Effect of supplementation with vitamin A or β carotene on mortality related to pregnancyBMJ 1999; 319 doi: https://doi.org/10.1136/bmj.319.7218.1201a (Published 30 October 1999) Cite this as: BMJ 1999;319:1201
Pooling of groups may not be appropriate
- K Vijayaraghavan, senior deputy director,
- Kamala Krishnaswamy, director
- National Institute of Nutrition, Jamai-Osmania, Hyderabad, 500 007, India email@example.com
- Division of Clinical Epidemiology, Department of Pediatrics, Maulana Azad Medical College, New Delhi 110 002, India firstname.lastname@example.org
- London School of Hygiene and Tropical Medicine, London WC1B 3DP
- Department of International Health, Johns Hopkins University, Baltimore, MD 21205, USA
- Kathmandu, Nepal
EDITOR—West et al reported that vitamin A and β carotene reduced mortality related to pregnancy up to 12 weeks post partum,1 but the conclusions cannot be considered to be unequivocal. Any differentials in ascertainment of pregnancies in the different groups could vitiate the results. About 6-7% of the pregnancies ended as miscarriages and stillbirths, and 2% had no reported outcome. Were there any differences between the groups in this outcome? If the 157 women were lost to follow up because they had died it would spuriously decrease mortality in the vitamin A group compared with the placebo group.
Although only half the women received ≥80% of the intended quantity of vitamin A or β carotene supplements, mortality was reduced substantially (40-50%). The authors report that the evidence of survival was noticed after 1.5 years of the trial, indicating a cumulative effect of supplementation, which depends on duration of feeding. Thus the women participating for shorter periods would have consumed lower amounts than those participating for longer periods even if they had consumed the intended quantities. In the absence of data on the relation between the total dose and maternal mortality, the authors' contention that “the risk of maternal death in vitamin A deficient populations could be substantially lowered with modest increase in vitamin A or β carotene intake” is hypothetical.
Several relative risks were not significantly different when tested separately for the vitamin A or β carotene groups but were significant when the groups were pooled. Is the significance spurious because of the increase in numbers due to pooling?
Supplementation was associated with protection from death due to injuries and other miscellaneous causes not attributable to it. The recalculated relative risks, excluding the deaths due to injuries, were significantly different in the β carotene group (0.523; 95% confidence interval …