Severe psychosocial deprivation may result in psychosocial dwarfism, including stunted growth, retarded somatic maturation, retarded cognitive development, and other psychosocial and behavioral problems (Money, 1992). These children are often hyposomatotrophic but usually regain normal levels of growth hormone with treatment (Albanese, 1994). Often, nutritional insufficiency accompanying the psychosocial deprivation has been suggested as an important, or even the primary factor, leading to the impacts on physical growth and maturation, although panhypopituitarism has also been suggested (Money, 1992). Most studies of these children are based on small clinical series and the nutrition hypothesis has not been satisfactorily evaluated.

We have reported extremely retarded physical growth and less retarded sexual maturation for a large series (n=255) of Romanian children 5-18 years of age who spent their entire lives in orphanages with severe psychosocial and nutritional deprivation (Himes et al., 2009). At the time that report was developed data concerning skinfold thickness that had been measured on the children were not included because skinfolds were only available for part of the sample (74%), and there were no skinfold reference data available allowing assignment of centile and Z-score rankings relative to age and gender. Since that time, reference curves for triceps and subscapular skinfold thickness have been published for US children (Addo and Himes 2010). These reference curves are based on the same samples as those included in the CDC 2000 Growth Charts (Kuczmarski et al., 2000) and therefore allow direct comparisons with other anthropometric indicators.

Herein, we investigate the patterns of development in triceps skinfold thickness as an indicator of long-term energy balance in 188 of the same Romanian children, and compare triceps skinfold status with growth in height, weight, BMI, and head circumference.


The sample comprises children living in six orphanages in Romania who were examined in 1997-2001 as part of the effort to close the worst institutions and accommodate the children elsewhere. Height, weight, triceps skinfold, and head circumference were measured using standard protocols (Lohman et al., 1988). Triceps skinfold thickness was measured using Lange calipers by the same examination team. More details concerning the sample and methods are presented elsewhere (Himes et al., 2009). As a group, the subset of 188 children with skinfold measurements included somewhat higher proportions of males and prepubescent children than the complete sample, although there were no statistically significant differences between the means of the growth variables for the two groups based on availability of triceps skinfold measurements.

Body mass index (BMI) was calculated as weight (kg)/ height (m)2. Age- and gender-specific centiles and Z-scores for height, weight, and BMI were calculated relative to the CDC 2000 Growth Charts for US children (Kuczmarski et al., 2000). Corresponding centiles and Z-scores for triceps skinfold thickness relative to US children were calculated according to Addo and Himes (2010), and those for head circumference were estimated relative to white Ohio children (Roche et al, 1987) because those from the CDC Growth Charts only extend to 36 months of age. Stages of development of pubic hair and breast in girls, and pubic hair in boys were assessed by inspection following conventional criteria (Tanner, 1962). Testicular volume was assessed using a Prader orchidometer (Zachmann et al., 1974).

Because the sample was drawn from six different orphanages, statistics were estimated from general linear models, when appropriate, using orphanage as a random effect to account for any orphanage-related intraclass correlations.


Means of height, weight, BMI, and head circumference were presented previously by sex and age groups (Himes et al., 2009). Means and standard deviations for triceps skinfold thickness are presented in Table 1 for each gender and within 2-year age groups. As points of reference, means of height and BMI also are presented for those 188 children for whom triceps skinfolds were measured.

The patterns of growth in triceps skinfold across age are presented as Z-scores in Figures 1 and 2 relative to other growth variables measured. The dramatic retardation in growth in height and weight is apparent in both genders, fluctuating between -3.0 to -5.0 Z in boys, and -4.0 to   -6.0 Z in girls. Head circumference was considerably retarded, approximating -2 Z in boys and -3 Z in girls throughout the age range, but apparently it was spared somewhat relative to more severe impacts on height and weight. With height and weight similarly retarded in each gender, the mean BMI Z-scores were substantially less affected, but still ranged from -1.0 to -2.0 Z in both genders across the age range.

