Early childhood growth in Daasanach pastoralists of Northern Kenya: Distinct patterns of faltering in linear growth and weight gain

Abstract

Objectives:

Investigations of early childhood growth among small-scale populations are essential for understanding human life history variation and enhancing the ability to serve such communities through global public health initiatives. This study characterizes early childhood growth trajectories and identifies differences in growth patterns relative to international references among Daasanach semi-nomadic pastoralist children living in a hot, arid region of northern Kenya.

Methods:

A large sample of height and weight measures were collected from children (N = 1756; total observations = 4508; age = 0–5 years) between 2018 and 2020. Daasanach growth was compared to international reference standards and Daasanach-specific centile growth curves and pseudo-velocity models were generated using generalized additive models for location scale and size.

Results:

Compared to World Health Organization (WHO) reference, relatively few Daasanach children were stunted (14.3%), while a large proportion were underweight (38.5%) and wasted (53.6%). Additionally, Daasanach children had a distinctive pattern of growth, marked by an increase in linear growth velocity after 24 months of age and relatively high linear growth velocity throughout the rest of early childhood.

Conclusions:

These results identify a unique pattern of early childhood growth faltering among children in a small-scale population and may reflect a thermoregulatory adaptation to their hot, arid environment. As linear growth and weight gain remain important indicators of health, the results of this study provide insight into growth velocity variations. This study has important implications for global public health efforts to identify and address sources of early growth faltering and undernutrition in small-scale populations.

INTRODUCTION

Patterns of somatic growth are important indicators of health and nutritional status in children and have significant consequences for adaptive variation in adult stature (Waterlow, 1994; Bogin, 1999; Cole, 2007). Linear growth serves as a measure of long-term nutritional status while weight gain acts as a measure of short-term nutritional status. The causes of growth faltering, either in linear growth or weight gain, are multifactorial and include disease and inadequate nutrition during infancy and childhood (WHO, 2006; Spencer et al., 2017; Gonzalez-Viana et al, 2017; Urlacher et al., 2018; Urlacher & Kramer, 2018). The detection of growth faltering by comparison against large international growth standards is common among studies of non-industrialized populations suffering malnutrition and disease (Frisancho et al., 2008; de Onis et al., 2007; Saha et al., 2009; de Onis & Branca; 2016; Spencer et al., 2017; Zhang et al., 2017; Martin et al. 2019). The use of such international growth references, with which large numbers of individuals can be easily assessed, has been an important global health tool as malnutrition is associated with myriad negative short- and long-term health conditions, such as reduced muscle or organ function, cognitive developmental deficits, and increased psychosocial stress (Saunders & Smith, 2010; Adair et al., 2013; Sudfeld et al., 2015).

Derived from large multinational surveys, the World Health Organization’s (WHO) Growth Standards are one of the most widely accepted and used international growth references for comparison across populations since their establishment in 2006 (WHO, 2006). Age-stratified z-scores (standard deviation values) for weight-for-height, weight-for-age, and height-for-age can be used to compare prevalence of wasting, underweight, and stunting, respectively, across mixed cross-sectional samples. Despite the reliance on the WHO standards for these assessments, significant issues arise with the analysis of population-specific growth, particularly in small-scale and genetically isolated populations. Body size and proportion vary substantially among populations worldwide, reflecting unique population histories and local ecologies (Pomeroy et al. 2021). For example, adult populations in hot, arid climates tend to exhibit relatively tall and thin body proportions, thought to reflect selection for thermoregulation (Pomeroy et al. 2021). The ontogeny of these differences in body size and proportion is poorly studied but could affect the assessment of growth and nutritional status. International growth standards do not account for potential genetic variation or adaptations to specific selective pressures within environments, like extreme heat, that could lead to differences in somatic growth pattern and timing (Cole, 2007; Ziegler & Nelson, 2012; Martin et al, 2019; Hruschka, 2020).

To address some of the limitations of international growth standards, large cross-population analyses to assess region specific variation in timing and magnitude of growth faltering from populations in developing nations have been employed (Shrimpton et al., 2001; Victoria et al., 2010; Alderman & Headley, 2018). Such work has found that most growth faltering begins at about 3 months of age and continues over the first 23 months of life, suggesting that the effects of poor nutrition and immune stress are most acutely felt in the first 1000 days of life (Shrimpton et al., 2001; Alderman & Headley, 2018). Specifically, data derived from nationally representative samples of 39 developing nations suggested that weight-for-age declined significantly between the ages of 3 and 12 months, with catch-up weight gain not occurring until about 18–19 months (Shrimpton et al. 2001). Simultaneously, weight-for-height faltering was found to be more tightly restricted to the first 15 months of life. Linear growth faltering was also found to occur in the first two years of life. However, little evidence was found for positive changes to linear growth between 2–5 years of age, and evidence for linear catch-up growth among these large-scale datasets from populations in developing nations remains disputed (Shrimpton et al., 2001; Leroy et al., 2015; Alderman & Headley, 2018). As these large-scale multinational analyses provide insight into global trends of somatic growth failure, they can be used for comparative investigations of variation in growth patterns and timing within distinct populations. Investigations of growth within small-scale populations that live in diverse and extreme environments remain scarce. Such investigations could provide us with valuable insights on adaptive variation in life history.

