Denominator issue - again!
Over the next few days we are going to seriously bump up against the “dominator issue, which I am guessing will be the source of public panic. With a move in New York and elsewhere to test only those people who are sick enough to go to the hospital, it is difficult to understand the mortality numbers. Here is an attempt to explain what I mean.
Many people in the general population have had or currently have COVID-19. Given most people will be asymptomatic or have mild symptoms only (around 80% per current estimates), we are encouraging people who have symptoms of COVID-19 to stay home and only go to the hospital if they feel really sick. The people who are asymptomatic or have mild symptoms are not reported in the system as a COVID case. (TRUE Denominator)
Only people who are sick enough to go to the hospital are being counted as cases (with symptoms, whether or not they are tested) (Current Denominator being used)
People sick enough to go to the hospital are more likely to need critical care. In public health terms this is called “selection bias”.
Some of these people will die (numerator).
Some of my colleagues are using the term “infection fatality ratio” instead of “case fatality rate.” At this point neither correspondents to what we are collecting. The currentdenominator is neither cases nor infections. It is actually “infections where people sick enough to go to the hospital”. As we ration hospital care to sicker patients (which is what is happening in NYC), it will seem as if a higher percentage of COVID-infected people are dying. But, because we don’t know the True Denominator (how many people have been infected with COVID), we don’t know what percentage of COVID-infected people are dying. We can expect that more people will die if our hospitals can’t handle the patient influx. We also don’t yet know the natural course of disease – which is a way of saying -what is your likelihood of dying if there is no hospital care at all.
The dominator issue:
How we are currently counting:
People who died / people who were sick enough to go to the hospital
How we would normally count:
People who died / People who were infected (which includes people that didn’t have any symptoms, people who had mild symptoms, or people who felt sick but not sick enough to go the hospital, and people that went to the hospital)
The other problem right now is that countries are not releasing disaggregated data (eg. by age, underlying medical conditions, etc). Without this type of detail, from a professional perspective, it very hard to sort through what we are hearing in the press.