Scientists have shown that airport screening for disease will often miss half or more of infected travellers, but can be improved. The findings are published in the journal eLife.
One of the biggest barriers to success is the lack of honest reporting by passengers about their risk of exposure - if being honest could put them at risk of delay.
The researchers from the University of California, Los Angeles (UCLA) and the London School of Hygiene and Tropical Medicine have identified ways to make current screening as effective as possible and highlighted what needs to be done next to improve it further.
They present options for policy makers; for example whether resources would be better spent on arrival screening – which will reduce but not eliminate the cases, or on tracing and containing potential cases highlighted by airport questionnaires. They used a mathematical model to analyse screening for six viruses: SARS coronavirus, Ebola virus, Middle East respiratory syndrome coronavirus (MERS-CoV), Marburg virus, Influenza H1N1, and Influenza H7N9.
“We found that for diseases with a long incubation period such as Marburg and Ebola, taking passengers’ temperature to test for fever is particularly ineffective at the start of an epidemic but does pick up more cases as it stabilises,” says graduate student Katelyn Gostic from the Lloyd-Smith Lab at UCLA.
For the early phase of these disease epidemics, questionnaires are the most effective detection method.
“With diseases such as swine flu that take a shorter time to incubate, fever screening is the most effective method throughout an epidemic,” she says.
Fever screening on arrival has been criticised for being particularly ineffective, but the scientists found it can catch cases missed at departure. Infrared non-contact thermometers will only ever pick up fevers at most 70% of the time. Also, symptoms of some diseases will progress during transit so can be easier to detect on arrival.
Understanding how each disease progresses can improve detection by making sure the right questions are asked in questionnaires. For example, exposure to a symptomatic patient is an established risk factor for contracting Ebola.
However, the researchers found that at best 25% of people honestly reported on exposure to influenza during the 2009 pandemic, and some might even have hidden symptoms by taking medication. This is the first time available information has been used to arrive at an estimated figure.
“Honest reporting can not only improve on-site detection but is essential to enable authorities to follow up with travellers who may have been exposed but have not yet developed symptoms,” says Gostic.
“We need to find ways to incentivise better self-reporting.”
The research was supported by the National Institutes of Health and the National Science Foundation in the US and by the Medical Research Council in the UK.
We spoke to Katelyn Gostic, Adam Kucharski and James Lloyd-Smith about their research.
1. What did you find most surprising about your results?
KG: I was surprised how many factors could influence screening effectiveness. One was the apparently high proportion of people who do not disclose exposure risks on traveller screening questionnaires. Using a back-of-the-envelope calculation, based on data from previous screening initiatives, we estimated that an absolute best-case-scenario estimate is that around 25% of travellers honestly report exposure risks. In reality this percentage is probably lower, but 25% as an upper-bound estimate is still pretty dismal.
We were also surprised by the high number of people who reportedly use over-the-counter drugs like acetaminophen, which conceal fevers and can make their symptoms undetectable. I’m not aware of any formal studies into the influence of fever-concealing drugs on airport screening, but there are lots of suggestions from past outbreaks that this could be an overlooked problem.
AK: It is surprising – and concerning – that we are still lacking crucial information about some emerging pathogens. For example, MERS-CoV has caused hundreds of cases over the past three years, yet many patients did not have a known source of infection. This means it would be near impossible to screen travellers based on potential risk of exposure.
JLS: Even under the best case scenarios we considered, arrival screening missed at least half of infected travellers for all pathogens. There is a clear message to anyone thinking about instituting these programs: they will reduce the rate of importing infections, but they will not stop it. Traveller screening by these methods is inherently leaky.
2. Do your findings make the case for airport screening to be abandoned for departures, arrivals, both?
KG: I think it is important for policy makers to recognize that airport screening will never definitively prevent case importations. There is strong political pressure to respond to global disease outbreaks, but it is critical to make sure that resources are appropriately allocated. Depending on the pathogen, the effectiveness arrival and departure screening can vary dramatically, and formal cost-benefit analyses of screening programs would help policy makers adequately weigh the economic and social impacts of such programmes.
