In my first Post on COVID-19, I wrote about the importance of putting the spread of the virus in perspective, which requires careful consideration of the data we have and the data we do not have. In this Post, I am encouraging everyone to ask the media and their elected officials to fill in these data gaps so that we can more carefully assess how fast the virus is actually spreading, whether the government’s modeling is accurate, and whether its public health measures are working.
I believe that if the media and government begin to answer the questions below, we will gain valuable perspective on the actual risks posed by the virus and put ourselves on a track to gradually return to more normal life in California sometime shortly after April 17th. Please put these questions to your local media and elected officials as soon as possible.
And please practice maximum social distancing for the next ten days! If we honor the stay at home order for a full two weeks, there is a very good chance the data will show a dramatic flattening of the curve and the ability to implement less disruptive social distancing measures.
1. Why aren’t we reporting the number of COVID-19 patients requiring hospitalization and daily hospital capacity?
The stated goal of the extraordinary stay at home order in California is to “flatten the curve” so that the number of hospital admissions for COVID-19 does not exceed healthcare capacity and our healthcare professionals can provide the optimal level of care for each patient (whether COVID-19 or otherwise). Given this rationale, these two pieces of data are the most important reporting the government could be doing, not only to reassure the public, but also to assess the extent to which social distancing measures and the stay at home order are actually flattening the curve. Yet the government is not reporting this data. We should insist that the government begin reporting these data immediately. Ideally, the government would also report average hospital stay duration for COVID-19 patients. This has a major impact on actual hospital capacity – the longer the average stay, the less capacity we have in the system (and vice-versa).
2. Why aren’t we reporting the rate of increase, rather than just the amount of increase, with respect to all COVID-19 statistics (number of tested, number of confirmed cases, number hospitalized)?
The same data regarding the outbreak can paint two very different pictures. In California, we confirmed 265 new cases on March 22nd. A week before, we confirmed new 57 cases on March 15th. That’s 4.6x growth in one week. It looks like an explosive trajectory. However, the daily growth rate was 17% on March 15th and 18% on March 22nd. This is more of a constant trajectory, which could be easily demonstrated graphically to help provide the public with more perspective on the spread of the virus.
3. Why aren’t we reminding the public that it takes roughly two weeks for the benefits of public health measures to appear in the data?
The virus has a roughly two-week incubation period. This effectively means that measures put in place on a given day will not produce their maximum public health benefits for two weeks. In California, the first stringent public health measures were put in place in various localities on March 8th. This means that the benefits of these measures should start appearing in the data on March 23rd. On March 19th, the Governor issued the statewide stay at home order. This means that from March 23rd to April 2nd, the data should demonstrate the impact of social distancing measures undertaken before the statewide order went into effect. From April 3rd to April 17th, we should see the impact of the statewide stay at home order. If hospitalizations remain below system capacity after April 3rd, we will have successfully flattened the curve.
4. Why are there far fewer fatalities in California when compared to Italy given our rough similarities in population and healthcare systems and the fact that COVID-19 has been present for roughly the same period of time in each area?
California and Italy both documented COVID-19 cases in January. Both have excellent healthcare systems. California has roughly two-thirds of the population of Italy, but as of March 23rd, California has 27 deaths from COVID-19, and Italy has more than 6800, tragically. To what extent can this enormous difference in fatalities be explained by demographic differences, such as differences in median age (California’s is younger) and cigarette smoking rates (California’s is less than half of Italy’s and one of the two lowest in the U.S.). What other factors could explain this dramatic difference in fatalities? What does this mean for our response efforts?
5. Will the government publish the model it is using to predict the potential curve that must be flattened (i.e., daily confirmed cases, daily hospitalizations, and average duration of hospital stays) and the flattening that should result from the stay at home order?
Peer review is critical for scientific advances. California is home to some of the finest universities and scientific minds in the world. We should be subjecting our State model to peer review by our premier epidemiologists and statisticians to test and refine its data and assumptions, thereby improving its overall accuracy and utility. This should be done at regular and short intervals (i.e., every three days).
To help people understand why the number of confirmed cases is nowhere close to overwhelming the State’s healthcare capacity today (but could be weeks from now under certain scenarios), I will be building and publishing a very rough model later this week to illustrate how many new daily confirmed cases would be required to generate hospitalizations that exceed system capacity under various assumptions. (I recognize that certain localities may face capacity constraints earlier; I’m focusing on the State).
6. How often is the government updating its model and refining assumptions with actual data from California and jurisdictions with demographics similar to California?
Models are only as good as the data and assumptions upon which they are built. In California, we are just beginning to see larger numbers of confirmed cases, which may make it possible to develop (1) a reasonably accurate hospitalization rate (i.e., number of hospitalizations/number of confirmed cases) and (2) average duration of hospital stays. Given limited testing for the virus, these are the two most important pieces of data for modeling the curve. We will have substantial changes in our data roughly every three days, so updating the model on a similar timeframe will be critical. Insofar as we seek to supplement California data with other data given our relatively small sample, it is critical that we pick data from jurisdictions that are very similar to California in terms of healthcare quality, average age, smoking rates, etc. Otherwise, we will substantially degrade the accuracy of the model.
7. What are the criteria for lifting the statewide stay at home order?
How will the state determine whether the stay at home order remains necessary? Presumably, the key factor will be whether there is sufficient healthcare capacity to begin gradually relaxing stay at home mandates while maintaining a flattened curve. But the government should be clear on the criteria it is going to use and the amount of capacity it deems necessary before the order will be lifted.
8. When will the state begin assessing whether those criteria have been satisfied?
The data needed to assess the effectiveness of the statewide stay at home order will be available beginning April 3rd. By April 17th, we will have the data we need to assess the maximum effectiveness of the stay at home order (i.e., the maximum rate of flattening of the curve). When will the government complete its analysis and inform the public whether the stay at home order will be lifted? We are asking Californians to make enormous sacrifices to ensure access to the healthcare system for COVID-19 patients and others who are seriously ill. The public deserves to know how long this period will last.
9. What is the long-term plan for managing the COVID-19 pandemic?
There is a popular misconception in the media and public that there are essentially only three options for managing this pandemic: (1) do nothing; (2) perform widespread and random testing with contact tracing and quarantining; or (3) issue stay at home orders for virtually all residents. We know that option 2 is unlikely in the U.S. given the limited availability of test kits and the fact we already have many cases. But this does not mean we default to option 3. Social distancing measures exist on a spectrum from not shaking hands (least social distancing) to everyone staying at home (maximum social distancing). It is quite possible that a combination of moderate and tailored social distancing measures may be sufficient to keep COVID-19 cases from overwhelming the healthcare system (and the data from March 23rd to April 2nd may help demonstrate this). We should insist that the government develop and model alternatives on the assumption that the stay at home order will successfully flatten the curve. We need to enter our “new normal” phase of managing the virus as soon as possible to reduce the massive collateral costs of near-maximum social distancing.