点击显示 收起
1 From the International Center for Diarrhoeal Disease Research, Bangladesh (ICDDR,B): Centre for Health and Population Research, Dhaka, Bangladesh; the Department of Pediatrics, Louisiana State University School of Medicine, New Orleans; the Division of Human Nutrition and Epidemiology, Wageningen University, Wageningen, Netherlands; the Institute of Nutrition & Food Science, Dhaka University, Dhaka, Bangladesh; and the Centre for International Child Health, Institute of Child Health, University College London.
See corresponding editorial on page 281.
ABSTRACT
Background: Zinc deficiency is thought to be common in young children in developing countries and some data suggest that it may detrimentally affect children's development.
Objective: Our goal was to assess the effect of zinc supplementation on the developmental levels and behavior of Bangladeshi infants.
Design: This was a randomized, double-blind, controlled trial conducted in Dhaka, Bangladesh. Three hundred one infants aged 1 mo were randomly assigned to receive either 5 mg elemental Zn or placebo daily for 5 mo, and subsequent growth and morbidity were observed. For the present study, developmental levels were assessed in a subsample of 212 infants at 7 and 13 mo of age with use of the Bayley Scales of Infant Development, and the infants' behavior during the tests was observed. The children's social backgrounds, weights, and lengths were also recorded.
Results: The children's nutritional status was generally poor. The zinc-treated group had slightly lower scores on the mental development index of the Bayley Scales than did the placebo group (ß = 3.7, SE = 1.3, P < 0.005). This effect remained significant when nutritional status and social background were controlled for. No other significant differences between groups were noted.
Conclusions: The mental development index scores of the zinc-treated group were slightly but significantly lower than those of the placebo group. This finding may have been due to micronutrient imbalance. Caution should be exercised when supplementing undernourished infants with a single micronutrient.
Key Words: Zinc deficiency • zinc supplementation • infants • mental development • behavior • Bangladesh • clinical trial • Bayley Scales of Infant Development
INTRODUCTION
Zinc deficiency has been recognized as a public health problem (1). It is known to occur in children with frequent diarrhea and in those whose diets are high in phytates and low in bioavailable zinc (2). Many undernourished children in developing countries fit this description. Zinc deficiency affects the incidence and prevalence of diarrhea, the incidence of pneumonia (3), and growth (4). There is some concern that zinc deficiency may also detrimentally affect children's development and behavior (5).
Research in animals showed that severe zinc deficiency affects brain development (6), and that deficiency in postnatal life is associated with increased emotional response to stress (7), reduced motor activity, and less accurate performance on measures of attention and short-term memory (6, 8). Relatively few studies have examined the effect of zinc supplementation on behavior and development in children. An observational study showed an association between maternal dietary zinc intake and children's motor development (9). In studies of low-birth-weight infants, zinc supplementation improved motor development in infants in Canada (10), but had no effect on the development of infants in Brazil (11). In the latter study, however, some improvement in behavior was found.
Six other studies investigated the effect of zinc in school-aged children, with inconsistent results. In stunted children in Guatemala (12) and Canada (13), zinc supplementation had no effect on cognition. In urban and rural Chinese children (14, 15), zinc combined with micronutrient supplementation benefited a range of neuropsychological functions more than did treatment with micronutrients alone. The zinc-alone group showed some benefits compared with the micronutrients-alone group. Reasoning and recognition memory benefited in both urban and rural children. A preliminary report from the United States showed that Mexican American children benefited from zinc and micronutrient supplementation on a test of reasoning (15).
In 2 other studies, the behavior of young children was observed in randomized controlled trials of zinc supplementation. Supplemented children in India showed higher activity levels than did the control group (16). In Guatemala, supplemented stunted children sat and played more and cried less than did the placebo group (17), but did not differ in motor development. Supplementation of pregnant women produced changes in patterns of heart rate and activity of the fetus (18). In summary, there is limited but inconsistent evidence that zinc affects children's development.
In 1997 a randomized controlled trial was conducted with Bangladeshi infants to investigate the effects of zinc supplementation from 1 to 6 mo of age on infants' growth and morbidity. We took the opportunity presented by this study to determine the effect of zinc supplementation on the infants' mental development and behavior at 7 and 13 mo of age.
SUBJECTS AND METHODS
Subjects
In the original study, a house-to-house survey was conducted in 3 slum areas of Dhaka in which all infants <4 wk of age were identified. Three hundred one infants were enrolled at 4 wk of age and were randomly assigned to a treatment or placebo group. The treatment comprised 5 mg elemental Zn/d [the recommended dietary allowance (19)] as zinc acetate in a syrup and the placebo was a cellulose substance given in an identical syrup. Every week, health workers provided a week's supply of the syrup to the homes and quantified what had been consumed the week before. An average of 85% of the zinc and placebo mixtures were given, a similar amount in both groups. The health workers also recorded information on any morbidity every week and obtained anthropometric measurements of the children monthly. The original study was completed when the infants reached 6 mo of age (n = 272); the results of that study will be reported elsewhere. In the present study, the infants' psychomotor development and behavior were assessed at 7 and 13 mo of age.
