点击显示 收起
1 From the Southeast Asian Ministers of Education Organization–Tropical Medicine Regional Center for Community Nutrition, University of Indonesia, Jakarta, Indonesia (MW-E, JU, EK, and LW); the Institute of Biological Chemistry and Nutrition, University of Hohenheim, Stuttgart, Germany (JGE); and the Deutsche Gesellschaft für Technische Zusammenarbeit (German Technical Cooperation), Eschborn, Germany (RG)
2 Any opinion, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the views of the supporting agency. 3 Rainer Gross died on 30 September 2006. 4 Supported by UNICEF (New York, NY). 5 Reprints not available. Address correspondence to M Wijaya-Erhardt, SEAMEO-TROPMED Regional Center for Community Nutrition, University of Indonesia, Jalan Salemba Raya 6, Jakarta 10430, Indonesia. E-mail: mwijaya{at}seameo-rccn.org.
ABSTRACT
Background: There is still uncertainty about the best procedure to alleviate iron deficiency. Additionally more reliable methods are needed to assess the effect of iron intervention.
Objective: We examined the efficacy of daily iron (10 mg), daily and weekly multiple-micronutrient supplementation (10 and 20 mg Fe, respectively) in improving body iron stores of Indonesian infants.
Design: Infants aged 6–12 mo were randomly allocated to 1 of 4 groups: daily multiple-micronutrients (DMM) foodlike tablets (foodLETs), weekly multiple-micronutrient (WMM) foodLETs, daily iron (DI) foodLETs, or daily placebo. Hemoglobin, ferritin, transferrin receptors, and C-reactive protein data were obtained at baseline and 23 wk.
Results: Body iron estimated from the ratio of transferrin receptors to ferritin was analyzed for 244 infants. At baseline, mean iron stores (0.5 ± 4.1 mg/kg) did not differ among the groups, and 45.5% infants had deficits in tissue iron (body iron < 0). At week 23, the group DI had the highest increment in mean body iron (4.0 mg/kg), followed by the DMM group (2.3 mg/kg; P < 0.001 for both). The iron stores in the WMM group did not change, whereas the mean body iron declined in the daily placebo group (–2.2 mg/kg; P < 0.001). Compared with the daily placebo group, the DMM group gained 4.55 mg Fe/kg, the DI group gained 6.23 mg Fe/kg (both P < 0.001), and the WMM group gained 2.54 mg Fe/kg (P = 0.001).
Conclusions: When compliance can be ensured, DI and DMM foodLETs are efficacious in improving and WMM is efficacious in maintaining iron stores among Indonesian infants.
Key Words: Iron deficiency • infants • multiple micronutrients • foodlike tablets • foodLETs • Indonesia • body iron stores
INTRODUCTION
Iron deficiency (ID) in the 6–24-mo-old population impairs the normal mental development of 40–60% of the developing world's infants. ID is highly prevalent in Indonesia, which has an estimated 38% prevalence of iron deficiency anemia (IDA) in children <5 y old (1).
Supplementation with iron tablets is the most widely used approach for controlling the global problem of ID and IDA. Because several micronutrients can improve the hemoglobin response to iron, it is reasonable to assume that multiple micronutrients would be more effective in reducing anemia than would iron alone. One important reason for the relative lack of programmatic success with iron supplements has been the perceived need for these supplements to be taken daily over relatively long periods. For this reason, and because of concern about the potential toxicity and intolerance of daily supplementation, there has been interest in the efficacy of weekly iron supplementation as compared with daily supplementation (2).
Anemia measured by hemoglobin concentration is used as a proxy indicator of ID. However, with regard to iron status, this is neither a specific measure—because anemia also may be caused by malaria, intestinal parasites, or other factors (3-5)—nor a sensitive measure—because ID can exist without anemia (4, 5). Although the plasma ferritin (PF) concentration is one of the most widely used indicators of iron status, it is known to be affected by several factors, including infection and inflammation. In contrast, some studies show that serum concentrations of soluble transferrin receptors (TfRs) accurately reflect iron status in the presence of inflammation (6-12). The ratio of TfR to PF is a more sensitive and specific indicator of iron status than is either measurement alone (5, 8, 11). Quantitative measurements of body iron based on the ratio of TfR to PF were shown to measure absorption of the added iron in trials of iron supplementation in pregnant Jamaican women and iron fortification in Vietnamese women (5). We aimed to evaluate the role of body iron measurements in comparing the efficacy of daily iron (DI) to those of daily and weekly multiple-micronutrient (WMM) foodlike tablets (foodLETs) in improving the iron status of Indonesian infants 6–12 mo old.
