Flattening the Curve: An Analysis
Like many of you, I have spent the last several days learning as much as possible about COVID-19. An overwhelming number of articles, graphs and analysis come to my inbox each day, both from my home country (Spain) and from the United States, where I work now.
Being able to identify what is true from what is not and separate the emotion from the facts supported by data becomes harder in situations like this one. However, I have found that understanding data and being capable of creating models helps me get true insight into what is happening, and how we could help others quickly stop the spread.
My colleague Armaan Shah published a dashboard that visualizes COVID-19 data provided by Johns Hopkins University. Using this dashboard, I was able to quickly get insight into the COVID-19 data and model some curves that show the spread across different countries.
At Kenway, we always talk to our clients about using information from their data to make informed, proactive decisions instead of reacting, and why that is so critical for business. Keeping that advice in mind, I want to share my analysis with everyone to stress why it is so important to stay home (if you can).
Flattening the Curve
We have heard this phrase a lot over the past week. We have seen several drawings with two curves, and explanations about why we need to flatten the curve. I want to first start with the “Why.”
The curve shows the rate at which the virus spreads and how it evolves over time. When the curve flattens, the health system (which is not prepared for this pandemic) is better positioned to take care of the number of cases coming in. Also, personal protective equipment (PPE), which is now in short supply, can be produced at a pace that matches the number of cases coming in so that hospitals can receive new shipments that will ultimately prevent doctors and nurses from being exposed and therefore quarantined. A reduction to the spread pace also translates into more time for researchers to develop to a vaccine.
On February 2, China started their national lockdown. The slope of the curve was already steep, and it kept increasing for the next 15 days.
On February 19, after 17 days, the curve started to flatten. Since then, the number of cases has increased at a slower pace. On March 19, for their first time since the pandemic began, China did not report any new cases and hasn’t reported any since. Implying that the lockdown proved beneficial.
The Chinese curve has now turned into a downtrend. The implications of this modeled forecast appear to bring some hope to other countries that are still in the first phase of this crisis, as the number of Chinese cases continue to appear to be decreasing.
*Note that the modeled chart includes a 30 days analysis, within a 90% confidence interval.
South Korea has also been able to flatten the curve. South Korea’s approach was slightly different than other countries: the curve was flattened by increasing the number of tests and by a strict tracking of the population. In fact, these measures allowed them to identify the first 30 cases and isolate immediately, limiting the spread. However, the case was not identified on time and did not self-isolate (despite the presence of a high fever), therefore spreading the virus to approximately 1,000 more individuals.
Since then, South Korea has approached the flattening by a strict monitoring of exposed people to COVID-19, tracking credit card movements and location to ensure that the spread does not keep increasing.
COVID-19 has hit some European countries hard, especially Spain with close to 30k confirmed cases and Italy with almost 60K.
Italy and Spain have both implemented lockdown measures that started March 9 and March 15 respectively. Meanwhile, Germany ordered closures but has not yet reached “shelter-in place.”
If we use the Chinese curve model (as explained above), Italy and Spain should expect to see results from their measures by the end of March. The curve would decrease its steepness closer to that date, and the number of cases would stabilize.
It is important to notice that, currently, there is just one available model (the Chinese curve) and it is heavily impacted by measures taken in each country. Stricter measures would potentially result in a faster turn on the steepness of the curve. Therefore, this model may change from country to country, but it can be used as a reference to get insight and make preventive decisions.
How is the United States doing? U.S. cases started increasing at a rapid speed later than in Europe or China. However, the current curve appears to be steeper than the Spanish curve and follows a similar path as the Italian one.
As of late, the number of U.S. cases has been dramatically increasing. The bright side is that U.S. companies and state and local governments started social distancing policies earlier in the spread timeframe than some European countries, and certain states (including California, New York and Illinois) have implemented lockdown measures.
Trend analysis shows that the number of U.S. cases will continue to increase for the next 15-20 days following an exponential curve. Using the Chinese model as a guide, if measures are strictly followed in the United States, we could potentially see the U.S. curve flattening sometime around mid-April. If not, the curve will continue to increase at its current pace.
The most important takeaway from my analysis is that the United States needs to anticipate outcomes and implement preventive, rather than reactive, measures toward fighting this pandemic. Analyzing existing data from other countries will help get us speed to insight and enable us to make the best decisions in a timely manner – saving lives in the process.
Below please find the full dashboard that allows you to interact with data. If you have any questions, contact us at firstname.lastname@example.org.
Stay safe and healthy!