The study was reviewed and approved by the research and ethical review committees of the International Center for Diarrhoeal Disease Research, Bangladesh. Prior informed consent was obtained from the children's parents.
Measurements
Developmental assessment
The infants' development was assessed by using the Bayley Scales of Infant Development (20), which includes 2 subscales: the mental development index (MDI) and the psychomotor development index (PDI). The Bayley Scales are the most commonly used infant development test. The children were brought to the International Center for Diarrhoeal Disease Research, Bangladesh, where they were tested in a quiet room in the presence of their mothers. Any sick child was treated and tested after he or she recovered. Two testers who were unaware of the children's group assignment carried out the tests at 7 mo of age, whereas only one tester assessed all the children at 13 mo. Before the study began, interobserver agreement for the MDI and PDI between each of the testers and the trainer was high (n = 21; r = 0.98). During the study, 10% of the tests were observed by the trainer and the level of reliability was similar.
Behavior rating
We rated the children's behavior during the tests on five 9-point scales that were modified versions of the scales developed by Wolke et al (21). These ratings included infants' activity (very still = 1, overactive = 9), emotional tone (unhappy = 1, radiates happiness = 9), approach or responsiveness to the examiner in the first 10 min (avoiding = 1, friendly and inviting = 9), cooperation with the test procedure (resists all suggestions = 1, always complies = 9), and vocalization (very quiet = 1, constant vocalization = 9). During the study, 10% of the tests were observed by the trainer and the interobserver correlations for all the scales with each of the testers were >0.90, except for the activity scale for which r = 0.82 for one of the testers.
Home stimulation
The Caldwell Inventory of Home Observation for Measurement of Environment (22) was modified for Bangladesh and completed by a female interviewer at each mother's home in the presence of her infant. The inventory comprised questions and observation items. Before and during the study, 20 interviews were observed by the trainer and scored independently and the interobserver correlation was high (r = 0.98).
Socioeconomic status
As a measure of wealth, the family's possessions were observed (eg, beds, other furniture, radio, cassette player, television, refrigerator, cabinet, and fan) and an asset score was calculated. Parental education and occupation were also recorded.
Anthropometry
The children's weight and length were measured by using standard techniques (23). Weight-for-age and length-for-age z scores were calculated according to the National Center for Health Statistics reference data for age and sex (24).
Statistical analyses
Independent-sample t tests or chi-square tests were used for comparisons between the treatment and placebo groups. Bivariate correlations between the outcome variables and the socioeconomic variables, age, and sex were calculated. To examine the treatment effect, multiple linear regression analyses of each of the behavioral variables were computed, with control for sex and age at testing. Any variable that was measured at enrollment and was correlated with an outcome variable was also entered stepwise in the regression. Because there were 2 testers at 7 mo, tester was controlled for in the relevant analyses. Analyses were performed with SPSS (version 9.0 for WINDOWS; SPSS Inc, Chicago).
RESULTS
Loss from the study
Of the 272 infants who completed the original study at 6 mo of age, 213 (104 in the zinc group and 109 in the placebo group) were tested at 7 mo; 28 children in the zinc group and 31 in the placebo group were lost to follow-up. Of those lost to follow-up, 1 died, the mothers of 2 refused to come for further assessment, and 31 were not located in time to be tested at 7 mo or had an acute illness at that age. The families of another 25 infants had moved and could not be located. At 13 mo, 14 more infants were lost to follow-up: 8 died (4 from each group), the mothers of 2 refused to come for further assessment, and 4 could not be traced. The data from one of the tested infants were excluded from the analyses because the infant had a history of seizures and very low scores (<50) on the MDI and PDI. We compared the enrollment data of all children lost since enrollment with those of the retrieved sample to determine whether the loss may have biased the findings. There were no significant differences between the lost and the retrieved sample in any of the variables examined (parental education, assets, and weight and length at 1 mo of age). The difference in sex approached significance at both 7 and 13 mo, with a higher proportion of boys lost to follow-up in the zinc group than in the placebo group (P < 0.1). We therefore controlled for sex in all analyses.
