Contents - Previous - Next


INCAP longitudinal study (1969-77) and its follow-up study (1988-89)

Data from studies carried out by the Institute of Nutrition of Central America and Panama (INCAP) in Guatemala were analyzed to address a number of issues unresolved by previous studies. These are the INCAP Longitudinal Study and its Follow-up, described in detail elsewhere (Martorell et al, 1995). The follow-up permits linking data from a comprehensive longitudinal study of pregnancy and early childhood growth and development conducted in 1969-77 to later information gathered in 1988-89 when the subjects were adolescents and young adults. One of the studies reviewed in Table 1 was based on data from the INCAP Longitudinal Study (Villar et al, 1984).

The studies took place in four villages located in the eastern part of Guatemala. The populations were "Lading" (Spanish speaking; Spanish/Indian ancestry) and relatively poor. Subsistence agriculture was the principal livelihood in the 1970s. Poor nutrition and high rates of infection were common in young children.

The 1969-77 longitudinal study focused on women during pregnancy and on preschool children. The data available includes birthweight, age, and postnatal growth. The 1988-89 follow-up study provided information on body size and composition, strength, intellectual achievement and other factors.

1) Hypotheses

Based partly on the literature review, the analyses presented here tested the following hypotheses.

Relative to control subjects born with adequate birthweights:

- IUGR cases experience partial catch-up growth in the first three years of life.

- IUGR cases nonetheless are smaller as adolescents and adults and this is independent of postnatal growth retardation.

- IUGR cases with lower ponderal indices experience greater catch-up growth than those with larger ponderal indices.

- IUGR cases are fatter and have greater waist/hip ratios at adolescence.

- The long-term consequences of IUGR are similar in males and females.

- Finally, better conditions in developed countries are expected to ameliorate the effects of IUGR. Therefore, differences between IUGR cases and controls in height and weight at adolescence are predicted to be larger in Guatemala than found in developed countries.

2) Variables

The variables used are given in Table 2. All subjects with complete information from the longitudinal study and the follow-up were selected. Sample sizes were 20 male and 19 female IUGR cases and 149 male and 143 female, non IUGR controls. The sample used (n = 331) represents 44% of all subjects (n = 752) whose birthweights were determined during the longitudinal study (1969-77). The requirement that subjects also have length measurements at 15 days and at or around three years of age, and exclusion of twins and cases with gestational age less than 37 weeks or greater than 43 weeks reduced the sample from 752 to 337 children. Among those with missing information are children born at the end of the longitudinal study (1975-77) who could not have been measured at 3 years of age by the time the study ended in 1977. Anthropometric data, from the follow-up study were available on 292 of the 3,37 subjects selected before, after excluding 7 cases who were pregnant when measured at follow-up. The sample of 292 cases with complete data did not differ significantly from the entire sample of 752 in terms of birthweight, gestational age, ponderal index, length at 15 days, length at 3 years, and percent male.

3) Analysis

Three birthweight groups were defined. One was made up of IUGR cases (< 2500 g) and the other two were partitions of the control sample into a "middle" (2500-3000 g) and upper group ( 3001 g). In some analyses, the IUGR sample was divided into those above and below the median PI. Gestational age was used as a covariate when contrasting cases to controls; this method was preferred for its simplicity to defining cases and controls in reference to birthweight distributions for gestational age, a procedure which uses an external reference population. The WHO/NCHS reference population was used in computing Z-scores for length at 15 days and height at 3 years of age. The difference between these Z-scores (DZ=Z36-ZB) was used as a measure of postnatal growth retardation.

Table 2. Variables selected for study

Longitudinal Study (1969-77)

BW

Birthweight (kg)

Measured with a beam balance within first 48 hrs to the nearest centigram (xx.x kg).

GA

Gestational age (wks)

Measured from date of last menstruation to date of delivery. Visits to the home every 2 wks provided the menstruation data. Range restricted to 37 to 43 wks in this analysis.

