Transfusions of blood products might help to cut deaths in a future flu pandemic, research suggests.
Spanish flu killed millions
US researchers examined records from the Spanish flu pandemic of 1918-1920 which killed up to 100m worldwide.
They found transfusions taken from people who had recovered may have improved the condition of others hospitalised by the virus.
The study, a joint project by various groups including the US Navy, will appear in Annals of Internal Medicine.
Experts fear that the H5N1 strain of bird flu responsible for the deaths of 140 people in Asia since 2003 could mutate to gain the ability to pass easily from human to human.
If this happens, they fear the lives of millions of people worldwide could be under threat.
The latest research suggests that blood transfusions might be an effective addition to the treatment arsenal, alongside vaccines and anti-viral drugs.
The World Health Organization has warned the world is not ready for a flu pandemic, and that a vaccine will take time to prepare and distribute following the first outbreaks. Anti-viral drugs might also initially be in short supply.
The researchers say a single recovering patient could donate enough blood plasma to treat many others.
They examined the Spanish flu outbreak because the virus behind it is a close cousin of H5N1.
However, they admit the records on which they based their study were limited.
The number of patients was small, the work was not scientifically well controlled, and dosages were not standardised.
Disruption caused by World War 1 probably made record keeping very difficult, and wartime censorship may have affected publication of research.
The researchers are calling for experts to examine the potential for plasma therapy - known technically as serotherapy - in greater detail, and to consider the benefits of drawing up treatment guidelines.
Writing in the same journal, Dr John Treanor, an expert in infectious diseases at the University of Rochester, said a similar approach had been used to treat other viral diseases.
He said plasma therapy might pose logistical difficulties, such as how to obtain, classify and prepare blood materials in the midst of an outbreak.
In addition, more work was needed to refine suitable dosages.
However, he added: "Although many logistical hurdles exist, controlled clinical studies done now will probably pay a considerable dividend when the pandemic begins.
"We can, should, and must explore these issues about serotherapy now, in advance of the pandemic."
Dr John Wood, a virology expert at the National Institute for Biological Standards and Control, said the research was intriguing, and demanded further investigation.
He said: "Our present armoury against pandemic flu is limited. We can probably make vaccine, but it will be five months before it is ready, and there are logistical problems with anti-virals.
"It is better to have as many strings in your bow as possible."
However, Dr Andrew Hayward, an immunologist at University College London, said it might prove difficult to carry out research into the effectiveness of the treatment before a pandemic strain of flu emerged.
He also warned that any treatment involving blood products carried the potential for transmission of other infections.
He said vaccines, anti-virals were likely to be the front line defences, along with basic public health measures, such as hand washing and wearing masks.
Dr Iain Stephenson, a consultant in infectious diseases in Leicester, said any potential treatment for pandemic flu was worth exploring.
However, he said: "Immunotherapy is likely to work only in the very early stages of infection and it will be difficult to identify patients that might benefit."