“Early warning system” tracking coronavirus hospital admissions

New online tool to trace the pandemic is showing a rise in hospital admissions of suspected coronavirus cases among people 20 to 50 years of age

A new tool to track potential coronavirus cases and help public health experts make critical healthcare decisions in Ontario is now available to the public.

Thursday marked the debut of the Acute Care Enhanced Surveillance Pandemic Tracker, a real-time system to monitor daily admissions to 97 per cent of hospitals in Ontario for suspected COVID-19 and related syndromes. Those syndromes – pneumonia, asthma, influenza-like illness and sepsis – are also expected to rise during the COVID-19 pandemic

“I’m not aware of anything like this publicly available anywhere in North America,” says Kieran Moore, the Medical Officer of Health for Kingston, Frontenac and Lennox & Addington (KFL&A), and creator of the tracker. “We’re being transparent and accountable to give the public as well as decision-makers situational awareness.”

Up to 80 or 90 per cent of actual COVID-19 cases can be resolved without hospital admission, so by focusing on admissions, the tracker draws attention to the 10 to 15 per cent of potential cases that would likely be the most acute.

Since there is often a lag between admission and any potential intensive care need, the tracker could be used to redeploy healthcare resources across Ontario to regional hot spots.

“It’s an early warning system,” Moore says. “It allows us to see what’s going on and prepare and respond to a rise in admissions a week or two before they have the capacity issue. Then we can look at things like bed capacity and ventilator capacity and monitor that as we go with a real-time epidemic curve. Given how vast Ontario is, it won’t light up everywhere at the same time.”

Getting resources where they are needed most will be critically important if, as Moore expects, the number of admissions rises in the weeks ahead.

“The seven-day moving average will cross all the historical averages and stay up for potentially several months,” Moore predicts, judging by the current numbers. “It’s going to go up and stay up to several hundred admissions, way above 250 a day, although I’m not sure how much higher.”

As of April 1, the seven day moving average for the province was greater than one standard deviation above the historical average. When the seven-day moving average rises above two standard deviations of the historical average, it signals what is termed “high abnormal activity.” A FAQ page defines these and other terms.

Moore says he will add an interface to the tracker in the coming weeks that will show affected age groups, adding he’s already observed a rise in admissions of people in the 20 to 50 age bracket.

The tool will also provide insight into the efficacy of social distancing and other public health measures. “If the black line plateaus and continues that way, that supports our suppression tactics,” he notes.

While the tracker does not include confirmed COVID-19 cases, Moore says it is reasonable to infer a high degree of accuracy from the tracker.

“We have confidence we’re over our typical winter respiratory season, and any rise is an indication of COVID-19,” he says. “We’ve got 10 years of this data, so we know what total pneumonia counts are expected on April 2, so that gives us some confidence this is different from what’s expected.”

KFL&A Public Health earlier disabled its flu-season tracker, which captures influenza-like illness (ILI) reported in emergency rooms. That tracker showed a steep, sudden rise in ILI in early March.

Because that data was based on reports of patients showing up in emergency rooms, not hospital admissions, Moore believes it was more susceptible to changes in public awareness of the pandemic. The new tracker also captures a broader set of COVID-19-related syndromes.

Some epidemiologists see a strong potential signal of COVID-19 in ILI activity. U.S. researchers have been reviewing whether ILI activity not caused by the flu has been higher than average since the pandemic began. They say that some of the rise in ILI activity could guide public health investigations like testing.

Two weeks ago, the Public Health Agency of Canada’s weekly flu report identified a decline in confirmed flu cases in tandem with a rise in flu-like activity. Public Health Ontario also reported an unseasonal rise in ILI activity in the last two weeks.

“Everything is suspected COVID right now until proven otherwise,” says David Fisman, who leads epidemiology at University of Toronto’s Dalla Lana School of Public Health. He says the previous ILI tracker was working well as a surveillance tool by capturing likely COVID-19 cases, but that the new system potentially underplays the scale of the current pandemic.

“We use the standard deviations to account for normal and expected fluctuation,” says Allison Maier, a KFL&A Public Health research associate. “We don’t want to unnecessarily alarm users of the tool when numbers of admissions are within the range of expected fluctuations.”

Various methods of tracking and projecting COVID-19 have proliferated recently. In addition to the new KFL&A Public Health tracker and existing flu trackers, a dashboard developed at York University attempts to project the number of new COVID-19 cases using machine learning.

There are efforts underway to track COVID-19 through social media, even by analyzing Google trends for the search “can’t smell” (loss of smell is a symptom commonly reported by people infected with COVID-19).

Moore is confident in the power of his new tracker. “Most other modellers are looking at test results alone, but this looks at the potential impact on the health system,” he says. “It’s not theoretical, it’s real-time live data coming in from the health system, and it’s highly accurate.”


Brand Voices

Leave a Reply

Your email address will not be published. Required fields are marked *