Abstract
Correction to Article Published in March 2008
In our paper ‘High-grade cervical abnormalities and screening intervals in New South Wales, Australia‘ 1 we estimated the number of extra cases of high-grade cervical abnormality annually if the screening interval policy in New South Wales were to be changed from two to three years as 267 by cytology and 225 by histology, assuming the same number of women participated triennially as biennially. These estimates were based on modelled increased risk of a high-grade lesion with a three-year versus two-year screening interval of OR = 1.36 (95% CI:1.28–1.44) for cytology and OR = 1.47 (95% CI:1.36–1.58) for histology, and also on pro-rating the existing two-yearly screening rate to three-yearly. Regrettably, we erroneously neglected also to pro-rate the original baseline abnormalities to three years along with the screens. The correct estimates should therefore be 144 (95% CI: 60–220) fewer cases of high-grade cytology, and 19 (95% CI: −50–89) fewer cases of high-grade histology.
Accordingly, we wish to make the following corrections to our paper:
Abstract, Results (p 36)
Replace: ‘We estimate that if the screening interval were increased from two to three years, and the number of women participating in triennial screening participation was the same as for biennial participation in NSW, then 267 (95% CI 186–347) extra cases of high-grade abnormalities would be detected annually by cytology and 225 extra cases (95% CI 160–291) confirmed by histology, mostly confined to women aged 20–49 years.’
with: ‘Despite the higher risks, we estimate that if the screening interval were increased from two to three years, and the number of women participating in triennial screening was the same as for biennial screening in NSW, then 140 (95% CI 60–220) fewer high-grade cytology results would occur per year, and there would be around 19 (95% CI −50–89) fewer high-grade cases confirmed by histology. As the 95% CI covers zero in the latter estimate, there is no significant difference in expected confirmed high-grade abnormalities between biennial and triennial screening if the same number of women are screened triennally as biennially.’
Abstract, Conclusion (p 36)
Replace: ‘Accordingly, our study provides evidence in support of retaining the recommended cervical screening interval at two years for HPV unvaccinated, well women.
with: ‘In terms of increased risk, our study provides evidence to support retaining the recommended cervical screening interval at two years for HPV unvaccinated, well women. However, in terms of extra high-grade abnormalities found if the numbers of women participating in triennial screening were the same as in biennial screening, fewer cases of high-grade cytology would be detected.’
Results (2nd para, p 40)
Replace: ‘These ORs translated into 267 (95% CI 186–347) extra cases of high-grade abnormality from cytology, and 225 (95% CI 160–291) extra cases of histologically confirmed high-grade abnormality, to be expected annually in NSW women if the screening interval were increased from two to three years and assuming the number of women screening three-yearly to be equal to the number screening biennially.’
with: ‘These ORs translated into 140 (95% CI 60–220) fewer cases of high-grade abnormality from cytology, and 19 (95% CI −50–89) fewer cases of histologically confirmed high-grade abnormality, to be expected annually in NSW women if the screening interval were increased from two to three years and the same number of women participated in triennial as biennial screening.’
Discussion (final para, pp 42–43)
Replace: ‘Finally, it is worth considering what high-grade cervical abnormalities actually imply. It may be argued, for instance, that it is immaterial if more high-grade abnormalities are detected because these will be treated and cancer is still prevented. However, harms found to be associated with treatment of high-grade cervical abnormalities, particularly by cone biopsy and loop electrosurgical excision, include higher risk of birth complications such as pre-term birth, low birth weight and premature rupture of membranes.13–16
with: ‘Finally, it is worth considering what increased detection rates (i.e. per woman screened) of high-grade cervical abnormalities would actually imply if the screening interval were to be extended. Harms found to be associated with treatment of high-grade cervical abnormalities, particularly by cone biopsy and loop electrosurgical excision, include higher risk of birth complications such as pre-term birth, low birthweight and premature rupture of membranes. 13–16 The present study implies that if the likelihood of these harms is spread over a longer period then the net effect is slightly in favour of extending the screening interval, as long as the same number of women screen triennially as biennially. However, this changes if women participate at significantly higher rates triennially than biennially. For instance, in NSW the triennial participation rate is ~73–76%, which is predominantly made up of women who screen biennially. We would expect the number of high-grade abnormalities detected annually to increase if this triennial participation rate were to remain the same, but the underlying mean screening interval approached three years rather than two years. Put another way, in the case of NSW, approximately 11,000 extra women with a negative result would need to screen three-yearly for (annual) confirmed high-grade abnormalities to exceed the number occurring under biennial screening. This increase would still produce less high-grade cytology, which would require 52,000 more negative women to screen triennially to equalize, but the net reduction in confirmed high-grade lesions would be short-lived if screening were to increase only by a relatively small amount.’
An easier way to compare two-yearly with three-yearly screening is to consider what would happen over six years, as below:
Two-yearly screening → 3 screens each with expected annual high-grade lesions p = 3p
Three-yearly screening → 2 screens each with pR = 2pR
Expected numbers of annual high-grade lesions in NSW women aged 20–69 years
Relative risk
For more details of the calculation, please contact SM.
We thank Mark Clements of the National Centre for Epidemiology & Population Health, Australian National University for bringing the error to our attention.