We identified 218
847 women in the older group and 445
382 in the younger group who met our inclusion criteria (table 1). The average interval between the three consecutive negative results (that is, between the first and the second, and the second and the third smear) was 40 months in the older group and 39 months in the younger group, reflecting the fact that women could accumulate the three registered negative smears either in about the 20 years before 1996, when recommendations tried to limit screening to once in three years, or in the seven years since 1996 to 2002, with the recommendation of one smear per five years. In the period between 1 January 1994 and 31 December 2002, 1.3 and 2.6 million person years in follow-up accrued in these groups, respectively, an average of 5.84 and 5.83 years per woman (table 2). The two groups had a similar rate of screening after the third negative smear (table 3): about a third had none, about a third had one, and the remaining third had more than one further primary test registered. Forty two women in the older group and 105 women in the younger group developed cervical cancer (table 2).
Table 1 Description of study population by five year age groups
Table 2 Incidence of invasive cervical cancer after third consecutive negative smear result for two age groups
Table 3 Percentage of primary screening tests after third consecutive negative smear result*
During follow-up, the difference in the cumulative incidence rate between the age groups was never significant (fig 2 and table 2) . Table 4 shows the average absolute yearly incidence rates.
Fig 2 Cumulative incidence rate for invasive cancer by age group and time since third consecutive negative smear result
Table 4 Incidence rates per 100 000 woman years at risk (95% CI) by age group and year in follow-up
The average age of women was 37.3 years in the younger group and 48.7 years in the older group. Pooling women into two large age groups does not seem to have affected this result, as the cumulative incidence rate for cervical cancer at 10 years in smaller five year age groups also did not differ significantly (table 1, P=0.24).
The overall hazard ratio was 0.84 (95% confidence interval 0.59 to 1.21) for the older compared with the younger group. The test for time dependency of the relative hazards was non-significant (P=0.86).
We also varied the criterion for study eligibility from requiring three to requiring either two or four consecutive negative results. This changed the absolute level of risk but not the relation between the two age groups. After two consecutive negative results, the 10 year cumulative incidence rate for cancer was 45/100
000 (39 to 52) in the younger group and 48/100
000 (38 to 61) in the older group. Within 10 years after four consecutive negative smears, the cumulative incidence rate in the younger group was 47/100
000 (34 to 65), whereas in the older group it was 26/100
000 (14 to 47). By 15 years, however, the cumulative incidence rate after four consecutive negative smears in the older group caught up with that in the younger group (74/100
000 (45 to 121) in the younger group and 82/100
000 (26 to 257) in the older group).
We also calculated the cumulative incidence rate with cervical intraepithelial neoplasia grade I+ as the end point (table 5, fig 3) . By 10 years, the cumulative incidence rate was 1258/100
000 (1209 to 1308) in the younger group and 594/100
000 (547 to 645) in the older group. The difference between both groups was significant throughout the entire follow-up. Use of cervical intraepithelial neoplasia grade II+ or grade III+ as the end point instead of grade I+ also showed that preinvasive lesions are more commonly detected in the younger groups. The cumulative incidence rate of cervical intraepithelial neoplasia grade II+ was 721/100
000 (684 to 759) among younger and 258/100
000 (227 to 293) among older women. The cumulative incidence rate of cervical intraepithelial neoplasia grade III+ was 445/100
000 (417 to 476) among younger and 165/100,000 (140 to 194) among older women.
Fig 3 Cumulative incidence rate for cervical intraepithelial neoplasia grade I+ (CIN I+) by age group and time since third consecutive negative smear result
Table 5 Incidence of cervical intraepithelial neoplasia I+ (CIN I+) after third consecutive negative smear result for two age groups