Multivitamin Supplements Have No Effect on Growth of Tanzanian Children Born to HIV-Infected Mothers^,,,

Abstract

Growth faltering and micronutrient deficiencies commonly coexist in HIV-exposed children in sub-Saharan Africa, and correcting deficiencies, such as those of vitamins B-complex, C, and E, may improve HIV-related endpoints and child growth. We therefore examined the effect of daily oral supplementation of vitamins B-complex, C, and E on growth among 2341 children born to HIV-infected mothers in Tanzania. HIV-infected women pregnant at ≤32 wk of gestation were enrolled in the study. Children were randomized at age 6 wk to receive multivitamins or placebo until age 104 wk. All women received the same types of vitamins pre- and postnatally. At 6 wk, 256 children (11.1%) were HIV infected and the mean (SD) Z-scores for length for age (LAZ), weight for length (WLZ), and weight for age (WAZ) were −0.39 ± 1.20, −0.21 ± 1.23, and −0.52 ± 1.11, respectively. There was no overall treatment effect on LAZ, WLZ, or WAZ profiles during the follow-up (P ≥ 0.15). There was no treatment effect from 6 to 104 wk on LAZ [(95% CI: −0.14, 0.13); P = 0.94], WLZ [(95% CI: −0.17, 0.13); P = 0.78], or WAZ [(95% CI: −0.15, 0.16); P = 0.97] or on the incidence of growth failure, defined as respective Z-scores < −2 (P ≥ 0.29). Among the subgroup of HIV-uninfected children, there was no treatment effect from 6 to 104 wk on LAZ, WLZ, and WAZ (P ≥ 0.71) or on the incidence of growth failure (P ≥ 0.16). Multivitamin supplements had no effect on growth among children born to HIV-infected women who were themselves receiving multivitamins.

Introduction

Poor child growth is an important public health problem among children living in low- and middle-income countries. Stunting, a failure to reach linear growth potential due to chronic nutritional deficits, increases the risks of child mortality and morbidity (1) and predicts lower adult height, education achievement, and economic productivity in later life (2). Wasting, a deficit of weight for length as seen during severe disease or starvation, is highly predictive of child mortality and morbidity, especially in its severe form (1). The combined effects of poor child growth account for >20% of the global mortality and morbidity burdens in children under age 5 y (1).

In poor countries, the prevalence of linear growth failure rises quickly in the first 24 mo of life (3). A major determinant is suboptimal feeding practices: only 37% of children in the developing world are exclusively breastfed for the recommended first 6 mo of life and merely 55% continue breastfeeding until age 24 mo (4). Furthermore, complementary and weaning foods that are microbiologically safe and that meet the child’s needs for micro- and macronutrients often remain inaccessible to large parts of the population (5).

To address the high prevalence of micronutrient deficiencies among young children and examine the role of micronutrients in child growth, the effect of providing multiple micronutrients has been tested (6–8). Summary estimates obtained from presumably HIV-unexposed children indicate that multiple micronutrient interventions lead to small but significant increases in length (6, 8) and weight gain (6). However, these results may not be generalizable to HIV-exposed children. Compared with those without HIV exposure, HIV-exposed but uninfected children may have poorer immune function (9) and higher morbidity (10), whereas those infected with HIV are at a higher risk of nutrient malabsorption, impaired storage, and altered metabolism (11) and thus micronutrient deficiencies (12) and growth faltering (13–16).

There is little evidence on the effect of micronutrient supplementation among HIV-exposed or -infected children. Among children born to HIV-infected women in Tanzania, maternal supplements containing high doses of vitamins B-complex, C, and E and given during pregnancy and lactation improved birth weight, child ponderal growth, and immune status (17, 18). The mechanisms of action may include improved RBC formation and oxygen delivery to tissues, increased nucleic acid synthesis, and less infection-related growth failure (19–23).

Given the high risk of morbidity and nutritional deficiencies among HIV-exposed children and evidence indicating that maternal vitamins B-complex, C, and E may improve child growth, we tested the effect of directly providing these vitamins from 6 wk onwards to children born to HIV-infected women.

Materials and Methods

Study design, study population, and setting.

