Epidemiological Sleuthing During the Pandemic

By Roger Chriss, PNN Columnist

How do we figure out what is really going on with Covid-19? Amid the ongoing surge of Covid cases from the delta variant of SARS-CoV-2, this is very important.

As hospitals are overwhelmed with patients, beds are unavailable, supplies of monoclonal antibody drugs are limited, and frontline healthcare workers are strained. Understanding what is really happening becomes vital to an effective response.

One of the tools most widely used to track the severity of the pandemic is the number of patients that are hospitalized. According to the CDC, we’re currently seeing an average of 9,636 new admissions every day, which is about 14% less than a week ago.   

New Hospital Admissions for Covid-19

hospitalizations_09242021.jpg

But how reliable are these numbers? And just how sick are the patients?

A recent preprint study of VA patients before the delta surge suggested that almost half of hospitalized Covid-19 cases had mild or asymptomatic symptoms. This finding prompted The Atlantic to conclude that the significance of the hospitalization gauge “can easily be misinterpreted.”

As often happens with public health data in the US, information is missing. SFGate recently investigated Covid-19 hospitalizations in the Bay Area and found that simply tracking patient numbers was unreliable. Three major hospital systems -- San Francisco General, Kaiser Permanente and Sutter Health -- could not say if mild and asymptomatic cases made up a large proportion of their Covid patients.

In other words, in many cases hospitals simply don’t have the data needed to assess the severity of outbreaks. We need better ways to validate hospitalization counts and assess local conditions in near real time.  

Fortunately, we have them. Broadly speaking, these include community surveillance, hospital monitoring and indirect signals. Each of these comprises multiple sources of information, and when combined, the result is a deep assessment of Covid conditions.

Carnegie Mellon University’s COVIDCast gives early indicators, including doctor’s visits and symptoms in communities across the nation. This can be paired with public behaviors like masking and distancing, visits to bars or restaurants, local vaccination rates, and searches on Google for Covid symptoms. All of this helps build a risk profile for a community that can be confirmatory for hospitalization data.

Similarly, wastewater epidemiology provides a close look at something all communities produce. For instance, the Sewershed Surveillance Project in Missouri has tracked virus levels in effluent for a year, with spikes often warning of possible outbreaks in cities and counties throughout the state. Wastewater epidemiology is already used in Europe and many parts of the US to help track drug use and disease activity.  

Next comes personnel. Hospital staff is in constant flux, and job postings provide valuable clues about what hospitals are expecting or confronting. For instance, travel nurses move around the US in response to calls for supplementary staff, so job boards like TravelNurse Source offer valuable insight for assessing a hospital’s needs.

The same for respiratory therapists, imaging technicians, and other frontline healthcare workers. Hospitals know their everyday needs, which can be readily found in reports from the American Hospital Association’s Data & Insights database. These reports include the number of staffed beds, admissions, and outpatient visits for most hospitals in the US. If we know what hospitals typically have and see, that allows us to check if they are experiencing a Covid surge.

Further, we have drug and equipment orders. Hospitals have to track all orders for financial and regulatory reasons. So new orders for fentanyl to sedate patients who need intubation, requests for ECMO respiratory machines, or just reordering PPE equipment and other supplies are all readily monitored via federal and state databases.

There are also indirect signals. Hospitals redirecting patients to other facilities, requesting ambulance or helicopter transport, arranging for at-home care for patients they might otherwise admit, or activating emergency plans are all publicly visible, as are signals such as requests for mobile morgues or National Guard assistance.

The above information can be combined to create a reasonably accurate profile of a hospital’s situation. Better, however, is to use such information proactively in order to avoid the severe surges seen in the South over the summer or in Idaho and Montana at present.

Early attempts to use wastewater epidemiology proactively have met with some success. When a wastewater sample from a dorm at the University of Arizona came back positive last year, the school quickly tested all 311 people who live and work there and found two asymptomatic students who tested positive.

All of the above is not limited to the pandemic. The same approach could be used for other infectious diseases, drug overdoses and other areas of public health. If the data streams and information analysis are combined and coordinated, the results could be that much better and more useful.

We cannot and should not rely on a single number, not with so many other tools available to be used. Hopefully we’ll see these tools put to use fast.

Roger Chriss lives with Ehlers Danlos syndrome and is a proud member of the Ehlers-Danlos Society. Roger is a technical consultant in Washington state, where he specializes in mathematics and research.