Error on the side of caution, but caution must be backed by science

My initial motivation for this post comes from the sequence of announcements by CDC about whether Ebola can be transmitted through the air instead of direct contract with body fluids (wet areas on some surface).   After initially declaring that the transmission was only possible by touching fluids from the body, they later released an announcement that fluids can be expelled from sneezing and coughing but the latest announcement is that there is no evidence that transmission occurs through sneezes and coughs.

Their current Q&A page states the following:

Can Ebola be spread by coughing or sneezing?

There is no evidence indicating that Ebola virus is spread by coughing or sneezing.  Ebola virus is transmitted through direct contact with the blood or body fluids of a person who is sick with Ebola; the virus is not transmitted through the air (like measles virus). However, droplets (e.g., splashes or sprays) of respiratory or other secretions from a person who is sick with Ebola could be infectious, and therefore certain precautions (called standard, contact, and droplet precautions) are recommended for use in healthcare settings to prevent the transmission of Ebola virus from patients sick with Ebola to healthcare personnel and other patients or family members.

This statement supports the conclusions that it is safe for people to use public transportation and for non-symptomatic Ebola infections to be free from quarantine or isolation.

If Ebola could spread by coughing or sneezing, then people would have legitimate concerns about being in public where Ebola-exposed individuals may be present.   The policy would have to change to recommend quarantine and isolation for exposed individuals.

These policies depend on the fact of whether coughing and sneezing can spread the virus.  Because we have no evidence that this is possible, we should not impose safety measures of quarantining non-symptomatic persons who have been exposed to Ebola.   We have no reason to be concerned about sharing public spaces with people who have been exposed to the virus.

I want to explore the meaning behind the initial statement that “there is no evidence”.   This seems to place the burden of proof on the proposition that sneezing or coughing that can spread the virus.   To say there is no evidence means that for now we reject this hypothesis because the evidence has not yet met the burden of proving it to occur.

We recognize that Ebola virus becomes abundant in the blood before symptoms first appear.  The lungs are optimized for close contact between the blood and the air.   The fluids along the respiratory tract have close proximity to infected blood.  There are multiple fluids involved including those in the mouth and nose.   There are plenty of opportunities to transport the virus by ways of sneezing and coughing, but so far we lack sufficient documented evidence that this has actually occurred.

This lack of evidence may partly come from our bias toward direct contact explanations.   If someone who contracts Ebola has in the past come in direct contact with a symptomatic patient, we conclude the infectious was the result of that contact.   Evidence of direct contact eliminates the need to consider the coughing or sneezing may have been involved.

In order to prove coughing and sneezing transmits the disease, we need to identify an infected individual who certainly had no direct contract but observed being nearby during a cough or sneeze.   On human terms, it is very likely that such a person would at some point come in direct contract with the earlier infected person and that contact would be ruled as the cause of transmission, thus eliminating from consideration the evidence of sneezing or coughing.

For whatever reason, we lack evidence that coughing and sneezing transmits Ebola virus.  Without this evidence, we can not prove that coughing and sneezing can transmit the virus.  Evidence-based decision-making obligates us to make decisions that are justified by scientific proof.  Thus we have no basis to raise concerns about mingling Ebola-exposed and unexposed people.  As of now there is no evidence that there is any risk of transmission from a public cough or sneeze.

The scientific determination of whether coughing and sneezing can transmit Ebola virus can be used in different contexts.

For example, one may want the scientific explanation to incorporate into computer models to estimate how quickly and widely the disease will spread in the future.   In that case, we want the best case scenario that coughing and sneezing (or the more dangerous airborne mechanism) does not contribute to the spread.   With this assumption, the models are predicting very troubling projections for how serious this epidemic will become within a few months.    Even without sneezing/coughing transmission, simulations show that this is a dangerous disease.

Alternatively, we may want to set public policy about whether to impose quarantines on exposed populations.   In this case we seek to optimize economic outcomes that encourage normal commerce without fear from contracting the disease from nearby individuals who may have been exposed.

These objectives have very different considerations, one for resource planning for future and the other for protecting the current population.   However, we demand that both decisions should be based on the same scientific evidence.   If science can not justify including coughing and sneezing as a transmission route for simulation modeling of future Ebola cases, then it can not justify imposing present-day quarantines on exposed individuals.

My personal opinion is that this does not make sense.   We need a different kind of scientific conclusion for these two different types of decisions.   For the purpose of including coughing and sneezing in our computer models, we want the burden of proof that coughing and sneezing does transmit the disease.   For the purpose of setting the current day quarantine policies, we want the burden of proof that coughing and sneezing does not cause transmission.  These decisions use two different scientific conclusions where neither of the contradictory premises can be disproved.

I think the presumption that coughing and sneezing can spread the virus is intuitive.   As mentioned, coughing and sneezing involves body fluids from the respiratory tract that brings air very close to the blood supply.  The mechanics of transmission are present.   To justify the alternative presumption we need some explanation of how the respiratory tract has some naturally disinfectant property for the virus.

The purpose of a public policy is to protect the population.  In the case of the Ebola virus, the purpose is also to protect out healthcare system that can easily be overwhelmed with just a few dozen cases.   With these stakes and with the knowledge of probable mechanisms for propelling infected droplets from coughs and sneezes, it is reasonable to error on the side of caution.  Instead we are told that the such policies are invalid because they contradict the science we use in our simulations.

Modern policy practice appears to demand there being just one scientific observation that must be used both for planning purposes (in simulations, for example) and for present day policy purposes (to protect the public).   The science has already been settled for the purposes of simulation of future spread of the epidemic.   That science failed to prove with high confidence that coughing and sneezing can transmit the disease.   We must set present day cautionary policies using this same conclusion we used for simulations.