Details of the trial were described elsewhere (24). In brief, pregnant women who provided written informed consent were invited to enroll in the study and were followed up throughout the antenatal, delivery, and postnatal periods to assess whether the mother-infant pair met the eligibility criteria for the child to be randomized at age 5–7 wk. The study regimen contained 60 mg vitamin C, 8 mg vitamin E, 0.5 mg thiamine, 0.6 mg riboflavin, 4 mg niacin, 0.6 mg vitamin B-6, 130 μg folate, and 1 mg vitamin B-12. Children <6 mo received 1 capsule and older children 2 capsules/d. The dosages administered represent 150–600% and 200–400% of the US Adequate Intake for children aged 0–6 mo and 7–12 mo, respectively, and 133–800% of the US RDA for children aged 1–3 y. For these nutrients, tolerable upper intake levels have not been defined for children <12 mo; the tolerable upper intake levels (if defined) were not exceeded for children ≥12 mo (25, 26). The regimen was provided to caretakers in addition to small, disposable plastic dosing cups and containers holding 5 mL of sterile water.

At monthly visits, the caretakers received the number of capsules, cups, and water containers required for the upcoming month in addition to a surplus to allow for continued administration of the regimen if the caretakers returned late to the study clinic. Those traveling out of Dar es Salaam were provided with sufficient regimen and supplies until the next scheduled research visit. Caretakers were instructed on how to push the capsule through the back of the blister pack, open it, decant the powder into the cup, add the sterile water to the powder, and observe the child consuming the full content of the cup.

The study statistician (R.J.B., based in Boston, MA) prepared a computer-generated randomization list using permuted blocks of size 20. Onsite study pharmacists stored the coded randomization list in a locked file cabinet and concealed the allocation by covering the numeric regimen code on each blister pack with a sticker. Infants enrolled at the study clinic were provided the next consecutive number in the series. Study physicians, research nurses, and participants were unaware of the assignment groups. The supplement used was a powder encapsulated in an opaque gelatinous capsule manufactured by Nutriset, which was not involved in the study design, implementation, analysis, or reporting of findings. The study protocol was approved by the Institutional Review Boards at Muhimbili University of Health and Allied Sciences and the Harvard School of Public Health and study data were reviewed by a Data Safety Monitoring Board twice annually.

Data collection.

At the enrollment visit during pregnancy, trained research nurses collected data on women’s socioeconomic and health status, educational level, obstetric history, date of last menstrual period, and anthropometry. One month after enrollment, research nurses obtained a blood sample to measure maternal immunologic and hematologic parameters. Immediately after delivery, trained research midwives measured birth weight to the nearest 50 g with a standard mechanical scale and infant length to the nearest 0.1 cm with a length board. Gestational age at birth was calculated from the date of the last menstrual period, obtained at the time of study enrollment.

After randomization, caretakers were asked to return with the child to the study clinic for monthly research visits. During these visits, nurses interviewed caretakers about the child’s morbidity history, determined compliance with the study regimen by counting the unused supplies, and measured child anthropometry using standard techniques (27). They measured the child’s weight on a digital infant balance with 10-g precision (Tanita) and length with 1-mm precision using a rigid length board with a movable foot piece. The study physicians examined the child every 3 mo.

HIV infection status was determined at 6 wk and 18 mo. The launch of the President’s Emergency Plan for AIDS Relief program in July 2005 allowed for prospective HIV testing as well as increased access to antiretroviral treatment (ART)¹³. Children who were HIV uninfected at 18 mo and still breastfed were tested again before they were discharged from the study after 24 mo of follow-up.

Study outcomes and sample size.

The trial aimed to determine whether the daily oral administration of multivitamins to Tanzanian children born to HIV-infected women reduces the risks of child mortality and diarrheal morbidity as the primary aims and growth faltering as a secondary aim.

We calculated age- and sex-specific Z-scores for the anthropometric indexes length for age (LAZ), weight for length (WLZ), and weight for age (WAZ) using the 2006 WHO growth standards and related software (28). The study outcome was the difference in attained LAZ, WLZ, and WAZ between treatment and control groups. Our secondary study outcome measures were incidence of stunting (LAZ < −2), wasting (WLZ < −2), and underweight (WAZ < −2) among those free of these conditions at baseline.

