"You can't fix by analysis what you bungled by design." -Light, Singer, and Willett (page v)

Over the past couple of weeks I've been engineering a spreadsheet to automatically walk through a series of calculations on the entire Our World in Data dataset in order to explore the progression of the pandemic, but in particular with respect to vaccination. I am currently working with the dataset through October 20, 2021.

I'm looking for it...

While the project takes a lot of grunt hours, I wanted to share some preliminary observations. Understand that these observations are very basic at the moment. I am not removing confounders, accounting for population or immunity changes, or trying to project epidemiological models, the totality of which is essentially impossible.

I am also not performing the absurd task of filtering between nations based on which vaccines were used. The over 200 nations in the OWID have very little overlap in selection of vaccines and protocols. It's almost as if nobody wanted clean enough data to study afterward. And who knows how we might correct for motivated data engineering by interested parties. You never know what kinds of schemes Big Hydroxychloroquine might hatch.

Case and Mortality Numbers Pre/Post Vaccination

The following is the progression of the number of COVID-19 cases per day by continent since vaccination began, measured on a per nation basis up to the start date of the vaccination program in that nation.

Cases have increased 26.1% in Africa.
Cases have increased 169.6% in Asia.
Cases have increased 690% in Australia
Cases have increased 56.6% in Europe.
Cases have increased 67.6% in North America.
Cases have increased 59.9% in South America.
Cases have increased 76.3% worldwide.

Out of 217 nations reporting vaccination statistics in OWID, 189 have seen an increase in per day cases, while 28 have seen a decrease. Most of the nations that have seen decreases include nations with few if any cases at all during the pandemic (in total, I have not yet normalized per million). Never mind that a few of the reporting entities are not independent nations. I may or may not sort that out on a priority basis. These include the following 17 nations with very few cases among them.

Burkina Faso (fewer than 15,000)
Chad (just over 5,000)
Comoros (fewer than 5,000)
Djibouti (fewer than 14,000)
Hong Kong (fewer than 13,000)
Kiribati (2 cases)
Liechtenstein (fewer than 4,000)
Madagascar (fewer than 45,000)
Mali (fewer than 16,000) cases are down 0.5%
Niger (just over 6,000) cases
Somoa (3 cases)
San Marino (just over 5,000)
Solomon Islands (20 cases)
South Sudan (fewer than 13,000)
Sudan (fewer than 40,000)
Tajikistan (fewer than 18,000)
Vanuatu (4 cases)

Edit: Apologies. I found an error in a formula, so the originally published numbers below have changed.

The following is the progression of the number of COVID-19 deaths per day by continent since vaccination began, measured on a per nation basis up to the start date of the vaccination program in that nation.

Deaths have increased 27.1% in Africa.
Deaths have increased 122.8% in Asia.
Deaths have increased 155.3% in Australia
Deaths have increased 30.2% in Europe.
Deaths have increased 37.3% in North America.
Deaths have increased 55.7% in South America.
Deaths have increased 46.3% worldwide.

Addendum: Out of 217 nations reporting vaccination statistics in OWID, 183 (186) have seen an increase in per day COVID-19 deaths, while 34 (31) have seen a decrease. In parentheses are the numbers that include projected deaths of active cases using the spot case fatality rated (smoothed on an 18-day case-to-death lag). The nation list is highly similar.

The next articles I write making use of this data will likely be more interesting and statistically sound, but prima facie these numbers do not make a compelling case for high vaccine efficacy rates.

No, I do not trust all of the data. There are those who argue that in many nations, a portion of the data is difficult to collect. But even when there are problems in data collection, there is more often consistency than accuracy, so taking ratios across time periods with enough data usually is a good first order starting point. It is noteworthy that China is nearly singular among large economy nations in showing numbers that point to substantial vaccine efficacy, though nearly all their cases reports still stem from the early days of 2020. Saudi Arabia shows improvement as well, but reports fewer than 9,000 deaths during the entire pandemic. Belgium, Spain, Switzerland, and the UK show declines in deaths per day.

It is noteworthy that epidemiological models generally predict declines in spread 20 months into a pandemic, regardless of the transmission speed of a viral outbreak. Muller's ratchet predicts a steeper decline in mortality.

Addendum: The day after this article was published, The Expose published an article using analysis from the same OWID database showing that the highest COVID-19 death rates are generally in the most vaccinated countries.

Share

Rounding the Earth Newsletter

Systemic COVID-19 Vaccine Efficacy Basics