Lockdown efficacy in America

Is the Lockdown Working for America? A data-driven approach

Let’s explore some data to find out how America has fared so far with the lockdown measures.

Rajesh Mukherjee
6 min readApr 27, 2020

--

TL;DR:

  • US is still far from “flattening the curve” of confirmed cases. Although daily new case numbers are plateauing, the numbers are not declining yet like Spain or Italy
  • Death Rate in US is lower than Italy and Spain in terms, but the curve is yet to stabilize
  • Washington state has done an exceptional job of controlling the spread
  • The virus seems to be less deadly than Ebola and SARS

Background

It’s been almost a month since the lockdown so it is a good time to understand the impact of lockdown and social-distancing. For the purpose of our discussion, we will measure impact in two dimensions — the spread and fatality.

First, we will explore how much of the curve we have flattened through lockdown and social-distancing. The absolute number of confirmed cases will flatten out when the daily new cases go down to zero.

Next, we will try to explore how Death Rates across different states and nations that went in lockdown.

Let’s dive right in.

How fast are the cases growing every day?

Fig 1. Daily New Cases by State — Source jarinsea@

Here is the good news. Overall most growth in “daily new cases” seems to be plateauing. Specifically, the daily cases have plateaued for New York at roughly 7,000 cases per day, followed by New Jersey and California. Whereas the daily cases have declined for Washington by 80% from 500 to 100 cases per day.

Daily case numbers in Michigan has gone down by 60% from 600ish from1500 cases a day.

Fig 2. Total Confirmed Cases (Cumulative) — copyright jarinsea@

This is the bad news is the absolute number of cases is over 720,000 now and is still growing.

How does the daily case numbers compare to other countries?

Fig 3. Daily Cases Global — source jarinsea@

Unfortunately, US has 30,000 cases per day which is 3x higher than its successor Spain. Although US has reached its peak, the curve is not declining as it did for Spain or Italy. The number of cases in Italy started declining after 15 days from the lockdown start date of March 9th, probably because Italy had stricter lockdown enforcement compared to the US. I imagine US will end up with the slowest trajectory compared to Spain or Italy given how the current things are.

Let’s look at the daily cases after normalizing it with the country’s population.

Fig 4. Daily cases normalized by Population — source jarinsea@

Spain seems to be worst when the population is taken into account. The peak number of daily cases per capita per day in Italy and US is the same which is about 100 cases per capita per day. The numbers in China look quite questionable

What is the current Death Rate in US from Covid-19?

Fig 5. US Deaths per Confirmed Cases- 04/18/20 — copyright: Jarinsea@

From the above chart, we can conclude that the current Death Rate (Total Deaths / Total Confirmed Cases) is around 5.2%. This number is mostly inflated due to the low number of testing America has performed so far.

What are the worst affected States?

Fig 6. Death Rate By State copyright: Jarinsea@

Northern Mariana Islands state has the highest death rate of 15%! but that might be driven by the low number of cases — 4 per day. Michigan, New York, and Connecticut seem to lead the Death Rate — 7.2%, 7%, and 6% respectively. Let’s compare the rates among states with high fatalities.

The Death Rate varies from 3.75% to 7.5% in the states with the highest fatalities. Michigan seems to be the worst place to be right now.

Although California has done well in following the lockdown orders, the curve is bending up now. Louisiana delayed the peak by almost 15 days, so that’s a win.

What about Death Rates in other nations?

Fig 8. Death rate top Countries — source jarinsea@

Italy has the highest Death Rate so far — 13% whereas the Death Rate in America is about 5.5%. Italy has an aging population and most people smokes, which could have attributed to the disproportionately high Death Rate.

So how likely a person might die of COVID-19?

Depends on which state that person lives.

Fig 9. Deaths per Capita by State — Source jarinsea@

Looks like a person living in New York has around 0.8% chance of death whereas someone living in California has less than 0.2% chances of death from the virus.

Comparison with other deadly Viruses

COVID-19 can be 400x more dangerous than the common flu for someone living in New York. SARS -COVID had a Mortality Rate (Total Deaths/ Total Population) of 9.6% whereas the African Marburg Virus had a Mortality Rate of 25%. Whereas influenza has a Mortality Rate of 0.02%. The states with the highest Mortality Rates from the virus are New York, New Jersey, and Louisiana with 8%, 5%, and 4% rates respectively.

In terms of the Fatality Rate (CFR) (Total Deaths/ Total Diagnosed Cases), meaning death among the total diagnosed population, COVID-19 is not as deadly as Ebola (CFR-90%) but deadlier than HongKong flu (< 1%), which killed over a million people. We are currently not sure what is the true CFR for COVID-19, but it is estimated to be around 3.8%.

However, COVID-19 more infectious than Ebola and SARS where a patient infected 2.53 and 2.75 persons respectively, but a COVID-19 patients infect 3.5 persons.

So what?

Overall the lockdown has had a positive impact in terms of slowing the spread but the victory is still far. The two important metrics to remember are :

  1. Daily new cases
  2. Total confirmed cases

When #1 goes down to zero # 2 will flatten, and then we can declare victory. We have done a decent job so far, however, we should not get complacent. A really good article about “When should we reopen?” is published in the NY times please check it out here. The US government has also shared an estimated number of days to reopen. We should all keep the fight on till then.

Thanks for reading and please be safe.

Caveats

Please note that the analysis doesn’t take into account (1) underlying health conditions, (2) basic reproduction rate (R0) (3) Health Care facilities (4) virus mutations (5)biases in data collection.

Data Collection:

All data is pulled from John Hopkins open dataset available on Github.

https://github.com/CSSEGISandData/COVID-19/tree/master/csse_covid_19_data/csse_covid_19_time_series

--

--

Rajesh Mukherjee

Tech professional in the US. Background in data and product.