What Factors Lead to Mortality Rate in the U.S. in 2020 During the Pandemic?

Written by David Hatfield (@DavidC_Hatfield, ), May 16th, 2021 (updated Nov 7th)

Click here to open the 2021 Addendum.

Contents:


Disclaimer

I am not a profressional statistician and I did this report on my free time. Some of the math may be a little "fuzzy", but I am confident the gist of the report is fairly sound and has merit.

Introduction

It is irritating when people say that lockdowns and mask mandates aren't a good idea and haven't worked and when they base their statements on insufficient data or flawed analyses. It's especially vexing when this comes from people such as Joe Rogan and Kim Iversen who consider themselves to be more to the left. So let's take a look at the numbers and see what pops out. (Many studies probably already exist, but I'd rather play with numbers than read reports. ;-) )

COVID Deaths Vs. Excess Deaths

Kim Iversen used COVID death count per million, among other things, to conclude that California didn't have enough fewer deaths than Florida to justify the much more restrictive measures that limited personal freedom and hurt the economy. Or something to this effect. However, there are potential problems with this data and logic. It seems likely that COVID death counts aren't highly accurate and weren't necessarily done in a consistent manner from state to state. Moreover, there were presumably many factors in addition to governmental regulations that affected mortality rates. To address the first point, some people prefer using excess death counts. And to look at the second point, we'll consider twelve different factors using a multivariate analysis.

Excess death count is the number of deaths above the average. According to the CDC, there were 500,000 excess deaths in 2020. By contrast, the CDC reported 360,000 people who died with COVID in 2020. Although there is some variation in death rate from year to year, the variation is usually a lot less than half a million. Therefore, the excess deaths in 2020 presumably were mostly a result of the pandemic. The excess death count is a lot greater than the COVID count. Part of this gap is surely because some people died with COVID but were not tested. There were also deaths related to the pandemic that were not directly caused by COVID, such as the deaths of people who did not or could not receive medical care who would have in a normal year and suicides due to the increase in stress and isolation. But it can be noted that some of the COVID deaths are not included in the excess death count, for instance people who died with COVID whose primary cause of death was something else or who would have died later in the year of something else. All things considered, when quantifying the effect of COVID on mortality rate, excess death count is a good addition to COVID death count and is arguably better, expecially when comparing one state to another.

Table 1 displays COVID deaths versus excess deaths by state:

