JF Insights

A quick update on the diverging paths of the US & Brazil versus Europe.

Lets look at fatalities – comparing US with a basket of European countries that in total have a similar population to that of the US. See chart below.

To keep the time axis equivalent to the different starting points of the virus in each county, the clock starts for each country on the day of the 100th death.

Europe has been below 50 deaths per day for the last 2 weeks. The US is over 500 (even ignoring the NJ adjustment noted below). Brazil has been over 1,000 deaths per day for the past week.

The point where Europe starts to diverge from the US is around day 55. From this point you see the European covid deaths decline sustainably. The big message here is simply that US deaths have not fallen to the same level as those of Europe. We never reached those low levels. Essentially – Europe kept the focus – but the US seemingly lost its focus.

Lets look at the case data to illustrate the problem further. The US is now at over 50,000 new cases per day. That’s higher than the numbers recorded in March or April.

Europe has been well below 1,000 per day for the past week – and for the past few days has been below 300 per day.

Brazil daily new case aveage is around 37,000 (Brazil has a population of 208m – so prorating that for the US would give around 57,000 daily cases).

So – there we have it – the US has more than 50x the daily case load as the European Countries in this group – and more than 10x the average daily deaths.

Notes on the data:

1) – European cases spike negative around day 107 due to UK decision to change COVID methodology and remove about 20,000 fatalities and cases from their data set.

2) The large spike in the US curve around day 100 – is due to NJ adding 1877 deaths as part of a methodology adjustment. Ignore that spike and look at the trend. It has been declining – but we are now starting to see it uptick around day 115.

Here is the list of worst counties based on new cases per million. The criteria is all counties with population greater than 100,000 people. In the list below – all counties have over 400 new cases per day, per million of population. Arizona is not relinquishing it’s spot as the new COVID epicenter of the US.

State	County	New Cases per million	New Fatalities per million
Arizona	Yuma	1034	19
South Carolina	Charleston	788	5
California	Imperial	784	26
Texas	Nueces	740	3
Florida	Miami-Dade	699	5
Louisiana	Calcasieu	664	3
Texas	Jefferson	620	2
South Carolina	Horry	602	4
Arizona	Maricopa	552	4
Florida	Duval	530	1
Florida	Escambia	515	1
Tennessee	Davidson	497	2
Texas	McLennan	495	3
Florida	Broward	484	2
Louisiana	Lafayette	483	4
Georgia	Muscogee	450	7
Florida	Orange	447
Georgia	Bibb	444	3
Texas	Galveston	444	3
Arizona	Navajo	442	12
Georgia	Whitfield	440	4
Arkansas	Washington	435	1
Florida	Hillsborough	429	2
Florida	Lee	428	2
Florida	Osceola	428	1
South Carolina	Dorchester	418	2
Florida	Leon	412
Tennessee	Shelby	410	3
Texas	Brazos	410	3
Idaho	Canyon	405

The fatality data lags cases, but as you can see Arizona cases have spiked for so long that in the case of Yuma, the fatalities are now growing significantly.

Here’s the full list of all counties with population greater than 100,000 and more than 200 new cases per day, per million of people. In total 141 counties meet those criteria. The list is organized by state for easy reference.

Notice the Flu Ratio column – this shows total fatalities for the county – expressed as a percentage of a typical flu level of deaths. Navajo, Arizona for example has a level of deaths 7 times greater than typical flu levels.