Standard of care.

Children and their caretakers had access to standard pre- and postnatal care, including nutritional counseling and support. Prenatal care included daily doses of ferrous sulfate (200 mg, equivalent to 60 mg ferrous Fe) and folic acid (0.25 mg) as well as malaria prophylaxis with sulfadoxine-pyrimethamine tablets (Fansidar; Roche Pharmaceuticals) at 20 and 30 wk of gestation. On the basis of earlier findings of the benefits of prenatal multivitamins among HIV-infected women who were not receiving ART (29, 30), women received supplements containing high doses of vitamins B-complex, C, and E during and after pregnancy. A subset of women (71 in the treatment and 67 in the control group; P = 0.72) had been enrolled in an efficacy trial comparing these vitamin dosages with those of the RDA (31). Women who started on ART were changed to multivitamin supplements containing RDA levels of these nutrients. As part of standard medical care, all children received immunizations, routine medical care for illnesses, and periodic large doses of vitamin A at 9, 15, and 21 mo of age. Children received cotrimoxazole prophylaxis until 6 mo of age; after that, only breastfeeding or HIV-infected children continued to receive cotrimoxazole. Mothers were counseled on the risks and benefits of breastfeeding. The provision of the study regimen was in line with the WHO recommendation on exclusive breastfeeding, which allows for the administration of oral rehydration salts, drops, or syrups in addition to breast milk (32).

When the study was designed, routine medical care for pregnant women with HIV infection included malaria prophylaxis, diagnosis and treatment for sexually transmitted diseases and prophylaxis, and diagnosis and treatment of opportunistic infections. Antiretroviral medication to curb maternal-to-child transmission of HIV was limited to nevirapine prophylaxis (one dose given to the mother at the onset of labor and one dose given to the infant within 72 h of birth) (33). As the study progressed, the availability of ART substantially increased through the President’s Emergency Plan for AIDS Relief and other governmental and nongovernmental programs. Beginning in July 2005, women and children in the study were screened for ART eligibility and treated according to Tanzanian Ministry of Health guidelines (34). For adults, eligibility was based on: WHO stage IV HIV disease, or CD4 cell count <200 cells/mm³, or WHO stage III and CD4 <350 cells/mm³. Eligibility for children was defined as follows: for children <18 mo, CD4% <20 or Pediatric WHO Stage III, or any child ≥18 mo of age with Pediatric WHO Stage III or CD4% <15%. For adults, the standard first-line regimen was stavudine, lamivudine, and nevirapine; for children, it was zidovudine, lamivudine, and nevirapine. Alternative drug regimens were available for special circumstances.

Laboratory analyses.

Maternal HIV-1 serostatus was determined by 2 sequential ELISAs using Murex HIV antigen/antibody (Abbott Murex) followed by the Enzygnost anti-HIV-1/2 Plus (Dade Behring); discordant results were resolved by a Western-blot test (Bio-Rad Laboratories). Children <18 mo were tested for HIV infection using the Amplicor HIV-1 DNA assay version 1.5 (Roche Molecular Systems) and older children using HIV ELISAs. Hemoglobin concentrations were measured using an AcT5 Diff AL hematology analyzer (Beckman Coulter) and T-cell subsets using the FACSCalibur system (Becton-Dickinson).

Statistical analyses.

We used intent-to-treat analyses to estimate treatment effects. We fitted mean LAZ, WLZ, and WAZ curves using mixed effects models with restricted cubic splines and placed knots at 3- mo intervals after randomization (35). An automatic knot selection procedure was employed using P < 0.05 to determine a parsimonious model. To improve model fit for WLZ, an additional knot at age 6 wk was included. The variables included in each outcome model were the treatment group, linear and spline terms for time (in wk), and interaction terms between the treatment group and time. We determined the overall significance of the treatment during the 98-wk follow-up period by testing a model with main effects for linear and nonlinear terms for time against one with additional interaction terms between treatment group and time. We estimated the treatment effect as the difference in attained LAZ, WLZ, and WAZ scores between the multivitamin and placebo groups every 3 mo and from 6 to 104 wk. Because antiretroviral therapy in line with the latest recommendations can reduce mother-to-child transmission to <5% in breastfeeding populations (36), we analyzed the multivitamin effect over time among the subgroup of children who were HIV uninfected at baseline. Results were similar among children remaining HIV uninfected during follow-up and are therefore not shown.

