Landing MA Jarjou, Ann Prentice, Yankuba Sawo, M Ann Laskey, Janet Bennett, Gail R Goldberg, and Tim J Cole

Transcription

1 Randomized, placebo-controlled, calcium supplementation study in pregnant Gambian women: effects on breast-milk calcium concentrations and infant birth weight, growth, and bone mineral accretion in the first year of life 1 3 Landing MA Jarjou, Ann Prentice, Yankuba Sawo, M Ann Laskey, Janet Bennett, Gail R Goldberg, and Tim J Cole ABSTRACT Background: Growth and bone mineral accretion in Gambian infants are poorer than those in Western populations. The calcium intake of Gambian women is low, typically mg Ca/d, and they have low breast-milk calcium concentrations, which result in low calcium intakes for their breastfed infants. A low maternal calcium supply in pregnancy may limit fetal mineral accretion and breast-milk calcium concentrations and thereby affect infant growth and bone mineral accretion. Objective: We investigated the effects of calcium supplementation in Gambian women during pregnancy on breast-milk calcium concentrations and infant birth weight, growth, and bone mineral accretion. Design: A randomized, double-blind, placebo-controlled supplementation study was conducted in 125 Gambian women who received 1500 mg Ca/d (as calcium carbonate) or placebo from 20 wk of gestation until delivery. Infant birth weight and gestational age were recorded. Breast milk was collected, and infant anthropometric and bone measurements were performed at 2, 13, and 52 wk after delivery. Infant bone mineral status was assessed by using singlephoton absorptiometry of the radius and whole-body dual-energy X-ray absorptiometry. Results: Compliance with the supplement was high. No significant differences were detected between the groups in breast-milk calcium concentration, infant birth weight, or growth or bone mineral status during the first year of life. A slower rate of increase in infant whole-body bone mineral content and bone area was found in the supplement group than in the placebo group (group time interaction: P 0.03 and 0.02, respectively). Conclusion: Calcium supplementation of pregnant Gambian women had no significant benefit for breast-milk calcium concentrations or infant birth weight, growth, or bone mineral status in the first year of life. Am J Clin Nutr 2006;83: KEY WORDS Bone mineral accretion, breast milk, calcium, Gambia, infants, pregnancy INTRODUCTION An intake of 200 mg Ca/d is required during pregnancy and in the postpartum period for fetal skeletal mineralization, secretion into breast milk, and growth during infancy (1). The calcium for skeletal mineralization is supplied by the mother across the placenta during fetal life and through breast milk during infancy. At birth, an infant s body contains g Ca, almost all of which is in the skeleton (1, 2). Most of this calcium is deposited during the second half of pregnancy; the rate of fetal bone mineral accretion (BMA) increases from 50 mg/d at 20 wk to 330 mg/d at 35 wk of gestation (2). After birth, BMA averages 140 mg/d during the first year of life; the rate is highest in the first months and slows progressively with age (1, 3). We showed in our previous studies that the calcium intake of rural Gambian women is low during pregnancy and lactation typically, mg/d (4 6). This low intake may constrain calcium supply to the fetus and affect fetal BMA. Furthermore, breast-milk calcium concentrations are significantly lower in Gambian women than in British women (5 7). Although our group showed through a randomized, controlled, calcium supplementation study that increasing the calcium intake of lactating Gambian women does not affect breast-milk calcium concentration (8), we found observational evidence to suggest that calcium intake in pregnancy may influence breast-milk calcium concentrations in Gambian women during the subsequent lactation (6, 9). Infants in rural areas of The Gambia have lower birth weights and less growth and BMA than do infants in Western populations (5, 10). Gambian infants are breastfed on demand for 2y,and breast milk is their main source of calcium during this time (5, 11). As a result, their calcium intake averages 200 mg/d throughout the first 12 mo of life, and breast milk provides 90% and 50% of total calcium intake at 3 and 12 mo, respectively (11, 12). This amount is close to the theoretical biological requirement for BMA in infancy (1, 3) and suggests that a constrained calcium supply through breast milk may limit skeletal growth in Gambian infants. 1 From the Medical Research Council Keneba, Keneba, The Gambia (LMAJ, AP, and YS); MRC Human Nutrition Research, Cambridge, United Kingdom (AP, MAL, JB, and GRG); and the Centre for Paediatric Epidemiology and Biostatistics, Institute of Child Health, London, United Kingdom (TJC). 2 Shire Pharmaceuticals and Nycomed Pharma donated the supplement and placebo tablets. 3 Address reprint requests to A Prentice, MRC Human Nutrition Research, Elsie Widdowson Laboratory, Fulbourn Road, Cambridge CB1 9NL, United Kingdom. ann.prentice@mrc-hnr.cam.ac.uk. Received September 21, Accepted for publication December 12, Am J Clin Nutr 2006;83: Printed in USA American Society for Nutrition 657

2 658 JARJOU ET AL The aim of the current study was to test, by means of a randomized, placebo-controlled, calcium supplementation study, whether an increase in calcium intake by Gambian women in the second half of pregnancy promoted fetal growth and BMA, as judged by weight at birth, by bone mineral status and body length 2 wk after delivery, by breast-milk calcium concentration in the subsequent lactation, and by infant growth and bone mineral status in the first year of life. SUBJECTS AND METHODS Sample size and statistical power The research described in this report was developed as a subsection of a larger study designed to investigate the effects of calcium supplementation on blood pressure in pregnancy; those findings will be reported separately. To facilitate attendance at the Medical Research Council (MRC) Keneba Clinic for bone scanning, the subset of subjects was restricted to residents of Keneba and Manduar, The Gambia. Because 50 women give birth in the 2 villages each year, the hypotheses that could be tested with a sample size of 50 per group were considered. To examine the power of the study, typical between-subject CVs in this population for breast-milk calcium concentration (13%) and for forearm bone mineral content (BMC) adjusted for bone width (BW) (9%), taken from previous Gambian studies that used the same technologies (5, 7, 8, 10, 13), were used. At 5% significance and 80% power, a sample size of 50 per group would give a minimum detectable between-group difference in breast-milk calcium concentration of 7%, which is equivalent to 15 mg/l when the mean concentration is 210 mg/l, and in radial shaft BMC of 5% after adjustment for BW, which is equivalent to g/cm when the mean BMC is g/cm. Such differences were similar to or smaller than those that have been