Switching to Online Shopping
To comply with government regulations on social distancing, people are encouraged to shop online rather than in-store. These present different possible contamination routes which should be assessed and compared.
These contamination routes are, in cyber security and intelligence terms, known as ‘threat vectors’. They combine two things: the agent that intends to cause harm (the threat) and the means by which they can reach you (the vector).
In this article we will examine the threat vectors for on-line shopping with home delivery, and compare them with those of shopping in person.
The Threat
“Coronavirus Disease” is a respiratory disease caused by the SARS-CoV-2 virus. While the fatality rate of the disease is still being assessed, it appears to be serious.
The Vector
The vectors are the means by which the virus goes from being expelled to being absorbed. In this case it appears to be expelled orally in mucus or saliva, and is absorbed through the eyes, mouth and nose. The amount of virus required to cause overwhelming infection is unknown.
When we open our mouths and exhale, we also expel droplets with our breath. These tiny droplets typically will land on surfaces within a meter or so (see below) and if we stand close together, these surfaces include people’s faces. This is what normally happens when we laugh and talk, and particularly when we cough and sneeze. Kissing, obviously, exchanges considerably more droplets, but this is not likely to be a factor in this assessment.
Other droplets that land on surfaces can remain viable for anything between a few hours to several days, although the half life on smooth surfaces is around 7 hours. This means that if people handle the surface some of these viruses will be picked up on their hands, and if they then touch their mouths some of the virus will be absorbed. If each of these transfers is sufficient then infection is possible. Touching an infected surface and then washing your hands strongly reduces the transfer rate and so strongly reduces the likelihood of infection.
Some viruses can remain suspended in the air for up to 2 hours, infecting people who breathe them in. In this case we say that the virus has been aerosolized.
We should therefore consider three main contamination routes: ‘Droplet Transmission‘, ‘Surface Transmission‘ and ‘Aerosol Transmission‘.
Droplet Transmission
Talking, Coughing, Sneezing and Laughing
In 2009, the US National Library of Medicine investigated the number and size of respiratory droplets released during talking and coughing. Individual results varied considerably, but on average each cough released 315 droplets, 85% of which fell within 40cm. Just speaking, by counting from one to one hundred, released 108 droplets, 90% of which fell within 30cm.
Other studies report vapour droplets reaching up to 2 meters after coughing. In 2014, MIT recorded coughs reaching up to 27 meters with an average of 3000 droplets, but this study considered ‘aerosolized droplets’ smaller than 50µm, which the previous study did not.. Measuring these smaller droplets finds many more of them. However it is not clear that these smaller droplets are a threat vector; if coronavirus had a range of 27 meters there would be a much higher infection rate.
Sneezing does not appear to be a key symptom for coronavirus, and so can be omitted from this assessment.
Threat Vectors: Droplet Transmissions
In-Store
In-store shopping has each customer interacting with many others and with staff for on average 40 minutes each week. Aisles must legally be more than four feet (1.2m) wide and therefore people crossing paths in aisles are likely to be within a meter of each other, and droplet transmission is possible. Interactions with checkout and floor staff are also usually within a meter. Self-service checkout can reduce interactions with staff, although there is typically closer contact with support staff when they are needed to interact with the same display as the customer.
Online
In online deliervies, all interactions happen on the doorstep within a 40cm range. However, if ‘contactless delivery’ is an option then the threat of droplet transmission is negligable.
People with symptoms are more likely to order oline, so the threat of droplet transmission for the delivery driver is higher.
Surface Transmission
Surface transmissions only occur when a sufficient level of coronavirus travels from the host to the surface, from the surface to the victim’s hand, and then from their hand to their face. This depends on:
- Viral Load: The amount of virus on a surface immediately after contamination, which in turn depends on the amount of droplets on that surface and the virus load in each one.
- Surface Material: How long the virus can survive on a specific surface
- Environmental Factors: Such as temperature and humidity
- Surface to Hand Transmission Rates: The proportion of droplets transmitted between the surface and the victim’s hand.
- Hand to Face Transmission Rates: The proportion of droplets transmitted between the victim’s hand and face
- Infection Requirements: The amount of coronavirus droplets required to cause an infection
We look at these in more detail below
Viral Load
The amount of virus transferred onto a surface depends on how the droplets got there, and how far they had to travel. If only small amounts of virus are present, it is not infectious. At the moment we don’t know how much this amount is.
Surface Material
At room temperature, coronavirus can survive on metals such as aluminium and steel for up to five days. The virus can survive on wood, paper and glass for up to four days, and on plastics and silicone rubber for up to five days. However the amount of viable virus left at the end of these trials is very small; the ‘half life’ tends to be in the order of several hours which means the ‘viral load’ is halved every few hours.
