As COVID-19 spreads around the world, more and more things such as conferences, schools, and large events such as SXSW are being canceled – an effort to halt the spread of the virus and reduce the strain on our healthcare system. Meanwhile, dangerous notions circulate: the idea that nearly everyone will get COVID-19 so distancing measures are irrelevant or the idea that we might as well just get it and be done with it.
In the last few days, different versions of a simple, yet powerful, graph show us why it’s worth trying to keep the number of cases low. The takeaway: protective measures, such as self-isolation and canceling large gatherings, will delay and decrease the outbreak peak, reduce the burden on hospitals at a given time, and decrease the overall number of cases.
But do social distancing measures actually work?
1. A very short thread on the power of data graphics and scientific communication.
Roughly a week ago, some very smart person* sat down, drew this graph, and saved lives.
(*It’s 2 AM. Without an economist subscription, I can’t quickly discover whom. Maybe someone can help.) pic.twitter.com/eU71Eu60eS
— Carl T. Bergstrom (@CT_Bergstrom) March 6, 2020
Social distancing during the 1918 influenza pandemic
To answer this question, let’s take a look at the response of several cities during the 1918 influenza pandemic. A 2007 paper in PNAS documented the effects of the 1918 pandemic in various US cities based on when public health interventions began, what the interventions were, and how many interventions they enforced. Examples of public health interventions include isolation policies, closures of schools, churches, and other venues, bans of public gatherings, and more.
Cities that began interventions earlier had significantly lower peaks of pneumonia and influenza-related mortality. And cities that implemented four or more interventions had a lower median peak weekly death rate (65/100,000 people) versus 146/100,000 people from cities with three or fewer interventions.
The response between Philadelphia and St. Louis made a great case that social distancing does work. In Philadelphia, the first case was reported on Sept 17 and authorities downplayed the significance of the case. They even allowed a city-wide parade to happen on Sept. 28. School closures and bans on public gatherings did not happen until Oct.3, 16 days since the first case. Meanwhile, St. Louis had its first case on Oct 5 and the city implemented social distancing measures two days later.
What was the effect? The 14-day difference in response time between the two cities represents approximately 3-5 doubling times for the epidemic. The peak weekly death rate from pneumonia and influenza-related deaths was 257/100,000 people in Philadelphia. The same metric in St. Louis was 31/100,000.
From the graph, you can see that there was a second peak towards the end of the study period. This occurred only after the city relaxed on intervention measures. No cities in the study experienced a second wave while the interventions were still in place.
Lessons from 1918 influenza pandemic for today
The authors from the 1918 influenza study highlighted some lessons from the cities responses that could be applied to future pandemics:
- The results stress that action from public health officials need to happen sooner rather than later
- Communities that implement more interventions proactively have better outcomes that communities that introduce interventions reactively
- In cases of severe pandemics, cities should maintain interventions longer than 2-8 weeks (the norm in 1918)
Don’t panic, be prepared, and think about collective changes in behaviors can have big impacts.
Coronavirus resources
- This guide from Julie McMurry highlights the state of COVID-19 and some of the things you can do to lower your risk, what not to do, and what to do if you get sick.
- ASM – Novel Coronavirus (COVID-19) Resources
- CDC – Coronavirus Disease 2019 (COVID-19)
- WHO – Coronavirus disease (COVID-19) outbreak
Shouldn’t the legend on the Hatchett et al. graph refer to mortality and not to ‘number of cases over time’?
Yes thanks. Will update
Note in particular how the second graph contrasts an infection rate that exceeds the healthcare system’s capacity versus one that doesn’t. Once the number of serious cases exceeds the capacity to handle how this disease kills—via the Acute Respiratory Distress Syndrome (ARDS)—care will have to be rationed. That means some will die who might have been saved.
Hospitals can’t make additional ICU beds, but they can prepare to repurpose existing beds. Time is probably too short to acquire new ventilators, although they can make sure that all the ones they have are functioning.
Finally, there’s the need for staff who know how to handle all the complexities of ARDS. That’s likely to prove the critical factor. Rooms won’t grow tired nor will ventilators, but overworked staff will make mistakes and can come down with this virus themselves, complicating matters greatly. Now is the time to be training experienced doctors, nurses and respiratory therapists in the specifics of treating ARDS. That can be done in mere days.
OK! The instructions are: 1) spray hair with hair spray. 2) light on fire. 3) run outside screaming Trump will ruin the universe.
U. Toole, thanks for the helpful commentary.
We will be watching for your demonstration.
Were the total numbers of fatalities different?
Ray, about seven times as many people died in Philadelphia as in St. Louis.
Here are the numbers with the references they came from:
Philadelphia: nearly 12,000 city residents died.
https://www.history.com/this-day-in-history/flu-epidemic-hits-philadelphia
St.Louis: 1700 people died
https://www.stltoday.com/news/archives/st-louisans-died-from-the-spanish-flu-in-the-city/article_aa4b3141-696f-5849-bf37-ad1fcacb7f3f.html
Ray, I had the same thought – that the number of deaths per 100,000 looks about the same, but that they were more clumped together versus spread out over time. Admittedly, spread out allows the health care system to handle the load better.
I think that is the overarching point. What the system can bear.
” I had the same thought – that the number of deaths per 100,000 looks about the same”
Actually if you look closer, Philadelphia had almost 2.5x the per capita death rate of St. Louis. Much of it probably due to overwhelming the medical system with the huge early spike in infection when the peak weekly rate in Phila was 8x St. Louis.