Table 1. COVID Deaths Vs Excess Deaths By State in 2020

Rank State COVID Deaths Per Mil Change State Excess Deaths Per Mil Predicted Residual
1 New Jersey 1974 2 Mississippi 2377 2103 274
2 New York 1935 New York 2340 2301 39
3 North Dakota 1825 3 New Jersey 2296 1587 709
4 South Dakota 1805 3 Arizona 2179 1634 545
5 Rhode Island 1682 24 Louisiana 2158 1759 399
6 Connecticut 1671 12 North Dakota 2151 1593 558
7 Mississippi 1635 6 South Dakota 2067 1610 457
8 Louisiana 1468 3 Alabama 1953 1834 119
9 Iowa 1462 11 New Mexico 1913 1752 161
10 Massachusetts 1431 27 Tennessee 1845 1752 93
11 Pennsylvania 1346 23 Arkansas 1809 1858 -49
12 Indiana 1340 4 South Carolina 1801 1860 -59
13 Illinois 1275 4 Michigan 1750 1605 145
14 Alabama 1274 6 Missouri 1731 1808 -77
15 Arkansas 1255 4 Washington D.C. 1724 1842 -118
16 Washington D.C. 1250 1 Indiana 1710 1528 182
17 New Mexico 1248 8 Illinois 1685 1503 182
18 Missouri 1236 4 Connecticut 1628 1118 510
19 Oklahoma 1204 8 Nevada 1627 1661 -34
20 Ohio 1195 5 Iowa 1603 1351 252
21 Michigan 1188 8 Wyoming 1595 1799 -204
22 Arizona 1166 18 Texas 1594 1443 151
23 Kansas 1128 10 Montana 1565 1223 342
24 Nebraska 1114 14 Georgia 1559 1773 -214
25 Montana 1092 2 Ohio 1550 1524 26
26 Texas 1086 4 Florida 1531 1170 361
27 Tennessee 1070 17 Oklahoma 1530 1982 -452
28 Maryland 1066 3 Delaware 1505 1584 -79
29 Nevada 1049 10 Rhode Island 1497 1268 229
30 Wisconsin 1032 Wisconsin 1493 1325 168
31 South Carolina 1001 19 Maryland 1491 1692 -201
32 Minnesota 993 7 Kentucky 1490 1547 -57
33 Florida 970 7 Kansas 1489 1574 -85
34 Kentucky 963 2 Pennsylvania 1433 1729 -296
35 Georgia 915 11 Colorado 1366 1249 117
36 Delaware 866 8 West Virginia 1279 1523 -244
37 Colorado 837 2 Massachusetts 1259 1117 142
38 Idaho 780 2 Nebraska 1257 1673 -416
39 California 758 3 Minnesota 1241 1372 -131
40 West Virginia 752 4 Idaho 1226 1614 -388
41 Virginia 675 Virginia 1200 1547 -347
42 Wyoming 663 21 California 1150 1208 -58
43 North Carolina 521 1 Utah 917 1353 -436
44 New Hampshire 506 1 North Carolina 823 1635 -812
45 Utah 467 2 New Hampshire 780 1023 -243
46 Washington 454 1 Oregon 762 757 5
47 Oregon 340 1 Washington 683 586 97
48 Maine 222 Maine 667 1301 -634
49 Hawaii 200 1 Vermont 305 485 -180
50 Vermont 151 1 Hawaii 287 739 -452

  • State - State colored to differentiate Red states (Republican governor & won by Donald Trump in 2020), Blue states (Democratic governor & won by Joe Biden), and Purple states (party mismatch between governor and presidential election).

  • COVID Deaths Per Mil - Number of deaths involving COVID-19 per million people. https://data.cdc.gov/NCHS/Provisional-COVID-19-Death-Counts-by-Week-Ending-D/r8kw-7aab

  • Change - Change in Rank from COVID Deaths Per Mil to Excess Deaths Per Mil.

  • Excess Deaths Per Mil - Number of deaths in excess of the normal amount per million people. https://data.cdc.gov/NCHS/Provisional-COVID-19-Death-Counts-by-Week-Ending-D/r8kw-7aab

  • Predicted - Number of excess deaths per million people predicted by a multiple linear regression of excess deaths relative to 12 variables.

  • Residual - Residual is margin between the Excess Deaths Per Mil and the Predicted value.

  • Conspiracy Alert I! The Ratio of Excess Deaths to COVID Deaths Was 9% Greater in Red States Than Blue states. Did Red States Underreport COVID Deaths? Did Blue States Overreport?

    Looking at COVID Deaths Per Mil as Kim Iversen did could lead one to believe that the rate of mortality was about the same in red and blue states, with more blue states at the top and bottom of the ranking and more red states in the middle. In fact the average was almost identical: 1,115 deaths per million in red states versus 1,116 in blue states. And by extension, this suggests that the more restrictive measures practiced more often in blue states were not very effective, if at all. Specifically, California (39th with 758 per million), one of the most restrictive states, lands only a little better than Florida (33rd with 970), one of the least restrictive states. However, when considering Excess Deaths Per Mil, red states tend to move up the list with higher death counts while blue states move down. California improves three spots to 42nd (1,150) and Florida worsens seven places to 26th (1,531). So the gap between California and Florida is larger, but some would argue it is still not enough to offset the cost of the California measures. Overall the average excess deaths in red states was 1,627 per million relative to 1,489 in blue states.

    Trump Voters (or Would-Be Voters) Were Twice as Likely to Die as a Result of the Pandemic in 2020?!

    1,627 to 1,489 is not a very large margin (9%), but considering some states are rather purple, the gap between Republicans and Democrats is probably larger. A linear regression of excess deaths relative to 2020 presidential election voting by state produces the formula below. (The regression was performed using https://www.statskingdom.com/410multi_linear_regression.html.)