PI

Ponderal Index

[(Birthweight (g)/Newborn Length (cm))] x 100. Length measured at 15 days (see LB)

LB

Newborn Length (cm)

Assessed at 15 days using a measuring board. Measured to the nearest millimeter (xx.x cm).

ZB

Newborn Length z-score (cm)

Length for age in standard deviation units using the WHO/NCHS reference population.

Z36

Height z-score at 36 months

Actual or predicted (from data at 24-60 months). Height-for-age at 36 months was used if available or, when not available, it was estimated using regressions with data taken at 24, 30, 42, 48, or 60 months of age. Height-for-age Z-scores in this population are highly stable after 24 months of age (Martorell et al, 1995) and the R for all regression models predicting height-for-age Z-scores at 36 months was above 80%.

Dz

Change in height z-score from birth to 36 months.

Dz = Z36-ZB.

Follow-up Study (1988-89)

Age

Age in years at follow-up.

From recorded births and date of examination.

SA

Skeletal age.

From x-rays of left-hand and wrist using the method of Tanner-Whitehouse (see Pickett et al, 1995 for details).

Ht

Height at follow-up (cm).

Measured with stadiometer (xxx.x cm)

Wt

Weight at follow-up (kg)

Measured with a beam balance (xxx.xx kg)

FFM

Fat-free-mass (kg)

Predicted from anthropometry+

FM

Fat-mass (kg)

Predicted from anthropometry+

% F

% body fat

Predicted from anthropometry+

WHR

Waist/Hip ratio

Ratio of circumferences x 100

SR

Strength of the right arm

Measured by a hand-held dynamometer (xx kg)

SL

Strength of the left arm

Measured by a hand-held dynamometer (xx kg)

+ Predicted from anthropometry from equations based on hydrostatic weighings of Guatemalan subjects of similar age and body composition as the sample (see Conslisk et al, 1992 for details).

The anthropometric characteristics during early childhood and at follow-up of IUGR, middle and upper groups were assessed using analysis of covariance. These analyses adjusted for gestational age and age at follow-up when appropriate. Multiple regression analyses were also carried out for which dummy variables for the IUGR (IUGR = 1, others = 0) and upper groups (Upper = 1, others = 0) were created. These analyses also included Dz and gestational age as independent variables.

4) Results

Results are presented in two parts: by whether the outcomes were assessed during the 1960-77 longitudinal study or the 1988-89 follow-up.

Early childhood results

Selected characteristics at birth and at 3 years are given for males and females in Table 3. There were no significant differences in gestational age by birthweight group. Nonetheless, all other variables in Table 3 were adjusted for gestational age. Twenty of 169 male newborns (11.8%) were IUGR; for females, the number was 19 of 162 newborns (11.7%). As expected, birthweight, birthlength and ZB differed significantly across birthweight group. Z36 was most negative in the IUGR group, about - 3.0 S.D., was about - 2.5 S.D. in the middle group and was least negative in the upper group, about - 2.3 S.D. Postnatal growth retardation, as measured by Dz was least in the IUGR group, about - 1 S.D. and greatest in the upper group, about -1.5 S.D. The ponderal index was strongly, positively, and linearly related to birthweight (Figure 1); the correlation between PI and BW was .545 in males (p < .001) and .642 in females (p < .001).

The IUGR sample (sexes pooled) was divided according to whether values were above or below the median PI.

Table 3. Mean characteristics of sample in early childhood by birth weight group*


IUGR (x)

Middle (x)

Upper (x)

Pooled S.D.

