2 charts estimate the true scope of the US's coronavirus infections and deaths. The portrait is grim.

• Coronavirus cases and deaths are undercounted in the US, primarily due to a lack of testing at the beginning of the outbreak.
• Several recent studies estimate the true scope of the pandemic.
• The charts below compare estimates of the true number of US cases and deaths to the official tallies.

No matter how you slice it, coronavirus cases and deaths are severely underreported in the US.

"The debate is by how much are cases and deaths undercounted," Dr. Howard Koh, a professor at the Harvard TH Chan School of Public Health, told Business Insider. "We probably won't know the real number of deaths and numbers infected until this is all over."

Over the last several months, however, researchers have attempted to model and estimate the true scale of the outbreak.

Robert Redfield, director of the Centers for Disease Control and Prevention (CDC), said on June 25 that the real number of Americans who've gotten the coronavirus is probably 10 times higher than the agency's official count.

"Our best estimate right now is that for every case that's reported, there actually are 10 other infections," he said.

By that day, the US had recorded 2.3 million cases. Redfield's assessment would make the true number 23 million.

MIT researchers, similarly, calculated that by June 18, around 17.5 million people in the US had gotten COVID-19 — around eight times the official number on that date. And a study published in the journal Science found that the US probably saw around 8.7 million coronavirus cases in the last three weeks of March alone, though only 100,000 were officially counted during that time.

Why the US numbers are off

The primary reason US numbers are so low is that the country's testing capacity was severely hampered at the beginning of the pandemic. The CDC initially distributed faulty tests, then lingering shortages of accurate tests meant that only patients with severe symptoms were eligible.

Plus, many people with mild or asymptomatic cases are unlikely to seek out a test at all.

"During the early part of the pandemic, laboratory-confirmed cases captured only an estimated 10% to 15% of all infections," the authors of a recent study estimating the US's true death toll wrote. "As the pandemic has progressed, official statistics have become better aligned with excess mortality estimates, perhaps due to enhanced testing and increased recognition of the clinical features of COVID-19."

The country is now testing far more widely: around 62 out of every 100,000 people per day. That means more cases are being diagnosed and reported than in March and April.

To estimate cases that were missed during the early phase of the US outbreak, researchers have taken varying approaches. The MIT researchers' model assumed a true infection rate of around 5.3%, and extrapolated from there. The study published in Science analyzed reports of flu-like illnesses from the CDC's influenza surveillance system. They found that many states reported a surge of illnesses in March — about 2.8 million — that weren't attributable to the flu or seasonal respiratory diseases. The researchers think most of those were coronavirus cases.

That study also suggested that only one-third of coronavirus patients in the US sought medical care in the first place, which would mean the actual surge of coronavirus cases was likely three times larger than those extra flu-like cases. That's how they arrived at an estimate of 8.7 million cases in March, instead of the 100,000 officially documented during that time.

A quarter or more of coronavirus deaths may not have been counted

Reported deaths also fall short of real-life tallies, albeit by a smaller margin.

Estimates from three sources — MIT, a July 1 study published in the Journal of the American Medical Association, and The New York Times — suggest that actual coronavirus deaths could be 23% to 44% higher than reported.

The MIT researchers estimated that the coronavirus case fatality rate was around 1.6% in the US as of June 18. That means deaths may have reached 174,000 by that time. About 120,300 were officially reported.

Meanwhile, a study in the Journal of the American Medical Association analyzed excess deaths from March 1 to May 30 that couldn't be attributed to the flu, and found that the US may have seen around 122,300 COVID-19 deaths. Around 95,200 coronavirus deaths were recorded during those three months. That would suggest the actual death toll was 28% higher than the reported figure.

Similarly, a New York Times analysis of excess deaths from March 15 to June 6 found 113,800 more deaths than expected based on the CDC's historical trends. Only 92,600 of those deaths were recorded as COVID-19 fatalities, leaving 21,200 unaccounted for.

Some coronavirus deaths could still be counted

It's still possible that some excess deaths will eventually be added to official numbers as states revise death certificates or submit new ones to the CDC. Under normal circumstances, death tallies from the CDC can take up to eight weeks to be reported. Death counts for COVID-19 are likely to take even longer.

But some deaths will likely never factor into the official total.

"There are a lot of deaths that we probably will never know if they were coronavirus deaths or not," Elizabeth Halloran, a biostatistician at Fred Hutchinson Cancer Research Center and University of Washington, previously told Business Insider.

The challenge of keeping an accurate record of coronavirus cases and deaths is likely to be exacerbated by flu season in the fall. Models from the University of Washington's Institute for Health Metrics and Evaluation (IHME) suggest that the US could see more than 200,000 additional coronavirus deaths from now to November.