This month, the US passed 5 million cases of coronavirus, and the domestic outbreak is already edging perilously close to 6 million. How did this unthinkable catastrophe get so far out of control?

The sad truth, scientists say, is we likely had no idea how widespread and pervasive COVID-19 infections already were in the early stages of the US outbreak, long before a national emergency was declared.

These hidden cases – missed by broad failures in testing and surveillance – could have meant that by the time multiple American states began declaring emergencies and enforcing shutdowns in March, it was already too late to contain the worsening infection already spreading undetected.

According to new estimates, almost 109,000 people in the US might actually have been infected by 12 March, at which point only 1,514 reported cases of local transmission had been reported.

One day later, a national emergency was proclaimed, but the findings suggest health authorities at the time were only glimpsing the tip of a vast viral iceberg – completely missing tens of thousands of unobserved infections already circulating in their midst.

"We weren't testing enough," says infectious disease epidemiologist Alex Perkins from the University of Notre Dame.

"The number of unobserved infections appears to be due to very low rates of case detection during a critical time, when the epidemic was really starting to take hold in this country."

According to Perkins and his team, the unprecedented spread of SARS-CoV–2 is the result of a perfect storm of things you don't really want to see in a virus: high transmissibility combined with both presymptomatic transmission and asymptomatic transmission.

Even with a best-case testing and surveillance regime, containing such a virus would be a challenge. And the US did not demonstrate a best-case testing and surveillance regime.

"Unlike other countries where testing and containment measures were pursued aggressively, rollout of testing in the United States was slow, and widespread social-distancing measures did not go into effect until several weeks after the first reported case," the researchers write in their paper.

In addition to a widespread lack of testing kits and facilities across the US, strictly enforced rules about who was allowed to get tested in January and February meant a huge number of people with serious, COVID-19-like symptoms would have been unable to get tested for the virus – creating a massive blindspot in real-time outbreak data.

"Until 27 February, testing criteria in the United States were limited to close contacts of confirmed cases and those with recent travel to China," the researchers explain.

"This means that any local infections resulting from an unobserved imported infection would have gone unnoticed. Community transmission occurred without notice while testing was still being rolled out, albeit to an unknown extent."

To go back in time and try to quantify just how many COVID-19 infections might have been taking place in this time-frame, Perkins and his colleagues modelled the spread, using a stochastic simulation fed with real-world data on the outbreak.

It's a method that produces inherently uncertain results, as the researchers acknowledge, but nonetheless it generates quick provisional estimates that help to explain how the US got to where it is today.

On the whole, the estimates suggest that there could have been about 108,689 cumulative infections in the US between the start of the year and March 12, with over 17,000 infections occurring on March 12 alone – itself, over 10 times the official cumulative count known at the time.

The estimates can also vary significantly, though, depending on the parameters of the simulations run. Some of the run-throughs suggested only about 10,000 cases by March 12, but others suggested much more: hundreds of thousands of infections by March 12, or even in excess of a million.

Despite the lack of precision, the researchers say their findings generally indicate the US was well past the possibility of containment by 12 March – and that at that time, large-scale mitigation efforts would have been necessary to effectively fight the virus, rather than reactionary measures rolled out on a state-by-state basis.

"I think the fact that there were so many infections by the second half of February speaks to the importance of and what we could have done in terms of containment," Perkins says, pointing out that early, aggressive action to test people, isolate cases, and conduct rigorous contact tracing, could have slowed the spread and ultimately saved lives.

"I think a lot of us were hoping that would be the situation with this disease. By the time we got to February, the problem had grown so big, containing the virus wasn't possible."

The findings are reported in PNAS.