We thank Dr Thornton for his interest in our article ‘Aspirin in the prevention of pre-eclampsia in high-risk women’ published in the January 2013 issue of BJOG. He raises four points we would like to respond to: the strength of effect; sample size; exclusion of participants from the main analysis; and trial registration.
Dr Thornton concludes the commentary by writing that the effect of aspirin we reported in the meta-analysis for women who also have abnormal uterine artery Doppler waveforms at 14 weeks of gestation differs little from its effect in other high-risk groups. We have difficulty in agreeing with this; in our opinion the risk ratio in our meta-analysis (RR 0.55; 95% CI 0.37–0.83) differs from those reported by the Cochrane review.[2, 3] The differences are even better encapsulated by comparing the numbers needed to treat (NNTs). In our meta-analysis, the substantial reduction in risk together with the high initial risk (36%) resulted in an NNT of six for preventing pre-eclampsia, whereas the Cochrane review reported an NNT for pre-eclampsia of 19 for women at high risk (20% risk) and 119 for women at moderate risk (6%), and concluded that ‘further information is required to assess which women are most likely to benefit, when treatment is best started, and at what dose’. The PARIS meta-analysis reported a 10% reduction and an NNT of 56 for women at high risk (18% risk) and 167 for women at moderate risk (6%).
In other respects our study represents both the strengths and weaknesses of clinician-initiated trials. With limited funding, from non-commercial sources only, the participating clinicians in ten centres were able to screen 947 women whose history indicated an increased risk of pre-eclampsia, to find those whose uterine artery flow indicates a particularly high risk. As we discuss in the article, in hindsight the criterion chosen was too strict, and only 152 women could be randomised. Although Dr Thornton is correct in stating that 31 of these women discontinued the treatment for various reasons and were excluded from the main analysis, he seems to overlook our description of the intention-to-treat analysis of all randomised women (except those who had a miscarriage), which gave a similar result. We do agree about the shortcomings of the information included in the ISRCTN registration (http://www.controlled-trials.com/ISRCTN14030412/), submitted by one member of the study team. For example, variables that were eventually listed in the outcomes section include predictor variables such as biochemical measurements during pregnancy. It should also be noted that the planned number of participants stated on the ISRCTN website, of 1000 women, refers to the women to be screened by Doppler ultrasound, and not those to be eventually randomised, as Dr Thornton stated in his commentary.
In our conclusion we echoed the words of the Cochrane review that ‘further information is required to assess which women are most likely to benefit’, and proposed that emerging biochemical risk markers, possibly in combination with early uterine artery Doppler ultrasound, are promising candidates to identify such women for future trials.