observed between populations or between persons (6, 10, 13), and therefore a study using 50 subjects per group was considered likely to generate results of biological interest. A minimum target sample size of 50 per group was therefore set for this study. Subject recruitment The study was conducted in the rural villages of Keneba and Manduar, in the province of West Kiang, The Gambia. Recruitment began in May 1995 and ended in June Potential subjects were pregnant women with no history of any medical condition known to affect calcium or bone metabolism, who presented at the antenatal clinic at MRC Keneba before 20 wk of gestation. The midwife used her estimation of fundal height to ascertain the week of gestation. All women who attended the clinic during the period of the study (n 250) were considered for recruitment. Women were approached for participation if they were identified by the midwife as having an uncomplicated singleton pregnancy and if they lived locally and were unlikely to be away from the area for prolonged periods. Twelve women were excluded because they did not meet these criteria; 83 others either declined to participate or did not attend the clinic early enough in the pregnancy for baseline measurements to be made. The remaining 155 women (62%) agreed to take part and were randomly assigned to 1 of 2 intervention groups. For the analysis of infant outcomes, 125 woman-infant pairs were included in the final dataset. This dataset included all women who had been correctly FIGURE 1. Flow chart of recruitment, exclusions, and losses for the randomized, controlled, calcium intervention study in pregnant Gambian mothers. *Subjects who delivered term babies before the predicted date of 36 wk of gestation, as assessed by fundal height at recruitment, and who therefore did not meet the inclusion criterion of wk gestation at the start of the intervention. diagnosed as being at wk of gestation at the start of the intervention period, who gave birth to a healthy infant, and whose infant had been measured anthropometrically 2 wk after delivery. Exclusion and discontinuations after randomization occurred for a variety of reasons, including maternal, fetal, or infant death and misclassification of the length of gestation at recruitment. It was a condition of the study that any mother who developed complications during the second half of pregnancy, such as pregnancy-induced hypertension, would be excluded from further participation. No subject was lost to the study for that reason. The flow of participants through the study and the reasons for exclusions and discontinuations are described, in accordance with the guidelines of the Consolidated Standards of Reporting Trials (14), in Figure 1. The 125 women who participated and were included in the final dataset did not differ significantly in age or parity from the 125 who were eligible but either did not participate or were lost to follow-up of infant outcomes (x SD age: y and y, respectively; P 0.06; parity: and , respectively; P 0.1). Subjects attended the clinic at MRC Keneba for study measurements at wk (P20, baseline) and wk (P36) gestation. The interval between P20 and P36 was d. After birth and during the first week after delivery, fieldworkers and medical staff visited the mother at home to

3 INFANT OUTCOMES AFTER CALCIUM SUPPLEMENTATION IN PREGNANCY 659 record infant weight, length, head circumference, and gestational age. Thereafter, during lactation, the mother and infant attended the clinic for measurements at 2 (12 4 d), 13 (92 4 d), and 52 (368 9 d) wk after delivery (L02, L13, and L52, respectively). Written informed consent was obtained from all subjects after an oral explanation in the local language. The study was approved by the MRC/Gambian Government Ethics Committee. Calcium supplementation Subjects were randomly assigned in a double-blind fashion to receive a calcium supplement or placebo from P20 until delivery. Assignment was by random permuted blocks of 4 to ensure that equal numbers of subjects were allocated to the supplement and placebo groups in each month and thereby to minimize the potential for seasonal confounding. Randomization was achieved by using published sets of tables. The code was held by a member of the study team (AP) who was not directly involved with the collection of data in the field or the laboratory and who had no contact with the study participants. The calcium supplement provided 1500 mg elemental Ca/d. This intake was selected because studies in South American women with a low calcium intake have suggested an effect of calcium supplementation at mg/d on blood pressure and the risk of pregnancy-induced hypertension (15, 16), which is the main outcome under test in the larger study of which the current study is a subsection. The supplement consisted of 3 chewable calcium carbonate tablets (Calcichew; Nycomed Pharma AS, Asker, Norway; distributed in the United Kingdom by Shire Pharmaceutical Development Ltd, Andover, United Kingdom), each containing 500 mg elemental Ca. The placebo consisted of 3 tablets of similar shape, taste, and texture in which the calcium carbonate was replaced with microcrystalline cellulose and lactose (Nycomed Pharma AS). Each participant began to receive the supplement after all P20 measurements were completed and continued with supplementation until delivery, when supplementation was stopped. The tablets were taken to the participants by fieldworkers each day of the week and were consumed in a fieldworker s presence. Intake of the supplement was noted each day by the fieldworker as a marker of compliance. The supplement was consumed between 1700 and 1900; this period between lunch ( ) and dinner (after 2000) was chosen to minimize possible interference with the absorption of other minerals, such as iron (17). During the Ramadan month of daytime fasting, tablets were consumed later in the evening, immediately after the subjects had broken their fast but before the main meal was eaten. Tablets that were not consumed because of illness or absence from the village were counted as missed doses. The mean number of days that subjects received supplementation was d. The tablets were well accepted, and there were no reported adverse effects. Tablet compliance, in terms of the number consumed relative to the number provided between P20 and delivery, was high (range: %); 97% of participants consumed 95% of tablets offered. There were no significant differences between the 2 groups in supplementation period (supplement group: d; placebo group: d) or compliance (range: 86% 100% for the supplement group and 92% 100% for the placebo group; participants consuming 95% tablets: 97% for both groups). Anthropometric measurements Each woman was weighed to the nearest 0.1 kg while wearing light clothing but no shoes (Wylux scales; CMS Weighing Equipment Ltd, London, United Kingdom). Standing height without shoes was recorded to the nearest 0.1 cm (Magnimetre stadiometer; CMS Weighing Equipment Ltd). Height