Main

Short stature in adulthood is associated with an increased risk of health problems, especially cardiovascular diseases (1–3). Early prediction of short adult stature is therefore of potential clinical interest should preventive measures or intervention strategies become available.

Several studies have been published on preterm birth and low birth weight in relation to long-term effects on adult height and weight. Most of the older studies investigated infants with a birth weight less than 2500 g (4–6) or less than 1500 g (7, 8) and followed them to adolescence or adulthood. In some more recent studies, birth weight and length have been corrected for gestational age (9–11), allowing infants to be grouped according to their birth weight and/or birth length for gestational age. Although these studies allowed relatively detailed analysis of the various groups, they were restricted to term infants and included a relatively low number of small infants within a limited geographical area, making generalizations difficult. To our knowledge, no nationwide population-based study has been published that has assessed whether well-defined birth data can be used to predict adult stature.

The aim of this study was to determine whether gestational age, birth weight, and birth length in a single-year birth cohort of Swedish males can be used to predict height, weight, and body mass index (BMI) at the age of 18–21 y.

Methods

The Swedish Birth Register, held by the National Board of Health and Welfare, was started in 1973 and contains data on more than 99% of all births in Sweden (12). The Register includes data on maternal demography, reproductive history, and complications during pregnancy, delivery, and the neonatal period. All births and deaths are validated every year against another central population register, using the mothers' and the infants' unique personal identification numbers that are assigned to each Swedish resident at birth. Information about mortality in birth cohorts is updated yearly through individual record linkage between the Birth Register and the Cause of Death Register, using the infants' personal identification number.

The Swedish Conscript Register includes information about Swedish males who are conscripted for military service. Most men (about 98%) are conscripted at 18–25 y of age. Those with known severe handicaps, congenital malformations, or chronic diseases are not conscripted. At conscription, all men have a thorough health examination, including height and weight measurements. The information is computerized, and the personal identification number of each subject ensures individual record linkage to other information sources.

In 1976, 50 343 male live births were recorded in the Birth Register, of whom 680 died before 18 y of age. In our analysis, multiple births (n= 860) and boys with congenital malformations (International Classification of Diseases, n= 2919) were excluded. To increase the homogeneity of the study population, we also excluded births to women who did not hold citizenship of a Nordic country (Sweden, Denmark, Finland, Norway, and Iceland;n= 1989). At least one of these exclusion criteria was present among 5915 boys. Of the remaining 44 428 boys eligible for inclusion, 39 901 (89.8% of the study cohort at birth) were conscripted between 1994 and the spring of 1997.

Information about birth weight, birth length, and gestational age was obtained from the Birth Register. Gestational age was estimated from the date of the last menstrual period. Small for gestational age was defined as a birth weight more than 2 SD below the mean birth weight for that gestational age, and large for gestational age was defined as a birth weight more than 2 SD above the mean, according to Swedish reference standards (13). Short for gestational age and tall for gestational age were defined similarly (13). Ponderal index at birth was calculated by dividing the birth weight in g by the cube of birth length in centimeters. A ponderal index below −2 SD was defined as low, and a ponderal index above +2 SD was defined as high.

The birth characteristics of the conscripts were compared with those of boys who had not been conscripted. Of those conscripted, 2.6% had a low birth weight for gestational age, 2.5% had a low birth length for gestational age, 4.5% were born preterm (<37 wk), and 2.7% had a high ponderal index. Of those who were not conscripted, 4% had a low birth weight for gestational age, 3.2% had a low birth length for gestational age, 5.7% were born preterm, and 2.6% had a high ponderal index.

At conscription, BMI was calculated by dividing the weight in kg by the square of the height in meters. Height SD score (SDS), body weight SDS, and BMI SDS at conscription were computed as (individual measurement minus cohort mean measurement)/cohort SD. Short stature at conscription was defined as a height below −2 SD (i.e.<166.3 cm), and individuals were classed as overweight if their BMI was above +2 SD, which corresponds to a weight of 91.9 kg for men with a height of 179.5 cm (mean height of the cohort).

Statistical methods.

The risk of the measured birth variables leading to short stature and excess weight was calculated by means of logistic regression. Odds ratios (ORs) are presented with 95% confidence intervals within parentheses. The standard statistical package SPSS for Windows, V8.0, was used in the statistical calculations.

