Literature
首页医源资料库在线期刊美国临床营养学杂志2004年80卷第4期

Comparison of the use of body mass index percentiles and percentage of ideal body weight to screen for malnutrition in children with cystic fibrosis

来源:《美国临床营养学杂志》
摘要:ABSTRACTBackground:TheCysticFibrosisFoundation(CFF)recommendsusingthepercentageofidealbodyweight(%IBWCFF)andbodymassindexpercentiles(BMIp)toassessweight-for-heightstatusandtoscreenformalnutrition。Objective:Theobjectivewastoexaminetheagreementanddiscrepancy......

点击显示 收起

Zhumin Zhang and HuiChuan J Lai

1 From the Department of Nutrition Sciences, University of Wisconsin, Madison.

2 Supported by National Institutes of Health grant K01 DK02891.

3 Address reprint requests to HCJ Lai, Department of Nutrition Sciences, University of Wisconsin, Madison, 1415 Linden Street, Madison, WI 53706. E-mail: lai{at}nutrisci.wisc.edu.


ABSTRACT  
Background: The Cystic Fibrosis Foundation (CFF) recommends using the percentage of ideal body weight (%IBWCFF) and body mass index percentiles (BMIp) to assess weight-for-height status and to screen for malnutrition.

Objective: The objective was to examine the agreement and discrepancy between the use of %IBWCFF and BMIp for screening malnutrition.

Design: Data from 13 021 children reported to the 2000 CFF Patient Registry were analyzed.

Results: In children of average stature (ie, height-for-age between the 25th and 75th percentiles) and aged <10 y, %IBWCFF corresponded closely to BMIp, and the prevalence of underweight estimated by %IBWCFF < 90% was similar to that by BMIp < 15th percentile. However, in children with short stature (ie, height-for-age < 25th percentile), %IBWCFF reflected significantly better weight-for-height status than did the BMIp, whereas the opposite trend was observed in children with tall stature (ie, height-for-age > 75th percentile). Such discrepancies averaged 8–12 percentage points when BMIp was reexpressed to the same unit and scale as %IBWCFF. Consequently, the prevalence of underweight estimated by %IBWCFF < 90% was significantly lower (7.3%) than that estimated by BMIp < 15th percentile (25.7%) in children with short stature, whereas the opposite trend was found in children with tall stature (47.7% and 14.4%, respectively). Additional analyses showed that BMIp was more sensitive to, and had stronger associations with, the percentage of predicted forced expiratory volume in 1 s.

Conclusion: Compared with BMIp, %IBWCFF underestimated the severity of malnutrition in children with short stature and overestimated the severity of malnutrition in children with tall stature.

Key Words: Cystic fibrosis • malnutrition • anthropometry • growth • weight-for-height • percentage of ideal body weight • body mass index • lung function • children


INTRODUCTION  
Assessment of the relative proportion of weight for height and its application in identifying the prevalence of underweight or overweight are widely used in both clinical and public health settings (1–8). One common approach for estimating the weight-for-height proportion is to express body weight as a percentage of an "ideal weight for that stature." However, various methods are used to define ideal weight for a given stature, including the median weight-for-age (9), median weight-for-height (2), and median weight-for-height adjusted for age (4, 5). In a study by Wright et al (10), the number of children classified as malnourished and the degree of undernutrition were found to vary substantially when different weight-for-height indexes were used. Similar findings were reported in other studies (5, 11).

In children with cystic fibrosis (CF), the CF Foundation (CFF) Nutrition Consensus Committee (12) recommends the use of percentage of ideal body weight [%IBWCFF; calculated by the Moore method (5)] to assess weight-for-height proportion, based on findings from observational studies linking percentage of IBW and prognosis (13–16). However, according to a study by Poustie et al (17), only 20% of the 44 CF centers in the United Kingdom routinely calculated %IBWCFF at each clinic visit. The limited use of %IBWCFF in routine clinical settings applicable to the CF population can be explained by several factors. First, calculation of %IBWCFF requires several steps; hence, it is relatively time consuming. Second, accurate calculation of %IBWCFF by hand is difficult because the ideal weight used in this method is generally determined by ways of visual estimation. As a result, large interindividual and intraindividual variation is common, making %IBWCFF values unreliable (17). Third, no readily available clinical resource tracks %IBWCFF over time, making longitudinal monitoring difficult.