The objective of this study was to gain more insight on variation in growth patterns in small-scale populations living in a relatively extreme environment. We analyzed a large longitudinal dataset of early childhood nutritional status among Daasanach pastoralists in northern Kenya, a population living in a relatively remote and low-nutrition environment. The aims of this study are two-fold. First, we aimed to characterize early childhood growth, identifying the potential presence and severity of growth faltering over the early childhood age-course. Second, we aimed to analyze the Daasanach-specific pattern of growth, both in timing and magnitude, to identify distinct differences relative to large-scale studies of early childhood growth faltering. Such growth variation, identified within a genetically homogenous population, may shed light on the broader adaptive variation in adult body size and shape that is expressed across human populations, the study of which has been central to human evolutionary biology since the 19^th century (Bergmann, 1847; Allen, 1877; Ruff, 1994).

MATERIALS & METHODS

Study Population: Daasanach

Daasanach communities live in the semi-arid and arid regions of southwestern Ethiopia and northwestern Kenya (Almagor, 1978). Based on the most recent national Kenyan census, about 19,300 Daasanach live in Kenya (KNBS, 2019). A majority of Daasanach, about 48,000, live in the Omo River Valley region of Ethiopia (Mwamidi et al., 2018). Most Daasanach in Kenya live in Illeret (4.314° N, 36.227° E) or the surrounding area. Illeret is in the northwestern corner of Marsabit County, and borders Lake Turkana to the west and Ethiopia to the north (Figure 1). Encroaching markets and the related phenomenon of sedentarization have led some Daasanach to live more settled lifestyles, living in permanent structures with greater access to market goods and infrastructure. Despite this, many Daasanach continue to practice a traditional semi-nomadic movement strategy and pastoral subsistence strategy (Mwamidi, 2018).

Figure 1:

Illeret, Marsabit County, Kenya: Map of study area and the surrounding region.

The region around Illeret experiences a bimodal seasonal cycle with mean temperatures ranging from 20°C and 37°C and yearly average rainfall of about 217mm (Opiyo et al., 2014; Liebmann et al., 2014; Mwamidi et al., 2018). The historical classification of agropastoralism given to Daasanach communities (Almagor, 1987), results largely as a function of ethnographic work conducted with Daasanach living in Ethiopia, as they have increased agricultural access. Daasanach in and around Illeret are far more reliant on herding practices and access to market food staples, such as beans, rice, and maize, with very little ability to perform sustainable agriculture.

Health issues related to long-term water and food insecurity caused by increased threat of drought, flash flooding, drinking water salinity, and climate variability have also been exacerbated by longstanding economic and political isolation (Little et al., 2001; VSF RSIPA, 2011–2013; Boru & Koske, 2014; MoALF, 2017; Rosinger et al., 2021). Childhood malnutrition remains a major concern in the region, and recent work has found high levels of both food and water insecurity across Daasanach communities at the household level (Bethancourt et al., 2021). Further, work has found that water insecurity may be leading to higher food insecurity for Daasanach (Bethancourt et al., 2022). The challenges facing Daasanach living in northern Kenya have led to a number of relationships with NGOs, though they operate with varying degrees of success. To date, NGOs have focused primarily on increasing the quality of maternal healthcare, increasing access to clean water, and supplying nutritional supplementation to combat childhood malnutrition. Despite such efforts, Daasanach community members living in and around Illeret had little involvement in scientific or medical research.

Ethical Approvals

Study protocols were approved by Penn State Institutional Review Board (IRB#STUDY00009589) and the Kenya Medical Research Institute (#KEMRI/RES/7/3/1) prior to all data collection. These approvals include permissions for the collection of health records from the Illeret Health Clinic, in addition to anthropometric, survey, and biological sample collection. In-person consent for data collection was also obtained from all relevant sources, including community elders, the Illeret Ward, and medical officials from the IHC. Additionally, permitting for human biology research was obtained from the National Commission for Science, Technology and Innovation (NACOSTI).

