key: cord-0916426-q2qrkoau authors: barnett, a. title: Covid-19 Risk Among Airline Passengers: Should the Middle Seat Stay Empty? date: 2020-07-05 journal: nan DOI: 10.1101/2020.07.02.20143826 sha: 3c6e863d5cc5bd45b717b25e4616aa00653d2a12 doc_id: 916426 cord_uid: q2qrkoau Recent research results and data generate the approximation that, when all coach seats are full on a US jet aircraft, the risk of contracting Covid-19 from a nearby passenger is currently about 1 in 7,000. Under the middle seat empty policy, that risk falls to about 1 in 14,000. Risks are lower in flights that are not full. These estimates imply Covid-19 mortality risks to uninfected air travelers are higher than those associated with plane crashes but probably less than one in one million. Introduction As of 7/1/20, the US air carriers American, Spirit, and United Airlines will fill all seats on their flights when demand warrants, while Alaska, Delta, jetBlue, and Southwest Airlines will keep middle seats empty. While Delta Airlines plans to continue its policy well beyond September 2020, United Airlines CEO Scott Kirby stated that there is no such thing as social distancing on a plane, implying that limited distancing confers no real benefit compared to none. What does the evidence suggest about the wisdom of a "middle seat" policy as a safety measure? Answering that question entails major complications and uncertainties, which can easily lead one to throw up one's hands. But even a rough approximation of the risks at issue seems preferable to clashes of unsubstantiated conjectures. This paper strives for such an approximation, with an emphasis on the word "rough." To estimate the risk to an uninfected passenger from a passenger experiencing Covid-19, it is necessary to consider three questions: • What is the probability that a given passenger on board is contagious with Covid-19? • What is the probability that universal masking can prevent a contagious passenger from spreading the disease? • How does the risk of infection depend on the locations on the aircraft of both the contagious and uninfected passenger? . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 5, 2020. . https://doi.org/10.1101/2020.07.02.20143826 doi: medRxiv preprint The general formula for combining the answers to these questions is: where P= the probability that a particular uninfected passenger contracts Covid-19 during the flight Q = the probability that a given passenger on the flight has covid-19 (It is assumed the Q is small enough that having two or more contagious passengers near the uninfected one is a remote risk.) QM = the probability that universal mask-wearing on aircraft fails to prevent transmission of Covid-19 QL = the conditional probability that a contagious passenger transmits Covid-19 to the uninfected one QL and thus P can depend on whether the operating policy is "fill all seats" or "middle seat empty" The Estimation of Q For a given passenger from a particular American state, the risk of contagiousness is estimated in several steps: • First, one finds N7, the number of confirmed new Covid-19 infections in that state over the last seven days (1) Seven days is chosen because that is the approximate length of the contagiousness period for someone experiencing Covid-19. (The average such period is a bit below seven days in asymptomatic cases and higher than seven in symptomatic ones.) • Then, in accordance with recent findings from the US Centers for Disease Control (2) , one multiplies N7 by ten to approximate the actual number of new infections in the state over the previous week. • Then one recognizes that people with Covid-19 who board airplanes are presumably either asymptomatic, pre-symptomatic, or mildly symptomatic. (Those with severe symptoms are unlikely to be flying.) Because of evidence that asymptomatic Covid-19 carriers are only . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 5, 2020. . https://doi.org/10.1101/2020.07.02.20143826 doi: medRxiv preprint about half as contagious as others (3), one multiplies the prior product by a factor of ¾. (This factor of ¾ arises if one assumes that asymptomatic Covid-19 passengers constitute about half of those Covid-19 passengers who board a flight, while the other half have the usual level of contagiousness.) • Then one multiplies by a factor of ½ to reflect the likelihood that passengers who fly are generally more affluent (and less likely to encounter Covid-19 risks) than the citizenry at large. • Finally, one divides by NPOP, the state's estimated population in 2020, yielding N7/NPOP as the state's per capita rate of new confirmed cases over the last week. The estimate of Q consistent with these specifications is: For Texas, which has a high number of recent Covid-19 cases, + as of 6/28/20 was 38,642, while 565 was 29.1 million. With those numbers, ≈ : ;<: . In New York, which is well past the peak of its epidemic, the corresponding numbers are + = 5,523 and 565 = 19.5 million , . ℎ & For QM, a meta-analysis in The Lancet (4) estimated that mask wearing cuts transmission risk given contagiousness from 17.4% to 3.1%, a reduction of 82%. Ignoring the possibility that the masks under study were more effective than those worn by airline passengers, one can estimate QM as 1 -.82 = .18. This quantity depends on the airline's seating policy, as well as the . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2020. Using similar reasoning, one can likewise determine that: ' ( ) ≈ .282 under "fill all seats" . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2020. Readers uncomfortable with this assumption can linearly adjust the results presented, in a form of sensitivity analysis. If one believes, for example, that (2) is appropriate for one hour of exposure, then one might consider doubling the transmission-risk estimates used here. The various estimates can be combined via (1) . Even on a flight from Dallas to New York City, there will be Texas natives, New York natives, and transfer passengers who originated elsewhere, so a mid-range estimate seems suitable. As noted, QM is estimated as 0.18, while . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2020. The first of these risk estimates is the average for passengers in the six filled seats in each row. The second is the average for the four seats occupied under "middle seat empty." For a coach passenger who is infected on a full flight and has a 1% chance of dying from the virus, then the mortality risk based on the estimates above would be about 1 in 700,000 under "fill all seats" and about 1 in 1.4 million under "middle seat empty." Both these estimates are higher than the risk of perishing in a US air crash unrelated to Covid-19, which is about 1 in 34 million (6)). However, data from late June 2020 imply that approximately 1 in 120 Americans have Covid-19 on a given day (i.e., 40,000 confirmed cases per day x 10 x 7 days is about 1/120 of the US population of 330,000,000). Thus, it is not at all clear that the risk of getting infected during a flight is any higher than the risk associated with everyday activities during the pandemic. Moreover, there is an extremely important caveat to these calculations: they are all contingent on a flight being as full as possible. Even on airlines that will fill every seat if they can, many flights will operate with empty seats. Thus, middle seats (and others) could often be empty even without a policy requirement. It follows that a 1 in 700,000 mortality risk is an . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2020. . https://doi.org/10.1101/2020.07.02.20143826 doi: medRxiv preprint upper bound to the actual risk to a randomly-chosen passenger; one suspects that actual death risk does not exceed one in 1 million. The risk estimates presented above do not consider the possibility of infection during boarding and leaving the plane, from contagious passengers who walk down the aisle to the lavatory, or within the lavatory itself. There have been reports that, even during the pandemic, passengers form crowds trying to enter and leave the plane quickly (e.g. (7)), which could pose risks beyond those considered here. (2) suggest that the factor of 0.13 in Equation (2) is too low, and thus that the overall risk estimate P is too low. It is possible though not certain that these two opposite effects largely cancel one another. This analysis offers a baseline risk estimate using the Chu et al. results at face value, which is a reasonable starting point absent more detailed information about the nature and duration of the exposure to someone contagious. Calculations like the ones here are highly approximate and, as has been evident during this pandemic, projections about it often fall far of the mark. It is therefore all the more . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 5, 2020. . https://doi.org/10.1101/2020.07.02.20143826 doi: medRxiv preprint important that attempts be made to use actual passenger outcomes to estimate what fraction of travelers contracted Covid-19 on their flights. If, averaged over US carriers, the risk level per passenger is estimated as (say) 1 in 12,000, then approximately 50 cases of Covid-19 should emerge each day at a time (like now) when 600,000 passengers are traveling daily. Determining how many such cases actually arise will not be easy: travelers who get asymptomatic Covid-19 (especially younger ones) may never know it, while some passengers who subsequently get Covid-19 may have been infected elsewhere than the airplane. Collating records over widelydiverse localities would be challenging. But when safety is at stake, it is worth some effort to substantiate or refute projections that are tied to strong assumptions. The calculations here, however rudimentary, do suggest a measurable reduction in Covid-19 risk when middle seats on aircraft are deliberately kept open. The question is whether relinquishing 1/3 of seating capacity is too high a price to pay for the added precaution. Age, Sex, Existing Conditions of COVID-19 Cases and Deaths, data for confirmed cases by day in American states Seroprevalence of Antibodies to SARS-Cov-2 in Six Sites in the United States Defining High-Value Information for Covid-19 Decision Making Physical Distancing, Face Masks, and Eye Protection, To Prevent Person-to-Person Transmssion of SARS-Cov-2 and COVID-19: A Systematic Review and Meta-Analysis The Risks: Know Them, Avoid Them Aviation Safety: A Whole New World? Five Flights in Four Weeks: What It's Like to Fly During the Coronvirus Pandemic I am grateful for immensely valuable suggestions from Edward Kaplan, Richard Larson and Amedeo Odoni.