State	County	New Cases per million	New Fatalities per million	County Population	Flu Ratio
Alabama	Montgomery	340	6	226,486	272.94%
Alabama	Jefferson	334	4	658,573	137.65%
Alabama	Tuscaloosa	301	4	209,355	123.53%
Alabama	Lee	280	2	164,542	135.88%
Alabama	Etowah	264		102,268	74.71%
Alabama	Mobile	227	1	413,210	196.47%
Alabama	Morgan	226	2	119,679	24.71%
Alabama	Shelby	216		217,702	62.35%
Alabama	Madison	201		372,909	12.35%
Arizona	Yuma	1034	19	213,787	313.53%
Arizona	Maricopa	552	4	4,485,414	115.88%
Arizona	Navajo	442	12	110,924	705.29%
Arizona	Coconino	342	4	143,476	393.53%
Arizona	Pinal	320	4	462,789	95.29%
Arizona	Pima	286	2	1,047,279	158.24%
Arizona	Cochise	270	9	125,922	65.29%
Arizona	Mohave	254	2	212,181	244.12%
Arkansas	Washington	435	1	239,187	78.82%
Arkansas	Pulaski	204	1	391,911	90.00%
California	Imperial	784	26	181,215	367.06%
California	Marin	363	1	258,826	42.94%
California	Stanislaus	343	1	550,660	48.24%
California	San Joaquin	320	1	762,148	42.35%
California	Riverside	306	2	2,470,546	114.12%
California	Tulare	302	4	466,195	171.76%
California	Kings	301	8	152,940	127.06%
California	Orange	283	2	3,175,692	67.06%
California	Los Angeles	243	3	10,039,107	204.12%
California	Fresno	233	1	999,101	45.29%
California	Merced	230		277,680	23.53%
California	San Bernardino	224	1	2,180,085	72.35%
California	Santa Barbara	219		446,499	38.24%
Florida	Miami-Dade	699	5	2,716,940	225.88%
Florida	Duval	530	1	957,755	41.76%
Florida	Escambia	515	1	318,316	87.06%
Florida	Broward	484	2	1,952,778	124.12%
Florida	Orange	447		1,393,452	24.71%
Florida	Hillsborough	429	2	1,471,968	60.00%
Florida	Lee	428	2	770,577	125.88%
Florida	Osceola	428	1	375,751	42.35%
Florida	Leon	412		293,582	15.88%
Florida	Pinellas	368	6	974,996	119.41%
Florida	Martin	360	3	161,000	102.35%
Florida	Collier	346	5	384,902	130.00%
Florida	Manatee	345		403,253	191.18%
Florida	Polk	335	2	724,777	85.29%
Florida	Palm Beach	329	4	1,496,770	212.94%
Florida	Bay	315	1	174,705	17.06%
Florida	Santa Rosa	309		184,313	28.82%
Florida	Seminole	295	1	471,826	24.71%
Florida	St. Johns	261	1	264,672	17.65%
Florida	Lake	253		367,118	37.06%
Florida	Pasco	251	2	553,947	25.29%
Florida	Sarasota	249	1	433,742	132.94%
Florida	Indian River	244	1	159,923	62.35%
Florida	Alachua	238		269,043	26.47%
Florida	St. Lucie	228	2	328,297	89.41%
Florida	Clay	214		219,252	91.18%
Florida	Brevard	211		601,942	18.82%
Florida	Okaloosa	204	1	210,738	22.35%
Georgia	Muscogee	450	7	195,769	156.47%
Georgia	Bibb	444	3	153,159	157.65%
Georgia	Whitfield	440	4	104,628	72.94%
Georgia	Chatham	363		289,430	75.29%
Georgia	Lowndes	349	4	117,406	60.00%
Georgia	Richmond	301	4	202,518	157.06%
Georgia	Gwinnett	299	1	936,250	108.82%
Georgia	Fulton	274		1,063,937	173.53%
Georgia	DeKalb	274		759,297	134.12%
Georgia	Clarke	249		128,331	68.82%
Georgia	Clayton	222		292,256	158.82%
Georgia	Cobb	216	1	760,141	190.59%
Georgia	Douglas	212		146,343	144.71%
Georgia	Hall	201	1	204,441	172.35%
Georgia	Carroll	200	1	119,992	195.88%
Idaho	Canyon	405		229,849	15.29%
Idaho	Ada	295		481,587	28.24%
Indiana	Elkhart	213	3	206,341	142.35%
Iowa	Black Hawk	251	1	131,228	260.00%
Iowa	Johnson	245		151,140	31.18%
Kansas	Wyandotte	326	4	165,429	302.35%
Louisiana	Calcasieu	664	3	203,436	170.59%
Louisiana	Lafayette	483	4	244,390	115.29%
Louisiana	Ouachita	372	7	153,279	307.06%
Louisiana	Livingston	313	1	140,789	158.82%
Louisiana	Terrebonne	308		110,461	324.71%
Louisiana	East Baton Rouge	295	2	440,059	366.47%
Louisiana	Caddo	291	7	240,204	595.29%
Louisiana	Tangipahoa	252		134,758	187.65%
Louisiana	Jefferson	224	1	432,493	662.35%
Louisiana	Ascension	205		126,604	269.41%
Louisiana	Bossier	205	7	127,039	162.35%
Mississippi	Hinds	332	2	231,840	101.76%
Mississippi	DeSoto	281	1	184,945	51.18%
Mississippi	Madison	207	3	106,272	188.24%
Missouri	Jasper	256		121,328	4.71%
Nevada	Clark	285	2	2,266,715	114.12%
North Carolina	Mecklenburg	285	1	1,110,356	82.35%
North Carolina	Durham	240	2	321,488	122.35%
North Carolina	Johnston	220	3	209,339	92.94%
North Carolina	Gaston	200		224,529	28.82%
South Carolina	Charleston	788	5	411,406	51.76%
South Carolina	Horry	602	4	354,081	90.00%
South Carolina	Dorchester	418	2	162,809	18.24%
South Carolina	Berkeley	369	2	227,907	61.76%
South Carolina	Greenville	306	2	523,542	97.65%
South Carolina	Richland	289	2	415,759	122.94%
South Carolina	Florence	253	7	138,293	238.24%
South Carolina	Sumter	253	3	106,721	137.65%
South Carolina	Beaufort	250	4	192,122	67.65%
South Carolina	Lexington	234	4	298,750	112.35%
South Carolina	Pickens	221	1	126,884	22.94%
Tennessee	Davidson	497	2	694,144	103.53%
Tennessee	Shelby	410	3	937,166	125.29%
Tennessee	Bradley	342		108,110	16.47%
Tennessee	Rutherford	268		332,285	61.76%
Tennessee	Williamson	218	1	238,412	37.06%
Tennessee	Sumner	214	1	191,283	160.00%
Tennessee	Wilson	207	1	144,657	69.41%
Texas	Nueces	740	3	362,294	22.94%
Texas	Jefferson	620	2	251,565	81.76%
Texas	McLennan	495	3	256,623	25.29%
Texas	Galveston	444	3	342,139	78.82%
Texas	Brazos	410	3	229,211	87.06%
Texas	Travis	360	2	1,273,954	63.53%
Texas	Bexar	328	1	2,003,554	38.24%
Texas	Webb	311	2	276,652	59.41%
Texas	Hays	308	1	230,191	17.65%
Texas	Dallas	307	2	2,635,516	88.24%
Texas	Comal	282	2	156,209	34.12%
Texas	El Paso	272	1	839,238	94.71%
Texas	Lubbock	264		310,569	98.24%
Texas	Ellis	254		184,826	60.59%
Texas	Hidalgo	250	4	868,707	39.41%
Texas	Kaufman	250	1	136,154	21.76%
Texas	Ector	241	2	166,223	31.76%
Texas	Harris	222	1	4,713,325	50.00%
Texas	Brazoria	211		374,264	34.71%
Texas	Cameron	201	4	423,163	92.94%
Utah	Salt Lake	235	1	1,160,437	57.06%
Washington	Yakima	347	5	250,873	375.29%