    Y = 2000 - 10.31 * X

    where Y stands for excess deaths per million and X is the percent of the 2020 presidential votes that were for Biden. So hypothetically, using this formula, a state with all Trump voters would have experienced 2,000 excess deaths per million people while a state with all Biden voters would have had only 969 excess deaths per million. However, among other limitations, the above regression is not statistically significant. None the less, the examination of excess deaths by state does lead me to believe that blue states, where COVID regulations tended to be more restrictive and people presumably were more cautious, had significantly fewer deaths.

    Conspiracy Alert II! Andrew Cuomo and New York State May Have Had Plenty of Company When It Came to Underreporting COVID Deaths

    As noted above, California and Florida had considerably more Excess Deaths Per Mil than COVID Deaths Per Mil, but there were some states with much larger margins. Leading the list are Arizona (1,166 moved up to 2,179), Wyoming (663 up to 1,595), South Carolina (1,001 to 1,801), and Tennessee (1,070 to 1,845). In these states excess deaths per capita doubled the COVID deaths, give or take. On the other end of the spectrum, Rhode Island (1,682 moved down to 1,497), Massachusetts (1,431 down to 1,259), and Connecticut (1,671 to 1,628) all have fewer excess deaths per capita than COVID deaths. Although there is some variation in mortality rate from year to year, the percent change in recent years has been around 1% but ballooned to 15.9% in 2020 (https://www.cdc.gov/mmwr/volumes/70/wr/mm7014e1.htm). Therefore, it seems reasonable to believe that the majority of the excess deaths in 2020 were related to COVID. So why were reported COVID deaths in some states so much fewer than excess deaths? Not all excess deaths were directly caused by COVID, such as the deaths of people who did not or could not receive medical care who would have in a normal year and suicides due to the increase in stress and isolation. But why would such indirect COVID deaths be higher in redder states (with Republican governors) like Arizona, Wyoming, South Carolina and Tennessee where COVID restrictions tended to be less severe? My hunch is that the discrepancy between COVID deaths and excess deaths is more a result of undercounting COVID deaths than it is a result of indirect COVID deaths. Certainly some people who died had COVID that went undiagnosed or unreported. But was any undercounting intentional or not? In New York, Democratic Governor Andrew Cuomo's administration has been accused of hiding some nursing home deaths. However, the excess death rate in New York (2,340 per million) was "only" 405 more than the COVID death rate (1,935 per million), a number far less than the margins in Arizona (1,013), Wyoming (932), South Carolina (800) and Tennessee (775). So although New York likely cooked the books, I wouldn't be surprised if there were other states, especially some under Republican leadership, that did likewise. On the other hand, it is possible that some bluer states like Rhode Island, Massachusetts, and Connecticut may have knowingly overreported COVID deaths. However, the larger margin between COVID and excess death rates in red states versus blue states may have been more unintentional than a product of treachery. Red states may have not devoted as much attention and as many resources to the pandemic as did blue states and consequently they may have recognized fewer of the actual COVID deaths.

    Were the Shutdowns and Mask Mandates Effective? What Factors Had the Greatest Impact on Mortality Rate?

    Joe Rogan and Kim Iverson have used California to demonstrate that the lockdowns were ineffective and Joe mocked the use of masks. So did the lockdowns and mask mandates have an effect or not? A regression analysis of Excess Deaths Per Mil versus total days of lockdown and total days of mandated mask use yields the formula:

    Y = 1617.3511 - 0.6850 * X1 - 0.3818 * X2

    where Y is excess deaths per million, X1 is lockdown days, and X2 is mask-mandate days. The X1 coefficient is negative (-0.6850), therefore more days of shutdown result in fewer deaths. 0 shutdown days (and 0 mask-mandate days) result in 1617 excess deaths per million. On the other end of the spectrum, 295 shutdown days (if shutdown started when the pandemic was declared in early March and stayed in effect the remainder of the year) result in 202 fewer excess deaths per million. Likewise, 295 mask-mandate days result in 113 fewer excess deaths. This suggests that shutdowns and mask mandates did "work" by leading to fewer deaths. However, some people might contend that the cost to the economy and individual freedom outweighs the benefit of fewer deaths. Moreover, the relationship between mask mandates and excess deaths is not statistically significant. (Statistical significance is achieved when there is a less than 5% chance that the result can be explained by chance.) Furthermore, even when there is a correlation, it does not prove causation. In other words, the reduction in excess deaths actually might be caused by other factors that are correlated with increased shutdowns and mandates. So let's do another regression analysis, this time of excess death rate (per million people) versus 12 different variables. It yields the formula:

    Y = 4521 - 61.98 * X1 - 29.19 * X2 + 8.297 * X3 - 2805 * X4 + 44.89 * X5 + 15.01 * X6 - 23.90 * X7 - 8.229 * X8 + 17.98 * X9 + 0.1457 * X10 - 0.01620 * X11 - 8.096 * X12

    The dependent variable Y is excess death rate. The 12 independent variables are listed in Table 2:

    Table 2. Independent Variables Sorted by Decreasing Impact on Excess Death Rate

    Variable Description Worst State Worst Value Best State Best Value Decrease in Excess Death Rate--Worst to Best
    X7 % Biden Voters Wyoming 27 Washington D.C. 92 1554
    X6! Population Density New York 105 Montana 5 1501
    X4# % White People Washington D.C. 47 ME, NH, VT 96 1374
    X3 Clear Days Arizona 193 Vermont, Washington 58 1120
    X2 Mean Temperature North Dakota 41 Hawaii 73 934
    X5* % Seniors 65+ Maine 21.3 Utah 11.5 440
    X8* Median Income West Virginia 44K Washington D.C. 85K 337
    X9* % Obese People Mississippi 41 Colorado 24 306
    X12* % Uninsured Massachusetts 3 Texas 18.4 125
    X1* Dem Governor? many 1 - Yes many 0 - No 62
    X10* Mask Mandate Days New Jersey 266 many 0 39
    X11* Lockdown Days California 257 many 0 4

    * - Correlation with excess death rate is not statistically significant.
    ! - The metric of population density used here is the average number of people per mile, assuming all people within a county are evenly distributed.
    # - For this metric, black people are considered 0% white and Latinxs and Native Americans are considered 50% white.

    America: Land of the Free and Home of the Fiercely Independent

    Surprisingly, lockdowns (X11) and mask mandates (X10) are not correlated with excess death rate. In fact, using the formula, more days of lockdowns and mask mandates actually increase the excess death rate slightly, but the results are not significant. So maybe the restrictions didn't help, as Rogan stated. Common sense and data from the Asia and Oceania lead me to believe that restrictions can help. So what explains the results? The lockdowns and mandates weren't the same from state to state, making it a little like comparing apples to oranges. They also weren't enforced thoroughly. But an even better explanation of the results might be that Americans don't like to be told what to do by the government.

    Being Old May Not Be a Death Sentence

    Also surprising is that the correlation between excess death rate with percentage of seniors (X5) and median income (X8) also are not statistically significant. Using the formula, having a higher percentage of seniors leads to more excess deaths, but not by a very large amount. Based on mortality tables by age, seniors had a much greater chance of dying. So why didn't states with higher percentages of seniors experience significantly greater excess deaths? Perhaps states with more seniors (e.g. Maine and Florida) attract seniors who are wealthy enough to move to retirement states and their wealth made it easier for them to shelter-in-place and get medical care. However, states with higher median incomes didn't experience a lot fewer deaths, so maybe wealth wasn't a big factor. Another thing to consider is that the typical death rate among seniors is already much higher, so the rate at which people died "with COVID" but not "from COVID" was presumably also higher among seniors, inflating the appearance that seniors were more vulnerable.