was measured on 8 separate occasions during the study, and the mean of the measurements was taken. Birth weight of the infant was measured within 24 h by medical staff either in Keneba or at the Royal Victoria Teaching Hospital (Banjul, The Gambia). It was not possible to record birth weight for 3 infants: 2 in the supplement group and 1 in the placebo group. In addition, in line with standard practice at the MRC Keneba Clinic, weight, crown-heel length, and head circumference were recorded within5d( 120 h) of birth by medical personnel, and gestational age was assessed by using the score of Dubowitz et al (18). Twenty-one mothers spent the traditional 8-d confinement period away from the study villages, and the set of measurements at 5 d could not be made (n 12 and 9 for the supplement and placebo groups, respectively). Anthropometric measurements of the study infants were performed at 2, 13, and 52 wk of age. Infants were weighed to the nearest 0.01 kg while naked (Seca baby-weighing scale; CMS Weighing Equipment Ltd). Supine crown-heel length was measured by using a length board (Kiddimate; Raven Equipment Ltd, Dunmow, United Kingdom) that was checked against a known reference measure before each measurement. Head circumference was measured by using a nonstretchy measuring tape. SD scores for weight and height were calculated relative to British reference data (19). Breast-milk calcium and phosphorus At L02, L13, and L52, breast milk was collected by maternal manual expression of 1 2-mL samples directly into low-calcium tubes (Z5 tubes; Bibby Sterilin, Stone, United Kingdom). Samples were frozen immediately at 20 C and later transported on dry ice to MRC Human Nutrition Research for analysis. A validated semiautomated micromethod (20) was used to measure calcium and phosphorus concentrations in whole-milk samples after lyophilization, ashing, and reconstitution in 0.3 mol HCl/L (specific gravity: 1.18; VWR, Lutterworth, United Kingdom). Calcium was assayed by the methyl thymol blue method and phosphorus by the ammonium molybdate method (both kits supplied by Roche Ltd, Lewes, United Kingdom). Quality assurance was performed by including the following reference materials in all runs: Randox Assayed Urine (Randox Laboratories Ltd, Crumlin, Northern Ireland); Lyphochek Quantitative Urine Control (Bio-Rad Laboratories Ltd, Hemel Hempstead, United Kingdom), and NIST 1549 nonfat milk powder (National Institute of Standards and Technology, Gaithersburg, MD). Standards and reference materials were prepared with 0.3 mol HCl/L. Maternal urine samples Twenty-four hour urine samples collected from each participant at P20 and P36 were measured for calcium and phosphorus and titratable acidity to provide a measure of compliance with the calcium carbonate supplementation. Urine collections were not obtained for 1 subject (in the placebo group) at P20 or for 8 subjects (n 4 in each group) at P36. All urine collection containers and apparatus were acid-washed to minimize calcium

4 660 JARJOU ET AL contamination. Subjects were supplied with urine bottles, a funnel, and a cooler containing frozen cold packs into which to place filled bottles to keep the urine cool. A fieldworker visited the subject at the start and end of each collection and at regular periods during the 24-h period to refresh the frozen cold packs and to return bottles to the laboratory refrigerator. At the end of the 24-h collection, the urine fractions were pooled and mixed, the total volume was recorded, and aliquots were taken into 30-mL low-calcium tubes (Universal tubes; Bibby Sterilin). One aliquot was analyzed immediately for titratable acidity by using direct titration to ph 7.4 with mol NaOH/L. A second urine aliquot was acidified with HCl to obtain a final acid concentration of 0.3 mol/l (Spectrosol; BDH, Poole, United Kingdom) and stored at 20 C. The acidified samples were transported on dry ice to MRC Human Nutrition Research in Cambridge for calcium and phosphorus analysis with the use of the same commercial kits and reference materials as described for breast-milk analysis. Maternal calcium intake Maternal calcium intake was assessed by 2-d weighed dietary record. Each subject was visited by a fieldworker before and after each meal and on several other occasions during the assessment period. All food items consumed and any leftovers were weighed to the nearest gram by using a small, portable scale, and recipes for all dishes were recorded. Particular attention was paid to whether dishes contained cow milk, baobab leaves, fish, or locust beans, the richest sources of calcium in the Gambian diet (4). Consumption of snacks between meals was ascertained by recall at the next visit. Computation of nutrient intakes from the food records was carried out by using a version of the computer program Diet In Data Out (21) that was adapted for Gambian foods. The coded records were analyzed by using an in-house suite of programs and a nutrient database for Gambian foods compiled from analytical work conducted in previous studies combined with recipe information (4, 22). Calcium from drinking water was not quantified because the calcium concentration of Keneba and Manduar water is low ( 10 mg/l) (4). Infant bone mineral status measurements Measurements of infant BMC (g/cm), BW (cm), and bone mineral density (BMD; g/cm 2 ) at the midshaft radius were made by single-photon absorptiometry [(SPA) Lunar SP2 scanner; Lunar Radiation Corporation, Madison, WI] at 2, 13, and 52 wk of age. The adult platform was replaced with one specially made for infants. The infant was laid supine, with the left arm extended along the platform and the palm facing down, and was gently held in position by the operator. The midshaft was identified by marking halfway between the olecranon and styloid process. The marked area was wrapped with a bag of tissue-equivalent material of appropriate size and placed in the measuring path. Three transverse scans were made at the same position, and the mean was recorded. The instrument was calibrated daily, and longterm stability was assessed regularly by using phantoms. The CV of BMC, BW, and BMD over the study period for the small phantom (BMC: g/cm) was 1.1%, 0.9%, and 0.8%, respectively, and that for the large phantom (BMC: g/cm) was 0.5%, 0.4%, and 0.5%, respectively, all of which indicated satisfactory instrument stability with no sign of drift. Partway through the study, a dual-energy X-ray absorptiometry (DXA, Lunar DPX ; Lunar Corporation, Madison, WI) instrument was installed at MRC Keneba. This instrument enabled the measurement of whole-body BMC (g), bone area (BA; cm 2 ), and BMD (g/cm 2 ) in a subset of infants at 2, 13, and 52 wk of age. Infants were measured while lightly wrapped in cotton material that did not substantially attenuate the X-ray beam; all clothing, jewelry, and amulets were removed. Scans were analyzed by DXA software (version 4.7b; Lunar Corporation). The pediatric small whole-body software was used for measurements