Characteristics of the tested sample
There were no significant differences between the groups in anthropometric measurements, household assets, parental education, or scores on the Home Observation for Measurement of Environment (Table 1). The mean weight-for-age and length-for-age of the infants in both groups were approximately –1 z score. Most (95.6%) of the infants were breast-fed up to 6 mo; only 5.4% were exclusively breast-fed. The average age of introducing complementary foods was similar in both groups: 2.1 mo in the zinc group and 2.3 mo in the placebo group.
View this table:
TABLE 1. Characteristics of the infants at enrollment1
At 7 and 13 mo of age, there were no significant differences in anthropometric measurements between the groups (Table 2); in general, the children's nutritional status deteriorated throughout the study. There were also no significant differences in any of the behavioral and developmental variables at 7 mo between the groups; at 13 mo, however, the mean MDI score was significantly higher in the placebo group than in the zinc group. The test-retest correlations between the 7-mo and 13-mo tests were as follows: placebo group, r = 0.23 (MDI), r = 0.40 (PDI); zinc group, r = 0.46 (MDI), r = 0.40 (PDI). None of the r values were significantly different (Fisher's z transformation).
View this table:
TABLE 2. Infants' nutritional status, test scores on the Bayley Scales of Infant Development (20), and behavior ratings at 7 and 13 mo of age by treatment group1
Associations with developmental outcomes at 7 and 13 mo
At 7 mo the MDI was significantly correlated with age at testing (r = –0.22, P < 0.01), mother's education (r = 0.17, P < 0.05), and weight-for-age at enrollment (r = 0.16, P < 0.05); the PDI was also significantly related to age at testing (r = –0.16, P < 0.05) and weight-for-age at enrollment (r = 0.23, P < 0.005). At 13 mo the MDI was significantly correlated with age at testing (r = –0.16, P < 0.05), father's education (r = 0.14, P < 0.05), and weight-for-age at enrollment (r = 0.21, P < 0.005); the PDI was significantly associated with weight-for-age at enrollment (r = 0.26, P < 0.005).
Treatment effect
To examine the treatment effect while taking into account other variables, we carried out a series of multiple regressions of the MDI, PDI, and behavior ratings at 7 and 13 mo. Because of the tendency for a sex difference between the children lost to follow-up and those tested and because of the significant correlation with age at testing of both the MDI and PDI, we entered sex and age in the first step of all regression analyses. Weight-for-age z score at enrollment and mothers' and fathers' education were also significantly correlated with the outcome variables and were therefore entered in the second step. Treatment (coded as zinc = 1, placebo = 2) was entered in a third step. The regression coefficients and P values for all significant independent variables in the analyses of the MDI and PDI at 7 and 13 mo are shown in Table 3. The age of the infants at the time of testing significantly affected both the MDI and PDI. Girls had lower scores on the MDI than did boys at both ages and the enrollment weight-for-age z score was significant in all regressions. When all these variables were controlled for, there was an independent effect of zinc treatment on the children's MDI at 13 mo, with the treated group having lower scores than the placebo group.
View this table:
TABLE 3. Regression coefficients (SEs) of the treatment effect and other significant covariates from multiple regression analyses of the mental development index (MDI) and psychomotor development index (PDI) of the Bayley Scales of Infant Development (20) at 7 and 13 mo of age
Many of the children were undernourished at 13 mo and weight-for-age was correlated with their development. To examine the treatment effect while controlling for concurrent weight-for-age, we conducted a second series of multiple regressions in which we added concurrent weight-for-age z score to the second step, before entering the treatment variable. Concurrent weight-for-age z score was highly significant in regressions of the MDI and PDI at both ages, and once the current weight-for-age z score was entered into the model, weight-for-age z score at enrollment was no longer significant. The treatment effect for the MDI at 13 mo of age was slightly larger and more significant when current weight was controlled for (ß = 3.7, SE = 1.3, P < 0.005). We repeated similar models with all of the behavior ratings, but no significant treatment effects were observed.
DISCUSSION
Our finding that zinc-treated infants had slightly but significantly lower MDI scores than did the placebo group is the opposite of our hypothesis; we were expecting beneficial effects. The study was reasonably rigorous in that it was a randomized, double-blind treatment trial. In addition, we controlled for any differences found between the infants lost to follow-up and those tested so that loss from the study would not bias the results. The findings were not confounded by the socioeconomic conditions of the families because we controlled for any difference in socioeconomic status between the groups.
