.A.

Detroit metro: <2% of Michigan with at least one vaccine shot; 20% I.C.U. capacity in Detroit. Gov. Whitmer relaxes some priorities on vaccines. Michigan hits demographic hiccup. State presents its plans and statistics for contagion and re-opening.

New York METRO: 1% of the State vaccinated with first shot; 37% I.C.U. availability. NYC vaccine distribution among boroughs, including local prisons. Westchester status report. More contagious Euro-mutation now into Connecticut and New York. Despite being ahead of the other states, Empire State still slow with the needle. Northern Jersey still sorting out vaccine priorities; half a million have signed up across Garden State.

Pennsylvania: Philadelphia and Pittsburgh; <2% of the Commonwealth with one vaccine; I.C.U availability of 24% in Philly and 12% in PGH. Philadelphia facing vaccination constraints; up to a year for fully vaccinated population. PITT vaccine being developed. Euro-virus into Philadelphia. Steelers country on COVID lock-out this Sunday. (😭 ¡¡¡BUMMER!!! 😷)

Tunisia and México.
Tunisia's COVID-19 infection rate remains mysterious due to little testing. At this point, la République Islamique has few cases and deaths recorded. She is twice as safe as the United States, since her mortality rate relative to the population is 40% of the U.S. level (i.e., restated deaths of 144,229 versus 371,886). The static pool death rate (i.e., fatalities relative to cases resolved in death or full recovery) of 4.2% is higher than the American rate, likely attributable to fewer tests.

The mortality run rate of la Tunisie, at 3.3%,  is almost twice as high as the U.S. percentage, again derivative of having only 7% of the American testing level. Intensive care availability appears to be strained but manageable at a level of 20%, perhaps more. The country would rank in the upper quartile of U.S. states from an overall perspective; this may change as this surge continues. Where Tunisia suffers currently is in the drying up of tourist revenues.

México has suffered greatly during this winter spike, despite her warmer climes. Most of her epidemic clusters lie in the Northern region, which borders the United States. There is likely some cross-border spilling over of the U.S. epidemic into México, and a unified hot-spot in the Tijuana-San Diego super-region. Nevertheless, the los Estados Unidos Mexicanos have begun to bring the surge in cases under control. 

México's fatality rate relative to population is roughly 9% below that of her northern neighbor. The static pool and mortality run rates however remain concerning at 10.4% and 8.7% since the world's largest Spanish speaking country basically tests people once they are almost certainly infected; that is to say, México's testing capacity is less than 4% that of los Estados Unidos de Norteamérica. México's recent performance would place her in the middle third of U.S. states based on their recent performance.

METHODOLOGY 
Since this data sweep serves as a information supplement without very little research, this week provides an opportunity to clean up the presentation through the following changes. For informational purposes, this text will follow the second table on thirty-eight states and five territories in subsequent weeks. For comprehensive explanations on methodology and purpose, please revert to Appendix I. 

1st, the two averages of weekly growth rates for the ten bellwether states are clarified in the Appendix. Essentially, the geometric or compound average growth rate is a smoothed average that allows for growth-on-growth increases (i.e., similar to compounding interest). The time weighted weekly averages are a trend-weighted average of each week's particular growth rates. While the compound rate is theoretically more defensible, comparing the two averages gives one a sense of more recent trends and volatilities.

2nd, in the 3table (of thirty-eight states and five territories), the risk classifications  -- of very low; low; moderate; high; and, very high -- remain the same. The parameters, however, are loosened to reflect nine months of experience. The new parameters center upon the first year base case fatality level of 335,301 souls that I forecast eight months ago, together with the updated projection from the University of Washington of 470,974 deaths. On 04dec20, the old and new parameters are applied to facilitate  transition.

3rd,changes in how positivity rates are presented in the 38+ now align the data with the original intention behind presenting them. The intent here is to show whether positivity is trending up or down and to what degree. The parameters are loosened to match the practical reporting constraints and data lags facing most states. 

States with changed positivity rates of less than 10% up (i.e., worsening) or down (i.e., improving) are deemed unchanged and the information unformatted (i.e., appearing in plain black font). The formatting differentiates deteriorations from improvements in the color of the font between 10-20%. Bold fonts indicate material deteriorations or improvements of more than 20%. 

Keep in mind that these percentage changes are based on percentages; percentages of percentages can attenuate the utility of data.

4th, a refresher on the assessment of a state’s testing capacity, again on the 38+ table. The data pivot off of the tests per million people expressed as a percentage. The symbology uses hand gestures to assessment the degree of testing capability and commitment relative to the national average for the week under review. Colors indicate whether testing is declining (red font) or improving (blue font) when a states results place its commitment to a new category of, specifically:

• 👎👎 meaning a state's testing level materially below the concurrent national benchmark (i.e., > 15 points below);
• 👎 meaning a state's testing is noticeably below the weekly national average (i.e., 5-15 points below);
• 👈 meaning a state's testing activity is slightly lower than average (i.e., < 5 points below);
• 👉👈 meaning a state's testing level is basically equal to that of the nation;
• 👉 meaning a state's testing level is < 5 points above the benchmark;
• 👍 meaning a state's testing activity is 10-15 points above the weekly national level; and,
• 👍👍 meaning a state's testing commitment is > 15 points above the concurrent average.