made at 2 and 13 wk, and the pediatric medium software was used for measurements made at 52 wk. The pediatric small whole-body software is recommended for infants weighing 5 15 kg. It takes 10 min to scan a 5-kg infant but 20 min to scan a larger child. The faster pediatric medium software, therefore, was selected for measurements at 52 wk to minimize the possibility of movement artifacts. We previously showed from in vitro data that results are not compromised by the use of the faster software (23). All the DXA scans were closely scrutinized by an experienced member of the research team (MAL), and those judged to be of insufficient quality were not included in the final dataset. The infants were not sedated for either the SPA or DXA measurements. Mothers were invited to breastfeed their infants before the scan to encourage the infant to sleep or remain placid during the measurements. Difficulties with movement artifacts, periodic technical problems with the SPA, and the arrival of the DXA instrument after the start of the study meant that a complete set of 3 scans was not obtained for all infants. At least one successful SPA scan was obtained for 117 (n 57 and 60 in the supplement and placebo groups, respectively) and 1 DXA scan was obtained for 71 (n 34 and 37 in the supplement and placebo groups, respectively) of the 125 infants. Scans at 2 timepoints were obtained on the SPA for 114 infants (n 56 and 58 in the supplement and placebo groups, respectively) and on the DXA for 46 infants (n 24 and 22 in the supplement and placebo groups, respectively). Successful measurements at all 3 timepoints were achieved with SPA for 89 infants (n 46 and 43 in the supplement and placebo groups, respectively) and with DXA for 20 infants (n 9 and 11 in the supplement and placebo groups, respectively). Statistical analysis Descriptive statistics are reported as means SDs, and differences are reported as means SEs for all variables, unless otherwise stated. Statistical analysis was performed by using Student s t test, analysis of variance (ANOVA), analysis of covariance (ANCOVA), and multiple regression analysis with DATADES software (version 6.1.1; Data Description Inc, Ithaca, NY). The data were transformed into natural logarithms to allow the investigation of power relations between continuous variables and proportional (percentage) effects of discrete variables (24). When the dependent variable is in natural logarithms, the regression coefficient for a discrete variable, once multiplied by 100, corresponds closely to the percentage effect as defined by (difference/mean) 100 (25). All percentages reported here were derived in this way. In all cases, the distribution of logtransformed variables approximated normality. Transformation to natural logs also corrected a marked positive skewness in the urinary output data. Geometric mean urinary outputs were derived by taking the anti-logarithm of the means of the logged data.

5 INFANT OUTCOMES AFTER CALCIUM SUPPLEMENTATION IN PREGNANCY 661 The possible effect of calcium supplementation on the infant skeleton was examined in 4 ways: 1) the effect on BMC was examined to ascertain whether bone mineral mass had been altered; 2) the effect on bone size (BW for SPA and BA for DXA) and on body length was examined to ascertain whether skeletal size had been affected; 3) the effect on BMD, a commonly used marker of bone status derived by the ratio of BMC to BW for SPA or that of BMC to BA for DXA (which, however, can be influenced by bone size and is prone to size-related artifacts), was examined (24); and 4) the effect on BMC, and hence on BMD, after full correction for BW or BA, weight, and body length (size-adjusted BMC) was examined to ascertain the effect on the skeleton independent of bone and body size (24). For the maternal data, when a measurement was available before supplementation, the effect of the supplement was ascertained by conditional regression analysis, in which the dependent variable was the change in the value since baseline, and the independent variables were intervention group (n 1and0inthe supplement and placebo groups, respectively), baseline value, and potential confounders. Baseline value was always included to minimize regression toward the mean. For the infant and breast-milk data (ie, when no data were possible before supplementation), ANOVA or ANCOVA with Scheffé post hoc tests was used to examine the differences between the intervention groups at each timepoint. Effects over time within persons were examined by using repeated-measures ANOVA, or ANCOVA as appropriate, performed with the use of hierarchical linear models that included subject (nested by intervention group) and timepoint. An interaction term (intervention group timepoint) was introduced to consider the possible effect of supplementation on the rate of change in the dependent variable over time. Because, in these analyses, each subject acts as his or her own control, a full set of scans per subject was not required, and the models were constructed with the use of all available data. When the analyses were restricted to infants with no missing scans, we obtained similar results, both in the magnitude of the differences between the supplement and placebo groups at each timepoint and in the influence of supplementation in change over time, but the statistical significances were lower because of the smaller sample sizes (data not presented). All models were constructed initially to include possible confounders, and, for size-adjusted BMC ANCOVA models only, to include weight, height, and BW (or BA). Nonsignificant variables were removed by backward elimination to produce parsimonious models. Gambia has 2 distinct seasons wet (July to December) and dry (January to June) and marked seasonal influences on pregnancy weight gain, birth weight, and infant growth are well characterized in this population (26). Season, infant sex, birth order (parity of mother), and maternal calcium intake were included as potential confounders in all models, and interaction terms with intervention group were examined as appropriate. With one exception (see Results), no significant differences between the 2 groups were noted in any of these variables or interactions, and these findings are not discussed further. RESULTS The subject characteristics at P20 for the 125 women who completed the study are shown in Table 1. The differences TABLE 1 Subject characteristics at 20 wk of pregnancy (P20) 1 Calcium-supplemented group Placebo group (n 64) Age (y) Parity Nulliparity [n (%)] 9 (15) 11 (17) Primiparity [n (%)] 10 (16) 11 (17) Parity 2 5 [n (%)] 26 (43) 30 (47) Parity 6 10 [n (%)] 16 (26) 12 (19) P20 in dry season [n (%)] 3 28 (46) 22 (34) Weight (kg) Height (m) Calcium intake (mg/d) Urinary calcium output, log e (mg/d) Urinary phosphorus output, log e (mg/d) Urinary titratable acidity, log e (mmol/d) Log e, natural logarithm. There were no significant differences in any variable between the 2 groups at P20 (Student s 2-tailed t test). 