Note: This Post has been revised since its initial publication in response to reader questions and comments.
As I have written previously, many in the media and the government often fail to present and interpret data correctly. With very few exceptions, such as the careful reporting by a handful of reporters at the LA Times and the online newspaper Noozhawk in Santa Barbara, this is proving to be especially true during the COVID-19 pandemic. Despite the fact that it has no bearing on our healthcare system capacity, the media tracks the pandemic primarily by reporting the inexorable rise in the total number of confirmed cases. Instead of carefully analyzing trends in the data, the media constantly reports on “surges” or “spikes” that they claim have just started or will soon be coming. They are wrong on both counts. How do I know this? It’s all in the data. You just have to look at it correctly.
Since the inception of the pandemic, California has been reporting two statistics on a daily basis (with certain gaps): (1) new fatalities; and (2) daily confirmed cases. Although the State and County have started publishing hospitalization data (in response to calls that I and others made for the State to start doing so), it will be most useful for assessing the trajectory of the virus in the first half of April (I have consistently predicted the virus will decelerate by April 9th). So today, as we look back on March, the best way to chart the trajectory of the outbreak is to look at the daily increase in both fatalities and new cases compared to the trajectory of an uncontrolled virus. The charts below use 5-day moving averages to plot the actual trajectory of the virus (to minimize the impact of gaps in the State’s data reporting).
Because of the variability in testing and new case reporting, epidemiologists have stated the most accurate way to chart the course of the pandemic in California is to examine the fatality rate. Accordingly, the chart below provides the 5-day moving average for fatalities since the pandemic began.
The most important thing to understand in reviewing this chart is that a fatality reported today likely reflects an infection that occurred 15-25 days ago (this is why fatalities will likely peak 5 to 10 days after the other data show that we have bent the curve).
What can we learn from this simple chart?
Critically for today, this means that sometime between March 1st and March 10th the pandemic began to lose steam. Since then, the virus has spread at a much lower rate, which likely corresponds to the implementation of decentralized public health measures beginning on March 8th. In short, the slow and steady rate of increase in daily fatalities confirms that we started flattening the curve before the stay at home order was issued and there has been no “surge” or “spike” in fatalities.
Because it is the most widespread metric used by the media for assessing the trajectory of the pandemic, I am including a chart with the daily increase in confirmed cases notwithstanding the challenges I previously noted with respect to trying to make sense of daily case data. I am including a chart with this data primarily to see if it paints a different picture of the virus trajectory when compared to the fatality data and because it provides a window into one driver of likely hospitalizations – people with serious symptoms who have tested positive for the virus.
Does anything stand out from this simple chart that would alter the conclusions drawn from the fatality data? No. It paints the same picture even if we should not read too much into it, except that a certain percentage of these cases are likely to end up in the hospital.
First, the virus does not appear to have been spreading at an uncontrolled rate.
Second, the flattening appears to be the result of the public health measures put in place before the stay at home order was issued on March 19th.
Third, there has not been a “surge” or “spike” in COVID-19 cases. To the contrary, our hospital admissions curve is flattening dramatically and may be demonstrably flat (or bent) within a few short days (as I have repeatedly predicted based on the results of models I began running ten days ago). I will be publishing hospital admission curves instead of case curves as soon as sufficient data are available.
I will have much more to write in the next few days about what this means for California going forward, especially if the data confirm my predictions that we will have flattened and potentially bent the curve the hospital admissions curve. But as you start your week, please do so knowing that we are absolutely on the right track in California and have been for weeks. Things are getting better, not worse. The story of California’s success is already written in the data.
Be safe. Stay healthy
Based on substantial comment and feedback, I revised this Post on April 6th to explain that Italy was the basis for my uncontrolled growth curve given its parallels to California in terms of the initial outbreak of the virus. I also re-ordered my two charts to make it clearer that fatality data is the basis for the conclusions reached in this Post and to acknowledge the limitations inherent in using confirmed case data pending the availability of sufficient hospital admission data.