Journal of Pediatric Nursing
Volume 18, Issue 1 , Pages 52-59, February 2003

Selection of growth parameters to define failure to thrive☆

Colleges of Nursing and Medicine, The Ohio State University, Columbus, OH.

Article Outline

Abstract 

Failure to thrive (FTT) is a syndrome of growth failure due to undernutrition. Determining whether an infant has FTT is based on the use of an anthropometric indicator and a selected cutoff value for that indicator. These anthropometric indicators include weight for age, weight for length, and length for age, and the cutoff values include the 10th, 5th, and 3rd percentiles. Each indicator and selected cutoff value provide unique information about an infant's growth. However, these parameters are often used interchangeably to explain the same growth phenomenon. The sensitivity and specificity of each anthropometric indicator are a function of the cutoff value selected and dictate which infants will be classified as having FTT and which infants will be classified as healthy. Depending on the sensitivity and specificity of the indicator, some infants with FTT will be classified as healthy, and some healthy infants will be classified as having FTT. A clear rationale for the selection of an anthropometric indicator and a cutoff value for defining FTT are important for increasing the generalizability of research findings and thereby expanding the current knowledge base related to FTT. Copyright 2003, Elsevier Science (USA). All rights reserved.

 

During infancy, growth is primarily a nutrition-dependent process (Karlberg, Jalil, Lam, Low, & Yeung, 1994). Age-appropriate growth of an infant reflects the adequacy of the infant's nutritional intake (Bessler, 1999). Therefore the attained body size of an infant can serve as an indirect marker of current nutritional status, past nutritional history, or both (Berhane & Dietz, 1999). Anthropometric indicators (weight for age, length for age, or weight for length) are the most practical method for the quantitative measurement of infant nutritional status (Gorstein et al., 1994; WHO Working Group, 1986). The value of a selected anthropometric indicator in relation to a predetermined cut-off value (such as 10th, 5th, or 3rd percentile) provides information about an infant's growth and nutritional status, in relation to a healthy, well-nourished reference population (Chumlea & Guo, 1999; Sherry, 1999).

Failure to thrive (FTT) in infants occurs when an infant's weight gain deviates from an established pattern of growth and is directly attributed to undernutrition. Undernutrition occurs for a variety of nonorganic and organic reasons, including the inability of the infant to ingest the proper amount of nutrients, inadequate nutrition provided to the infant, incomplete absorption of nutrients, and/or increased calorie requirements (Metallinos-Katsaras & Gorman, 1999; Zenel, 1997). Ultimately, it is the interactive effects of multiple factors on the infant's nutritional intake that result in undernutrition (Bithoney, Dubowitz, & Egan, 1992; Kessler, 1999). Defining FTT, for clinical or research purposes, is typically based on the use of an anthropometric indicator and specified cutoff value for that indicator. When the infant's attained growth falls below the cutoff value for a specified anthropometric indicator, the infant is considered to be undernourished and not growing normally and is thus classified as having FTT (Frank, Silva, & Needlman, 1993; Peterson, 1993; Peterson & Chen, 1990). However, the determination that an infant has FTT cannot be made on the basis of 1 isolated measurement of an infant's weight and/or length. Rather, it is based on the assessment of serial measures of weight and/or length that deviate from previously established growth channels (Berhane & Dietz, 1999; Corrales & Utter, 1999).

While there is no disagreement that inadequate growth is the hallmark of FTT, the use of different growth parameters to define FTT has been problematic in determining the presence of this condition. There is a lack of agreement in clinical and research literature about which anthropometric indicator and cut-off value should be used to define FTT (Raynor & Rudolf, 2000; Wilcox, Nieburg, & Miller, 1989; Wright, Ashenburg, & Whitaker, 1994). The use of various indicators and cut-off values to define FTT implies that they all represent the same phenomenon. The dilemma is that, while there are some relationships among the indicators, each indicator and selected cut-off value provides distinctly different information about an infant's nutritional status, growth, and growth failure (WHO Expert Committee, 1995). The lack of consistent usage of anthropometric indicators and cut-off values to define FTT limits identification of infants with FTT in clinical practice and limits the generalizability of research findings. This ultimately hinders our understanding of this syndrome. The purposes of this article are to describe which aspect of the growth process each anthropometric indicator represents and to describe how the selection of a cut-off value influences the sensitivity and specificity of the indicator.

Back to Article Outline

Anthropometric indicators 

Anthropometrics are the use of body measurements to assess an infant's growth and nutritional status (Chumlea & Guo, 1999; Gorstein et al., 1994). Thorough assessment of an infant's growth begins with the use of precise methods for accurately measuring the infant. When done correctly, anthropometric measurements should be reproducible (Pomerance, 1995). In relation to FTT, each indicator provides different information about the status of the infant's growth failure (WHO Working Group on Infant Growth, 1995) (see Table 1).