P

Males


n=20

n=61

n=88



GA (wks)†

39.3

39.8

40.0

1.4

.169

BW (kg)

2.31

2.79

3.43

0.28

< .001

LB (cm)

47.6

49.7

50.9

2.1

< .001

ZB

-2.15

-1.23

-0.73

0.86

< .001

Z36

-3.19

-2.51

-2.27

0.93

< .001

Dz

-1.04

-1.28

-1.53

1.01

.149

PI (w/L)

2.16

2.30

2.62

0.34

< .001

Females


n=19

n=72

n=71



GA (wks)†

40.3

40.0

40.2

1.4

.591

BW (kg)

2.36

2.81

3.38

0.22

< .001

LB (cm)

47.7

49.0

50.1

1.7

< .001

ZB

-1.87

-1.26

-0.75

0.73

< .001

Z36

-2.84

-2.51

-2.36

0.88

.143

DZ

-0.97

-1.25

-1.61

0.98

.032

PI (w/L)

2.20

2.41

2.69

0.28

< .001

*IUGR £ 2500g, Middle 2501-3000g, Upper 3001 g. Gestational age 37 wks and £ 43 wks for all cases.

†Values below dotted line adjusted for gestational age.

Mean characteristics are shown for the lower half ponderal index group (LPI) and the upper half ponderal index group (UPI) in Table 4. The LPI group was longer at birth and at 3 years of age but had less catch-up growth than the UPI group.

Linear growth dynamics from birth to three years are displayed in Figure 2. Patterns were similar in boys and girls and for clarity the results were pooled by sex; values, however, are adjusted by sex as well as by gestational age for IUGR, middle and upper birthweight groups (right half of Figure 2) and by sex, gestational age and birthweight for LPI and UPI groups (left half of Figure 2). The prenatal component is of greater importance for size at 3 years for the IUGR group and of least importance for the upper group. As suggested by the literature, there was greater catch-up in length in the IUGR group than in the upper group. In the middle group, prenatal and postnatal components were of equal magnitude. When the IUGR group is divided by PI, two distinct patterns emerge. The LPI group follows the pattern of the middle birthweight group and is similar in size to this group at 3 years of age. The UPI is the shortest at birth but has the least degree of postnatal growth retardation of any group (i.e. greatest degree of catch-up growth).

Follow-up results

Means for body size and composition and for strength at adolescence are given in Table 5. The mean age at follow-up was 14.9 1.5 yrs in males and 14.8 1.5 yrs in females and did not differ by birthweight group. Nonetheless, values were adjusted for age at follow-up as well as for gestational age. Skeletal age also did not differ significantly by birthweight group in either sex. Age, rather than skeletal age was used as a covariate because the latter is truncated at maturity whereas age is continuous. Age is a better covariate because it captures information about body composition changes which occur beyond skeletal maturity.

A representation of the findings in Table 5 is given in Figure 3. Differences between IUGR and the upper birthweight group (i.e. mean values for IUGR group less mean values for upper group) were expressed as effect sizes ("d"), that is, as percent of the pooled standard deviation (Cohen, 1977). An effect is considered large if d=.8, medium if d=.5 and small if d=.2 (Cohen, 1977). Effect sizes were large to medium for height, weight, FFM and FM, medium to small for strength, and small to non-significant for %F and WHR (Figure 3).

The characteristics at follow-up of the LPI and UPI groups are given in Table 6. Differences are not statistically significant for any variable, largely due to the low statistical power. However, there is a tendency for the UPI group to be shorter, lighter and leaner than the LPI group. The mean stature (sexes pooled) of the UPI group was 145.9 cm, lower than that of the LPI (147.5 cm), Middle (150.9 cm) and Upper (151.7 cm) groups (Tables 5 and 6). Weights were 37.5, 40.0, 42.5 and 43.7 kg respectively for UPI, LPI, Middle and Upper groups (Tables 5 and 6).

Multivariate analyses were carried which included dummy variables for IUGR and UPPER groups (the middle group was used as the reference), gestational age, and age at follow-up as covariates. The results are given Table 7 for males and Table 8 for females. Gestational age was not a significant predictor of any of the outcomes but as expected, age at follow-up was strongly related to all variables. In males, IUGR was associated with a reduction of 5.9 cm in height, 3.5 kg in weight, 0.7 kg in fat and 2.9 kg in FFM, all relative to the middle group (Table 7). IUGR males were also weaker and there was a tendency for them to have a lower %F. Males of the upper group had a tendency, but never statistically significant, to be larger than the middle group. A decline of one standard deviation in length Z-score between birth and 3 years of age was associated with a reduction of 2.6 cm in height, 1.9 kg in weight, 0.4 kg in FM, 1.4 kg of FFM as well with lower % F and reduced strength.

