We defined cases of SARS-CoV-2 infection according to Vietnam Ministry of Health guidelines in place at the time of our investigation (12). Specifically, we defined suspected flight-associated COVID-19 cases as passengers or crew members on board flight VN54 landing in Hanoi on March 2 who reported fever and cough, with or without shortness of breath, during March 1–16. We defined confirmed flight-associated COVID-19 cases as passengers or crew members on flight VN54, regardless whether signs or symptoms developed, who had positive SARS-CoV-2 real-time reverse transcription PCR results from nasopharyngeal swab samples (13). Flight-associated cases were considered to have very likely acquired infection on board VN54 and were hence classified as probable secondary cases in this analysis if the following 3 criteria were met: 1) they experienced signs/symptoms 2–14 days after arrival or if they were SARS-CoV-2 positive by PCR 2–14 days after arrival in the absence of signs/symptoms; 2) in-depth investigation did not reveal any potential exposure to SARS-CoV-2 before or after the flight during their incubation period; and 3) they had shared cabin space with the probable index case during the flight (14–17).

At the time of flight VN54 arrival, all passengers from COVID-19–infected areas, including the United Kingdom, had their body temperature screened by thermal imaging and were required to declare any COVID-19 symptoms; only passengers arriving from China, South Korea, Iran, or Italy were required to undergo SARS-CoV-2 testing and 14-day quarantine. At that time, the use of face masks was not mandatory on airplanes or at airports (18).

As soon as the travel history of the probable index case became evident, the passenger list and flight manifest for flight VN54 was obtained from the Bureau of Immigration and the Civil Aviation Administration and sent to all provincial Centers for Disease Control with instructions for local health staff to trace all passengers and crew members of flight VN54. All successfully traced passengers and crew members were interviewed by use of a standard questionnaire, tested for SARS-CoV-2, and quarantined in designated facilities or at home. Any symptomatic person was isolated immediately until the test result was received. In-depth interviews were conducted with all persons with suspected or confirmed flight-associated cases; the specific focus was detecting any potential SARS-CoV-2 transmission events before and after the flight to investigate potential alternative scenarios for transmission other than during the flight. Furthermore, all persons with suspected or confirmed flight-associated cases were asked to identify persons with whom they had had close contact (<2 meter distance for >15 minutes) between arriving in Vietnam and the start of quarantine/isolation. These close contacts were also contacted, tested, and quarantined for 14 days. All persons in quarantine were checked twice daily for clinical signs/symptoms and fever; oropharyngeal swabs were collected on the day of admission, after 3–5 days, and on day 13, unless signs/symptoms developed, in which instance a specimen was collected immediately and the person was isolated and monitored until receipt of the test result.

Initial investigations of the probable index case generated our working hypothesis of in-flight transmission and guided further investigations. In particular, we investigated all possible exposures of all persons with flight-associated cases during their incubation period in relation to the timing of the flight, including locations where flight-associated cases may have crossed paths before and after the flight. To identify factors associated with in-flight infection risks, we calculated risk ratios and 95% CIs.

Among the 217 passengers and crew members on a direct flight from London to Hanoi in early March 2020, we identified a cluster of 16 laboratory-confirmed COVID-19 cases. In-depth epidemiologic investigations strongly suggest that 1 symptomatic passenger (case 1) transmitted SARS-CoV-2 infection during the flight to at least 12 other passengers in business class (probable secondary cases).

Case 1 was the only symptomatic person on board and was the only person with a flight-associated case who had established contact with a person with a confirmed case (her sister) during her incubation period. The incubation periods for all persons with confirmed flight-associated cases overlapped with the timing of the flight (Figure 2). Our interviews did not reveal that any of the additional persons with flight-associated cases had been exposed to SARS-CoV-2 before or after the flight during their incubation periods other than having taken the same flight as case 1, nor did they suggest exposure for any of the 4 travel companion couples after the flight (Appendix). Similar intervals between arrival and positive SARS-CoV-2 test results among 3 case-pairs suggest a common exposure event rather than subsequent infection from one partner to the other. Last, we found a clear association between sitting in close proximity to case 1 and risk for infection (Table 2).

In the absence of genomic analysis, we were unable to completely rule out alternative transmission routes. However, all persons with flight-associated cases departed from the United Kingdom (none transited from other countries); and until the departure date of flight VN54, only 23 COVID-19 cases had been recorded in the United Kingdom. Although testing had not been implemented on a large scale nationwide at that time (19), community transmission in the United Kingdom was not yet widely established (20), making the presence of multiple persons on board incubating the illness unlikely. Similarly, for case 4, who reported having visited India before the United Kingdom during his incubation period, the possibility of preflight transmission remains slim because by March 1, only 3 cases of COVID-19 had been reported in India, although testing in India was still limited (20–22). Furthermore, none of the 30 colleagues of case 4, who shared the same preflight travel history but were all seated in economy class, were infected (Appendix).

We consider local transmission after arrival in Vietnam unlikely. As of March 1, 2020, only 16 cases of COVID-19 had been reported in Vietnam, and 17 days had passed since the last reported case (case 1 reported here became Vietnam case no. 17) (18). At that time, 1,593 persons had tested negative for SARS-CoV-2 infection in Vietnam, and according to official policy at that time, another 10,089 contacts and travelers returning from COVID-19–affected areas overseas were under preemptive quarantine directly at the time of arrival. In early March 2020, there was no evidence of community transmission of SARS-CoV-2 in Vietnam (18). We also note that cases 3 and 14 experienced symptom onset 17 days after flight VN54. Whether these cases reflect unusually long incubation periods or symptoms caused by conditions other than COVID-19 is unknown.

