‘We in Ontario are flying blind’
How an ambitious, volunteer-driven think-tank is giving us COVID-19 data the province couldn’t
What do you get when you take an engineerturned-doctor, an epidemiologist, a data geek, an astrophysicist and add in some keen public health students?
You get an ambitious, volunteer-driven Ontario think-tank that aims to help provincial leaders and the public understand, in real time, if the province is winning or losing the COVID-19 battle. Each day, the #HowsMyFlattening group analyzes and publicly posts travel patterns, hospital capacity, the effects of public health restriction and loosening decisions, new cases and many other data points. Soon they will add data that shows how far people are moving from their homes, analyses of job postings, spikes in emergency calls for food and shelter, and more, all in an attempt to “forecast” where the pandemic is going.
“I think what the team is doing is a superb effort in transparency,” said Adalsteinn (Steini) Brown, dean of the Dalla Lana School of Public Health at the University of Toronto. He said this transparency helps people buy into the restrictions we are living under, and the recovery plan. “In the same way that people don’t like to see their daily lives change in such a profound way, they also want to have confidence in the system as it starts to open up.”
As to Ontario’s battle, we are doing better today than a week ago, the team’s current analysis of a seven-day trend shows, a bold feature at the top of its unique website. In short, the virus was again spreading at an alarming rate last Monday — now the spread has slowed. More on that later.
The people behind this project call it #HowsMyFlattening in recognition of its central question. Many team members are affiliated with U of T, which is now partially funding the project. #HowsMyFlattening has been compared to the data “dashboard” put out by Johns Hopkins University in Maryland, both examples of academia — not government — providing some of the best pandemic information. Here’s the story of #HowsMyFlattening’s origin, what they have learned and plans for the near future.
Take yourself back to the middle of March. Ontario had just gone on lockdown — days after Premier Doug Ford encouraged people to travel and have fun on March break, then reversed his advice. Still, overnight summer camps seemed possible, the school year was just on pause, and surely, most thought, a return of youth and pro
fessional sports was just a few weeks away.
Enter Dr. Ben Fine. He’s an MITtrained engineer turned radiologist. He works at Trillium Health Partners and ww lectures at the University of Toronto. Like everybody on the “flattening team” (the name refers to the goal of “flatten- (( ing” the growth of the pandemic) he is under 40, some are under 30. Fine was a university student when the SARS outbreak hit in 2002.
When Fine recalls early March at his hospital in Mississauga, he thinks of all of the ICU doctors he was chatting with. They were watching Italy, a well-funded health-care system with capacity “that was totally overrun to the point that ww there was wartime triage occurring over t who gets a ventilator and who doesn’t.”
“The prospect of that coming here as a physician was the worst nightmare,” says Fine.
He started poking around. He wanted to do something to help. He found a dearth of data out there related to Ontario and COVID-19. What he did find on the provincial website was cumbersome, not reader- or analysis-friendly. Like the Toronto Star at roughly the same time, he noticed Ontario’s ministry of health was only posting a daily total of cases and tests in static, pdf form, meaning it was not easily analyzed. And like the Star, the fledgling team he was starting to assemble had to use the internet archive called “the waybackmachine” to get the previous day’s reports cc to get a handle on trends in Ontario.
At the time, Ontario put each day’s tally of tests, positives, negatives, and deaths on its website and then replaced it the next day. The province had told the Star it would be “too difficult” to dig up the previous day’s reports.
Once Fine’s group assembled its own data, what popped out was the first of many insights. With each test performed, there was a percentage increase ff in positive cases and a decrease in negatives. Also, there was an enormous testing backlog.
“For so many reasons we in Ontario are flying blind every day,” said Dr. Laura Rosella, one of the leaders of the flatten- RR ing group. She’s an epidemiologist at the Dalla Lana School of Public Health. Population and public health are her specialties. Her analogy is the captain of an airplane trying to determine the right direction.
“Think of the pilot and all of the instruments. Half of them are covered at any point in time. Either because we don’t collect the information, it’s not coming in fast enough, it has to be manually collected, it’s missing altogether and that’s the story of this response. It’s no one individual’s fault, everyone is trying their best.”
Rosella believes “when the post mortem on COVID-19” in Ontario is eventu ally done, issue one will be the outbreaks aa in long-term care, but close behind will be data, and the lack of data. Ontario needs to quickly learn from this experience, she said.
“We knew that having any chance to respond faster would require rapid collaborative sharing of data across the health-care system,” said Fine. In his experience, there are too many “silos” in ee the world of health-care data.
Joining Fine and Rosella in March was Ali Vahit Esensoy, a data scientist at a AA company called Klick Health. Esensoy used to run the data science team at Cancer Care Ontario. Esensoy knew Fine from a project they have been working on to understand why it takes ww so long for people in Ontario to get MRI scans. He credits Fine for vision, and bringing them all together. “Ben is a great connector, he definitely has the energy for it.”