Data Collection

The Illeret Health Clinic (IHC) has conducted systematic collections of anthropometric data for children aged 0–5 years as part of a large-scale multiyear malnutrition survey of children living across 14 communities in the greater region surrounding Illeret, Kenya since May of 2018. The distances from the IHC to communities sampled ranged from >1km to ~16km. Anthropometric measures were collected at the IHC for communities living within a short distance, typically those located within a 2km radius. Communities located farther from the IHC were visited for data collection. Families with children aged 59 months and younger were encouraged to participate, but free to decline participation. When ages were not available through IHC records, ages were estimated to the month by the IHC nutritionist.

Height (or length) was measured to the nearest 1 mm using UNICEF provided stadiometers (or infantometers), weight was measured to the 0.1kg using a digital scale (Tanita), and MUAC was measured to the nearest 1 mm using a fabric measuring tape. IHC health records were used to determine deworming status, micronutrient powder (MNP) supplementation status, and disability status. Additionally, the date of data collection, the first and last names of the subject, sex, the community of residence, and malnutrition survey re-measure status were collected. Our study team conducted medical chart abstraction on all records from the inception of the program (May 2018) through December 2019. All anthropometric data were transcribed into digital form from the hand-written IHC’s malnutrition survey records.

In total, 8,669 records were abstracted from 5,858 Daasanach children in communities within and around Illeret. We restricted our analysis to those with repeated (≥ 2) observations, thus excluding 4,082 children because they had a single observation. Initial age records from the IHC were used for children’s first age in the analysis. For all subsequent observations, we used the time since first measurement to calculate age. The ages for follow-up measurements were calculated as the age at first observation (from IHC records) plus the calendar difference in days between the first observation and all following measurements. Our reasoning with this approach was that age estimates were likely to be more accurate and precise at younger ages. The applied age correction yielded subjects’ ages that were on average 1.5 months older than their recorded ages in the IHC records at follow-up. This correction helps address underestimation of age at follow-up visits due to potential malnutrition. Further, we excluded 20 individuals with z-scores ≥ 5 or ≤ −5 for HAZ, WAZ, or WHZ measures as outliers potentially due to measurement error. These steps reduced the sample to 4,508 measures from 1,756 subjects, with a mean of 2.6 (range: 2–8) observations per individual (Table 1).

Table 1:

IHC Sample statistics – Ages (months) of subjects by community.

	Male (Age)			Female (Age)			Total (Age)
Community	N	Mean	SD	N	Mean	SD	N	Mean	SD
1	70	32.12	14.16	97	30.53	14.56	167	31.20	14.37
2	13	29.97	15.37	15	35.83	15.92	28	33.11	15.66
3	157	30.12	14.01	150	32.57	14.65	307	31.32	14.36
4	145	30.01	14.20	119	32.05	14.66	264	30.92	14.42
5	158	27.68	15.69	149	28.18	14.13	307	27.92	14.93
6	199	31.57	12.96	176	31.94	12.86	375	31.75	12.90
7	214	32.98	14.50	202	31.70	13.66	416	32.36	14.10
8	242	32.92	12.48	187	33.16	12.69	429	32.02	12.56
9	259	34.86	13.36	217	33.41	14.01	476	34.20	13.66
10	164	31.08	14.06	132	31.15	14.08	296	31.11	14.04
11	229	29.85	13.97	203	27.64	14.92	432	28.81	14.45
12	163	28.42	13.51	156	28.24	13.99	319	28.33	13.72
13	136	32.68	13.77	180	29.77	14.27	316	31.02	14.11
14	180	33.20	14.50	196	30.04	14.46	376	31.55	14.54
Total	2329	31.52	13.99	2179	30.86	14.16	4508	31.20	14.07

Analysis

All statistical analyses were completed in R (4.1.2). For the first objective of comparing Daasanach to WHO references, height-for-age (HFA) was calculated as cm/month, weight-for-age (WFA) as kg/month, weight-for-height (WFH) as kg/cm, and MUAC-for age as cm/month for all subjects. These values were plotted against WHO standards to identify the relative pattern and magnitude of growth among Daasanach infants and young children. WHO specific z-scores for height-for-age (HAZ), weight-for-age (WAZ), and weight-for-height (WHZ) were also generated using the R package ‘localgrowth’, which was previously developed and used for growth analyses of small-scale (Shuar and Tsimane’) populations (Urlacher et al., 2016; Blackwell et al., 2017; Martin et al., 2017). Additionally, MUAC-for-age values were created using the ‘MUACz’ package for all individuals for whom the measurement was taken between the ages of 3 and 59 months.

To examine growth faltering as a function of change in cross-sectional z-score over the early childhood age-course (Aim 1), sex-specific local polynomial regression (LOESS) models (α=0.25) of HAZ, WAZ, or WHZ were generated as a function of age for boys and girls. Additionally, height-for-age differences were calculated for boys and girls by assessing the difference, in centimeters, between a child’s height and the median WHO height standard for their age. This follows a similar procedure previously used for the identification of potential population-level catch-up growth among children less than five years old (Leroy et al., 2015).