SUBJECTS AND METHODS
The present study was part of a multicenter study of International Research on Infant Supplementation (IRIS) trial that was conducted in populations from 4 quite different countries: Indonesia, Peru, South Africa, and Vietnam. The aim of the IRIS study was to investigate whether multiple micronutrient supplementation can prevent growth faltering, anemia, and micronutrient deficiency during infancy in Indonesia.
Study location and population
The details about participants, study design, and other results from the Indonesian part of the multicountry study were reported previously (13, 14). Briefly, we conducted the study from June to December 2000 in 12 villages in the Salam subdistrict and in 6 villages in the Ngluwar subdistrict, both of which are located in Magelang district. Magelang lies in the center of Java, the most populous island of Indonesia. We selected the study area according to the requirements of the IRIS protocol—ie, that the prevalence of anemia (hemoglobin: < 110 g/L) and of vitamin A deficiency (serum retinol: < 0.7 µmol/L) should be 30% in the population <5 y old. Deficiencies of several micronutrients were found in our subjects: 58.1% were anemic, 34.5% had PF concentrations < 12 µg/L, 10.4% had plasma zinc concentrations < 10.7 µmol/L, and 24.3% had plasma retinol concentrations < 0.7 µmol/L (13, 15). Approximately 12% of the subjects were underweight, and 8% were stunted (15).
We conducted a screening study in 356 infants 6–12 mo old from a list of infants in the study area and excluded those with premature birth (<37 wk gestation), low birth weight (<2500 g), severe wasting (<–3 z scores), severe anemia (hemoglobin: < 80 g/L), or fever (>39 °C) on the day of blood sampling. We gave additional WMM foodLETs to the placebo group for 3 mo after the end of the supplementation trial.
The purpose and procedures of the study were explained to the parents at enrollment, and only the infants whose parents who gave written informed consent were recruited. The IRIS trial was guided by the ethical considerations recommended by the Council for International Organizations of Medical Sciences (16). The Ethical Committee for Studies on Human Subjects, Faculty of Medicine, University of Indonesia approved the study protocol.
Procedures
In advance of recruitment, we used a simple computer program (a random number generator) for the randomization process, which was done by household. At baseline, we then enrolled 284 (80%) eligible infants and randomly allocated them to 4 groups as follows: daily multiple micronutrients (DMM; n = 72), WMM (n = 70), DI (n = 72), or daily placebo (DP; n = 70). Through the use of coding, the subjects' families and the investigators were blinded to the specific composition of the foodLETs assigned. One family had twin boys who fulfilled the requirements to participate in the study; therefore, both boys were included and treated as separate cases, but they were allocated to the same intervention group. The allocation codes 1, 2, 3, and 4 were kept centrally by UNICEF New York and opened at the end of the study, before statistical analyses (13).
The multiple-micronutrient supplement and placebo were produced in the form of foodLETs and were provided in blister packs of 7 tablets. The micronutrient content of each foodLET had been formulated according to the daily Recommended Nutrient Intake for infants 1–2 y old, with the exception of zinc, which was the equivalent of 1 Recommended Nutrient Intake for children 6–12 mo old: 375 µg RE vitamin A (as retinyl acetate), 5 µg vitamin D, 6 mg -tocopherol equivalents of vitamin E (as -tocopherol), 10 µg vitamin K, 35 mg vitamin C, 0.5 mg thiamine, 0.5 mg riboflavin, 0.5 mg vitamin B-6, 0.9 µg vitamin B-12, 6 mg niacin (as niacinamide), 150 µg folate, 10 mg iron (as ferrous fumarate), 5 mg zinc (as zinc gluconate), 0.6 mg copper (as cupric gluconate), and 59 µg iodine (as potassium iodide). The dose of the WMM (ie, weekly) foodLETs was double that of the DMM. DI foodLETs contained 10 mg iron (as ferrous sulfate), whereas DP foodLETs contained no added micronutrients. The ingredients in all foodLETs were similar (ie, milk solids, confectionery sugar, fructose, flavor, citric acid, calcium carbonate, and magnesium stearate). Milk solids (1100 mg) were added to each foodLET (total weight: 2800 mg) to prevent chemical interactions between the micronutrients. Roche Laboratories (Nutley, NJ) developed the product, and Hersil (Lima, Peru) produced the crushable and dissoluble foodLETs (14, 17, 18).