RESULTS

At conscription, mean height (±SD) was 179.5 ± 6.6 cm, mean weight was 72.1 ± 11.2 kg, and mean BMI was 22.3 ± 3.1 kg/m2. Mean height, weight, and BMI at conscription increased with birth length, birth weight, and gestational age, although this tendency was more obvious for birth length and birth weight than for gestational age (Table 1). Birth length was positively correlated primarily with height at conscription, and birth weight was related primarily to weight at conscription. Specifically, there was a mean difference of almost 10 cm in height at conscription between boys who were short and those who were long at birth (<48 and >55 cm, respectively). Correspondingly, there was a mean difference of almost 10 kg in weight at conscription between boys who had a low birth weight and those who had a high birth weight (<2500 and >4500 g, respectively). When birth weight and birth length were analyzed using SD for gestational age, the differences in final height and weight were even more pronounced.

Table 1 Height, weight, and BMI according to birth weight, birth length, and gestational age of Swedish male conscripts born in 1976** Values are means ± SD. Missing values: birth length, n= 62; birth weight, n= 7; gestational age, n= 104; birth weight SDS, n= 112; birth length SDS, n= 165; ponderal index SDS, n= 65.
Full size table

As expected, a short birth length or a low birth weight was much more common among preterm births (Table 2). Preterm births were also more common among boys who were short or long for gestational age at birth as well as among boys with a low or high birth weight for gestational age. A low ponderal index was much more common in preterm compared with term births.

Table 2 Birth length and birth weight for gestational age SDS and ponderal index according to distribution of gestational age in Swedish male conscripts born in 1976** Missing values: birth length, n= 164; birth weight, n= 110; birth length SDS, n= 165; birth weight SDS, n= 112; ponderal index SDS, n= 167.
Full size table

The association between birth length and final height is illustrated in Figure 1. When birth length and mean height at conscription were expressed as SDS, there was a consistent increase in mean height at conscription with increased birth length. Figure 2 illustrates that mean height at conscription increases with increasing gestational age until term. However, the difference in mean final height is especially pronounced between boys born very preterm (<32 wk) and boys born at a later gestational age.

Figure 1
figure 1

Mean height at conscription in relation to birth length for gestational age SDS.

Full size image
Figure 2
figure 2

Mean height at conscription in relation to gestational age.

Full size image

In total, 1.8% of all the boys were considered short at conscription (i.e.≤166.3 cm). The prevalence of short stature at conscription increased primarily with decreasing birth length but also with decreasing birth weight for gestational age as well as with decreasing gestational age (Table 3). Multivariate analysis illustrates that the risk of short stature at conscription is influenced primarily by birth length for gestational age but also, to some extent, by birth weight for gestational age and gestational age. Compared with boys with a normal birth length for gestational age, boys with a short birth length faced a sixfold increase in the risk of being short at conscription, whereas a protective effect was found among boys with a high birth length. After adjusting for birth length for gestational age, boys with a low birth weight for gestational age were also at increased risk of short stature at conscription (OR = 1.8). Compared with boys born at term, those born very preterm (<32 wk) faced an almost threefold increased risk of being short as adults; however, the confidence intervals were wide, and the results should be interpreted with caution.

Table 3 Rates and odds ratios (95% CI) of short final stature (<166.3 cm) according to birth characteristics of Swedish male conscripts born in 1976* Estimates are adjusted for the other variables included in the table using multiple logistic regression.† Reference group.
Full size table

Of the 708 boys who were short at conscription, 113 were also short for gestational age at birth (sensitivity = 16.0%) (Table 4). A short birth length was associated with short stature at conscription in 113 of 968 boys (positive predictive value = 11.7%). Using low birth weight for gestational age as a predictor of short adult stature lowered both the sensitivity and the positive predictive value (11.7 and 8.3%, respectively). If both a short birth length for gestational age and a low birth weight for gestational age were used as predictors of short adult stature, the sensitivity was lower (8.2%); that is, only 8.2% of all infants being short at conscription had a low birth weight as well as a low birth length. Using this narrower definition of smallness at birth as a predictor only slightly increased the positive predictive value (12.3%).

Table 4 Validity of birth length as a predictor of adult short stature Sensitivity, (113/708) × 100 = 16%; specificity, (37509/38369) × 100 = 97.8%; positive predictive value, (113/968) × 100 = 11.7%.
Full size table

The risk of being overweight (BMI > +2 SD) at conscription was analyzed in a multivariate model that included ponderal index and gestational age at birth. Compared with boys with a normal ponderal index, boys with a high ponderal index were at increased risk of being overweight at conscription (OR = 1.8; 95% confidence interval = 1.4–2.2).