    Six People Walk into a Bar

    Six people walk into a bar. A month later they die of COVID. Joe Rogan says: "Look, four of them were obese. If they had been in shape, not all would have died." And Jimmy Dore adds: "I talked to them; none of them were insured. If we had Medicare for all, not all would have died." Saint Peter overhears this and says: "Tell you what fellows, I'll give them another chance. Do your best!" So the six people once again walk into a bar. Joe approaches them and says, "Come work out with me and lose the weight and you'll have better immunity against the virus." They agree and lose the weight. And Jimmy hands them insurance cards and says, "Now you have medicare and you don't have to pay for healthcare." A month passes and once again the six people die of COVID. Joe and Jimmy are incredulous. And they say to Saint Peter: "You must have made a mistake! They're all in shape and insured. Some of them would have survived." To which Saint Peter says, "You noticed their weight and lack of insurance, but oddly neither of you of were struck by the fact that four of them were wearing MAGA hats."

    Ice Cream Tied to Shark Attacks - A Confounding Tale

    The Seaside Gazette proclaims "Rise in Ice Cream Sales Tied to Increase in Shark Attacks!" Statistics tells us that correlation does not imply causation. Ice cream sales are infact correlated to shark attacks, but the relationship is not causal. If the mayor of Seaside bans all ice cream sales, the number of shark attacks are unlikely to decrease. What is at play is that there is a third variable, known as a confounding variable, that does affect both of the other two variables. In this example, a confounding variable is weather. The heat of summer causes people to both buy more ice cream and go swimming more. And more swimming leads to more shark attacks. It is important to identify confounding variables so that... well... ice cream sellers are not unfairly blamed for shark attacks... and so that the influence of obesity and the lack of health insurance on COVID mortality is not exaggrerated.

    Cause of Death Among the Uninsured: Being Trump Voters

    On The Jimmy Dore Show, Jimmy has repeatedly said things to the effect that "200 thousand people who died from covid would be alive today if they had Medicare for all." (clip from The Jimmy Dore Show. Start at 17:45.) On Twitter he added the quote "About one-third of COVID-19 deaths and 40% of infections were tied to a lack of insurance;" referencing the report "Unprepared for COVID-19 How the Pandemic Makes the Case for Medicare for All" by Eagan Kemp. The actual quote in this report uses the word linked not tied, as did Jimmy, and is derived from another report titled "The Catastrophic Cost of Uninsurance: COVID-19 Cases and Deaths Closely Tied to America’s Health Coverage Gaps" by Stan Dorn and Rebecca Gordon, who use the words associated as well as linked when describing the relationship between insurance and pandemic mortality. And Dorn and Gordon drew their conclusion from the original study titled "County-Level Predictors of COVID-19 Cases and Deaths in the United States: What Happened, and Where Do We Go from Here?" by John M McLaughlin et. al. of Pfizer. In this study they preform a regression analysis of mortality relative to many variables, but never use the words linked or tied and use proper caution when drawing conclusions. So the progression from the original study to two reports to Jimmy's mouth was like a game of Whisper Down the Lane, with the conclusions being expressed with ever greater certainty.

    200 thousand seemed too high, given that only about 10% of the population is uninsured, and that emergency care is for everyone, even the uninsured. So I added % Uninsured (X12) as an independent variable in the regression analysis. The result suggests, quite contrary to what is suggested in the reports, that mortality actually goes up when more people are insured, but the correlation is not statistically significant. Why the huge difference? A large part of the difference appears to be the result of a confounding variable that Pfizer did not include in their study, and that variable is politics. If I remove % Biden Voters (X7) and the variables involving the governor (X1), masks (X10) and lockdowns (X11), then % Uninsured (X12) goes from predicting ca. 30,000 more deaths if everyone were insured to predicting ca. 60,000 fewer deaths if everyone were insured. 60,000 (or 12% of total mortality) is significantly less than the ca. 1/3 (33%) stated in the reports, but because my original regression showed no advantage to being insured and because there was still a big swing of 90,000 lives when I removed politics from the equation, I suspect that politics, as a confounding variable, explains for most or all of the correlation between insurance and pandemic mortality. Why did Pfizer leave out politics? Is it that I'm smarter than a team of scientists and mathematicians from Pfizer? Or was it that Pfizer didn't want to touch politics with a 10-foot poll? A boy can dream. (It's worthwhile to note that there are a lot of differences between the two studies. Some of our variables were similar, some different. The Pfizer study used county-level data, COVID deaths, and a binomial regression. I used state-level data, EXCESS deaths, and a linear regression. So Pfizer used higher resolution data and their model presumably was better, but I think they'd also see a big drop in the benefit of being insured if politics were included in their analysis.)

