Why is Colorado’s COVID-19 situation so much less clear than last year?

If it seems like Colorado’s COVID-19 picture is murkier than it’s been since the pandemic’s early days, it’s not your imagination — and it’s not just happening here.

In previous surges, there was a relatively predictable cadence to COVID-19 data: the percentage of tests coming back positive would increase, cases would go up and hospitalizations and deaths would follow. Data from wastewater monitoring wasn’t widely available for much of 2020, but once it came online, it became an even earlier warning sign when infections were starting to rise.

Now, those measures still exist, but they’re far less informative as fewer and fewer people get tested. Some people have even started looking at one-star reviews of scented candles online — since a diminished sense of smell is a characteristic COVID-19 symptom — for a very rough idea of how cases might trend, though that’s not useful for guessing what’s happening in any particular geographic area.

Colorado’s COVID-19 positivity rate and cases have been rising since mid-October, but public health experts weren’t sure the trend was real, given the data’s limitations. The uncertainty largely was dispelled this week, when hospitalizations rose significantly for the first time since June — though they remain relatively low compared to previous waves.

Being able to interpret the early data matters because knowing what’s coming allows hospitals to plan for an increase in patients and regular people to decide if they want to take additional precautions. When hospitalizations started taking off in Colorado during the virus’s delta wave in October 2021, the state had about two weeks of warning from case and positivity data, with a similar lead time when the omicron variant hit in December.

Now, though, it’s not clear how much warning Colorado would have if another wave were building, Beth Carlton, an associate professor of environmental and occupational health at the Colorado School of Public Health, said last week, when hospitalizations were still plateaued. Those who study the data need to figure out how to develop a clear picture so people know when they need to take more precautions and when they can relax a bit, she said.

“We need to really be thinking strategically about what are the key data,” she said. “I think that this is the question that is most important to be asking right now.”

Wastewater monitoring isn’t skewed by whether people decide to get tested, since anyone who’s infected will shed the virus in their stool, said Jude Bayham, an assistant professor at Colorado State University and Colorado School of Public Health. That said, utilities collect their samples at different points in the treatment process, so some may be checking up on a larger segment of their population than others.

Another thing to consider is how to interpret the results in places where the number of people using the facilities tends to swing, like a business district that’s mostly empty on the weekend or a ski town whose population jumps over the winter, he said.

“It’s always a challenge to know what the real trend is,” he said.

Bayham and others developed an algorithm to try to sort through the noise. The filtered results are posted on the Colorado Department of Public Health and Environment’s wastewater dashboard, with viral levels in each utility characterized as increasing, decreasing or flat. The results are most useful for detecting when an area is seeing increasing infections, because people who are no longer contagious keep shedding viral particles in their stool for some time, meaning it takes longer to see a decline than a rise, he said.

People who want to know the general trajectory of infections can look at how their area’s wastewater is trending over time, but it doesn’t tell you the precise number of infections on any given day, Bayham said.

“That’s pretty useful, but it’s still a noisy signal,” he said.

Dr. Ajay Sethi, an infectious disease epidemiologist at the University of Wisconsin Madison, said similar caveats apply to most types of data. Cases have always been underestimated, initially because tests weren’t available, and now because so many people take home tests, he said. And of course, some people don’t feel sick and therefore don’t know to get tested.