Friday, April 3, 2020

Why We Need Face Masks and More

Many countries, including the US, have started to recommend the use of facemasks by everyone in public places. This post explains why this may have a very large impact on managing the epidemic, especially when supported by other measures. Let's start with a graph that shows some simulation results:
The graph shows the total number of predicted deaths in the US from the COVID-19 epidemic under three different scenarios. This time, I used a public website to create the simulations, and I will show you how you can use the website yourself to create "what-if" scenarios.

The first scenario is the "do nothing" scenario, which would lead to more than 2 million deaths from COVID-19. The second scenario assumes that we can reduce virus transmissions that lead to new infections by 60%.  In this particular computer model, that would still lead to almost one million deaths over the course of one year. You can see that such measures would lead to a "flattening of the curve":
But have closer look at the green bar in the first picture: if we can reduce the transmission rate by 80%, then the number of predicted deaths drops to 150,000 - still a large number, but a huge drop compared to the other scenarios.

In reality, it is likely that the differences would be even larger, because the models above do not take hospital capacity into account. Severe COVID-19 cases need intensive care and ventilators, and if not enough ICU beds and ventilators are available, death rates increase dramatically. The following graph shows how many days in a row the number of new ICU admissions would exceed 10,000:
Note that 10,000 daily admissions is likely to be way more than the hospitals in the US can handle, since patients typically require ICU care for at least a week. The total number of ICU beds is somewhere around 100,000, but most of these beds are typically in use,

An interactive computer model

So, let us look more closely about how I got to the results above. I used the simulator on the web site - here is a screen shot:
 From what I can tell, the site uses a pretty standard epidemiological model. What is quite nice is that you can easily change parameters on the left, and see the results on the right. For my simulations, I left most parameters at their defaults, except for these:
  • Population section
    • Population size: 327 million (to match the US population)
    • Initial infections: 1000 (this has no effect on the results, other than making numbers go up faster)
  • Severity section:
    • Sick patients that are hospitalized: 10% (close to actual numbers in the US)
    • Sick patients die from the disease: 1% (in the middle of common estimates)
  • General Contact Reduction section:
    • General contact reduction: 0%, 60%, and 80% for the three scenarios
    • Contact reduction begin: 70 days (matching the total deaths a few days from now)
    • Contact reduction days: 365 (must be set; short numbers let you see "second flare ups" if measures are only temporary)
I encourage you to go to the web site and play around with it yourself, but please keep reading a bit more to understand a few important issues.

"Contact reduction": a dangerous name

The only parameter I changed was the "general contact reduction" number. The term "contact reduction" is often used in this context, but is dangerous because it is very easily misunderstood. What it really means is this:
The reduction in probability that an non-infected person becomes infected.
Now if the only way to become infected is by direct contact,  then calling this parameter "contact reduction" makes sense. But this is clearly not the case for COVID-19; there is an increasing body of evidence that COVID-19 can be transmitted without close contact through aerosols and "fomites", as I explained in a previous post. A very recent study from researches in Nebraska provided further evidence of this, and confirms what researchers in Singapore had previously reported.

What makes the term "contact reduction" dangerous is that it creates the illusion that simply avoiding close contact should be sufficient. If all we have to do is to reduce close contact by 80%, it should be sufficient if people reduce their contact with others by 80%, right? In the case of corona virus, this is clearly wrong. The corona virus can also be transmitted by "long distance transmission", so avoiding close contacts is not enough, and the effect of social distancing can very easily be over-estimated.

Unfortunately, there are multiple examples where researchers published computer studies and fell into this trap. Even the World Health Organization is still clinging to transmission ideas that have originated in the 1950s, and do not reflect what we have learned in the recent years about virus transmission in general, and the SARS-CoV-2 virus in specific.

In my eyes, using the term "contact reduction" to describe the effect on reducing transmissions in COVID-19 is not just dangerous - it is deadly because it leads to delaying actions that can reduce long-distance transmissions.

Getting to 80% transmission reductions: how facemasks can protect you

I talked about facemasks in a separate post, but let's quickly recap how even self-made facemasks can protect you:
  • They keep you from touching your mouth and nose with fingers that have picked up virus, for example when you touch a door handle while shopping, or a touch pad when paying by debit or credit card
  • When other people wear facemasks, the amount of virus they spread is significantly reduced, which makes it less likely that you touch a virus-contaminated surface at work or when shopping
  • While self-made facemasks and surgical masks are not great at removing virus from air you breathe in, they do remove some of the virus-containing droplet. The level of protection from this is likely to be a lot lower than from the two previous points, but it is larger then zero.
Let's look at this from the point of mathematical probabilities and computer models. If we assume that 60% of COVID-19 infections are from close-contact transmissions, and 40% are from long-distance transmissions, than the maximum effect we can get from avoiding close contact is a 60% reduction. So even with perfect social distancing, that's all we could hope for. The transmissions would drop to a level of 40%, which is a great step, but still leads to many, many deaths and a long period of suffering.

To reduce transmissions below 40%, we need additional measures that specifically address long-range transmission. Facemasks are one example of such measures: by absorbing virus-laden droplets from breath, coughs, and sneezes, and reducing fomite-base transmissions, they are likely to help in reducing long-distance transmissions. If they were effective in reducing long-range transmissions by 50%, that would reduce overall deaths in our example by hundreds of thousands.

How to reduce transmissions by 80% (or more)

While computer models like the one I used here have some limitations, there is absolutely no doubt that we need to achieve a reduction in virus transmissions by 80%. Many countries have tried to contain the epidemic by focusing largely on social-distancing type measures, but have had very little success. In contrast, countries and regions that had more success used a vast array of "stop the spread" measures, including measures that specifically affected long-distance transmissions. Here is an (incomplete!) list of some of the measures that must be at least seriously considered to stop the spread:
  • General use of facemasks, with the goal of making higher-efficiency masks like surgery masks available to the entire population, once the issues with sufficient high-quality masks for medical staff are solved
  • Infection stations at public places, including multiple locations in grocery stores, restaurants, and public transportation
  • Frequent general disinfection of public areas, including subways, buses, and work spaces
  • The use of UV lamps for overnight disinfection of work spaces, similar to what is used in laboratories
  • Anti-profiteering laws for health safety items like facemasks
  • Nationally coordinated distribution of facemasks and disinfectant until they are available in sufficient quantities.
The list is largely based on what countries like Taiwan have done to effectively contain COVID-19. It certainly can and should be expanded to include other measures. Perhaps implementing many such measures would be overkill to reach the goal of reducing transmissions by 80% - but if so, the only effect would be that the epidemic would stop sooner. But the alternative of potentially not implementing sufficient measures as soon as possible will lead to tens or hundreds of thousands of avoidable deaths from COVID-19.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.