- TRUMP IS CORRECT ON MORTALITY RATES
The problem with Mr. Trump’s approach is that it is simply impossible at this time to calculate the ratios that he would like to use.
Those of you who follow my essays understand that I look at death statistics because they are the best gauge of the effects of infections. These data are well documented and available on a state-by-state basis as well as on a country-by-country basis. Much more difficult to find are similar statistics on hospitalizations. The most difficult statistic to find is the actual number of cases. Most epidemiologists believe that the numbers of identified cases are only a fraction of the actual numbers of cases, underestimating the true infections by as much as 5 to 10 times.
As Mr. Trump has said, a very useful “mortality” statistic would be, “How many deaths result from a population of infected individuals?” This deaths/case ratio is a very good measure of how a particular hospital, state, or country is handling those infected individuals, how well the healthcare system is responding and how likely an infected person is to die after being infected.
A similar statistic would be deaths/hospitalizations; however, as I said, hospitalization statistics are extremely difficult to find. And this is complicated by the fact that each hospital, city, state and country may use different gating criteria for admitting an individual to a hospital, or for sending them home.
The other measure of “mortality” is to ask, “How many people have died from the disease as a fraction of the total population in the country?” This deaths/millions of people is a very good measure of how a particular state or country is experiencing the pandemic and how likely a person in a population will die this week or next.
Let’s take just a moment to look at how reliable these 3 potential metrics are: deaths, cases and population.
POPULATION
This metric is the most reliable of the three. We have strong confidence that populations of states and countries are accurate. There may be some variation based on the ability of governments to count their people, but in general, potential errors are in the single digits.
DEATHS
This metric is pretty strong but has potential errors.
There are potential errors in overcounting deaths. For example, it is possible that a patient is admitted to a hospital in critical condition; that patient is tested for COVID-19 and found to be positive; but the patient does not survive the critical illness. That patient will be counted as a COVID-19 death since he/she had been tested and found to be positive; however, in some situations the critical condition would have occurred whether the patient had been infected or not. It is possible, particularly with the fact that many people who become infected have mild or symptom-free disease, that there are people who die in the hospital whose positive test for virus was coincidental, not causative.
Overcounting is most likely true, even if the numbers are small. We will not know how many of these events occurred until epidemiologists have the ability to look historically at statistics in the future. Retrospective analysis is not only common; it is the norm. Comparison of expected or predicted death rates from particular causes or conditions will be compared to observed or actual death rates and the difference between those two will be used to retrospectively characterize the death rates from COVID-19.
There are also errors in undercounting deaths. These errors come from multiple sources. For example, a patient is admitted to a hospital in critical condition and expires before a COVID-19 test can be completed. Perhaps he/she waited too long to go to the hospital, suffered irreversible organ damage and died too quickly to allow for testing. Although actually a death from COVID-19, without tested confirmation it would not be recorded as such. Additionally, there will be patients who die at home before being tested, confirmed and recorded. Finally, there may be institutions whose internal processes reduce the number of cases that can be recorded as COVID-19 deaths (or more correctly, deaths associated with COVID-19 infection). The absence of national standards for attribution of COVID-19 as a complication resulting in death may result in inconsistent reporting of appropriate statistics.
When taking into account both the potential overcounting and potential undercounting, the vast majority of healthcare professionals, virologists and epidemiologists conclude that these errors when combined likely result in an undercount of the actual number of deaths.
Since variations may occur in hospitals, states and by countries, it is always important to keep these potential counting errors in mind when looking at data.
True death statistics will only be available in the future after retrospective analyses.
All of that being said, deaths, as a statistic are reasonably solid. It is a good metric to use. One may question whether someone is infected or not, that testing has error rates, but death is a clear binary choice.
CASES
This measurement is extremely difficult to make. The total number of cases in a population is enormously difficult to determine. Even were it possible to test every single individual today and thus to determine exactly how many people are infected today, that number would be irrelevant tomorrow because of the mobility of the virus. There have been identified large error rates in the tests being performed today, particularly with the rapid tests. Further, each state and each country use different screening methods to allow people to qualify for testing.
The bottom line with testing is that although it identifies cases, it does not provide any viable statistical measurements of the level of infection in a community, a state or a country.
MORTALITY
As you can see from the above, looking at Deaths/Case, it is virtually impossible at this time to make this measurement. We have a pretty good idea of deaths, but no way to determine cases. If we stopped testing today, the denominator would drop to almost zero, making the death rate huge.
Measuring Deaths/million is a valid metric. We have a pretty good idea of deaths and a pretty good idea of population. Although there is some uncertainty in both the numerator and the denominator, those variations are small, so the metric, particularly in comparing one area or country to another is quite valuable.
In the future, should hospitalizations or overall cases become metrics that are clear, we may actually be able to judge our healthcare system’s response to this virus and compare hospitals, states and countries to each other on how their healthcare system was able to respond to patients’ needs.
Please remember, that when we can do those comparisons, we will be looking at the ability of healthcare professionals to treat the disease, not on whether the infection is better or worse in compared locales.
DEATHS PER MILLION PEOPLE
There are 88 countries with populations over 10 million, of those ,79 have lower death rates than the US.
There are 28 countries with populations over 50 million, of those 24 have lower death rates than the US.
There are 14 countries with populations over 100 million, of those all 13 have lower death rates than the US.