With the availability of body mass index percentile (BMIp) charts as part of the new growth charts (18), the CFF Nutrition Consensus Committee also recommends using BMIp to assess weight-for-height proportion (12). In a study by Cole (6), BMI was shown to be superior to other weight-for-height indexes for controlling the effect of sex, height, and age on weight. In addition, the use of BMIp as a screening tool to identify both overweight and underweight in children was also validated recently (8).

In view of the abovementioned observations, we conducted the present study to examine the agreement and discrepancy between the use of %IBWCFF and BMIp for screening malnutrition and to determine whether %IBWCFF or BMIp has stronger association with lung disease indexes in children with CF.


SUBJECTS AND METHODS  
Study population
The CFF Patient Registry documents the diagnosis and annual follow-up evaluations of patients with CF who are seen at accredited centers in the United States, as described in detail elsewhere. CFF Registry data reported in the year 2000 were used. Among the 22 303 patients documented in the registry, 13 280 were children aged 2–20 y. Children aged <2 y were excluded from the present study because BMI references are not available in the 2000 Centers for Disease Control and Prevention (CDC) growth charts (18). Of the 13 280 observations, 0.4% (n = 57 observations) had heights or weights less than zero and another 1.5% (202 observations) had height or weight or BMI measurements that deviated >4 SD from the reference mean. These height and weight values were likely outliers resulting from measurement or recording errors and were excluded. The remaining 13 021 children were included in the present study.

Computation of ideal body weight recommended by the Cystic Fibrosis Foundation
The steps involved in calculating %IBWCFF by hand for an individual child with CF are as follows:

  1. Obtain the child’s sex, age, height, and weight.
  2. Plot the child’s height on the growth chart and determine his or her height-for-age percentile.
  3. Determine the weight that corresponds to the same percentile ranking as the child’s height-for-age. (This is his or her ideal weight.)
  4. Divide the actual weight of the child by his or her ideal weight and multiply that value by 100%.

These steps are illustrated in an example here and in Figure 1A. For example, a 10-y-old boy with a height of 147.0 cm and a weight of 30.0 kg would be in the 90th percentile for height-for-age. His ideal weight = 90th percentile value of weight at age 10 y = 42.0 kg and his %IBWCFF = (actual weight/ideal weight) x 100% = (30/42) x 100% = 71.4%.


View larger version (35K):
FIGURE 1.. An example illustrating the calculations of ideal body weight recommended by the Cystic Fibrosis Foundation (%IBWCFF) (A) and body mass index percentiles (BMIp) (B). To calculate %IBWCFF, plot the child’s actual height (ie, 147 cm) on the height-for-age growth chart from the Center for Disease Control and Prevention (CDC), estimate his height-for-age percentile (ie, 90th percentile), estimate his ideal weight by using the weight value corresponding to the 90th percentile point (ie, 42.0 kg), then divide his actual weight by the ideal weight and multiply by 100% (ie, 71.4%). To estimate BMIp, calculate the child’s BMI (ie, 13.9), plot his BMI on CDC’s BMI-for-age growth chart, then estimate the percentile ranking (ie, 3rd percentile)

 
In the present study, %IBWCFF calculations were performed in a computerized program in SAS (SAS Institute Inc, Cary, NC) by using the reference values downloaded from the CDC’s website for the new growth charts (Internet: http://www.cdc.gov/growthcharts).

Computation of body mass index percentile
How BMIp can be estimated by hand by using the new CDC growth charts is illustrated in Figure 1B. By using the same example described in the previous section, a 10-y-old boy with height of 147.0 cm and weight of 30.0 kg has a BMI of 13.9 (ie, 30/1.472 = 13.9), which corresponds to approximately the 3rd percentile. In the present study, BMIp was calculated in a computerized program in SAS by using the reference values downloaded from CDC’s website for the new growth charts (Internet: http://www.cdc.gov/growthcharts).