Blisters were labeled with the printed subjects' names and allocation code only. Each supply of 7 tablets was wrapped in identical coded blister packs, and all foodLETs had the same taste, color, and flavor. The foodLETs were given daily at home—on days 1–6, the consumption of the foodLET was under the close supervision of a trained fieldworker, and, on day 7, it was under the supervision of the child's mother. Thus, compliance was recorded directly for the first 6 d, and that on day 7 was recorded on the next day's visit. Twenty-three infants did not complete the study: during the study period, 12 infants' families moved to another area (n = 2, 5, 2, and 3 in the DMM, WMM, DMM, and DP groups, respectively), 9 discontinued the trial (n = 4, 3, and 2 in the DMM, WMM, and DP groups, respectively), and 2 refused the second blood collection (n = 1 each in the DMM and WMM groups) and were not included in the data analysis. The losses to follow-up were equally distributed across the 4 treatment groups (13). Infants also received vitamin A as part of a national program; those 12 mo old were given 200 000 IU vitamin A, and those <12 mo old were given 100 000 IU.
We collected all information by using a structured questionnaire according to survey methods (19). We measured the weight and length of infants monthly and the weight and height of mothers at baseline; the weight was measured to the nearest 0.1 kg with the use of an electronic weighing scale (model 770; SECA, Hamburg, Germany). We measured infant length on a length board (model 210; SECA) and maternal height to the nearest 0.1 cm by using a standing height measurement microtoise. The mothers' perception of the intervention was evaluated in mid-November 2000 by administering a questionnaire that asked about the mothers' perceptions of the effects and side effects of the supplementation during the intervention. Blood samples were taken from infants at baseline and at 23 wk of supplementation. Because of a Muslim festival celebrated by most of the subjects studied, we deviated in one instance from the protocol, reducing the study period by 2 wk. However, we continued giving the foodLETs to the infants up to week 25. A 2-mL blood sample was withdrawn via venipuncture into vacuette heparin tubes using a butterfly-plus-luer adapter (Greiner, Kremsmuenster, Austria). The tubes were put in a cooled styropore box with coolpacks. We measured the hemoglobin concentration by using the cyanomethemoglobin method (# 4010l; Boehringer, Mannheim, Germany) and centrifuged (Hettich, Tuttlingen, Germany) the whole blood within 6 h at 3600 rpm for 15 min at room temperature at the Muntilan General Hospital Laboratory, Magelang. The plasma was transferred into two 500-µL Eppendorf cups and was kept frozen at –76 °C before being shipped to Germany in December 2000. The blood samples were analyzed in the Micronutrient Laboratory of the Institute of Biological Chemistry and Nutrition at the University of Hohenheim (Stuttgart, Germany). PF and C-reactive protein (CRP) were measured by using an enzyme-linked immunosorbent assay (20). Originally, the TfR was not included in the protocol; at the Southeast Asian Ministers of Education Organization–Tropical Medicine Regional Center for Community Nutrition (Jakarta, Indonesia) in 2004, it was analyzed in 259 samples with the use of the same method (20).
A hemoglobin concentration < 110 g/L was used as a cutoff for defining anemia, a PF concentration < 12 µg/L was used as the cutoff for ID, and a hemoglobin concentration < 110 g/L and a PF < 12 µg/L were used as the cutoff for IDA (21, 22). Tissue ID was defined as a TfR concentration > 8.5 mg/L and iron overload with a PF concentration > 400 µg/L (22). The cutoff for inflammation was a plasma CRP concentration of 12 mg/L (23). A body iron store was estimated by using the following equation (5, 24):
RESULTS
Of the 261 infants who completed the study, the sample from 1 infant was not measured for PF and that from another infant was not measured for TfR (in both cases, because of a low volume of plasma); 1 blood sample was missing. Fifteen infants had plasma CRP concentrtions > 12 mg/L at baseline or at week 23, which indicated inflammation; those infants were excluded from the statistical analysis for PF and body iron stores. A detailed description of the study profile was presented elsewhere (13).