Figure 1. Relationship between birth weight (BW) and ponderal index (PI).

Table 4. Mean characteristics of IUGR sample in early childhood by ponderal index group+


LPI

UPI

Pooled S.D.

P


n=20

n=19



GA (wks)

39.6

40.0

1.62

.445

BW (kg)

2.31

2.36

0.17

.408

LB (cm)

49.15

45.98

1.23

< .001

ZB

-1.42

-2.65

0.47

< .001

Z36

-2.68

-3.40

0.87

.016

Dz

-1.25

-0.74

0.90

.093

PI (w/L)

1.97

2.40

0.15

< .001

+Lower Half Ponderal Index (LPI) < 2.14w/L; Upper Half Ponderal Index (UPI) 2.14w/L. IUGR £ 2500g for all cases. Gestational age 37 wks and £ 43 wks for all cases. Values below the dotted line are adjusted for sex, gestational age and birthweight. Omission of birthweight as a covariate has little or no effect on the means or p values.

The findings in females were similar to those in males for most variables (Table 8). IUGR cases were smaller by 3.6 cm in height, 4.5 kg in weight, 1.1 kg in FM and 3.4 kg in FFM. IUGR cases had a tendency to have larger WHR. Strength was reduced in IUGR cases. The upper group tended to be more distinct from the middle group than was the case in males. A Z-score reduction of 1 standard deviation from birth to 3 years was associated with a loss of 1.8 cm in height, 1.6 kg in weight, 0.4 kg in F, and 1.2 kg in FFM. There were no significant associations between Dz and % F, WHR or strength.

Figure 2. Postnatal linear growth (Birth to 3 yrs by birthweight group.

Table 5. Body size, compositions and strength at follow-up*



Males
Birthweight Group

Females
Birthweight Group

IUGR (n=20)

Middle (n=61)

Upper (n=88)

Pooled S.D.

P

IUGR (n=19)

Middle (n=72)

Upper (n=71)

Pooled S.D.

P

Ht (Cm)

147.7

153.0

154.0

7.4

.004

145.7

148.8

149.3

5.1

.023

Wt (kg)

39.1

42.2

42.9

5.9

.041

38.8

42.8

44.4

6.1

.003

FFM (kg)

34.8

37.3

37.9

4.9

.046

29.9

32.9

33.8

3.8

.001

FM (kg)

4.3

4.9

5.0

1.3

.130

8.9

9.8

10.6

2.7

.038

% F

10.6

11.5

11.4

2.1

.205

22.6

22.6

23.4

3.4

.332

WHR

91.2

90.6

91.0

4.1

.771

90.8

88.6

89.3

5.1

.243

SR (kg)

25.0

27.5

26.7

6.0

.289

17.1

21.1

20.5

5.1

.011

SL (kg)

22.5

26.3

25.9

5.8

.038

17.5

20.2

19.7

4.8

.102

*Adjusted for age at follow-up and for gestational age.

Regressions were carried out to test whether effects differed by gender. The variable "sex" (males = 1, female = 0) was added to the models used in Tables 7 and 8 in addition to three interaction terms (Sex X DZ Sex X IUGR, Sex X Upper). The results (not shown) provide little evidence that effects vary by sex. Only one interaction term was statistically significant: the term Sex X DZ (p=.037) for the regression predicting height. Three other terms had p-values <.15: Sex X DZ in FFM regressions, Sex X DZ in SL regressions, and Sex X Upper for FM regressions. No interaction terms involving the variable IUGR were statistically significant.

