QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Black, M. M. Articles by Krishnakumar, A. Search for Related Content PubMed PubMed Citation Articles by Black, M. M. Articles by Krishnakumar, A.

---

Supplement

Predicting Longitudinal Growth Curves of Height and Weight Using Ecological Factors for Children with and without Early Growth Deficiency

Department of Pediatrics, University of Maryland School of Medicine, Baltimore, MD 21201

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Growth curve models were used to examine the effect of genetic and ecological factors on changes in height and weight of 225 children from low income, urban families who were assessed up to eight times in the first 6 y of life. Children with early growth deficiency [failure to thrive (FTT)] (n = 127) and a community sample of children without growth deficiency (n = 98) were examined to evaluate how genetic, child and family characteristics influenced growth. Children of taller and heavier parents, who were recruited at younger ages and did not have a history of growth deficiency, had accelerated growth from recruitment through age 6 y. In addition, increases in height were associated with better health, less difficult temperament, nurturant mothers and female gender; increases in weight were associated with better health. Children with a history of growth deficiency demonstrated slower rates of growth than children in the community group without a history of growth deficiency. In the community group, changes in children's height and weight were related to maternal perceptions of health and temperament and maternal nurturance during feeding, whereas in the FTT group, maternal perceptions and behavior were not in synchrony with children's growth. These findings suggest that, in addition to genetic factors, growth is dependent on a nurturant and sensitive caregiving system. Interventions to promote growth should consider child and family characteristics, including maternal perceptions of children's health and temperament and maternal mealtime behavior.

KEY WORDS: • hierarchical linear models • height • weight • growth deficiency • failure to thrive

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Growth deficiency, or failure to thrive (FTT),⁴ affects millions of children worldwide (UNICEF 1995 ). Most long-term growth studies are from developing countries where early growth deficiency is often severe, malnutrition is endemic and poverty is chronic (Allen 1994 ). Little is known about the growth patterns of children raised in industrialized countries with mild forms of early growth deficiency.

Growth is a complex process that varies across individuals and may be influenced by multiple interacting factors. Although genetic factors such as parental height and weight form the blueprint for children's growth (Himes et al. 1985 ), child factors such as gender, age, early growth, health or temperament may be particularly important. For example, children with difficult temperaments may be more reactive to challenges in the household than children with easy temperaments and thus more difficult to feed (Black et al. 1996 ). Family factors, including maternal depression and passive or unresponsive interactions between mother and child, have also been associated with growth deficiency (Black et al. 1994 , Drotar et al. 1990 , Drotar 1991 ). Unfortunately, little is known about the relationship between ecological factors and growth. The objectives of this investigation were to examine ecological factors associated with longitudinal changes in weight and height among low income children with and without early growth deficiency.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Participants.

Two groups of children (FTT and community) were recruited from inner-city pediatric clinics serving low income families. Recruitment criteria included chronological age <25 mo, full-term birth, birth weight appropriate for gestational age and no congenital problems or handicapping conditions. Children in the FTT group had weight-for-age or weight-for-height below the 5th percentile, using the National Center for Health Statistics (NCHS) growth charts (Hamill et al. 1979 ) and no identified medical problems, as determined by medical record review and physical examination. Children in the community group had weight-for-age and weight-for-height above the 10th percentile.

The sample included 127 children in the FTT group and 98 in the community group. The majority of the families were African American (92%), with few financial resources (Aid to Families with Dependent Children 76%), headed by single women (89%) who had not completed high school (58%).

Evaluation schedule. Families were invited to participate in a longitudinal research project, using consent procedures approved by the Institutional Review Board at the University of Maryland at Baltimore. A baseline evaluation included growth measurements, a video-taped observation of the parent and child eating lunch and a maternal interview. Children with growth deficiency were enrolled in a multidisciplinary Growth and Nutrition Clinic for 2 y where they received growth monitoring, nutrition counseling and interaction coaching regarding mealtime behavior (Black 1995 ). The evaluation was repeated three times at 6-mo intervals and then annually at ages 3, 4, 5 and 6 y (8 evaluations). Children in the community group received no intervention and were evaluated at baseline, and ages 3, 4, 5 and 6 y (5 evaluations). Families were compensated financially at each evaluation.

