Coronavirus cases in England were doubling every seven to eight days at the beginning of September, new data suggests.
It is estimated that between August 22 and September 7, 13 people per 10,000 were infected, compared with four people per 10,000 between July 24 and August 11.
According to the latest round of the Real-Time Assessment of Community Transmission (REACT-1) study, Covid-19 cases are shown as no longer clustering in healthcare or care home settings, as seen in May and June.
Experts say this suggests the virus is now spread more widely in the community.
Out of 152,909 swab results, 136 were positive, and prevalence doubled every 7.7 days, the figures suggest.
Professor Paul Elliott, director of the programme at Imperial from the School of Public Health, said: “I would say the prevalence is still quite low.
“It’s higher than it was in our second round, which was coming into June and July, so it’s gone back up.
“It was very high, we had the lockdown, it came down during May, continued to go down into August into really quite low levels.
“Now it’s gone back up again.
“I think the really important thing here is that this system was set up as an early warning system. And I think it has picked up the signal early. And that’s being fed in to Government.”
Steven Riley, professor of infectious disease dynamics at Imperial College London, said: “The current prevalence is also key for the context, that it is much lower now than it was estimated to be at the start of lockdown.
“But it is comparable, so we’re estimating it to be a substantially slower rate than was observed at the beginning of lockdown.”
At the end of March, it was estimated the number of cases were doubling every three to four days.
Researchers also found that infections are increasing across all adult age groups younger than 65, and across all areas of the country.
Higher rates were seen in young people aged 18 to 24 years, and infection was highest in Yorkshire and the Humber, the North East and the North West.
More than 300,000 volunteers were tested across England between July 24 and September 7, as part of the study.
In the latest findings from Imperial College London the reproduction number (R) – the average number of people an infected person is likely to pass the disease on to – was estimated to be 1.7.
The R rate published by Imperial has been estimated based on a cohort of 150,000 volunteers within a specific time frame.
The weekly official government R rate is produced by Sage and uses many data sources and models to produce a consensus view in the scientific community of the likely R number over a longer time frame.
Prof Elliott said: “Our large and robust dataset clearly shows a concerning trend in coronavirus infections, where cases are growing quickly across England and are no longer concentrated in key workers.
“What we are seeing is evidence of an epidemic in the community and not a result of increased testing capacity.
“This is a critical time and it’s vital that the public, our health system and policy-makers are aware of the situation as we cannot afford complacency.”
The study was commissioned by the Department of Health and Social Care (DHSC) and carried out by researchers at Imperial College London, Imperial College Healthcare NHS Trust and Ipsos Mori.