Comparing Attained Weight and Weight Velocity during the First 6 Months in Predicting Child Undernutrition and Mortality

ABSTRACT

Background

The first 6 mo of life are critical for subsequent risk of undernutrition and mortality. The predictive abilities of attained weight at the end of each month and monthly weight velocity for undernutrition and mortality need to be compared.

Objectives

This study aimed to examine the predictive abilities of different weight metrics during the first 6 mo of life in predicting undernutrition and mortality.

Methods

This study used a cohort of infants in Tanzania. Weight and length were measured monthly from birth to 18 mo of age. Three weight metrics during the first 6 mo of life were considered as predictors, including attained weight-for-age z score (WAZ) at the end of each month, monthly change in WAZ, and monthly weight velocity z score (WVZ). Logistic models were used with undernutrition (at 6 or 12 mo) and mortality (over the first 18 mo) as outcomes. AUC values were compared across metrics.

Results

For predicting wasting at 6 mo, WVZ (AUC: 0.80) had a greater predictive ability than attained WAZ (AUC: 0.76) and change in WAZ (AUC: 0.71) during the second month of life. After 2 mo, attained WAZ (AUC: 0.81–0.89) had greater predictive abilities than WVZ (AUC: 0.71–0.77) and change in WAZ (AUC: 0.65–0.67). For predicting stunting at 6 mo, attained WAZ (AUC: 0.75–0.79) had consistently greater predictive abilities than WVZ (AUC: 0.56–0.66) and change in WAZ (AUC: 0.50–0.57). The weight metrics had similar abilities in predicting mortality, with the AUC rarely reaching >0.65.

Conclusions

Attained weight at the end of each month had greater abilities than monthly weight velocity in the same month in predicting undernutrition. Attained weight remains a useful indicator for identifying infants at greater risk of undernutrition.

Introduction

The global progress in improving linear growth and reducing wasting and stunting for children <5 y old remains slow (1). In resource-limited countries, the distributions of weight-for-height and height-for-age are still shifted toward undernutrition (1). Child undernutrition remains a major global health problem, and an array of maternal and child interventions are effective in treating child undernutrition (2). However, scarce evidence and guidance exist on preventing growth faltering as early as possible in low- and middle-income countries. In the United Republic of Tanzania, 34% of children <5 y of age had stunting, and 5% had wasting in 2015–2016 (3). Partially due to the high burdens of child undernutrition, Tanzania's <5-y-old child mortality rate of 67/1000 live births (3) is one of the highest in sub-Saharan Africa.

The period from conception to the second birthday (i.e., the first 1000 d of life) is commonly referred to as “the window of opportunity” for ensuring adequate growth and development (4). Whereas the prevalence of growth faltering peaks at ∼24 mo of age (5), recent evidence shows that the incidence of wasting and stunting also reaches a peak within the first 3 mo of life (1, 6). Many young children may not recover from this early undernourishment and may enter later childhood with elevated risks of undernutrition, morbidity, and death (6–8). Therefore, the first 6 postnatal months is an especially critical period for the early identification of children at greater risk (9). The timely identification of future risk of undernutrition and mortality will, in turn, inform the initiation of intensive support for successful breastfeeding or alternative interventions such as micronutrient supplementation and supplementary feeding (2, 10–12).

Attained weight (i.e., weight measured at specific time points) is a summary measure of growth history up to a specific point. Weight velocity (i.e., change in weight between 2 time points) reflects the more recent growth trajectory (13, 14). Both attained weight and weight velocity have been suggested as markers of undernutrition, and previous studies have compared the abilities of attained growth and growth velocity in predicting child undernutrition (13–16) or mortality (17–21) later in childhood. However, only a few studies (14, 20, 21) have derived growth metrics using established growth standards. Further, to the best of our knowledge, no studies have compared the attained weight at the end of each month with the weight velocity during the same month.

The first 6 mo of life is a critical period for growth surveillance and intervention initiation, and a granular assessment of the predictive abilities of attained weight and weight velocity is needed. Therefore, using data from a cohort of infants enrolled from urban Tanzania, we aimed to examine the abilities of different weight metrics during the first 6 mo of life in predicting undernutrition and mortality.

Methods

Study design and study population

This study used data from a randomized, double-blind, placebo-controlled trial in Tanzania. The primary aim of the trial was to evaluate the effects of prenatal maternal multiple micronutrient supplementation (MMS) on fetal loss, low birth weight, and preterm birth. Pregnant women who attended antenatal clinics in Dar es Salaam, Tanzania, were recruited between 2001 and 2004. Around the time when the study was conducted, >99% of pregnant women in Dar es Salaam received antenatal care from a skilled provider, 92% had ≥4 antenatal visits, and 94% delivered in health facilities (22).

All participants of this study were confirmed to be seronegative for HIV infection based on antibody detection tests. After recruitment during the second trimester (12–27 weeks of gestation), women were randomly assigned to receive daily oral MMS or placebo until 6 wk postpartum. The supplement included 8 vitamins (thiamin, riboflavin, niacin, pyridoxine, folic acid, cyanocobalamin, ascorbic acid, and tocopherol); the specific composition and dosage have been described elsewhere (23). The study enrolled 8428 eligible women, of whom 6 died, and 43 were lost to follow-up before delivery. Among the remaining 8379 women with data on birth outcome, 156 were pregnant with twins or triplets, resulting in 8223 women giving birth to singleton infants, 343 of whom died during the follow-up period. Among the 8223 singleton infants, 440 did not have any postnatal anthropometric measurements, for 33 of whom this was due to deaths during the first month of life. The current analysis focuses on the 7783 singleton infants who had any postnatal anthropometry data available (Supplemental Figure 1), with varying sample sizes applicable to each specific analysis. The study was approved by the institutional review boards at Muhimbili University of Health and Allied Sciences and Harvard TH Chan School of Public Health. All women enrolled in the study provided written, informed consent to participate.

Data collection

Women (before birth) and mother–child dyads (after birth) had monthly study visits at the clinic (or were followed up at home for missed clinic visits), during which nutritional and health outcomes were assessed until ≤18 mo of age. Trained research midwives measured birth weight at delivery to the nearest 10 g using digital scales. Child postnatal anthropometry measures, including weight and length, were collected by trained study nurses using calibrated instruments and standard operating procedures. Child weight was measured using a weighing scale with the child placed on the scale. Child length was measured using a length board placed horizontally on the ground, with the child gently placed on the board with their head against the headrest. Weight and length measures were taken twice, and any discrepancy was resolved by a third measurement. Child mortality was ascertained using verbal autopsy forms during follow-up visits. The research midwives and nurses were monitored by study supervisors. All study staff, including research midwives, research nurses, and study supervisors, received training and periodic retraining on data collection and quality assurance.

Statistical analysis

We computed the attained child growth z scores, including weight-for-age z score (WAZ), length-for-age z score (LAZ), and weight-for-length z score (WLZ), at birth and each monthly measurement, using established child growth standards. Specifically, we computed z scores at birth using the INTERGROWTH-21^st newborn size standards (24). For term-born children, we computed postnatal attained z scores using the WHO 2006 child growth standards (25) and the WHO's SAS macro (26). For infants born preterm (i.e., before 37 completed weeks of gestation), we computed postnatal attained z scores using the INTERGROWTH-21^st standards for postnatal growth of preterm infants (27). Postnatal z scores that were outliers were set to missing based on the WHO flagging system, defined as WAZ < −6 or >5, LAZ < −6 or >6, and WLZ < −5 or >5 (28); 0.08% (66 records) of WAZ values, 0.28% (241 records) of LAZ values, and 0.35% (296 records) of WLZ values were set to missing accordingly. Because monthly weight measurements might not occur on the exact date corresponding to the end of the month, we treated the measurements within 14 d of the target date as a proxy for the measurement for that month (e.g., WAZs measured between 77 and 105 d were treated as the WAZ at 3 mo). This approximation would have minimal impacts on the analyses because the postnatal child growth standards (WHO and INTERGROWTH-21^st) accounted for the date of the actual measurements in the calculation of z scores.

