How We Know What We Know
Through years of fighting infectious diseases we have developed many models that allow us to predict how a disease spreads. With a brand-new virus such as COVID-19, however, we often have difficulty in getting accurate statistics. The lack of testing in most nations has skewed results, and these numbers may be critical to the survival of hundreds of thousands of people. Fortunately there are a few areas where testing is prevalent, namely some random testing given in New York in recent weeks.
(Many of these numbers are fictional and for demonstration purposes only)
The process of finding errors in our statistics is called Bayesian Statistics. The more we factor in our errors, the more accurate we are. For example, take the 3000 confirmed cases from New York near the start of the pandemic. If it was known that 6,500 tests were administered total, and all of these were to symptomatic persons, this implies that 46% of all symptomatic persons were symptomatic COVID-19 patients. Now we take a look at how many people exhibit these symptoms, and we find it to be 13,000. We can now extrapolate that there were 6,000 symptomatic cases, and 8,000 total cases. With just a little math, we have taken a number completely skewed by our testing issues and transformed it into a reasonably good guess as to the number of cases. By taking the population of New York as 8,623,000, you can calculate the probability that any random person on the street is contagious by dividing the total cases by this number and yielding 0.09%. Please keep in mind however that the number of cases is constantly changing, and so these exact numbers are probably not accurate.
It is crucial that we use mathematics like this for a better idea of the extent of the disease. This allows us to predict where cases will occur several days in the future, allowing us to redistribute medical supplies as needed. Failure to recognize the true extent of this virus can lead to poor policy decisions, as could be seen in Italy, Spain, and the US. These nations spent much of the critical early period of this pandemic ignoring it or profiting off of it. If all nations had acted swiftly, then this virus may never have become a pandemic like it is.
There are several different ways to measure how dangerous a virus is. The most obvious, and most frequently referenced one, is mortality rate. Simply put, the mortality rate is the rate at which infected persons die from this virus. If there were 50,000 cases of COVID-19 total in Suffolk County, and 500 people died, then this virus would have a 1% mortality rate in Suffolk County. Keep in mind, however, that mortality rate is by no means a constant. You may even know that for COVID-19, the mortality rate is far higher amongst older people than it is amongst children. To account for this, epidemiologists will look at a population-adjusted mortality rate, or a mortality rate for a certain population. If that population were people ages 10-19, then it would be said that the age-adjusted mortality rate for people ages 10-19 years is 0.25%.
Mortality rate is however only one measure of severity. If an elderly person were to die, it would seem less damaging than if a young healthy person with their whole lives ahead of them were to die. Instead of looking at risk of death, it may be more meaningful to look at the total expected years of life lost. This is a good judgement of how harmful the virus is globally and over a long-term period. It basically measures how much expected life the disease has killed. It would be impossible to give an accurate count for total years of life lost until this pandemic has run its course, however you can measure how many years of life are lost per year.
For a far more in-depth look at the severity of this virus, we recommend reading this paper, which is where many of our numbers come from.
Note to reader: this post will be updated soon with the next part: Simulating the Pandemic. Please check back!