Thursday, October 30, 2014

Is There Any Risk of Ebola Transmission from an Asymptomatic Person?

by Michael Dorf

Absent breaking news that takes me in a different direction, I expect that next week I'll have another Ebola quarantine column on Verdict that attempts to put some of the legal controversy in a broader context, but here I want to note my frustration over the failure of even the relatively responsible media coverage of the issue to address what seems to be an important question in evaluating the rationality of the state quarantine measures that are stricter than the recommendations of the CDC. The question is this: Is there a non-negligible risk of spreading the virus from an Ebola-infected person who is just on the cusp of developing symptoms? The issue is relevant to the wisdom of the quarantine policies and presumably would also be relevant to potentially imminent litigation between nurse Kaci Hickox and the state of Maine.

According to a literature review on the CDC website, the Ebola virus level in the blood of a person with an ultimately fatal infection peaks about five days after the onset of symptoms, as revealed in the following chart included in the CDC paper:


Now it's important to note that the y-axis is the log of the viral load, which means that viral load increases exponentially. Another way to say that is that if the graph above plotted viral load rather than log(viral load) as a function of time, it would be much more sharply peaked.

According to the CDC and every other source I could find, an asymptomatic Ebola patient is not contagious, while a patient who is several days into the symptomatic phase of infection is contagious to people who are in close proximity, for two reasons: (1) as reflected in the graph above, his viral load will have peaked; and (2) the later symptoms include vomiting, bleeding and other discharge that will more likely expose others to infection.

These facts seem to support the CDC approach as against the stricter quarantines in Maine, New Jersey, and some other states. If a person cannot spread the disease before developing symptoms, the logic goes, then he or she should not be subject to the serious deprivation of liberty of a quarantine (even a home quarantine) until the onset of symptoms.

But we might worry about what the patient does while not quarantined. Under the CDC protocol, a patient takes his temperature twice per day, and if he detects a fever within the 21-day incubation period, he calls himself in for treatment. (This was the protocol followed by Dr. Spencer in NYC.) Let's suppose that a health care worker returning from Liberia follows this protocol and gets a normal temperature reading at 8 am. She goes out about her daily routine, returning home to take her temperature again at 8 pm. If the health care worker was just on the cusp of developing symptoms right after she took her temperature, her viral load could roughly triple before she takes her temperature again.

Here's why: As I read the chart above, the log(viral load) just before symptoms develop is 4.6. A day later, the log(viral load) is 7.2. Thus, (assuming linearity to first order) 12 hours after symptoms develop, the log(viral load) is 5.8. That's a change of 1.2 in log(viral load), meaning that the viral load itself more than triples (because e to the 1.2 power is 3.32.)  If you didn't follow that, just trust me. I used to be a scientist.

Note that the detection threshold is lower than the viral load at the onset of first symptoms, so it's also possible that our hypothetical infected patient would have tested negative for Ebola just two days before the onset of symptoms.

Accordingly, it is possible for a health care worker to have a negative Ebola test on (say) Day 15, a normal temperature reading on the morning of Day 17, and yet have a sufficiently high viral load that her bodily fluids would spread the disease if they came in contact with others while she is out and about on Day 17 unknowingly having Ebola.

It still seems quite unlikely that such a person actually would have her bodily fluids come into contact with others, but even that's possible. Suppose she is in a car accident and paramedics responding to the scene (and not taking Ebola precautions because they do not know that she is infected or even at risk of being infected) have contact with her blood. Again, it's highly unlikely that any particular Ebola-exposed individual would be in such an accident on just the one day when her viral load goes from undectable to detectable and symptomatic, but it's not impossible.

Whether that small but non-zero risk of transmission from an asymptomatic Ebola-exposed individual justifies a quarantine depends on a number of factors, including, of course, the infringement on liberty and the deterrent effect of an overbroad quarantine on health care workers going to west Africa to help the people most at risk (both for their sake and for the sake of people in the rest of the world who benefit from stopping Ebola's spread at the center of the outbreak).

In addition, we might also consider population density in evaluating whether a quarantine can be justified. Contact tracing has to be much easier in Fort Kent, Maine than in New York City, northern New Jersey, or Chicago. The fact that Hickox is known to people in her community will mean that in the extremely unlikely event that she is in an accident of the sort described above in the crucial window of time, local authorities will know who has been exposed. By contrast, with orders of magnitude more people, the likelihood of any particular person exposing others is greater in a high-population density locale, as is the difficulty of containing that exposure. Accordingly, even if the New York, New Jersey, and Illinois quarantine measures are ultimately unjustified, they do seem easier to justify than the one in Maine and similar places.


***** Update: A reader points me to a very interesting study that found that in West Africa, risk of general transmission is much higher after Day 4 of symptoms than earlier. The finding is consistent with my analysis above because the study does not say (and as an epidemiological study probably cannot say) whether the effect is due to the later symptoms that make it much more likely for body fluids to come out of an infected patient or due to higher viral load. I pretty strongly suspect it's the former. But whatever the mechanism, as the authors conclude, if waiting for the onset of symptoms before isolating people is an effective strategy for containing Ebola in Africa, it's even more likely to be effective here, even in cities.