Introduction: The extent of early cervical rescreening, defined as rescreening earlier than the recommended interval for a given initial test result, is difficult to determine mainly because the testing history of women is incomplete, especially with newer population based screening registers.
Methods: Estimation of early rescreening is based on analysis of a 1 month cohort of women (February 1997) recorded on the New South Wales Pap Test Register (PTR) who initially tested negative and who had no recorded history of a positive test result. For the purposes of estimating early rescreening rates and sources of multiple screening occurring within the recommended 2 yearly screening interval for New South Wales, the cohort excluded the estimated proportion of women with an unrecorded history of a positive result. Approaches to exclusion were different for women with a history of a high grade result (CIN2 or higher) or a low grade result (CIN1 or lower).
Results: Characteristic rescreening peaks occurred at 6, 12, 18, 24 and 27 months according to negative result category. The rescreening peak at 27 months illustrates the effect of the PTR late reminder system. After adjusting the cohort for estimated proportions of women with a history of a lesion, the number of women estimated to have rescreened early was approximately 156 000 over 1997-8, and the early rescreening rate was estimated as 15.3% of women who have a Pap test. A feasible target for reducing rescreening through service provider interventions was estimated to be a reduction of 7.4%, based on reducing the overscreening proportion from 15.3% to 7.9% through truncating characteristic rescreening peaks to background levels of rescreening. This represents just under 200 000 screens that could be performed over 2 years on New South Wales women who are underscreened or unscreened without incurring additional costs due to screening.
Conclusions: Reasonable estimates of early cervical rescreening can be derived but some assumptions in estimating the proportion of women with a positive test result history may be needed particularly if a screening programme is new and universal recording of screening data is only recent. Characteristic peaks, as departures from a background of random noise in time plots of rescreening indicate that a component of early rescreening is systemic and the chief source of such variation in screening behaviour would be service providers.