We used Cox proportional hazard models (37) with the exact method for ties (38) to investigate the treatment effects on incidence of stunting, wasting, and underweight. We set extreme values for LAZ (< −6 or >6), WLZ (< −5 or >5), and WAZ (< −6 or >5) to missing as recommended (28).

We compared compliance with the study regimen across treatment groups with the Wilcoxon rank-sum test. All P values reported are 2-sided and significance was defined as P < 0.05. No adjustments for multiple comparisons were performed. Statistical analyses were carried out by using SAS system version 9.2 (SAS Institute).

Results

From a pool of 103,290 pregnant women screened for trial eligibility, 6063 were HIV infected and 3500 consented to participate in the current trial (Fig. 1). Mothers were enrolled from February 2004 to June 2007 and infants randomized from August 2004 to November 2007. In total, 1193 infants were randomized to receive multivitamins and 1194 to receive placebo. Of those, 1170 in the multivitamin group had valid baseline and follow-up height measurements and 1171 in the placebo group. The mean follow-up time was 76 ± 34 wk and did not differ between treatment groups (P = 0.87). The administrative end of the follow-up was 31 May, 2008.

FIGURE 1.

Study profile of randomized trial of multivitamin supplementation to infants born to HIV-infected mothers in Tanzania. Study exclusions shown in italics.

Child baseline characteristics, including mean birth weight, HIV status, hemoglobin concentrations, anthropometric indices, and maternal characteristics did not differ between the multivitamin and placebo groups (Table 1). Among children known to be HIV uninfected at baseline, 73 (3.1%) became HIV infected during follow-up. The median reported compliance with the study regimen was 96% (IQR = 91–99%) and did not differ between treatment groups (P = 0.64). At 6 and 9 mo, 660 (28.2%) and 114 (4.9%) children, respectively, were still breastfed and there were no differences between study groups (P ≥ 0.44).

TABLE 1.

Baseline characteristics of enrolled children and their HIV-infected mothers by supplementation group in Tanzania¹

Characteristic	Multivitamin group (n = 1170)	Placebo group (n = 1171)
Children’s characteristics at 6 wk2
Male sex, %	54.4	53.5
Birthweight <2500 g, %	6.7	7.3
Born <37 wk gestational age, %	14.6	15.6
HIV-infected at 6 wk, %	10.6	11.6
Breastfeeding at 6 wk, %	91.0	89.7
CD4 count, cells/mm3	1874 ± 798	1869 ± 825
Hemoglobin, g/L	10.4 ± 1.5	10.3 ± 1.6
LAZ	−0.35 ± 1.17	−0.43 ± 1.23
WLZ	−0.21 ± 1.24	−0.22 ± 1.21
WAZ	−0.48 ± 1.10	−0.55 ± 1.12
Mothers’ characteristics at enrollment3
Gestational age, wk	24.6 ± 5.5	24.6 ± 5.6
Age, y	28.2 ± 5.0	28.2 ± 4.9
Years of education, n	7.3 ± 2.7	7.1 ± 2.8
Prior pregnancies, n	2.5 ± 1.3	2.6 ± 1.3
Household food expenditure,4 Tsh/person	494 ± 416	502 ± 431
Mid-upper arm circumference, cm	25.9 ± 3.3	26.0 ± 3.2

¹Values are means ± SDs or percentages. LAZ, length-for-age Z-score; WAZ, weight-for-age Z-score; WLZ, weight-for-length Z-score.

²Missing values for child characteristics ranged from 0 to 7%.

³Missing values for maternal characteristics ranged from 0.1 to 6%.

⁴Tsh = Tanzanian shillings. At the start of the study, 1250 Tsh = 1 US$.

