Fragile figures

Sir Nigel Crisp, the NHS chief executive, freighted his annual report - published last week - with statistics to demonstrate that "something really significant is happening in the NHS".

Expect those statistics, others like them and exhaustive accompanying analysis to saturate the media between now and the general election. Performance data on the whole of the public sector, and the NHS in particular, will be integral to judging the government's effectiveness when it faces the voters.

But just how robust is this information? No one has suggested any figures in Sir Nigel's report were wrong, although he was of course selective in what he presented. And while it would be unfair to say the publication gave a misleading impression, it didn't necessarily paint the whole picture.

That is perhaps in the tradition of annual reports everywhere, and maybe Sir Nigel can be excused for blowing the NHS's trumpet, given journalistic reluctance to ever highlight its successes.

At least the days of deliberately fiddling the figures seem to be over, according to the Audit Commission. "We are encouraged to find that dishonest manipulation of waiting-time data is rare and is a marginal problem," it said earlier this year.

Yet public confidence in public sector performance data remains fragile. The Conservatives have already tried to exploit this, accusing the prime minister of pressurising civil servants into proving that additional spending is reaching the front line rather than being swallowed up by bureaucracy.

Tony Blair was quoted in a leaked memo claiming there was "a mismatch between the public's experience of local service improvement and of their knowledge of the statistics which made up the national picture".

Calls from the independent Statistics Commission last week for a new statutory code of practice and changes in the way official statistics are managed reinforce the doubts.

For the NHS, the problem is deeper than simply having to suffer politicians' contentious interpretations of facts and figures: the service has a long history of producing poor-quality data in the first place. Returns that should show similar results may vary widely. Coding can be inaccurate or incomplete, and data sets can be missing or entered twice into a system.

Senior staff may often show a striking lack of awareness about important data, while those on the front line may not know why they have to collect particular information or what its benefits are, so they perceive it as irrelevant to patient care and demanded on a Whitehall whim. Even where data is fit for a specific purpose, it may often be used for another for which it is not.

All this suddenly matters more than ever, and not just because of the next election campaign. Modern healthcare routes patients through complex pathways between a range of professionals in hospitals and the community. Treatment may easily go awry if information about them is inaccurate, imprecise or slow to arrive.

Appraisal and revalidation of doctors, currently being introduced across the NHS, will rely heavily on robust clinical information, evidence and data. Meanwhile, the Healthcare Commission is clearly not going to rest until it has replaced the star ratings with more meaningful performance information, and so looks unlikely to tolerate any woolly or approximate data.

But it is the advent of 'payment by results' which makes accurate data collection imperative for NHS organisations as never before. Trusts will be paid for work they do according to a tariff, regardless of what it has cost them to carry out the work. To guard against going into the red they will therefore need to know their costs very accurately, and for that they will have to enter into their computer systems precise information about each case diagnosis.

Various staff share this responsibility at the moment, among them clinical coders whose main role is to enter data from doctors' case notes and discharge summaries. In the United States and Australia this is largely a graduate profession; in the NHS it suffers from low pay and low status, with attendant problems of recruitment and retention.

Along with inculcating in all staff the importance of collecting accurate data, upgrading the status of clinical coders could make a significant strategic improvement in the NHS.

But it's unlikely to happen: any such move would play into the hands of the opportunist politicians forever ready to accuse the NHS of squandering its resources on pen-pushing bureaucrats.

· Peter Davies is a freelance writer specialising in health. You can contact him at petergdavies@ntlworld.com