The Read clinical codes fiasco showed up weaknesses in NHS funding and project management. But the fatal flaw was a stubborn insistence on ignoring US developments and going it alone, argues Peter Mitchell

In April, the Department of Health finally admitted the inevitable: the Read codes - aka Clinical Terms 3 - cannot survive on their own and are to be merged into the US SNOMED clinical vocabulary.

Now everybody makes mistakes, and big organisations like the NHS sometimes make very big mistakes purely because of the scale of their operations. That is inevitable, and it is also excusable: otherwise no-one would ever learn by experience.

But when an organisation spends tens of millions of pounds on a project that many experts advise from the start is impracticable; when it persists in the development even though no-one else shows any interest in buying into it, and when the pilot testing phase is halting, bug-ridden and unpopular with users. Is that excusable?

It is tempting to say no, it is not, that CT 3 was foreseeably a fiasco from start to finish, in which money was carelessly washed down the drain through dismal management practice. That it was another manifestation of what logicians call the Concorde fallacy: once you have dug yourself deep enough into a hole, and buried so much money in it that there is no way back, then you might as well keep digging.

But is that fair? Was CT 3 really a no-hoper from the start, or was it just bad luck? In many ways, it was only a brave effort gone wrong. US terminology expert Dr Chris Chute of the Mayo Clinic - who has now been called in to design an independent assessment of the codes - says that the content and design of Clinical Terms 3 'constituted one of the major clinical terminology efforts on the planet'.

It was not technical problems, but political, personality and pricing issues that scuppered Read, says Dr Chute. As long ago as 1994, he says, he tried to get the US medical establishment interested in the system. He failed.

'SNOMED's future was quite fragile at that time', he says. 'In any event the editor of SNOMED and James Read were, to put it mildly, not the best of friends. Few people wanted to hear of 'us' working with 'them'. It did not help at all that there was a widespread perception of unreasonable pricing for non-UK users of the Read codes.'

At the same time, HSJ readers will recall, the facts about Dr Read's unacceptable financial arrangements with the NHS Executive became public. Two years later, when Dr Read was essentially out of the picture at the NHS's coding centre, Dr Chute paid a visit to evaluate the progress of CT 3. He was impressed.

'The codes had very well developed content and they had intelligently adopted academic work for the most part in their structures and design,' he says. But what they lacked was a robust maintenance environment - essential to such a complex field - which the SNOMED people were beginning to develop.

Along with the international rejection of CT3, this was a major factor leading to the merger. There was no way the NHS could fund - and project- manage - such a wide-ranging job. The very nature of the system, with its hierarchy of qualifiers - described by one expert as a 'perpetual motion machine' - meant that it would be forever afflicted with the need to describe finer and finer coding detail.

So we should be celebrating, not lamenting, the end of Read. Common sense has prevailed. Should we therefore write it all off as a natural mistake that anybody could make?

No. One of the reasons for CT3's international failure was its inappropriate licensing contract - and that was at least partly attributable to the financial arrangements between Dr Read and the NHS. That could have been foreseen.

The second big mistake was to play at the Not Invented Here game. What was the point of developing a national system of clinical coding, different to that being developed in the US? If anything in human experience is universal it is our diseases.

Where there are national differences, there can be ways of 'reading across' between them - we know this because the vocabularies are now to be merged after all. Now, with a common international vocabulary, the price of medical software and knowledge bases can be reduced dramatically.

It is no use saying that the Americans were just as obstinate about SNOMED as we were with Read. That's true, but we have to recognise that the US, with its enormous internal market, is in the driving seat regarding technical standards. Every other industry accepted this long ago, and the NHS Executive should have done the same.

The lesson? The NHS might be big, but it is weak. Weakly managed and weakly funded. It can't afford to develop its own tools: it needs to take advantage of universal standards and products wherever it can.

Information for Health doesn't say much about this, which is its own single biggest weakness. It should have said: use the Internet for communications, not NHSnet; use HTML for your documents, not Word, PDF or any other proprietary format; use off-the-shelf software for data processing, even if it's American. Whatever you do: don't develop your own. That way lies the Concorde fallacy.