For constructing Daasanach-specific centile curves for HFA (Aim 2), WFA, and WFH were generated for sex-specific samples between the ages of 0–59 months using Generalized Additive Models of Location, Shape, and Scale (GAMLSS) as part of the ‘GAMLSS’ package in R (4.0.2) (Rigby and Stasinopoulos, 2005). Modeling procedures were based on previous work modeling the growth of the Shuar and Tsimane’ farmer-horticulturist populations in South America (Urlacher et al., 2016; Blackwell et al., 2017). All models were fit with Box-Cox Power Exponential (BCPE) distributions with varying model parameters with samples that fell outside ±5 standard deviations of the Daasanach-specific curve being dropped (Cole & Green, 1992).

GAIC scores were used to determine the best fit model for the sample populations and centile models were produced using the best fit GAMLSS model parameters. Additionally, pseudo-velocity curves were produced from the 1^st derivate of mu (median) of the GAMLSS models of male and female linear and weight growth. These models allow for the identification quantification of age-related changes in growth across early childhood. WHO pseudo-velocity references based on the 1^st derivative of the median growth standards were added to facilitate comparisons in timing and magnitude of growth variation with Daasanach subjects.

RESULTS

WHO Reference Growth Curves

Rates of stunting (≤ −2SD HAZ) among Daasanach children between the ages of 0–5 years were 17.9% and 10.4% for boys and girls, respectively. Rates of underweight (≤ −2SD WAZ) were higher, with 42.4% of boys and 34.3% of girls at or below the threshold, while rates of wasting (≤ −2SD WHZ) were highest, with 56.1% of boys and 51.0% of girls observed to be wasted (Figure 2 & 3; Table 2 & 3). About half of children surveyed had mid-upper arm circumference (MUAC) values at or below −2SD, 53.1% of boys and 42.8% of girls, but only 3.9% of children had MUAC values below 115mm, the WHO defined threshold for severe acute malnutrition (WHO, 2009). However, cases of severe wasting, as defined by incidences where individuals fall at or below −3SD for WHZ, was observed in 18.5% of boys and 15.4% of girls (Table 2 & 3).

Figure 2:

Daasanach growth relative to WHO standards (total measurements = 4508; male = 2329; female = 2179). Left: Males (A-D: HFA, WFA, WFH, MUAC-for-age). Right: Females (E-H: HFA, WFA, WFH, MUAC-for-age). Black lines represent WHO standards. Solid black line = 0SD, long dashed = ± 1SD, short dashed = ± 2SD, dash dot = ± 3SD, dotted = ± 4SD. Yellow dots = individuals above −2SD, purple dots = subjects at or below −2SD (stunted, underweight, or wasted).

Figure 3:

Combined sex density distribution of HAZ, WAZ, and WHZ for Daasanach children aged 0–59 months (measurements = 4508).

Table 2:

WHO Z-Scores and growth faltering rates among Daasanach male children aged 0–59 months.

Male	Age (months)			HAZ - WHO				WAZ - WHO				WHZ - WHO
Age (months)	N	μ	SD	μ	SD	% < −2	% < −3	μ	SD	% < −2	% < −3	μ	SD	% < −2	% < −3
0–5	41	3.56	1.42	−0.39	1.34	7.3% (3)	0% (0)	−0.45	1.37	12.2% (5)	2.4% (1)	−0.18	1.52	4.9% (2)	2.4% (1)
6–11	200	8.51	1.58	−1.06	1.24	23.0% (46)	4.5% (9)	−1.94	1.16	48.0% (96)	17.0% (34)	−1.79	1.19	40.0% (80)	15.0% (30)
12–17	235	14.44	1.56	−1.53	1.18	35.3% (83)	9.8% (23)	−2.33	1.05	65.1% (153)	25.1% (59)	−2.23	1.19	55.7% (131)	26.4% (62)
18–23	217	20.39	1.91	−1.9	1.11	50.2% (109)	16.6% (36)	−2.53	1.03	70.0% (152)	28.6% (62)	−2.29	1.04	59.4% (129)	23.5% (51)
24–29	307	26.4	1.77	−1.62	0.83	29.3% (90)	4.9% (15)	−2.27	0.79	62.5% (192)	16.6% (51)	−2.06	0.93	54.4% (167)	14.3% (44)
30–35	288	32.24	1.59	−1.31	0.91	19.4% (56)	3.8% (11)	−2.07	0.83	52.4% (151)	14.2% (41)	−2.06	0.96	54.9% (158)	16.3% (47)
36–41	372	38.3	1.77	−0.62	0.87	4.0% (15)	1.1% (4)	−1.69	0.76	33.1% (123)	4.8% (18)	−2.06	0.94	54.8% (204)	15.6% (58)
42–47	338	44.33	1.87	−0.05	0.85	2.7% (9)	1.5% (5)	−1.49	0.69	19.5% (66)	3.0% (10)	−2.25	0.86	62.4% (211)	18.0% (61)
48–53	247	50.24	1.82	0.15	0.79	2.0% (5)	0.8% (2)	−1.39	0.66	15.4% (38)	1.6% (4)	−2.31	0.84	66.0% (163)	20.2% (50)
54–59	84	55.55	1.38	0.32	0.74	2.3% (2)	1.2% (1)	−1.42	0.66	14.3% (12)	2.4% (2)	−2.58	0.86	72.6% (61)	31.0% (26)
0–59 (Total)	2329	31.5	14	−0.88	1.2	17.9% (418)	4.6% (106)	−1.88	0.96	42.4% (988)	12.1% (282)	−2.11	1.04	56.1% (1306)	18.5% (430)