We calculated monthly weight velocity z scores (WVZs) using the WHO 2009 child growth velocity standards (29). For the weight velocity standards, the target dates of the measurements from 1 to 6 mo were 28, 61, 91, 122, 152, and 183 d, respectively, with a tolerable range of ±3 d (29). Weight measurements taken outside this tolerable range yet still within 14 d of the target date were corrected to the target date using linear interpolation (29, 30). We used this strategy because the WHO child growth velocity standards do not automatically account for the exact timing of weight measurements. In addition to the WHO flagging system used for attained weights, for the analysis of monthly changes of WAZ and monthly weight velocity, intervals spanning <15 d or >45 d were removed (30); 1 child, 2 children, 64 children, 46 children, 22 children, and 51 children had their intervals for the analysis of birth to 1 mo, 1–2 mo, 2–3 mo, 3–4 mo, 4–5 mo, and 5–6 mo removed from the analysis, respectively.

We considered the following undernutrition outcomes: wasting, severe wasting, stunting, and concurrent wasting and stunting, measured at 6 mo and 12 mo of age. Wasting was defined as WLZ < −2; severe wasting was defined as WLZ < −3; stunting was defined as LAZ < −2; and concurrent wasting and stunting was defined as the simultaneous presence of wasting and stunting. All 4 undernutrition outcomes were treated as dichotomous variables. We chose these as the primary undernutrition outcomes because they are the major predictors of mortality and remain a high burden in low- and middle-income countries (1, 6, 31). For child mortality, we considered all-cause mortality that occurred during various time intervals of the first 18 mo of life, including between 1 wk and 3 mo, between 3 and 6 mo, between 6 and 12 mo, and between 12 and 18 mo.

We compared the predictive abilities of 3 categories of early-life weight metrics—attained WAZ at the end of each month, monthly changes in attained WAZ, and monthly WVZ—all measured during the first 6 mo of life. We treated all weight metrics as continuous variables. We used logistic regression models with weight metrics as the predictors and undernutrition (at 6 or 12 mo) and mortality (over the first 18 mo) as the outcomes. To facilitate the comparison of predictive abilities across metrics, we computed the AUC of the receiver operating characteristic for each predictor–outcome combination, with a higher (i.e., closer to 1.0) AUC indicating a greater predictive ability and a lower (i.e., closer to 0.5) AUC indicating a weaker predictive ability (32). For the undernutrition outcomes at 6 mo, we conducted subgroup analyses by 1) birth weight for gestational age [small for gestational age (SGA) compared with appropriate/large for gestational age (AGA/LGA)]; 2) gestational age at birth (preterm birth compared with term birth); and 3) birth weight (<3000 g or ≥3000 g). We used 3000 g instead of 2500 g as the cutoff because the prevalence of low birth weight (<2500 g) was very low (<7%) in the study population. All analyses were conducted using SAS version 9.4 (SAS Institute Inc.).

Results

Table 1 summarizes the general characteristics of the children and their mothers. Of the infants, 48% were female. The proportions of preterm births and SGA births were 17%, and the proportion of low birth weight was 6.6% in the analytical sample. At 6 mo of age, the prevalences of wasting, severe wasting, stunting, and concurrent wasting and stunting were 3.9%, 0.88%, 10.6%, and 0.40%, respectively; at 12 mo of age, these prevalences were 6.8%, 1.2%, 21.7%, and 2.4%, respectively. Approximately 4% of the infants died during the follow-up, and 49% of the deaths occurred during the first week of life.

TABLE 1.

Maternal and child characteristics among infants in Dar es Salaam, Tanzania (2001–2006)¹

Characteristics	Values
Maternal characteristics
Women, n	7783
Maternal age at enrollment, y	25.2 ± 5.1
Maternal education,2 y
0–4	877 (11.3)
5–7	5177 (66.8)
8–11	1299 (16.8)
≥12	393 (5.1)
Maternal marital status3
Married or cohabiting	6846 (88.6)
Single	877 (11.4)
Maternal occupation4
Housewife	5604 (74.6)
Employed	1908 (25.4)
Gestational age at birth, wk	39.4 ± 3.3
Child characteristics
Children, n	7783
Girls5	3725 (48.3)
Preterm birth6	1315 (16.9)
Small for gestational age7	1160 (17.3)
Large for gestational age8	973 (14.5)
Birth weight,9 g	3134.1 ± 503.2
Low birth weight10	492 (6.6)
Undernutrition at 6 mo11
Wasting	226 (3.9)
Severe wasting	51 (0.88)
Stunting	612 (10.6)
Concurrent wasting and stunting	23 (0.40)
Undernutrition at 12 mo12
Wasting	213 (6.8)
Severe wasting	36 (1.2)
Stunting	682 (21.7)
Concurrent wasting and stunting	75 (2.4)
Child mortality13	301 (3.9)
Child mortality during the first week of life14	147 (48.8)
Child mortality between 1 wk and 3 mo14	30 (10.0)
Child mortality between 3 and 6 mo14	26 (8.6)
Child mortality between 6 and 12 mo14	67 (22.3)
Child mortality between 12 and 18 mo14	31 (10.3)

¹Values are means ± SDs for continuous variables and n (%) for categorical variables. LAZ, length-for-age z score; WLZ, weight-for-length z score.

²Maternal education was missing for 37 women.

³Maternal marital status was missing for 60 women.

⁴Maternal occupation was missing for 271 women.

⁵Child sex was missing for 68 children.

⁶Preterm birth was defined as gestational age at live birth <37 wk.

⁷Birth weight < 10^th percentile for gestational age based on the INTERGROWTH-21^st newborn size standards; missing for 1057 children.

⁸Birth weight > 90^th percentile for gestational age based on the INTERGROWTH-21^st newborn size standards; missing for 1057 children.

⁹Birth weight was missing for 361 children.

¹⁰Low birth weight was defined as birth weight <2500 g; missing for 361 children.

¹¹Stunting, wasting, and severe wasting were defined as LAZ < −2, WLZ < −2, and WLZ < −3, respectively; information on wasting, stunting, and concurrent wasting and stunting at 6 mo was missing for 1994, 1985, and 1994 children, respectively.

¹²Stunting, wasting, and severe wasting were defined as LAZ < −2, WLZ < −2, and WLZ < −3, respectively; information on wasting, stunting, and concurrent wasting and stunting at 12 mo was missing for 4639, 4636, and 4641 children, respectively.

¹³Information on child mortality was missing for 3 children.

¹⁴The percentages were among children who died during the follow-up period.

