We all know that we should give it. But where does it go - and are our donations being used wisely? Jeremy Laurance investigates

For Richard Titmuss, in his classic sociological study The Gift Relationship: From Human Blood to Social Policy, it was the purest expression of altruism: men and women queueing at church halls and clinics across the country to donate blood. Even Tony Hancock yielded his "armful". About 10,000 gifts of blood are made each day, without expectation of any return (nobody wants a transfusion). Altruism is not yet dead.

But even the best-intended gift can carry a sting. Blood can transmit infections such as HIV, hepatitis B and C and variant CJD (vCJD). The idea that the life-giving blood supply might be tainted is terrifying. Since the late 1990s, when such hazards came to light, huge sums have been spent to ensure blood is "safe". But now voices in the blood service are asking whether these sums are securing the safe service we deserve. Millions are being invested in obscure tests of little benefit, while much more obvious risks are ignored.

There are four blood groups - O, the most common, A, B, and AB, the least common. They are not of equal value. Blood is also either rhesus positive (about 80 per cent of people) or rhesus negative. People with O-negative blood are known as universal donors because they can give to anyone, so this is the most valuable blood. For other groups, donor and recipient must be matched, and this is where errors can occur. So those with AB blood - 3 per cent of the population - can receive blood from any group but can only give to group AB.

Blood has four components - red blood cells, white blood cells, platelets and plasma - separated by centrifuge immediately after it is given. Whole blood is almost never used.

Red cells are given to replace blood lost during surgery. They carry oxygen around the body, but they only keep for 35 days - so new donors are needed all the time. Most red cells are used in routine surgery, with just 8 per cent given to accident victims. White blood cells - the leucocytes that provide the body's defence system - are filtered out and discarded because of the (low) risk that they could transmit vCJD.

Platelets, which help to form clots, are given to cancer patients having chemotherapy and others with clotting disorders, but they only have a shelf life of five days. Plasma is the yellow liquid in which the other parts float, which contains proteins. It is processed to extract albumin (used for burns), factor VIII (the clotting factor for haemophiliacs), and other substances.

However, because of the risk of vCJD (associated with eating BSE-infected beef), plasma given by British donors is discarded: all plasma for making blood products is imported.

Every unit donated in the UK is screened for infections, including HIV and hepatitis, and new screening tests are being identified all the time. Blood donors must answer elaborate questionnaires designed to exclude those at possible risk of infection. All this comes at a price. The cost of collecting blood has doubled since 1998 from £47 a "unit" - half a litre - of red cells to £90. The overall cost of the blood service has doubled in real terms since 1995 from £250m to £500m.

Two directors of the National Blood Service (NBS) have now broken cover to criticise the sums spent to reduce already tiny risks when much more serious problems - such as transfusion mix-ups in which blood is given to the wrong person - remain unaddressed.

Professor Marcela Contreras, the NBS's director of development and research in London, and Brian McClelland, its strategy director in Scotland, say that, in some cases, millions are being spent to remove a risk of infection that is so small it is negligible.

An example they cite in the British Medical Journal is the hepatitis C virus RNA test, introduced in 1999 to detect infection missed by the routine test carried out on all donations. So far, this sophisticated RNA test has been used on 1.4 million donations already screened for hepatitis C, at a cost of £8.3m, and one positive sample of hepatitis C has been discovered - a lot to spend to protect a single patient. Yet there is no specific funding to tackle the much bigger problem of errors in transfusion.

Dudley and Patricia Green know how catastrophic these can be. Their son James was left severely brain-damaged after being given the wrong blood because of a surname mix-up. James needed the transfusion because he fell seriously ill with jaundice at four days old. Instead of O-positive, he was given A-negative, triggering convulsions and heart failure. This happened at a hospital in Essex in the 1990s. James was left unable to walk, talk or do anything for himself. The North Essex Health Authority accepted liability and paid compensation.

The case was among those that led to the Serious Hazards of Transfusion (Shot) Group being set up in 1997 to monitor the safety of transfusions. It found the most common error in hospitals was the one that devastated the Green family - when blood intended for one patient is given to another.

These risks are far higher than the risks of infection. Shot's data for 1996 to 2003 spell this out; the chances of being seriously harmed by an error in transfusion are 80 times higher than the risk of contracting HIV, 45 times higher than of contracting hepatitis C, and six times higher than of contracting hepatitis B. In the period, 11 patients suffered serious injury and two died as a result of a transfusion error, for every million units of blood used.

Contreras says: "We should be spending more wisely on measures that are most effective. Once we start doing something we never reconsider it. We are doing the RNA test for hepatitis C for very little return. Commercial companies convince the politicians that a new test will work and it is introduced. We are led by the media and politicians."

Worse is to come. The process of leucodepletion - to remove white blood cells from all blood donations to cut the risk of vCJD - costs £70m a year, and there are plans to upgrade it, at even greater expense. In the UK, 17 people who gave blood before leucodepletion are known to have developed vCJD later and could in theory have passed the disease on. Fifty people have been identified who received blood from these donors, and two later turned out to be infected - though it is impossible to show that they contracted the disease from the transfusion.

We cannot know how many cases of vCJD leucodepletion may be preventing. It was necessary to introduce it because there were fears of a blood-borne epidemic. But it is inefficient, removing only about half the infectious agent. New processes involving "prion filters" will soon be available, which are claimed to reduce vCJD infectivity 10,000-fold. But they could add at least £100m a year to the nation's blood bill.

It might be unthinkable to remove a test, but serious thought needs to be given before tests costing millions are added, while simple errors in administering blood pose a greater threat.

There's a more fundamental question: do we need all the blood that is being used? Doctors give blood to patients in surgery because it has been the custom, not because it is based on good evidence of benefit. There is very little research on the use of blood, but what there is suggests it is overused. If patients are given unnecessary transfusions, they are being needlessly exposed to risks.

One trial cited by Contreras and McClelland found that critically-ill patients did better with less transfusion. Another found that giving two units to intensive-care patients with low haemoglobin had no effect on oxygenation levels. Another on sick babies showed that giving blood freely was no more beneficial than giving it sparingly.

Giving patients less blood, or giving blood to fewer patients, would reduce costs to the service and risks to the patients. It seems that this message is now getting through. Despite the increasing work-rate of the NHS, the demand for blood is falling. "We are using less this year (2004-05) than we did last year and we used less last year than two years ago," says Contreras.

But there is still much to learn. Rather than throwing money at reducing exceedingly small risks of infection, it is now vital that some of that money is spent investigating whether we need the blood we use.