Early attained weight and length predict growth faltering better than velocity measures in infants with CF

Abstract

Background

CF infants often do not grow as expected which adversely affects later clinical outcomes, thus sensitive early measures of growth deficiency are important. This study compared attained growth for age with velocity standards to determine which better predicts growth deficits at 24 months of age.

Methods

Growth deficiency in infancy based on weight and length velocity, and attained growth was calculated for 1,992 infants in the US CF Foundation National Registry using the World Health Organization (WHO) and US growth standards. One, two, and three month increments were used for calculating velocity and pooled for each age interval. Sensitivity and specificity of early indicators to predict growth deficiency at 24 months were calculated.

Results

Observed prevalence of weight deficiency (<10^th percentile) during the first year of life was 26.8% higher (95%CI=(25.6, 28.1%), p<0.001) on average when measured by attained weight for age than velocity. Attained weight for age at four months was a more sensitive predictor of diminished weight for age (<10^th percentile) at 24 months (sensitivity=100%, 95%CI=(87, 100%)) than weight velocity (sensitivity=40%, 95%CI=(23, 59%)). Attained length at four months was more sensitive to detecting subsequent stunting (<10^th percentile length for age) (77%, 95%CI=(62, 87%)) than length velocity (30%, 95%CI=(19, 45%)).

Conclusions

In CF infancy, attained weight or length are more sensitive than velocity-based definitions for predicting subsequent diminished growth.

BACKGROUND

Cystic fibrosis (CF) is an autosomal recessive genetic disease caused by a mutation in the CF transmembrane conductance regulator (CFTR) gene (1). Malabsorption and malnutrition are among the earliest disease manifestations (2, 3). Inadequate weight gain and growth during infancy and childhood are important indicators of subsequent growth delay (4, 5), cognitive development (6) and morbidities in a myriad of disease settings (7–9). Growth deficiencies in CF, as early as 2 years of age (10), have been associated with both decreased lung function and survival (11, 12), with those falling below the 10^th percentile in weight for age having the poorest prognosis (13). Recently published CF Foundation (CFF) infant care guidelines (14) recommend the use of age and gender specific expected weight velocity to evaluate growth throughout infancy (15); and while there are other proponents of using incremental growth measures (i.e. rate of change or velocity) as an alternative to attained growth for age cut-points to monitor growth (15, 16), there is no consensus regarding criteria for defining growth failure (17), nor has growth velocity ever been formally characterized in the CF population.

Early monitoring of nutrition, weight and length is a foundational component of clinical care for at-risk infants to prevent growth deficiencies before they occur (18). In this study, we sought to compare traditional attained growth for age with velocity metrics, and evaluated the ability of these different measures in early infancy to predict growth at 24 months of age in a cohort of infants born with CF. Two sets of standards are available for assessing incremental growth in infancy: the US Center for Disease Control recommended (19) World Health Organization’s (WHO) infant growth standards (20, 21), representing international current feeding and nutrient supplementation practices, and the Guo et. al. US (Guo-US) velocity standards (15), recommended by the CFF infant care guidelines (14). In this paper we described and compared both sets of standards for weight and length velocity in addition to attained growth for age in infants with CF in an attempt to identify the most sensitive, early indicators of growth failure at 24 months of age. We hypothesized that velocity measures would be more sensitive than attained weight and length for predicting subsequent growth deficits.

METHODS

Study population and variables

This retrospective cohort study utilized the US CFF Patient Registry which contains longitudinal data on more than 27,000 patients at over 110 accredited CF care centers (22). We included newborns diagnosed with CF who born between Jan 1, 2004 through Dec 31, 2008 and entered in the registry before 4 months of age. Follow-up data through Dec 31, 2009 were included to allow for two year follow-up on most individuals. The Institutional Review Board at Seattle Children’s Hospital, Seattle, Washington approved this study.

