Originally Published: 10 AUG 22 06:49 ET

By Deidre McPhillips, CNN

(CNN) -- The United States seems to have hit a Covid-19 plateau, with more than 40,000 people hospitalized and more than 400 deaths a day consistently over the past month or so.

It's a dramatic improvement from this winter -- there were four times as many hospitalizations and nearly six times as many deaths at the peak of the first Omicron wave -- but still stubbornly high numbers.

And there are big question marks around what might happen next, as the coronavirus' evolution remains quite elusive 2½ years into the pandemic.

"We've never really cracked that: why these surges go up and down, how long it stays up and how fast it comes down," said Dr. Eric Topol, a cardiologist and professor of molecular medicine at Scripps Research. "All these things are still somewhat of a mystery."

BA.5 remains the dominant subvariant in the US for now, causing most new cases as it has since the last week of June.

Data from the US Centers for Disease Control and Prevention, published Tuesday, shows that the Omicron offshoot accounted for 87% of new cases in the first week of August, inching up a few percentage points from the week before.

That slight increase in prevalence is a sign that no other variants are outcompeting it -- and promising for future trends.

BA.5 "has been very formidable because it's so transmissible and has so much immune evasion," Topol said. But the plateau in hospitalizations is "encouraging" because it means the subvariant probably has worked its way through most of the hosts it can find.

"Right now, the question is what comes as we descend from BA.5. It could take weeks."

CDC ensemble forecasts predict stable trends in hospitalizations and deaths over the coming weeks, and experts agree that the worst of the wave has probably passed.

But it's hummed along at a high level because it continues to find people whose immunity from vaccination or infection has waned over time -- something that will continue to happen, said William Hanage, an epidemiologist and associate professor at the Harvard T.H. Chan School of Public Health.

And with children going back to school, a change in seasons and other variants on the horizon, it's unclear when the plateau will drop -- and by how much.

"I would expect things to decline at least for the next month or so," said Trevor Bedford, an epidemiologist and genomic scientist at the University of Washington's School of Public Health.

"But, of course, there are other things waiting in the wings. If it's not variant-driven, it will be seasonality-driven," he said, with case rates likely to rise as more people head indoors for colder weather.

Even if trends don't improve as expected, though, it's unlikely that potential future waves will be as devastating as with Delta or the original Omicron variant.

Alpha, Delta and Omicron "are cousins, not children," Hanage said. They each made a significant impact on the population because they were so different from each other.

But the turnover that's happened this year -- from BA.2 to BA.2.12.1 to BA.5 -- has all "emerged from within Omicron," and the subvariants are much more similar to each other. If the next variant is as different as the change from Delta to Omicron, it would be "quite a change" from what's been happening recently, he said.

That's not impossible, and close attention is being paid to the latest subvariants that have popped up.

Over the past week, BA.4.6 grew from about 4% of cases in the US to just under 5%, according to CDC data.

Its mutations aren't "especially worrisome" when it comes to immune evasion, Topol said, but we just don't know yet. Even a small increase in prevalence is growth, and "if it keeps growing, that means it's got an advantage. The more it grows, the more we'll have to be concerned about it."

Another Omicron offshoot, BA.2.75, has not made it to the CDC's variant tracker. It accounts for less than 1% of cases in the US, according to gene sequencing company Helix.

But people are watching "with a degree of trepidation" because it has more changes on the spike gene that could lead to more substantial immune escape or transmissibility, Hanage said.

Still, so much is unknown. Combinations of changes like this have sometimes led to the next variant of concern, and other times they've gone nowhere, he said.

Looking ahead, Hanage says, "there's likely to be one step forward, two steps back" when it comes to progress in the Covid-19 pandemic.

And if deaths stay above 400 a day for a full year, that's more than twice as bad as the worst recent flu seasons, he said.

"So those are numbers we've got at the moment, at a point when things are, relatively speaking, good," he said. "This is something which I think a lot of people don't grasp."