Table 3:

WHO Z-Scores and growth faltering rates among Daasanach female children aged 0–59 months.

Female	Age (months)			HAZ - WHO				WAZ - WHO				WHZ - WHO
Age (months)	N	μ	SD	μ	SD	% < −2	% < −3	μ	SD	% < −2	% < −3	μ	SD	% < −2	% < −3
0–5	53	3.45	1.37	0.16	1.52	9.4% (5)	0% (0)	−0.24	1.12	3.8% (2)	0% (0)	−0.39	1.57	13.2% (7)	3.8% (2)
6–11	228	8.66	1.67	−0.68	1.15	11.0% (25)	2.2% (5)	−1.6	1.12	35.5% (81)	9.2% (21)	−1.61	1.17	38.2% (87)	9.6% (22)
12–17	191	14.54	1.65	−1.04	1.29	20.9% (40)	6.8% (13)	−1.88	1.11	44.0% (84)	16.2% (31)	−1.89	1.05	49.2% (94)	14.1% (27)
18–23	209	20.37	1.78	−1.53	0.87	28.2% (59)	3.3% (7)	−2.19	0.87	57.9% (121)	17.2% (36)	−21.97	0.91	50.2% (105)	12.0% (25)
24–29	290	26.39	1.83	−1.15	1.01	14.4% (42)	3.4% (10)	−1.97	0.93	46.9% (136)	12.4% (36)	−1.88	0.98	43.4% (126)	11.0% (32)
30–35	269	32.45	1.64	−0.81	0.88	8.2% (22)	0.7% (2)	−1.85	0.82	42.8% (115)	7.8% (21)	−2.01	0.94	53.5% (144)	12.6% (34)
36–41	341	38.37	1.78	−0.39	1.04	6.5% (22)	2.3% (8)	−1.57	0.86	29.3% (100)	5.6% (19)	−1.97	1.03	49.9% (170)	15.5% (53)
42–47	306	44.39	1.88	0.12	0.81	2.0% (6)	0% (0)	−1.4	0.76	21.9% (67)	2.3% (7)	−2.19	0.91	58.8% (180)	19.3% (59)
48–53	231	50.12	1.74	0.28	0.86	2.6% (6)	1.7% (4)	−1.36	0.75	16.0% (37)	3.0% (7)	−2.36	0.87	65.8% (152)	24.2% (56)
54–59	61	55.39	1.33	0.45	0.60	0% (0)	0% (0)	−1.29	0.6	9.8% (6)	0% (0)	−2.52	1.13	75.4% (46)	41.0% (25)
0–59 (Total)	2179	30.84	14.16	−0.56	1.59	10.4% (227)	2.2% (49)	−1.66	0.96	34.3% (749)	8.2% (178)	−1.97	1.06	51.0% (1111)	15.4% (335)

Relative to WHO standards, Daasanach children were born at or near normal lengths and weights, with mean HAZ and WHZ values for children aged 0–5 months > −0.5SD (Figure 4; Table 2 & 3). However, Daasanach children experienced significant growth faltering beginning shortly after birth. Like previous work examining growth faltering across global populations, Daasanach faltering began before 6 months of age and was most severe between the ages of 18–24 months (Figure 4) (Shrimpton et al., 2001; Alderman & Headley, 2018). In this 6-month age range, the prevalence of stunting was 50.2% for boys and 28.2% for girls, with rates of underweight at 70.0% and 57.9% for boys and girls, respectively. Mean HAZ for both boys and girls returned to and then exceeded the 0SD threshold, or the median, of WHO HFA standards by about 4 years of age, while the prevalence of weight faltering continued to increase beyond the third year of life (Figure 4).

Figure 4:

LOESS models of HFA, WFA, WFH z-scores by age for Daasanach children aged 0–59 months (total measurements = 4508 male = 2329; female = 2179).Solid grey line = 0SD standard, dashed grey line = −2 SD.