These three are the leaders of the volunteer group (they all have day jobs and do much of this work in off hours) and their skills are complementary. Esensoy, tt like Fine, is an engineer by training — both like to use engineering principles to optimize systems in health care. Esensoy has a lot of ministry contacts and his time at Cancer Care Ontario aa gave him plenty of exposure to health- gg care bureaucracy. Rosella knows public health “like nobody else,” says Esensoy. An epidemiologist and public-health AA professor at U of T, Rosella is also the Canada Research Chair in Population Health Analytics.
Her goal is to make data usable for
“We are working know the so front- hard, line and workers we are trying to contribute in whatever way we can.” FARBOD ABOLHASSANI HEALTH INFORMATICS GRADUATE
public-health decision making.
“I just love when data can make its way to decision making,” says Rosella.
As the team grew, it moved from blog postings to creating its own dashboard at howsmyflattening.ca. Using a Slack aa channel and Zoom, the team is in con- cc stant brainstorm mode. What data do they need? Where do they get it? Doc- tt tors on the team use connections to get hospital data showing capacity of beds and ventilators. People with ministry connections dig up public health data. One researcher has compiled data on every public health decision and intervention (and now the loosening of re- vv strictions), and the date of the intervention, and when it ended (if it has).
Nothing has come easy. Through a grant from U of T, they have been able to hire several recent grad students to assist. About 200 people are now contributing to #HowsMyFlattening. Their ad- uu visers include Dr. Jenn Kwan, Dr. David v Fisman, Dr. Marzyeh Ghassemi and Dr. Isaac Bogoch, all of whom are playing key roles in fighting the pandemic.
This week, the team is preparing to publish new streams of data, and also to make a broad appeal for more. They hope to have the data from the GTA’s “211 Central” service, where people in need of food banks, housing support or mental- health services turn. That will help the public and leaders understand more about the effects of the pandemic on vulnerable people.
The website already has data showing travel patterns in Ontario, courtesy of Apple and Google direction-look-up da- AA ta. Soon the team will add anonymized data that will show how far people are moving from their home on a given day. Compare that to spikes and troughs of COVID-19 cases and it will add more to the Ontario picture.
One of the first things a visitor to #HowsMyFlattening will see is what is referred to as the “Rt value.” Using a calculation created by Rosella and astrophysicist Alf Whitehead (he works with Esensoy), it is an estimate of virus spread from one sick person based on the past seven days of test results in Ontario. Rt is the actual transmission rate at a given time. Esensoy says “everything we are doing in Ontario is designed to keep the Rt value under one.” In the middle of March, the Rt value was estimated to be 2.2, then it steadily dropped to below one. Last Monday it was 1.11 and had been steadily rising for ww several days, but by end of week was falling. Sunday, it was .94.
Esensoy explains how the Rt value works:
“If I am sick and I get one other person sick and that person gets one other person sick, we are at one. If two people are sick and they only get one person sick, that’s a point-five. The moment that number goes above one (which it is now), it means things are speeding up. A very important indicator. Over the next vv little while, we are going to be looking at this number.” Should the number reach tt two, that means that one sick person has tt infected two people, Esensoy explains.
Of course, given that Ontario’s testing numbers fluctuate, this is just an estimate, the team says.
Esensoy compares what they are doing to a weather forecast. “We need to prepare people to think of this pandemic as aweather condition,” he says.
“We are toying around with ways to report it that way. Sharing the reality as much as possible, not sugarcoating it. mm People are giving up a lot and they are expecting transparency. If you are giving up so much, we are going to give you uu reliable clear data so that people continue to participate in this, because it is critical.”
Farbod Abolhassani, 23, a recent graduate with a master’s degree in health informatics, was pulled in by Fine and now leads the group’s website and technology development initiative.
“We are trying to centralize the data in one location and make it open. There are so many smart people across the province who just want to help and that is what HMF (#HowsMyFlattening) is all about,” sais Abolhassani. “We know aa the front-line workers are working so t hard, and we are trying to contribute in whatever way we can.”
Why is government not doing what #HowsMyFlattening is doing?
“It’s puzzling,” Rosella says. “Out of all the people (in government) who were responding, nobody thought to say we need a transparent, interactive dashboard to let the public know what is happening all the time in real time. It may have just fallen through the cracks. They have the data, obviously.”
She surmises it may be that government was and is worried it would put out information that was not perfect. Instead, the #HowsMyFlattening team is focused on being transparent and getting as much information out as possible.
Fine chuckles when he thinks that, at age 38, he is one of the “adults” in the a group.
Like Rosella, he has young children under the age of 10. In one of the Star’s interviews during Fine’s drive home from the hospital, it was interrupted by an explosion of young voices when he pulled into the driveway.
The “real adults” are the advisory team that checks their work, and senior leaders in the health-care system.
“The adults have been supportive,” Rosella laughs. “Nobody has been mad at us for doing this.”
Will it continue?
“Ben and I have this philosophy that if everyone steps up their game and someone takes this over, we are totally happy for that to happen. We are just a stopgap measure.”
For now, they will keep at it.