The baseline characteristics of the 4 groups did not differ significantly (Table 1). The average (± SD) household size was 5.0 ± 1.6 persons, and the mean number of children < 5 y age was 1.2 ± 0.4. Most (>98%) households had electricity, either directly by public supply (173/244) or via a neighbor (67/244). The mean body mass index (in kg/m2) of mothers was normal at 21, and 28 (12%) were overweight (body mass index > 25). Among all infants, 60 (25%) had IDA (hemoglobin: < 110 g/L and PF < 12 µg/L); 83 (34%) had both a low PF concentration (<12 µg/L) and an elevated TfR concentration (>8.5 mg/L); 204 (79%) had tissue ID (TfR > 8.5 mg/L). One hundred thirty-three (55%) infants had body iron stores, and the other 111 infants (45.5%) had a mean tissue iron deficit [3.5 and –3.1 mg/kg, respectively (95% CI for the difference: –7.3, –6.1)]. Boys (128/244) on average had a tissue iron deficit, whereas girls (116/244) had body iron stores [–0.3 and 1.4 mg/kg, respectively (95% CI: –2.8, –0.7)]. None of these values were indicative of iron overload (PF > 400 µg/L) at the beginning or the end of the study.
View this table:
TABLE 1. Baseline characteristics1
Before treatment, there were no significant differences among the 4 groups in iron status (PF, TfR, and body iron) (Table 2). The geometric means of PF and TfR were 17.44 µg/L (95% CI: 15.38, 19.77 µg/L) and 10.08 mg/L (9.81, 10.35 mg/L), respectively, and the mean (± SD) of iron stores was 0.5 ± 4.1 mg/kg. During the 23-wk trial, the DMM and DI groups had significantly higher mean PF and significantly lower TfR concentrations than did the WMM and DP groups. The proportion of tissue ID (TfR > 8.5 mg/L) decreased from 72% to 54% and from 75% to 45% in the DMM and DI groups, respectively, and remained unchanged in the WMM and DP groups (from 85% to 80% and from 83% to 85%, respectively). Time x treatment interactions were significant for the proportion of infants with TfR concentrations > 8.5 mg/L (P < 0.001). The mean body iron increased 2.3 mg/kg in the DMM group and 4.0 mg/kg in the DI group (P < 0.001 for both). Thus, the average daily iron gain was 0.015 and 0.025 mg/kg in the DMM and the DI group, respectively. In the WMM group, body iron stores did not change significantly (0.3 mg/kg; P = 0.468). In the DP group, the mean body iron declined 2.2 mg/kg (P < 0.001); this average daily iron loss was 0.014 mg/kg. Compared with the mean iron absorption in the DP group, that in the DMM group over 23 wk was 4.55 mg/kg or 41 mg total body iron, which represented the absorption of 2.5% of the 1610 mg Fe that the DMM foodLET provided; the DI group gained 6.23 mg Fe/kg or 55 mg total body iron, which represented the absorption of 3.4% of the 1610 mg Fe that the DI foodLET provided (P < 0.001 for both). Infants in the WMM group gained 2.54 mg Fe/kg or 22 mg total body iron during 23 wk (P = 0.001), which represented the absorption of 4.8% of the 460 mg Fe that the WMM foodLET provided. The mean iron absorption in the DMM and DI groups was significantly (P < 0.01) higher than that in the WMM group.
View this table:
TABLE 2. Iron status indicators in Indonesian infants 6–12 mo old at baseline and after 23-wk supplementation1
By the end of the study, the proportion of infants who had a tissue iron deficit (body iron < 0) had decreased significantly in the DMM and DI groups, but the proportion in the WMM group did not change significantly (Table 3). The proportion of infants with body iron deficit in the DP group increased significantly (P = 0.02). In anemic infants (hemoglobin: < 110 g/L), the proportion of those with a tissue iron deficit at week 23 decreased significantly (P < 0.001) in the DMM and DI groups. In infants with IDA (hemoglobin < 110 g/L and PF < 12 µg/L), the proportion of those with tissue iron deficit decreased in the 3 supplemented groups: P = 0.002, P = 0.031, and P < 0.001 for the DMM, WMM, and DI groups, respectively.