Table 1. Comparison of anthropometric indicators
Weight-for-AgeWeight-for-HeightHeight-for-Age
DefinitionTotal body mass in relation to chronological age*Total body mass in relation to height*Attained linear growth in relation to chronological age*
Relationship to nutritionIndicative of long-term nutritional history*Indicative of current nutritional statusCumulative effects of past nutritional history
Low score suggestsDecrease in total body mass†Decrease in soft and fat tissues‡Slowing in linear growth§
Problem with the measureA decrease cannot discriminate between acute and chronic undernutrition‡Height measurement must be availableProvides no information about current nutritional status
Usefulness in FTTProvides information about general changes in weight over time‡Indicator of choice because it is sensitive to current nutritional status∥Indicates that undernutrition has been occurring for a significant period of time*
*WHO (1995). †Keller (1982). ‡Gorstein et al (1994). §WHO (1986). ∥Sherry (1999).
A weight for age lower than the 5th percentile on the National Center for Health Statistics (NCHS) growth charts is the most common anthropometric indicator and cutoff value used to define FTT (Corrales & Utter, 1999).

Weight for age 

The weight of an infant is a composite measure of total body mass, which comprises fat mass, fat-free mass, and skeletal tissues (Chumlea & Guo, 1999; Gorstein, 1989). Weight for age is the most common anthropometric indicator used in the measurement of infant growth for research and clinical purposes. Although weight for age is a composite measure of total body mass, it actually gives no indication of the body's structure. This measure does not distinguish between infants who are long and thin from infants who are short and fat (Gorstein et al., 1994). A low weight for age indicates that the infant is either gaining insufficient weight relative to age or actually losing weight (WHO Expert Committee, 1995). Low weight for age may be caused by changes in soft tissues, skeletal growth, or both. Despite the predominant use of this indicator in clinical settings and research protocols, weight for age is not the best indicator of nutritional status. Because of its inability to discriminate between infants of various sizes, the use of weight for age to define FTT can potentially result in false-positive and false-negative errors in classifying an infant as having FTT or being healthy (Keller, 1982). A fall in weight for age should be interpreted cautiously. If an infant is underweight, further consideration should be given to whether this is the result of changes in soft tissue, skeletal growth, or both (Gorstein et al., 1994).

Weight for length 

The anthropometric indicator weight for length provides a more accurate measurement of infant growth and current nutritional status. This indicator separates out the contribution of bone mass to body weight and provides information about the amount of soft tissue and fat tissue expected in relation to an infant's length. For example, a weight for age at the 10th percentile provides no information about whether the infant appears underweight or overweight, but a weight for length at the 10th percentile provides a visual picture of an infant who is long and thin. In defining FTT, a deficit in weight for length is the result of acute undernutrition and is indicative of wasting (Gorstein et al., 1994; Keller, 1982). Similar to weight for age, a fall in weight for length is related to either an infant's inability to gain weight or actual loss of weight relative to length (Gorstein et al., 1994; WHO, 1986). Weight for length is better able to discriminate infants with FTT from healthy infants than weight for age (Trowbridge, 1979). If length measurements are available, weight for length should be the anthropometric indicator of choice for defining FTT. Every effort should be made to obtain a measure of length when the infant is weighed.

Length for age 

Skeletal or linear growth is a slower process than the growth of other tissues seen during the rapid growth of infancy. Length does not fluctuate with short-term undernutrition (Frank & Zeisel, 1988). A low length for age (≤10th percentile) may be indicative of genetic influences. If this is the case, the infant's growth trajectory will remain near the 10th percentile. In defining FTT, a decrease in length for age over time reflects a slowing in linear growth, as an infant cannot “lose” length (WHO Working Group, 1986). In the progression of FTT, a low length for age follows either a loss of weight or a slowing in weight gain that has been occurring for an extended period of time. It is assumed that when an infant experiences a slowing in linear growth, he or she is passing from acute undernutrition into a more chronic phase. A decrease in length for age is viewed as a physiologic adaptation to long-term undernutrition that preserves adequate weight for length at the expense of linear growth (Berhane & Dietz, 1999).

