Sibling absence and body mass index: From adolescence to adulthood

Summary

Objectives:

To examine if sibling absence is associated with higher BMI and to identify potential lifestyle factors underlying this effect; to determine if sibling effects on BMI persist into adulthood.

Methods:

We used data from all five waves of the National Longitudinal Study of Adolescent to Adult Health to study the sibling factors and BMI of 3,563 participants who were in grades 7–12 at first wave (1994–95). These participants were measured again in the second wave (1996), the third wave (2001–2002), the fourth wave (2008–2009), and most recently, the fifth wave (2016–2018). We identified categories of siblings, comparing the BMI of those without siblings either to birth order or sibship size. BMI was calculated based on direct measurements of height and weight; underlying mechanisms were self-reported.

Results:

Participants without siblings had significantly higher BMI than those with siblings, across waves, regardless of sibship size. Those without siblings had consistently higher BMI than middle-born, but not last-born participants. Adolescents without siblings reported eating fast food more frequently and spending more screen time.

Conclusions:

Sibling absence is associated with higher BMI in adolescence and this difference persists into adulthood.

Reviewers of the literature^1–4 have concluded that children growing up without siblings are more likely to be overweight or obese. Yet there has been little research to understand the mechanisms or factors behind the differences observed and whether this difference persists from childhood to adulthood. Our goal was to use national, longitudinal data to determine if this no-siblings effect on obesity risk persists into adulthood and to identify lifestyle factors that may help to explain the related increase in risk.

We used data from all waves of the National Longitudinal Study of Adolescent to Adult Health (Add Health).⁵ Add Health study participants were included in our longitudinal sample (N = 3563, 61% female) if they had sibling status data available at the first and fifth wave. At the first wave, the participants were in grades 7–12 (1994–95), and 1 year later (1996) during the second wave, their height and weight were measured for the first time. Later, these measurements were made again as part of the third (2001–2002) and fourth waves (2008–2009), and finally, as part of the fifth wave (2016–2018).

The outcome of interest was defined as the mean body mass index (BMI) for each study wave. BMI was calculated in terms of weight (kg), divided by the square of height (m), and is a continuous representation of body mass, indicating degrees of underweight to overweight. Identifying participants without siblings was only possible based on responses to Add Health questions posed during the first and fifth waves. During the first wave, all participants were asked, ‘How many children have your biological parents had together?’ If participants indicated that their parents had solely one child, then they were labelled as first-wave single children. If participants indicated that their parents had two (or more) children, then they were asked to indicate their order of birth. During the fifth wave, all participants were asked, ‘Do you have any siblings, either living or deceased? Include any biologically related, adoptive, and step-brothers or sisters?’ If participants indicated that they had no siblings, then they were labelled as fifth-wave single children. No birth order information was collected. If participants indicated that they had siblings, then they were asked about the number of siblings.

We examined each of these two definitions of single-child status in separate analyses. When we used the first-wave designation as the single-child variable, categories of first born, middle born, and last born were created and the mean BMI of single children was compared to the mean BMI of birth-order groups within each wave.

In contrast, when we used the fifth-wave designation as the single-child variable, we compared the mean BMI of single children to those of their peers with siblings, categorized by sibship size. In these analyses, within each wave, the mean BMI of single children was compared to the mean BMI of participants who reported having one, two, or three or more siblings.

By considering these two definitions of single children in separate sets of analyses, we were able to determine whether birth order, sibship size, or both was associated with a difference in mean BMI between those with and without siblings. First-wave single children may have had adopted, half, or step-siblings at the first wave. In addition, they may have acquired siblings after the first wave. In contrast, fifth-wave single children, identified an average of 24 years after the first wave, can be regarded as never having a sibling. First-wave single children were a larger percentage of the sample (17.6%) when birth-order analyses were conducted; fifth-wave single children were a smaller percentage of this sample (4.3%) when sibship-size analyses were conducted.

BMI scores were adjusted by covariates representing individual and background factors known to influence BMI.⁶ The covariates were assessed at the first wave and included participants’ age, gender, race/ethnicity, Special Education status, family income, family structure, and financial distress. Within each wave, we used ordinary least square (OLS) models to compare the BMI of single children to the BMI of their peers with siblings, categorized in terms of birth order or sibship size. In each wave’s OLS model, we added the sampling weights, which were created by Add Health to generalize the results nationally.^7,8

Lifestyle factors known to be associated with obesity were examined, focusing on screen time and fast-food consumption. The screen time variable was a composite of reports of time spent watching TV and videos and playing video games (hours a week); longer periods spent on these activities are known to promote obesity.⁹ The fast-food variable represented responses about the frequency of eating at fast-food restaurants (times a week), a dietary practice associated with obesity.¹⁰ Because identical indicators of screen time and fast-food consumption were collected only at the second and third waves, we were able to create variables from these data for use in our analyses.

We placed the two lifestyle variables as dependent variables in OLS models identical to those used to evaluate sibling effects on BMI. Separate analyses were conducted comparing the means of respective lifestyle factors for first-wave single children to birth-order groups and comparing fifth-wave single children to sibship size groups. The same control variables were used in these analyses as in the BMI analyses.

Adjusted mean BMI for each sibling group across waves of Add Health is presented in the figure, and indicates that the mean BMI was higher at each subsequent wave for all groups. The comparison of the first-wave single children to the other birth order groups is presented in the top panel (Figure 1A) and the comparison of the fifth-wave single children to other sibship sizes is presented in the bottom panel (Figure 1B). Note that the mean BMI for all groups surpassed the cut-off score for obesity (BMI ≥ 30) by Wave V, except for middle borns.

FIGURE 1.