For the prediction of wasting at 6 mo (Figure 1), during the first month of life (Figure 1A), WVZ and attained WAZ (P = 0.93), and change in WAZ and attained weight (P = 0.10), had similar predictive abilities, yet WVZ had a greater predictive ability than the change in WAZ (AUC: 0.66 compared with 0.61; P < 0.001). During the second month (Figure 1B), WVZ (AUC: 0.80) had a greater predictive ability than attained WAZ (AUC: 0.76; P = 0.034) and change in WAZ (AUC: 0.71; P < 0.001), whereas attained WAZ and change in WAZ had similar predictive abilities (P = 0.12). After 2 mo (Figure 1C–E), attained WAZ (AUC: 0.81–0.89) had greater predictive abilities than WVZ (AUC: 0.71–0.77), which in turn had greater predictive abilities than the change in WAZ (AUC: 0.65–0.67).

FIGURE 1.

ROC curves of weight metrics during the first 6 mo of life and wasting at 6 mo among infants in Dar es Salaam, Tanzania (2001–2006). (A) Comparison of attained WAZ at 1 mo with change in WAZ and WVZ from birth to 1 mo; (B) comparison of attained WAZ at 2 mo with change in WAZ and WVZ from 1 to 2 mo; (C) comparison of attained WAZ at 3 mo with change in WAZ and WVZ from 2 to 3 mo; (D) comparison of attained WAZ at 4 mo with change in WAZ and WVZ from 3 to 4 mo; (E) comparison of attained WAZ at 5 mo with change in WAZ and WVZ from 4 to 5 mo. Attained z scores at birth were computed using the INTERGROWTH-21^st newborn size standards. For term-born children, postnatal attained z scores were computed using the WHO 2006 child growth standards (25). For infants born preterm, postnatal attained z scores were computed using the INTERGROWTH-21^st standards for postnatal growth of preterm infants. Wasting was defined as weight-for-length z score < −2. ROC, receiver operating characteristic; WAZ, weight-for-age z score; WVZ, weight velocity z score.

For the prediction of severe wasting at 6 mo (Figure 2), during the first month (Figure 2A), WVZ and attained WAZ (P = 0.94), change in WAZ and attained WAZ (P = 0.48), and WVZ and change in WAZ (P = 0.065) had similar predictive abilities. During the second (Figure 2B) and third (Figure 2C) months, WVZ and attained weight, and change in WAZ and attained WAZ, had similar predictive abilities, whereas WVZ had greater predictive abilities than the change in WAZ (P = 0.0092 in the second month and P = 0.0016 in the third month). After 3 mo (Figure 2D, E), attained WAZ (AUC: 0.85–0.88) had greater predictive abilities than WVZ (AUC: 0.71–0.76), which in turn had greater predictive abilities than the change in WAZ (AUC: 0.65–0.68).

FIGURE 2.

ROC curves of weight metrics during the first 6 mo of life and severe wasting at 6 mo among infants in Dar es Salaam, Tanzania (2001–2006). (A) Comparison of attained WAZ at 1 mo with change in WAZ and WVZ from birth to 1 mo; (B) comparison of attained WAZ at 2 mo with change in WAZ and WVZ from 1 to 2 mo; (C) comparison of attained WAZ at 3 mo with change in WAZ and WVZ from 2 to 3 mo; (D) comparison of attained WAZ at 4 mo with change in WAZ and WVZ from 3 to 4 mo; (E) comparison of attained WAZ at 5 mo with change in WAZ and WVZ from 4 to 5 mo. Attained z scores at birth were computed using the INTERGROWTH-21^st newborn size standards. For term-born children, postnatal attained z scores were computed using the WHO 2006 child growth standards (25). For infants born preterm, postnatal attained z scores were computed using the INTERGROWTH-21^st standards for postnatal growth of preterm infants. Severe wasting was defined as weight-for-length z score < −3. ROC, receiver operating characteristic; WAZ, weight-for-age z score; WVZ, weight velocity z score.

For the prediction of stunting at 6 mo (Figure 3), attained WAZ (AUC: 0.75–0.79) had consistently greater predictive abilities than WVZ (AUC: 0.56–0.66), which in turn had greater predictive abilities than the change in WAZ (AUC: 0.50–0.57). For the prediction of concurrent wasting and stunting at 6 mo (Figure 4), during the first month (Figure 4A), WVZ and attained WAZ (P = 0.47), change in WAZ and attained WAZ (P = 0.68), and WVZ and change in WAZ (P = 0.061) had similar predictive abilities. During the second month (Figure 4B), WVZ and attained weight (P = 0.10), and change in WAZ and attained WAZ (P = 0.53), had similar predictive abilities, whereas WVZ had a greater predictive ability than the change in WAZ (AUC: 0.94 compared with 0.84; P = 0.021). After 2 mo (Figure 4C–E), attained WAZ (AUC: 0.95–0.99) had greater predictive abilities than WVZ (AUC: 0.77–0.87), which in turn had greater predictive abilities than the change in WAZ (AUC: 0.63–0.73). Similar results were found for undernutrition at 12 mo, although the AUC values for all weight indexes were lower for predicting undernutrition at the more distant time point of 12 mo than at 6 mo (Supplemental Tables 1, 2).

FIGURE 3.

ROC curves of weight metrics during the first 6 mo of life and stunting at 6 mo among infants in Dar es Salaam, Tanzania (2001–2006). (A) Comparison of attained WAZ at 1 mo with change in WAZ and WVZ from birth to 1 mo; (B) comparison of attained WAZ at 2 mo with change in WAZ and WVZ from 1 to 2 mo; (C) comparison of attained WAZ at 3 mo with change in WAZ and WVZ from 2 to 3 mo; (D) comparison of attained WAZ at 4 mo with change in WAZ and WVZ from 3 to 4 mo; (E) comparison of attained WAZ at 5 mo with change in WAZ and WVZ from 4 to 5 mo. Attained z scores at birth were computed using the INTERGROWTH-21^st newborn size standards. For term-born children, postnatal attained z scores were computed using the WHO 2006 child growth standards (25). For infants born preterm, postnatal attained z scores were computed using the INTERGROWTH-21^st standards for postnatal growth of preterm infants. Stunting was defined as length-for-age z score < −2. ROC, receiver operating characteristic; WAZ, weight-for-age z score; WVZ, weight velocity z score.

FIGURE 4.

ROC curves of weight metrics during the first 6 mo of life and concurrent wasting and stunting at 6 mo among infants in Dar es Salaam, Tanzania (2001–2006). (A) Comparison of attained WAZ at 1 mo with change in WAZ and WVZ from birth to 1 mo; (B) comparison of attained WAZ at 2 mo with change in WAZ and WVZ from 1 to 2 mo; (C) comparison of attained WAZ at 3 mo with change in WAZ and WVZ from 2 to 3 mo; (D) comparison of attained WAZ at 4 mo with change in WAZ and WVZ from 3 to 4 mo; (E) comparison of attained WAZ at 5 mo with change in WAZ and WVZ from 4 to 5 mo. Attained z scores at birth were computed using the INTERGROWTH-21^st newborn size standards. For term-born children, postnatal attained z scores were computed using the WHO 2006 child growth standards (25). For infants born preterm, postnatal attained z scores were computed using the INTERGROWTH-21^st standards for postnatal growth of preterm infants. Wasting and stunting were defined as weight-for-length z score < −2 and length-for-age z score < −2, respectively. ROC, receiver operating characteristic; WAZ, weight-for-age z score; WVZ, weight velocity z score.