Until the evidence exists that coughing and sneezing transmits the virus, our policies must be consistent with the fact that coughing and sneezing can not transmit the virus.   We need to see actual transmissions by coughing and sneezing before we will allow policies designed to limit this risk of exposure.

This reminds me of my last post about the minimum viable product.   The concept of the minimum viable product is put something into production in order to gather real-world data of its failings.   We need the failures to provide the data to improve the next iteration of the product.   In my last post, I described how this works out with modern commercial rocket launches where launch failures are a form of success in that it provides data for what can go wrong.

In this case, the minimum viable product is a policy the forbids quarantines of Ebola exposed individuals, and advises us of no risk to participate normally in public and commerce.   The data we are looking for is actual transmissions that can only be explained by the sneezing and coughing route.   When that happens, we will have the proof we need to come out with the new-improved product of an updated policy that advises more precautions in public and recommends quarantines of exposed individuals.    The minimum viable product succeeds best when it is fails because that is how we obtain the evidence we need to make a better product (in this case a policy).

As I mentioned in the last post, we place a primacy of observed data in real world operation over our faulty ability to figure out what may happen from first principles.  First principles of rocketry may be how fuels burn and fluid dynamics of the burning fuel and the surrounding air.   First principles of Ebola in the body is that virus is in the blood, the blood is very close to the surface inside the lungs, and coughs and sneezes can propel droplets for significant distances.  In both cases, our modern expectation is to dismiss our ability to figure things out from first principles but instead demand observation evidence in the real world from observing the minimum viable product in production.

In the case of the two recent rocket failures, the failures that delivered to us this superior form of conclusive data resulted in one case the loss of resupply mission (with impacts on space station operations) and in the other case the loss of at least one human life.

In the of the minimum viable product of Ebola public policies, the failure that will deliver us the superior form of conclusive data that confirm our first-principle suspicions that coughing and sneezing can transmit the virus.   The cost will be the new patients who provided us this valuable service of giving us the evidence we need for a strictly evidence-based decision making for setting public policies.   By the time we figure this out, there may be a few thousand such individuals, but their suffering is for the public good of adhering to our evidence-based decision-making ideals.

This is another example of what I described earlier as the population’s obligation to participate in order to support the decision-maker’s obligation to follow the recommendations supported by data.   The initial policy of avoiding quarantines is another form of obligation to endure suffering for the sake of gathering evidence.

I argued in these earlier posts that we make a choice between automated decision making (based on scientific evidence) and human decision making (that allows for doubts and fears).  In that latter, we demand accountability from the human decision maker when we encounter bad consequences from a decision.   This accountability requires the human decision-maker to convince us he made a good decision at the time based not only on the facts but also on reasonable fears and doubts.

In contrast, automated evidence-based decision-making does not involve any accountability.   When we encounter bad outcomes from evidence-based decision-making, we must be appeased by the fact that our suffering has produced new data that will become evidence for future decision making that will avoid repeating those consequences.

In the current policy-making environment, we need some people to contract Ebola in a way that can only be ascribed to coughing or sneezing.  This will provide the greater good of now having sufficient hard evidence to support quarantines of Ebola-exposed individuals.   Their suffering is for a good cause.

In those earlier discussions, I describe two parties with obligations to follow evidence-based decision-making.   Those parties are the decision makers and the population affected by those decisions.  In order to work, evidence-based decision making obligates both parties to cooperate.   In the case of public policies for managing the Ebola epidemic, there are additional parties that are affected.   For example, our entire healthcare system lacks capacity to manage more than a few simultaneous cases of Ebola patients, and it is not clear how long this can be sustained for a succession of cases.

The risk of getting the public policy wrong is the we can suddenly have too many patients at the same time.   This could collapse our health care system not only for its ability to handle Ebola cases but for it to handle it normal operations.   Hospital staff may come down with the disease, or they may refuse to work where it is possible to contract the disease due to lack of appropriate protective gear.   This would affect the entire hospital operation, not just the part devoted for handling Ebola.   The most vulnerable operation of the hospital is it emergency care centers.   One patient arriving to the emergency room before being diagnosed with Ebola can shut down the entire emergency room or discourage people from using the facility in the future.

This case of setting public policy concerning Ebola-exposed individuals illustrates a conflict in the goals of human-accountable decision-making that wants to err on the side of caution, and science-based decision making that demands any caution to be backed by evidence.   Our current environment discounts unsubstantiated fears and doubts of human decision-makers.   The concept of err on the side of caution loses any relevance when we demand that any concept of caution has to meet the burden of proof with scientific evidence.   With such evidence there is no possibility of doubt and thus the decision does not involve making an error.   We demand decisions to be made free of any fear or doubt.  We must allow nature to provide us actual observational evidence for the cautious approach before we can permit enacting the cautious approach.

Implicit in this trust in evidence-based decision making is that we can survive the failure that provides the successful consequence of providing us the evidence we need to justify the more cautious policy.


2 thoughts on “Error on the side of caution, but caution must be backed by science

  1. Pingback: Authoritarianism by data, settled science obligates action with available resources | kenneumeister

  2. An interesting cautionary example is what happened with Typhoid Mary.

    This month marks 100 years since Typhoid Mary was apprehended for the second and final time, living the next 23 years—the rest of her life—under quarantine.


    So why was Mallon alone among carriers imprisoned for life? “That’s the million-dollar question that nobody can answer,” says James Colgrove, a professor of sociomedical sciences at Columbia. It might have been because she was female, Irish, uncooperative and without a family.

    This account illustrates the need to be careful about being too careful.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s