When examining mean z-scores by 6-month age cohorts, the mean HAZ for boys reached 0.15 ±0.79 for those aged 48–53 months, rising to a mean of 0.38 ±0.74 for boys aged 54–59 months. Prevalence of male stunting dropped to its lowest rate for this age cohort as well, with only 2.3% being found to be stunted (Table 2). For girls, mean HAZ becomes positive (0.12 ±0.81) in the 42–47 months cohort, slightly earlier than found in boys. Mean HAZ amongst the oldest female age cohort, 54–59 months old, was 0.45 ±0.60 and there were no observed instances of stunting (Table 3). These results were consistent with the polynomial regressions of HAZ by age, which found girls and boys surpassing the 0SD threshold at about 45 and 47.5 months of age, respectively (Figure 4). Polynomial regressions for height-for-age differences, calculated as the difference in centimeters to the median height-for-age standard showed a similar pattern to HAZ analyses, with boys and girls aged 59 months having median height-for-age differences values ~2 cm and ~2.5 cm greater than the median standard values, respectively (Figure 5).

Figure 5:

Height-for-age differences (WHO standards) for male and female Daasanach children under five years of age (total measurements = 4508; male = 2329; female = 2179).

While mean HAZ increased dramatically among Daasanach children over the course of early childhood, mean WAZ only increased slightly from the lowest observed means of −2.53 ±1.03 for boys aged 18–23 months and −2.19 ±0.87 for similarly aged girls to −1.42 ±0.66 for boys and −1.29 ±0.60 for girls in the 54–60 months cohort. However, this change of about +1SD was matched by a large decrease in underweight prevalence, from 70.0% to 14.3% for boys aged 18–23 months and 54–59 months, respectively, and from 57.9% to 9.8% for similarly aged girls (Table 2 & 3). While mean HAZ and WAZ both increased after initial growth faltering in the first 24 months of life, mean WHZ continued to decrease over the course of early childhood, matched by a marked increase in wasting prevalence after 41 months of age (Figure 4). Mean WHZ among boys (−2.58 ±0.86) and girls (−2.52 ±1.13) aged 54–59 months were the lowest observed for any sex-specific age group across all measures. Similarly, wasting prevalence increased to 72.6% and 75.4% for boys and girls, respectively, after a period of relative stability in wasting prevalence between 12–41 months (Figure 4; Table 2 & 3).

Daasanach Specific Growth Curves

Sex-specific centile curves for Daasanach height, weight, and weight-for-height for children aged 0–5 years are presented in Figure 6. Pseudo-velocity curves of linear growth and weight gain constructed using GAMLSS found positive acceleration for linear growth between ~22–36 months old for males and ~20–36 months old for females (Figures 7). Likewise, there was an increase in weight gain velocity for males between the ages of 18–35 months and females between the ages of 17–34 months (Figure 8). Linear growth velocity peaked at ~0.98cm/month for boys and girls aged ~36 months, a rate greater than the median WHO velocity standards for children aged 22–24 months (WHO, 2006). Weight gain velocity peaked slightly earlier than linear growth, at a value of nearly 0.20 kg/month for boys and girls aged ~35 months.

Figure 6:

Centile curves for Daasanach children aged 0–59 months (total measurements = 4508; male = 2329; female = 2179). Red lines = 5^th and 9^th centiles; Blue lines = 25^th and 75^th centiles; Black = 50^th centile. Measures from the Illeret Health Clinic malnutrition survey.

Figure 7:

Pseudo-velocity curves produced from GAMLSS modeling of linear growth for Daasanach children aged 0–59 months (total measurements = 4508; male = 2329; female = 2179. Black line = median of spline model of linear growth (HFA), orange line = 1^st derivative of median linear growth spline parameter, green line = loess model of WHO 1^st derivative of median HFA.

Figure 8:

Pseudo-velocity curves produced from spline modeling of weight gain for Daasanach children aged 0–59 months (total measurements = 4508; male = 2329; female = 2179). Black line = median of spline model of linear growth (WFA), orange line = 1^st derivative of median weight gain spline parameter, green line = loess model of WHO 1^st derivative of median WFA.

DISCUSSION

This study aimed to characterize early childhood growth among Daasanach living in northern Kenya and compare Daasanach-specific patterns of growth to international standards for children aged 0–59 months. We found that Daasanach children displayed patterns of early childhood growth that were not congruent with those established by the WHO. While the pattern of faltering in Daasanach children in the first 24 months of life largely match those described for other populations in developing regions of the world, they deviate beyond 24 months (Dewey, 1997; Victoria et al., 2008; Hermanussen et al., 2016; Alderman & Headley, 2018). Daasanach children are born at relatively normal heights and weights relative to WHO standards, but quickly begin to show both linear and weight growth faltering before 6 months of age (Figure 4). As linear faltering and weight gain faltering continue throughout the first 24 months of life, prevalence of stunting and underweight peaks between the ages of 18–23 for both males and females. Unlike other populations in low- and middle-income countries, both Daasanach boys and girls show significant gains in linear growth relative to WHO standards after about 24 months of age and throughout the rest of early childhood. This results in Daasanach children achieving average HAZ greater than WHO standards by 5 years of age, and stunting rates below 5% after 42 months of age.

