About 23 per cent of people infected with COVID will become "long hauliers", finds a study that identifies predictors of who is likely to develop the sometimes-debilitating symptoms that can last for months.

The study finds that obesity and hair loss at the time of infection are predictors of long COVID, but that other underlying conditions such as diabetes or smoking status have no discernable link to long-lasting symptoms.

The research, which appears in Scientific Reports, is unique because it accounts for preexisting symptoms such as fatigue and sneezing that are common to other conditions and may be mistaken for COVID symptoms.

"Long COVID is a major public health concern. Twenty-three per cent is a very high prevalence, and it may translate to millions of people," said Qiao Wu, a doctoral candidate at the University of Southern California.

"More knowledge on its prevalence, persistent symptoms and risk factors may help health care professionals allocate resources and services to help long hauliers get back to normal lives," Wu added. While SARS-CoV-2 is typically an acute illness lasting about three weeks, some people with COVID have symptoms that last months or longer.

The World Health Organisation (WHO) defines long COVID as symptoms that last 12 weeks or longer. The team used an Internet-based national survey and interviewed 308 infected, non-hospitalised individuals one month before, around the time of infection and 12 weeks later.

After accounting for pre-existing symptoms, about 23 per cent of the participants reported that they had experienced new-onset symptoms, including headache (22 per cent); runny or stuffy nose (19 per cent); abdominal discomfort (18 per cent); fatigue (17 per cent); and diarrhoea (13 per cent).

In addition, the researchers found that people had significantly higher odds of experiencing long COVID if, at the time of infection, they were obese; experienced hair loss; headaches; and had a sore throat.

Unexpectedly, the odds of long COVID among people who experienced chest congestion were lower. There was also a lack of evidence relating the risk of long COVID to pre-existing health conditions such as diabetes or asthma, age, gender, race/ethnicity, education or current smoking status.