Encounter based weight and length measures were used for calculating weight and length velocity as change in grams and centimeters per unit time, respectively. The velocities were then standardized to the WHO cohort (23) and the Guo-US standards to generate age and sex specific percentiles. WHO provides weight and length velocity charts for increments of 2, 3, 4, and six months duration (1 month increments are also given for weight velocity only) at every age interval from zero to 24 months. Guo-US provides weight standards at 1 month increments from 0 to 6 months of age, 2 month increments from 0 to 12 months of age, and 3 month increments from 0 to 24 months; length standards are published for 2 month increments from 0 to 6 months of age and 3 month increments from 0 to 24 months. We utilized only the WHO 1, 2, and 3 month increments that correspond with the Guo-US reference data. There are no guidelines for handling anthropometrics measured between ages or at imprecise increment lengths. To limit bias and noise while retaining enough data to perform the analysis, we restricted all analyses to include only observations where a patient’s weight or length was recorded at an age no more than ±9 days (30% of a month) from expected age at interval start (0–23 months), and follow-up measures of weight or length recorded at 1 or 2 or 3 months ±6 days later (20% of a month). More details in Supplemental Methods section online. Attained weight and length for age were scored to the WHO standard growth charts (20).

The 10^th percentile of weight for age (12, 13, 18), and length for age (13) were chosen as the primary measures of growth failure at 24 months. The following thresholds were chosen as potential early markers of deficits at 24 months in order to compare velocity to attained growth for age and identify the most sensitive predictor: 1) Guo-US 50^th percentile for weight velocity (CFF Infant Care guidelines recommendation (14)) and length velocity, 2) WHO standardized 2.5^th, 5^th, 10^th, and 50^th percentiles for weight and length velocity, 3) WHO standardized 2.5^th, 5^th, and 10^th percentiles for weight and length for age. For the velocity measures, increment durations (1, 2 or 3 month) were pooled by age interval and individuals who fell below the threshold for any of the increments were considered to have met the criteria. The 50^th percentile threshold was only applied to the velocity measures and was chosen as the closest comparator to the CFF Infant Care Guidelines’ recommendation. The more conservative 2.5% threshold for defining growth deficiency is recommended by WHO, acknowledging that their standard cohort of infants is considered optimal as opposed to typical (24). The 5^th and 10^th percentile thresholds for growth deficiency are commonly used clinically and reported in the literature (17, 25, 26).

Statistics

Rates of growth deficiency by each definition were tabulated cross-sectionally at each month and reported as a percentage. McNemar’s test for paired binary data was used to test for differences between velocity and attained growth failure rates averaged over the first year to avoid multiple testing, and Guo-US versus WHO standards at each month. Sensitivity and specificity estimates were used to examine the association between the early metrics and weight and length for age at 24 ±1 months of age (defined as < WHO 10^th percentile) in the subset of data with eligible velocity measures. Wilson’s 95% confidence intervals (CI) were calculated for sensitivity and specificity measures. Analyses were performed using SAS (version 9.2, SAS Institute Inc., Cary, NC, 2009), and R (version 2.15, The R Foundation for Statistical Computing, Vienna, Austria, 2012).

RESULTS

The cohort was comprised of 1,992 infants from whom over 11,000 length and weight measurements were derived from the CFF Patient Registry from 2004 through 2009 meeting eligibility criteria. This group had a median of 4 visits (interquartile range = 3–6) where either weight or length velocity were calculated (Supplement Figure 1). The demographic characteristics of the cohort were: 50.6% male, 91.9% Caucasian, 9.6% Hispanic, and 47.7% homozygous F508del. Newborn screening was responsible for 49.6% of diagnoses. Meconium ileus was present in 26.5%, and 90.1% of the cohort were classified pancreatic insufficient as defined by use of enzyme replacement therapy. At least 9 months of follow-up was observed for 98.9% of the infants and 85.0% had 2 years of data in the Registry. Characteristics of the analysis set did not differ from the overall sample of babies born during this time period (Supplement Table 1), and weight and length measurements meeting the specific velocity calculation criteria were similar to those from the entire set (Supplement Figure 2).