2 x SD (all such values). 3 Dry season, January June; wet season, July December. 4 Mean of 8 measurements made during the study. between the supplement and placebo groups in age, parity, anthropometry, calcium intake and urinary mineral and acid outputs at baseline were not significant. At P36, the difference between the 2 groups in dietary calcium intake ( and mg/d for supplement and placebo groups, respectively) were not significant, but the total daily calcium intake in the supplement group was increased by the supplement to mg/d (P 0.001). The difference in maternal weight gain from P20 to P36 was not significant between the 2 groups ( and kg in the supplement and placebo groups, respectively; P 0.2 after adjustment for baseline and other confounders). At P20, the geometric mean urinary calcium output of the mothers was 67 mg/d, and no significant difference was observed between the 2 groups (Table 1). As anticipated, calcium supplementation had a significant effect on urinary calcium, phosphorus, and titratable acid outputs. At P36, the geometric mean urinary calcium output was higher in the supplement group (89.1 mg/d) than in the placebo group (49.6 mg/d). The difference was 59 15%, after adjustment for baseline value (log e values: and mg/d in the supplement and placebo groups, respectively; P 0.001). The difference was accounted for by an increase of 29 9% in the supplement group between P20 and P36 (P 0.03) and a decrease of 32 9% in the placebo group over the same period (P 0.02). The difference between the 2 groups in geometric mean urinary calcium output at P36 equated to 39.5 mg Ca/d, or 2.7% of the ingested calcium supplement. The maximum calcium output recorded at P36 was 347 and 358 mg/d in the supplement and placebo groups, respectively, which suggests that supplementation had not resulted in urinary calcium outputs above the normal range. Both urinary phosphorus and titratable acidity were lower in the supplement group than in the placebo group at P36 (log e urinary phosphorus output: and mg/d in the

6 662 JARJOU ET AL TABLE 2 Breast-milk calcium and phosphorus concentrations in the calcium-supplemented group () and the placebo group () 1 L02 L13 L52 (n 63) (n 58) (n 59) Calcium concentration (mg/l) Phosphorus concentration (mg/l) Calcium:phosphorus (mg/mg) All values are x SD. L02, L13, and L52, lactation at 2, 13, and 52 wk. Hierarchical repeated-measures ANOVA with subject (nested by intervention group) and timepoint showed significant decreases over time within persons (P 0.001) but no significant differences between the 2 groups at any timepoint or any significant intervention group timepoint interaction for all 3 variables. supplement and placebo groups, respectively; P after adjustment for baseline value; log e urinary titratable acidity: and mmol/d in the supplement and placebo groups, respectively. P after adjustment for baseline value). These differences were due to a significantly greater decrease in these variables in the supplement group than in the placebo group over the same period, at a time when both groups were experiencing significant (P 0.001) decreases in urinary phosphorus and acid output. The breast-milk calcium and phosphorus concentrations of the 2 groups at L02, L13, and L52 are shown in Table 2. Calcium supplementation had no significant effect on breast-milk calcium concentration, phosphorus concentration, or the calcium-tophosphorus ratio at any timepoint or on the rate of change over time. In the placebo group, the breast-milk phosphorus concentration was lower in those mothers who were at P20 in the dry season (wet dry: phosphorus concentration 9.3% 3.0%; P 0.003). There was no evidence of a significant seasonal effect in the supplement group (wet dry: phosphorus concentration 2.2% 2.7%; P 0.4), and the difference in seasonal response between the 2 groups was significant (season intervention group interaction: P 0.006). There was no indication of a significant seasonal effect or interaction with intervention group for breast-milk calcium concentration (P 0.1). Of the 125 infants born to the study mothers, 56 were female (n 27 and 29 in the supplement and placebo groups, respectively) and 69 were male (n 34 and 35 in the supplement and placebo groups, respectively). More infants were born in the wet (n 75; 33 in the supplement and 42 in the placebo group) than in the dry (n 50; 28 in the supplement and 22 in the placebo group) season. The mean birth weight of those infants born in the wet and dry seasons was 2.97 and 3.03 kg, respectively; the difference was not statistically significant (P 0.5). The data on infant anthropometry according to the group allocation of the mother are shown in Table 3. The differences between the 2 groups in birth weight, infant weight, body length, or head circumference within the first 5 d of life or at any time during infancy were not significant. In addition, the difference in gestational age at birth between the 2 groups was not significant ( and wk in the supplement and placebo groups, respectively; P 0.7). The bone mineral status data for all infants with a successful scan at any timepoint are given in Table 4. Within persons, whole-body BMC, BA, BMD, and size-adjusted BMC, as measured with DXA, increased significantly over time (P 0.001). At the midshaft radius, however, although there were significant increases over time in BMC and BW, as measured with SPA (P 0.001), there also was a tendency for BMD and size-adjusted BMC to decrease from 2 to 13 wk and then to increase slightly, but these changes were not statistically significant (P 0.09 and 0.1, respectively). The differences between the 2 groups in infant bone variables at any timepoint were not significant. However, both in the whole body and at the midshaft radius, there was a trend for supplement group infants to have slightly higher BMC and BA (BW) than did placebo group infants at 2 wk but for the converse to be the case at 52 wk, which suggests a slower velocity of skeletal growth during infancy in the supplement group. This trend was statistically significant in the whole body (group timepoint interaction: P 0.03 for BMC and 0.02 for BA), but not at the midshaft radius (P 0.1). The intervention group TABLE 3 Anthropometric measurements in the infants 1 Birth 5 d 2 2wk 13wk 52wk (n 59) (n 63) (n 49) (n 55) (n 64) (n 62) (n 60) Weight (kg) Weight-for-age SDS Length (mm) Length-for-age SDS Head circumference (mm) All values are x SD., calcium-supplemented group;, placebo group; SDS, SD score relative to the British reference data (27). Hierarchical repeated-measures ANOVA with subject (nested by intervention group) and timepoint by using data from 2, 13, and 52 wk showed significant increases over time within persons (P 0.001) but no significant differences between the 2 groups at any timepoint or any significant intervention group timepoint interaction for all 3 variables. 2 Obtained at the medical examination for gestational age assessment within 5dofbirth.