Adjusted mean body mass index (BMI) scores for each sibling group across waves of Add Health. (A) BMI comparison of the first-wave single children to the other birth order groups. (B) BMI comparison of the fifth-wave single children to other sibship sizes. The regression information (i.e., comparing different birth orders/sibship sizes to single children on BMI across waves) is shown in Table 1. Significant group differences in BMI between single children and other sibling groups are indicated in (A) and (B) by ●. Specifically, single children (the top line marked with ◆)had higher BMI than the respective sibling group marked with ●. In (B), single children had significantly higher BMI than all other sibship size groups across all waves

As shown in the top panel of Table 1, the results of the OLS analyses indicated that participants without siblings were found to have higher BMI than middle borns at each wave, higher than first borns at the fourth and fifth waves, but never significantly different from the last borns. Similarly, as shown on the bottom panel of Table 1, fifth-wave single children had significantly higher BMI than their peers with one, two, or three or more siblings, and this difference was observed at all waves.

TABLE 1.

Estimates of ordinary least square (OLS) coefficients examining sibling effects on body mass index (BMI) and lifestyle factors

	Sibling status and BMI					Sibling status and lifestyle factors
First-wave single children	BMI					Screen time		Fast food
Dependent variable	%	Wave II	Wave III	Wave IV	Wave V	Wave II	Wave III	Wave II	Wave III
Birth order (Ref: single-child)	17.6%
First born	33.0%	−0.07 [0.04]	−0.07 [0.04]	−0.14** [0.05]	−0.14** [0.05]	−1.42** [0.46]	−1.5** [0.53]	−0.28* [0.13]	−0.34 [0.16]
Middle born	18.9%	−0.11* [0.04]	−0.11* [0.05]	−0.21** [0.06]	−0.18** [0.06]	−1.90** [0.54]	−1.45* [0.58]	−0.17 [0.16]	−0.24 [0.18]
Last born	30.6%	−0.02 [0.04]	−0.04 [0.05]	−0.07 [0.06]	−0.09 [0.06]	−0.48 [0.49]	−0.11 [0.51]	−0.07 [0.13]	−0.06 [0.18]
	Sibling status and BMI					Sibling status and lifestyle factors
Fifth-wave single children	BMI					Screen time		Fast food
Dependent variable	%	Wave II	Wave III	Wave IV	Wave V	Wave II	Wave III	Wave II	Wave III
Sibship size (Ref: single-child)	4.3%
One sibling	27.4%	−0.17* [0.07]	−0.28** [0.10]	−0.18* [0.09]	−0.17* [0.08]	−0.21 [0.79]	0.40 [0.90]	−0.48* [0.20]	−0.31 [0.34]
Two siblings	28.2%	−0.20** [0.06]	−0.28** [0.09]	−0.25** [0.09]	−0.22** [0.08]	−0.39 [0.78]	0.28 [0.89]	−0.56* [0.22]	−0.06 [0.33]
Three or more siblings	40.2%	−0.16** [0.06]	−0.24** [0.08]	−0.21* [0.08]	−0.18* [0.08]	−0.31 [0.75]	0.74 [0.91]	−0.47* [0.22]	−0.34 [0.32]

Note: BMI = kg/m²; estimates are OLS coefficients with SEs in brackets. The covariates include age, gender, race/ethnicity, Special Education status, family income, family structure, and financial distress fromWave I.

^*p < 0.05.

^**p < 0.01.

In terms of analyses aimed at potential lifestyle factors driving the emergence of these sibling effects, the results, also presented in Table 1, indicated that first-wave single children reported spending more time with screens than first borns and middle borns, but not last borns, during both waves. They also reported eating more fast food than first borns but not middle or last borns at both waves. In contrast, no differences were found between fifth-wave single children and their peers with siblings in terms of screen time. However, fifth-wave single children reported eating fast food more often than each of the other sibship sizes, but only at the Wave II, not Wave III.

The consistency of the sibling absence effects stands out within the longitudinal context of higher and higher BMI across waves, representing adolescence through adulthood. Add Health sample members who reported never having had any sort of sibling consistently had higher BMI on average than did participants who had siblings. By the last wave, the average BMI had advanced over the line demarking obesity for all sibling groups, except for middle-born participants, who just barely stayed below this threshold. Yet the differences that emerged in adolescence between those who grew up with siblings and those who did not persist at all waves, spanning adolescence through adulthood.

The middle-born effect found here across all waves provides nuance to our understanding of the effects of the sibling absence on BMI. This middle-born effect suggests that middle borns add to their BMI more slowly than the other birth orders. Middle borns are distinct in that they were more likely than other birth orders to be living with both older and younger siblings. First borns may live as only children for years before the birth of a sibling. Similarly, last borns may live alone at home without siblings after their older siblings leave. Note that the mean BMI of last borns was never significantly different from that of first-wave single children.

The results of this study point to potential lifestyle factors contributing to the sibling absence effect on BMI. That is, adolescents who never had a sibling reported eating fast food more frequently than did any of the groups growing up with siblings, although this effect was not found during early adulthood. This finding suggests that the presence of a sibling within families is associated with less frequent fast-food consumption. In addition, the results indicated that participants who reported being the only child of their biological parents spent more time on screens than did first borns or middle borns, during adolescence and early adulthood.

Although the measurements of height and weight were taken only after the first wave and only for a subset of study participants, our results can be generalized nationally because sampling weights created by Add Health were adopted to compute population estimates.⁶ Nonetheless, the loss of data from some study participants can be regarded as a limitation.

These findings add to our understanding of the development of obesity over the life course. Since the number of single-child families is expected to increase in the United States,¹¹ public health campaigns that bring attention to the obesity risks associated with the sibling absence are warranted.