Analyzing Arizona – Poverty may be the key

All eyes seem to be on Arizona these days as this state has the highest rate of new cases per capita in the country at 222 per million per day.

Taking a deeper dive, the counties with extremely high fatality rates (practically off the charts) are Santa Cruz, Navajo and Yuma – with respectively 22, 21, and 17 fatalities per million per day. Granted the populations in these counties are not very large, which shows how this virus can rip through small populations very quickly.

CountyPopulationNew Cases / Day / millionFatalities Per Day / millionDominant Ethnicity
Santa Cruz47,000148422White (73%)
Navajo111,00063121Native American (47%)
Yuma214,00072517Hispanic / Latino (63%)
Apache71,00048716Native American (73%)
Various data for the top 3 counties with fatalities per million in Arizona

The counties with the highest populations in Arizona – Maricopa (4.5 million) and Pima (1 million) are broadly in line with the national average for fatalities – about 2 fatalities per million per day. However even in those counties, new case creation is far too high (at 204 and 159 per million per day respectively). It’s out of control!

For the worst affected counties (Santa Cruz, Navajo, Yuma, Apache) – the level of infection and fatalities is even further out of control.

Not Ethnicity – more likely Poverty

I expected to see ethnicity being a big factor – so was surprised to see that Santa Cruz is predominantly white – although I do not know if the portion infected or dead from COVID in Santa Cruz is in line with the overall demographics. We have to assume that it is.

One factor that is consistent in all of these counties is the high poverty rate. According to – the poverty rate of all counties is in excess of ~20%. Here are the results:

CountyPoverty Rate
Santa Cruz21.8%

By contrast, the largest county – with a fatality rate of 2 per million per day (one of the best in Arizona) has a poverty rate of 15.7%. Yavapi County with a population of 235,000 has about 1 fatality per million per day and a poverty rate of 14.7% – one of the lowest (it also has 80% white ethnicity which could also be a factor).

Why is COVID taking hold in these communities?

Perhaps there is less protective equipment (e.g. facemasks, hand sanitizer) etc in these communities, or perhaps they reopened too quickly thinking the peak had passed, or perhaps there is more crowded living accomodations in these communites; or perhaps a less healthy population in general as a result of lower income.

In Arizona – aside from seemingly poor covid management in general, locations of high poverty seem to be disproportionately driving the huge surge in cases and fatalities in those locations.

There has to be a POVERTY lessons learned from this state that can be applied to other states also.

South leads the way in cases, but the North leads the way in fatalities… for now

New Case Data

A lot of data to cover in this analysis. Starting with cases – as of the data I have downloaded (to June 16th) – you can clearly see an explosion of cases in the South. See below:

Arizona has been widely reported is leading the way, but as you can see from this picture, many states are exceeding 100 new cases per million per day.

Context – from my previous posts – the fatality rate for a new case today is approximately 3.5% – so anything greater than 30 new cases per day – will exceed 1 fatality per million in the near term. If that trend holds – then look for possibly 7 fatalities per million per day in Arizona in the next few weeks.

Also recall that at New York’s peak in early April – over 500 cases per day, per million were being reported. Rhode Island, Massachusets, and NJ all hit a peak exceeding 360 cases per million per day in late April. All now feature on the top 3 for recent fatalities. Case data matters and seemingly is a good predictor of fatalities 3 to 6 weeks in the future.

Fatality Data

Today’s fatalities however are driven by cases from a few weeks ago. Here the North – and in particular the North East leads the way – in paricular Rhode Island. Only 16 states are at, or below the level of 1 fatality per million per day.

Hospital Capacity

Its time to relook at hospital capacity. Looking at the past few weeks of case data, and dividing that by an internet search of hospital bed capacity – we get the following estimate of which states may start to experience constraints against capacity. The good news – nothing looks too severe right now – but the media focus on Arizona is appropriate (but they should also consider Maryland, Utah, Nebraska, Indiana etc).

The assumptions behing the expected bed need are 15% of current caseload – which is estimated as the past 1 month of cases. Both are estimates / assumptions – but I’m sure the general direction of the analysis is sound. 15% was used by NY for capacity planning in the early stages of COVID.