Comparison of Attained Weight and Length for Age to Velocity

In this analysis, we examined the percentage of infants with CF who met traditional definitions of growth deficiency with attained weight or length less than the 10^th, 5^th or 2.5^th percentiles for age. This was compared with corresponding percentages using the WHO velocity standard percentiles. Figure 1 and Table 1 show that the majority of anthropometric deficits occur at the youngest ages. Attained weight for age identified more at risk infants (<10^th percentile) on average during the first year than the equivalent velocity based measure (26.8%; 95%CI=(25.6, 28.1%); p <0.001). Similarly, length for age failure rates were on average 27.3% (95% CI = (25.6, 29.1%); p <0.001) higher during the first year than the corresponding 10^th percentile threshold for WHO length velocity. Supplemental material online and Supplemental Figure 3 provides a comparison of the US versus WHO velocity standards in this population.

Figure 1.

Percent of the CF infant cohort falling below attained growth for age (solid lines) and growth velocity (dashed lines) thresholds. The 10^th, 5^th, and 2.5^th WHO standard percentiles (thickest, medium, and thinnest lines, respectively) for each the attained and velocity growth charts at 2, 4, 6, 12, 18 and 24 months. All increments for velocity measures (1, 2 and 3 month) are pooled by age interval.

Table 1.

Percentage of cohort in nutritional failure by age for each anthropometric outcome, and thresholds for defining growth failure

Outcome	Threshold	Age (months)
		2 (n=425)	4 (n=741)	6 (n=671)	12 (n=374)	18 (n=403)	24 (n=317)
Weight for age	< 10%	44.2	43.3	34.7	11	6.9	6.9
	< 5%	33.6	33.5	25.5	8.3	3.7	2.8
	< 2.5%	25.2	26.5	18.6	3.7	2.5	1.9
Weight Velocity (Guo -US)	< 50%	61.4	50.6	42	48.1	56.1	59.3
Weight Velocity (WHO)	< 50%	75.1	52.9	38.2	30.5	45.7	51.1
	< 10%	31.5	18.5	12.5	6.4	11.2	18.3
	< 5%	21.2	12	10.4	4.5	6.7	12.6
	< 2.5%	14.1	8.2	8.2	3.5	3.2	8.2
		(n=60)	(n=494)	(n=520)	(n=164)	(n=403)	(n=317)
Length for age	< 10%	30	37.9	32.7	26.8	23.3	24.9
	< 5%	27	26.9	26	17.7	15.9	17
	< 2.5%	20	20.2	17.9	10.4	12.2	11.4
Length Velocity (Guo-US)	< 50%	40	56.5	41.9	45.7	53.1	57.4
Length Velocity (WHO)	< 50%	72	54.3	38.3	38.4	53.3	58.7
	< 10%	43	23.3	17.5	20.7	25.8	30.3
	< 5%	38	18.8	14	13.4	19.9	24.3
	< 2.5%	33	16	9.6	13.4	16.9	19.9

Association between Early Growth deficiencies and Subsequent Weight and Length Deficiencies

At 24 months of age 6.9% fall below the 10^th percentile for weight and 24.9% fall below the 10^th percentile for length (Table 1). Among those ‘at risk’ in weight, 81.8% were also failing in length; whereas among those failing in length only 22.8% were also failing in weight. At 24 months 5.7% fall below both weight and length thresholds. Table 2 shows the sensitivity and specificity of growth failure information at early ages (4, 6, 12, and 18 months) to predict weight for age below the 10^th percentile at 24 months of age. Attained weight and length for age growth deficiencies at 4 months (using 2.5^th, 5^th and 10^th percentile thresholds) are sensitive predictors of weight for age <10^th percentile at 24 months (range = 76% – 100%) with specificity range of 56% to 83%. Sensitivity measures were highest at 4 months of age. Figure 2A shows that attained weight for age (<10^th percentile) at 4 months was a more sensitive predictor of diminished weight at 24 months (sensitivity=100%, 95%CI=(87, 100%)) than weight velocity (<10^th percentile) at 4 months (sensitivity=40%, 95%CI=(23, 59%)). A higher weight velocity threshold at 4 months (< WHO 50^th percentile) improves the sensitivity (68%, 95%CI=(48, 82%)) and specificity (42%, 95% CI=(37, 47%)), but it remains less sensitive than weight for age (p=0.04).

