We discuss methodological issues arising in a recent evaluation trial of a new antenatal care programme, as sponsored by the Special Programme of Research, Development and Research Training in Human Reproduction, and WHO's Division of Reproductive Health (Technical Support). The randomisation unit for the trial is the antenatal care clinic, with 53 clinics located in four countries randomly allocated to provide either the new programme or the traditional programme currently in use. Approximately 24,000 women presenting for antenatal care over an average period of 18 months will have been recruited.
PIP: The World Health Organization (WHO) Antenatal Care Randomized Controlled Trial is evaluating the impact of a new program of prenatal care on the health of mothers and newborns. Study subjects will receive either the standard prenatal care program currently offered at participating sites or a new regimen comprised of scientifically evaluated, objective-oriented prenatal care services. A total of 24,000 pregnant women from 53 prenatal care clinics in Argentina, Cuba, Thailand, and Saudi Arabia have been enrolled and stratified on the basis of the number of pregnant women enrolled in each clinic during the year preceding the study, the type of clinic (free-standing or hospital), and the administrative health system to which they belong. This article discusses methodological issues related to the study's design, with emphasis on sample size considerations, planned approaches to the statistical analysis, and data quality control. The rationales for selecting clinics as the unit of randomization are to reduce the risk of treatment contamination, encourage participation, and facilitate administrative and logistic convenience in the implementation of the intervention. Randomization of intact clinics to different intervention groups with predefined strata reflects the fact that the aim of the trial is to show the equivalence, not necessarily the superiority, of the new prenatal care program with the existing program of standard care. The two major adverse outcomes, a high maternal morbidity index and low birth weight, are expected to be in the range of 10%. To ensure that a statistically nonsignificant effect can be interpreted meaningfully, the trial has been designed to have a 90% power for ruling out an absolute difference of at least 0.02 in the expected incidence of the primary end points. A confidence interval approach was selected for sample size estimation, as recommended for equivalence trials, to provide additional assurance that the sample size is adequate.