In this study we evaluated the timeliness of contact tracing (CT) of flight contacts in daily practice. We conclude that the prevailing policy to provide close contacts antiviral PEP during the early phase of the influenza pandemic is very difficult to implement effectively and therefore has little effect to control disease spread. Active case finding through contact tracing of exposed persons is an important procedure during the containment phase of an emerging communicable disease. However, our data show that, even in a small-industrialized country with modern communication tools, tracing of flight contacts exceeds the required maximum of 48 h after exposure.
For influenza, close contacts of contagious index cases are entitled to receive antiviral PEP within 48 h after exposure to prevent them from becoming ill and further spreading of the disease. Starting oseltamivir within 48 h does not prevent disease but shortens the disease period, mitigates symptoms and might decrease further transmission. Awareness among contacts to seek medical evaluation when influenza-like (ILI) symptoms occur, for both proper antiviral treatment and (home-) isolation advice, reduces further spreading. As influenza has a relative short latent period, for influenza A(H1N1)/2009 varying between 0,7-3,1 days [16, 17], contacts ideally should be informed within 1 day. Oseltamivir postexposure prophylaxis for this pandemic strain is reported to be effective even when administrated more than 48 h after exposure in household settings [18], however, delays in administration are not specified. We cannot exclude the possibility that in our study, even delayed administration of oseltamivir prophylaxis may have prevented some people from becoming ill, although we anticipate the effectiveness of the intervention overall to be less in this setting than in households.
Our study among 17 contact investigations showed an average total delay of 3,9 days between flight arrival and identification of contacts by passenger list, which is too late for effective PEP, and late for alerting on first symptoms of disease. Only in three contact investigations (18%), contact details were obtained within 48 h. However, after identification of passenger details, health authorities need time to actually trace the contact and administer PEP. It is highly unlikely that this was achieved within the same 48 h. We therefore conclude that contact investigation for provision of PEP as conducted here was ineffective.
Regarding the awareness of ILI symptoms, Schiphol Airport handed all passengers on flights arriving from Mexico information leaflets on influenza A/H1N1 2009 with information on early symptoms and requesting them to seek medical advice in case of fever and respiratory symptoms such as coughing. Posters with this information were placed in passenger halls, to inform passengers arriving indirectly from Mexico via transit through other airports, or arriving from non-endemic areas with higher transmission (e.g. USA). As contact details were identified on average 3.9 days after exposure, however not contacted yet, we conclude that CT did not have additional value for timely achievement of increased awareness.
It is not a new finding that contact tracing of flight passengers is a time-consuming procedure [8]. In one study among flight passengers during the pandemic in 2009, 52% (53/95) of the contacts were reached within 72 h [5]. In a measles contact investigation, 75% (202/275) of responding passengers were contacted within 72 h. In this study however, the diagnosis of measles was already suspected during the flight, and laboratory confirmation was initiated immediately after landing [19]. It also helped that many contacts were tourists staying at the same hotels, which facilitated tracing them.
Our study shows that the longest delay before identification of contact details for an influenza index case is caused by the time between arrival and laboratory diagnosis (interval I, 2,6 days). This delay is a result of patients delay in seeking medical care, and doctor's delay, including laboratory confirmation. For influenza, the indicated laboratory test was Polymerase Chain Reaction, which takes several hours to obtain the result and in the beginning of the pandemic, the PCR test was not yet available in many laboratories.
Patients delay was considerable however. It even took the seven passengers with date of onset before the flight, and therefore symptomatic during the flight, 1 to 2 days after arrival before laboratory confirmation was made. Also, none of the airline reported that these patients already were identified during the flight, nor that infection control measures were taken. For the indexes that became ill on the day of arrival, delay until laboratory confirmation still lasted 3 days (range 1-6 days). A pre-pandemic study by Sharangpani et al. among flight passengers showed that they are more willing to seek physicians care in case they developed flu-like symptoms when the perceived the pandemic as serious [20]. Leggat et al. demonstrated during the pandemic that only a minority (35,5%) of Australian citizens would cancel their air travel in case of cough and fever lasting more than 1 day. This was higher among persons who were more concerned about the pandemic [21]. In the Netherlands, the perceived severity of the disease decreased significant during this study period [22]. We expect that the delay until laboratory diagnoses in this study considerably is affected by patients delay seeking medical care, which might be better in diseases experienced as more threatening.