At this point, what has Arizona got to lose?

Its really hard to see what’s happening in the USA – and in particular states like Arizona. As widely reported – Arizona has been shooting up the league table for new cases of Covid – since early June. By the end of June, also the end of Q2 – it could hardly have been in a worse position imaginable. While during Q1 – all eyes were on the North East states in particular, many other states enacted lock downs and kept their numbers in check.

Here’s the data in league table format. Arizona is the #1 state for new cases per capita – and is moving ever higher on fatalities per capita also.

• 

• 

Note – the recent spike in NJ fatalities is due to a recent decision to add those people who died of presumed COVID but were not tested at the time into their stats. 1,877 deaths were added into their population of numbers on June 25th. Expect the NJ ‘new fatalities’ number to fall back in line with their trend in about a week.

Across the US – we have a sea of RED – showing many states with case levels far too high to support an acceptably low level of fatalties.

The result in aggregate for the US – is a very disappointing level of cases – that exceeded those recorded earlier in the crisis. Granted that much more testing is being done now, than was the case in early April – but lets not forget that prior to Memorial Day – the case trend was definitely downwards.

Whatever we were all doing, prior to Memorial Day, we need to return to that. There’s no cause for complacency here.

So that brings me back to apps that enable users to sign up for contact tracing. The latest updates from IOS provides the operating system support to enable apps to support contact tracing, without violating privacy data. Some states like Alabama, North Dakota, South Carolina have signed up to use these features. Many states have not – including ARIZONA. Read about this here.

Now – why would the state with the largest growth in COVID cases, say no to technology solutions to help people understand if they may have been exposed? It defies belief.

Starting with an assumption that all cases in the last 2 weeks are contagious – we get these startling stats on which states have the most contagious populations. We hope this population (after all they have been tested postive), are self isolating – but there is also an unknown population of untested, asymptomatic people. Anyhow – knowing what we know, which is the number of new positive COVID tests over the last 2 weeks here’s what we get.

Now, some states have bigger populations than others; after all California is home to 40 million people. So lets adjust for that and look at all states on a per capita basis. Here’s what we get:

Arizona is the #1 state – per million of residents with the number of known cases, assumed to be contagious. South Carolina is #2, Arkansas #3, Alabama #4, Florida #5.

Lets expand this a bit – to identify the people who we assume to be ‘coping with covid’ – those who may, at some point, be in need of hospital care. If we assume that coping window to be approximately 1 month, we get the following CASELOAD stats – per million.

For states in the top of this leage table, the potential stress on the health care system are going to be very significant.