During the follow-up period, LAZ changed by −0.78 (95% CI: −0.85, −0.72), WLZ by −0.45 (95% CI: −0.52, −0.37), and WAZ by −0.66 (95% CI: −0.74, −0.59) (Fig. 2). LAZ decreased rapidly from wk 38 to 86 and WLZ from wk 18 to 50, while WAZ decreased steadily during follow-up. There was no overall treatment effect on profiles of LAZ (P = 0.15), WLZ (P = 0.87), or WAZ (P = 0.86) during follow-up (Fig. 2) and no effect at any time point tested, including no treatment effect on Z-score change from 6 to 104 wk (Table 2). There was no treatment effect on the incidence of stunting [HR = 0.92 (95% CI: 0.79, 1.08); P = 0.30], wasting [HR = 0.92 (95% CI: 0.79, 1.07); P = 0.29], and underweight [HR = 0.95 (95% CI: 0.81, 1.10); P = 0.48] (Table 3).

FIGURE 2.

LAZ (A), WLZ (B), and WAZ (C) by supplementation group among 2341 children born to HIV-infected mothers in Tanzania. Based on mixed effects models with restricted cubic splines. Each model contained a term for the treatment group, linear and spline terms for time (in wk), and interaction terms between the treatment group and time. Curves represent mean Z-scores during follow-up and bars represent 95% CI. LAZ, length-for-age Z-score; WAZ, weight-for-age Z-score; WLZ, weight-for-length Z-score.

TABLE 2.

Changes in Z-scores from 6 wk to 104 wk by supplementation group among 2341 children born to HIV-infected mothers in Tanzania¹

Index	Multivitamin group2	Placebo group2	Difference (95% CI)3	P-difference
LAZ	−0.78 (−0.69, −0.87)	−0.78 (−0.68, −0.87)	−0.005 (−0.14, 0.13)	0.94
WLZ	−0.46 (−0.56, −0.35)	−0.43 (−0.54, −0.33)	−0.02 (−0.17, 0.13)	0.78
WAZ	−0.66 (−0.77, −0.55)	−0.67 (−0.78, −0.56)	−0.003 (−0.15, 0.16)	0.97

¹From mixed-effects models with restricted cubic splines. LAZ, length-for-age Z-score; WAZ, weight-for-age Z-score; WLZ, weight-for-length Z-score.

²Values are mean Z-score (95% CI) changes from 6 to 104 wk.

³Values are differences (95% CI) in mean Z-score changes from 6 to 104 wk between the multivitamin and placebo groups.

TABLE 3.

Incidence of stunting, wasting, and underweight from 6 to 104 wk by supplementation group among children born to HIV-infected mothers in Tanzania¹

Endpoint	Multivitamin group	Placebo group
Stunting (LAZ < −2)
Events/child-years, n/n	312/1282	333/1257
HR (95% CI)	0.92 (0.79, 1.08)	1.00 (Ref)
Wasting (WLZ < −2)
Events/child-years, n/n	331/1256	354/1233
HR (95% CI)	0.92 (0.79, 1.07)	1.00 (Ref)
Underweight (WAZ < −2)
Events/child-years, n/n	316/1308	329/1284
HR (95% CI)	0.95 (0.81, 1.10)	1.00 (Ref)

¹From Cox proportional hazards models for the time to Z-score < −2. In each model, only children with baseline Z-score ≥ −2 were eligible to enter the analysis. LAZ, length-for-age Z-score; WAZ, weight-for-age Z-score; WLZ, weight-for-length Z-score.

Among children who were HIV uninfected at baseline, the profile of LAZ during follow-up differed slightly between treatment groups (P = 0.04) but not for WLZ (P = 0.79) and WAZ (P = 0.73) (Supplemental Fig. 1). Nevertheless, for all 3 endpoints, there were no significant differences between treatment groups at any time point during the follow-up period, including no treatment effect in change from 6 to 104 wk (Supplemental Table 1). There was no treatment effect on the incidence of stunting [HR = 0.96 (95% CI: 0.81, 1.14); P = 0.63], wasting [HR = 0.89 (95% CI: 0.75, 1.05); P = 0.16], or underweight [HR = 0.95 (95% CI: 0.80, 1.13); P = 0.54] among HIV-uninfected children at baseline (Supplemental Table 2).

Discussion

In this randomized, placebo-controlled trial among children born to HIV-infected mothers, daily micronutrient supplements containing vitamins B-complex, C, and E given to young children for 24 mo did not improve mean anthropometric Z-scores or reduce the risks of growth faltering. The large sample size, the regular and high rate of follow-up during 24 mo, and the randomized, controlled design of the study allow for the drawing of causal inferences about the observed effects.