More needs to be done to limit case growth in the South. Anything more than 30 new cases per million per day, and without a declining trend, should be a cause for concern.

Brazil (and Mexico)- can we rely on the data?

I’m puzzled by Brazil. It is the 2nd in the list of new case creation globally – at 122 new cases, per million, per day. It is now the country with the 2nd highest death toll at 42,720 deaths – just edging the UK into 3rd place.

Brazil interests me because it has a very large population (208 million) – and some very densely populated cities (Sao Paulo, its largest city has 12.8 million people). Its GDP per head is around $10k – making it poorer than the US (at $60k). It’s COVID wave started much later than the US, or Europe making it an interesting case study. In the early days of its crisis, the country’s leadership seemed to be in denial.

My expectations therefore were for Brazil to have a very bad experience with COVID. While then 2nd highest number of fatalities might correspond with that narrative it still seems less than I would expect.

Let’s look at the pace of fatalities. From the date of the 100th death – to reach the 1000th death it took Brazil 14 days. It took the UK 12 days, US, Germany and France 11 days, Italy 9 days, and Spain 8 days. Everything seems broadly in line here.

Now, looking at the date of the 10,000th death, it took US and Spain 22 days, France 25, Italy 26, UK 27. Well you get the picture. All of these countries experienced a COVID progression roughly in line with each other.

Now consider this for Brazil – the time to reach the 10,000th death is 43 days. That’s 70% slower than the progression in all of the other countries I have mentioned here. I find that hard to understand. How can Brazil have kept up with a high rate of new cases – and at the same time slowed the progression of the fatality of the disease so quickly? One answer could be improved health outcomes – another could be under reporting of fatalities.

Mexico is showing the same trend as Brazil. Mexico reached its 1000th death 18 days after recording its 100th death. It reached the 10’000th death 57 days later. Mexico has a GDP per head similar to that of Brazil – around $9k per head.

Another key stat – Mexico’s current fatality rate exceeds 11%. It has average daily deaths of 468, and average daily new cases of 4,167. This means Mexico is likely under-reporting cases (the longer term fatality rate is closer to 3.5% in the US and is currently about 3.7% globally). The fact that the data shows that Mexico has taken 57 days to reach the 10’000th death with the current high fatality rate is inconsistent.

If the result is that the data is under reported – then this means the COVID risk is much greater in these countries. This should influence travel restrictions imposed by other countries.

The US currently has travel restrictions from Brazil and Mexico. Other countries should follow suit.

Declining Fatality Rate in the US, but cases are not falling fast enough

I had an intuition, or rather expectation, that as the COVID crisis continues, the fatality levels would fall. The drivers behind my reasoning were as follows:

  1. Medical care would get better through the learning curve. Examples of this include reports of not putting people on ventilators immediately when oxygen levels are low – but monitoring closely instead
  2. Attempts to protect the more vulnerable segments of the population would become more effective. Example better ring fencing of nursing homes etc.
  3. More testing means people who previously had COVID but were not sick enough for hospital treatment and were not included in the COVID case stats – would now be captured in the data. Previously these people recovered at home, many of whom were untested. The US is now producing over 400,000 tests per day.

Taking data since April 21st – and comparing 7 day average fatalities with the 7 day average number of cases (lagged 3 weeks) results in the chart below. You can clearly see that from the early peak as the virus impacted the weaker segements of society, and testing data was incomplete / inadequate – we can see the number of fatalities stabilizing at around 3.5% of those confirmed with COVID.

US fatality rate is declining – but very slowly

3.5% is still a high number. To continue my theme on acceptable risk. If we assume an acceptable level of COVID fatality is 1 per million per day – that means the number of new cases per day must be less than 28 per million. Only 12 states meet that criteria. The US aggregate numbers show 64 new cases per million per day. Slightly more than twice the level needed to remain within an acceptable level of fatalities.

On the global basis here are the large countries with lower levels of new case creation

CountryNew Cases Per million
Some countries with current low levels of COVID case creation. US for comparison

US COVID Hotspots Today

Here’s the latest data showing the rate of new case creation in the US today by state. The good news – NY is less than half the level of some of the worst states for current case creation.