Growth. Children were weighed by a study nurse using Seca Scales that were calibrated regularly. Weights and heights were plotted using NCHS growth charts and verified by a nutritionist. Stunting and wasting were defined as 2 SD below the median for height-for-age and weight-for-height. Growth was analyzed using raw height and weight assessed in centimeters and kilograms.

Parental height and weight. Biological parents were weighed and measured at each visit. For parental height, we used the mean height of both parents; for weight, we used maternal weight averaged over visits.

Child health. Children's health was assessed by the Rand Health Questionnaire (Eisen et al. 1980 ) at baseline and ages 4, 5 and 6 y, and averaged over multiple administrations.

Children's temperament. Children's temperament was assessed at baseline by the Fussy-Difficult factor of the Infant Characteristics Questionnaire (Bates et al. 1979 ).

Maternal depression. Maternal depression was measured at baseline using the depression subscale of the Brief Symptom Index (Derogatis and Spencer 1982 ).

Maternal nurturance. Maternal nurturance during feeding was measured by a modified version of the Parent Child Early Relational Assessment (Black et al. 1996 , Clark 1985 ) that includes six items (growth fostering, enthusiasm, social initiative, child-oriented language, amount of verbalization and involvement with child) with a coefficient > 0.84. The video tapes were scored by two raters who were unaware of the children's growth history. Raters were trained until they reached >90% agreement; interrater reliability was reviewed through weekly checks.

Socioeconomic status. The three socioeconomic indicators were maternal education, maternal age at delivery and household density (rooms divided by household members).

Data analytic strategy.

We used hierarchial linear modeling (HLM) to examine the effects of child and family variables on changes in children's growth from baseline through age 6 y. HLM is a two-stage procedure for estimating intraindividual change and correlates of change (Bryk and Raudenbush 1992 , Burchinal et al. 1994 and 1997 ). In the Level 1 model (within-subject), growth curves were formulated for each child as a function of time-varying covariates (e.g., age), represented by the model Y_ti = ß_0i + ß_1ia_ti (linear) + ß_2ia²_ti(quadratic) + e_ti where Y_ti is the outcome variable for child imeasured at time t; ß_0i is the mean outcome for child i; ß_1ia_ti is the average linear growth rate in Y for child i over time and a_ti is the age of child i at time t; ß_2i is the rate of acceleration in Y; and e is the random error term.

In the Level 2 model (between-subjects), time-invariant covariates are used to predict the values of ß from Level 1. The Level 2 model for the intercept is as follows: ß_0i = _0i + ₀₁ (predictor 1)i + ₀₂ (predictor 2)i + ... . + e_0iwhere _0i is the expected intercept for child with values of zero for predictors 1 and 2;_{0 i} and _0iare regression coefficients associated with predictors 1 and 2; and e_0i is the unique random effect associated with child i. The Level 2 model for the linear slope is as follows: ß_1i = ₁₀ + ₁₁ (predictor 1)i+ ₁₂ (predictor 2)i + ... . + e_1i where ₁₀is the expected slope for child i with values of zero for predictors 1 and 2; ₁₁ and₁₂ are regression coefficients associated with predictors 1 and 2 and e_1i is the unique random effect associated with child i.

The dependent variables were height and weight collected at eight age points. We examined these outcomes within three additive models. The first model identified the effect of genetic factors (parental height and maternal weight) and children's age at recruitment. The second model examined any additional variance accounted for by child factors such as health, temperament, gender (girls = 1 and boys = 0) and growth history (FTT = 1 and community = 0). In the third additive model, we examined the role of maternal nurturance, depression and sociodemographic characteristics. To explore whether the relationship between growth and the child and family variables differed by the children's growth history, we included interaction terms (e.g., group x child health) in the second and third additive models.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Children in the FTT and community groups did not differ on most sociodemographic characteristics (see Table 1 ).However, parents of children in the community group were taller and mothers had more years of schooling. Mothers in the FTT group were less nurturant during feeding. They perceived their children in better health and temperamentally less difficult than community group mothers. At baseline and at each subsequent time point, there were significant differences in height and weight between children with and without an early history of growth deficiency (see Figs.1 and 2).Although children in the FTT group continued to lag behind children in the community group in weight and height at age 6 y, the number of children who were stunted and/or wasted at baseline (Z-scores below -2 SD) was reduced by age 6 y (from 18 to 3% and from 31 to 3%, respectively).