7 INFANT OUTCOMES AFTER CALCIUM SUPPLEMENTATION IN PREGNANCY 663 TABLE 4 Infant bone mineral status measurements, all available measurements 1 2wk 13wk 52wk Midshaft radius by SPA n BMC (g/cm) BW (cm) BMD (g/cm 2 ) Whole-body by DXA n BMC (g) BA (cm 2 ) BMD (g/cm 2 ) All values are x SD., calcium-supplemented group;, placebo group; BMC, bone mineral content; BW, bone width; BMD, bone mineral density; BA, bone area; SPA, single-photon absorptiometry; DXA, dual-energy X-ray absorptiometry. Hierarchical repeated-measures ANOVA with subject (nested by intervention group) and timepoint showed significant increases over time within persons (P 0.001) but no significant differences between the 2 groups at any timepoint. There was a significant intervention group timepoint interaction for whole-body BMC and BA (P 0.03 and 0.02, respectively), which suggests a slower velocity of bone growth between 2 wk and 52 wk in the supplement group than in the placebo group, but not for BMD or size-adjusted BMC or for any of the variables at the radius. timepoint interactions for BMD or size-adjusted BMC either in the whole body or at the midshaft radius were not significant. DISCUSSION This study showed that an increase in calcium intake during the second half of pregnancy in mothers who are accustomed to a very low calcium intake does not increase the transfer of calcium to the offspring, during either fetal life or subsequent breastfeeding. This finding supports evidence from well-nourished populations suggesting that physiologic mechanisms support human pregnancy independent of maternal calcium intake and, consequently, that no increase in intake by pregnant women is required (1, 28, 29). The dietary calcium intake of the women in this study, at 350 mg/d, was low compared with the intakes in Western countries and was similar to that recorded in previous studies in this population (4 6, 8). The calcium supplement provided an extra 1500 mg elemental Ca/d in the form of calcium carbonate. This high intake was chosen because the study was part of a larger trial in which blood pressure was the primary outcome, and doses of mg Ca/d have been used in previous investigations relating to pregnancy-induced hypertension both in wellnourished populations and in those with a low calcium intake (15, 16, 30). The supplement was well accepted, and no adverse effects were noted, either by the participants or by the finding of an increase in the maximum urinary calcium output. As expected, the mean urinary calcium output of the Gambian mothers at 20 wk of pregnancy was low compared with that of Western women (31, 32). This output had decreased by one-third at 36 wk in the placebo group. The difference in urinary calcium output between the supplement and placebo groups at P36 was equivalent to 3% of the ingested supplemental calcium, a value within the range typically observed in men and nonpregnant women during intervention studies with calcium carbonate, other calcium salts, or dietary calcium (8). This finding confirmed that the calcium carbonate supplement had been consumed and was biologically available. The lower urinary acid output of the supplement group than of the placebo group at P36 was further evidence of the uptake of the calcium carbonate supplement. The mean breast-milk calcium concentration of the Gambian women in the study was also similar to that recorded previously in this population (8). These concentrations are substantially lower than those observed in the United Kingdom and other Western countries (6, 7). As a consequence, the average calcium intake of Gambian infants during breastfeeding is lower than that of British infants, because breast-milk calcium concentration is independent of breast-milk volume (6). Increasing the calcium intake of the mothers in the second half of pregnancy had no significant effect on breast-milk calcium concentration at any time during the first year of lactation, which is contrary to our original hypothesis that was based on observations over several years in this population (9). The results also do not support the findings of an observational study of Spanish women that described an association between dietary calcium intake in the third trimester of pregnancy and breast-milk calcium concentrations (33). The fact that calcium supplementation in pregnancy did not increase breast-milk calcium concentrations during the subsequent lactation resembles the results of calcium supplementation studies in lactation (8, 34), which suggests that breast-milk calcium concentrations are determined by factors that are not related to maternal calcium intake during pregnancy and lactation. Calcium supplementation had no significant effect on fetal growth or BMA, as was shown by the lack of significant differences in birth weight and gestational age; in weight, crown-heel length, and head circumference in the first 5 d of life; and in anthropometric measurements and BMC of the whole body and midshaft radius at 2 wk of age. This finding is similar to the results of a study in India of pregnant women from a low socioeconomic background in which supplementation with either 300 or 600 mg Ca/d (n 24 and 25, respectively) did not significantly increase neonatal bone density, as measured by radiographic densitometry of the arms and legs, birth weight, or crown-heel and crown-rump lengths, compared with placebo (35). However, it contrasts with the results of a study in adolescent African Americans in which 2 servings of dairy products per day during

