Monday, December 9, 2019

More Life Expectancy


Top fifteen Michigan counties for life expectancy highlighted in Yellow.

Lowest fifteen Michigan counties for life expectancy highlighted in Pink.

List sorted by per-capita-income, a common marker for "poverty".

Wayne County, home of Detroit, has the very lowest life expectancy and has a per-capita-income 37% higher than Lake County.

Leelanau County has the highest life expectancy and has a per-capita-income 14% lower than Oakland County. A reader might comment that Leelanau County is a small county and therefore not a good sample.

Dropping down to the second highest life expectancy, Ottawa County, which has a per-capita-income that is 30% lower than Oakland County. Ottawa County is not a micro county in terms of population.

There are two "Top Fifteen" counties that have lower per-capita-incomes than three "Bottom Fifteen" counties.

Shared Underlying Causality
Picture in your head a foundation made of cinder blocks. Two of the walls were laid by one mason and the other two were laid by a second mason. A sill plate is placed on top of the wall. Two of the sill plates came from Menards and two of the sill plates came from Home Depot.

The height of the tops of the sill plates don't match up. Was it the actual dimension of the lumber or the skill of the masons?

Let's stretch the analogy. Suppose two foundations are built in the same fashion. One foundation is only two courses high while the other is 12 courses high. Which foundation is likely to be the most out-of-level?

Economists have a droll description for many of the chronically unemployed. They have "Employment resistant personalities". That is, they don't show up for work, they get in fights with bosses, other employees and customers. They don't accept authority. Are either incapable or unwilling to follow written instructions.

If you went to a nurse and asked them, "Who will have the less favorable outcome: A poor person or a person who does not comply with treatment?" Wanna bet who they will pick?

The point of this tangent is that there is much overlap between the traits of the person who has "Employment resistant personalities" and who will be non-compliant with medical treatment.

Stretching the foundation analogy to the limit: Life outcomes, both economic and for health, are cumulative. One bucket of fried chicken or one cigarette or shaking salt on your food once, missing one doctor's appointment will probably not kill you. Telling the boss he is full of BS ONCE probably won't get you fired.  But brick-by-brick the outcomes diverge between counties with different cultures.

Non-compliant personalities and lack of trust in authorities could be a shared, underlying cause for both poor economics and poor health outcomes.

12 comments:

  1. I pick and chose the authorities I trust. Politicians? Precious little. My docs, or my mechanic? Considerably more. I do try to be that "educated patient" or "educated consumer" as much as possible, however, as the old man said, "You can't beat a man at his trade." I research, but I ask them questions so I can understand their reasoning behind a suggestion. I then make a decision.

    A great example is my neurologist. He is dead set against allowing my chiropractor to perform adjustments on my neck. He's a great doc, and he explained why in depth. Good enough. I did my research, and then talked to my chiropractor about the subject. Weighing everything I read and was told, I made the decision to let the chiropractor do her thing on the neck and haven't breathed a word of that decision to the neurologist. Yes, something could possibly go wrong with an adjustment and I could die. The risk is small, and one I'm willing to accept in order to continue turning my head without pain.

    The difference between me and the folks you're addressing is that they don't think their decisions through, they simply react in a knee-jerk manner to some stimulus (the boss) or fail to react to one (medical treatments). When it goes badly for them, the wailing begins. For them, my sympathy is limited.

    I noticed a good while back that these folks seem to cluster geographically. My concept is this: Humans learn by example. We learn behaviors from our family, friends and peers. What we do reinforces those behaviors in others near us. A trend toward poor behaviors such as smoking, or poor dietary choices, or bad attitudes toward "da Man", will, over generations, reinforce itself to the point you can see it in demographic numbers, whether lifespan or crime rates or both. Probably others if we look into it more.

    ReplyDelete
    Replies
    1. I applaud your approach.

      My lament is against the automatic, knee-jerk opposition to anything an authority says.

      Don't want a vaccination? Study it up. Make your choice. If you are too busy or lazy to study, go with what the authorities recommend. If you have a bad reaction...well, you now have more information. Don't get the vaccination.

      I enjoy the occasional snort of whiskey. Is it good for me? Probably not. But it is a risk I acknowledge and accept.

      However, I avoid french fries, potato chips and salad dressing like the plague. To me it is a fair trade.

      Delete
  2. Is there a correlation between rural vs urban locales?

    ReplyDelete
    Replies
    1. Looking at the data, Detroit has a "crater" profile. Life expectancy is lowest in the center of the crater, highest where mobile, insured workers live in the first ring of counties and then to base-rate in more rural areas.

      First ring counties have many people with those good habits and easy/quick access to hospitals.

      An alternative explanation, one that may be a player, is that a high density of privately insured patients pay for enough lab and imaging capacity to expedite quick turn-around. All patients benefit. Go to most rural counties and you lose that density.

      It should be obvious that the loss of diagnostic speed does not out-weigh the compliance issue.