Surprisingly, mean Z-scores for triceps skinfold thickness across the ages did not approximate the extreme patterns of weight retardation, but demonstrated a sparing of subcutaneous fat so that in boys skinfolds were at approximately -1.5 Z at 6 and 7 years , and increased throughout the age range to exceed the reference median in the oldest age group, reaching 0.5 Z. For girls, Z-scores for triceps skinfolds approximated -2 Z throughout the age range, but still demonstrated substantial sparing of effects compared with height and weight.

Considering all children and ages, the mean triceps Z-score for boys (-0.76; SD, 1.07) was significantly larger than that for girls (-1.62; SD, 1.17), with a t-value of 5.21 (p<.001). The right-hand skew usually characteristic in distributions of skinfold thicknesses were not evident in these Romanian children, such that gender-specific Z-scores for triceps skinfold thickness pooled across age did not depart significantly from Gaussian normality using the Kolmogorov-Smirnov test (boys p=0.301; girls p=0.581).

Expected hormone-mediated patterns of subcutaneous fatness associated with sexual maturation were evident in the Romanian children. As an example, median triceps skinfold thicknesses are presented by stage of pubic hair development in Figure 3 for both genders. For girls, median triceps skinfolds increased steadily from approximately 6 mm at stage 1 to approximately 12 mm at stage 5. For boys, median triceps skinfold thickness increased from 6 to 8 mm from pubic hair stages 1 to 3, and decreased thereafter to approximately 6.5 mm at stages 4 and 5. Patterns of development of triceps skinfold relative to testicular volume in boys and stages of breast development in girls proceeded in corresponding fashion appropriate to gender (data not shown).

Pearson correlation coefficients for Z-scores for triceps skinfold and age, and correlations with the Z-scores for other growth variables are presented in Table 2 for boys and girls. The pattern of increasing Z-scores for triceps skinfold across the ages seen in Figure 1 for boys is statistically significant by correlation coefficient ( r=.396, p< .001), although no corresponding systematic pattern with age is observed in girls. Triceps skinfold Z-score is positively and significantly correlated with Z-scores for height in girls, and with weight, BMI, and head circumference in both genders, although at slightly higher levels in girls than boys.


These Romanian children present dramatic patterns of growth retardation, reflecting lives of severe deprivation, almost certainly including nutritional insufficiency and psychosocial neglect. Mean Z-scores of -3.0 to -6.0 for height and weight are extreme even for children with severe protein-energy malnutrition (Waterlow, 1992). There are few corresponding data for head circumference in severe undernutrition, but it appears that head growth in the Romanian children was considerably spared from the effects of depravation compared with the extreme effects on height and weight. While the reference curves for head circumference were derived from the Fels Longitudinal Study, rather than the US population, the Fels data correspond quite closely to the US national data (Roche, 1992). Actually, any small differences between Fels and the national data are such that the resulting Z-scores based on the Fels data may slightly underestimate Z-scores for head circumference because the Fels children are slightly taller than the national US sample.

The relatively high levels of Z-scores for triceps skinfold thickness given the dramatic retardation in height and weight was unanticipated. While absolute values of mean skinfold thickness are low relative to the reference data, especially in girls, the disparity between the Z-score levels for weight and triceps in both genders suggests eitiologies for the extreme retardation in height and weight other than simple chronic protein-energy deficiency. Because the reference data for skinfold thickness (Addo and Himes, 2010) were based on the same sample of children used for the CDC Growth Charts (Kuczmarski et al., 2000), the observed disparities between Z-scores for triceps skinfold and height and weight are not due to the reference data used.

It is significant that expected patterns of development of triceps skinfold relative to stages of sexual maturation are qualitatively similar to those seen in better-off populations (Buckler, 1990), even though the median levels of subcutaneous fatness are relatively low in the Romanian children. The steady increase in median triceps skinfold in girls across pubic hair stages, and initial rise then decrease in median triceps skinfold in boys in the latter pubic hair stages suggest that the Romanian children did not have pan hypopituitarism because of the pituitary hormones that would have played roles in the subcutaneous fatness changes and progress of pubertal stages (Rogol et al., 2002). Growth hormone was not available for these children so it is unknown whether they were hyposomatotrophic.