Comparison between ideal body weight recommended by Cystic Fibrosis Foundation and body mass index percentile
Because %IBWCFF and BMIp are calculated by different methods and also expressed in different units (ie, percentage in %IBWCFF and percentile in BMIp) and scales (ie, 100% is optimal in %IBWCFF and 50th percentile is optimal in BMIp), direct comparison between the values of %IBWCFF and BMIp are difficult. There are 2 fundamental differences in the derivation of %IBWCFF and BMIp. First, the ideal weight for a given stature is defined differently, as shown in Figure 1 and Figure 2. In %IBWCFF, the ideal weight for a given stature is assumed to be the weight value that corresponds to the same percentile ranking as the child’s height-for-age (eg, ideal weight corresponding to 42.0 kg in the example illustrated in Figure 1A). In other words, the child with an ideal weight for his or her stature should have height-for-age and weight-for-age at exactly the same percentile ranking. In BMIp, the ideal weight for a given stature is assumed to be the weight value that corresponds to the 50th percentile of BMI at that age (eg, ideal BMI corresponding to 16.6 in the example illustrated in Figure 1B, which is equivalent to assigning ideal weight to be 16.6 x 1.472 = 35.9 kg). Second, deviation from the actual status to the ideal status is expressed differently. In %IBWCFF, this deviation is expressed as a percentage relative to the ideal value, whereas in BMIp, this deviation is expressed in percentile ranking.


View larger version (35K):
FIGURE 2.. Comparison of the ideal weights calculated by the ideal body weight recommended by the Cystic Fibrosis Foundation (%IBWCFF) method and the body mass index percentile (BMIp) method in children with CF who were reported to the 2000 CFF Patient Registry. Average stature: height-for-age between the 25th and 75th percentiles; short stature: height-for-age < 25th percentile; tall stature: height-for-age > 75th percentile.

 
To facilitate the numerical comparisons between %IBWCFF and BMIp, we reexpressed BMIp into the percentage unit with 100% being optimal. For instance, in the example illustrated in Figure 1, a 10-y-old boy with height of 147.0 cm and weight of 30.0 kg has a %IBWCFF of 71.4% and a BMIp at the 3rd percentile. To reexpress his BMIp, his absolute BMI is divided by the median value of BMI at his age obtained from CDC’s BMI chart, ie, 13.9/16.6 x 100 = 83.7%. This calculation is equivalent to dividing his actual weight by the ideal weight derived from the BMIp method, ie, 30/35.9 = 83.6%. Therefore, the reexpressed BMIp is referred to as %IBWBMIp, as shown in Figure 3 and throughout this report.


View larger version (45K):
FIGURE 3.. Comparison of the nutritional status as reflected by ideal body weight recommended by the Cystic Fibrosis Foundation (%IBWCFF) with that indicated by body mass index percentile (BMIp) in children with CF who were reported to the 2000 CFF Patient Registry. Average stature: height-for-age between the 25th and 75th percentiles; short stature: height-for-age < 25th percentile; tall stature: height-for-age > 75th percentile.

 
Criteria of malnutrition by ideal body weight recommended by Cystic Fibrosis Foundation and body mass index percentile
The criterion of %IBWCFF < 90% recommended by CFF Pediatric Nutrition Consensus Report (12) was used to identify malnutrition. Alternatively, BMIp < 15th percentile was chosen as an equivalent indicator for malnutrition and compared with %IBWCFF < 90%. The cutoff value of 15th percentile for BMIp was chosen on the basis of the study by Waterlow et al (2) that 10% below the median corresponded approximately to 1 SD below the median, which translated to approximately the 15th percentile in normalized distribution. In addition, we found that among children with average stature in which the ideal weights obtained from the BMIp and the %IBWCFF methods were similar, 15th percentile BMIp corresponded more closely to 90% IBWCFF than 10th percentile BMIp at most ages. This similarity is clearly demonstrated in the middle 2 panels of Figure 4, in which the percentages of patients with BMIp < 15th percentile (ie, the black dotted lines) are closer to those with %IBWCFF < 90% (ie, the black solid lines) than those with BMIp < 10th percentile (ie, the gray lines) at most ages.


View larger version (38K):
FIGURE 4.. Comparison of the prevalence of malnutrition as defined by ideal body weight recommended by the Cystic Fibrosis Foundation (%IBWCFF) < 90% with that defined by body mass index percentile (BMIp) < 15th percentile in children with CF who were reported to the 2000 CFF Patient Registry. In the panels representing children with CF of average stature, the prevalence of malnutrition as defined by BMIp < 10th percentile is also shown (gray lines) for comparison with the prevalence of malnutrition as defined by BMIp < 15th percentile. Average stature: height-for-age between the 25th and 75th percentiles; short stature: height-for-age < 25th percentile; tall stature: height-for-age > 75th percentile.

