[ad_1]
If we don’t make a living analyzing the statistics, it is easy to be misled by misinformation about Covid-19 statistics on social media, especially if we don’t have the proper context.
For example, we can select the statistics that support our point of view and ignore the statistics that show us that we are wrong. We also need to interpret these statistics correctly.
It is easy for us to share this misinformation. Many of these statistics are also interrelated, so misunderstandings can multiply quickly.
Here’s how we can avoid five common mistakes and impress friends and family by getting the stats right.
1. What is scary is the infection rate, not the death rate
Social media posts Comparing Covid-19 to other causes of death, like the flu, implies that Covid-19 isn’t really that deadly.
But these posts miss the infectivity of Covid-19. For that, we must look at the infection mortality rate (IFR): the number of deaths from Covid-19 divided by all those infected (a number that we can only estimate at this stage, see also point 3 below).
While the jury is still out, Covid-19 has a higher IFR than the flu. Posts that imply a low IFR for Covid-19 certainly underestimate it. They also lose two other points.
First, if we compare the typical flu IFR of 0.1 percent with the more optimistic Covid-19 estimate of 0.25 percent, then Covid-19 is still more than twice as deadly as the flu.
Second, and most importantly, we must look at the basic reproduction number (R₀) of each virus. This is the estimated number of additional people an infected person can infect.
The R₀ for influenza is approximately 1.3. Although Covid-19 estimates vary, its R₀ is around a median of 2.8. Due to the way infections grow exponentially (see below), the jump from 1.3 to 2.8 means that Covid-19 is much more infectious than the flu.
When you combine all of these statistics, you can see the motivation behind our public health measures to “limit the spread.” It’s not just that Covid-19 is so deadly, it’s deadly and highly infectious.
2. Exponential growth and misleading charts
A simple graph could represent the number of new Covid cases over time. But since new cases can be reported erratically, statisticians are more interested in the growth rate of total cases over time. The steeper the upward slope on the chart, the more we should be concerned.
For Covid-19, statisticians seek to track exponential growth in cases. Simply put, unrestricted Covid cases can lead to increasing numbers of cases. This gives us a graph that follows slowly at first, but then curves sharply upward over time. This is the curve that we want to flatten, as shown below.
However, social media posts routinely compare Covid-19 figures to other causes of death showing:
- more linear patterns (numbers increase over time but steadily)
- much slower growing flu deaths or
- low numbers from the early stages of the outbreak and thus lose the impact of exponential growth.
Even when researchers talk about exponential growth, they can still be misleading.
The widely shared analysis by an Israeli professor claimed that the exponential growth of Covid-19 “fades after eight weeks.” Well, I was clearly wrong. But why?
Their model assumed that Covid-19 cases grow exponentially over several days, rather than a succession of transmissions, each of which can take several days. This led him to trace only the erratic growth of the initial phase of the outbreak.
The best views truncate those first erratic cases, for example, starting from case 100. Or they use estimates of the number of days it takes for the number of cases to double (between six and seven days).
3. Not all infections are cases
Then there is the confusion between Covid-19 infections and cases. In epidemiological terms, a “case” is a person who is diagnosed with Covid-19, primarily from a positive test result.
But there are many more infections than cases. Some infections show no symptoms, some symptoms are so mild that people think it is just a cold, the tests are not always available to everyone who needs them, and the tests do not detect all infections.
Infections “cause” cases, tests find cases. US President Donald Trump was close to the truth when he said the number of cases in the US was high due to the high testing rate. But and others I still totally got it wrong.
More tests do not result in more cases, they allow a more accurate estimate of the actual number of cases.
The best strategy, from an epidemiological point of view, is not to perform fewer tests, but to test as widely as possible, minimizing the discrepancy between cases and infections in general.
4. We cannot compare deaths with cases from the same date.
Estimates vary, but the time between infection and death could be up to a month. And the variation in recovery time is even greater. Some people get very sick and take a long time to recover, others show no symptoms.
Thus, deaths recorded on a given date reflect deaths from cases recorded several weeks earlier, when the case count may have been less than half the number of current cases.
Rapid case duplication time and long recovery time also create a large discrepancy between case counts. active and recovered cases. We will only know the true numbers in hindsight.
5. Yes, the data is messy, incomplete, and may change.
Some social media users get angry when statistics are adjusted, they fuel conspiracy theories.
But few realize how gigantic, chaotic, and complex the task of tracking statistics on a disease like this is.
Countries and even states can count cases and deaths differently. It also takes time to collect the data, which means retrospective adjustments are made.
We will only know the real numbers of this pandemic in hindsight. Similarly, early models weren’t necessarily wrong because the modelers were misleading, but because they didn’t have enough data to work with.
Welcome to the world of data management, data cleansing, and data modeling, which many chair statisticians don’t always appreciate. Until now.
– Jacques Raubenheimer is Principal Investigator in Biostatistics at the University of Sydney
– This story was first published on The Conversation.
[ad_2]