The US total is currently 64 new cases per day, per million people. Fatalities are 2.6 per million people. The US number of cases per million per day peaked in early April at 96 – so we are 33% less than that – however the progress has stalled. The analysis below – shows those states preventing the US average new cases figure from falling further.

Arizona currently has the highest levels of new case creation

Lets see how this is impacting fatalities:

Rhode Island leads the way in fatalities per million per day – at over 10. Unfortunately it’s too small to see in this US map – so here’s the data in a table. As I’ve posted on my blog – my view of acceptable risk is 1 fatality per million per day; 3 is on the path to acceptable and anything above 5 is far from acceptable.

Here’s a look at my assessment of the US by state – with a focus on the current worst 20 in terms of current 7 day average fatalities per million.

Although Rhode Island is the worst – it’s actually improved week over week both in terms of daily fatalities and rate of new case creation.

Illinois has improved on the rate of new case creation – but its fatality rate is proving to be very stubborn and not coming down at all. In fact, recently the opposite has occured – fatalities have jumped.

Why you need DRIVE to be successful in Risk Management

Determination:  Be prepared to make tough decisions including those that constrain business activity.  Data Driven:  Transparency is the key to identifying risks and avoiding the bias that can come with gut feel.

Resourceful: be creative in how you use your time and resources to focus on analysis and decisions that have the most impact. Respected:  Cultivate respect by being measured in what you say; using data to support your case.  Listen and be open minded.   Risk/Reward:  Understand the Risk/Reward tradeoffs – be commercial in your decisions. 

Insight/ Intuition:  Follow your intuition to identify potential future risks.    Challenge success.  Be proactive. Independence:  You have the independence and freedom to focus on the areas of risk you think could manifest a number of years down the road.  Many catastrophic risks are years in the making.

Vigilant: Defend against the recurrence of risks that have occurred in the past or in other organizations.  Complacency kills!  . Visionary:  look for the next risk that could threaten your business.  Be creative – think about the worst case scenario and use this as an opportunity to strengthen controls to prevent that scenario from occurring.

Empowered & Empowering:  You are empowered to chase down risk.  Empower your team with the same mindset.

What behaviors do you think are critical for success in risk management today?

Vermont Sets a Very Tough Standard for Vacationers

Photo by William Alexander on

This post is inspired by a question from a subscriber. Thank you!

Ah – the peace and tranqulity of Vermont. Who wouldn’t want to escape the heat of the south, or the bustle of the city with a relaxing vacation in the countryside.

Well before you pack for that trip consider the states “CROSS STATE TRAVEL” advisory. If you come from a higher risk state or country forget it (or for my Brooklyn friends fuggedaboutit!

This is a sensible step for Vermont – it is already one of the lowest risk states in the US – with virtually zero average death rate in the past week or so. Lets look at how Vermont has defined it. It says if your home county “caseload” is greater than 400 per million of population – you can’t come in.

I don’t see a definition for caseload – but I’m going to assume this is the likely number of contagious people. If we assume a contagious period of 2 weeks – and look at the average daily number of new cases over the past 2 weeks for each state – we can begin to see who qualifies for that. Answer not many.

By my calculations – the states that would qualify to travel to Vermont are (in lowest order of caseload first)

  1. Hawaii
  2. Montana
  3. Oregon
  4. West Virginia
  5. Wyoming
  6. Alaska
  7. Oklahoma
  8. Idaho
  9. Maine

But bear in mind – the calculations are done at the county level

Over 1 million COVID Recoveries in the US

Here’s a number you don’t see grab the headlines much…. the number of people in the US now recovered from COVID is likely in excess of 1 million!!!

The number of daily cases of COVID in the US peaked on April 24th – around 46 days ago (at the time of writing this blog). At that time 31,430 cases per day were being identified. Today that number is closer to 22,000 – also at a time when testing has vastly increased. (see my blog entry on 400,000 tests per day).

With the first cases diagnosed in February this year, how long do we have to wait for a case diagnosed today – to be an officially ‘cured’ case? Here’s some assumptions that I’ve made to help us better understand the active case load today.