View larger version (11K): [in this window] [in a new window]   Figure 1. Height of children with and without early growth deficiency. Median = 50th percentile, based on NCHS standards.

As an initial step, a random effects HLM model, with height as the outcome and no other predictors was run. The deviance statistic from this run was used to compare against a second model in which age and weight were entered as random effect predictors of height. The results indicated an improved fit [²(7) = 3910.87, P < 0.001]. The linear parameter of age fit the data better than other polynomial models.

Child weight was included as a random level predictor of height because it correlated with height and differed for children in the FTT and community groups. Entering weight as a predictor of height in Level 1 and as a predictor of the intercept in Level 2 should eliminate spurious relationships confounded by weight (Bryk and Raudenbush 1992 ). The test for homogeneity of Level 1 variance was not significant, suggesting that the model was appropriately specified (Bryk and Raudenbush 1992 ).

Results from the final model are shown in Table 2 .At baseline, taller children were older, with taller and older mothers, no history of growth deficiency, mothers who were more nurturant during feeding and perceived their children as healthier and temperamentally less difficult. In addition, there were significant interactions between group (FTT or community) and child health, temperament and maternal nurturance. For the community group, the effect of child health on height was stronger than in the FTT group (t = -2.24, P < 0.05; unstandardized ß for community group: 5.25 + 0 [-4.05] = 5.25; unstandardized ß for FTT group: 5.25 + 1 [-4.05] = 1.2). Similarly, the effects of maternal nurturance and child difficult temperament were stronger in the community group than in the FTT group (maternal nurturance: t = -2.26, P < 0.05; ß = 1.52 and ß = -0.08; child difficult temperament: t = -3.46, P < 0.001; ß = -2.34 and ß = 0.45).

Prediction of growth in weight.

Using a similar analytic strategy, the Level 1 model with age and height had a good fit [² (7) = 3086.04, P < 0.001] and was homogenous.

In Level 2, the genetic and child characteristics models indicated a good fit. Family characteristics did not improve the fit. At baseline, heavier children had parents who were taller and heavier. In addition, heavier children were older, with better health, less difficult temperament and no growth deficiency. There were significant interactions between children's growth history and measures of their health and temperament. The associations were stronger for children in the community group than in the FTT group (child health: t = -2.32, P < 0.05; ß = 1.72 and ß = 0.05; child difficult temperament: t =2.41, P < 0.05; ß = -0.62 and ß = 0.16).

The rate of change in weight was stronger for children who were younger at recruitment, with heavier mothers, no history of growth deficiency and better health. The interaction between maternal perception of child health and growth history was significant, indicating that the effects of child health were stronger for the community group than the FTT group (t = -2.97, P < 0.01; ß = 0.05 and ß = 0.00).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Children with an early history of growth deficiency demonstrated catch-up growth during their preschool years; by age 6 y, only 3% were stunted or wasted. However, they gained both height and weight at a slower rate than their community peers, suggesting that growth deficiency during infancy and toddlerhood increased the likelihood of slower growth in both height and weight through early childhood.

As expected, genetic factors were important contributors to children's growth. Children of tall parents were taller at baseline and experienced a faster rate of increase in height over time. Similarly, children of mothers who were heavier were heavier at baseline and gained weight at a faster rate. However, beyond the influence of parental size, both child and family factors played important roles.

One of the most striking findings involved the differences in determinants of growth between the community and FTT groups. In the community group, the patterns linking children's growth with maternal perceptions of children's health and temperament and maternal behavior during feeding were consistent with general theories of child behavior and development (Bronfenbrenner 1993 ) and have been described previously (Black et al. 1996 , Engle et al. 1996 ). Mothers who reported that their children were healthy had children who were taller and heavier at baseline and gained weight faster through their preschool years. Similarly, children of mothers who were nurturant and responsive during feeding were taller at baseline and demonstrated accelerated growth in height. Growth may have provided cues that mothers used to reassess their perceptions of their child's health, and presumably their behavior toward their child. Thus, healthy growth may be dependent on a nurturant caregiving system based on accurate perceptions of children's health and temperament.

