For a coronavirus data analysis, there are several reasons that a COVID-19 pandemic analysis of the number of deaths and new cases can deceive.
Reasons for this potential deception include:
- Data accuracy because of availability of test kits
- Measurement system’s false positive and false negative test results
- Availability of test kits to all sectors of the population
- Potential financial bias because more monies may be provided to a healthcare facility when a case is considered COVID-19.
- Someone who had coronavirus at death may have been in the late stages of their life, where, for example, COVID-19 shortened the life span of someone who had terminal cancer by only a few days or weeks.
A coronavirus data death is similar in many respects to past pneumonia or influenza deaths and may now even be a cause of these types of death. Because of this, an alternative measurement to determine whether the impact from the coronavirus data pandemic is currently lessening, stable, or increasing would be to analyze the percentage of pneumonia, influenza or coronavirus deaths to total deaths.
If this percentage increases, one could presume that the increase was caused by coronavirus issues. Similar, if there were a decrease, one could presume that coronavirus issues were decreasing.
The coronavirus data analyses provided below will focus on the percentage of pneumonia, influenza or coronavirus deaths to total deaths.
Source of Coronavirus Data
Data for this analysis is from a weekly Centers for Disease Control (CDC) coronovirus published report. The most recent data when this paper was written is from https://www.cdc.gov/coronavirus/2019-ncov/covid-data/covidview/06122020/nchs-mortality-report.html. The data in the following table is from this report.
The coronavirus data analyses provided below will focus on the percentage of pneumonia, influenza or coronavirus deaths to total deaths.
The analysis approach that his paper will suggest is a time series plot of the percentages (far-right column in the table) versus weekending dates. However, before creating this time series plot, one needs to make sure that there are no issues with the reported data values.
When examining this table, one notes:
• The total number of deaths for weeks numbered 22 and 23 was much smaller than previous weeks, which does not seem correct.
• The percentage value in the far-right column is much less than previous weekly-reported values. This would be good news from a coronavirus pandemic reduction perspective; however, is this table-number value a valid percentage estimate?
From these observations, one might wonder if CDC improves the accuracy of individual reported values over time. Let’s look into this possibility.
Previous CDC weekly reports can, in general, be obtained by subtracting multiples of seven days from the “06122020 date coding” in the above URL.
Using this process for obtaining previous CDC reports, the following table was created, which shows that reported CDC data values do change over time.
From this table, it appears that the table’s values start to approach similarity if the last two weeks from an individual weekly report were ignored. To highlight this observation, the last two weeks of each reporting in this table are shaded gray.
Because of this data assessment, the last two weeks from the most recent CDC data will not be included the time series plots I will be providing in future blogs.
Data analysis from these analysis plots and my past analyses can be found at COVID-19 Analyses.