Numerous trials have examined the effect of micronutrient interventions on child growth. A meta-analysis of 20 studies conducted among children <5 y indicated that supplements providing at least 3 micronutrients led to small but significant increases in height but not weight gain (8). In a review of 36 studies among children up to school age but restricted to studies with at least 4 micronutrients, supplements led to slightly greater beneficial effects on length gain compared with the first meta-analysis and also increased weight gain (6). The trials reviewed in both analyses were conducted among presumably HIV-unexposed children and generally provided 1–2 RDA of at least 6 micronutrients, most commonly including iron, zinc, vitamins B-complex, A, and C (7).

The current trial was conducted among HIV-exposed children, because they have higher risks of micronutrient deficiencies (11), mortality (39), and growth failure (13–16), especially if subsequently infected, compared with their unexposed peers. A trial regimen composed of vitamins B-complex, C, and E was chosen, because these vitamins regulate immune function and other important metabolic pathways. Vitamin C is required for phagocytosis as well as T-cell function (40) and controls inflammatory responses (41), whereas vitamin E is needed for T-cell differentiation (19). B vitamins function in RBC formation and oxygen delivery to tissues as well as nucleic acid synthesis and other metabolic pathways relevant for growth (23). Other support to these biologic mechanisms is that although single-nutrient trials with vitamin A, iron, and possibly zinc may have limited or no effect on child growth, trials that combined these with other nutrients (generally including vitamin B-complex and vitamin C) led to significant improvements in growth (8). Data from our group suggest that among adult women from Tanzania, vitamins B-complex, C, and E given pre- and postnatally slowed HIV disease progression, reduced mortality risk (30), and lowered the risk pregnancies resulting in low infant birth weight (29).

The same intervention also improved postnatal ponderal growth (18) among children born to these women, possibly by improving immunological status and reducing diarrheal morbidity (17). It is therefore unclear why the direct administration of the same vitamins in the current trial had no effect on child growth. One potential explanation is that because women in the current study received these same vitamins as clinical standard of care, maternal supplements may have offset potential benefits of the trial regimen by improving infant dietary intake through breast milk. This hypothesis is supported by previous data on the beneficial effect of maternal supplementation on infant vitamin B-12 and E concentrations in the first 6 mo of life (42). However, although breastfeeding was common at 6 wk, the majority of children were weaned by 6 mo and the effect of maternal supplementation would therefore have been limited to the first months of life. Alternatively, the lack of treatment effect and the high prevalence of growth faltering observed in the current study may be partially explained by common deficiencies in other nutrients required for growth (43). This may include zinc, which we did not include in the experimental regimen because of safety concerns at the time of trial design (44, 45), essential fatty acids (46), or the larger group of so-called type II nutrients (43).

In other micronutrient trials among HIV-affected populations, supplements containing the RDA of 14 micronutrients had the same effect on growth as supplements containing the RDA of 6 vitamins among HIV-infected children from Uganda (47). In a trial of maternally HIV-unexposed and -exposed Zambian infants, a locally made infant food that was fortified to meet infant micronutrient did not improve linear growth compared with a food containing lower micronutrient amounts (48). However, there was evidence that the high-micronutrient food improved linear growth in infants of HIV-infected mothers who stopped breastfeeding before 6 mo.

Our study has several limitations. The findings may not be generalizable to the increasing number of mother-infant pairs receiving ART, because only a subset of children (n = 143) received ART, or to children whose mothers did not receive multivitamin supplements pre- and postnatally. As with most long-term studies among high-risk populations, the study experienced loss to follow-up during the study period due to death or inability or refusal to continue participation. This study loss may have introduced bias and reduced power to detect differences between groups.

In conclusion, in this large, randomized, placebo-controlled trial among children born to HIV-infected women receiving multivitamins, we found no evidence that the provision of vitamin B-complex and vitamins C and E to HIV-exposed children, including infants who remained HIV uninfected, reduces growth faltering. More research is needed to develop effective nutritional and non-nutritional interventions to prevent growth faltering among young children in resource-poor settings.