Subgroup analyses by birth weight for gestational age (Table 2) showed that monthly WVZ had greater predictive abilities for severe wasting and concurrent wasting and stunting among infants born AGA or LGA than among those born SGA. For example, for severe wasting at 6 mo, the AUC of WVZ for 2–3 mo was 0.86 among AGA/LGA infants compared with 0.74 for SGA infants; for concurrent wasting and stunting at 6 mo, the AUC of WVZ for 4–5 mo was 0.85 among AGA/LGA infants compared with 0.68 for SGA infants. Subgroup analyses by gestational age at birth (Table 3) showed that monthly WVZ had greater predictive abilities for concurrent wasting and stunting among preterm infants than among term infants. For example, for concurrent wasting and stunting at 6 mo, the AUC of WVZ for 2–3 mo was 0.99 among preterm infants compared with 0.84 for term infants. Subgroup analyses by birth weight (Table 4) showed that monthly WVZ had greater predictive abilities for concurrent wasting and stunting among infants born ≥3000 g than among infants born <3000 g. For example, for concurrent wasting and stunting at 6 mo, the AUC of WVZ for 4–5 mo was 0.85 among infants born ≥3000 g compared with 0.70 among infants born <3000 g. For the prediction of child mortality, all 3 types of weight metrics had similar AUC values. However, none of the weight metrics predicted child mortality particularly well, with the AUC values rarely reaching >0.65 (Table 5).

TABLE 2.

Weight metrics during the first 6 mo of life and undernutrition at 6 mo among infants in Dar es Salaam, Tanzania (2001–2006), stratified by birth weight for gestational age¹

	Undernutrition at 6 mo2
	Wasting		Severe wasting		Stunting		Concurrent wasting and stunting
	n	AUC (95% CI)	n	AUC (95% CI)	n	AUC (95% CI)	n	AUC (95% CI)
Small-for-gestational-age births3
Attained WAZ at the end of each month
At birth	845	0.56 (0.47, 0.65)	845	0.49 (0.28, 0.69)	847	0.56 (0.52, 0.61)	845	0.63 (0.40, 0.85)
1 mo	536	0.61 (0.51, 0.71)	536	0.64 (0.48, 0.81)	537	0.70 (0.64, 0.75)	536	0.66 (0.31, 1.00)
2 mo	731	0.73 (0.65, 0.81)	731	0.79 (0.69, 0.89)	733	0.71 (0.66, 0.75)	731	0.91 (0.84, 0.98)
3 mo	743	0.76 (0.68, 0.83)	743	0.85 (0.75, 0.95)	745	0.74 (0.70, 0.78)	743	0.95 (0.93, 0.98)
4 mo	782	0.81 (0.73, 0.88)	782	0.86 (0.77, 0.94)	783	0.75 (0.71, 0.79)	782	0.97 (0.95, 0.99)
5 mo	785	0.84 (0.77, 0.90)	785	0.89 (0.81, 0.98)	787	0.76 (0.72, 0.80)	785	0.97 (0.94, 0.99)
Monthly change in WAZ
Birth–1 mo	536	0.57 (0.46, 0.67)	536	0.43 (0.24, 0.62)	537	0.68 (0.62, 0.73)	536	0.36 (0.03, 0.69)
1–2 mo	501	0.72 (0.63, 0.80)	501	0.68 (0.53, 0.83)	502	0.57 (0.51, 0.63)	501	0.85 (0.71, 0.99)
2–3 mo	680	0.62 (0.53, 0.72)	680	0.68 (0.47, 0.90)	682	0.54 (0.49, 0.60)	680	0.74 (0.52, 0.97)
3–4 mo	711	0.68 (0.60, 0.76)	711	0.61 (0.42, 0.80)	712	0.54 (0.49, 0.59)	711	0.68 (0.48, 0.87)
4–5 mo	740	0.65 (0.57, 0.73)	740	0.63 (0.44, 0.82)	741	0.53 (0.48, 0.58)	740	0.62 (0.40, 0.84)
5–6 mo	774	0.75 (0.68, 0.82)	774	0.73 (0.57, 0.89)	776	0.51 (0.47, 0.56)	774	0.79 (0.65, 0.93)
Monthly WVZ
Birth–1 mo	537	0.59 (0.48, 0.69)	537	0.43 (0.26, 0.59)	538	0.65 (0.60, 0.71)	537	0.32 (0.00, 0.66)
1–2 mo	503	0.76 (0.67, 0.85)	503	0.79 (0.70, 0.89)	504	0.67 (0.61, 0.73)	503	0.94 (0.90, 0.99)
2–3 mo	520	0.71 (0.62, 0.80)	520	0.74 (0.58, 0.89)	521	0.61 (0.55, 0.66)	520	0.83 (0.63, 1.00)
3–4 mo	682	0.72 (0.65, 0.79)	682	0.67 (0.50, 0.84)	683	0.58 (0.53, 0.63)	682	0.71 (0.54, 0.87)
4–5 mo	723	0.69 (0.62, 0.76)	723	0.64 (0.48, 0.80)	724	0.56 (0.51, 0.61)	723	0.68 (0.49, 0.86)
5–6 mo	739	0.73 (0.65, 0.81)	739	0.74 (0.61, 0.87)	741	0.49 (0.44, 0.53)	739	0.75 (0.55, 0.94)
Appropriate4- and large5-for-gestational-age births
Attained WAZ at the end of each month
At birth	4295	0.54 (0.49, 0.59)	4295	0.53 (0.43, 0.62)	4302	0.64 (0.61, 0.67)	4295	0.54 (0.35, 0.74)
1 mo	2854	0.66 (0.60, 0.71)	2854	0.59 (0.44, 0.74)	2859	0.72 (0.69, 0.76)	2854	0.75 (0.55, 0.94)
2 mo	3760	0.74 (0.70, 0.79)	3760	0.68 (0.57, 0.79)	3766	0.76 (0.74, 0.79)	3760	0.84 (0.69, 1.00)
3 mo	3839	0.82 (0.78, 0.86)	3839	0.78 (0.68, 0.88)	3845	0.78 (0.76, 0.81)	3839	0.94 (0.88, 1.00)
4 mo	3924	0.87 (0.84, 0.90)	3924	0.82 (0.74, 0.91)	3931	0.78 (0.75, 0.80)	3924	0.97 (0.94, 1.00)
5 mo	4009	0.90 (0.88, 0.93)	4009	0.87 (0.81, 0.94)	4016	0.77 (0.75, 0.80)	4009	0.99 (0.97, 1.00)
Monthly change in WAZ
Birth–1 mo	2854	0.67 (0.62, 0.72)	2854	0.68 (0.57, 0.80)	2859	0.63 (0.59, 0.67)	2854	0.84 (0.72, 0.97)
1–2 mo	2665	0.71 (0.66, 0.77)	2665	0.72 (0.60, 0.84)	2670	0.55 (0.50, 0.59)	2665	0.83 (0.61, 1.00)
2–3 mo	3516	0.69 (0.64, 0.73)	3516	0.79 (0.72, 0.86)	3522	0.54 (0.50, 0.57)	3516	0.75 (0.57, 0.93)
3–4 mo	3616	0.66 (0.61, 0.71)	3616	0.70 (0.60, 0.80)	3622	0.51 (0.47, 0.54)	3616	0.64 (0.41, 0.88)
4–5 mo	3738	0.66 (0.61, 0.71)	3738	0.69 (0.59, 0.79)	3745	0.51 (0.47, 0.54)	3738	0.82 (0.71, 0.92)
5–6 mo	3979	0.66 (0.60, 0.71)	3979	0.82 (0.74, 0.90)	3986	0.54 (0.50, 0.57)	3979	0.83 (0.68, 0.99)
Monthly WVZ
Birth–1 mo	2861	0.70 (0.66, 0.75)	2861	0.69 (0.57, 0.81)	2866	0.66 (0.62, 0.69)	2861	0.87 (0.74, 1.00)
1–2 mo	2671	0.81 (0.76, 0.85)	2671	0.79 (0.69, 0.89)	2676	0.66 (0.62, 0.70)	2671	0.92 (0.83, 1.00)
2–3 mo	2726	0.79 (0.74, 0.83)	2726	0.86 (0.81, 0.91)	2731	0.65 (0.61, 0.68)	2726	0.88 (0.78, 0.99)
3–4 mo	3433	0.74 (0.70, 0.79)	3433	0.77 (0.69, 0.85)	3439	0.57 (0.53, 0.60)	3433	0.79 (0.68, 0.90)
4–5 mo	3641	0.72 (0.69, 0.76)	3641	0.75 (0.67, 0.82)	3648	0.55 (0.52, 0.59)	3641	0.85 (0.77, 0.93)
5–6 mo	3801	0.68 (0.63, 0.73)	3801	0.83 (0.76, 0.90)	3807	0.52 (0.49, 0.55)	3801	0.86 (0.76, 0.96)

