点击显示 收起
1 From the Departments of Nutrition (LCM), Paediatrics and Child Health (PJI and KJN), Immunology (LSZ), and Obstetrics and Gynaecology (PZ), University of Zimbabwe, Harare, Zimbabwe; the Johns Hopkins Bloomberg School of Public Health, Department of International Health, Baltimore (LHM and JHH); the Research Institute of the Montreal General Hospital, Montreal (BJW); and the ZVITAMBO Project, Harare, Zimbabwe (LCM, PJI, KJN, EM, LHM, LSJ, PZ, BJW, and JHH)
2 The ZVITAMBO Project was primarily supported by the Canadian International Development Agency (R/C Project 690/M3688), the US Agency for International Development (cooperative agreement no. HRN-A-00-97-00015-00 between Johns Hopkins University and the Office of Health and Nutrition of the USAID), and a grant from the Bill and Melinda Gates Foundation (Seattle); additional support was provided by the Rockefeller Foundation (New York) and BASF (Ludwigshafen, Germany). 3 Reprints not available. Address correspondence to JH Humphrey, ZVITAMBO Project, 1 Borrowdale Road, Borrowdale, Harare, Zimbabwe. E-mail: jhumphrey{at}zvitambo.co.zw.
ABSTRACT
Background: Young infants are at risk of vitamin A deficiency. Supplementation of breastfeeding mothers improves the vitamin A status of their infants, but there are no data regarding its effect on infant mortality, and data on the effect of directly supplementing infants during the first few weeks of life are conflicting.
Objective: The objective was to measure the effect on infant mortality of supplementing neonates and their HIV-negative mothers with single, large doses of vitamin A during the immediate postpartum period.
Design: A randomized, placebo-controlled, 2-by-2 factorial design trial was conducted in 14 110 mothers and their infants; 9208 of the mothers were HIV-negative at delivery, remained such during the postpartum year, and were retained in the current analysis. The infants were randomly assigned within 96 h of delivery to 1 of 4 treatment groups: mothers and infants received vitamin A (Aa), mothers received vitamin A and infants received placebo (Ap), mothers received placebo and infants received vitamin A (Pa), and both mothers and infants received placebo (Pp). The vitamin A dose in the mothers was 400 000 IU and in the infants was 50 000 IU. The mother-infant pairs were followed to 12 mo.
Results: Hazard ratios (95% CI) for 12 mo mortality among infants in the maternal-supplemented and infant-supplemented groups were 1.17 (0.87, 1.58) and 1.08 (0.80, 1.46), respectively. Hazard ratios (95% CI) for the Aa, Ap, and Pa groups compared with the Pp group were 1.28 (0.83, 1.98), 1.27 (0.82, 1.97), and 1.18 (0.76, 1.83), respectively. These data indicate no overall effect. Serum retinol concentrations among a subsample of women were similar to reference norms.
Conclusion: Postpartum maternal or neonatal vitamin A supplementation may not reduce infant mortality in infants of HIV-negative women with an apparently adequate vitamin A status.
Key Words: Vitamin A • neonatal vitamin A supplementation • maternal vitamin A supplementation • infant mortality • Zimbabwe
INTRODUCTION
Vitamin A supplementation programs for children aged 6 mo to 6 y living in places where vitamin A deficiency is a public health problem have been implemented in 50 countries, reaching >200 million children annually and averting an estimated 350 000 deaths each year (T Goodman, WHO, personal communication, 2004). Programmatic policy on vitamin A supplementation for infants during the first 6 mo of life has been more tentative because of conflicting data on its effect, despite clear evidence that young infants are at high risk of deficiency (1–4). Two approaches have been tested. First, provision of a single large dose to breastfeeding mothers within the first 6–8 wk postpartum increases breast-milk vitamin A concentrations and improves infant vitamin A status (1, 4), although its effect on infant mortality has not been studied, except in one retrospective analysis (5). Given its benefit on vitamin A status, feasibility, and low cost, supplementation of women with 200 000 IU vitamin A postpartum is currently being implemented in 15 countries as national programs and is recommended by the WHO for all areas where vitamin A deficiency is a public health problem (6). Recently, an Informal Technical Consultation convened by the WHO recommended that this postpartum dose should be increased to 400 000 IU, given as two 200 000-IU doses, within the safe infertile postpartum period (7). The recommendation was subsequently endorsed by the International Vitamin A Consultative Group (8)
Data on the effect of the second approach, direct supplementation of young infants, have been more inconsistent. In Nepal, a 50 000-IU dose given to infants at 2–3 wk of life had no effect on infant mortality (relative risk: 1.07; 95% CI: 0.66, 1.72) and a 100 000-IU dose given to infants at 1, 2, and 3 mo of age resulted in a nonsignificant 26% excess risk of death (9). In a trial conducted in Peru, India, and Ghana, no effect on infant mortality was observed from giving mothers 200 000 IU vitamin A postpartum and infants 25 000 IU vitamin A at the time of immunization (6, 10, and 14 wk) (10). Conversely, supplementation with 50 000 IU within the first 24 h of life was associated with a 64% reduction in mortality in a small (n = 2067) hospital-based trial in Indonesia (11). Surprisingly, the serum retinol distribution of the mothers of those infants was similar to that of a reference population of healthy women, which suggests that the vitamin A status of the mothers of these infants was adequate. It was hypothesized that vitamin A may have a particular benefit when given immediately after birth and that even relatively well-nourished populations in which infections remain the leading cause of infant death may benefit.
