Vitamin a supplementation does not improve maternal survival
Methods: A cluster-randomized, double-blind, placebo-controlled trial was undertaken in 7 districts of the Brong Ahafo Region of Ghana to assess the effect of weekly, low dose vitamin A supplementation in women of reproductive age on pregnancy-related mortality, all-cause female mortality and survival of their babies. Eligibility criteria consisted of all women 15-45 years of age at the time of recruitment, who gave informed consent and who planned to live in the trial area for at least the next 3 months. There were a total of 272 field worker areas with each field worker area containing 4 contiguous clusters. Block randomization was used with 2 clusters in each fieldwork area allocated to vitamin A supplementation and 2 to placebo. There were 1086 clusters (2 field worker areas had only 3 clusters) which contained an initial maximum of 100-120 enrolled women. Based on the cluster of residence, the women were randomly assigned to receive either vitamin A in 544 clusters (25 000 IU/7500 μg retinol equivalents in soya bean oil) or placebo in 542 clusters (only soya bean oil). The capsule distribution started in December 2000 in a phased manner and ended in September 2008. Field workers visited women in one cluster every week over a 4-weekly cycle and they were given 4 capsules to be taken over the next 4 weeks. There was no direct observation of capsule consumption and compliance was supported by an extensive information, education and communication (IEC) programme. Data collection started in January 2004 and ended in October 2008. Four-weekly home visits by field workers, verbal postmortems and hospital-based data collection were done to gather information on pregnancy-related morbidity. Additionally, a serum retinol estimation sub-study was done to assess the vitamin A status of women in the trial along with a survey to assess the extent of night blindness in study participants. Specially trained IEC supervisors gathered data on sociodemographic characteristics, adherence to study treatment, exposure to IEC activities and any perceived sideeffects. The trial was supervised by the trial Steering Committee. Yearly blinded safety analyses and monitoring of vitamin A content in the capsules was done by the data monitoring and ethics committee. Intention-to-treat analysis was used to compare the treatment groups. Random-effects regression was used as the study had a cluster randomized design. For outcomes with the denominator as pregnancies or births such as perinatal mortality rate and pregnancyrelated mortality, logistic models were used and for those with the denominator as person-years such as infant mortality rate or all-cause female mortality, a Poisson model was used. Mean serum retinol concentrations were compared by t-test and the proportions in each group with serum retinol concentrations of <0.7 μmol/L were compared with the chi-square test. The number of participants recruited and followed up, and their characteristics, were similar in both intervention and control arms (104 484 women in the vitamin A arm and 130 297 in the placebo arm).
Results: Overall, adherence to all 4 capsules was high and similar in both the arms (84.3% in the vitamin A arm v. 84.1% in the placebo arm, for all 4 capsules taken in the past month). A combination of successful visit and adherence rates gave an overall estimated adherence of more than 70% in both arms. Vitamin A supplementation did not have any protective effect on 'pregnancy-related deaths' (adjusted odds ratio of 0.92; 95% confidence interval [CI] 0.73-1.17), all-cause female mortality (1.01; 0.93-1.09), pregnancy-related hospital admissions (0.98; 0.89-1.09), stillbirths (1.04; 0.96-1.13), perinatal deaths (1.01; 0.94-1.08), neonatal deaths (0.95; 0.87-1.04) and infant deaths (0.98; 0.91-1.05).