 
Association of ideal body weight recommended by Cystic Fibrosis Foundation and body mass index percentile with lung disease status
The lung function indicator forced expiratory volume in 1 s (FEV1) was used to assess whether %IBWCFF or BMIp has stronger association with lung disease status in children with CF. FEV1 was expressed as a percentage of predicted values according to the Dockery equations [found in Wang et al (19)]. For this analysis, only patients who were aged >6 y and had lung function measurements were included (n = 9770).

Statistical analysis
SAS (version 8.2) and R (Internet: http://www.r-project.org) statistical software programs were used to perform statistical analyses. All statistical analyses were performed separately for each sex or stratified by sex because of the well-documented sex gap in growth status and clinical outcomes between male and female patients with CF (15, 20, 21).

Paired t test was conducted to test the difference between the ideal weights calculated by the %IBWCFF and the BMIp methods. Comparisons among categorical outcomes were assessed by chi-square contingency table methods. Agreement between the 2 criteria in identifying children with malnutrition, as presented in Figure 5, was assessed by the statistic (22). A < 0.4 was considered to be in poor agreement according to Landis and Koch (23). A locally weighted scatter plot smoother (the LOESS function in R) was used to estimate trends in relation to age in Figures 2–4.


View larger version (37K):
FIGURE 5.. Agreement between using ideal body weight recommended by Cystic Fibrosis Foundation (%IBWCFF) < 90% or body mass index percentile (BMIp) < 15th percentile as the criterion for identifying malnutrition in children with CF who were reported to the 2000 CFF Patient Registry. Average stature: height-for-age between the 25th and 75th percentiles; short stature: height-for-age < 25th percentile; tall stature: height-for-age > 75th percentile. White areas represent the percentages of children classified as normal by both criteria; gray areas represent the percentages of children classified as malnourished by both criteria; hatched areas represent the percentages of children classified as malnourished by BMIp < 15th percentile but normal by %IBWCFF < 90%; black areas represent the percentages of children classified as normal by BMIp < 15th percentile but malnourished by %IBWCFF < 90%.

 
For the analyses comparing the relative strengths of associations between %IBWCFF and BMIp and lung disease outcomes presented in Figure 6, multiple regression analyses (by using the PROC GLM procedure in SAS) were performed by using the following model: percentage of predicted FEV1 = ß0 + ß1(%IBWCFF or BMIp) + ß2(age in y). The scales and units of %IBWCFF and BMIp were adjusted such that ß0 represents the estimated value of percentage of predicted FEV1 at optimal nutritional status (ie, %IBWCFF = 100% or BMIp = 50th percentile) and ß1 represents the estimated change of percentage of predicted FEV1 for every unit change in the nutritional index (ie, a 10% change in %IBWCFF or a 1 SD change in BMIp). More specifically, for %IBWCFF, the model percentage of predicted FEV1 = ß0 + ß1[(%IBWCFF – 100)/10] + ß2(age in y) was fitted; thus, ß0 represents the estimated value of percentage of predicted FEV1 when %IBWCFF is at 100% and ß1 represents the estimated change of percentage of predicted FEV1 for every 10% change in %IBWCFF. For BMIp, the model percentage of predicted FEV1 = ß0 + ß1(BMIp reexpressed as Z score) + ß2(age in y) was fitted. It should be noted that BMIp and BMI Z score are mathematically interchangeable on the basis of the properties of normal distribution, eg, BMIp of 15th, 50th, and 85th percentiles are equivalent to BMI Z scores of –1, 0, and 1, respectively. Therefore in this model, ß0 represents the estimated value of percentage of predicted FEV1 when BMI Z score is 0 (equivalent to BMIp of 50th percentile) and ß1 represents the estimated change of percentage of predicted FEV1 for every 1 SD change in BMIp. Age was included as a continuous covariate in all models to adjust for its influence on percentage of predicted FEV1.


View larger version (42K):
FIGURE 6.. Associations of ideal body weight recommended by the Cystic Fibrosis Foundation (%IBWCFF) and body mass index percentile (BMIp) with the percentage of predicted forced expiratory volume in 1 s (FEV1), as estimated by regression analyses using the model percentage of predicted FEV1 = ß0 + ß1(%IBWCFF or BMIp) + ß2age as explained in details in Statistical Analysis. Data from children with CF who were reported to the 2000 CFF Patient Registry were used. Average stature: height-for-age between the 25th and 75th percentiles; short stature: height-for-age < 25th percentile; tall stature: height-for-age > 75th percentile.