Two larger trials were planned to follow up the Indonesian trial. In the first trial, 6-mo mortality decreased by 22% after 11 619 infants in South India were given 48 000 IU vitamin A within 2 d of birth (12). In the second trial, the Zimbabwe Vitamin A for Mothers and Babies (ZVITAMBO) Project, the effects of single large doses of vitamin A given to infants, their mothers, or both on several maternal and infant outcomes were investigated. This article reports the infant mortality results in infants born to women who were HIV-negative to ascertain the effect of the vitamin A supplements uncomplicated by HIV. The results in infants born to HIV-infected women will be reported separately.
SUBJECTS AND METHODS
Enrollment of subjects
The study was conducted in Harare, Zimbabwe. Zimbabwe is categorized by the WHO as a high-risk area for vitamin A deficiency (6, 13). The common diet consists of maize-meal porridge and relishes of vegetables (especially tomatoes), legumes, dried small fish, small amounts of milk, and meat for those who can afford it. At the time of the study, there were no national vitamin A supplementation programs for postpartum women or neonates. Supplementation of postpartum women with 200 000 IU was initiated by the Zimbabwe Ministry of Health 18 mo after recruitment ended. The program was initiated because of the findings of a nationally representative survey conducted in 1999 in which 35.8% of preschool children and 7% of reproductive-aged women had serum retinol concentrations < 0.7 µmol/L (14). Compared with South India, malnutrition and vitamin A deficiency are less prevalent and severe in Zimbabwe, but HIV is endemic in Zimbabwe: 25% of adults are infected (15).
From 25 November 1997 to 29 January 2000, 14 110 mother-infant pairs were enrolled within 96 h of delivery at 1 of 14 maternity clinics and hospitals. Pairs were eligible if neither of the pair had an acutely life-threatening condition, the infant was a singleton with a birth weight > 1500 g, and the mother planned to stay in Harare after delivery. Written informed consent was obtained from the mother. Socioeconomic and demographic characteristics were collected by interview, and obstetric details of the pregnancy and delivery were transcribed from hospital records. Gestational age was estimated with the Capurro method (16). Infant birth weight was measured with an electronic scale (model 727; Seca, Hanover, MD), and infant length, infant head circumference, and maternal midupper arm circumference were measured according to methods described by Gibson (17). Addresses were recorded for the mother's urban and rural residences (it is common for urban Zimbabweans to travel frequently to extended family rural homesteads), for the mother's place of work, for the mother's husband's place of work, and for a relative of the mother who would always know of her whereabouts.
Blood collection and processing
The study nurses collected whole blood into EDTA and plain (serum) tubes from mothers and infants by venipuncture and heel prick, respectively. Blood for plasma was stored at room temperature (20 °C) and that for serum in a cool box (10–15 °C) before being transferred to the laboratory for processing within 2 h of phlebotomy. Plasma and serum were separated and stored in aliquots at –70 °C until used.
Randomization to treatment groups
Mother-infants pairs were randomly assigned to 1 of 4 treatment groups: mothers received 400 000 IU vitamin A (as retinyl palmitate) and infants received 50 000 IU vitamin A (Aa group), mothers received 400 000 IU vitamin A and infants received placebo (Ap group), mothers received placebo and infants received 50 000 IU vitamin A (Pa group), and both mothers and infants received placebo (Pp group). Treatment and placebo capsules appeared identical and both contained a soy oil base with vitamin E as a preservative (50 IU per maternal capsule; 10 IU per infant capsule) (Tishcon Corporation, Westbury, NY).