View this table:
TABLE 3. Proportion of infants who had tissue iron deficit at baseline and at week 23 by treatment group1
The relation between iron status indexes and age was examined by pooling the observations from the DP group (n = 120) at baseline and at week 23 (Table 4). In infants <1 y old, age was negatively correlated with PF and body iron and positively correlated with TfR but was not correlated with hemoglobin. No significant correlation was found between age and iron status indexes in children >1 y old. Children >1 y old had a mean tissue iron deficit (body iron < 0) of –2.1 mg/kg body wt, whereas those <1 y old had iron stores (body iron > 0) of 0.5 mg/kg body wt (P < 0.001). Of the 38 children with IDA, all had a tissue iron deficit (body iron < 0; data not shown). Similarly, of the 60 infants with IDA in the 4 groups at baseline, all had body iron < 0 (Table 3).
View this table:
TABLE 4. Correlation coefficients between iron status indicators and age in the infant1
The rate of compliance for consuming 121 (75%) foodLETs was 54 (89%), 52 (90%), 59 (91%) and 54 (90%) in the DMM, WMM, DI, and DP groups, respectively (P = 0.98). The most common reasons for not taking the foodLETs was illness, such as diarrhea, fever, or stomatitis aphthosa in the DMM group (n = 9) and forgetting to take supplements in the WMM (n = 5), DI (n = 6), and DP (n = 4) groups. When the mothers were asked about their experience of the intervention for their infants, 45 (74%) in the DMM group and 47 (72%) in the DI group reported that their infants had constipation or dark stool after taking the foodLETs, whereas 30 (52%) and 10 (17%) did so in the WMM and DP groups, respectively (P < 0.001). The DMM, WMM, DI, and DP groups did not differ significantly in reported vomiting (39%, 41%, 23%, and 30%, respectively; P = 0.11) or diarrhea (18%, 14%, 20%, and 10%, respectively; P = 0.42). If the intervention were to be repeated, most mothers in the DMM, WMM, and DP groups would prefer to receive supplements weekly [n = 25 (41%), 28 (48%), and 24 (40%), respectively], whereas the mothers in the DI group did not prefer weekly over daily delivery [n = 23 (35%)]. There was no significant difference among groups in frequency preference (P = 0.82).
DISCUSSION
The results from the present study, which featured a double-masked study design and strict monitoring of compliance, showed that multiple micronutrient and iron foodLETs on a daily basis are efficacious in improving body iron stores and reducing the proportion of a mean tissue iron deficit (body iron < 0) in Indonesian infants. When the change in body iron estimates was used, a difference in each of the 4 groups was significant. In our previous report (13), the change in ferritin in DMM group did not differ significantly from that in the DI and WMM groups, and that in the WMM group did not differ significantly from that in the DP group. Our results agree with those of Cook et al (5), which indicated that body iron measurements showed a difference between the 2 groups given iron supplements that was not identified by the TfR measurement alone. More important, body iron measurements can be used to calculate iron absorption in iron intervention trials (5). In populations in whom the prevalence of ID is high—eg, infants, children, and pregnant women—the serum ferritin concentration usually is close to the iron-deficient range, and it does not accurately portray differences in functional iron (6). The equation for estimating body iron stores from the ratio of TfR to PF was derived from an adult phlebotomy study (5, 24) and has not been validated in children, which could be a limitation with respect to our infant subjects. However, that equation performed well in recent iron trials in schoolchildren (26, 27).
A WHO/CDC meeting evaluated the performance of 5 indicators of iron status (ie, hemoglobin, zinc protoporphyrin, mean cell volume, TfR, and serum ferritin) and recommended 3 indicators for predicting changes in and assessing iron status: hemoglobin, serum ferritin, and TfR. If possible, the acute-phase protein CRP or –1 acid glycoprotein (or both) should also be measured (28).
During the 23-wk intervention, the DI and DMM groups had an iron absorption of 3.4% and 2.5%, respectively. With respect to improving the iron status of infants, the DI regimen is more effective than is the DMM regimen. A possible explanation for this effect is that the iron absorption from ferrous fumarate in the DMM group was lower than that from ferrous sulfate in the DI group. A previous study showed that the mean iron absorption from ferrous fumarate relative to ferrous sulfate in Bangladeshi children was only 27–35% (29). In Mexican children aged 2–4 y, iron absorption from complementary food fortified with ferrous sulfate was higher (7.9 ± 9.8%) than that from ferrous fumarate (2.43 ± 2.3%) (30). Another explanation could be that the inclusion of zinc in the DMM supplement may have interfered with the absorption of iron. As was shown, supplementation with iron and zinc (each: 10 mg/d) had a smaller positive effect than did iron alone, but the effect was still significant in improving the iron status in infants (31, 32). However, not all body iron deficits were fully controlled in the DI group infant population after 23 wk of treatment. Global guidelines for iron supplementation have been published by the International Nutritional Anemia Consultative Group/WHO/UNICEF. The recommendations are that, in persons 6–24 mo old, 12.5 mg Fe/d plus 50 µg folic acid/d until the age of 12 mo (2); this dose is slightly higher than the supplementation dose in the present study.