 

RESULTS  
Comparison between ideal body weight recommended by the Cystic Fibrosis Foundation and body mass index percentile
Initial analyses were performed to compare whether the ideal weight for a given stature obtained by the %IBWCFF was similar to that obtained by the BMIp method. Because the ideal weight for a given stature depends on sex, age, and stature, results from these analyses are presented by sex, age, and stature. As shown in Figure 2, in children with CF of average stature (ie, height-for-age between the 25th and 75th percentiles), the ideal weights obtained by the %IBWCFF method were nearly identical to those obtained by the BMIp method up to age 12 y. However, in children with CF of short stature (ie, height-for-age < 25th percentile), the ideal weights obtained by the %IBWCFF method were significantly lower than those obtained by the BMIp method (P < 0.0001). In contrast, in children with CF of tall stature (ie, height-for-age > 75th percentile), the ideal weights obtained by the %IBWCFF method were significantly higher than those obtained by the BMIp method (P < 0.0001). The discrepancy in the value of ideal weight between the %IBWCFF and the BMIp methods increases with age and increases as the children’s stature deviates from the 50th percentile point. In addition, this discrepancy was significantly greater in girls than in boys (P < 0.0001).

As a result of the aforementioned discrepancy in ideal weight, the relative percentage of actual weight to ideal weight also differed between the %IBWCFF and BMIp methods. This difference is clearly demonstrated in Figure 3. In children with CF of average stature, %IBWCFF and %IBWBMIp patterns were similar at all ages, although such small differences remained statistically significant (P < 0.0001). However, in children with CF of short stature, %IBWCFF was substantially higher than %IBWBMIp at all ages by an average of 8.1 percentage points (P < 0.0001). This finding implies that, in children with shorter than average stature, nutritional status would always appear better when %IBWCFF is used than when %IBWBMIp is used. In contrast, in children with CF of tall stature, %IBWCFF was substantially lower than %IBWBMIp at all ages by an average of 11.9 percentage points (P < 0.0001). This finding implies that, in children with taller than average statue, nutritional status would always appear worse when %IBWCFF is used than when %IBWBMIp is used.

Further comparison between %IBWBMIp and BMIp reveals the difference between using the percentage unit compared with the percentile unit to quantify the deviation from actual to ideal weights. In Figure 3, the plotting range of 90–110% in the percentage unit is equivalent to the range of the 15th–85th percentile in the percentile unit, as described in Subjects and Methods. This relation was verified by the nearly overlapping lines of %IBWBMIp and BMIp up to age 10 y in children with average stature and up to age 5–10 y in children with short or tall stature (Figure 3). However, in children aged >10 y, discrepancy between %IBWBMIp and BMIp was apparent and was particularly noticeable in children with short stature. This observation indicates that, in children aged >10 y, the relation of 10% below the median corresponding approximately to 1 SD below the median no longer holds.

Prevalence of malnutrition as indicated by ideal body weight recommended by the Cystic Fibrosis Foundation < 90% or body mass index percentile < 15th percentile
The prevalence of malnutrition on the basis of %IBWCFF < 90% and BMIp < 15th percentile is shown in Figure 4. In children with CF of average stature, the prevalence of malnutrition estimated by the %IBWCFF < 90% was similar to that estimated by the BMIp < 15th percentile method at all ages (P = 0.34). However, in children with CF of short stature, the prevalence of malnutrition identified by the %IBWCFF < 90% was significantly lower at all ages by an average of 10 percentage points compared with that identified by BMIp < 15th percentile (P < 0.0001). The opposite trend, and greater differences (with an average of 33 percentage points), was observed in children with CF with tall stature (P < 0.0001). To summarize, the prevalence of underweight estimated by %IBWCFF < 90% was significantly lower (ie, 7.3%) than that estimated by BMIp < 15th percentile (ie, 25.7%) in children with short stature, whereas the opposite trend was found in children with tall stature (47.7% with %IBWCFF < 90% and 14.4% with BMIp < 15th percentile).

The agreement between using %IBWCFF < 90% and BMIp < 15th percentile for identifying malnutrition was further assessed by the statistics. In children with average stature, >90% of children were consistently classified by the 2 criteria, and statistics indicated good agreement (Figure 5). However, large discrepancy between using %IBWCFF < 90% and BMIp < 15th percentile for identifying malnutrition was found when the children’s stature deviated from the median.