Discussion

The use of the term 'catch-up' growth in this study may be viewed as inappropriate because all groups, whether IUGR or not, experienced postnatal growth retardation in length. Thus, relative to the NCHS/WHO reference population, no group can be said to have experienced catch-up growth. Rather, the term catch-up growth is used to contrast postnatal growth patterns of the IUGR group relative to the non-IUGR groups.

As others have found, IUGR infants have better postnatal linear growth (i.e. less negative DZ values) than either middle or upper birthweight groups. In other words, their position in the distribution relative to non-IUGR controls is improved at 3 years compared to what it was at birth. At birth, the length Z-score (sexes pooled) was-2.19 for the IUGR group and-0.72 for the upper group, a difference between groups of 1.47 S.D. units at birth. At 3 years of age the length Z-scores were -3.22 for the IUGR group and -2.23 for the upper group, a difference of 0.99 S.D. units.

Figure 3 Effect sizes [(IUGR-Upper)\S.D.] for body size, composition and strength.

Table 6. Body size, composition and strength at follow-up of IUGR sample by ponderal index group+


LPI

UPI

Pooled S.D.

P


n=20

n=19



Ht (cm)

147.46

145.88

6.89

.501

Wt (kg)

40.03

37.45

6.27

.231

FFM (kg)

33.25

31.36

4.92

.263

FM (kg)

6.78

6.09

1.82

.271

% F

16.82

15.78

2.92

.300

WHR

91.44

90.37

4.37

.475

SR (kg)

22.09

19.96

5.24

.238

SL (kg)

21.73

18.48

4.94

.061

+Lower Half Ponderal Index (LPI) < 2.14w/L; Upper Half Ponderal Index (UPI) 2.14w/L. IUGR £ 2500g for all cases. Values adjusted for sex, age at follow-up, gestational age and birthweight. Omission of birthweight as a covariate has little or no effect on the means or p values

Examination of absolute gains in length between birth and 3 years of age rather than on DZ leads to a different impression about catch-up growth. Values for Z36 in Table 3 are equivalent to mean lengths of 82.8, 85.4 and 86.3 cm respectively for IUGR, Middle and Upper groups in the case of males and to 83.4, 84.6 and 85.2 cm respectively in the case of females. From these values, absolute increments in length (between birth and 3 years) are 35.2 for IUGR, 35.7 for Middle and 35.4 cm for Upper groups for males and, correspondingly, 35.7, 35.6, and 35.1 cm for females. These absolute increments are very similar across groups. How can DZ values suggest catch-up growth in the IUGR group relative to non-IUGR groups but absolute increments suggest similar growth across groups? The explanation lies in the fact that the standard deviation of the reference curve increases with age with the slope of change being greater from top to bottom of the distribution. In other words, a greater increment is required to keep a tall child in its trajectory than is the case for a small child. Thus, readers should note that the conclusion that there is relative catch-up in growth in IUGR infants in this paper refers to distributional criteria (z-scores) and not absolute increments.

IUGR cases with lower PI values, contrary to expectations, experienced lower relative catch-up growth in length than IUGR cases with greater PI values (i.e. DZ was-1.25 for the LPI and-0.74 for the UPI). The opposite conclusion (i.e. LPI led to greater relative catch-up growth) would be obtained if weight is used instead as the indicator. For example, Table 6 shows that weight at adolescence is greater in LPI than in UPI groups (40.0 vs 37.5 kg); however, at birth, birthweights were similar (2.31 vs 2.36 kg respectively). As others have shown, long but thin IUGR infants (LPI) catch-up in weight for height quickly during infancy but length gains are similar to those of controls.

Thus, IUGR groups, despite less growth retardation post-natally, end up being shorter, lighter and weaker at adolescence compared to non-IUGR groups. The UPI group was the smallest, followed by the LPI, middle and upper birthweight groups. Also, effects of IUGR were somewhat magnified when controlling for postnatal growth retardation (i.e. adjusting for differential catch-up growth).