Summary 

Each anthropometric indicator is the result of different biologic processes and provides different information about an infant's growth (WHO Expert Committee, 1995). These indicators should not be used interchangeably. In selecting the appropriate indicator, consideration should be given to the validity of the indicator—that is, to what extent does the indicator measure what it is intended to measure (Keller, 1982; Nunnally & Bernstein, 1994), and is the measure indicative of the phenomenon of interest? In the case of FTT, the most suitable indicator for clinical and research purposes will be the one that is most useful in correctly discriminating between infants with FTT and healthy infants (Brownie, Habicht, & Cogill, 1986).

Cutoff values 

Following decision making for the selection of an appropriate anthropometric indicator, consideration must be given to the selection of a cutoff value or the percentile below which growth is considered abnormal. In the clinical and research literature, cutoff values of less than the 10th, 5th, or 3rd percentile have been used to define FTT. Whether for clinical or research purposes, the level of the cutoff value will determine which infants are classified as having FTT and which are not. It is important to remember that weight and length are continuous variables. The use of a cutoff value transforms an anthropometric indicator into a dichotomous variable (Brownie & Habicht, 1984; Habicht & Pelletier, 1990). The most precise definition of FTT should be determined by coupling the best indicator with the best cutoff value (Habicht, Meyers, & Brownie, 1982). The cutoff value of choice should be the one that best discriminates between infants who most likely will have FTT and those who will not (Brownie et al., 1986; Gorstein et al., 1994).

The sensitivity of an FTT indicator is defined as the proportion of infants who actually have FTT who are correctly classified as having it. Specificity refers to the proportion of healthy infants classified as healthy, i.e., non-FTT (Habicht et al., 1982; Keller, 1982). However, sensitivity and specificity of any proxy measure rarely reach 100%. There will always be some infants above the cutoff value who have FTT (false negative) and some infants below the cutoff value who do not have FTT (false positive) (Chen, Chowdhury, & Huffman, 1980). A cutoff value is often determined in relation to the median value of the reference population. For assessment of infant growth, the 50th percentile on the NCHS growth chart is the median reference. The sensitivity of an indicator is increased as the cutoff value is moved farther below the median value of the reference population (WHO Expert Committe, 1995). An inverse relationship exists between sensitivity and specificity so that changing the cutoff value to increase sensitivity or specificity will cause a decrease in the other (Brownie et al., 1986; Habicht et al., 1982). The cutoff value could be set far enough below the reference median that only infants with FTT are selected, but many other infants with FTT would not be selected. The dilemma is that as the cutoff value is moved closer to the median, infants who are small but do not have FTT will be incorrectly included (WHO, 1995). Therefore the cutoff value with the most sensitivity (3rd percentile) will result in the least number of healthy infants being incorrectly classified as having FTT but would increase the number of infants with FTT being classified as healthy.

In defining FTT, the best cutoff value for an anthropometric indicator is the one that has the most sensitivity. Ideally, this cutoff value should be able to completely separate infants with FTT from healthy infants (Habicht, 1980; Mora, 1989). However, in infants there will always be some overlapping of growth parameters between the two groups. For example, by definition, 5% of infants in the reference population will grow below the 5th percentile on the NCHS growth chart, but the majority of infants below the 5th percentile are most likely not growing normally (Peterson, 1993). These particular healthy infants will have weights that have followed a stable pattern of growth, and their heights will be in proportion to their weight (Berhane & Dietz, 1999) (see Figure 1).

Another example is the infant whose weight trajectory has been following the 75th percentile and then experiences a faltering in weight gain. The infant's weight falls to the 15th percentile but is above the traditional cutoff value of the 5th percentile to define FTT. This infant would be inappropriately classified as healthy (see Figure 2). In this second example the infant would be considered as having FTT because of an alternative definition that indicates that FTT is present if an infant's weight falls across two major percentiles (95th, 90th, 75th, 50th, 25th, 10th, and 5th) over a period of 3 to 6 months. However, the infant's weight never falls below a specified cutoff value.

Back to Article Outline

Summary 

Reference data from a healthy population of infants serve as the basis for selecting a specific cutoff value. However, reference data do not provide the necessary information about sensitivity that is required for the selection of a cutoff value that must discriminate between healthy infants and infants with FTT (WHO, 1995). The selection of a cutoff value is a purely arbitrary decision (Brownie & Habicht, 1984; Mora, 1989) representing a statistical separation of infants with FTT from healthy infants (WHO Working Group, 1986) based on statistical probabilities. Ideally, the cutoff value should be set at a level that has some physiologic relevance, such as functional impairment (WHO, 1995). For the infant with FTT, functional impairment might include evidence of developmental delay or feeding problems. However, because infants with FTT are considered a heterogeneous group (Drotar, 1989), it may be difficult to identify functional indicators of growth failure that are common to all infants with FTT. This may be further hindered because the duration and severity of an infant's growth failure are rarely reported (Drotar, 1990). Infants with FTT and healthy infants are classified by the cutoff value used. The cutoff value should be set at a level that is relevant to the concerns identified in clinical practice or to the purposes of the research. For example, if the interest was in identifying infants with FTT who have acute growth failure, then weight for length that has fallen below the 5th percentile and length for age that has remained stable would be the indicators of choice. If long-term or chronic FTT were the measure of interest, then length for age that has fallen below the 5th percentile would be the appropriate choice.