Association of ideal body weight recommended by the Cystic Fibrosis Foundation and body mass index percentile to percentage of forced expiratory volume in 1 s
As shown in Figure 6, in children with CF of average stature, the values of ß0 were similar between the %IBWCFF model and the BMIp model for both sexes. However, the BMIp model had a significantly larger ß1 (slope) than the %IBWCFF model, P < 0.0001. This observation indicated that BMIp was more sensitive to changes in percentage of predicted FEV1 than %IBWCFF. The BMIp model also had larger R2 than the %IBWCFF model, indicating that BMIp had stronger association to percentage of predicted FEV1 than %IBWCFF.

In children with short or tall stature, differences in the values of ß0 were noted. Specifically, in children with short stature, the %IBWCFF model had smaller ß0 than the BMIp model, whereas the opposite trend was observed in children with tall stature. Nevertheless, both ß1 and R2 were consistently larger in the BMIp models than the %IBWCFF models, indicating that BMIp had stronger associations to percentage of predicted FEV1 than %IBWCFF. Finally, it should be commented on that statistics showed that the agreement between %IBWCFF < 90% and BMIp < 15th percentile as criterion for malnutrition was similar between children aged <6 y (who did not have lung function data) and children aged >6 y (who had lung function data) in children with short or average stature. In children with tall stature, the agreement between these 2 indicators was better in children aged <6 y compared with children aged >6 y.


DISCUSSION  
In this report, we presented comprehensive analyses to compare 2 weight-for-height indexes that are commonly recommended for screening malnutrition in children with CF. Our results demonstrated a good agreement between these 2 indexes, namely, %IBWCFF and BMIp, among prepubertal children (ie, aged <10 y) who had average stature (ie, height-for-age between the 25th and 75th percentiles). However, in older children and children with shorter or taller than average stature, significant discrepancy was found between %IBWCFF and BMIp. Specifically, when compared with BMIp, %IBWCFF underestimates the severity of malnutrition in children with short stature and overestimates the severity of malnutrition in children with tall stature. This discrepancy increases with age and also increases as the children’s stature deviates from the median value.

The discrepancy between %IBWCFF and BMIp as a malnutrition index is primarily due to their differences in defining the ideal weight for a given age and stature. As described previously, in the %IBWCFF method, the ideal weight for a given age and stature is assumed to be the weight value that corresponds to the same percentile ranking as the child’s height-for-age (5). This assumption can be valid in children with average stature, as evidenced by the observation from the reference population of CDC’s growth charts (18) that children with both height and weight at the 50th percentile values have weight-for-height or BMI or both values that fall around the 50th percentile point during prepubertal period. However, the same assumption is likely not valid in children with short or tall stature, because the amount of weight gain per increment of height gain, and vice versa, is not constant as the stature deviates from the median. This is evidenced by the observation from the reference population of CDC’s growth charts (18) that children with both height and weight values at the 90th percentiles have weight-for-height or BMI or both values that fall substantially greater than the 50th percentile point (ie, around the 80th percentile). However, children with both height and weight values at the 10th percentile have weight-for-height or BMI or both values that fall substantially lower than the 50th percentile point (ie, around the 25th percentile). These observations explain why ideal weights calculated by the %IBWCFF method are substantially greater than those calculated by the BMIp method in children with tall stature, whereas the opposite results are obtained in children with short stature. Alternatively, in the BMIp method, the ideal weight for a given age and stature is assumed to be the weight value that corresponds to the median BMI at that age. Although the validity of this approach to estimate ideal weight remains to be proven, the ideal weight estimated by this method is directly derived from several observed reference populations representative of the US population (18).

In addition to differences in defining ideal weights, differences in expressing the deviation from the actual to the ideal status also contributed to the discrepancy between %IBWCFF and BMIp in identifying malnutrition. In the %IBWCFF method, the deviation is expressed as a percentage relative to the ideal value. This method of standardizing measurements to reference values has a major drawback in that a given percentage of reference does not correspond to the same SD score across age (2, 24). Furthermore, the original intent of percentage of reference categories was to approximate decrements in SD scores (2). Therefore, classification based on SD score is preferable. In the BMIp method, deviation from the actual to the ideal statue is expressed in the percentile system, which is mathematically interchangeable with the SD score system when the reference curves were normalized, as in the case of CDC’s growth charts (18). Therefore, the percentile or the SD score system is preferable to the percentage of reference system in expressing the deviation between the actual and the ideal status.