Table 7. Birth and postnatal growth influences on body size, composition and strength at follow-up: males

Variables

F-value

p-value

Adj R

Age ß se, p

GA ß se, p

IUGR ß se, p

Upper ß se, p

Dz ß se, p

Ht (cm)

35.26

0.0001

0.5048

4.15 0.36, p < .0001

-.58 0.38, p =.1288

-5.94 1.81, p =.0012

1.60 1.17, p =.1720

2.60 .54, p < .0001

Wt (kg)

42.05

0.0001

0.5514

3.88 0.30, p < .0001

-.53 0.31, p=.0838

-3.53 1.46, p =.0167

1.18 .95, p =.2162

1.85 43, p < .0001

FFM (kg)

42.18

0.0001

0.5522

3.29 0.25, p < .0001

-.46 .26, p =.0774

-2.85 1.23, p =.0213

.98 .80, p=.2189

1.47 .36, p < .0001

FM (kg)

20.39

0.0001

0.3673

0.59 0.07, p< .001

-0.08 .07, p=.2833

-.68 .33, p=0.0447

.19 .22, p=.3729

.38 .10, p=.0002

% F

4.07

0.0017

0.0841

0.34 0.11, p < .0021

-0.05 0.11, p =.6820

-1.05 0.54, p =.0551

.02 .35, p=.9555

.40 .16, p =.0145

WHR

06.99

0.0001

0.1512

-1.20 0.21, p < .0001

.25 .22, p=.2613

.64 1.06, p=.5443

.39 .68, p=.5674

-.05 .31, p=.8843

SR (kg)

25.33

0.0001

0.4200

3.34 0.31, p < .0001

-.50 .33, p=.1293

-2.65 1.56, p =.0910

-.60 1.01, p =.5529

.86 .46, p=.0655

SL (kg)

29.13

0.0001

0.4586

3.33 0.30, p < .0001

-.83 .31, p=.0088

-4.07 1.51, p=.0079

-.15 .96, p=.8763

.96 .44, p=.0325

+DZ is (Z36 - ZB) A positive coefficient, say 2.60 for height, means that catch-up growth equivalent to 1 standard deviation between birth and 3 years of age is associated with an increase of 2.6 cm in adolescent height. The same relationship can be interpreted to mean that growth retardation equivalent to 1 standard deviation between birth and 3 years of age is associated with a loss of 2.6 cm in adolescent height. The latter terminology is preferred in the text.

Table 8. Birth and postnatal growth influences on body size, composition and strength at follow-up: females

Variables

F-value

p-value

Adj R

Age ßse, p

GA ßse, p

IUGR ßse, p

Upper ßse, p

Dz ßse, p

Ht (cm)

11.92

0.0001

0.2533

1.02.25,
p <.0001

-.01 .28, p=.9597

-3.64 1.24, p=.0039

1.25 .82, p=.1293

1.80 .40, p=.0001

Wt (kg)

16.50

0.0001

0.3249

1.99 .32,
p <.0001

-.05 .34, p=.8893

-4.45 1.55, p=.0047

2.20 1.02,
p =.0327

1.58 .50,
p =.0018

FFM (kg)

17.87

0.0001

0.3438

1.20 .20,
p <.0001

-.02 .21, p =.9103

-3.38 .96,
p =.0006

1.30 .63,
p =.0413

1.15 .31,
p =.0003

FM (kg)

10.96

0.0001

0.2362

.79 .14,
p <.0001

-.02 .15, p=.8752

-1.07 .68, p=0.1198

.89 .45, p=.0483

.43 .22, p=.0514

% F

05.13

0.0002

0.1137

.80 0.18,
p <.0001

-.01.20, p=.9484

-.03 .89,
p =.9713

.87.59, p=.1422

.131.29,
p =.6399

WHR

01.10

0.3637

0.0030

.43 .27, p=.1165

-.04 .29, p=.9042

2.31 1.32, p=.0832

.50 .87, p=.5646

-.35 .42, p=.4166

SR (kg)

