In this analysis, a total number of 187,755 flights were simulated across 24 airports and ten countries estimating around 33.9 million passenger seats. The importation risk indices of the individual countries for various airports estimated are shown in Fig. 1. As evident from Fig. 1 the Middle Eastern countries posed the highest risk indices followed by UK, Italy, and USA. It is important here to note that the current simulation does not take an account of transit passenger; for example, a passenger originating from JFK arriving in BOM via DXB, would be assigned as a passenger from DXB and not from JFK. The estimated importation risk index is consistent with observed data obtained from various government sources that 7 airports from middle east brought in the maximum number of 2019-nCoV infected passengers (~ 300) in India until the international flights were suspended. It is important to note that Emirates Airlines, a Dubai based airlines, operates 172 flights per week to India serving nine destinations (Emirates and India; partners in economic growth 2018), which is by far highest number of flights and destination by any other foreign airlines. Further, in 2017–18 Emirates has reported 87% of the seat factor and assuming that 50% of them are transit passengers arriving from Europe and USA, thus by far, it is an extremely fair and valid assumption that Dubai airport (and other transit airports) in absence of adequate preventive measures, could be a potential hotspot in future for transmitting infection during COVID-19 like outbreaks. As shown in Fig. 2 the importation risk index for various countries (airports clubbed together) clearly showing the middle east having the highest risk index. A certain amount of risk, although significantly lower than middle east airports, was also posed by passengers originating at LHR and taking the direct flights to India. As expected, the third highest importation risk was posed by Italy, which was one of the worst affected countries during our study window. It is also to be noted that Indian Government has operated special flights to bring back the stranded citizen from Italy. Surprisingly and contrary to expectation the direct travelers from China posed the lowest risk (Fig. 2), this could be due to potentially much lower numbers of passenger exchange with China as compared to other countries. While importation risk index does not directly represent the number of infected passengers imported in a country it is important to have the robust validation of the methodology. For this purpose, a scatter plot between number of confirmed imported cases from various countries and importation risk index is shown in Fig. 3. As evident from the figure a correlation (R2 = 0.99) clearly indicates the strong agreement between number of passengers imported from a specific country and corresponding importation risk index. There is approximately factor of 6 difference between highest number of passengers (from middle east; ~ 300 passengers) and second highest from United Kingdome (~ 50 passenger). To avoid any bias in the representation of correlation we have also plotted the scatter plot between all the countries (excluding middle east) and importation risk index (show in inset in Fig. 3), and observed the strong relation (R2 = 0.91).
Concluding remarks
We used the various online sources along with the data from government organizations for ten countries to screen the flights and passengers form 24 airports connecting various international airports in India. As of 4th March 2020 India has started the thermal screening of passengers arriving from 12 countries, viz. China, South Korea, Japan, Italy, Iran, Singapore, Thailand, Malaysia, Hong Kong, Vietnam, Nepal, and Indonesia. It is, however, not clear if the passengers either travelling and/or transiting through any of the airports form middle east, especially Dubai, were also being subjected to thermal screening. If that was not the case, it is very important to mention that non-screening of passengers arriving through large transit hub like Dubai (DXB) and were infected could have remained undetected, which later on went on to became positive cases. On the other hand, the stamping of the passengers initiated by authorities in India came as one of the most effective steps, as long as it was stringently executed by authorities and followed by passengers for the self-quarantined for the prescribed period. We strongly emphasize that this was one of the most effective way to contain the spread due to importation of the infected passengers in two following ways. a. a full database of the passenger was being obtained at the time of the screening, and b. it is much easier and feasible to track the very same passenger for next 14 days, which is the advisable quarantine period. Such a screening and additional measures like stamping need not be restricted for the passengers arriving only from largely affected countries but must be extended to the passengers coming from large transit hub in future. For example, at the time of initiating the thermal screening for the passengers arriving from Singapore, there were not many cases registered in Singapore or other east Asian countries. As per our understanding such a measure was implemented primarily to identify the passengers arriving from China, which was epicenter of pandemic, transiting through one of the east Asian countries particularly Singapore airport. Overall, we believe that the effective measures taken by the government and authorities at various international airports in India were adequate and effective, although they were restricted only to the passengers arriving from largely infected countries. We, however, also note that the screening and quarantining of the passengers arriving from major transit hub could also have been strictly implemented from the very beginning. We also recommend that international flight operations in India could be cautiously opened in phased manner. We are also of the opinion, based on our analysis, that only one airport can be considered for international operations until the global pandemic is substantially brought under control. Neverthless, these policies are just suggestions and can be reviewed periodically.
It is very important to note that analyses presented here involve the modeling and estimation based on combination of various different data sources. Under such a situation, where large datasets from various sources is combined to derive certain index, requires a lot of assumptions to consider, and pose various limitations. All these simulations and subsequent estimation heavily rely on the accuracy and reasonable assumption of passenger seats and flight data. In this way, we are not aware of any particular methodology to validate our model, except that comparing the derived importation risk index with actual number of passengers, which for our study is in good agreement. Nevertheless, our results provide data and methodology, which in a nominal way could be used for effective measures and mitigation policies to be implemented at travel hubs for effective containment of travel related risk of infections. Lastly the study presented here completely relies on the data available at this period of time and can be further tuned in future as and when additional data is available, which may substantially alter the results/conclusions presented here.