28 April 2020

When Does COVID Kill?

DISCLAIMER: I am not a doctor or an epidemiologist. I am someone with a mathematics background who likes to analyze large data sets. I'm making a conjecture based upon what I've read about COVID-19 and this post is meant to be thought provoking, especially for those who have a better capacity to evaluate the data properly, and not as a definitively conclusion that should be acted upon. I am engaged in hypothesis generation, not in empirical proof of a hypothesis at this point.

Are COVID deaths primarily due to the interaction of select pre-existing conditions and a COVID infection?

COVID death risk is very strongly influenced by age. But, the association is made partially because it is easiest to analyze statistically.

Another way of conceptualizing the data, and I don’t know how much this has been formally analyzed, is that the true relationship could be between select pre-existing conditions and morbidity and morality risk, with age serving as proxy for the likelihood of having one or more of the risk enhancing pre-existing conditions.

Hints For A Pre-Existing Condition Theory Among The Young

In this frame, it makes a lot more sense that the young, who are normally viewed as having weaker overall immune systems than those in the prime of life, are so little affected. It also is a good fit when you dig down to who in particular among the young dies or suffers serious impairment.

For example, among the young who have died in Colorado are premature infants already in ICU, the child with severe epilepsy only responsive to CBD after whom the strain of marijuana known as Charlotte’s Web is named, a girl a couple of years behind my son in high school who was in a wheel chair and had respiratory issues, etc.

Anecdotal and small sample studies seem to show that mortality is heavily concentrated among those with a short list of pre-existing conditions (ca. 70%-90%+). It also isn’t unreasonable to suspect that many or most of those who die without any of the pre-existing conditions on the short list also have pre-existing conditions that are less common in the general population (and hence don’t make up a huge share of the total), or have pre-existing conditions that are undiagnosed, or not recognized as relevant.

Those conditions are most rare in the 5-14 age group, which has none of the congenital conditions that kill within the first five years of life (especially the first year or two), but are at an age before anything else with a genetic cause crops up. This is because this is an age group that faced strong continuing selection from infectious diseases or other causes, prior to vaccination and antibiotics, even in non-epidemic periods. Those who die in that age range overwhelmingly die without reproducing first, unlike those who die in their 30s or later, so genetic conditions that increase mortality in that age range are subject to especially strong selective pressure.

Hints For A Pre-Existing Condition Theory Among The Elderly

It is also worth noting that deaths among the elderly, despite age itself being statistically a huge risk factor, are heavily concentrated among those in nursing homes and assisted living facilities, i.e. among those who almost by definition have some impairments or serious pre-existing conditions already. Admittedly, some of the risk is from high density institutional living in what are predominantly pretty non-medically and sloppily run facilities that do not enough to prevent infections from spreading. But, the percentage of the dead in these institutions seems to be running at 35%-70% (in Colorado where I live it is 63%), but they are a much smaller share of the total elderly population. They make up about 5% of the elderly population, which in turn makes up about 15% of the total population, so about 0.75% of the total population accounts for about 35% to 70% of the dead. 

The mortality rate of the roughly 14% of the population that is over age 65 but is not in a nursing home is much, much lower. So, it is’t just the elderly who are dying, it is the elderly who have pre-existing conditions, often serious ones. Pre-existing conditions are much more common in the elderly, whose bodies are breaking down due to old age on multiple fronts. But, these conditions are not universal or uniformly distributed by any means.

How frail are people in nursing homes? 