10.66

0.0001

0.2307

1.42 .27,
p <.0001

-.26 .29, p=.3746

-4.24 1.32, p=.0016

-.34 .87, p=.7004

.76 .42, p=.0755

SL (kg)

08.25

0.0001

0.1838

1.33 .25,
p <.0001

.07 .27, p=.7886

-2.79 1.24,
p =.0251

-.21 .81, p=.7952

.50 .40,
p =.2078

+Dz is (Z36 - ZB) A positive coefficient, say 2.60 for height, means that catch-up growth equivalent to 1 standard deviation between birth and 3 years of age is associated with an increase of 2.6 cm in adolescent height. The same relationship can be interpreted to mean that growth retardation equivalent to 1 standard deviation between birth and 3 years of age is associated with a loss of 2.6 cm in adolescent height. The latter terminology is preferred in the text.

Skeletal age in our population was similar in IUGR and non-IUGR groups as reported in the literature. Also, there were no statistically significant differences between IUGR and non-IUGR groups in %F or in the WHR. However, the sample, aged 14.5 years, may be too young to assess relationships with fatness indicators. In other analyses which focus on data at adulthood collected since 1991, a relationship between growth failure and greater fatness has been observed in females (Schroeder et al, in preparation).

In this study, differences between the IUGR and upper groups at adolescence were -6.3 cm and - 3.6 cm in height for males and females respectively and -3.8 and -5.6 kg in weight respectively. This is not very different from -5 cm and -5 kg, the differences found between IUGR and non-IUGR cases at 17-18 years of age in studies from developed countries. Thus, it does not appear that the effect of IUGR on later outcomes is magnified by poverty. However, the entire study population is stunted, reflecting an environment of poverty. In all respects, the findings concerning the consequences of IUGR for future body size and other characteristics were similar in males and females.

Maternal height is related to birthweight in the study sample. Pooling the sexes, maternal heights were 146.9 5.6 cm (n = 39) for the IUGR, 148.9 5.2 cm (n = 133) for the middle and 149.7 5.0 cm (n = 159) for the upper group. Differences in maternal height also appear with respect to the ponderal index groups: 148.1 5.7 cm (n = 20) for the LPI vs 145.6 5.7 cm (n = 19) for the UPI group. Interpretation of these data is difficult. Maternal height cannot be said to represent exclusively a woman's genetic potential in rural Guatemala because of the malnutrition and disease to which she was exposed as a young child. Indeed, ethnically similar populations, such as Mexican-American women measured in the first round of NHANES III (198-91), are over 10 cm taller than women in the study sample (unpublished analyses). More likely is that maternal height reflects, to unknown degrees, both a woman's history of malnutrition and her genetic endowment. Short maternal stature has an effect on the next generation through its influence on intrauterine growth and adjustment for maternal height attenuates the apparent effect of intrauterine growth retardation on adolescent size. In analyses similar to those presented in Table 5 but which included maternal height as an additional covariate, adjusted height at follow-up was 153.4 for the upper group, 153.2 for the middle and 149.5 for the IUGR group in the case of males and 149.2, 148.8 and 146.1 cm for the corresponding groups in females. Significance levels for the effect of birthweight group were reduced from .004 to .079 in males and from .023 to .044 in females. Thus, intergenerational influences are among the causal factors of IUGR in this population.

In conclusion, the findings presented illustrate that intrauterine growth retardation has important long-term consequences on body size, composition and strength. These are the first results to be reported from a developing country, a setting where such effects may have important functional implications. For example, lean body mass and strength are important determinants of work capacity and productivity, key assets in agricultural societies. Among women, short stature is a risk factor for delivery complications, including death, and maternal body size and composition predicts birth size and survival.

Acknowledgements - Supported by grants from NIH (HD22440 and HD29927) and from UNICEF. The assistance of Morgen Hughes in data management and programming is gratefully acknowledged.


Contents - Previous - Next