Lost in DataMay 13, 2020 -
Categorized in: Uncategorized
Accurate information is an extremely valuable asset when making life-altering decisions.
Excessive or inaccurate information is a liability that leads to confusion instead of clear-headed decisions.
While I am neither an epidemiologist nor a professional statistician, it appears to me that we may be getting lost in excessive and potentially inaccurate data in making COVID-19 response decisions.
Understanding both current and historical decision-making often requires two sets of data. One data set documents the evidence on which decisions are made. It is always incomplete and full of information gaps. Decision-makers never have access to real-time accurate data.
The second data set uses additional contemporary information to document what was actually occurring. This is the evidence which many people use to evaluate whether a decision was correct. It is far more complete and accurate, but only available in hindsight.
My observation is that these two sets of data are being jumbled together in the current crisis which is creating confusion and, perhaps, false impressions. Here are a few specific public data points for you to ponder:
Total number of positive tests
- This is typically reported as the total number of people who have ever tested positive for COVID-19. It is useful to gauge the overall spread of the virus.
- This is getting muddled because “new” cases that are reported every day may include newly diagnosed cases from weeks or even months ago.
- The solution would be to keep two different sets of data: A real-time set that would only include new positive tests within the last 48 hours and another historical set that would include both old and new positive tests by the date of the test.
Percentage of positive tests
- In some states this data point is being used to gauge how restrictive social distancing measures should be.
- In addition to suffering from some delayed data issues, this data point could be misinterpreted due to selection bias and the difference between past and current availability of tests.
- It would be helpful to divide the data into two sets – people who presented with COVID-19 symptoms and people who were asymptomatic.
- If the percentage of timely positive tests from people who have COVID-19 symptoms changes we would know in real-time whether the virus was spreading more or less rapidly.
- Testing large numbers of asymptomatic people is useful and may give us an idea of how likely a second wave may be later this year, but it also could provide a false sense of progress if they are included in the percentage of positive tests at this point.
- Our baseline for the percentage of positive tests should continue to be based on those people presenting with symptoms.
- The number of deaths attributed to COVID-19 are staggering and sad.
- This data point suffers from both a timely reporting deficiency and potential “cause of death” variability.
- As time passes, states are doing additional testing on people who passed away earlier this year and revising the number of deaths attributed to COVID-19. Once again, the daily data would be far more useful if divided into deaths in the last 48 hours and an historical total number of deaths by day of passing.
- This virus has a significantly larger impact on people with pre-existing conditions which makes it difficult to assess a true cause of death. Often there are many contributing health factors, but if a person tests positive for COVID-19 that is most likely listed as the primary cause for their passing.
- It would be helpful to have two sets of data – the number of deaths from those with pre-existing conditions and those without any pre-existing conditions. This would help decision-makers assess the health risks to the general population more clearly.
- I find this to be one of the more timely and helpful data points.
- The data is somewhat delayed as people who are hospitalized likely were infected and contagious for at least a week before seeking care.
- However, one of the primary reasons for social distancing measures is to ensure our medical facilities are not pushed beyond their capacity like Italy. Gauging hospital capacity is vital to being able to treat people effectively.
- If the number of hospitalizations is rising immediate additional social distancing is likely required. If the number of hospitalizations is declining social distancing measures could likely be relaxed gradually.
It is very easy to list the data points I would like to see in a perfect world, but clearly the world we live in is far from perfect. We are blessed to have the freedom to criticize those entrusted with making difficult decisions, but it is only after we work together to collectively take necessary precautions that the risks presented by this virus will decline to acceptable levels.
Quote of the week: Mark Twain: “Facts are stubborn, but statistics are more pliable.”
Securities offered through LPL Financial. Member FINRA/SIPC. Investment advice offered through IAG Wealth Partners, LLC, (IAG) a registered investment advisor and separate entity from LPL Financial.
Past performance is no guarantee of future performance. In fact, the opposite can be true. The information contained in this report does not purport to be a complete description of the securities, markets, or development referred to in this material. Investing involves risk including loss of principal.
Any opinions are those of IAG and not necessarily those of LPL Financial. Expressions of opinion are as of this date and are subject to change without notice. This information is not intended as a solicitation or an offer to buy or sell any security referred to herein. No strategy assures success or protects against loss. Investing involves risk including loss of principal.