Cervical screening

Two expensive healthcare practices have recently been endorsed as policy in England and Wales.¹ One is continuing to invite women over 50 for cervical screening; the other is shortening the screening interval from five years to three for younger women. National decisions on single issues disregard competing needs and force local decision makers to neglect other, more pressing, problems.

To inform its decision the NHS cervical screening programme commissioned a case-control analysis.² The difference between three yearly and five yearly screening is too small to measure, which is why we are having to use estimates, despite a huge natural experiment involving widely differing screening intervals throughout the nation and worldwide. The analysis estimated that, for women under 40, the risk reduction is 30% with five yearly screening and 41% with three yearly screening.² For women aged 40-54 years it is 63% and 69% respectively. The paper mentioned that three yearly screening costs 60% to 66% more than five yearly, and harm from over-diagnosis and over-treatment increases as screening interval decreases. This seems to have had no influence on the recommendations.

What does the new guidance mean for a typical local programme? The Avon programme offers five yearly screening to 250 000 eligible women. Each year 59 000 women are tested, of whom 39 000 are aged 25-49.³ Switching to three yearly intervals means nine routine tests by age 50 instead of six. Allowing for the number already having tests for follow up, the actual change in workload for women aged 25-49 will be around 40%, or 15 500 additional women screened each year, together with consequent investigations, treatments, and counselling. At a conservative estimate, with liquid based cytology, of £25 ($46; €36) per woman per year,⁴ this will cost an extra £385 000 each year.

The estimates from the commissioned case-control study² suggest this could at best add one woman to the existing 24 women annually (assuming proportional share of national benefit) in whom death from cervical cancer is prevented by our local programme.⁵ Given known problems with case-control analyses it could be less.⁶ With five yearly screening, for each death prevented at least 150 women have abnormal results, and at least 50 are treated.⁷ During any one year in Avon 3000 women aged 25-49 receive abnormal results.³ Changing to three yearly screening could add another 1000 women to this group.

So for districts efficiently regulating a five yearly programme the new policy requires diversion of £385 000 annually into an activity that possibly helps one, and adds 1000 to the “worried sick” category. If the NHS were flush with money and no more beneficial activities were in the queue needing funding then this would be acceptable. But there is no spare money and there are other unmet needs.

How unstoppable is three yearly screening? Avon dealt with it in 1987, described its approach in the BMJ in 1991,⁸ and has been cited as an example of good practice.⁹ During 15 years with a strict five year interval, returning non-indicated tests to the sender, we receive perhaps three queries a year from women, their partners, their members of parliament, or the press. We respond promptly, with a full explanation, and have never encountered disagreement. I have attended many meetings, seminars, and workshops with women's groups to discuss the issue. Always there is keen interest and real readiness to understand the prioritisation issues we face in health care.

The new guidance recommends no screening for women under 25. This is welcome, as the harm to benefit ratio makes screening unethical; for one woman who could be helped by screening in this age group there are tens of thousands who have abnormal results. Controlling this work will help offset the extra cost of three yearly screening. Stopping routine screening beyond 50 would also help, but the NHS cervical screening programme has not sanctioned this. It takes 17 793 three yearly tests to detect one new high grade abnormality in well screened women over 50,¹⁰ and at most one death is prevented for each 36 high grades.⁷ To help one person older than 50 through routine five yearly screening therefore takes 420 000 tests, costing over £8m.

Suppose you are faced with the decision about investing £385 000 annually to benefit your local population. You can prevent one death every 22 years by routine five yearly screening beyond age 50, one death a year and harm an extra 1000 by switching to three yearly screening under 50, or 10 deaths a year through support that helps smokers stop,¹¹ and have enough spare to provide first rate nursing care and family support at home for 183 patients facing death from cancer.¹²

Which would you choose?