Increased lab automation means that the tedious job of cross-matching blood samples will soon be passed on to computers, writes Peter Mitchell

Checking a blood sample from a patient against donor blood has been a routine pre-transfusion precaution for many years, even though banked blood is already categorised by ABO and Rh factors when it is collected.

The thinking has been that a lastminute cross-matching test can remove the small risk that atypical antibodies in the patient's blood, perhaps due to a previous transfusion or pregnancy, will cause an incompatibility reaction in the patient.

But is a physical match really necessary, given the rarity of positive results - most of which are false positives anyway? The conventional wisdom in haematology appears to be changing to the view that computers can do the job just as well.

'Mismatches hardly ever show up at this stage, ' says Ed Didsbury, haematology service manager of Peterborough District Hospital. For example, one of the atypical antibody types - anti-Kpa - is present in only one in 5,000 samples, with the corresponding blood group in only 2 per cent of donor bags.

Mr Didsbury calculates a Kpa mismatch will appear only once in 13 years at Peterborough. And one of the standard cross-match tests - antiglobulin - has already been dropped by 10 per cent of UK labs because of the rarity of a positive match and because of the improved quality of the antibody screen, he says.

So hospital laboratories are now being given the all-clear to move towards electronic cross-matching or EXM, saving staff time and money, as well as reducing delays in providing blood to trauma victims.

Peterborough went over to EXM last August, one of the first district general hospitals to do so. 'With this method, we can use computer software to check the patient doesn't have the [rogue] antibodies, ' says Mr Didsbury.

The patient's blood still has to be typed and screened for atypical antibodies, of course. This information is stored on the laboratory management system, which then automatically picks a bag of donor blood that matches it. The 'final check' is simply to positively identify the donor blood type by scanning the bar-code on the donor bag.

Only if the patient does have atypical antibodies will the lab system give a warning that a physical crossmatch must be done before issuing the blood bag - just as it would stop the issue of a bag of the wrong ABO type.

There is a big clinical saving, says Mr Didsbury. 'If the patient is one of the vast majority - perhaps 95 per cent - who don't have atypical antibodies, and do have a confirmed blood group, it becomes much quicker to treat them, ' he says.

'Because of today's large and increasing workload, and because most analysers do tests in large batches, it can take hours to do a physical cross match, or perhaps 30 minutes if it is urgent. With EXM it can be ready in two to three minutes.'

In conjunction with the introduction of automation and the EXM, Peterborough has been able to take on a 4 per cent increase in workload while at the same time reducing staff by one whole-time equivalent. It should also save up to£5,000 a year in consumables, he says.

But fewer than one in 20 UK hospital labs are using EXM at the moment. Partly this is because it relies on the lab having a high enough degree of automation. 'Very few hospitals have automated equipment with electronic transfer of results from their analysers into their laboratory management or pathology systems, ' says Mr Didsbury. Typically this could cost up to£100,000 (Peterborough uses an AutoVue blood analyser and the TelePath laboratory information management system).

However, the main issue is fear of an error - what if the wrong result was recorded in the patient's earlier blood test, or the donor bag contains the wrong blood type? Mr Didsbury thinks the risk is minimal: 'The NBA blood bags have a guaranteed blood group to one in 4 million. And we test all patients twice so that we are convinced we have recorded the right blood group for them.'

The only worry is that the patient does have undetected atypical antibodies. This is very rare and the consequences are not usually life threatening - typically a minor reaction, says Mr Didsbury.

The Institute of Biomedical Science has now accepted the general principle of EXM, provided that a 'secure system of control' is used alongside it. It also insists that humans are excluded from as much of the checking and data entry processes possible - for example, barcode scanners must always be used to read the bag labels.

The remaining debate is whether the patient's blood type needs to be determined by testing two separate blood samples - to guard against the risk that a single sample might be labelled with the wrong patient's name, or the blood analyser makes an error. The IBMS says it should; others, including Mr Didsbury, say that is too strict and that duplicate blind testing of a single sample is good enough.

After all, if two blood samples are used, the one used for cross-matching could just as easily be the wrongly labelled one, and give a wrong 'allclear' message. The only really satisfactory solution is to get humans out of the process altogether.