¹Attained z scores at birth were computed using the INTERGROWTH-21^st newborn size standards. For term-born children, postnatal attained z scores were computed using the WHO 2006 child growth standards (25). For infants born preterm, postnatal attained z scores were computed using the INTERGROWTH-21^st standards for postnatal growth of preterm infants. WAZ, weight-for-age z score; WLZ, weight-for-length z score; WVZ, weight velocity z score.

²Stunting, wasting, and severe wasting were defined as length-for-age z score < −2, WLZ < −2, and WLZ z score < −3, respectively.

³Birth weight <10th percentile for gestational age and sex based on the INTERGROWTH-21st newborn size standards.

⁴Birth weight between 10th and 90th percentiles for gestational age and sex based on the INTERGROWTH-21st newborn size standards.

⁵Birth weight >90th percentile for gestational age and sex based on the INTERGROWTH-21st newborn size standards.

TABLE 3.

Weight metrics during the first 6 mo of life and undernutrition at 6 mo among infants in Dar es Salaam, Tanzania (2001–2006), stratified by gestational age at birth¹

	Undernutrition at 6 mo2
	Wasting		Severe wasting		Stunting		Concurrent wasting and stunting
	n	AUC (95% CI)	n	AUC (95% CI)	n	AUC (95% CI)	n	AUC (95% CI)
Preterm births3
Attained WAZ at the end of each month
At birth	826	0.61 (0.53, 0.70)	826	0.57 (0.39, 0.75)	827	0.71 (0.64, 0.78)	826	0.77 (0.53, 1.00)
1 mo	488	0.74 (0.65, 0.84)	488	0.69 (0.48, 0.91)	488	0.77 (0.68, 0.86)	488	0.95 (0.87, 1.00)
2 mo	716	0.77 (0.69, 0.86)	716	0.68 (0.50, 0.86)	717	0.86 (0.80, 0.91)	716	0.94 (0.83, 1.00)
3 mo	720	0.84 (0.77, 0.91)	720	0.78 (0.63, 0.93)	721	0.86 (0.80, 0.92)	720	1.00 (0.99, 1.00)
4 mo	762	0.88 (0.83, 0.94)	762	0.84 (0.72, 0.96)	763	0.85 (0.79, 0.90)	762	1.00 (1.00, 1.00)
5 mo	780	0.93 (0.89, 0.96)	780	0.90 (0.82, 0.98)	781	0.84 (0.77, 0.90)	780	1.00 (0.99, 1.00)
Monthly change in WAZ
Birth–1 mo	478	0.73 (0.63, 0.83)	478	0.71 (0.50, 0.92)	478	0.60 (0.50, 0.71)	478	0.92 (0.80, 1.00)
1–2 mo	458	0.58 (0.46, 0.69)	458	0.55 (0.29, 0.81)	458	0.52 (0.40, 0.64)	458	0.63 (0.13, 1.00)
2–3 mo	645	0.70 (0.61, 0.79)	645	0.83 (0.73, 0.93)	646	0.55 (0.46, 0.65)	645	0.75 (0.50, 1.00)
3–4 mo	675	0.68 (0.56, 0.79)	675	0.70 (0.53, 0.88)	676	0.56 (0.45, 0.66)	675	0.63 (0.25, 1.00)
4–5 mo	717	0.65 (0.55, 0.75)	717	0.72 (0.57, 0.87)	718	0.48 (0.38, 0.57)	717	0.52 (0.28, 0.76)
5–6 mo	769	0.62 (0.51, 0.73)	769	0.72 (0.57, 0.87)	770	0.57 (0.47, 0.66)	769	0.63 (0.40, 0.87)
Monthly WVZ
Birth–1 mo	483	0.76 (0.67, 0.84)	483	0.74 (0.56, 0.91)	483	0.67 (0.56, 0.77)	483	0.94 (0.88, 1.00)
1–2 mo	454	0.81 (0.72, 0.89)	454	0.70 (0.48, 0.93)	454	0.69 (0.60, 0.79)	454	0.87 (0.67, 1.00)
2–3 mo	473	0.83 (0.74, 0.91)	473	0.90 (0.83, 0.97)	473	0.62 (0.52, 0.72)	473	0.99 (0.97, 1.00)
3–4 mo	635	0.79 (0.70, 0.88)	635	0.83 (0.72, 0.94)	636	0.58 (0.47, 0.68)	635	0.84 (0.69, 0.99)
4–5 mo	690	0.75 (0.67, 0.83)	690	0.80 (0.67, 0.92)	691	0.54 (0.45, 0.64)	690	0.77 (0.66, 0.87)
5–6 mo	734	0.66 (0.56, 0.77)	734	0.75 (0.61, 0.89)	735	0.52 (0.43, 0.61)	734	0.78 (0.65, 0.90)
Term births
Attained WAZ at the end of each month
At birth	4314	0.60 (0.55, 0.64)	4314	0.56 (0.46, 0.66)	4322	0.68 (0.65, 0.70)	4314	0.71 (0.54, 0.89)
1 mo	3308	0.71 (0.66, 0.75)	3308	0.70 (0.57, 0.83)	3314	0.74 (0.71, 0.77)	3308	0.69 (0.47, 0.91)
2 mo	4323	0.79 (0.76, 0.82)	4323	0.78 (0.69, 0.86)	4330	0.76 (0.74, 0.79)	4323	0.92 (0.86, 0.98)
3 mo	4436	0.84 (0.81, 0.86)	4436	0.86 (0.79, 0.92)	4443	0.78 (0.76, 0.80)	4436	0.95 (0.91, 0.99)
4 mo	4535	0.87 (0.85, 0.90)	4535	0.88 (0.82, 0.94)	4542	0.78 (0.76, 0.80)	4535	0.97 (0.95, 0.99)
5 mo	4608	0.89 (0.87, 0.92)	4608	0.90 (0.84, 0.96)	4616	0.78 (0.75, 0.80)	4608	0.98 (0.97, 0.99)
Monthly change in WAZ
Birth–1 mo	2912	0.58 (0.53, 0.64)	2912	0.55 (0.43, 0.68)	2918	0.57 (0.53, 0.60)	2912	0.61 (0.38, 0.83)
1–2 mo	3084	0.73 (0.68, 0.78)	3084	0.74 (0.64, 0.84)	3090	0.56 (0.52, 0.59)	3084	0.90 (0.82, 0.98)
2–3 mo	4062	0.67 (0.62, 0.71)	4062	0.75 (0.66, 0.84)	4069	0.54 (0.51, 0.57)	4062	0.71 (0.57, 0.85)
3–4 mo	4199	0.67 (0.62, 0.71)	4199	0.67 (0.57, 0.77)	4205	0.52 (0.50, 0.55)	4199	0.66 (0.53, 0.79)
4–5 mo	4319	0.65 (0.61, 0.69)	4319	0.61 (0.51, 0.71)	4326	0.52 (0.49, 0.54)	4319	0.70 (0.56, 0.83)
5–6 mo	4571	0.69 (0.65, 0.74)	4571	0.79 (0.70, 0.88)	4579	0.51 (0.48, 0.54)	4571	0.74 (0.61, 0.87)
Monthly WVZ
Birth–1 mo	3273	0.65 (0.60, 0.70)	3273	0.60 (0.48, 0.71)	3279	0.63 (0.60, 0.66)	3273	0.30 (0.11, 0.48)
1–2 mo	3053	0.80 (0.76, 0.84)	3053	0.82 (0.77, 0.88)	3059	0.66 (0.62, 0.69)	3053	0.96 (0.94, 0.98)
2–3 mo	3161	0.76 (0.72, 0.80)	3161	0.81 (0.73, 0.88)	3167	0.62 (0.59, 0.65)	3161	0.84 (0.73, 0.95)
3–4 mo	3992	0.73 (0.70, 0.77)	3992	0.74 (0.66, 0.83)	3998	0.57 (0.54, 0.60)	3992	0.75 (0.66, 0.84)
4–5 mo	4222	0.70 (0.67, 0.74)	4222	0.68 (0.60, 0.76)	4229	0.56 (0.53, 0.59)	4222	0.78 (0.67, 0.89)
5–6 mo	4361	0.71 (0.67, 0.75)	4361	0.82 (0.75, 0.89)	4368	0.53 (0.51, 0.56)	4361	0.77 (0.65, 0.89)