The most likely route of transmission during the flight is aerosol or droplet transmission from case 1, particularly for persons seated in business class (23). Contact with case 1 might also have occurred outside the airplane at the airport, in particular among business class passengers in the predeparture lounge area or during boarding. Although Vietnam Airlines keeps business class passengers separated from economy class passengers during most procedures before and during the flight, contact with the 2 economy class cases might have occurred after arrival during immigration or at baggage claim. We also note that 2 passengers, in the seats between the 2 cases in economy class, were lost to follow-up. Whether either of these passengers could represent a separate index case in economy class is unknown.

The role of fomites and on-board surfaces such as tray tables and surfaces in toilets remains unknown. For example, airline crew often use business class toilets while on board, which might explain the case among the crew serving in economy class, for whom no other potential source of infection could be established. Of note, the temporal sequence of symptom onset among cases in economy class and the crew member serving in economy class also allows for the possibility of a second in-flight transmission event, independent of the cluster in business class (Figure 2).

Our study has several limitations. First, we did not have genomic sequencing data available to support our hypothesis of in-flight transmission. However, the conclusiveness and unambiguity of our in-depth epidemiologic upstream and downstream investigations coupled with extensive laboratory testing make us confident of our main findings. Second, we lacked detailed data on activities of the cases while on board (e.g., movements or seat changes, use of toilets, or sharing meals), which might have enabled us to pinpoint the precise route of transmission. Third, our assessment of passengers’ preflight exposure to other confirmed cases relied on interviews only. Fourth, we had no data available on individual passenger use of face masks while on board, which would have enabled a more refined risk analysis. Face masks were neither recommended nor widely used on airplanes in early March, in particular not among travelers from Europe (24–26), who constituted the majority of passengers on flight VN54. Last, given the delay between arrival and confirmation of the probable index case, no environmental samples could be collected from the airplane.

Our findings have several implications for international air travel, especially because several countries have resumed air travel despite ongoing SARS-CoV-2 transmission. First, thermal imaging and self-declaration of symptoms have clear limitations, as demonstrated by case 1, who boarded the flight with symptoms and did not declare them before or after the flight. Second, long flights not only can lead to importation of COVID-19 cases but also can provide conditions for superspreader events. It has been hypothesized that a combination of environmental factors on airplanes (humidity, temperature, air flow) can prolong the presence of SARS-CoV-2 in flight cabins (27). No evidence indicated that the regular air conditioning and exchange system on flight VN54 were malfunctioning. The number of probable secondary cases detected in our study is on the upper end of hypothesized estimations for SARS-CoV-2 transmission on airplanes in the absence of face mask use, although the movement of aerosols and droplets in the specific conditions of a flight cabin remains poorly understood (27). A study of a COVID-19 cluster with 16 infected flight passengers from Singapore in February 2020 identified only 1 instance of potential in-flight transmission (28). In-flight transmission has been hypothesized but not substantiated sufficiently in a non–peer-reviewed report of a cluster of 10 flight-associated cases in China in February (N. Yang et al., unpub. data, https://www.medrxiv.org/content/10.1101/2020.03.28.20040097v1.full.pdf). In January 2020, no secondary cases were detected after a 15-hour flight to Canada with a symptomatic person with COVID-19 on board (29), although contact tracing and monitoring were limited (30). Similar results with similar limitations have been reported from flights arriving in France (31,32) and Thailand (33) in January and February. All of these studies limited contact tracing to passengers within 2 rows of the index cases, which could explain why secondary flight-related transmission was not detected by those studies.

The latest guidance from the international air travel industry classifies the in-flight transmission risk as very low (34) and recommends only the use of face masks without additional measures to increase physical distance on board, such as blocking the middle seats (7,35). Our findings challenge these recommendations. Transmission on flight VN54 was clustered in business class, where seats are already more widely spaced than in economy class, and infection spread much further than the existing 2-row (36) or 2 meters (37) rule recommended for COVID-19 prevention on airplanes and other public transport would have captured. Similar conclusions were reached for SARS-CoV superspreader events on a flight in 2003, in which a high risk for infection was observed for passengers seated farther than 3 rows from the index case (4). This finding also concurs with transmission patterns observed for influenza virus (38) and is generally in line with the mounting evidence that airborne transmission of SARS-CoV-2 is a major yet underrecognized transmission route (39,40).

Our findings call for tightened screening and infection prevention measures by public health authorities, regulators, and the airline industry, especially in countries where substantial transmission is ongoing (37). Making mask wearing obligatory and making hand hygiene and cough etiquette standard practice while on board and at airports seems an obvious and relatively simple measure (27). Blocking middle seats, currently recommended by the airline industry (7,35), may in theory prevent some in-flight transmission events but seems to be insufficient to prevent superspreading events. Also, systematic testing, quarantine policies, or both, for inbound passengers at arrival might be justified for countries with low levels of community transmission, high risk for case importation, and limited contact tracing capacity (5). In Vietnam, for example, as a result of this investigation, national policy was changed toward mandatory testing at arrival irrespective of departure location and 14-day quarantine irrespective of test result or clinical signs/symptoms (41). This policy change eliminated the need for resource-intensive contact tracing of flight passengers altogether and enabled detection of another 106 cases among »5,000 passengers on 44 flights until all international flights were halted on March 28. However, given the logistic and economic implications of such policies, developing a quick and reliable point-of-care test that covers the entire infectious period remains paramount.

We conclude that the risk for on-board transmission of SARS-CoV-2 during long flights is real and has the potential to cause COVID-19 clusters of substantial size, even in business class–like settings with spacious seating arrangements well beyond the established distance used to define close contact on airplanes. As long as COVID-19 presents a global pandemic threat in the absence of a good point-of-care test, better on-board infection prevention measures and arrival screening procedures are needed to make flying safe.