A case of mistaken identity How did the recent disaster at Addenbrooke's Hospital, Cambridge, happen, in which a woman died after being given the wrong blood group during a brain operation?

Addenbrooke's does use electronic cross-matching, though by a different name, preferring to call it 'computer issue'. But Mr Didsbury points out that the source of the error - misidentification of the patient - would not have been eliminated if a manual cross-match had been done.

'The cross-match is only one of the steps in the process leading to compatible blood products being issued, and it depends on the validity of the pre-transfusion sample, ' he says.

'There are many steps in the process that lead to a patient having a blood transfusion, a failure in any one of which, if undetected, could lead to an incompatible blood transfusion reaction.

'All hospitals and their laboratories will have procedures in place to minimise the risk of such errors.

'The art here is in finding a system that is simple yet secure. If there are too many steps and double-checks, then you can introduce an element of complacency.'

The error at Addenbrooke's, he says, was one of basic patient misidentification. The patient identity labels on the bags of blood (which would match the labelling of the pre-transfusion sample) were not compared to the wristband of the patient.

Nor is it the first time this error has been made: it is the same type of error that led to the removal of a patient's only healthy kidney in west Wales earlier this year.

But one of the downsides of maintaining a strict audit trail of patient identity from the bedside to the laboratory and back to the bedside is that pathology staff will come under criticism for daring to reject unsuitable requests and pre-transfusion samples, says Mr Didsbury.

'Unfortunately some non-laboratory staff consider such protocols an infringement of their medical judgement and take it personally, ' he says.

'If I have any doubt as to the validity of a pre-transfusion sample then it is disposed of, and the doctor is asked to start again.

'The electronic cross-match is a suitable alternative to a manual bench test and is perfectly acceptable. I, as a professional, would not use it otherwise.'

Cross-matching - what and why?

Three tests are carried out on a patient's blood before they receive a transfusion. They are:

a blood group test to determine the ABO and rhesus (D) type;

an antibody screen to detect irregular blood group antibodies, and the cross-match, to detect any remaining incompatibility between the patient's blood and that in the donor bag.

The first test is used to pick a donor bag of the correct type: everyone knows that being given blood of the wrong ABO type, and sometimes D type, can be clinically disastrous or even fatal. So ABO type is checked by testing the patient's blood sample twice, usually on an automatic analyser. If the lab is automated, the results are uploaded to the lab computer, which checks the two results against each other and against any previous test done on the same patient. The machine highlights differences to the operator before the transfusion is authorised. So errors are extremely rare.

The antibody screen is the test that shows whether a patient has acquired sensitivity to certain blood groups, either because of previous transfusions or from pregnancies. The failure to detect some blood group antibodies, such as Kell, Kidd, Duffy, Lewis, Lutheran or MNS types, can lead to blood transfusion reactions.

In the cross-match, the lab technician mixes a small amount of the prospective donor blood with a sample of the patient's blood, and watches for a reaction. Two tests are done: the antiglobulin test, to make sure there are no irregular antibodies that have been missed in the antibody screen; and the saline spin, to make sure there is no ABO mismatch. Such reactions are very rare, and can arise only if there has been a previous mistake in identifying the patient's ABO type, or if there has been some other human or technical error (eg mislabelling a donor bag), or if irregular antibodies are present that were not picked up at the antibody screen stage.

Even the very few positives that occur at this stage are usually false, caused by 'cold' antibodies with a raised thermal range and not considered to be clinically significant.

The only true positives are due to very rare (atypical) antibodies such as anti-Kpa or anti-Cw.

By far the most likely cause of a serious transfusion mistake is human error, but the risk has been greatly reduced with the introduction of lab automation and computerisation.

A typical blood analyser performs both blood group checks and antibody screens.

Such machines are highly accurate. In the five months after installation, the Autovue analyser at Peterborough did not make a single blood group error in 9,000 samples analysed and did not miss a single antibody in the evaluation phase of 1,000 duplicate samples.

By monitoring the bar codes that are attached to all the reagents and cards, as well as the patients' blood samples, it can also detect whether someone has tampered with the working area, or whether some of its reagents are expired. It then automatically discards the work and starts again.

There still needs to be a formal inspection of patients' samples, to make sure that they are correctly labelled.

Aside from that, humans can be almost entirely excluded from the process.