      Delete
  3. What races are represented in those countries? Black and White life expectancies are different enough that race might be a factor. How many churches per capita, how many bars and how many meth labs per capita?

    Income is a good predictor because it correlates well to IQ, which is the best single predictor of most things.

    ReplyDelete
    Replies
    1. Wayne, Genessee, and St Joseph have significant African-American populations. Clare, Lake...are Wonder Bread white.

      The single most authoritative source I have regarding the real-world usefulness of a high IQ claimed that the only thing a high IQ was useful was to predict the successful completion of the first year (or first 25%) of college.

      That said, it is difficult to complete one's second year or graduate if you flunk-out in your first year.

      If you have a low "IQ" and successfully complete your first year of college, then you found coping mechanisms and work-arounds. For example, you wrote down every word the professor used that you did not understand and then you looked them up as soon as you were out of the lecture hall.

      I tried to avoid the race angle. There are groups of white people who foster a "non-compliant culture".

      There are African-Americans...often from the Caribbean or Africa or from military families, who resisted the siren song of victimhood and do not foster "non-compliant personalities".

      Sadly, the race hustlers sell the idea that white CIA agents created HIV and Sickle Cell and drugs to keep the Black Man down. The race hustlers, more than anybody else, poison the rational, reasoned consideration of what "authorities" are saying.

      Delete
    2. IQ is so correlated to everything that it is confounded with everything, and often linearly dependent as well. Run a regression with anything and IQ, and you will find that IQ adds little explanatory power. Remove the other explanatory variable, and you will find that IQ explains most of the variance which the other explained. IQ remains the best single factor predictor of almost everything related to human action or achievement.

      Delete
    3. Wayne, 78.07 years/$22.4k=3.485 years/$k, Genovese, 78.33 years/$22.3k=3.513 years/$k, and St. Joseph, 79.43/$20.6k=3.856 years/$k.

      Clare, 78.99/$18.1k=4.364 years/$k, Lake, 79.38/$16.2k=4.900 years/$k.

      That's only a few data points, but it's interesting. The White counties get slightly more years of life expectancy per dollar of income than the predominantly Black counties. It's almost an additional year per thousand dollars. If that holds up after correcting for things like crime rate, it's probably because Blacks die at a higher rate from things like hypertension. If you know the Black and White percentages in each county, you could apply the know death rates for each race and give that hypothesis a rough test.

      I'd look at things like drug busts in the counties, number of liquor licenses, number of churches, to try to explain the variance we're seeing.

      I'm not suggesting you should do research for me, I'm trying to suggest places you might look to satisfy your curiosity. Don't feel that you need to report back - that's not what I expect.

      Delete
    4. I forgot to mention the 800 pound gorilla of life expectancy: what's the infant mortality rate in those counties?

      Delete
  4. Hello all,
    Since I don't know the distribution of the underlying population (parametric vs. nonparametric) I used your life expectancy groups in ordinal fashion (1 longest, 3 shortest) and ran a Kruskal-Wallis analysis.
    Unsurprisingly, longest life group has significantly higher income (P < 0.0001).
    Correlation (using Spearman's rho nonparametric test) is -0.89 (negative since long life is 1 and short life is 3) p < 0.001. My math shows if we square r we get 0.7921. so 79% of the life expectancy model is explained by income.
    In terms of population p = 0.065, fails the statistical test for a difference, practically YMMV.
    Correlation between population and LE = -0.05: essentially random (population size explains less than 2/100th % (0.0025) of the model.
    Interesting stuff, thanks for sharing!
    Mongo

    ReplyDelete
    Replies
    1. Wow, Mr Mongo:

      Thanks for digging into that.

      Your analysis is a good springboard for how a politically motivated statistician can jimmy the numbers. For example, if you wanted to force the population correlation you could have selected the top 13 from each end. That would have dropped off Genessee County (Flint) and Wayne County (Detroit) off the list.

      In light of the Socialist's plans for health insurance, I wish I had the motivation and access to look at the % Medicaid hypothesis. Anecdotal data suggests that Medicaid patients are more likely to clog Emergency Rooms with trivial complaints, bog down the triage process, saturate the 9-1-1 system with requests for ambulance rides the day their "check" comes in.

      My brother the firefighter often wished that there were no bars or liquor stores within a half mile of hospital emergency rooms. He called the Lansing FD the $500 Taxicab service.

      If Socialized medicine becomes the law of the land, what will prevent every county in the US from sliding into Wayne County's dysfunctions?

      Delete
    2. "If Socialized medicine becomes the law of the land, what will prevent every county in the US from sliding into Wayne County's dysfunctions?"

      (1) Getting home from the ER, several counties over. (I', looking at YOU, Upper Peninsula, or Montana, or South Dakota, or...)
      (2) retiring docs/PAs/NPs. If my license lapses, how will they draft me to work in Bugtussle (Or, for that matter, Flint)?

      Delete

Readers who are willing to comment make this a better blog. Civil dialog is a valuable thing.