    Would you expect there to be a causal relationship between politics and insurance? It makes sense to me. Aren't Trump voters more likely to want control of their hard-earned money, especially if their medical bills are low, and wouldn't they have been more likely to opt out of insurance when given a chance? An article titled "Senate Republicans Represent States With Highest Uninsured Rates" reports that of the top 22 most uninsured states, 16 have two Republican Senators, 5 are split, and only 1 has two Democratic Senators. So there clearly appears to be a strong correlation between red states and higher rates of uninsured people. (I guess I buried the lead.) But wait. Red states have more poor people, you say, and you'd expect them to have more uninsured people? Maybe, but in my regression analysis, the correlation of % Uninsured (X12) is greater (but inverted) with % Biden Voters (X7) than it is with Mean Income (X8), although it's close. (Actually, the strongest correlation was with Clear Days (X3). Interesting. More sunny days and less insurance? Hmm...)

    As for the other side of the confounding relationship, namely that of politics and pandemic mortality, it does appear that Trump voters put themselves at greater risk and did experience a greater rate of mortality, as will be discussed later (The Big Three: Political Party, Population Density, and Race).

    Therefore, it appears more likely that pandemic deaths among the uninsured can be attributed to political philosophy than to the lack of insurance. If Jimmy actually did hand out insurance cards to everyone, what reactions do you think he would get? Some people might tell him, "we don't want your benefits, we don't want your welfare..., you leave us the hell alone" (Lauren Boebert, CPAC 2021), and some people, not trusting the medical establishment, wouldn't use the service. If Jimmy insists on continuing to say Medicare for all would have saved 200,000 lives, in fairness to the original study, he should also mention that getting all Americans to smoke would have saved an additional 200,000 lives! No kidding; check out the study. Put that in your pipe and... well, you know. And Jimmy, don't forget to mention that the study, which lead to the reports, and gave you 200,000 Medicare-saved lives, was conducted by your archenemy, Big Pharma! (Okay, one of many archenemies.) Moral of the story: be careful with numbers (and words).

    Cause of Death Among the Obese: Being Trump Voters

    As with seniors, higher percentages of obese people (X9) only lead to moderately more excess deaths. Rogan stated that hospitals reported that 70% of the people who died with COVID were obese. If true, it seems that the correlation between obesity percentage and excess deaths would be highly significant. Maybe the obesity percentage numbers are not that accurate. Or perhaps the emphasis that Rogan puts on physical fitness to protect yourself from COVID is greatly exaggerated. Let's take a closer look. The correlation between obesity and excess deaths is statistically significant when considered without the other 10 independent variables. But when all 12 variables are included, the impact of obesity is no longer significant. However, the correlation between obesity and candidate preference (X7) is significant, with Trump voters twice as likely to be obese. And unlike with obesity, when all 12 variables are considered, the correlation between candidate preference and excess death rate is significant. Therefore, the relationship between obesity and death rate largely may not have been causal. In other words, although more obese people were dying, they were dying not as a result of obesity so much as a result of being Trump voters, who presumably were less cautious concerning COVID and experienced greater exposure to the virus.

    Weather is a Significant Factor

    The next surprise is the apparent effect of weather on mortality. The correlation between the number of clear days (X3) and mean temperature (X2) and excess deaths were both significant. But for me, the big surprise is that having more sunny days is correlated with more deaths rather than fewer deaths. People have reported that sunshine kills the virus quickly and that the vitamin D produced by sunshine helps to fight the virus. But I guess something about cloudy climates serves as better protection against the spread of COVID. As for temperature, warmer states did better.