8 664 JARJOU ET AL TABLE 5 Comparison of dual-energy X-ray absorptiometry (DXA) measurements of infants 3 wk after birth 1 Location and instrument Age Weight BMC BA BMD Reference d kg g cm 2 g/cm 2 The Gambia, Lunar DPX 2 (n 20, ) Current study The Gambia, Lunar DPX (n 24, ) Current study United Kingdom, Lunar DPX (n 26) United Kingdom, Hologic QDR (n 81, M) United Kingdom, Hologic QDR (n 64, F) Canada, Hologic QDR 4500A (n 18, D ) NA NA 43 Canada, Hologic QDR 4500A (n 32, D ) NA NA 43 USA, Hologic QDR 1000/W (n 65) 1 8 not given USA, Hologic QDR 1000/W (n 128, ) USA, Hologic QDR 1000/W (n 128, ) USA, Hologic QDR A (n 63, PMF) USA, Hologic QDR A (n 64, MF) BMC, bone mineral content; BA, bone area; BMD, bone mineral density;, calcium-supplemented group;, placebo group; M, males; F, females; D, infants with low vitamin D status (defined as plasma 25-hydroxyvitamin D 37.5 nmol/l); D infants with adequate vitamin D status (defined as plasma 25-hydroxyvitamin D 37.5 nmol/l); PMF, palm olein enriched milk-based formula; MF, palm olein nonenriched milk-based formula; NA, not available. 2 Lunar Corporation, Madison, WI (all Lunar instruments listed). 3 x SD (all such values). 4 MA Laskey, A Prentice, unpublished observations, Hologic Inc, Waltham, MA (all Hologic instruments listed). 6 Weight at birth. 7 BA derived from BMC and BMD given in the publication cited. 8 SDs derived from SEM and sample size data given in the publication cited. pregnancy was associated with lower fetal femur length measured by prenatal ultrasound but not with other fetal anthropometric variables or birth weight (36). It is possible that calcium supplementation in our study affected femur length without influencing total body length, but this could not be measured with the methods used. The finding of a lack of significant differences also contrasts with a US study in which a higher whole-body BMC, measured by DXA, was observed 2 d after delivery in the offspring of women supplemented with 2000 mg Ca/d during pregnancy than in the offspring of unsupplemented women, but the difference was seen only in those participants (n 25) whose dietary calcium intake was in the lowest fifth of the distribution, ie, 600 mg Ca/d (37). Birth weight and gestational age were not significantly affected. The difference in the DXA results between the US and Gambian studies may reflect differences in the timing of the scan relative to delivery, when the exposure to the supplement effectively ceased (2 and 12 d after delivery, respectively). It is possible that the greater neonatal BMC in the US study represented a transitory effect related to the suppressive effects of calcium on bone resorption (38, 39). Alternatively, it may reflect differences between US and Gambian mothers in the extent to which they are adapted to a low calcium intake, or it may relate to the relatively small sample sizes in both studies. The pregnancy supplement also had no significant effect on growth or BMA in the infant during the first year of life, except for a possible modest reduction in the rate at which whole-body BMC and BA increased. On average, the infants in the study showed the typical growth pattern of Gambian children (40, 41) in that, relative to reference data, they were small at birth, grew well until 13 wk, but experienced severe growth faltering by 52 wk in terms of both weight-for-age and length-for-age. Gambian infants have also been shown to have a lower radial shaft BMC and size-adjusted BMC than do British infants. These bone measures diverge with increasing age, which suggests a slower rate of skeletal growth and BMA in infancy and early childhood (10). The SPA data of the study infants were in line with those in this earlier investigation. The DXA results from the current study, when compared with the limited data available from other studies (Table 5), also suggest that whole-body BMC, and hence the total-body calcium content, of Gambian infants is lower than that in Western populations. This comparison must be viewed with caution because DXA scanning of infants is a relatively new and largely unvalidated technique. Differences in software algorithms and bone edge detection and problems caused by shallow tissue depth and low X-ray attenuation mean that such comparisons should be strictly limited to those between scans conducted on crosscalibrated instruments from the same manufacturer. As can be appreciated from Table 5, between-instrument differences in DXA measurement are particularly pronounced for BA and BMD. However, despite these difficulties, infant DXA scanning appears to be consistent within studies and seems to provide useful information about differences in BMA within persons and between groups in the same population. The limited evidence from the current study, however, supports the possibility that fetal calcium accretion is lower in the fetuses of underprivileged mothers from developing countries than in those of mothers from developed countries (46), but the lack of an effect of the calcium supplement suggests that a low maternal calcium intake is not the primary determining factor. The results of this study support the view that metabolic adaptations occur during human pregnancy and lactation to provide

9 INFANT OUTCOMES AFTER CALCIUM SUPPLEMENTATION IN PREGNANCY 665 sufficient calcium for fetal growth and breast-milk production, such that the outcome is independent of maternal calcium intake. These adaptive processes are likely to include effects on 1 of the following: intestinal absorption, renal conservation, and mobilization of calcium from the maternal skeleton (1). In lactation, it was shown that maternal physiologic responses to breastfeeding are not influenced by current calcium intake (1, 47). It is, however, possible that the metabolic changes in pregnancy may be greater in women with a very low calcium intake than in those with a recommended calcium intake and that an increase in calcium intake may have health benefits for the mother. The effects of the calcium supplement on maternal bone mineral status, calcium metabolism, and blood pressure are currently being investigated and will be reported separately. In summary, this randomized, placebo-controlled supplementation study showed that an increase in calcium intake of 1500 mg/d in the second half of pregnancy by Gambian women accustomed to a very low calcium intake does not provide significant benefits to their offspring in terms of higher breast-milk calcium concentrations, infant birth weight, or growth and BMA in the first year of life. We thank the mothers and their infants for their patience and enthusiastic participation, and we acknowledge the contributions in Keneba of midwives Frances Foord, Ndey Haddy Bah, and Fatou Sosseh; fieldworkers Ebou Jarjou, Morikebba Sanyang, Lamin Sanneh, and Mariama Jammeh; and clinic and laboratory staff Abdoulie Jaiteh, Karamo Camara, Bakary Darboe, and Musa Colley; and in Cambridge of Dot Stirling, Shailja Nigdikar, and Ann Laidlaw for laboratory management; Alison Paul and Celia Greenberg for dietary coding and analysis; Sheila