In contrast, the weak and sometimes negative associations in the FTT group between children's growth and maternal perceptions of health and temperament and maternal behavior, provide evidence that maternal perceptions of children's health and temperament may be distorted among children with growth deficiency. For example, at baseline when there were clear group differences in height and weight, mothers in the FTT group reported better health ratings and less difficult temperamental ratings than mothers in the community group. Previous investigators have described distortions in parental perceptions and observations of parent-child interactions among parents of children with FTT, including denial of growth problems (Ayoub and Milner 1985 ), low maternal responsivity and insensitivity to cues (Drotar et al. 1990 , Hutcheson et al. 1993 , Wolke et al. 1990 ). If parents in the FTT group misperceived their children's health and temperamental status and were unable to respond to the children's cues or to provide nurturant caregiving, especially during feeding, their household environment may have been less conducive to optimal growth. Thus, caregiving differences may explain the differential growth between the FTT and community groups.

Other child and family characteristics also contributed to children's growth, including child's age at recruitment, child's gender and maternal age. At baseline, older children were taller and heavier, as expected. However, when rate of change was considered, younger children gained in height and weight at a faster rate. This finding has important implications for the development of future intervention programs because research on children's potential for catch-up linear growth beyond the early years has been inconclusive (Martorell et al. 1994 ). Although supplementation programs have often targeted preschool or school-age children, infancy and toddlerhood may be a critical time in which to intervene to promote growth.

Although there were no gender differences at baseline, girls gained height, but not weight, at a faster rate. The data offer no explanation, but these findings are similar to reports from Peru in which previously malnourished females grew faster than males (Graham et al. 1982 ).

Maternal age was a protective factor for baseline status, but not for changes in height and weight. Mothers who were older at delivery were more likely to have children who were taller and heavier. These findings are consistent with reports of increased behavioral and developmental problems among children of adolescent mothers, often related to disturbances in caregiving (Miller and Moore 1990 ).

Household density and maternal education served as indices of poverty, but they were not related to baseline status or to growth after more proximal variables were considered. Most families in this study were living in low income, urban communities, often supported by public assistance; thus there was little variability in indices of poverty.

	CONCLUSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
These findings have important implications for theory and practice. Growth should be conceptualized from an ecological perspective, and future investigations of growth should include measures of child and family characteristics, along with genetic factors. In addition, intervention programs for growth-deficient children should pay close attention to the caregiving environment. Intervention strategies are likely to be most successful when caregivers do not deny the child's growth deficiency, have accurate perceptions of the child's health and temperament, and are nurturant and responsive to the child.

View larger version (11K): [in this window] [in a new window]   Figure 2. Weight of children with and without early growth deficiency. Median = 50th percentile, based on NCHS standards.

	FOOTNOTES

¹ Presented at the symposium "Causes and Etiology of Stunting" as part of Experimental Biology 98, April 18–22, 1998, San Francisco, CA. The symposium was sponsored by the American Society for Nutritional Sciences and the Society for International Nutrition Research. Published as a supplement to The Journal of Nutrition. Guest editor for the symposium publication was Edward A. Frongillo, Jr., Cornell University, Ithaca, NY.

² Support for this research was provided by grants MCJ-240568 and MCJ-240621 from the Maternal and Child Health Program (Title V, Social Security Act) and grant 90CA1401 from the National Center on Child Abuse and Neglect, Health Resources and Services Administration, Department of Health and Human Services.

³ Abbreviations used: FTT, failure to thrive; HLM, hierarchical linear modeling; NCHS, National Center for Health Statistics.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 

1. Allen L. H.. Nutritional influences on linear growtha general review. Eur. J. Clin. Nutr. 1994;48:S75-S89.

2. Ayoub C., Milner J.. Failure to thriveparental indicators, types, and outcomes. Child Abuse Neglect 1985;9:491-499.[Medline]

3. Bates J. E., Freeland C., Lounsbury M.L.. Measurement of infant difficultness. Child Dev 1979;50:794-803.[Medline]

4. Black M. M.. Failure to thrivestrategies for evaluation and intervention. School Psychol. Rev. 1995;24:171-185.

5. Black M., Hutcheson J., Dubowitz H., Berenson-Howard J.. Parenting style and developmental status among children with nonorganic failure-to-thrive. J. Pediatr. Psychol. 1994;19:689-708.[Abstract/Free Full Text]

6. Black M., Hutcheson J., Dubowitz H., Starr R. J., Jr, Berenson-Howard J.. The roots of competencemother-child interaction among low income, urban, African American families. J. Appl. Dev. Psychol. 1996;17:367-392.