Back to Article Outline

Conclusion 

Comparisons of anthropometric measures to a reference population serve as a marker of the process of growth failure. However, anthropometry is not diagnostic in and of itself and provides no information about the etiology of FTT (Peterson, 1993; Zenel, 1997). The reality of clinical practice or research in infants with FTT is that the NCHS growth chart is used as a standard, instead of a reference, and decisions about FTT are made based on where an infant's anthropometric measures are in relation to the established percentiles on the growth chart (WHO, 1995). A single definition of FTT does not exist that is capable of completely discriminating between infants with FTT and infants who are healthy. There will always be some misclassification of infants (Habicht et al., 1982). However, key to defining FTT is that the infant's growth deviated from an established pattern of age-appropriate growth. The final definition of FTT should maximize the number of infants correctly classified as having FTT and thus reduce the number of healthy infants classified as having FTT.

Anthropometric indicators and cutoff values cannot be used interchangeably to represent the same process of growth failure. The selection of the appropriate anthropometric indicator and cutoff value to define FTT should be driven by the outcomes desired in clinical practice or the purpose of the research. Ideally, the use of all 3 anthropometric indicators at a cutoff value of the 5th percentile would facilitate the most comprehensive understanding of the growth process in FTT. Because each indicator and cutoff value are representative of different processes, the rationale for the use of a selected anthropometric indicator and cutoff value should be clearly stated (WHO Working Group, 1986). Regardless of the anthropometric indicator and cutoff value selected, accurate interpretation of the growth data is dependent up serial measurements and not one point in time (Berhane & Dietz, 1999). There is much to be learned about FTT in both clinical practice and research. A clearly stated rationale for the selection of a specific anthropometric indicator and cutoff value will increase the ability to interpret growth data accurately and increase understanding of this complex syndrome of growth failure.

