The only way that the number of potential swine flu deaths can be accurately predicted is if the government adopts more stringent methods of monitoring the virus, researchers have said.
Scientists from Imperial College London believe that current methods make it difficult to plan suitable levels of health provision due to the fact that they either overestimate or underestimate the numbers.
The researchers, writing in the British Medical Journal (BMJ), said that the decision by many people to take care of themselves at home and put off visiting their GPs is the main source of the bias.
Working out a death rate based on the number of people dying compared with the number of cases officially being seen fails to take into account these people, they said.
Therefore, the actual number of people dying compared with the total number of cases will be lower than current figures suggest.
And extra uncertainty is produced because only a proportion of people are being tested and recorded as having the flu during the course of its spread.
The researchers said: "Initially increased awareness by patients and doctors may lead to high ascertainment, but as cases increase and systems are overwhelmed, only a proportion will be tested (potentially those with links to other confirmed cases), making it difficult to understand the scale of under-reporting."