The appreciable correlations between Z-scores for triceps skinfold and those for weight, BMI, and head circumference indicate there was sufficient variation in triceps skinfold thickness to covary significantly with other measures of somatic growth. This is an expected pattern in most adequately nourished and even marginally malnourished populations. Usually, however, with chronic protein-energy undernutrition there is such universal depletion of subcutaneous fat stores that skinfold thicknesses cease to be very useful as indicators of long-term energy balance because they are uniformly extremely low, and accompanied with very little variation (Himes, 1980).

These data for Romanian orphans indicate dramatic growth retardation associated with spending their lives in severely deprived circumstances. The data on triceps skinfold thickness, however, suggest that the extreme retardation in height and weight was probably due to factors beyond chronic dietary energy deficiency. Z-scores for triceps were consistently greater (less negative) than those for height and weight. Moreover, variation in triceps skinfold thickness was sufficient to significantly covary with other growth variables in both genders. We conclude that the extreme dwarfism was brought about by multifactorial psychosocial depravation cumulatively experienced by the Romanian children, probably including undernutrition, but with effects beyond what would be expected from chronic energy insufficiency per se. The fact that the extreme stunting observed with this population is accompanied with less severely delayed onset of puberty (Himes et al., 2009) also departs from what would be expected with less complicated chronic dietary deficiency.

We are aware of no other similar data for comparison. Because the available clinical data follow severely deprived children through the course of treatment (Money, 1992) the natural history of such long-term and untreated psychosocial deprivation is largely unknown. The data on the Romanian children are cross-sectional, with all the caveats for interpretation of findings that this entails. Nevertheless, all of the children had spent their entire lives in these orphanages, and it is likely that differences across ages primarily represent effects of cumulative exposure to the psychosocial depravation rather than differential cohort effects from sampling different children over the years covered.

It is difficult to describe the plight of children growing up in such extreme circumstances. Nevertheless, it is important to understand, as much as possible, the biological consequences for insights into the diagnosis and management of children in more common though less severe circumstances. For example many children, world-wide, still reside in institutions and probably encounter psychosocial deprivation, hopefully to a lesser degree. Overall, the biological adaptation and resilience of children remains intriguing scientifically, and hopeful from a humanitarian view.

Table 1. Mean measurements for Romanian orphanage children

Age (y) N Height (cm)   BMI (kg/m2)   Triceps Skinfold (mm)  
    Boys   Mean SE Mean SE Mean SE
5-6.99 10 97.7 4.91 14.5 0.77 5.75 1.10
7-8.99 11 100.6 3.90 15.2 0.61 5.51 0.87
9-10.99 11 112.7 4.91 14.7 0.77 6.46 1.10
11-12.99 24 119.9 3.06 15.6 0.48 6.99 0.68
13-14.99 24 130.3 3.57 16.2 0.56 7.01 0.80
15-16.99 14 145.1 4.10 17.4 0.64 6.64 0.92
17-18.99 5 156.3 6.32 18.6 0.99 10.11 1.41
5-6.99 3 90.0 7.47 13.5 1.74 5.00 2.20
7-8.99 9 98.7 5.30 14.7 1.24 6.20 1.56
9-10.99 13 112.8 4.01 14.6 0.94 5.67 1.18
11-12.99 26 125.8 2.91 15.2 0.68 6.83 0.86
13-14.99 17 130.5 3.81 16.8 0.89 8.72 1.12
15-16.99 14 141.1 4.42 18.8 1.03 8.93 1.30
17-18.99 7 144.7 6.40 22.6 1.49 9.27 1.89

Table 2: Pearson correlation coefficients for Z-scores for triceps skinfold with age and other growth variables

Variables Boys (n=99) Girls (n=89)
Age .396 *** .008
Height .195 .462 ***
Weight .303 ** .564 ***
BMI .365 *** .548 ***
Head Circ. .257 ** .391 ***

** p ≤ 0.01; ***p ≤ 0.001

Figure 1: Mean Z-scores for growth variables for Romanian Boys

Romanian Boys

Figure 2: Mean Z-scores for growth variables for Romanian Girls

Romanian Girls

Figure 3: Median triceps skinfold thickness by stages of pubic hair development in Romanian children

Romanian children