A one-time lockdown won't halt the novel coronavirus and repeated periods of social distancing may be required into 2022 to prevent hospitals from being overwhelmed, Harvard scientists who modeled the pandemic's trajectory said Tuesday.

Their study comes as the US enters the peak of its COVID-19 caseload and states eye an eventual easing of tough lockdown measures.

The Harvard team's computer simulation, which was published in a paper in the journal Science, assumed that COVID-19 will become seasonal, like closely related coronaviruses that cause the common cold, with higher transmission rates in colder months.

But much remains unknown, including the level of immunity acquired by previous infection and how long it lasts, the authors said.

"We found that one-time social distancing measures are likely to be insufficient to maintain the incidence of SARS-CoV-2 within the limits of critical care capacity in the United States," lead author Stephen Kissler said in a call with reporters.

"What seems to be necessary in the absence of other sorts of treatments are intermittent social distancing periods," he added.

Widespread viral testing would be required in order to determine when the thresholds to re-trigger distancing are crossed, said the authors.

The duration and intensity of lockdowns can be relaxed as treatments and vaccines become available. But in their absence, on and then off distancing would give hospitals time to increase critical care capacity to cater for the surge in cases that would occur when the measures are eased.

"By permitting periods of transmission that reach higher prevalence than otherwise would be possible, they allow an accelerated acquisition of herd immunity," said co-author Marc Lipsitch.

Conversely, too much social distancing without respite can be a bad thing. Under one modeled scenario "the social distancing was so effective that virtually no population immunity is built," the paper said, hence the need for an intermittent approach.

The authors acknowledged a major drawback in their model is how little we currently know about how strong a previously infected person's immunity is and how long it lasts.

Virus likely here to stay

At present the best guesses based on closely-related coronaviruses are that it will confer some immunity, for up to about a year. There might also be some cross-protective immunity against COVID-19 if a person is infected by a common cold-causing betacoronavirus.

One thing however is almost certain: the virus is here to stay. The team said it was highly unlikely that immunity will be strong enough and last long enough that COVID-19 will die out after an initial wave, as was the case with the SARS outbreak of 2002-2003.

Antibody tests that have just entered the market and look for whether a person has been previously infected will be crucial in answering these vital questions about immunity, they argued, and a vaccine remains the ultimate weapon.

Outside experts praised the paper even as they emphasized how much remained unknown.

"This is an excellent study that uses mathematical models to explore the dynamics of COVID-19 over a period of several years, in contrast to previously published studies that have focused on the coming weeks or months," Mark Woolhouse, an infectious disease epidemiologist at the University of Edinburgh said.

"It is important to recognize that it is a model; it is consistent with current data but is nonetheless based on a series of assumptions -- for example about acquired immunity -- that are yet to be confirmed."