Back to Article Outline

References 

  1. Berhane R, Dietz WH. Clinical assessment of growth. In:  Kessler DB,  Dawson P editor. Failure to Thrive and Pediatric Undernutrition: A Transdisciplinary Approach. Baltimore: Paul H. Brookes Publishing Co; 1999;p. 195–214
  2. Bessler S. Nutritional assessment. In: 2nd ed.  Samour PQ,  Helm KK,  Lang CE editor. Handbook of Pediatric Nutrition. Gaithersburg (MD): Aspen Publishers; 1999;p. 17–42
  3. Bithoney WG, Dubowitz H, Egan H. Failure to thrive/growth deficiency. Pediatrics in Review. 1992;13:453–459
  4. Brownie C, Habicht J. Selecting a screening cut-off point or diagnostic criterion for comparing prevalences of disease. Biometrics. 1984;40:675–684
  5. Brownie C, Habicht J, Cogill B. Comparing indicators of health or nutritional status. American Journal of Epidemiology. 1986;124:1031–1044
  6. Chen LC, Chowdhury A, Huffman SL. Anthropometric assessment of energy-protein malnutrition and subsequent risk of mortality among preschool children. American Journal of Clinical Nutrition. 1980;33:1836–1845
  7. Chumlea WC, Guo SS. Physical growth and development. In: 2nd ed.  Samour PQ,  Helm KK,  Lang CE editor. Handbook of Pediatric Nutrition. Gaithersburg (MD): Aspen Publishers; 1999;p. 3–15
  8. Corrales K, Utter S. Failure to thrive. In: 2nd ed.  Samour PQ,  Helm KK,  Lang CE editor. Handbook of Pediatric Nutrition. Gaithersburg (MD): Aspen Publishers; 1999;p. 395–412
  9. Drotar D. Behavioral diagnosis in nonorganic failure to thrive: A critique and suggested approach to psychological assessment. Journal of Developmental and Behavioral Pediatrics. 1989;10:48–55
  10. Drotar D. Sampling issues in research with nonorganic failure to thrive children. Journal of Pediatric Psychology. 1990;15:255–272
  11. Frank DA, Zeisel SH. Failure to thrive. Pediatric Clinics of North America. 1988;35:1187–1206
  12. Frank DA, Silva M, Needlman R. Failure to thrive: Mystery, myth, and method. Contemporary Pediatrics. 1993;10:114–133
  13. Gorstein J. Assessment of nutritional status: Effects of different methods to determine age on the classification of undernutrition. Bulletin of the World Health Organization. 1989;72:273–283
  14. Gorstein J, Sullivan K, Yip R, de Onis M, Trowbridge F, Fajans P, et al. Issues in the assessment of nutritional status using anthropometry. Bulletin of the World Health Organization. 1994;72:273–283
  15. Habicht J. Some characteristics of indicators of nutritional status for use in screening and surveillance. American Journal of Clinical Nutrition. 1980;33:531–535
  16. Habicht J, Pelletier DL. The importance of context in choosing nutritional indicators. Journal of Nutrition. 1990;120:1519–1524
  17. Habicht J, Meyers LD, Brownie C. Indicators for identifying and counting the improperly nourished. American Journal of Clinical Nutrition. 1982;35:1241–1254
  18. Karlberg J, Jalil F, Lam B, Low L, Yeung CY. Linear growth retardation in relation to the three phases of growth. European Journal of Clinical Nutrition. 1994;48(Suppl 1):S25–S44
  19. Keller W. Choice of indicators of nutritional status. In:  Schurch B editors. Evaluation of Nutrition Education in Third World Communities. Vienna: Hans Huber Publishers; 1982;p. 101–113
  20. Kessler DB. Failure to thrive and pediatric undernutrition: Historical and theoretical context. In:  Kessler DB,  Dawson P editor. Failure to Thrive and Pediatric Undernutrition: A Transdisciplinary Approach. Baltimore: Paul H. Brookes Publishing Co; 1999;p. 3–17
  21. Metallinos-Katsaras E, Gorman KS. Effects of undernutrition on growth and development. In:  Kessler DB,  Dawson P editor. Failure to Thrive and Pediatric undernutrition: A transdisciplinary approach. Baltimore: Paul H. Brookes Publishing Co; 1999;p. 37–63
  22. Mora JO. A new method for estimating a standardized prevalence of child malnutrition from anthropometric indicators. Bulletin of the World Health Organization. 1989;67:133–142
  23. Nunnally JE, Bernstein IH. Psychometric Theory. (3rd ed. New York: McGraw-Hill, Inc; 1994;
  24. Peterson KE. Failure to thrive. In:  Queen PM,  Lang CE editor. Handbook of Pediatric Nutrition. Gaithersburg (MD): Aspen Publishing; 1993;p. 366–383
  25. Peterson KE, Chen LC. Defining undernutrition for public health purposes in the United States. Journal of Nutrition. 1990;120:933–942
  26. Pomerance HH. Growth and its assessment. Advances in Pediatrics. 1995;42:545–574
  27. Raynor P, Rudolf MCJ. Anthropometric indices of failure to thrive. Archives of Disease in Childhood. 2000;82:364–365
  28. Sherry B. Epidemiology of inadequate growth. In:  Kessler DB,  Dawson P editor. Failure to Thrive and Pediatric Undernutrition: A Transdisciplinary Approach. Baltimore: Paul H. Brookes Publishing Co; 1999;p. 19–36
  29. Trowbridge FL. Clinical and biochemical characteristics associated with anthropometric nutritional categories. American Journal of Clinical Nutrition. 1979;32:758–766
  30. WHO Expert Committee . Physical status: The use and interpretation of anthropometry. WHO Technical Report Series, No. 854. Geneva: World Health Organization; 1995;
  31. Working Group WHO. Use and interpretation of anthropometric indicators of nutritional status. Bulletin of the World Health Organization. 1986;64:929–941
  32. Working Group on Infant Growth WHO. An evaluation of infant growth: The use and interpretation of anthropometry in infants. Bulletin of the World Health Organization. 1995;73:165–174
  33. Wilcox WD, Nieburg P, Miller DS. Failure to thrive: A continuing problem of definition. Clinical Pediatrics. 1989;28:391–394
  34. Wright JA, Ashenburg C, Whitaker RC. Comparison of methods to categorize undernutrition in children. Pediatrics. 1994;24:944–946
  35. Zenel JA. Failure to thrive: A general pediatrician's perspective. Pediatrics in Review. 1997;18:371–378

☆ Address correspondence and reprint requests to Deborah K. Steward, PhD, RN, Assistant Professor, The Ohio State University, College of Nursing, 1585 Neil Ave, Columbus, OH 43210.

PII: S0882-5963(03)00024-1

doi:10.1053/jpdn.2003.25

Journal of Pediatric Nursing
Volume 18, Issue 1 , Pages 52-59, February 2003

Article Tools

* Image Usage & Resolution