For the arthritic grandmother who divides her time between Yorkshire and her daughter's family in London, today's report by the National Audit Office on information technology in the NHS will make depressing reading. For the cancer sufferer seeking expert opinions at several different hospitals, it will be worse than that. The report estimates that the centrepiece of the National Programme for IT in the NHS (NPfIT) - a secure, shareable, constantly updated electronic medical records service for all 60 million NHS patients - will not be ready until 2014 at the earliest. Privately, experts worry that as presently designed it may never work at all.

At the heart of the Care Records Service's well-documented problems are two costly software products that are custom-built, national in their intended application and, so far, unworkable. There is an alternative. It would run on software bought off the shelf, that can be chosen and customised by individual health authorities, and would not crash. Not only is it not too late to adopt this approach; it is already being used.

An unmodernised NHS is unthinkable. Bold and intelligent use of information technology is inevitably at the heart of this modernisation, and not just by streamlining administration. In principle, electronic records will save lives, whether by cutting errors arising from doctors' scrawls or giving paramedics instant information on accident victims' blood types and allergies. They will also improve patient care. As digital downloads replace clipboards on hospital beds, nurses' efficiency will rise. As new drugs are approved, every patient who might benefit will know. The vision of a vast, beneficent healthcare computer network is not fanciful. It is what taxpayers have been paying for since 2002, and should expect.

When first outlined, too hastily, to Tony Blair, the national programme was not only the biggest non-military IT project in history, but also uniquely complex. Hence the assumption that only bespoke software could make it work. As specifications evolved, budgets and timelines stretched. Early versions of the care records software proved inadequate, recriminations flew and far too little was accomplished. Taxpayers, at least, have not yet suffered ruinously: of the programme's projected total cost of £12.7 billion, only £3.6 billion has actually been spent.

By no means all this has been wasted. The high-speed broadband network to which strategic health authorities must eventually connect their own systems is complete and working. The PACS system for sharing digital imaging is also up and running, as is the choose-and-book system designed to give meaning, at last, to patient choice. But the two main Care Records Service (CRS) programmes, known as Lorenzo and Cerner, are still years from full deployment.

Pilot versions have been plagued with glitches, which few expect will all be solved by 2014. NHS trusts required to use them have been customising the bespoke software rather than adapt their own systems to a national standard. Some foundation trusts, meanwhile, are abandoning the standard altogether to buy database software on the open market, with no cost in confidentiality or performance. Other trusts would like to opt out, but under current rules cannot afford to. These rules must be changed.

The national programme has been appallingly mismanaged. This does not invalidate the vision behind it, but it does oblige the NHS to trust its trusts to make it a reality.