In our previous report (13), we showed that DMM was more efficacious than was DI in improving hemoglobin concentration and in decreasing the proportion of anemia in infants. This finding may be due to the fact that anemia is caused not just by iron deficiency but also by a lack of other components that are in the DMM supplement, such as riboflavin, folic acid, vitamin C, and vitamin A, all of which are known to favor iron absorption or hematopoiesis (or both) (15).
Several studies in different population groups have shown that weekly dosing with iron alone or iron along with other micronutrient supplementation successfully reduced anemia and improved micronutrient status and that it was as effective as daily administration (33-40). The currently available evidence suggests that iron supplements should be taken daily to treat anemia, especially by pregnant women who are consuming low amounts of available iron. However, weekly supplements would reduce side effects (36, 39), and the lower costs may improve compliance. The daily foodLETs used in the present trial provided 70 mg Fe when calculated on a weekly basis, whereas the weekly foodLET provided 20 mg Fe. The present study showed that daily dosing of iron alone or with other micronutrient foodLETs is superior to weekly dosing in the treatment of ID in infants, but the WMM supplements had a greater effect in preventing the worsening of iron status than did the DP.
At baseline, almost half of respondents had a tissue iron deficit (body iron < 0). Infants receiving DP, who represented the healthy population without intervention, had an average iron loss of 0.01 mg/d and developed a mean deficit in tissue iron over 23 wk, although they were still being breastfed and were given complementary feeding. These findings can be generalized with regard to the importance of providing additional iron to infants. Iron is involved in many central nervous system processes that could affect infant behavior and development. The key liabilities of tissue ID relate to abnormalities in host defense, work performances, and neurologic function (41).
When stratified into age groups, we found that infants <1 y old had iron stores (body iron > 0), whereas those >1 y old had a tissue iron deficit (body iron < 0); these findings are consistent with earlier findings showing that infants <1 y old had body iron stores of 0.29 ± 1.04 mg/kg and those 1 y old body iron stores of –0.50 ± 0.74 mg/kg. This decrease in 1-y-old infants was followed by a linear increase in body iron after the age of 2 y, despite increasing body size (24). We also found that IDA occurred in infants at an iron stores deficit of <–3 mg/kg, which is slightly higher than the results of previous studies with a deficit of <–4 mg/kg in adults and in children (5, 24). The subjects in the present study developed mild anemia (hemoglobin: 100.5 g/L) at an iron deficit of –1.8 mg/kg, which supports the findings of a study in 6–10-y-old Moroccan children showing that children develop mild anemia (hemoglobin: 108 ± 4 g/L) at mean iron deficits of –1.1 mg/kg (26).
In conclusion, the present study found that both DMMs and DI foodLETs given for 23 wk increased the body iron stores of Indonesian infants aged 6–12 mo, and WMM foodLETs prevented the deterioration of iron stores. These findings support the use of the body iron stores calculations suggested by Cook et al (5) to improve the precision of the laboratory diagnosis of iron absorption, detect IDA, and find the most appropriate iron intervention strategy.
ACKNOWLEDGMENTS
We thank all the parents and their infants for participating in this study. We also thank Kristiani (head of Salam primary health center) and Nurul (head of Ngluwar primary health center) and their staffs for valuable advice, help, and cooperation in the field. We are grateful for the assistance of fieldworkers, blood analyst team (Muntilan General Hospital), personnel in subdistrict health office (Magelang), and laboratory staffs (Southeast Asian Ministers of Education Organization–Tropical Medicine Regional Center for Community Nutrition–University of Indonesia).
The authors' responsibilities were as follows—MW-E: supervised field activities, entered data, conducted the analysis, and drafted the manuscript; JGE: analyzed the blood samples and assisted in writing the manuscript; JU and EK: were principal investigators and took overall responsibility for the study in Indonesia and writing of the manuscript; LW: supervised field activities, entered data, and conducted the analysis; and RG: conceived of and designed the the multicenter study. None of the authors had a personal or financial conflict of interest.
REFERENCES