ObjectiveTo determine trends in pregnancy-associated cancer and associations between maternal cancer and pregnancy outcomes.DesignPopulation-based cohort study.SettingNew South Wales, Australia, 1994–2008.PopulationA total of 781 907 women and their 1 309 501 maternities.MethodsCancer and maternal information were obtained from linked cancer registry, birth and hospital records for the entire population. Generalised estimating equations with a logit link were used to examine associations between cancer risk factors and pregnancy outcomes.Main outcome measuresIncidence of pregnancy-associated cancer (diagnosis during pregnancy or within 12 months of delivery), maternal morbidities, preterm birth, and small- and large-for-gestational-age (LGA).ResultsA total of 1798 new cancer diagnoses were identified, including 499 during pregnancy and 1299 postpartum. From 1994 to 2007, the crude incidence rate of pregnancy-associated cancer increased from 112.3 to 191.5 per 100 000 maternities (P < 0.001), and only 14% of the increase was explained by increasing maternal age. Cancer diagnosis was more common than expected in women aged 15–44 years (observed-to-expected ratio 1.49; 95% CI 1.42–1.56). Cancers were predominantly melanoma (33.3%) and breast cancer (21.0%). Women with cancer diagnosed during pregnancy had high rates of labour induction (28.5%), caesarean section (40.0%) and planned preterm birth (19.7%). Novel findings included a cancer association with multiple pregnancies (adjusted odds ratio 1.52, 95% CI 1.13–2.05) and LGA (aOR 1.47, 95% CI 1.14–1.89).ConclusionsPregnancy-associated cancers have increased, and this increase is only partially explained by increasing maternal age. Pregnancy increases women’s interaction with health services and the possibility for diagnosis, but may also influence tumour growth.

Healthcare Standards (ACHS).2 The ACHS defines the 20 th centile as a "best practice rate" that is potentially achievable and uses it to identify and prioritise clinical areas in which research and quality improvement activity would have the greatest benefits. Using the 20 th centile rate for quantifying the potential impact on the overall caesarean rate of reducing practice variation is appealing, as it does not rely on an arbitrary target value but instead is data-driven, being influenced by the rates currently achieved by 20% of hospitals. Importantly, appropriately risk-adjusted caesarean rates, rather than the observed caesarean rates, should be compared with the 20 th centile.However, we urge caution in immediate use of the 20 th centile, as lower caesarean rates should not be achieved at the possible expense of increased adverse outcomes. Our initial study 1 aimed to identify clinically relevant groups with the greatest (and potentially most modifiable) variation in hospital caesarean rates, adjusted for casemix. A detailed analysis of outcomes was beyond the scope of that paper. We are currently exploring the Robson Groups with the greatest variation in hospital caesarean rates using more comprehensive data obtained through record linkage. Our aim is to identify further factors contributing to variation in each group, if any, and to examine the relationship between caesarean rates and maternal and neonatal outcomes.