  1. Assume that a newly diagnosed person is contagious for 2 weeks.
  2. Assume recovery takes around a month
  3. Assume fatality (if applicable) takes place within 1 month

The problem with assumptions is that you can always poke holes in them… but as a starting basis these seemed reasonable to me.

Using those assumptions for the US we have a total of 1.9m cases since inception; 109,000 fatalities. That leaves us with….. drum roll please….. 1.1m people RECOVERED from COVID; 388,000 coping with COVID; and 281,000 actively CONTAGIOUS.

As a sense check – at an average of 21,000 new cases per day – and following the self quarantine requirement for 14 days – that would give around 290,000 people in self quarantine.

A steadily falling level of fatalities.  Cases have not fallen as much, possibly due to more testing.
The US is #8 on the league table of current fatalities per million of population.

By the end of this week, the US will likely pass the milestone of 2 million COVID cases. You can bet that will be widely reported. But will the in excess of 1 million recovered be as widely reported? Lets see….

Time to Pump the Brakes in at least 6 states

For the following states, the level of fatalities from COVID is too high – and is trending in the wrong direction. Not enough is being done.

Masachusetts has the highest current level of fatalities per million of population per day coming in at 9.8 New cases are also very high at 134 per million of population per day. The fatality rate is stable, week over week, but that’s little consolation when the numbers are so high.

Michigan is also deteriorating. At 5.8 fatalities per million per day – and increasing rapidly, at 54% over the prior week.

Pensylvania, 4.9 fatalities per million, increasing at 8% week over week

Louisiana, 4.4 fatalities per million, increasing at over 40% week over week.

New Hampshire, 4.4 fatalities per million – increasing at about 15% week over week

Indiana 3.6 fatalities per million – increasing at about 6% week over week

See the detailed charts below.

All of these states have a current level of COVID fatality higher than NY when compared equivalently using per million of population statistics. Think about that for a second. NY was the epicenter. Now it is very definitely the case that for current data, 13 other states have a worse fatality level than NY. NY continues to improve.

A new way of looking at COVID Risk

It seems inevitable that COVID will not simply disappear so we need to turn our focus to “what is the acceptable level of COVID risk” that we are willing to live with. The fact that many states continue to reopen (or at least not roll back on some of the opening measures) means that our leaders are implicitly defining an acceptable level of risk.

Lets try to define an acceptable level of risk. My suggestion – 1 fatality per million per day. Here’s the reasoning. 1) It seems like a very small number 2) this is broadly equivalent to the US level fo FLU fatalities in a 6 month season 3) many states are already well below that level.

Using this measure, and looking at the week over week changes in fatalities – we can do some automated analysis of the states in the US. Here’s what we get:

Current Risk by State – June 2nd 2020

27 states (plus Washington DC) – are not where they need to be – by a long shot. Of these 27 – 15 are improving.

Massachusetts and Rhode Island are far in excess of any reasonable acceptable level for COVID fatalities and are worsening. By the flip side – many states including Kansas, Arkansas, Tennessee are well within an acceptable level of COVD fatalities – using the above measure.

There’s a risk in some of the current reporting that small spikes in countries with very low levels of COVID get represented as very large percentage increases. This methodolody corrects for that. A state with a fatality level of less than 1 person per million per day – will still be reported as acceptable – even if there is a short spike.

The worst states today – with an unacceptably high current COVID fatality rate AND trending in the wrong direction are:

Massachusetts, Rhode Island, Georgia, Colorado, Minnesota

Massachussets – the new Epicenter for new fatalities in COVID (per million of population)

Georgia was one of the first states to reopen quite widely. The data shows us that Georgia has not improved sufficiently, perhaps indicating that the lock down did not go for long enough, or that the Test, Trace, Isolate protocols are not sufficiently developed.

Georgia is struggling to get the COVID fatality rate under control

NY is improving rapidly – but current levels are still too high to be acceptable. Well done NY – keep it up! At that rate of progress, NY could be lower than Georgia within 2 or 3 weeks.

Great progress in NY. No longer the epicenter for new fatalities