Thank you for your interesting and relevant comments on the association between cervical conisations and obstetric outcomes.1,2 We agree that a randomised study would be needed to fully establish a cause and effect relationship between previous conisation and obstetrics outcome. However, we only had the option to do a population-based cohort study, as a randomised study in patients with severe cervical dysplasia is not feasible in Danish patients, as neither the patients nor the Danish Ethical Committee would accept any randomisation that included a ‘no-treatment’ option.
Our data were analysed using a wide variety of statistical methods such as adjusted or unadjusted odds ratios, adjusted or unadjusted relative risks, and adjusted or unadjusted hazard ratios. After statistical counselling we ended up using unadjusted and adjusted hazards ratios from Cox analysis. This choice enabled us to include women who were delivered electively preterm because of maternal or fetal pathology, allowing their time to delivery to be included in the analysis. These patients were censored at the time of delivery. However, independent of which statistical method we used, we found the same strong relationship between one or two conisations and preterm birth. We included adjustment for possible confounding factors such as maternal age, smoking status, parity, marital status, and educational level, as stated in the methods section. In the final analysis, only confounding factors that affected the hazard ratio by more than 10% were included, to minimise the exclusion of women with missing values. As also clearly stated in the results section, only the mothers’ age, parity, and smoking remained in the final Cox regression analysis. We realise that there can be unknown unadjusted confounding factors, for example in vitro fertilisation, and we acknowledged this in the discussion. We did not adjust perinatal mortality before 28 weeks of gestation for preterm delivery before 28 weeks of gestation, as all babies that died before 28 weeks of gestation had also delivered before 28 weeks of gestation.
The causal relationship between conisations and preterm delivery cannot be established by a population-based cohort study, as socio-economic status, sexual habits, dysplasia per se and conisations can all be contributing factors to the risk of preterm delivery. We tried partly to compensate for these factors by: (1) adjusting for known confounding factors; (2) including a control group with earlier dysplasia not treated with conisation; and (3) evaluating women delivering both before and after conisation. All these results pointed in the direction of a strong relationship between conisation and preterm delivery. We found a stronger association between a deep cone and preterm delivery, and also two cones and preterm delivery. We have not been able to think of any confounding factor that would cause this relationship. We can conceive of only two explanations for these findings: (1) the more you remove of the cervical canal (a deep cone or two cones) the higher the risk of preterm delivery; or (2) the severity or persistence of the CIN in itself increases the risk of preterm delivery.
Despite our inability to demonstrate causality between cone biopsy and preterm delivery, it is a fact that we, as well as many others, have demonstrated an increased risk of preterm delivery in women with severe dysplasia treated by conisation. We consider that these women should be counselled according to this knowledge in subsequent pregnancies. We did not intend to undermine in any way the secondary prevention of cervical cancer, but we were hoping to encourage the primary prevention of CIN by the introduction of national vaccination programmes, so that in future, conisations with all their attendant risks will no longer be necessary.