False economy flaws screening procedures

Each month brings news of yet more women being told their smear tests, previously reported as clear, are now suspect. Such sagas derive directly from the NHS cervical screening programme's fundamentally flawed quality control.

A laboratory undertaking 100,000 tests annually can expect 7,000 abnormal slides. For those 7,000 women, the worst error results in this false-negative reporting, the rate of which greatly exceeds zero. Indeed, the programme's target standard of excellence for final reporting is 5 per cent, with 15 per cent being deemed acceptable.

If this laboratory's screening performance were 10 per cent, 700 abnormal smears would remain undetected after a single examination. Until recently, internal quality control typically involved re-screening 10 per cent of slides. This would pick up a further 63 abnormalities, leaving 637 women still wrongly believing their slides were clear. External quality control of laboratories focuses on their routinely reported abnormality rates. If, for example, this laboratory were to report a figure of 3 per cent, concerns would be raised. Screeners are also routinely tested for proficiency. Identification of poorly performing staff triggers a recall with all the attendant media coverage, litigation and compensation.

An alternative would be to build in redundancy to cope with the test's disappointingly low reliability. Keyboard operators in banks achieve an error rate of around one per 1,000. Screening all slides twice would reduce the expected number of undetected abnormalities from 700 to 70, but cost pounds6m a year - double the current cost of pounds1 per slide. Reducing that to seven would triple the expenditure. A 10-fold or 100-fold improvement at double or treble the cost seems an attractive proposition.

Were additional funds available, it might, though, prove difficult to recruit staff for a relatively low-paid job involving a two-year training period. In that case, what about training members of the long-term prison population, a proportion of whom would surely find this a constructive use of their time? An imaginative bonus system linking performance with remission could prove both attractive and effective.

Mindful of budgetary constraints, cytologists have evolved rapid screening of all slides. This additional half-minute screening can detect around 80 per cent of false negatives which slip through a 15-minute examination. This would still leave 140 of our 100,000 women labouring under a possibly fatal misapprehension. Although it meets the programme's 5 per cent target, this is still woefully inadequate. Remember the IBM advertisement: if your service level is 99 per cent, what do you tell the 100th customer?

Nobody appears to have asked why these rapid tests can't constitute the primary testing procedure. If an 80 per cent performance can be achieved, each slide could be examined say, six times in five minutes. This would yield an error rate of 64 per million against the current 5 per cent target. At one-third the cost, that constitutes an impressive cost-benefit ratio.