Tuesday, March 31, 2020

Guest Post by "Louisiana": "Time to Assess Practical Risk"

[This guest post - the "high agency" approach to surviving this pandemic - is by our long time correspondent Louisiana. You might remember that he won the 2018 Prediction Contest, so read carefully. Also, be sure to read our previous posts and guest posts on coronavirus. -CBS]

It is time to assess practical risk. The economy will open back up soon, and the virus will not be contained (too late). We need to prepare to be infected and know the risk if it happens.

I took it upon myself to build a spreadsheet model comparing Covid risk to background risk of dying by age using SSA actuarial tables.

I direct everyone to the very solid statistics being kept by the Center for Evidence Based Medicine at Oxford. The current best estimate overall Case Fatality Rate for CV19 is 0.51%. The current best estimate overall Infection Fatality Rate is conservatively about 0.2% (probably lower). That is, about half of infections are asymptomatic or never severe enough to be tested. Thus the actual fatality risk is about 40% of the case fatality risk.

While the Chinese CFR is inflated as explained by the link, we can derive relative risks by age, comorbidity, and sex. For sex, best to look at Italian data since similar percentages of men and women smoke there, unlike in China where many more men smoke.

If you make a copy of the spreadsheet, you can input your own data. For myself, as a 40 year old male with no comorbidities, the best estimate is that I have a 1 in 4266 chance of dying if infected with Covid-19. The overall risk of dying for a man my age, according to the SSA life table, is 1 in 413.

Getting infected with Covid, then raises my overall death risk by 9.7%. That is, my death rate if infected over the next year is equivalent to a 42 year old instead of a 40 year old. This is not the sort of risk, for me, that is worth the lockdown measures being taken.

In fact, it’s arguably not worth it for anyone without comorbidities. For almost everyone, the relative risk is about 10%. For example, for a healthy 70-year-old, the odds of dying if infected are 1 in 416. The odds of dying in general are 1 in 43. Only a 10.4% elevated risk.

What if you have comorbidities?

The news isn’t that bad. Take a 70 year old male with some health problems. You would have a 1 in 100 chance of dying from covid compared to the general risk of 1 in 43 from social security data. However, if you have these comorbidities, your base risk of dying anyway is elevated and the SSA actuarial data is not accurate for you. At worst, you’re talking about a ~40% increase in death probability, more likely about 20%. Again, not worth shutting down the entire economy.

CV-19 seems to be killing people by temporarily knocking out some percentage of the lung's capacity to deliver oxygen. Thus, those with more "buffer" - younger, less weight, more trained cardiovascular systems - are much less likely to die. If you don't have enough buffer, you end up on oxygen or a ventilator, the latter of which makes you more vulnerable to bacterial infections.

No one is helpless. People can take action right now to fight this. Start exercising, hard. Aerobic capacity starts to improve almost immediately, and within a week you will have built a real buffer that could save your life. Second, if overweight start losing weight by eating much less food. All weight loss, regardless of the type of diet, involves burning body fat and putting the body into ketosis at least for part of the day, which boosts immune function.

The weight loss itself means less tissue for your infected lungs and heart to support, and less likely to end up on a vent. This is not the time to eat a bunch of comfort food.


Allan Folz said...

Thanks, Louisiana. Great research and great spreadsheet.

I'll offer that rather than using one's actual chronological age, one could use an estimated age based on your personal health and fitness level.

Think of it like a negative co-morbidity. Sadly, we won't ever get research breaking-out atypically healthy and fit people, but we can infer it a certain amount given how strongly age-correlated WuFlu is. If I compare myself to co-workers or to the rando's at the grocery store, I don't feel it's at all unreasonable to consider myself 10 years younger than my chronological age.

Cheers. @pdxsag

Anonymous said...

The bigger issue that you all / most people are missing is that hospitals are no different than bars. There is no antibody/vaccine for SARS-COV-2, so doctors are basically making cocktails for patients to try to treat side effects of the Covid-19 infection. If only a few people come in, then a cocktail can be carefully made for each person (ie based on a patient's health history, blood oxygen saturation, etc). If it gets busy, then only pitcher drinks can be made and key ingredients will be rationed / reduced until these run out. If everyone shows up all at once, then nobody can get in, let alone what they need when they need it, and supplies won't be restocked until after a shut down of one form or another. What makes this situation worse is that all healthcare professionals (not just those working in ICUs or emergency divisions) are running out of the PPE required to do their jobs. When doctors are unable to work, patients won't be able to access care in the event of an emergency, a critical situation, or even for routine visits. This is what is happening now all around the globe.

I would encourage you all to give some thought to this and to reassess the practical risks (as @Louisiana aptly titled this post).

Anonymous said...

Update from the OP:

I hope I'm wrong, but appears Italy is going to be the better model for the US, because the doubling rate of deaths is not slowing as I expected.

My earlier estimates were based on something in between Germany and Italy. Our higher incidence of obesity and metabolic syndromes, along with a more Italian-like extroverted culture in many of our subpopulations, seems to offset any benefit from being younger on average. Early numbers from the American South look grim, but only among those with these known comorbidities, which are more common here. The good news from Italy is that deaths are slowing and will top out below 20,000 if the peak in daily deaths has been reached.

However, at our current rate of doubling this implies around 120,000 deaths total in the US. This is about 2-3x as bad as a bad flu season. My spreadsheet model of personal risk remains valid and conservative:


If the average IFR is 0.2% as this model assumes and half of Americans get infected, that's a total of about 350,000 deaths. The 0.2% assumed IFR is likely overestimated. For most people under 70, this is statistically not a dire health risk compared to other causes of death.

Is this worth the economic damage being done? No. Could we get a similar result by wearing masks in public and asking the vulnerable to self-quarantine? Most likely yes. Are the models predicting the velocity of the disease accurately? No, it appears the peak of the curve would not have been as bad as predicted, likely because the early assumptions about the fatality rate were inflated.