One way to quantify that is by looking at death rates for nursing home residents pre-COVID.
In a study of elderly Americans who moved to a nursing home for their final months or years of life, 65 percent died there within one year, according to an investigation by researchers at the San Francisco VA Medical Center and the University of California, San Francisco. 
In the study, which appears in the online Early View section of the “Journal of the American Geriatrics Society,” the researchers found that length of stay before death in a nursing home was associated with differences in gender, net worth, and marital status. 
Men had shorter lengths of stay before death than women, residents with higher net worth had shorter lengths of stay than those with lower net worth, and residents who were married or otherwise partnered had shorter lengths of stay before death than those who were single, says lead author Anne Kelly, MSW, a social worker at SFVAMC. 
“It’s a matter of resources. People with more access to care and resources were able to stay in the community for longer before moving to a nursing home than those with less access,” explains Kelly. “One reason that men had shorter stays before death than women might be that women tend to outlive men, and so by the time a woman moves to a nursing home her partner is more likely to have died, whereas men are more likely to have a spouse or partner to care for them at home through the end of life.” . . .  
“One quarter of all deaths in the United States occur in nursing homes, and that figure is expected to rise to 40 percent by the year 2020,” says Smith. . . .  
Smith describes the average and median length of stay before death as “surprisingly brief.” The implication, he says, is that “we need to engage nursing home residents in planning conversations about end-of-life care and treatment preferences very soon after they are admitted. We have only a brief amount of time to address their concerns before they become seriously ill.” 
For the study, the authors analyzed data on 1,817 nursing home residents who died between 1992 and 2006. The residents were participants in the Health and Retirement Study, an ongoing nationally representative longitudinal study of health, retirement, and aging sponsored by the National Institute on Aging. 
The average age of participants when they moved to a nursing home was about 83. The average length of stay before death was 13.7 months, while the median was five months. Fifty-three percent of nursing home residents in the study died within six months. 
Men died after a median stay of three months, while women died after a median stay of eight months
From Steve Tokar, "Social Support is Key to Nursing Home Length of Stay Before Death", Patient Care (August 24, 2010) (emphasis added).

The mortality differences based on sex in COVID deaths (about 60% are men) also mirrors the sex differences in survival durations for nursing home residents.

The good news is that the proportionate impact of COVID on lost years of life may be much more modest than it is in the raw number of deaths. Lots of people who are dying of COVID had short life expectancies, in a large share of cases, median life expectancies of less than a year. Many people who die of COVID would have died of something else in the near future if they hadn't died of it.

Does This Help Explain The Geography Of COVID Outbreaks?

Western Europe and North America have been much harder hit by COVID than most of the rest of the world, especially when considering deaths per million people, a measure that is insensitive to testing rates and the extent to which testing is targeted well.

The pre-existing condition theory might help to explain this pattern.

First, the "age pyramid" in most of the rest of the world is much more bottom heavy. Most other countries have more young people and fewer older people, proportionately.

Second, many countries outside Western Europe and North America don't warehouse their sickest of the sick in nursing home-like facilities with the very least healthy of old and disabled in them. This reduces transmission and infection rates among the very vulnerable.

Third, it is also probably a reality in most of the world that most people unwell enough that they would be admitted to nursing homes in Western Europe and North America, instead die at home a few months to a year or so on average than they would have with the comparatively intensive paramedical attention that they receive in a nursing home. So, the group of people with an odds ratio of death from COVID on the order of 50 to 100 times that of the general population makes up a much, much smaller share of the total population in places outside Northern Europe, like India, Indonesia, and Africa.

Could very high exposure rates explain deaths among the quite small percentage of deaths among healthy younger people?

There are cases of seemingly healthy people dying of COVID with no apparent pre-existing conditions. But, the other common thread there in many of those cases seems to be extremely high levels of exposure. For example, ER doctors and COVID ward nurses, and public transit workers in areas with very high infection rates, seem to make up many of these cases.


Tom Bridgeland said...

This all seems quite reasonable.

My patients with covid have been among the less severely affected, as young as 37. So far as I know, none have died. It's still brutally hard on them, like the worst flu you've ever had.

Dave Barnes said...

Some statistics are garbage.

My father dies 3 weeks after moving into the old folks' home.
But, he:
1. was 95.5
2. only moved after his doctor said: "your heart is at 30%, put your affairs in order".
3. exhausted himself caring for my brain-rotted mother for 5 years.
He held out moving until he could no longer prevent it.

My mother died within 8 months of moving into "sunset living".
She died of body failure due to ever progressing brain rot.
She was 95.5.

AlanL said...

I expect we will see in the coming year or two that the covid-19 excess mortality charts show a bow wave effect, with periods of lower than normal mortality coming after the current alarming peaks.