I will be sure to post this information to all my woke friends.
#1 Community Health Pro-tip = Don’t be Racist.
About every 10/15 years there is some sort of medical crisis. I remember in the 30’s it was the Whooping Cough. I was about 5 years old, I knew for sure I was going to die. In our neighborhood every other person had it. We got home made medication from our Mother, not from a real Doctor. We did not have TV or internet, so I do not remember if anyone actually died. I’ve had a lot of childhood illnesses, but a vaccine came along, and solved the problem. Twenty years from now people will be talking about a new virus, but it may be more serious than the one we are dealing with now.
You mean when a high school level education was on par with a modern Bachelors degree? oh wait that only brings us to 1980.
“Social Distancing” needs a definition, is it 6 feet, 10? Is it self-quarantine at home? Isolate for a day, 2, 1 week? What does it mean?
What was the population breakdown of those two cities based on age? What is the age breakdown of infections and deaths? Industry, smoking rates, air quality? Available Healthcare? Are these all equal for these 2 regions?
This article leaves me with far more questions then Answers.
Here’s an article that may answer your question about Social Distancing:
https://www.mnn.com/health/fitness-well-being/stories/social-distancing-definition-and-why-it-matters
If you think those are important, Confused, please look those up and get back to us. They are all facts that are available.
First of all, the original graphs were of mortality rate, deaths per 100,000 population, not the number of cases. There is a great difference between how the flu was treated in 1918 versus today. Deaths are primarily due to secondary bacterial pneumonias, which today are treated by antibiotics, which were not largely available in this country until a few years after WWII, and ventilators, which weren’t invented until 1928. Also, the flu hit St. Louis later than Philadelphia when the dangers were better known. No breakdown is given on exactly what measures were taken by each city, the number of relative hospital beds, the general state of health in each city at the time. Philadelphia’s population in 1918 was 1.8 million. St. Louis was only a third of that. Population density makes a great deal of difference in the spread of a virus. So while I am not arguing that NO containment or mitigation efforts be made, the vast difference portrayed in the graphs may not have as much relevance to 2020 as people assume. Just because A follows B does not always mean that B causes A, this is the well known “post hoc, ergo propter hoc” fallacy.
Deborah, the citations in the comment I left above do show some of the measures taken by each city, and more importantly, the timing of those measures. I am sure these are not the only sources to highlight the effectiveness of social distancing in fighting the deadly 1918 epidemic. Overall, the weight of evidence makes it reasonable to believe in this case that A follows B, because B causes A – in other words, that social distancing lowers the rate of infection and death.
“Population density makes a great deal of difference in the spread of a virus.” In denser populations social distancing is more important, which is the point of the article. I saw it first hand, begin to take effect in Italy once the government expanded the red zone to the south. People-Italians (of all people) were in our aging and respecting a 1meter distance. The fear in the south was overburdening an already max capacity system, forcing turn aways and reducing available care for trauma or other illness.
Encouraging- not in our aging.
…at 85 years old, I’m staying home except for walking my dog. Outdoor air is. Probably OK.
20 minutes a day of sunlight is still needed. just don’t go near people and wipe down things you touch often with a good cleaner.
I am a physician, and I am looking for guidance on the duration of a persons ability to be infectious to other people after originally contracting Covid 19. We know that there is up to two weeks from exposure to onset of symptoms, but how long after symptoms arise does a person remain infectious. This is a critical question in our planning, if it is not yet known, it needs to be addressed as quickly and effectively as possible. I
await hearing more. Thank you
Have you seen this?
https://www.medrxiv.org/content/10.1101/2020.02.11.20021493v2?utm_source=fbia
Thank you for this clear and insightful piece. The subtext of your blog is the need for scientific literacy.
Facts matter because viruses aren’t influenced by opinion.
How can you just distance yourself from life? Your family,friends and loved ones. Its hard to just sit around today without out feeling alone ,now we have to stop our daily routines for this virus . When we where born our first senses are taste ,touch ,hear .Now these are being taking away from us .be blessed be safe but live people just live your life without fear because there is only one life and im going to live mines without fear.
Freddy,
Sometimes an appropriate amount of caution is simply prudent. In the case of COVID-19, it is a life-or-death issue for thousands, possibly millions, of people, similar to the 1918 flu that killed many, many people.
There is no need to stop living completely. If you see neighbors walk by, you can open your door to say hello – from a distance. You can call friends and family on the phone. You can use technology to have face-to-face conversations (by FaceTime, Skype, Zoom, and other methods). You can go for walks or to parks – being outside and in sunlight will help. My guess is that in this time of crisis, more people than usual will be willing to give a friendly wave, or have a friendly conversation – from a safe distance. And if your living circumstances allow it, a pet can help a lot to reduce the feelings of being alone.
The graphs are important to emphasize the need to “flatten” the curve during epidemics so that Health Facilities are not overwhelmed. A true representation of mortality difference between cities would also factor into the stats parameters such as; age demographics, population density, socioeconomic status, gender etc etc. Finally to NOT misrepresent the figures, the curves must be extended to the zero number of case deaths on the Y axis and the Area Under the curves for each city and statistical significance testing should be applied to understand whether the difference is significant.
I’ve misplaced a similar piece that contained 1918 epidemic graphs from a larger number of cities. I’d appreciate that being posted by anyone who has it in hand.
Thanks