AK: Our findings show that airport screening must be tailored to the outbreak in question. There is no one-size-fits-all solution. The effectiveness of departure and arrival screening will depend on the pathogen, the screening method used, and the current state of the epidemic. For example, questionnaires about potential exposure would be unlikely to catch MERS-CoV cases, but might be useful for other infections. Fever screening at arrival might pick up some influenza A/H7N9 cases, but would be unlikely to spot many Ebola infections.
3. Can you suggest how honest reporting could be increased?
KG: As things stand, anyone who reports honestly puts him or herself at risk of delay or detainment—this is a terrible incentive for truthful reporting. This is more a question for social scientists, but broadly I think that if we want people to report honestly we need to think more carefully about the incentives driving peoples’ decisions while answering screening questionnaires. In an ideal world, we would also develop more precise screening methods, so that people can feel more confident that they won’t be detained if they aren’t actually sick.
4. Will you make your findings available to airport authorities in any countries?
KG: The broad patterns we identify in this study are definitely relevant to airport authorities, however we would be hesitant to encourage anyone to develop policies around specific numerical outputs from this study. As with any model, our quantitative predictions depend on assumptions and inputs that might not generalize to all situations. Indeed, we have shown that optimal screening approaches for one pathogen may be very different from another. So if airport authorities or public health officials were to use this study, we would encourage them to focus on patterns rather than numbers.
I also want to remind readers that we consider the screening outcomes predicted in this study to be best-case scenario projections—in reality screening effectiveness would probably be lower, as it has been historically.
5. How could your findings be used to change current airport screening practices?
KG: This study highlights the need to learn more about ways to incentivize honest reporting on questionnaires. Ideally, the majority of travellers would report exposure risks honestly, and then individuals at risk of infection would be enrolled in a follow-up monitoring program at their destination. This is the current strategy for Ebola in the US. Such a strategy acknowledges that we can’t always know at the time of travel whether someone will get sick, but it enables us to implement treatment and containment quickly if symptoms do arise.
Such approaches would only be effective for diseases that are not transmissible before symptoms develop, like Ebola. In contrast, measles is highly contagious and can be transmitted before the characteristic rash develops. This makes measles notoriously difficult to contain.
AK: Researchers have previously criticised airport practices, particularly fever screening at arrival, as having little scientific basis. Our results expand on these studies, showing that a wide range factors can influence the effectiveness of screening at departure and arrival. When deciding on screening practices, these factors should be taken into account. Screening can be expensive and inconvenient, and it is important to provide a scientific case for it, rather than just political one.
6. What further evidence is needed to make more precise recommendations?
KG: Formal cost-effectiveness studies would allow policy makers to assess the social and economic impacts of screening policies at departure and arrival. We also need studies to quantify fever-suppressing drug use, as well as behavioural incentives to encourage honest reporting. Finally, it would be helpful to quantify how human error impacts screening efficacy; we used estimates of the machine efficacy of fever scanners, but because human operators are also fallible, fever screening efficacy may be lower in reality than in our model.
JLS: Another important gap is that we have little direct data on the efficacy of departure screening. This is needed to weigh the benefits of different screening policies, and areas for investment. For example, in the current Ebola outbreak, how many potential travellers were turned away before boarding airplanes to depart West Africa? Of these, how many were actually Ebola cases? There is broad agreement that departure screening is probably more efficient than arrival screening, but we don’t actually have any examples where we know how well it worked in practice.
7. What factors did you identify that could improve the effectiveness of screening programs for future emerging pathogens?
KG: Improving general epidemiological knowledge so that it is possible to screen for specific risk factors is an obvious opportunity for improvement. For example, risk factors for emerging pathogen Middle Eastern Respiratory Syndrome (MERS Co-V) are not yet well-characterized. As stated above, this would need to be accompanied by improved incentives for risk factor reporting and by systems for follow-up with individuals at risk for infection.
Screening for a broader set of symptoms than fever might improve screening effectiveness, but would likely increase the already high cost of screening and number of false positive case detections.
Media contacts
Emily Packer
eLife
e.packer@elifesciences.org
+441223855373