¹Attained z scores at birth were computed using the INTERGROWTH-21^st newborn size standards. For term-born children, postnatal attained z scores were computed using the WHO 2006 child growth standards (25). For infants born preterm, postnatal attained z scores were computed using the INTERGROWTH-21^st standards for postnatal growth of preterm infants. WAZ, weight-for-age z score; WLZ, weight-for-length z score; WVZ, weight velocity z score.

²Stunting, wasting, and severe wasting were defined as length-for-age z score < −2, WLZ < −2, and WLZ < −3, respectively.

³Gestational age at birth <37 wk.

TABLE 4.

Weight metrics during the first 6 mo of life and undernutrition at 6 mo among infants in Dar es Salaam, Tanzania (2001–2006), stratified by birth weight¹

	Undernutrition at 6 mo2
	Wasting		Severe wasting		Stunting		Concurrent wasting and stunting
	n	AUC (95% CI)	n	AUC (95% CI)	n	AUC (95% CI)	n	AUC (95% CI)
Birth weight < 3000 g
Attained WAZ at the end of each month
At birth	1568	0.57 (0.50, 0.64)	1568	0.55 (0.41, 0.68)	1571	0.63 (0.59, 0.66)	1568	0.72 (0.54, 0.90)
1 mo	1063	0.63 (0.55, 0.71)	1063	0.66 (0.51, 0.82)	1064	0.73 (0.68, 0.77)	1063	0.74 (0.44, 1.00)
2 mo	1464	0.73 (0.67, 0.79)	1464	0.78 (0.66, 0.91)	1467	0.76 (0.73, 0.79)	1464	0.94 (0.90, 0.99)
3 mo	1487	0.76 (0.71, 0.82)	1487	0.81 (0.69, 0.93)	1490	0.78 (0.75, 0.81)	1487	0.98 (0.96, 0.99)
4 mo	1549	0.82 (0.78, 0.87)	1549	0.85 (0.76, 0.94)	1551	0.78 (0.75, 0.81)	1549	0.98 (0.98, 0.99)
5 mo	1579	0.86 (0.82, 0.90)	1579	0.89 (0.83, 0.96)	1582	0.78 (0.75, 0.81)	1579	0.98 (0.96, 1.00)
Monthly change in WAZ
Birth–1 mo	984	0.62 (0.54, 0.70)	984	0.58 (0.43, 0.72)	985	0.64 (0.60, 0.69)	984	0.67 (0.37, 0.96)
1–2 mo	992	0.69 (0.61, 0.77)	992	0.73 (0.62, 0.85)	993	0.54 (0.50, 0.59)	992	0.86 (0.74, 0.98)
2–3 mo	1349	0.63 (0.56, 0.70)	1349	0.70 (0.54, 0.85)	1352	0.51 (0.47, 0.56)	1349	0.74 (0.54, 0.94)
3–4 mo	1420	0.68 (0.62, 0.75)	1420	0.63 (0.49, 0.78)	1422	0.54 (0.50, 0.58)	1420	0.63 (0.42, 0.83)
4–5 mo	1474	0.65 (0.59, 0.71)	1474	0.69 (0.56, 0.82)	1476	0.50 (0.46, 0.54)	1474	0.61 (0.40, 0.81)
5–6 mo	1558	0.70 (0.64, 0.77)	1558	0.73 (0.60, 0.86)	1561	0.55 (0.52, 0.59)	1558	0.76 (0.63, 0.89)
Monthly WVZ
Birth–1 mo	1069	0.65 (0.57, 0.73)	1069	0.62 (0.49, 0.75)	1070	0.67 (0.62, 0.71)	1069	0.26 (0.00, 0.56)
1–2 mo	998	0.76 (0.69, 0.84)	998	0.83 (0.75, 0.90)	999	0.67 (0.62, 0.71)	998	0.96 (0.93, 1.00)
2–3 mo	1024	0.75 (0.69, 0.81)	1024	0.79 (0.68, 0.90)	1025	0.61 (0.57, 0.66)	1024	0.87 (0.70, 1.00)
3–4 mo	1347	0.76 (0.71, 0.81)	1347	0.72 (0.60, 0.84)	1349	0.55 (0.51, 0.59)	1347	0.72 (0.56, 0.87)
4–5 mo	1438	0.71 (0.67, 0.76)	1438	0.71 (0.61, 0.82)	1440	0.57 (0.53, 0.60)	1438	0.70 (0.53, 0.87)
5–6 mo	1490	0.71 (0.65, 0.77)	1490	0.75 (0.64, 0.86)	1493	0.52 (0.48, 0.55)	1490	0.74 (0.57, 0.91)
Birth weight ≥ 3000 g
Attained WAZ at the end of each month
At birth	3572	0.55 (0.50, 0.60)	3572	0.54 (0.44, 0.65)	3578	0.63 (0.60, 0.67)	3572	0.51 (0.27, 0.75)
1 mo	2684	0.69 (0.63, 0.74)	2684	0.58 (0.42, 0.74)	2689	0.71 (0.67, 0.75)	2684	0.74 (0.55, 0.92)
2 mo	3493	0.77 (0.73, 0.82)	3493	0.69 (0.57, 0.81)	3498	0.75 (0.72, 0.78)	3493	0.87 (0.73, 1.00)
3 mo	3583	0.84 (0.80, 0.88)	3583	0.81 (0.70, 0.92)	3588	0.77 (0.74, 0.80)	3583	0.95 (0.91, 1.00)
4 mo	3652	0.88 (0.84, 0.91)	3652	0.83 (0.73, 0.92)	3658	0.77 (0.74, 0.80)	3652	0.98 (0.96, 1.00)
5 mo	3712	0.91 (0.88, 0.93)	3712	0.87 (0.80, 0.95)	3718	0.76 (0.73, 0.79)	3712	0.99 (0.98, 1.00)
Monthly change in WAZ
Birth–1 mo	2406	0.64 (0.58, 0.70)	2406	0.70 (0.56, 0.83)	2411	0.60 (0.56, 0.65)	2406	0.75 (0.53, 0.97)
1–2 mo	2506	0.73 (0.68, 0.79)	2506	0.69 (0.54, 0.84)	2511	0.58 (0.53, 0.63)	2506	0.83 (0.61, 1.00)
2–3 mo	3289	0.69 (0.64, 0.73)	3289	0.82 (0.75, 0.89)	3294	0.55 (0.51, 0.58)	3289	0.71 (0.55, 0.88)
3–4 mo	3376	0.65 (0.60, 0.70)	3376	0.71 (0.61, 0.81)	3381	0.54 (0.50, 0.58)	3376	0.67 (0.50, 0.84)
4–5 mo	3476	0.65 (0.61, 0.70)	3476	0.65 (0.54, 0.76)	3482	0.50 (0.47, 0.54)	3476	0.74 (0.62, 0.87)
5–6 mo	3687	0.67 (0.61, 0.72)	3687	0.81 (0.72, 0.90)	3693	0.50 (0.46, 0.53)	3687	0.73 (0.54, 0.92)
Monthly WVZ
Birth–1 mo	2687	0.69 (0.64, 0.74)	2687	0.68 (0.54, 0.82)	2692	0.63 (0.59, 0.67)	2687	0.78 (0.64, 0.93)
1–2 mo	2509	0.83 (0.79, 0.87)	2509	0.78 (0.66, 0.89)	2514	0.67 (0.63, 0.72)	2509	0.92 (0.82, 1.00)
2–3 mo	2565	0.78 (0.74, 0.83)	2565	0.86 (0.81, 0.92)	2570	0.65 (0.61, 0.69)	2565	0.87 (0.77, 0.98)
3–4 mo	3209	0.73 (0.69, 0.78)	3209	0.78 (0.69, 0.87)	3214	0.61 (0.57, 0.64)	3209	0.80 (0.72, 0.89)
4–5 mo	3393	0.72 (0.68, 0.76)	3393	0.73 (0.65, 0.82)	3399	0.56 (0.53, 0.60)	3393	0.85 (0.80, 0.90)
5–6 mo	3524	0.70 (0.65, 0.75)	3524	0.84 (0.76, 0.91)	3529	0.55 (0.51, 0.58)	3524	0.80 (0.67, 0.94)

