A cohort effect in cervical screening coverage?