Do you Read me?

Senior NHS officials are fighting to save what they see as a precious baby from being thrown out with some murky bathwater.

The baby is the Read system of clinical coding. It took a public battering following revelations about the system's inventor, James Read, and his company, Computer Aided Medical Systems. Newspapers have described the Read codes as a 'failed project'. A Commons public accounts committee report is likely to be scathing, portraying Read as the latest in a series of NHS computing fiascos.

NHS chief executive Alan Langlands and his head of IT, Frank Burns, say such attacks are unfair, and claim that the basic Read codes, version two, are an outstanding success. They are in use in 80 per cent of general practices and in some departments of about 150 acute hospitals. Mr Burns' own hospital, Wirral, relies on Read codes for its electronic patient record system.

Mr Burns is determined that the codes will survive. He told the Healthcare Computing 98 conference that the National Audit Office report 'did not say that Read codes did not work'. He was confident that NAO's call for new evaluations 'will allow us to eliminate any lingering doubts'.

But the real debate is over the codes' third version - 'Read 3'. It represents years of work by 2,000 medical professionals who developed a thesaurus of more than 270,000 terms used in the process of care. Although claimed to be of 'service strength', Read 3 is in use at only 12 pilot sites. Only one thorough evaluation has so far been published.

Mr Burns is likely to set a target for their adoption in his new information management and technology strategy - to be published imminently. There is, he says, no other way to produce the quality and detail of information needed for the performance measures set out in The New NHS.

The new strategy will probably set deadlines for Read 3 to be rolled out across the NHS. It will have to turn into reality the 'comprehensive marketing plan' called for in the critical 1996 report by consultants Silicon Bridge. However, ministers may prefer to wait for the independent review called for by the NAO. They will also have to weather the PAC's report, which is likely to be critical of Mr Langlands.

To sell Read 3 to his political masters and the wider NHS, Mr Burns will have to accomplish three difficult tasks. First, he must prove that the codes work well enough to justify the NHS's continuing support. The only rigorous evaluation published, carried out at Withybush Hospital in Wales, caused controversy. The Information Management Group says it shows that 'up to 95 per cent of the diagnosis and procedure terms needed by doctors' were in the thesaurus. But the report reveals that more work is needed. Critics say no single system of coding can capture the subtleties of medical language and any attempt to do so is doomed.

IMG officials admit there are shortcomings, but say that Read is the closest the NHS is going to get to an electronic language of health. And the arguments for using such a language are overwhelming.

The IMG's medical information group of clinicians says national performance indicators require a rich vocabulary of information if they are to compare like with like. 'This richness is simply not present in existing classification systems (such as ICD-10).' A single system is essential, the group says.

Second, Mr Burns must set up a publicly acceptable mechanism to develop and license the codes. The most controversial feature, the relationship with CAMS, comes to an end in March next year. The Silicon Bridge report comments that 'the unusual licensing arrangements with CAMS have contributed, at least in part, to the reluctance of the market to commit wholeheartedly to the use of Read codes'. The new contract will probably be put out to open tender.

Third, Mr Burns must show how the Read codes fit into coding systems being developed elsewhere. Despite extensive efforts by the NHS to sell Read codes, the only major overseas user is the New Zealand health service. The only market that would guarantee Read's long-term future is the US, which is highly unlikely to adopt it. US healthcare organisations are under pressure to standardise the way they encode electronic information and the Healthcare Insurance Portability and Accountability Act has set a series of deadlines, beginning this year. But the SNOMED clinical coding system has overwhelming support there.

Although it will be many years before healthcare data is routinely exchanged electronically around the world, the NHS must plan how Read will converge with SNOMED if it is to be taken seriously.

But such talk would inevitably raise suggestions that it might be easier to ditch Read and switch to SNOMED from the start. It will be interesting to see if Mr Burns will take the risk.