Background A protective association between parity and ovarian cancer has been reported by many investigators. Some have reported heterogeneity of the association by histological subtype, but few studies have enough cases to explore this fully.

Methods We used data from the Million Women Study, a prospective cohort of middle-aged women in the UK. Participants were recruited in 1998 on average when they were invited for routine breast screening, and completed a questionnaire on a range of exposures, including the number of children they had had. Follow-up was via routinely-collected data from central cancer registries. Using Cox regression models, we estimated adjusted relative risks of ovarian cancer amongst parous women compared to nulliparous women, and per birth amongst parous women, adjusted by nine potential confounding factors. We used a competing risks approach to explore variation by histological subtype. All analyses were conducted using STATA 13.

Results After excluding those with bilateral oophorectomy and unknown parity, the study population included 1.1 million women. There were 6952 incident cases of ovarian cancer, after a mean follow-up of 12.2 years. Overall, parous women had a 26% lower risk of ovarian cancer than nulliparous women (RR: 0.74, 95% CI: 0.69–0.79), but there was significant heterogeneity by histological subtype (p-het=0.0002). The reduced risk appeared to be significantly stronger for endometrioid (n = 631, RR: 0.57, 95% CI: 0.46–0.69) and clear cell tumours (n = 343, RR: 0.46, 95% CI: 0.36–0.60) than for serous tumours (n = 2912, RR: 0.84, 95% CI: 0.75–0.94). Amongst parous women, each additional birth was associated with a 6% reduction in relative risk of ovarian cancer (RR-per-birth: 0.94, 95% CI: 0.91–0.96); there was no significant heterogeneity by tumour subtype.

Conclusion We have strong evidence that the reduced risk of ovarian cancer in parous women varies by histological subtype. Our results are in keeping with those from other epidemiological studies, and are consistent with findings from histopathological and genetic studies, which suggest that the different subtypes of ovarian cancer have different aetiologies. The greater degree of heterogeneity by subtype when comparing parous vs nulliparous women, than for increasing parity amongst parous women, may suggest the influence of infertility.