7. Bronfenbrenner U.. Ecological systems theory. Wozniak R. Fisher K. eds. Specific EnvironmentsThinking in Contexts 1993:3-44 Erlbaum Hillsdale, NJ.. .

8. Bryk A. S. Raudenbush S. W. eds. Hierarchial Linear Models 1992 Sage Newbury Park, CA.. .

9. Burchinal M. R., Bailey D. B., Snyder P.. Using growth curve analysis to evaluate child changes in longitudinal investigations. J. Early Intervention 1994;18:422-442.

10. Burchinal M. R., Campbell F. A., Bryant D. M., Wasik B. H., Ramey C. T.. Early intervention and mediating processes in cognitive performance of children of low-income African-American families. Child Dev 1997;68:935-954.

11. Clark, R. (1985) The Parent-Child Early Relational Scale. Unpublished document..

12. Derogatis, L.R. & Spencer, R., eds. (1982) The Brief Symptom Inventory (BSI): Administration, Scoring, and Procedure Manual. Clinical Psychometric Research, Baltimore, MD. .

13. Drotar D.. The family context of nonorganic failure to thrive. Am. J. Orthopsychiatry 1991;61:23-34.[Medline]

14. Drotar D., Eckerle D., Satola J., Pallotta J., Wyatt B.. Maternal interactional behavior with nonorganic failure to thrive infantsa case comparison study. Child Abuse Neglect 1990;14:41-51.[Medline]

15. Eisen M., Donald C. A., Ware J. E., Jr, Brook R. H.. Conceptualizations and Measurement of Health for Children in the Health Insurance Study 1980 Rand Corporation Santa Monica, CA.. .

16. Engle P. L., Castle S., Menon P.. Child developmentvulnerability and resilience. Soc. Sci. Med. 1996;43:621-635.

17. Graham G. G., Adrianzen B. T., Rabold J., Mellits E. D.. Later growth of malnourished infants and children. Am. J. Dis. Child. 1982;136:348-352.[Abstract]

18. Hamill P.V.V., Drizd T. A., Johnson C. L., Reed R. B., Roche A. F., Moore W. M.. Physical growthNational Center for Health Statistics percentiles. Am. J. Clin. Nutr. 1979;32:607-629.[Abstract/Free Full Text]

19. Himes J. H., Roche A. F., Thissen D., Moore W. N.. Parent-specific adjustments for evaluation of recumbent length and stature of children. Pediatrics 1985;75:304-313.[Abstract/Free Full Text]

20. Hutcheson J., Black M., Starr R.. Developmental changes in interactional characteristics of mothers and their children with failure to thrive. J. Pediatr. Psychol. 1993;18:453-466.[Abstract/Free Full Text]

21. Martorell R., Khan L. K., Schroeder D. G.. Reversibility of stuntingepidemiological findings in children from developing countries. Eur. J. Clin. Nutr. 1994;48:S45-S57.

22. Miller B., Moore K.. Adolescent sexual behavior, pregnancy and parentingresearch through the 1980s. J. Marriage Fam. 1990;52:1025-1044.

23. . UNICEF. The State of the World's Children 1995 1995 Oxford University Press New York, NY.. .

24. Wolke D., Skuse D., Mathiesen B.. Behavioral style in failure to thrive infantsa preliminary communication. J. Pediatr. Psychol. 1990;15:237-243.[Abstract/Free Full Text]

This article has been cited by other articles:

				M C J Rudolf and S Logan What is the long term outcome for children who fail to thrive? A systematic review Arch. Dis. Child., September 1, 2005; 90(9): 925 - 931. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Black, M. M. Articles by Krishnakumar, A. Search for Related Content PubMed PubMed Citation Articles by Black, M. M. Articles by Krishnakumar, A.