A separate team at Johns Hopkins University prepared the study capsule packets. Study identification numbers were randomly allocated to the treatment groups by computer in blocks of 12. The numbers were printed on adhesive labels and affixed to amber-colored zip-lock plastic bags that were packed with the assigned capsules. Capsule packets were prepared separately for each of the 4 treatment groups and were then merged into numeric order before shipping to Zimbabwe, where a series of packets were distributed to each recruitment site. As each mother- infant pair was recruited, the capsules in the next sequential bag were administered, and the associated study number was assigned to the pair. Lists linking the study number to the treatment were kept in sealed envelopes and encrypted computer files.
Follow-up of the subjects
Follow-up visits were conducted in 3 designated follow-up clinics at 6 wk, 3 mo, and then every 3 mo thereafter up to 12 mo. Home visits were attempted for defaulting pairs to either their urban or rural home anywhere within Zimbabwe. Cause of death was determined from medical records for infants who died in a hospital or from a review of verbal autopsy information by a study pediatrician, who was masked to treatment group, for infants dying at home. Multiple causes of death were permitted and were not ranked hierarchically, in keeping with the recommendations of an expert group convened by the World Health Organization (18).
Laboratory analyses
Mothers were tested for HIV at baseline; those whose results were negative at baseline were tested again at 6 and 12 mo postpartum by enzyme-linked immunosorbent assay and Western Blot. Only infants of mothers who remained HIV-negative to 12 mo postpartum were included in the current analysis.
Serum retinol concentration was quantified at baseline and 6 wk for a representative subsample of 375 mothers. Extraction was effected by using a 50:50 ethanol:acetonitrile solvent containing retinyl acetate as an internal standard. The serum:extraction solvent ratio was 40:100. The extracted retinol was quantified by HPLC with the use of Resolve C18 analytic and guard columns (Waters Corp, Milford, MA) with a mobile phase of 95:5 acetonitrile:water. Ultraviolet detection was used, optimized at 325 nm. Internal quality control serum samples with low, medium, and high retinol concentrations that had been validated against Standard Reference Material 968b from the National Institute of Standards and Technology (Gaithersburg, MD) were run with every batch of test samples. The laboratory participated satisfactorily in an external quality assurance program organized by the National Institute of Standards and Technology.
Data analysis and statistical methods
Statistical analysis was conducted with the use of SAS (version 8.2; SAS Institute Inc, Cary, NC) and STATA (version 8; College Station, TX). Baseline characteristics were compared across the 4 groups by using analysis of variance for continuous variables and chi-square for categorical variables. The a priori primary objective of the study was to evaluate the effect of maternal and neonatal vitamin A supplementation on infant mortality at 6 and 12 mo by using Kaplan-Meier estimates. Survival curves were compared by log-rank test. Cox proportional hazards models were used to test for potential interaction between the 2 treatments (maternal and infant vitamin A supplementation) on mortality, to model potential effect modification, and to adjust for potential confounding. The primary planned analysis was for the 2 main effects (assuming there would be no interaction between the maternal and infant treatments), with secondary analyses to examine the effects of each of the individual treatment groups compared with the double placebo arm. Adjusted models were constructed by entering potential prognostic factors (maternal: age, parity, arm circumference, marital status, education, income, religion, and death of a previous child; infant: weight, length, head circumference, gestational age, Apgar score, and mode of delivery), retaining those that remained significant at P < 0.10 before forcing treatment groups into the models. Final models were rerun excluding infants whose mothers' HIV status was known to be negative at delivery but was not available at the endpoint. Interaction terms were considered statistically significant at P < 0.10; all other comparisons were considered statistically significant at the conventional 0.05 level or when the 95% CI did not include 1.0.
The Medical Research Council of Zimbabwe, The Medicines Control Authority of Zimbabwe, The Johns Hopkins Bloomberg School of Public Health Committee on Human Research, and the Montreal General Hospital Ethics Committee approved the study protocol.
RESULTS
We enrolled and randomly assigned 14 110 mother-infant pairs (Figure 1). Of these mothers, 4495 were HIV-positive and 54 were HIV-indeterminate at baseline, and 353 mothers seroconverted during the postpartum year, which left 9208 persistently HIV-negative mothers for the current analysis. Among infants born to HIV-negative mothers and for whom vital status was known at 6 and 12 mo, maternal HIV-negative status could be confirmed at those times for 90% and 86%, respectively.
FIGURE 1.. Flow of participants throughout the trial. Treatment groups: Aa, mothers received 400 000 IU vitamin A and infants received 50 000 IU vitamin A; Ap, mothers received 400 000 IU vitamin A and infants received placebo; Pa, mothers received placebo and infants received 50 000 IU vitamin A; Pp, mothers and infants received placebo. pos., positive; neg., negative; ind., indeterminate.