Besides the aforementioned considerations about the definition of ideal weights and the statistical properties of the indexes, our finding that BMIp is more sensitive to, and has a stronger association with, the lung function parameter percentage of predicted FEV1 in the subgroup of children with CF aged >6 y provide further evidence to support the use of BMIp in favor of %IBWCFF as a screening tool for malnutrition in children with CF. However, it should be noted that, despite BMIp < 15th percentile was used to compare with %IBWCFF < 90% in the present study, it does not imply that the 15th rather than the 10th percentile [as recommended by the CFF Pediatric Nutrition Consensus Report (12)] should be used to define nutrition failure. As explained in "Subjects and Methods," the rationale for using BMI < 15th percentile to compare with %IBW < 90% was primarily because, among children with average stature in which the ideal weights obtained from the BMIp and the %IBWCFF methods were similar, the 15th percentile BMIp corresponded more closely to 90% of %IBWCFF than the 10th percentile BMIp. However, to define nutrition failure by using the BMIp index, one should choose the cutoff based on its relation to functional outcomes and not based on how close it corresponds to 90% of %IBWCFF. One potential functional outcome that might be considered is how BMIp influences lung function in children with CF. Although previous observational studies reported that BMI was correlated to percentage of predicted FEV1 (25), quantitative relations were not clearly defined. Our cross-sectional regression analyses between BMIp and percentage of predicted FEV1 enabled us to identify potential BMIp cutoffs to define nutrition failure that would be expected to affect lung function. For example, what BMIp is needed to maintain percentage of FEV1 > 80%, which is a value indicative of mild lung disease? On the basis of our results, for children with CF aged 7 y, a BMIp of 6th percentile for boys and 9th percentile for girls is needed to maintain percentage of FEV1 > 80%. Higher BMIp cutoffs were found for older children because percentage of FEV1 worsens with age (eg, BMIp > 28th and BMIp > 43rd percentiles for 15 y-old boys and girls, respectively) and among children with shorter than average stature. Because no single BMI cutoff value can be applied to all children with CF but the lowest cutoff from our results is close to the 10th percentile value, BMIp < 10th percentile appears to be a justifiable indicator for defining nutrition failure that is likely to influence lung function in children with CF.

From a practical standpoint, the readily available BMIp charts from the new CDC growth charts offer an additional advantage for screening malnutrition in routine clinical settings. Furthermore, BMI is the one most commonly recommended and widely used anthropometric index for screening overweight and underweight in the adult population. This enables continuous monitoring of nutritional status more easily when transitioning from pediatric into adult care. However, calculation of %IBWCFF by hand is time consuming, prone to estimation errors, and difficult for longitudinal tracking.

In summary, findings from the present study led us to conclude that, among children with average stature (ie, height between the 25th and 75th percentiles), %IBWCFF and BMIp yields similar estimates of ideal weights and good agreement in classifying malnutrition. However, among children with short stature (ie, height < 25th percentile), %IBWCFF underestimates the severity of malnutrition when compared with BMIp. Among children with tall stature (ie, height > 75th percentile), %IBWCFF overestimates the severity of malnutrition when compared with BMIp. Considering the practical advantages of BMIp, ie, ease of calculation and readily available BMI charts for longitudinal monitoring, we recommend using BMIp rather than %IBWCFF to assess weight-for-height status. Nevertheless, it should be emphasized that the optimal use of the BMIp chart in a child with CF is to determine whether his or her weight is in an appropriate range for his or her stature, and that BMIp should not be used as the single indicator for screening malnutrition but used in combination with height-for-age percentile and other biochemical and clinical indexes to screen for malnutrition in children with CF.


ACKNOWLEDGMENTS  
We thank Preston W Campbell from the CFF for providing the Registry data.

ZZ contributed to the analysis of data and writing of the manuscript. HCJL contributed to the design of the study, the interpretation of the results, and the writing of the manuscript. The authors had no conflicts of interest.