Table 2.

Sensitivity and Specificity for measures and timepoints to predict weight for age at 24 months less than the World Health Organization (WHO) 10^th percentile

Predictor	Threshold	Sensitivity				Specificity
		Age (months)				Age (months)
		4 (n=353)	6 (n=343)	12 (n=146)	18 (n=81)	4 (n=353)	6 (n=343)	12 (n=146)	18 (n=81)
Weight for age	< 10%	100	91	64	33	56	63	83	94
	< 5%	96	82	45	33	68	73	93	95
	< 2.5%	92	76	36	33	78	81	95	95
Weight Velocity (Guo-US)	< 50%	68	56	55	100	46	55	40	33
Weight Velocity (WHO)	< 50%	68	56	45	100	42	60	67	50
	< 10%	40	26	18	0	81	84	90	91
	< 5%	32	21	18	0	88	88	91	94
	< 2.5%	24	15	18	0	92	90	93	95
		(n=223)	(n=229)	(n=72)	(n=66)	(n=223)	(n=229)	(n=72)	(n=66)
Length for age	< 10%	94	85	25	67	67	68	74	83
	< 5%	88	80	25	67	76	77	85	89
	< 2.5%	76	65	25	33	83	85	91	97
Length Velocity (Guo-US)	< 50%	65	75	25	33	46	56	50	46
Length Velocity (WHO)	< 50%	59	70	25	33	48	60	56	46
	< 10%	24	30	25	33	78	80	76	78
	< 5%	24	25	25	33	82	84	85	81
	< 2.5%	24	20	25	33	84	89	87	87

Figure 2.

Sensitivity and Specificity of 4 month anthropometric measures to predict weight (3A) and length (3B) for age at 24 months (falling below WHO 10^th percentile). Weight/length for age < WHO 10^th percentile at 4 months (square), weight/length velocity < WHO10^th percentile at 4 months (closed triangle), weight/length velocity < WHO 50^th percentile at 4 months (open triangle). Lines are 95%CI.

Attained length for age growth deficiencies (<10^th percentile) at 4, 6, 12 and 18 months are sensitive predictors of length for age <10^th percentile at 24 months (range = 73% – 83%) with specificity range of 72% to 91%. (Supplemental Table 2). Figure 2B indicates that attained length <10^th percentile at 4 months was more sensitive to detecting subsequent length failure (sensitivity=77%, 95%CI=(62, 87%)) than length velocity (30%, 95%CI=(19, 45%)). As with prediction of weight at 24 months, a higher length velocity threshold at 4 months (50^th percentile) improves sensitivity (63%, 95%CI=(48, 76%)) to detect length failure at 24 months, but at the expense of sensitivity (49%, 95%CI=(49, 57%)). Falling below the 10^th percentile in weight for age at 4 months was also a sensitive indicator of length faltering at 24 months (81%, 95% CI= (71, 88%)).

CONCLUSIONS

This retrospective analysis of growth data from nearly 2000 infants with CF failed to support this study’s hypothesis and the current CFF guideline’s assertion (14), that velocity based definitions of growth deficiencies are more sensitive to identifying early growth failure and predicting subsequent growth than traditional measures of growth attainment by age. Furthermore, the results indicate that early growth faltering (at four months of age) is most predictive of later growth deficiencies. While there have been assessments of failure to thrive in CF childhood and infancy (25–27) this is the first study we know of to evaluate the two velocity standards (WHO and Guo-US) with respect to attained growth and compare their sensitivity to detecting subsequent growth deficiency in CF infants.