Maternal, household, infant, and delivery characteristics at baseline did not differ significantly across the 4 treatment groups (Table 1). At baseline, 37.1% of the mothers had a serum retinol concentration < 1.05 µmol/L. About 24% of the mothers were teenagers and nearly half were primiparous; >94% of the mothers were married or in a stable relationship. The vast majority of mothers were literate; 82% had 8 y of formal education. Total household income was <US$1.20 and <US$4.75 per day for 12% and nearly 80% of the women, respectively. The mean birth weight of the infants was 3.0 kg; 12% weighed <2500 g. and 6% were born prematurely (<37 wk). Three-quarters of the pairs received their treatment dose within 24 h, and 94% received it within 48 h of delivery. All of the mothers initiated breastfeeding (97% within 12 h of delivery) and 99.5% were still breastfeeding at 6 mo (data not shown).
View this table:
TABLE 1. Baseline characteristics according to treatment group1
At 6 wk, maternal serum retinol concentrations rose in all 4 groups, which reflected the normalization of pregnancy-associated hemodilution (19), but the mean (±SD) increase was greater (P = 0.0096) in the 2 groups in which mothers had received vitamin A (0.58 ± 0.49 µmol/L) than in the 2 maternal placebo groups (0.42 ± 0.51 µmol/L). On the basis of these 6-wk values, there was little evidence of vitamin A deficiency in the study population. Among placebo-treated mothers, only 6.0% and <1% of all concentrations remained below 1.05 and 0.70 µmol/L, respectively (data not shown). This serum retinol distribution is very similar to that of a reference population of American women of reproductive age (20), is higher than that of a nationally representative sample of Zimbabwean women in whom 7% had a concentration <0.70 µmol/L (14), and is substantially higher than that of a group of Zimbabwean women living in an arid rural area of the country in whom 40% had a concentration <0.70 µmol/L (13).
During the first year of life, 170 infants died (148 by 6 mo and 22 between 6 and 12 mo), and the overall 12-mo mortality rate was 20.1/1000 infant years. Nine of these deaths were due to injuries or congenital defects (3 in the Aa group and 2 in each of the other arms) and were excluded from the analyses. There was no interaction between maternal and infant vitamin A supplementation on infant mortality at either 6 or 12 mo (P = 0.89 and P = 0.60, respectively). At 6 mo, there was no effect of either maternal supplementation [hazard ratio (HR): 1.10; 95% CI: 0.79, 1.53] or infant supplementation (1.11; 0.80, 1.55) on mortality. At 12 mo, Cox model point estimates of HRs associated with maternal and infant supplementation remained >1.0, but the wide 95% CIs were inclusive of 1.0 and estimates for each of the 3 individual treatment groups compared with the double placebo arm were somewhat higher, although not significantly so (Table 2). No substantial change was observed in these estimates when adjusted for additional baseline variables or when infants of mothers' whose negative HIV status could not be confirmed at the endpoint were excluded (data not shown). Kaplan-Meier estimates of cumulative mortality produced similar results (Figure 2). Thus, the results indicated no overall effect on mortality.
View this table:
TABLE 2. Mortality at 12 mo among infants according to whether they and their mothers received vitamin A supplementation within 96 h of delivery1
FIGURE 2.. Kaplan-Meier survival curves for infants according to treatment group: Aa, mothers received 400 000 IU vitamin A and infants received 50 000 IU vitamin A; Ap, mothers received 400 000 IU vitamin A and infants received placebo; Pa, mothers received placebo and infants received 50 000 IU vitamin A; Pp, mothers and infants received placebo. The infants included in the analysis were those born to HIV-negative mothers who did not seroconvert during the postpartum year and who made at least one follow-up visit (n = 2254, 2250, 2225, and 2248 for the Aa, Ap, Pa, and Pp groups, respectively). A log-rank test for homogeneity of the survival curves was conducted over the 12-mo period (P = 0.67).
The most common symptoms of disease associated with death were acute respiratory infection, acute diarrhea, sepsis, and malnutrition (Table 3). No cause-specific comparisons between treatment groups and the double placebo arm were statistically significant.