Comparisons with Other Populations and International Standards

Growth among the Daasanach contrasts with the globally observed pattern of faltering, which sees weight gain increasing after about two years of age when faltering is present, but consistent with rates of stunting continuing through early childhood. The Daasanach growth faltering patterns also contrast with those of Datoga pastoralists in Tanzania, which were found to have consistently low WFA and HFA values across childhood and adolescence when compared to National Center for Health Statistics (NCHS) growth reference (Sellen, 1999). Likewise, much higher rates of stunting have been observed among Maasai infants and children in Kenya (Galvin et al., 2015). However, like Daasanach, the rates of wasting and underweight among these same Maasai individuals also increase from infants to children to juveniles. The Daasanach pattern of linear growth velocity, marked by an increase in velocity at about two years of age, did not match those observed among young children in the neighboring Turkana (Little et al., 1983; Galvin, 1992, Gray et al., 2004; Gray et al., 2008). It is important to note, however, that the height velocity of Turkana children more closely matches US standards than does weight velocity, much like Daasanach (Little & Johnson, 1987). More anthropometric data for children under the age of five in other populations would enhance our understanding of linear growth among Daasanach and other East African pastoral populations.

Daasanach children continue to have relatively low cross-sectional mean WHZ across the first 59 months of life (male = −2.11 ±1.04; female = −1.97 ±1.06) and have a particularly high prevalence of wasting throughout early childhood (male = 56.1%; female = 51.0%). By WHO standards, rates of wasting are at their most severe between 54–59 months of age, the last age cohort measured here, with 72.6% of boys wasted (31.0% severely wasted) and 75.4% of girls wasted (41.0% severely wasted) (Table 2 & 3). These remarkably high rates of wasting appear to reflect the distinctive pattern of Daasanach growth rather than malnutrition. Despite wasting among nearly three quarters of all children aged 54–59 months, rates of underweight for those same children are 14.3% and 9.8% for males and females, respectively. Overall prevalence of stunting in our sample is less than half the estimated stunting prevalence for preschool children in the UN sub-region of Eastern Africa as predicted based on global trends in 2010 (mean = 43.9%). In fact, only the prevalence of stunting for male children aged 18–23 months (50.5%; 95% CI: 38.0, 50.0) falls within the 95% confidence interval for this prediction (de Onis et al., 2012). Additionally, there is a significant difference in the rate of severe acute malnutrition detected by the MUAC cutoff of 115mm and the WHZ cutoff of −3SD. While 17.0% of children in our sample were deemed severely wasted based on WHO guidelines, fewer than 4% have an MUAC that would categorize as severely wasted by those same guidelines (WHO, 2009). Together, these results suggest that patterns of wasting in Daasanach children is likely a result of their unusually high rates of linear growth.

Further evidence of important differences between early childhood growth in Daasanach children and model standards was identified through cross-sectional modeling. Daasanach-specific growth patterns exhibited an increase in both linear growth velocity and weight gain at ~24 months of age, which continued until about 48 months of age in both boy and girls (Figure 7 & 8). At its peak at ~3 years of age, linear growth velocity reaches nearly 1cm/month for boys and girls, which is ~0.3cm/month higher than the “normal” median growth velocity for children 3 years of age (WHO, 2006). Additionally, this pattern was temporally distinct from the mid-childhood growth spurt, which is generally associated with adrenarche and a moderate increase to growth velocity between the ages of 6–8 (Katz et al., 1985; Bogin, 1999). It is possible that this increase in velocity is associated with weaning, though the direction of the effect is opposite the expected decrease in growth at weaning (Rehman et al., 2009). While both linear growth velocity and weight gain velocity increased in a similar fashion temporally, the magnitude of change in weight gain is relatively lower to linear growth change. Additional analysis of the change in HAZ and WAZ relative to WHO standards also suggested a larger increase in linear growth than weight gain. This discrepancy between patterns of linear growth velocity and weight gain velocity helps explain the high prevalence of wasting – or low WHZ – across Daasanach children, especially those children aged 42–59 months. Overall, these findings suggest that international standards are often inappropriately applied to small-scale populations to identify growth faltering.

