Does this COVID life expectancy study use bad statistics?

I signed up for this forum because I was trying to estimate the age- specific change in life expectancy for people who have reported cases of COVID-19. I'm not very good with statistics, but after a few hours of research, I came up with a rudimentary model that seemed like a decent estimate. After coming up with my own results, I did a search for other studies which tried to answer the same question, and discovered this published journal article which came up with a result completely different from my own.

Here is the article, and my own spreadsheet is attached (sources included).
I used assumptions that death rate is constant during a reported age range, that a person can only become infected once, that all people are equally likely to become infected regardless of preexisting conditions, and that death rates return to normal after recovery for survivors.

Basically, my model is the same as asking, if 100% of people became infected, how would life expectancy change? In no age group did the life expectancy decrease by more than 1.7 years. However, the article linked assumes that in some countries, the overall life expectancy could decrease by as much as 6 years if 70% of people became infected. So did I make a mistake or did they?

P.S. - I had to convert my spreadsheet to CSV format so that this forum would let me upload it. To display it properly, just change the filename extension back to .csv and it should open in Excel or any spreadsheet program.



No cake for spunky
You are brave. I have spent 14 years working in statistics and still doubt much of what I do. There are a lot of complexities you don't know when you only spend a limited time on it.

They say this we built a discrete-time microsimulation which does not sound like a statistical method. Is this statistics or machine learning? Cox proportional hazard is probably the classical statistical approach to this issue.