Collecting passenger details from foreign airlines also caused considerable delay because of differences in time zones and the need to convince the concerned airline companies about the urgency to collect and hand-over passenger lists with contact details. Sometimes official request letters were necessary for legal reasons to release personal contact details. Dutch companies were easier to convince by Dutch health authorities to hand over passenger details. Our data show that contact details that were identified too late or not at all, indeed more often originated from non-Dutch than from Dutch airline companies. An internationally standardized contact tracing protocol, communicated with the International Civil Aviation Organization (ICAO) and International Air Transport Association (IATA), would facilitate the timeliness, and therefore effectiveness of contact tracing.
Although one might expect differently, timeliness of CT for flights where PLC's were available, was not better than CT for flights without PLC. However, PLC's reduces the effort, in terms of staff support for airline companies and the municipal health service to collect useful passenger information considerably. PLC's were only used by Dutch airlines, who already were able to provide passenger lists relatively quickly. This also explains the limited attributed shortening in timeliness. Contact details on PLC's might be more accurate to trace the passenger than details provided by the passenger list or booking station. This is further investigated.
This study has several limitations. As available data were recorded in days, and not in hours, it was not possible to determine the time intervals more precisely. As this was both with first and last date of the intervals, we expect no negative or positive bias. Secondly, the arrival date was used for date of exposure, while the actual exposure might have already taken place the day before at departure of the flight. This would imply an increase in delay and decrease the effectiveness of contact tracing. Also, we have no data if, and when contacts were actually reached and oseltamivir was administered. Since several steps were still required to reach the contacts after they were identified through passenger lists, this only would have lead to further delay in administrating prophylaxis. Further investigation into the timeliness of administration of prophylaxis among these contacts is initiated, to have insight in the delay of this last interval to facilitate future decisions on the effectiveness and necessity of contact tracing among flight passengers. Lastly, this study includes CT initiated at only one airport. CT procedures might be different at airports in other countries, which influences interval III. As this is not causing the main delay, we do not expect that in other countries CT would be much faster.
We conclude that tracing close contacts among flight passengers during the initial phase of pandemic A/H1N1 2009 was not effective, as timely provision of PEP could not be achieved in most cases. Most contacts came from an endemic area (Mexico) or areas with well known increased transmission during the first 2 months of the pandemic. The additional risk for those travelers of being a close contact during a long haul flight is small (3,5%) [5]. Furthermore, airline companies and/or Schiphol airport already provided contacts with information on the disease and its symptoms by. The benefit to inform them of the fact that they were contacts of a laboratory confirmed case did not justify the extra effort health authorities invested in contact tracing, especially during a period where public health officials, airports and airline companies were absorbed by efforts of other pandemic related control measures.
In hindsight, the limited burden of disease of influenza A/H1N1 2009 did not justify contact tracing efforts. The main reason for flight contact tracing is raising alertness for possible exposure to uncommon infectious diseases, enabling early recognition and treatment of the disease and timely installation of control measures (e.g. SARS and viral hemorrhagic fevers). For some diseases, PEP is indicated as well. The risk assessment upon which the decision to install contact tracing is based should incorporate - apart from an evaluation of the severity and rarity of disease - an assessment of the required timeliness of effective control measures [23]. The expected time for laboratory confirmation of index cases and identification and tracing of contacts should be related to the maximum period during which quarantine, PEP or other control measures are effective in order to decide on the benefit of this time consuming procedure. Lastly, also cabin crew should be aware of their role of signaling infectious patients. In consultation with medical professionals, direct control measures can be installed, as well as medical evaluation after landing.