Levitt for data entry and Steve Austin for assistance with organizing tablet and sample transport; and Jaime Wu for assistance with searching the literature. LMAJ and AP were the principal investigators and were responsible for the study design, data collection and analysis, interpretation of results, and for drafting the manuscript. AP conceived the study and supervised LMAJ, who conducted the work as part of his PhD program. MAL and YS were responsible for single-photon absorptiometry and dual-energy X-ray absorptiometry measurements and interpretation. JB was responsible for the urine and breast-milk analyses. GRG was responsible for drafting and critically reviewing the manuscript. TJC was responsible for expert statistical input and for critical review of the manuscript. None of the authors had a financial or personal conflict of interest. REFERENCES 1. Prentice A. Micronutrients and the bone mineral content of the mother, fetus and newborn. J Nutr 2003;133:1693S 9S. 2. Forbes GB. Calcium accumulation by the human fetus. Pediatrics 1976; 57: Prentice A, Bates CJ. Adequacy of dietary mineral supply for human bone growth and mineralisation. Eur J Clin Nutr 1994;48S: Prentice A, Laskey MA, Shaw J, et al. The calcium and phosphorus intakes of rural Gambian women during pregnancy and lactation. Br J Nutr 1993;69: Prentice A. Calcium intakes and bone densities of lactating women and breast-fed infants in The Gambia. Adv Exp Med Biol 1994;352: Prentice A, Laskey MA, Jarjou LMA. Lactation and bone development: implications for the calcium requirements of infants and lactating mothers. In: Tsang RC, Bonjour J-P, eds. Nutrition and bone development. New York: Vevey/Lippincott-Raven, 1999: Laskey MA, Prentice A, Shaw J, Zachou T, Ceesay SM. Breast-milk calcium concentrations during prolonged lactation in British and rural Gambian mothers. Acta Paediatr Scand 1990;79: Prentice A, Jarjou LMA, Cole TJ, Stirling DM, Dibba B, Fairweather- Tait S. Calcium requirements of lactating Gambian mothers: effects of a calcium supplement on breastmilk calcium concentration, maternal bone mineral content, and urinary calcium excretion. Am J Clin Nutr 1995; 62: Prentice A, Dibba B, Jarjou LMA, Laskey MA, Paul AA. Is breast-milk calcium concentration influenced by calcium intake during pregnancy? Lancet 1994;344: Prentice A, Laskey MA, Shaw J, Cole TJ, Fraser DR. Bone mineral content of Gambian and British children aged 0 36 months. Bone Miner 1990;10: Prentice A, Paul AA. Contribution of breast-milk to nutrition during prolonged breastfeeding. In: Atkinson S, Hanson L, Chandra R, eds. Human lactation 4. Breastfeeding, nutrition, infection and infant growth in developed and emerging countries. St John s, Canada: ARTS Biomedical Publishers, 1990: Jarjou LMA. The calcium intake of breast-fed Gambian infants and its influence on growth and bone mineral development in the first year of life. Master s thesis. Anglia Polytechnic University, Cambridge, United Kingdom, Prentice A, Shaw J, Laskey MA, Cole TJ, Fraser DR. Bone mineral content of British and rural Gambian women aged years. Bone Miner 1991;12: Moher D, Schulz KF, Altman DG, for the CONSORT Group. The CON- SORT statement: revised recommendations for improving the quality of reports of parallel-group randomised trials. Lancet 2001;357: Belizan JM, Villar J, Gonzalez L, Campodonico L, Bergel E. Calcium supplementation to prevent hypertensive disorders of pregnancy. N Engl J Med 1991;325: Villar J, Repke J, Belizan JM, Pareja G. Calcium supplementation reduces blood pressure during pregnancy: results of a randomised controlled trial. Obstet Gynecol 1987;70: Gleerup A, Rossander-Hulthen L, Gramatkovski E, Hallberg L. Iron absorption from the whole diet: comparison of the effect of two different distributions of daily calcium intake. Am J Clin Nutr 1995;61: Dubowitz LMS, Dubowitz V, Goldberg C. Clinical assessment of gestational age in the newborn infant. J Pediatr 1970;77: Freeman JV, Cole TJ, Chinn S, Jones PRM, White EM, Preece MA. Cross-sectional stature and weight reference curves for the UK, Arch Dis Child 1995;73: Laskey MA, Dibba B, Prentice A. A semi-automated micromethod for the determination of calcium and phosphorus in human milk. Ann Clin Biochem 1991;28: Price GM, Paul AA, Key FB, et al. Measurement of diet in a large national survey: comparison of computerised and manual coding of records in household measures. J Hum Nutr Diet 1995;8: Prynne CJ, Paul AA, Dibba B, Jarjou LMA. Gambian food records: a new framework for computer coding. J Food Compos Anal 2002;15: Laskey MA, Prentice A. Comparison of adult and paediatric spine and whole body software for the Lunar dual energy X-ray absorptiometer. Br J Radiol 1999;72: Prentice A, Parsons TJ, Cole TJ. Uncritical use of bone mineral density in absorptiometry may lead to size-related artifacts in the identification of bone mineral determinations. Am J Clin Nutr 1994;60: Cole TJ. Sympercents: symmetric percentage differences on the 100 loge scale simplify the presentation of log transformed data. Stat Med 2000; 19: Rayco-Solon P, Fulford AJ, Prentice AM. Differential effects of seasonality on preterm birth and intrauterine growth restriction in rural Africans. Am J Clin Nutr 2005;81: Cole TJ. Growth monitoring with the British 1990 growth reference. Arch Dis Child 1997;76: Department of Health. Nutrition and bone health: with particular reference to calcium and vitamin D. Report of the Subgroup on Bone Health, Working Group on the Nutritional Status of the Population of the Committee on Medical Aspects of Food and Nutrition Policy. Report on Health and Social Subjects 49. London: The Stationery Office, Food and Nutrition Board, Institute of Medicine. Dietary reference intakes for calcium, phosphorus, magnesium, vitamin D, and fluoride. Washington, DC: National Academy Press, Levine RJ, Hauth JC, Curet LB, et al. Trial of calcium to prevent preeclampsia. N Engl J Med 1997;337: Ritchie LD, Fung EB, Halloran BP, et al. A longitudinal study of calcium homeostasis during human pregnancy and lactation and after resumption of menses. Am J Clin Nutr 1998;67: Cross NA, Hillman LS, Allen SH, Krause GF, Vieira NE. Calcium homeostasis and bone metabolism during pregnancy, lactation, and postweaning: a longitudinal study. Am J Clin Nutr 1995;61: