People wear face masks to protect against the spread of the coronavirus as they walk along a street in Beijing. Photo / AP
People wear face masks to protect against the spread of the coronavirus as they walk along a street in Beijing. Photo / AP
The number of people infected with Covid-19 during the first wave of China's outbreak may be up to four times higher than official figures suggest, new analysis has found.
By February 20, authorities in China had reported roughly 55,000 cases – but according to modelling published in The Lancet, this figure could have been as high as 232,000 had a broader case definition been applied from the outset of the epidemic.
Between January 15 and March 3, China's national health commission issued seven different case definitions for classifying coronavirus patients.
Researchers at the University of Hong Kong calculated how the changing definitions had altered the trajectory of the country's epidemiological curve, based on data published by the World Health Organisation after a visit to China in late February.
The new analysis found that the changes in case definitions had a "substantial effect" on the figures, and adjustments increased the proportion of infections detected and counted by between 2.8 and 7.1 times.
But the report noted that changing case definitions are expected during an epidemic as scientific knowledge and laboratory capability evolves.
People wear face masks to protect against the spread of the coronavirus as they walk along a street in Beijing. Photo / AP
It added that countries without enough test kits should also include clinical diagnosis in their case definitions to gauge a better understanding of the number of infections – a necessity to determine the public health response.
The official figures in many countries, including the UK and the US, still rely on cases confirmed by testing alone.
The new estimates for coronavirus infections in mainland China are based on the fifth case definition used, published in early February. In this set of guidelines, authorities said official numbers should include cases diagnosed by symptoms only, rather than requiring infections to be confirmed by both clinical symptoms and laboratory tests.
The move provoked some controversy at the time and led to a substantial spike in cases, including a jump of roughly 15,000 in a single day. Authorities later reversed the decision to broaden the case definition.
But had this classification been in place since the start of January, researchers estimate that 232,000 people may have been diagnosed with Covid-19, including 127,000 in Wuhan.
By February 20 the city, where coronavirus first emerged, had reported 27,000 cases, according to official numbers.
"Still, this would be an underestimate of the number of infections up to that point because it would not have captured some mild or asymptomatic cases," the report's authors noted.
The impact of later versions of diagnosis guidelines cannot be gauged because no data based on symptom onset dated after February 20 has yet been released.
The study added that changing case definitions should be taken into account when looking at the rate of the epidemic's growth to prevent "substantially overestimating" the reproductive rate.
Dr Adam Kucharski, an epidemiologist and mathematician at the London School of Hygiene and Tropical Medicine, who was not involved in the study, said the modelling was a "useful study" and demonstrated "why we need to be very careful when interpreting the shape of the curve".
The publication of the study comes amid controversy around China's numbers – the country was accused of covering up the true extent of its outbreak when authorities revised Wuhan's death toll last week, adding 1,290 new fatalities. The government insisted the jump was due to reporting delays.
"This new article should not be used to promote views that there has been some kind of 'cover-up' in China," said Professor Chris Dye, an infectious disease expert at the University of Oxford.
"The paper probably offers a useful adjustment of the number of Covid-19 cases in China, but does not change the essential conclusion about the low incidence and effectiveness of Covid-19 control in China," he added.