REFERENCES  

  1. Waterlow JC. Classification and definition of protein-calorie malnutrition. BMJ 1972;3:566–9.
  2. Waterlow JC, Buzina R, Keller W, Lane JM, Nichaman MZ, Tanner JM. The presentation and use of height and weight data for comparing the nutritional status of groups of children under the age of 10 years. Bull World Health Organ 1977;55:489–98.
  3. Use and interpretation of anthropometric indicators of nutritional status. Bull World Health Organ 1986;64:929–41.
  4. McLaren DS, Read WW. Classification of nutritional status in early childhood. Lancet 1972;2:146–8.
  5. Moore DJ, Durie PR, Forstner GG, Pencharz PB. The assessment of nutritional status in children. Nutr Res 1985;5:797–9.
  6. Cole TJ. A method for assessing age-standardized weight-for-height in children seen cross-sectionally. Ann Hum Biol 1979;6:249–68.
  7. Cole TJ, Donnet ML, Stanfield JP. Weight-for-height indices to assess nutritional status–a new index on a slide-rule. Am J Clin Nutr 1981;34:1935–43.
  8. Mei Z, Grummer-Strawn LM, Pietrobelli A, Goulding A, Goran MI, Dietz WH. Validity of body mass index compared with other body-composition screening indexes for the assessment of body fatness in children and adolescents. Am J Clin Nutr 2002;75:978–85.
  9. Gomez F, Galvan RR, Frenk S, Munoz JC, Chavez R, Vazquez J. Mortality in second and third degree malnutrition. J Trop Pediatr 1956;77–83.
  10. Wright JA, Ashenburg CA, Whitaker RC. Comparison of methods to categorize undernutrition in children. J Pediatr 1994;124:944–6.
  11. Lai HC, Kosorok MR, Sondel SA, et al. Growth status in children with cystic fibrosis based on the National Cystic Fibrosis Patient Registry data: evaluation of various criteria used to identify malnutrition. J Pediatr 1998;132:478–85.
  12. Borowitz D, Baker RD, Stallings V. Consensus report on nutrition for pediatric patients with cystic fibrosis. J Pediatr Gastroenterol Nutr 2002;35:246–59.
  13. Kraemer R, Rudeberg A, Hadorn B, Rossi E. Relative underweight in cystic fibrosis and its prognostic value. Acta Paediatr Scand 1978;67:33–7.
  14. Sturgess JM, Czegledy-Nagy E, Corey M, Thompson MW. Cystic fibrosis in Ontario. Am J Med Genet 1985;22:383–93.
  15. Kerem E, Reisman J, Corey M, Canny GJ, Levison H. Prediction of mortality in patients with cystic fibrosis. N Engl J Med 1992;326:1187–91.
  16. Corey M, McLaughlin FJ, Williams M, Levison H. A comparison of survival, growth, and pulmonary function in patients with cystic fibrosis in Boston and Toronto. J Clin Epidemiol 1988;41:583–91.
  17. Poustie VJ, Watling RM, Ashby D, Smyth RL. Reliability of percentage ideal weight for height. Arch Dis Child 2000;83:183–4.
  18. Kuczmarski RJ, Ogden CL, Guo SS, et al. 2000 CDC Growth Charts for the United States: methods and development. Vital Health Stat 11 2002;(246):1–190.
  19. Wang X, Dockery DW, Wypij D, Fay ME, Ferris BG Jr. Pulmonary function between 6 and 18 years of age. Pediatr Pulmonol 1993;15:75–88.
  20. Rosenfeld M, Davis R, FitzSimmons S, Pepe M, Ramsey B. Gender gap in cystic fibrosis mortality. Am J Epidemiol 1997;145:794–803.
  21. Lai HC, Kosorok MR, Laxova A, Makholm LM, Farrell PM. Delayed diagnosis of US females with cystic fibrosis. Am J Epidemiol 2002;156:165–73.
  22. Fleiss JL. Statistical methods for rates and proportions. New York: Wiley, 1981.
  23. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159–74.
  24. Peterson KE, Chen LC. Defining undernutrition for public health purposes in the United States. J Nutr 1990;120:933–42.
  25. Nir M, Lanng S, Johansen HK, Koch C. Long-term survival and nutritional data in patients with cystic fibrosis treated in a Danish centre Thorax 1996;51:1023–7.
Received for publication August 7, 2003. Accepted for publication April 15, 2004.


作者: Zhumin Zhang
医学百科App—中西医基础知识学习工具
  • 相关内容
  • 近期更新
  • 热文榜
  • 医学百科App—健康测试工具