¹Attained z scores at birth were computed using the INTERGROWTH-21^st newborn size standards. For term-born children, postnatal attained z scores were computed using the WHO 2006 child growth standards (25). For infants born preterm, postnatal attained z scores were computed using the INTERGROWTH-21^st standards for postnatal growth of preterm infants. WAZ, weight-for-age z score; WLZ, weight-for-length z score; WVZ, weight velocity z score.

²Stunting, wasting, and severe wasting were defined as length-for-age z score < −2, WLZ < −2, and WLZ < −3, respectively.

TABLE 5.

Weight metrics during the first 6 mo of life and child mortality during the first 18 mo of life among infants in Dar es Salaam, Tanzania (2001–2006)¹

		Child mortality2
		Mortality between 1 wk and 3 mo	Mortality between 3 and 6 mo	Mortality between 6 and 12 mo	Mortality between 12 and 18 mo
	n	AUC (95% CI)	AUC (95% CI)	AUC (95% CI)	AUC (95% CI)
Attained WAZ at the end of each month
At birth	6595	0.59 (0.48, 0.70)	0.57 (0.46, 0.68)	0.50 (0.43, 0.58)	0.53 (0.42, 0.64)
1 mo	4449	—	0.53 (0.37, 0.69)	0.58 (0.48, 0.69)	0.54 (0.39, 0.68)
2 mo	5857	—	0.60 (0.48, 0.72)	0.55 (0.46, 0.65)	0.59 (0.49, 0.69)
3 mo	5859	—	0.62 (0.50, 0.73)	0.55 (0.46, 0.64)	0.57 (0.48, 0.66)
4 mo	5904	—	—	0.55 (0.47, 0.64)	0.56 (0.46, 0.67)
5 mo	5865	—	—	0.56 (0.48, 0.64)	0.60 (0.51, 0.70)
6 mo	5891	—	—	0.57 (0.47, 0.66)	0.64 (0.54, 0.74)
Monthly change in WAZ
Birth–1 mo	3986	—	0.68 (0.55, 0.82)	0.58 (0.48, 0.69)	0.56 (0.44, 0.69)
1–2 mo	4043	—	0.48 (0.31, 0.64)	0.53 (0.44, 0.62)	0.55 (0.43, 0.68)
2–3 mo	5297	—	0.52 (0.38, 0.67)	0.53 (0.44, 0.61)	0.50 (0.39, 0.61)
3–4 mo	5378	—	—	0.50 (0.42, 0.58)	0.53 (0.41, 0.64)
4–5 mo	5440	—	—	0.56 (0.48, 0.64)	0.59 (0.48, 0.70)
5–6 mo	5421	—	—	0.62 (0.52, 0.71)	0.64 (0.55, 0.74)
Monthly WVZ
Birth–1 mo	4404	—	0.61 (0.50, 0.72)	0.57 (0.47, 0.68)	0.60 (0.49, 0.72)
1–2 mo	4002	—	0.55 (0.40, 0.70)	0.58 (0.47, 0.69)	0.59 (0.47, 0.71)
2–3 mo	4074	—	0.54 (0.40, 0.69)	0.49 (0.39, 0.58)	0.52 (0.40, 0.64)
3–4 mo	5094	—	—	0.52 (0.44, 0.60)	0.47 (0.35, 0.59)
4–5 mo	5273	—	—	0.56 (0.48, 0.64)	0.59 (0.49, 0.70)
5–6 mo	5170	—	—	0.57 (0.46, 0.69)	0.62 (0.52, 0.72)

¹Attained z scores at birth were computed using the INTERGROWTH-21^st newborn size standards. For term-born children, postnatal attained z scores were computed using the WHO 2006 child growth standards (25). For infants born preterm, postnatal attained z scores were computed using the INTERGROWTH-21^st standards for postnatal growth of preterm infants. WAZ, weight-for-age z score; WVZ, weight velocity z score.

²Infants who died during the first week of life (147 of 301 deaths) were excluded from the analyses.

Discussion

In this analysis using a large cohort of infants from Dar es Salaam, Tanzania, we report that, compared with monthly weight velocity or change in attained weight, attained weight at the end of each month during the first 6 mo of life better predicted undernutrition at 6 and 12 mo. The weight metrics had similar abilities in predicting child mortality in the short- and long-terms, with none of the metrics predicting child mortality particularly well.

