As public-health units on the Niagara Peninsula have tried to track, project, and respond to COVID-19 transmission throughout the pandemic, they’ve turned to mobility data from cellular networks. “There's a couple of things that we're currently looking at mobility data on,” says Mustafa Hirji, acting medical officer of health for Niagara Region Public Health and Emergency Services. He and his team use it to work out where people are gathering, and, he says, “We've tried to do some very specific looking at mobility in areas of interest within Niagara where people might be travelling from outside of the region.”
While experts say that the datasets will continue to be a valuable tool during the pandemic — even after the current stay-at-home order ends and the province focuses on loosening restrictions — they do have their limitations.
Google and Apple both collect anonymized and aggregated user-mobility data and publish it publicly, and private companies have been analyzing it for public-health units throughout the pandemic. Hamilton Public Health Services, for example, has hired Scarsin, a Markham-based forecasting company, to analyze information — including that publicly available mobility data — and produce models predicting COVID-19 transmission. “Mobility data is one of a number of inputs within the model that helps to predict potential transmission of the virus,” reads an email statement forwarded to TVO.org by a Hamilton public-health spokesperson.
Our journalism depends on you.
You can count on TVO to cover the stories others don’t—to fill the gaps in the ever-changing media landscape. But we can’t do this without you.
Knowing where people are coming from is particularly important for Niagara Region, Hirji says, because it must balance its economic reliance on tourism with the risk of COVID-19 variants such as B.1617. He has consistently highlighted the risk posed by travellers from elsewhere in Ontario entering the region. Often, they’ve been GTA residents (mostly from the hot spots of Peel and Toronto), he says, seemingly trying to avoid local pandemic restrictions at home when Ontario had localized lockdown rules rather than a provincewide stay-at-home order. “Whenever cases go up in Toronto and Peel, not too much later, cases go up here in Niagara,” says Hirji. (To prevent this, he recommends treating the entire Golden Horseshoe region as one zone under any set of targeted pandemic restrictions.)
One main takeaway from mobility data, Hirji says, is that it suggests that compliance with stay-at-home orders and public-health guidelines has faded with time. “We assume that if there is more [mobility], presumably people didn’t need to travel before,” he says. “They probably now don’t still need to travel, but they’re choosing to travel.” In Niagara, recent Google mobility data, shows that shopping and recreation have “increased to basically a pre-lockdown level,” Hirji says, adding, “Probably a big part of that is actually people being outdoors because the weather’s nice and partaking in outdoor activities. So maybe that isn't as concerning to see that increasing now as it might have been three months ago, when the weather wasn't as good, and if people were going to shopping and recreation, they were more likely indoors, where there's going to be more risk of COVID-19.”
According to Scarsin CEO Paul Minshull, mobility data appears to indicate that each new set of restrictions has been less effective than the last. “During the first lockdowns back in March of last year, you would see community mobility dropping 40 to 45 per cent, and the second lockdown, it's more like 30 to 35 per cent, and then the third lockdown, it's more like 27 per cent,” he says. While he is cautious about speculating on the causes behind the shift, he suggests one possible explanation for the waning effectiveness of government measures: “You can call it pandemic fatigue over time.”
But there are many questions mobility data can’t answer. For example, says epidemiologist Colin Furness, it doesn’t show whether people are wearing masks or physically distancing when they go to parks: “It says nothing about safety.” When it comes to schools, “We can associate elevated risk with schools opening, but we can't say that it's what's going on in the classroom or what's going on in the school yard. The fact is, when schools are open, teachers and bus drivers and support staff are all going to work. So they're all getting out of their house,” he says. “But you also have parents who don't have kids at home who can go out a little bit more, maybe go to the office. So you have this whole ecology of movement around schools being open or closed. So we're no further ahead in understanding whether schools themselves are safe, but we know when they open schools, mobility goes up, and when mobility goes up, COVID goes up.”
To fill in the local-transmission picture more fully, municipalities such as Niagara and Hamilton draw on additional information sources: government policies, vaccination rates, and seasonal weather data are among the other inputs that Scarsin analyzes for Hamilton, says Minshull. When projecting COVID-19 transmission, it’s important to consider that people tend to be outdoors more often when it’s warmer, he explains — hence the inclusion of meteorological trends in modelling: “In Hamilton, you see very large growth in use of parks recently, which is unsurprising now that the weather's nice, and that obviously has a much lower risk than indoor activities.”
In addition to Google mobility data, Niagara acquires demographic data from Environics Analytics, a Toronto-based data-analytics firm. The data is typically used in marketing and includes postal codes, media preferences, and values, which the public-health unit can then use to “sell people a behaviour that's going to keep them healthy,” Hirji says (he likens the process to how a company might promote a product). Using Environics data, he says, his team has been able to determine that people travelling into Niagara are likely to be young, more affluent families seeking leisure and entertainment. The team then develops targeted messaging about taking precautions if you have to go out — and not going out at all if you can avoid it.
Mobility data will continue be a top indicator of COVID-19 transmission going forward, even as Ontario enters the first stage of its reopening plan on June 11, Minshull suggests, adding that vaccination rates likely follow in terms of importance. “As more and more people get vaccinated, particularly fully, it becomes more acceptable for them to congregate together and not even necessarily have to wear a mask,” he says. But, he adds, COVID-19 variants could bring mobility risks back up.
While such data does have its shortcomings, Furness says, "It's the one thing that we can find tracks really closely to transmission rates. What it really says is overall — and it's not surprising, because this is how communicable disease works — the more mobility means the more contact and exposure; the more contact exposure means greater transmission."
And, Furness warns, if the baseline of COVID-19 transmission gets high this summer, “There may really be hell to pay in the fall.”
Correction: An earlier version of this article indicated that Niagara acquires demographic data from Environics Research; in fact, it acquires such data from Environics Analytics. TVO.org regrets the error.
Ontario Hubs are made possible by the Barry and Laurie Green Family Charitable Trust & Goldie Feldman.