View this table:
TABLE 3. Cause-specific mortality of infants by vitamin A treatment group1
Cox models were constructed to determine whether the effect of either maternal or neonatal vitamin A supplementation on infant mortality was modified by nutritional status, as assessed by birth weight, birth length, or head circumference. The three-factor interaction (maternal vitamin A, infant vitamin A, and nutritional-status indicator) was not significant for length or head circumference but was significant for weight (P = 0.059). In a second model that included the 4 treatment groups, birth weight, and the interaction terms between each treatment group and birth weight, the HRs (and 95% CIs) per kg birth weight for the Aa, Ap, and Pa groups were 1.26 (0.49, 3.21), 1.72 (0.66, 4.44), and 2.62 (0.97, 7.04). Thus, whereas all 3 interaction terms were positive, only the term for the Pa group was significant. Finally, visual inspection of the HRs for mortality associated with each treatment group compared with the double placebo arm plotted by 200-g birth weight strata suggested an increase in risk with increasing birth weight for all 3 treatments (Figure 3). Thus, the data suggest that vitamin A supplementation (either maternal or neonatal) may have benefited small infants but harmed large infants. However, this evidence is weak because no dose-response relation was observed, and the interaction term was not significant for 2 of the 3 treatment groups.
FIGURE 3.. Hazard ratios of death, by birth weight, in infants in 3 treatment groups: Aa, mothers received 400 000 IU vitamin A and infants received 50 000 IU vitamin A; Ap, mothers received 400 000 IU vitamin A and infants received placebo; Pa, mothers received placebo and infants received 50 000 IU vitamin A.
DISCUSSION
In this study, neither maternal nor early neonatal vitamin A supplementation had an overall effect on the mortality of infants born to women who were HIV-negative. Our results are not consistent with those from the other 2 studies in which infants were supplemented at birth, both of which showed substantial and significant reductions in early infant mortality. The discrepancy between our findings and those of the South India study may have been due to differences in the vitamin A status of the mothers. In South India, 6% of the mothers had experienced night blindness during the index pregnancy, whereas in our study no woman reported night blindness, and the distribution of their serum retinol concentrations was similar to that of a reference population of American women of reproductive age. The discrepancy between our findings and those of the Indonesian study are more difficult to explain because the mothers in both study populations had apparently normal serum retinol concentrations.
The apparent modest increase in mortality resulting from either postpartum maternal or early neonatal supplementation may be due to random variation, given the wide CIs. In addition, it may reflect unknown but inherent baseline differences in the mothers, the infants, or both in the double placebo arm participants because it seems unlikely that maternal dosing alone, neonatal dosing alone, or both combined would have adverse effects of such similar magnitude. Several studies have suggested that general nutritional status may modify the effect of vitamin A supplementation on infant mortality and morbidity. In Nepal, in which the vitamin A supplementation of young infants (50 000 IU for neonates and 100 000 IU for children aged 1–5 mo) was associated with a nonsignificant increase in mortality, there was a dose-responsive increase in the relative risk of mortality among supplemented children with increasing arm circumference (9). Three other trials carried out among children aged 6 mo reported reduced morbidity symptoms among malnourished children but increased morbidity among better-nourished children after vitamin A supplementation compared with placebo (21–23). The infants in this study were larger than those in the South India study; the mean birth weight and proportion of infants weighing <2.5 kg were, respectively, 3 kg and 12% in our study compared with 2.7 kg and 31% in the South India study. However, we were unable to show significant interactions between any measure of nutritional status and effect on infant mortality of either maternal or neonatal supplementation. Further research on neonatal vitamin A supplementation should be designed to investigate this potential effect modification more carefully.
The findings of this study suggest that the vitamin A supplementation of neonates and postpartum women is unlikely to reduce infant mortality in relatively well-nourished populations, even when infectious diseases are the commonest cause of death. These findings have little relevance to vitamin A supplementation programs for neonates and postpartum women living in areas where vitamin A deficiency is more severe and no relevance to programs aimed at children aged 6-71 mo, for whom significant reductions in morbidity, mortality, or both were observed after vitamin A supplementation in 25 trials (24).
ACKNOWLEDGMENTS
We thank the Harare City Health Department, Harare Central Hospital, Mashonaland East Provincial Medical Department, and Chitungwiza Hospital for collaborating and providing access to its facilities.
Members of the ZVITAMBO Study Group (in addition to the authors) are as follows: Henry Chidawanyika, John Hargrove, Agnes Mahomva, Florence Majo, Michael Mbizvo, Kuda Mutasa, Faith Muzengeza, Mary Ndhlovu, Robert Ntozini, Ellen Piwoz, Lidia Propper, Philipa Rambanepasi, Andrea Ruff, Naume Tavengwa, and Claire Zunguza.
All authors participated in the design and implementation of the study and contributed to the writing of the manuscript. LCM directed the vitamin A laboratory. EM and LHM conducted the statistical analyses. PJI, KJN, and PZ were responsible for the clinical care delivered during the study. LSZ and BJW provided consultation on the HIV diagnostic activities. JHH drafted the manuscript. KJN and JHH were the coprinciple investigators. None of the authors had a personal or financial conflict of interest.
REFERENCES