During the first year of life, inadequate attained weight and length for age indicate greater deficits than thresholds based on weight and length velocity. Because adverse health outcomes are strongly associated with poor growth in infants with CF, the most sensitive measures should be used to prevent long-term consequences. Why is there a discrepancy in these two measures? We hypothesize that pancreatic enzyme replacement therapy and nutritional supplements lead to increased (”catch-up”) velocity yet it is inadequate to completely correct the deficit which is measured by attained growth by age. In addition, the high correlation between repeated weight and length for age percentiles in an individual (28) compared to the variability of velocity measures within an infant (21) could also account for the high sensitivity when using attained weight and length to predict growth at 24 months.

The CFF Infant Care Guidelines recommend using the medians for weight velocity to monitor infant growth based on US standards (14). We showed that the WHO velocity standards performed similarly to the US standards in their ability to predict growth deficits at 24 months, but neither was preferable to attained growth measures at the earlier time points. The WHO velocity standards were not available at the time the CFF Infant Care Guidelines were written; however, comparing the CF infant population using both methods indicate that the WHO standards are less likely to identify sub-optimal (<50^th percentile) weight gain throughout infancy. The differences in geographic and demographic composition (29) or statistical methodologies (15, 21) between the two standards may account for these findings.

Weight and length at 24 months of age is a benchmark because of its clinical implications in CF (10, 18) and other childhood diseases (7). Weight for length or body mass index were not considered in this population because stunting leads the weight for length measures to look normal or above average (12). The WHO 10^th percentile is a conservative criterion and has been shown to be associated with subsequent pulmonary function (12, 13, 18); choosing a more strict definition (the 5^th or 2.5^th percentile) would result in less stable estimates of sensitivity because so few would be considered poor growers at 24 months. The 50^th percentile was not evaluated as a 24 month endpoint because this is considered the ideal state, not an indicator of nutritional failure. The finding that 4 month anthropometric measures are most predictive of 24 month weight and length suggests that deficiency is pre-determined; however the question remains whether this result is driven by genetic or environmental factors (30). The discordance at 24 months between those below the 10^th percentile in length and those ‘at risk’ in weight suggests that there are modifiable as well as genetic factors at play. Given the advent of newborn screening, prioritization of nutritional interventions in the first few months of life may be a window of opportunity and these findings suggest that careful study of aggressive nutritional management early in life is warranted.

This study is based on an encounter based registry, where sicker, smaller infants may be more likely to come to regular and additional clinic visits, introducing a selection bias. However, clinic attendance is particularly high during infancy in the majority of CF care centers (22), and our strict age and increment requirements may have alleviated this bias. Great care was taken to ensure that age and length of the increments used to calculate velocity were precise, and doing so limited the analysis to a small subset of the cohort. Comparison of this subset to all of the anthropometric data does not indicate a bias or lack of generalizability. Neither the Guo-US nor the WHO documentation provides guidance on interpolating scores for an infant. For example, if a child is 3.5 months of age at baseline and 5.9 months old at the next clinic or study visit, should the child be assessed using the intervals that begin at three or four months of age? Subsequently, should the two or the three month increment standard be used? One month increments are known to be highly variable (21), but otherwise optimal increment length to evaluate rate of change is unknown. We employed only the incremental WHO standards that matched the frequency of the Guo-US standards for purpose of comparison; more research into the behavior of velocity measures under specific conditions may better inform clinicians and researchers in other disease settings.

Attained weight and length for age and velocity appear to highlight different aspects of anthropometric deficiencies among a large multi-center cohort of CF infants. Identification of growth deficiency in infancy by attained weight and length are more sensitive as clinical predictors of future growth faltering when compared to velocity based measures. Validation in an independent cohort is needed to confirm these findings and a prospective, observational study of CF infant growth and nutrition currently in progress will afford additional evidence to support or refute these results. Further research is under way to examine whether the utility of velocity versus attained growth are modified by CF complications or other factors known to impact growth such as pancreatic status, meconium ileus, and diagnosis through newborn screening. Our study demonstrates the importance of focusing on attained growth, particularly in the first few months of life, in infants with CF.