Recall my earlier post on the high bar set by Vermont. Vermont is using a caseload measurement to determine who is allowed to visit. A sensible measure to prevent Vermont’s numbers from skewing higher. Take a look at where Vermont sits, basically 49 out of 51 states (including Washington DC – which I know is not a state).

How did Arizona, and many of these states get so bad, so quickly? The answer has to be some very poor decisions by individuals and leaders. COVID is making us realize that we cannot assume our fellow citizens will act sensibly and rationally. That’s so disappointing.

In a few weeks time we’ll see how today’s caseload numbers translate into fatalities. If past trends hold, then expect very significant fatalities. I would not at all be surprised with possibly around 8 – 12 fatalities per person per day in the next few weeks in Arizona.

Up until Memorial day, Arizona was trending at about 45 new cases per day per million people – with a fatality rate of about 1.7 fatalities per day per million people and improving steadily.

In a little less than 1 month, new cases are now 8x higher at 360 per day, per million people. Fatality levels have already increased again to 3 fatalities per million per day and are on the rise.

We have to unpick the decisions taken in Arizona around Memorial Day and learn these lessons to apply to the next crisis.

Although the US has the highest number of COVID deaths there are often some very uneven comparisons. Like comparing the US to Spain, or Italy. The fact is the US is 5 times bigger than those countries so we would expect our numbers to be higher.

We can make a more even comparison by comparing the numbers in aggregate for a group of European countries – such as UK; Italy; France; Spain and Germany. Taken together these have a population of 345 million – almost in line with the US at 326m.

Here’s what we see on the fatalities. The measurement here is from the date of the 100th death. This allows us to compare each measurement at an equiavlent point of time – since we know that Italy and Spain started COVID earlier than in the US.

US deserves a lot of credit for avoiding the peak that Europe experienced. The US did flatten the curve. However starting in May – European countries started to reduce average daily deaths faster than the US.

By this current point – the US is running at a daily average number of deaths approximately 3x that of the European counties I mentioned earlier.

Cases

We can really see the driver for this in cases. Although the US is testing more than ever (over 400,000 tests per day) – the surge in cases can’t be explained simply by more testing. The fact that US deaths remain higher and are not flattening as quickly reinforces the fact more needed to be done.

US cases bottomed out at about 20,000 per day on average. Unfortunately they are now on the upswing again – driven by the surge in cases in the South.

The media images of people not practicing social distancing and not wearing face masks demonstrate that complaceny is at large. We need to be more vigilant at this point. The early US lead, effective at avoiding the peak of fatalities, has been squandered. While the US states hit hardest by the virus initially have kept their focus – many others are learning the lesson the hard way.

What happened to Test, Trace, Isolate – and the apps that can help us understand if we have been in proximity to someone who tested positive?

By now we know that to get the best view of the COVID data we have to identify the specific locations that are creating the excess number of cases that together are keeping the overall US case count too high.

This post focuses on the 35 worst counties in America – those with populations of over 100,000 – and with a current new case daily average in excess of 100 cases per million of population. The reason for only looking at counties greater than 100,000 population is that anything smaller can be skewed very easily. Data as of June 18th.

With those criteria in mind, here are the worst counties for current levels of fatalities and current rate of case creation.

Lets go through some of the numbers. First daily new cases. The US average case to fatality rate is about 3.5% – so any county today experiencing 100 new cases per day – could likely experience about 3.5 deaths in a few weeks. See my earlier post for the case to fatality rate in the US. That seems far too high to be acceptable (and is higher than the US average today of 2 fatalities per million per day.

I’ve taken the view that the only long term acceptable fatality rate is about 1 person per million per day (that’s broadly equivalent to seasonal flu over a 6 month period) – so if COVID stays at these levels for a whole year – that’s about double the seasonal flu level.

Secondly, daily fatalies. I’ve filtered this data set on those counties currently exceeding 3 fatalities per million per day. So that’s 3x what I would expect to be acceptable in the long term.

The combination of some of these stats gives some very dire projections. Take Yuma Arizona, currently 18 fatalities per million per day – and daily new cases of 861 per million per day. That means future fatalities (taking the 3.5% trend above) in this county could be closer to 30 fatalities per million per day. See my earlier post on Arizona – and the potential link between poverty and those counties with high caseloads.

Now – I’ll be the first to admit that Yuma only has a population of about 214,000 so – in real numbers (not per million) – the results will appear smaller. But expressing data per million of population is the only fair way of comparing very uneven data sets.

Many of these counties tend to have a picture like the one below. Basically at no point was COVID successfully brought under control. Now with the ending of restrictions in many states – we see an upsurge in cases and fatalities.

I welcome thoughts, comments on this article as well as suggestions for future analysis.

Be careful out there. Especially in the counties mentioned here.