Ontogeny of Adaptive Differences in Body Shape and Size

Previous studies of growth among East African pastoralists have suggested that the lean and tall statures of adults in these communities result from extended growth throughout adolescence and into early adulthood despite relatively slow overall linear growth rates (Little et al., 1983; Little & Johnson, 1987; Galvin et al., 1992). Results from the current study suggest the distinctive tall, thin adult phenotype has ontogenetic origins that are evident in earlier in childhood. Daasanach children appear to allocate relatively more energy to linear growth, and less to weight gain, than is typical for populations represented in WHO and other international growth standards. The allocation of energy to linear growth among the Daasanach is notable given the challenges of living in a relatively calorically sparse environment. Undernutrition remains a concern for Daasanach, with high rates of water and food insecurity (Bethancourt et al., 2021). Allocation of limited food energy to linear growth has also been noted in studies of Turkana pastoralists, in which individuals were reported to show positive linear growth despite caloric deficits and negative weight gain (Little & Johnson, 1987).

The growth pattern identified among Daasanach children may reflect broader forces of adaptive variation in body shape, as outlined by Bergmann’s and Allen’s “rules” (Bergmann, 1847; Allen, 1877). These “rules” consider the ecological relationship between body size (Bergmann) and body proportionality (Allen) variation as a function of thermoregulatory adaptation and suggests those that live in warmer climates evolved slimmer bodies with longer limbs, while humans living in colder climates evolved bodies with relatively higher mass and shorter limbs (Ruff, 1991; Ruff, 1994). The square-cube law, which describes the relationship between the surface area and volume of a three-dimensional object as it changes size, dictates the physiological basis for the fitness advantages of differing body sizes and shapes (Schmidt-Nielson, 1984). Daasanach adults match the adaptive expectations of both Bergmann’s and Allen’s rules, being relatively tall in adult stature, with very low adult body mass indices, while living in a hot and very arid environment (Opiyo et al., 2014; Wells et al., 2019; Bethancourt et al., 2021).

Limitations

A potential limitation of this work is its use of age estimation, which can lead to error in age-dependent growth measures (Diekmann et al., 2017). To reduce the effect of this factor, all age estimations were made by a single observer who was a government sponsored nutritionist familiar with the local population. This limitation was also mitigated through the use of an age correction method that produced new age estimates based on the difference in dates between the first and subsequent measurements. This method of calculating age at subsequent observation potentially addressed issues of underestimation of age at follow-up visits as the ages we calculated were 1.5 months older than the recorded follow-up ages on average. It is not surprising that the follow-up observed estimated ages were younger than the time passage recorded between visits as stunting and wasting through malnutrition can make children appear younger than they are. Had we used the recorded ages in the analysis rather than the calendar corrected ages, the growth measures would have been biased in such a way as to produce higher HAZ and WAZ scores due to the younger estimated ages relative to anthropometric measurement. Additionally, we limited the analytic sample to only those subjects with multiple recorded measurements, reducing our overall sample by about half. This correction is unable to account for potential cohort effects, however, which could account for similar recent population-wide variation caused by a shared stressor like a drought. This effect is, in part, mitigated by the relatively high number of communities from which these data are derived. Anthropometric measures were collected from children over the course of three years across 14 communities that ranged from <1 km from the IHC to ~16 km. Despite this, the potential effects of wide-spread environmental variations cannot entirely be accounted for within this sample. Patterns of growth for this sample were also analyzed in combination with anthropometric data collected by the Daasanach Health and Life History Project, finding that the IHC sample falls within expected anthropometric ranges. The robusticity of the results of this study are bolstered by the relatively large size of the sample and the statistical techniques used for all analyses. Future data collection and expanded maternal and childhood healthcare efforts by the Illeret Health Center will contribute to the longitudinal expansion of these already robust analyses.

CONCLUSIONS

This study presents growth analyses of a large dataset of population-specific measures from Daasanach children in early childhood, with additional comparisons to WHO standards for the identification of growth faltering. This study identified a pattern of growth in Daasanach children that is distinct from expected patterns of growth faltering or linear growth velocity (WHO, 2009; de Onis et al., 2012). We found that after initial faltering in linear growth and weight gain during the first 24 months of life, Daasanach children showed a significant decrease in linear growth faltering that resulted as a function of increased growth velocity after two years of age. Such patterns of linear growth and weight gain may be indicative of a thermoregulatory adaptation to the exceptionally hot, arid environment of northwestern Marsabit County, Kenya. Further collection of cross-sectional and longitudinal data, as well as the inclusion of anthropometric data from individuals in late childhood and adolescence, is vital for understanding how phenotypic plasticity and adaptation have affected early childhood growth patterns. Establishing local, population-specific patterns of growth will improve our understanding of human biological variation and aid in the assessment and mitigation of growth faltering and malnutrition.

DATA AVAILABILITY STATEMENT

Data available upon reasonable request due to privacy and ethical restrictions.