Although the Bayley Scales of Infant Development were standardized in the United States, they have been used in many other cultures, often in research on nutrition and development. Most researchers use control groups within the same population. It is possible that slight differences in administration of the test occur in different cultures. Additionally, children in different cultures are exposed to very different environments, which would be expected to affect their development. Therefore, it is difficult to interpret the differences in children's scores across cultures (25). Both the MDI and PDI scores of the infants in the present study were well within the normal range of the test standards. At 12 mo of age, the infants' PDI scores were slightly lower than the US norms, whereas their MDI scores were slightly higher than the US norms. In a previous study of poor infants in Brazil, higher MDI scores were found (26). Infants in traditional but poor societies often have high scores on developmental tests in the first year (27, 28). These high scores may result from a supportive environment in which many adults interact with the infant, the infant in frequently carried, and few physical restrictions, such as cribs and playpens, are placed on the infant. Scores often decline in the second year. It appears that the motor scores of the infants in this study had already declined by 12 mo of age, probably because of their poor nutritional status.
The scores were associated with maternal education and the infants' weight-for-age in a logical way and interobserver reliability was good. It should therefore be valid to compare the scores of groups of children within the same population.
The effect of zinc was not apparent at 7 mo but became obvious at 13 mo of age. A lag between treatment and developmental response was reported previously; for example, in a recent study, iron-fortified formula was found to benefit children's development 6 mo after the cessation of treatment (29).
In the present study, zinc was given alone, in between meals without other food or micronutrients, for 5 mo. In contrast, in 5 Bangladeshi studies that reported beneficial effects on growth and morbidity (30–34), children were treated for 2 wk and the zinc was administered with additional food and multivitamins. It is possible that the lower scores in the zinc-treated group in the present study were caused by an imbalance of micronutrients. In one Bangladeshi study (35), high doses of zinc (6 mg•kg-1•d-1) given to severely malnourished children were associated with increased mortality. The authors suggested that apart from having a direct detrimental effect on the immune system during sepsis, high-dose zinc supplementation may have aggravated deficiencies of other minerals. The competitive interactions between zinc and copper and between zinc and iron may further limit the availability of these trace metals (36, 37).
In one study, zinc supplementation was associated with low concentrations of plasma copper (38). Copper is important in myelination of the central nervous system and in catecholamine and cytochrome-c oxidase metabolism (39). Although iron deficiency is associated with poorer levels of development in children (40), we found no data on the effect of copper deficiency on children's development. Measures of iron and copper status would have been helpful in the present study; unfortunately, we had none.
The children in the present study were generally short at enrollment (mean length-for-age: –1.1 z score) and although they were not weighed at birth, many of them undoubtedly had low birth weights. In a concurrent study of pregnant women in the same area, 43% of infants had low birth weights; of these, 12% were delivered preterm and 31% at term (41). These characteristics suggest a risk of zinc deficiency; however, the children did not exhibit a reduction in morbidity or a growth spurt in response to zinc treatment, indicating that zinc was not growth limiting. The children were given 5 mg/d, which is the recommended dietary allowance, but most of the infants were partially breast-fed, which should facilitate zinc bioavailability. We cannot calculate the total intake of zinc because we lack details on the precise quantity of breast-feeding, but it is clear that these infants received more than the recommended dietary allowance.
Many of the children in the present study were moderately undernourished by 13 mo. At enrollment their mean weight-for-age and height-for-age were approximately -1 z score and declined one-half a z score during supplementation from 1 to 6 mo of age. Such children are probably deficient in energy and possibly protein and other micronutrients (42). The children were also very young and at this age are likely to be more vulnerable to biological insults than are older children (43). It is possible that giving only zinc in these circumstances will have unforeseen consequences.
No previous studies have reported any adverse effects of zinc supplementation on the mental development or behavior of children. Four studies assessed the effect of zinc supplementation on the development or behavior of children aged <2 y. In one study, iron-fortified formula and copper were also given (10); in another, multivitamins were given and the children were older than 12 mo (16). In the other 2 studies, zinc was given alone. In one of these, 5 mg Zn was given daily, but for only 2 mo (11). In the other study (17), 10 mg Zn was given daily, but the infants were slightly older and had a much higher incidence of diarrhea (44) and were thus more likely to be zinc deficient.
The difference in mental development scores between the zinc and placebo groups in the present study was small. However, in a preliminary analysis of data from a companion study of zinc supplementation of pregnant Bangladeshi women, we found a slightly larger difference in the same measure of mental development in the women's offspring (Hamadani et al, unpublished observations, 2000). The findings from the present study suggest a possible risk of giving zinc alone for a prolonged period; however, the results may be peculiar to this population. It is clear that the children in the present study were undernourished and required more than just zinc. In this respect, the association between weight-for-age and development is particularly worrying. There is an obvious need for more studies of the effect on children's development of giving zinc combined with other micronutrients.
ACKNOWLEDGMENTS
We are grateful to the children and their parents who participated in the study and thank Mrs Saadia, Sultana Parveen, Shireen Ali, MAR Patwary, and KM Rafique for their assistance.
REFERENCES