DISCUSSION

This study assessed the value of gestational age, birth length, and birth weight for predicting the risk of being short or overweight as a young man. The data clearly show that birth length is the most important of these variables for predicting adult height. The mean adult height of the boys who had a birth length below −2 SD for gestational age was 11.5 cm shorter than the adult height of the boys who had a birth length for gestational age above +2 SD. Compared with infants with a normal birth length, short infants faced a sixfold increased risk of being short in adulthood, whereas a substantial reduction in risk was obtained by being long for gestational age at birth. A short birth length for gestational age was recorded in 16% of the boys who were short at conscription, and among boys who were short at birth, almost 12% were also short at conscription. The independent influence of birth weight was less; after correcting for birth length, low birth weight for gestational age (< −2 SD) was associated with an 80% increase in the risk of short stature in adulthood. However, if information about birth length is not available, birth weight may be used as a predictor of short adult stature, as indicated by the high crude OR of 5.4. Combining information about birth length and weight does not add much information. Of the boys who were both short and light at birth, 12.3% were short at conscription, but only 8.2% of all short boys at conscription have this combination of birth characteristics. Very preterm birth (<32 wk) was associated with a more than twofold increase in the risk of being short in adulthood. A combination of very preterm birth with a low birth weight and length markedly increased the risk of being short in adulthood, but this combination occurred in only a small number of boys (n= 4) in the whole birth cohort.

The present study was large enough to permit powerful comparisons between several length and weight subgroups. There were about 1000 children in each of the two groups that were small for gestational age (more than −2 SD below the mean for length or weight) and the two groups that were large for gestational age (more than +2 SD above the mean for length or weight). Birth data (birth weight, birth length, and gestational age) were collected prospectively, which precludes recall bias. The birth cohort studied was relatively homogeneous, consisting of nonmalformed male singletons born in Sweden to Nordic mothers. About 10% of the birth cohort were not conscripted, however, and, among these, small-for-gestational-age infants and preterm infants were slightly over-represented. Thus, if anything, the increased risk of adult short stature associated with these birth characteristics might be somewhat underestimated in the present investigation.

It has been claimed that the measurement of length in newborn infants is less reliable and produces greater variability than the measurement of weight. However, any possible measurement error in birth length would have been nondifferential with regard to final height. The association between birth length and final height may therefore have been underestimated in the present investigation. Increasing the precision of length measurements of newborn infants, together with more precise measurements of gestational age (by general use of second trimester ultrasound screening), should make birth length for gestational age an even more useful predictor of final height.

Previous studies that have attempted to predict adult height from birth data for preterm or small-for-gestational-age children have had several shortcomings. In most, birth weight has generally been the only independent variable measured (4–8). Low birth weight, however, can be due to a short duration of pregnancy, reduced fetal growth velocity, or reduced fat deposition. Deviations from normal in terms of birth weight may thus have various etiologies that are likely to confound the results of any studies based only on birth weight.

We are aware of only two studies that have focused on the association between birth length and adult height (10, 11). In one of these studies, birth data were collected from a school cohort from a single city in Sweden (10). The study was restricted to 3650 babies born at term, of whom 198 were short or light for gestational age. Infants who were short for gestational age had a sevenfold increased risk of being short as adults, which is consistent with the findings of the present national study of a birth cohort. In the other study, growth among 213 infants (99 of whom were short) born small for gestational age (birth weight and/or birth length for gestational age <3rd percentile) in one district in France was compared with growth among infants who were appropriate for gestational age at birth (11). The difference in final height between those infants who were born small and those who were born appropriate for gestational age was less than 4 cm in both sexes, which is less than that in the present study and in the study by Karlberg and Albertsson-Wikland (10). This may be explained by the different definitions of “small for gestational age” used. In the study by Leger et al. (11), less than 50% of the infants were short for gestational age, and the majority were thus light but not short for gestational age. The study may also have been subject to selection bias, as 42% of the selected children did not participate in the study.

Although both the above studies have attempted to differentiate between the effects of birth length and birth weight on adult height, only infants born after a full-term pregnancy were studied. The present study is the first to demonstrate the relation between gestational age and final height in a large representative population. The duration of pregnancy seems to have only a limited impact on final height, except in very preterm births. The apparent lack of any previous data on the association between gestational age and final height is not surprising, as, in the past, there was a low survival rate for children born very preterm.

During this work, a study has been published that clearly shows that a high birth weight and ponderal index, but not birth length, are predictive of being overweight at 18 y of age (14). These results are in agreement with the findings of the present study.

Target (or mid-parental) height has also been used to predict final height. An attempt to improve the precision of predictions based on mid-parental height has recently been published by Luo and coworkers (15). Mid-parental height is at present recognized as the most important determinant of final height. However, evaluation of the predictive value of parental height was beyond the scope of the present study. Of the birth data that are readily available, we conclude that birth length is the most important variable for predicting adult stature. However, birth weight is closely associated with birth length and can be used for comparisons with earlier studies.