INVESTIGATION OF THE EFFECT OF OCCULT INVASIVE CANCER ON PROGRESS TOWARDS SUCCESSFUL CERVICAL SCREENING

Objectives-To describe the effect of occult invasive disease on progre ss towards meeting the Health of the Nation target for reducing the in cidence of cervical cancer, and to investigate the possible effect of a higher risk of cervical cancer in women born since 1940. Setting-Sou thampton and South West Hampshire (SSWH), with a total female populati on of 218 549 in 1990, the midpoint of a study period covering 1985 to 1995. Methods-Incidence was calculated per 100 000 women years at ris k in overlapping three year periods for symptomatic and screen detecte d cancers. The same method was used for cohorts of women born before a nd after 1940. Screen detected stage Ia, cancers were identified as a subgroup. Results-The incidence of invasive cervical cancer fell by 27 .4%, from 16.8 to 12.2 per 100 000 women years at risk between 1985-87 and 1993-95, which was a significant linear trend (chi(2)=4.494, df=1 , p=0.034). The corresponding figures adjusted for age in a standardis ed European population were 16.3 and 11.5: a fall greater than require d to meet the Health of the Nation target set for the year 2000. Incid ence remained relatively high until screen detected cancers, more than one third of which were stage Ia,, had passed a peak in 1992. When sc reen detected stage Ia, cancers were excluded, incidence fell by 41.2% , from 16.5 to 9.7 per 100 000 women years at risk: a highly significa nt linear trend (chi(2)=12.391, df=1, p<0.001). The incidence in the f irst three years of the study was higher in women born between 1940 an d 1954 than in those born between 1925 and 1939, though the reverse wo uld be expected by age and the natural history of the disease. In the 1940-54 birth cohort 44% (23/52) of screen detected cancers were stage Ia(1), with a peak in 1992. When these were excluded, incidence fell by 57.1%, from 31.7 to 13.6 per 100 000 women years at risk: a highly significant linear trend (chi(2)=13.704, df=1, p<0.001), whereas an in crease would be expected for a cohort aged from 30-45 to 40-55. In the 1925-39 cohort only 24% (8/33) of screen detected cancers were stage Ia,. When these were excluded, incidence fell by 35.3%, from 24.9 to 1 6.1, which was not a significant linear trend (chi(2)=0.409, df=1, p=0 .522). Conclusion-An overall decline in incidence was not achieved unt il prevalent occult invasive disease had been detected by improved scr eening. The data confirm the effectiveness of screening, particularly in a high risk cohort of women born between 1940 and 1954.