I wrote “Visualization of the COVID-19 infection rates” with two goals: to warn people about the upcoming pandemic and to provide insight into that pandemic.

The US took precautions within a couple months, and the length and intensity of the precautions has surprised me. Even four months later, individuals generally believe they should take actions to limit the spread of COVID-19. This includes wearing a mask and working remotely if possible.

But are these precautions justified? There’s no harm done if everyone gets an benign virus. Do the data justify mandating wearing masks and closing schools? Let’s look.

The hospital data from New York City (NYC) indicates that they are past the most intense part of the infection:

New COVID-19 cases/hospitalizations/deaths are down by 30–50$\times$ since the peak. By this measure, NYC has moved “flattened the curve” and are seeing minimal new cases, hospitalizations, and deaths.

However, the lockdowns are still continuing. The subway rides have been down below normal weekend levels for nearly 5 months:

How necessary are these lockdowns? Let’s look at some data to find out.

Case study: Sweden

Sweden has a different approach; the government made strong recommendations to the elderly to “limit close contact with other people” and

… [are] encourag[ing] citizens to use common sense, work from home if possible, and not gather in crowds over 50. Primary schools are open, as are bars and restaurants, with images showing people enjoying drinks and crowding streets.

—”Sweden Sticks With Controversial COVID-19 Approach”.

CNBC reports that “[Sweden] did not go into lockdown, instead issuing recommendations about social distancing and working from home while allowing many schools and businesses to stay open.”

Obviously, not having a lockdown has significant benefits: kids can see their friends at schools, restaurants/bars are still serving food and don’t have to lay people off, etc. In fact, the Sweden economy has performed well, at least when compared the US. Here’s a table on the annualized GDP growth rate:

Time Sweden US
2020, Q1 +0.1% -5.0%
2020, Q2 -8.6% -32.9%

There have even been stories written about how the Sweden economy has performed better than the economies of neighboring countries.

This must have come at a cost, right? Sure, they might have been able to keep their schools open and their economy functioning, but certainly more people contracted COVID-19? Absolutely:

But the number of infections is meaningless. No one cares if everyone contracts a harmless disease. Let’s look how harmful COVID is with the deaths attributed to COVID:

Clearly, far more elderly people have deceased from COVID than younger people when normalized by the population in that age group. The data from Sweden is high resolution – they specify the number people aged between (say) 75 and 80 years old that have died. The data from NYC are unfortunately too coarse to do any detailed comparisons; however, the general trend is clear: NYC and Sweden have the approximately the same number of deaths per population.

That’s right: NYC and Sweden have (approximately) the same number of deaths per population, even after normalizing for age. There’s no obvious difference as with the case count.

Maybe NYC is an outlier because of their population density.1 Let’s make the same plot for the US instead:

About 0.5% of the US population over 85 has deceased due to COVID. For the 40 year old, 0.005% of the US population has deceased due to COVID. For context, the US suicide rate is 150 per million or 0.015% for the population aged 35 to 44.

Let’s look at the various death rates for the US, and see how the number of deaths from COVID compare for each age group. Let’s plot these death counts relative to the number of COVID deaths:

A value of 20 on this chart means the death rate from (say) suicide is 20× greater than the death rate COVID-19 for that age group. I defined “death rate” for suicide/etc as low as it can be, n_dead / n_people. For COVID-19, death rate is defined as n_dead / n_infected.

This chart is a little misleading; this compares the deaths in 2015 to the number of COVID-19 deaths, not the deaths that occurred during the COVID-19 lockdowns from suicide/drugs/etc. I hypothesize that the number of suicides and drug overdoses have increased during the lockdowns. The suicide rate in 2015 is 20× the death rate of COVID-19 for the population aged 15–24; I suspect the suicide rate has increased, especially because the CDC director reports that deaths from suicide/drug overdoses are “far greater” than COVID deaths for high school aged students:

But there has been another cost that we’ve seen, particularly in high schools. We’re seeing, sadly, far greater suicides now than we are deaths from COVID. We’re seeing far greater deaths from drug overdose that are above excess that we had as background than we are seeing the deaths from COVID.

Robert Redfield, July 14th, 2020

COVID-19 lockdowns come with both economic costs and mental health costs. Let’s look at some data on COVID-19 and children.

COVID-19 and children

Iceland has performed a contact tracing study that studies infection and traces it back to it’s source, then recurses. Iceland tested 6% of their population in their contact tracing study before April 4th.2 Of the people randomly sampled, none of the children under 10 tested positive for COVID-19 despite a 0.8% positive rate for people older than 10 years. They also found that the infection probability increased (gradually) with age for the population under 20 years old. Iceland’s study included genetic tracing to determine the index cases, but unfortunately did not distinguish “school” and “work.”

Preliminary evidence from the NIH suggests that children are more likely to be missing the receptor for COVID-19, specifically because children are more susceptible to allergic asthma. The NIH is further funding this study to examine correlation the relevant gene and infection, and also COVID-19 in children:

One interesting feature of this novel coronavirus pandemic is that very few children have become sick with COVID-19 compared to adults. Is this because children are resistant to infection with SARS-CoV-2, or because they are infected but do not develop symptoms? The HEROS study will help us begin to answer these and other key questions.

Anthony S. Fauci, M.D., NIAID Director

Spreading without any symptoms, asymptomatic spread is rare; spreading before symptoms develop is “is believed to be far more common than asymptomatic spread” (source).


I presented data that provides evidence to support these hypotheses:

  • Elderly people have a significantly higher risk of contracting and dying from COVID-19.
  • The death rate for the population under 20 is minimal relative to suicide and drug overdose death rates.
  • Sweden and the US have similar death rates despite drastic differences in their public policy approach.

As an aside, here’s data from Minnesota on the age of various patient classes:

Population Median age
All MN residents 38.2
People who positive
for COVID (patients)
Patients not in hospital 34
Patients in hospital 59
Patients in ICU 61
Patients who die 83

This means that half the people in the ICU are over the age of 61, and half of the COVID-19 hospitalizations are older than 59.

Data sources

  1. NYC has about twice the population density of Stockholm and about 5× the population. 

  2. “Spead of SARS-CoV-2 in the Icelandic Population.” Gudbjartsson et. al. New England Journal of Medicine. DOI: 10.1056/NEJMoa2006100