The abilities of attained growth and growth velocity in predicting child undernutrition (13–16) or mortality (17–21) have been compared in previous studies. Only a few studies (14, 20, 21) have used the latest growth standards such as the WHO child growth (25) and growth velocity (29) standards or the INTERGROWTH-21^st growth standards for preterm infants (27). No studies, to our knowledge, have compared the predictive abilities of attained weight at the end of each month with the weight velocity during the same 1-mo interval. This study was an observational analysis of existing data instead of the prospective applications of different weight metrics in a programmatic setting. Nonetheless, our results may inform the design of future growth monitoring efforts during the critical period of the first 6 mo of life.

We report that attained weight has predictive performances greater than or comparable with weight velocity during all months and for all undernutrition outcomes, except for wasting, for which during the second month weight velocity slightly outperformed attained weight (AUC: 0.80 compared with 0.76). Monthly changes in WAZ, on the other hand, consistently had low predictive abilities compared with weight velocity and attained weight. Our finding that attained weight had generally greater predictive abilities than weight velocity is consistent with Schwinger et al. (14), who compared the capabilities of growth velocities and attained growth during the first 12 mo of life in predicting stunting, wasting, and underweight at 2 y of age, based on 240 children in Nepal. The study showed that attained WAZ at the end of 12 mo had high predictive abilities for undernutrition at 2 y of age, with an AUC of 0.81, 0.82, and 0.95 for stunting, wasting, and underweight, respectively (14). Schwinger et al. (14) concluded that attained growth measured at single time points during infancy was more strongly correlated with undernutrition than growth velocity was. An earlier study by Ruel et al. (16) reported that weight velocity from 3 to 6 mo was worse than attained WAZ at 3 or 6 mo at predicting stunting at 3 y of age. We used narrower monthly intervals instead of the 3-mo intervals used in Schwinger et al. (14) and Ruel et al. (16), both of which also suggested that attained weight is superior to weight velocity in predicting undernutrition later in childhood. In resource-constrained settings, where repeated anthropometric measures are not feasible, the simpler metric of attained weight has greater potential for assessment at the population level (13, 14). Findings from our study indicate that postnatal attained weight is a useful indicator for identifying infants at risk of undernutrition. In contrast, birth weight per se had poor predictive abilities, highlighting the importance of monitoring postnatal growth. However, birth weight in conjunction with postnatal attained weights may be informative for identifying growth faltering, as suggested by the National Institute for Health and Care Excellence guidelines (33).

Physiological and methodological factors may explain the better predictive abilities of attained weight than of weight velocity. Diseases during early infancy may cause temporary weight loss, and weight velocity may be more sensitive to the weight increase after treatment of the diseases. However, this “catch-up” weight increase after temporary weight loss may be inadequate to correct chronic deficits, which are better captured by attained weight (14, 34). Further, metrics of change such as weight velocity depend on 2 weight measurements, so they are more subject to the influence of measurement errors that may tend to attenuate the strength of the associations with undernutrition.

To the best of our knowledge, no previous studies have conducted subgroup analyses that compared weight velocity with attained weight by birth outcomes. Our subgroup analyses indicate that weight velocity may have better predictive performances among preterm infants, infants born AGA, and infants born with adequate birth weight. This finding may have implications for the targeted application of weight velocity, yet will need to be replicated in additional studies.

The first few weeks of life are critical to early nutrition, because breastfeeding is established within the first few days (35) and infants regain birth weight in the first 2 wk (36). Therefore, the predictive potential of weight velocity during narrower (weekly or biweekly) time windows in the first few weeks cannot be ruled out. However, shorter intervals of measurements would lead to greater measurement errors and increased false alarms due to the oscillations in weight (37). Using consecutive weight velocity measurements may alleviate these issues (30).

Despite the potential of weight velocity in predicting child undernutrition, several considerations may affect its practical utility in low- and middle-income countries. First, conducting the repeated anthropometry measures as frequently as needed in early life may not be feasible in many resource-constrained settings. Second, because growth velocity is subject to measurement errors of 2 anthropometric measures, it is more imperative to have high-quality measurements, which may be difficult to achieve in certain settings. Third, the anthropometry measures should ideally be taken within the maximum tolerable differences between the planned and actual measurement ages (i.e., ±3 d from 0 to 6 mo, ±5 d from 6 to 12 mo, and ±7 d from 12 to 24 mo) according to the WHO weight velocity standards, further adding to the logistical challenges (29). Fourth, tables exist that facilitate the identification of weight velocity that corresponds to certain z scores (−3, −2, −1, 0, 1, 2, and 3) or percentiles (e.g., 1st, 3rd, 5th, 15th, 25th, 50th) (29). However, no publicly available tools or software programs can readily calculate WVZ as a continuous variable, hampering its use in both practical and research settings. Standardized tools available on mobile platforms may accelerate the uptake of growth velocity metrics by health care providers and community health workers in low- and middle-income countries.

We report that attained weight and weight velocity generally have similar predictive abilities for child mortality. Among a cohort of infants in the Democratic Republic of the Congo, it was reported that 3-mo weight velocity had an AUC of 0.67 for predicting mortality within the next 3 mo, which was greater than the AUC of 0.57 of attained weight (21). The AUC values of weight velocity for mortality in our study are somewhat lower (<0.62), and none of the early-life weight metrics predicted child mortality well. This discrepancy may be due to the different study settings. There was no nutritional program and limited health care at the time of the study (1989–1991) in a rural setting in the Democratic Republic of the Congo (21). In contrast, all women in our study received standard antenatal care (including iron and folic acid supplementation), and some received MMS, which may be associated with higher birth weight and with lower incidence of SGA births (23).

Similar to findings of previous studies (20, 21), none of the weight metrics appeared to predict child mortality very well, which may be explained by 2 factors. First, mortality was rare (∼4%) during the follow-up period in our study. Nearly half of deaths occurred in the first week of life and were not included in the analyses, leading to diminished statistical power. Second, we used all-cause mortality and, during very early infancy, the proportion of deaths unrelated to nutrition may be greater, thus limiting the predictive ability of weight metrics. Other maternal and child determinants beyond postnatal anthropometry may be present. Such factors include maternal nutrition, dietary intake, concurrent illnesses, and quality of nurturing, ultimately determined by upstream socioeconomic and political factors (1).

The strengths of this study include a large sample size, a long period of follow-up, the comparison of 3 categories of early-life weight metrics derived using established child growth standards, and the presence of monthly anthropometric measurements. This study also has some limitations. First, child growth during the first several weeks has important implications for later-life nutrition. However, we could not examine the predictive abilities of attained weight and weight velocity during early infancy (e.g., the first few weeks of life) because we did not have frequent measurements during this period. Second, our study provided a comparative analysis of attained weight and weight velocity. We opted not to propose specific cutoffs for practical use, which may be determined by triangulating with additional evidence from diverse settings and considering the costs and benefits of potential interventions.

In conclusion, postnatal attained weight remains a useful indicator for identifying infants at risk of undernutrition. The predictive value of weight velocity over narrower windows during early infancy warrants further investigation.

Notes

Supported by Eunice Kennedy Shriver National Institute of Child Health and Human Development of the NIH grant NICHD R01 37701 (to WWF).

Author disclosures: The authors report no conflicts of interest.

Supplemental Figure 1 and Supplemental Tables 1 and 2 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn/.

Abbreviations used: AGA, appropriate for gestational age; LAZ, length-for-age z score; LGA, large for gestational age; MMS, multiple micronutrient supplementation; SGA, small for gestational age; WAZ, weight-for-age z score; WLZ, weight-for-length z score; WVZ, weight velocity z score.

Data Availability

Data described in the article, codebook, and analytic code will be made available upon request to the corresponding author, pending approval by the study team.