Enviromental Factors
We do not yet know how easily coronavirus is transferred by physical contact, but we can use studies of other water-borne viruses as a guide. An experimental study reported around a third of respiratory droplets were transferred from Influenza-contaminated surfaces to human skin after five seconds of contact. Coronavirus’s higher infection rate suggests this might be higher.
Surface to Hand Transmission
We do not yet know how easily coronavirus is transferred by physical contact, but we can use studies of other water-borne viruses as a guide. An experimental study reported around a third of respiratory droplets were transferred from Influenza-contaminated surfaces to human skin after five seconds of contact. Coronavirus’s higher infection rate suggests this might be higher.
Hand to Face Transmission
In 2015, a group of medical students were observed touching their faces on average 23 times per hour, nearly a half including the mouth, nose or eyes. This suggests that the time between touching a surface and absorbing the virus is likely only a few minutes. One of the main barriers to transmission that we can easily control is to wash our hands soon after touching possibly contaminated surfaces, and avoid touching our faces.
Note that wearing gloves makes little difference to virus transfer; droplets can be transferred via the surface of a glove as easily as on skin in the timescales given here . However wearing gloves can make you more conscious of touching your face, and so reduce such transmissions. And removing gloves, if done correctly, removes any surface contamination.
Threat Vectors: Surface Transmissions
In-Store
Given the number of people passing through and around shop surfaces, talking and coughing, we will assume all surfaces are contaminated unless specifically treated, although we do not know the viral load. The transmission routes are therefore from touching surfaces and then touching the face, without washing in between.
Common touch points include trolly handles, pin-number keypads, banknotes and packages that staff and other customers have already handled. Mitigations include staff cleaning trolley handles and pin-number keypads between customers.
Other in-store surface contamination include customers returning unwanted items, staff facing up stock on shelves, rotating stock, and moving stock to clean shelves.
Online
For home delivery from retail stores, produce is picked directly from the shop floor before being transferred into stacks and taken by the delivery team. This presents all the above risks to the picker, but not to the customer. The threat vector to the customer is from the surface of the food packages and the delivery box or bag handles. There is typically a few hours between the items being picked during the night shift and the deliveries the next day, which is not enough for the virus to lose more than a half of its viral load.
Normally the driver would take plastic crates to the doorstep, and then hand over the bags inside them to the customer; the customer would not touch the crate handles. As we saw above, to reduce the droplet transmission deliveries are now made to doorsteps and the crates left for the customer to take in, empty, and return to the doorstep for the driver to retrieve within the delivery slot. Each customer is now handling the crates that the driver has touched, who in turn handled the crates of other customers.
Mitigations include staff washing their hands between each delivery, which essentially removes this threat vector.
Aerosol Transmission
Coronavirus particles can remain suspended in air up to three hours after an infectious cough, according to a study by the New England Journal of Medicine, and coronavirus is reported to have an aerosol half-life of sixty-six minutes. These laboratory settings however were not real world conditions, and epidemiologists are unsure under which conditions these contaminated mists are infectious. Real-world reports of isolated infections (eg airplane [ref]) suggests to us that aerosol transmission is rare under most real-world conditions.
Threat Vectors: Aerosol Transmissions
We don’t know whether aerosol transmission is a major factor in this infection. If it is, then it is likely to be more contagious in shops and apartment hallways where the air is relatively still than on external doorsteps.
Conclusion
We have identified three main threat vectors: droplet, surface and aerosol transmissions. There are others; for example there are some reports of coronavirus within fecal matter which does not appear more relevant.
There are many unknowns, and some of what we know, we don’t know very well. Symptomatic people are more likely to order online. To optimise routes, drivers are assigned an area and as more households switch to online shopping, these areas become more condensed. An area with an infection outbreak will have more symptomatic customers and smaller, more concentrated and more infectious areas. All these and more will affect overall risk.
However this assessment strongly suggests that, for all three vectors with the current mitigations in place, home delivery is safer for the customer, though more risky for the picker.
TISG
This is an example of a ‘Tactical Intelligence Assessment’: We are faced with a real-world question that requires a timely answer, but the data to inform the answer is sparse, unreliable and cluttered. Relevant research requires specialist knowledge to parse. Some estimates are based on other estimates and we must beware of filling in gaps with reports that appear solid but are just other speculation. We should also beware of using unconfirmed statistics just because they are the only ones available. Sources must be critically assessed, as many studies are poorly executed and poorly reviewed but present themselves more confidently than they should.
We therefore need to understand what we don’t know as well as what we do, to find alternatives where we can, to note what cannot yet be answered and what we would need to investigate, and yet draw an informed conclusion in time for the next shop.