    The Big Three: Political Party, Population Density, and Race

    The three factors that appear to have had the greatest effect on excess mortality in 2020 are political party, population density, and race. As was previously seen when correlating 2020 presidential voting with excess deaths, even when considering 12 independent variables, candidate preference (X7) had a large impact on excess deaths. In the formula, each percentage point in favor of Biden is worth 23 less excess deaths per million people. The gap between the state with the highest Biden voters (Washington D.C. - 92% Biden) and that with lowest percentage (Wyoming - 27% Biden) is 1,528 deaths per million. So if D.C. was as red as Wyoming, the formula would predict 1,528 additional deaths per million in D.C., almost doubling its excess death toll. And if Trump won 100% of the vote in the U.S., the formula predicts there would have been a total of 890,000 excess deaths. Whereas, if all the votes were cast for Biden, there would have been only 120,000 excess deaths. This suggests that the behavior of Trump supporters may have been seven or more times as risky as that of Biden supporters. The reason that the pandemic death rate among Trump supporters was "only" twice as much is that Trump supporters have some lower risk factors, such as living in areas of lesser population density (X6) and having fewer people of color (X4). There is a lot of uncertainty in these numbers, but it illustrates the greater risk taken by Trump suporters. This risk is unsurprising since Trump and his supporters tended to downplay the danger of COVID-19 and thus presumably didn't take as many precautions, such as wearing masks and social distancing.

    The effect of political party is offset to a degree by population density (X6), since blue states tend to be more populous and greater population density is linked to more excess deaths. This is intuitive with a communicable disease. The gap between the state with the lowest population density (Montana - 5 people per mile) and that with the highest density (New York - 105 people per mile) is 1,496 deaths per million. So if the population of New York was as dispersed as Montana, the formula would predict 1,496 fewer deaths per million, leaving New York as one of the safest states, having only 844 excess deaths per million. The third big factor is skin color (X4). Greater mortality numbers have been reported among black people, Latinxs, and Native Americans. Higher mortality numbers are predicted by this regression analysis. So why are the numbers higher? One thought is that these minorities are poorer and/or more unhealthy and consequently less safe. However, wealth (X8) and obesity (X9) are independently considered as part of this analysis, so there's something else involved with race. It has been reported that people with darker skin generate less vitamin D and that vitamin D helps to protect against COVID. This is probably a significant factor. It's also seems possible that racial discrimination is involved whereby minorities have less access to healthcare.

    Conclusions

    As Joe Rogan and Kim Iversen said, lockdowns and mask mandates don't appear to have worked. But I think it's not valid to make such conclusions by comparing the COVID death rates in different states, such as California vs. Florida and Texas. Instead more variables should be considered. Although the average COVID death rates was similar in red and blue states, red states actually had a 9% greater excess death rate. This increase suggests that the underreporting of COVID deaths was greater in red states which raises the possibility that Andrew Cuomo and New York may have had company, especially among red states, cooking the books. The 9% gap between red and blue states actually translates to twice the death rate among Trump voters when analyzed by linear regression. The greater risk among Trump voters presumably is because they weren't as cautious and thus were more exposed to the virus even though they live in less populated areas. Rogan also reported the high percentage of obesity among COVID deaths and concluded obesity was a primary factor. Interestingly, the current study found no significant correlation between obesity and excess death rate. Instead the relationship is indirect and not causal. Obesity is correlated with Trump voters and Trump voters are correlated with greater excess deaths. If all voters were Biden voters, a linear regression with 12 independent variables predicts there would have been only 120,000 excess deaths, down from 500,000. And if we had taken a more coordinated and aggressive approach to dealing with COVID, it seems reasonable to believe the COVID death numbers could have approached zero, similar to that reported in China, New Zealand, Australia, and other Western Pacific countries, with less of an overall impact on our economy than we are experiencing. Even if the findings in this study aren't very accurate--the input data may have flaws and linear regressions are simplistic--they show the importance of taking a deeper look at the data, especially before drawing conclusions in front of millions of people.

    The other factors correlated to greater excess deaths rates are greater population density, darker skin, clearer days and colder weather.

    Go back to top for the Disclaimer.