Social Science & Medicine 123 (2014) 141e148

Contents lists available at ScienceDirect

Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed

Birth order and physical fitness in early adulthood: Evidence from Swedish military conscription data € b, c, d Kieron Barclay a, *, Mikko Myrskyla a

Department of Sociology, Stockholm University, 106 91 Stockholm, Sweden Department of Social Policy, 2nd Floor, Old Building, London School of Economics and Political Science, Houghton Street, London WC2A 2AE, UK c Max Planck Institute for Demographic Research, Konrad Zuse Str. 1, 18057 Rostock, Germany d Population Research Unit, Department of Social Research, University of Helsinki, P.O. Box 18, 00014, Finland b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 20 August 2014 Received in revised form 28 October 2014 Accepted 4 November 2014 Available online 5 November 2014

Physical fitness at young adult ages is an important determinant of physical health, cognitive ability, and mortality. However, few studies have addressed the relationship between early life conditions and physical fitness in adulthood. An important potential factor influencing physical fitness is birth order, which prior studies associate with several early- and later-life outcomes such as height and mortality. This is the first study to analyse the association between birth order and physical fitness in late adolescence. We use military conscription data on 218,873 Swedish males born between 1965 and 1977. Physical fitness is measured by a test of maximal working capacity, a measure of cardiovascular fitness closely related to V02max. We use linear regression with sibling fixed effects, meaning a within-family comparison, to eliminate the confounding influence of unobserved factors that vary between siblings. To understand the mechanism we further analyse whether the association between birth order and physical fitness varies by sibship size, parental socioeconomic status, birth cohort or length of the birth interval. We find a strong, negative and monotonic relationship between birth order and physical fitness. For example, third-born children have a maximal working capacity approximately 0.1 (p < 0.000) standard deviations lower than first-born children. The association exists both in small (3 or less children) and large families (4 or more children), in high and low socioeconomic status families, and amongst cohorts born in the 1960s and the 1970s. While in the whole population the birth order effect does not depend on the length of the birth intervals, in two-child families a longer birth interval strengthens the advantage of the first-born. Our results illustrate the importance of birth order for physical fitness, and suggest that the first-born advantage already arises in late adolescence. © 2014 Elsevier Ltd. All rights reserved.

Keywords: Birth order Physical fitness Cardiovascular health Early adulthood Swedish administrative registers Military conscription data Fixed effects Sibling comparison

1. Introduction Physical fitness is an important dimension of physical health, consisting of a number of health-related components, including cardio-respiratory endurance, muscular endurance, muscular strength, body composition, and flexibility (Caspersen et al., 1985). Higher levels of physical fitness have been shown to be associated with higher self-rated health (Shirom et al., 2008), as well as a lower risk of suffering from all-cause and cardiovascular-related mortality at all ages, even amongst otherwise healthy individuals (Blair et al., 1995). It is known that exercise, alcohol consumption, and smoking are important predictors of physical fitness (Marti

* Corresponding author. E-mail addresses: [email protected] (K. Barclay), m.myrskyla@lse. €). ac.uk (M. Myrskyla http://dx.doi.org/10.1016/j.socscimed.2014.11.007 0277-9536/© 2014 Elsevier Ltd. All rights reserved.

et al., 1988). Relatively little, however, is known about the relationship between physical fitness and early-life conditions. Previous studies have shown that early life adversity has negative consequences for cognitive development, educational attainment, and health in adulthood (Conley and Bennett, 2000). More particularly, studies have also shown that adverse conditions early in life can impact cardiovascular health (Alastalo et al., 2009), which would impede the ability of individuals to develop physical fitness. An important but thus far unanalysed potential factor influencing physical fitness is birth order. Several factors indicate that being a later born child means occupying a disadvantaged position within the household. Previous research indicates that parents spend less time caring for later born children (Price, 2008), and in Sweden there is a negative relationship between birth order and € m and Duvander, 2002). time spent on parental leave (Sundstro Recent research also shows that mothers are less likely to seek pre-

142

€ / Social Science & Medicine 123 (2014) 141e148 K. Barclay, M. Myrskyla

natal care for later pregnancies, and are also less likely to breast feed later born children (Buckles and Kolka, 2014). There is a long tradition of studying the relationship between birth order and outcomes ranging from personality to educational attainment and intelligence (Ernst and Angst, 1983; Black et al., 2005; Bjerkedal et al., 2007). However, relatively few studies have addressed the relationship between birth order and adult health. Studies on the relationship between birth order and suicide show that later borns have a higher risk of suicide (Bjørngaard et al., 2013; Rostila et al., 2014). All-cause mortality has also been reported to increase with birth order (Modin, 2002; Barclay and Kolk, 2013), though Modin (2002) did not use a within-family comparison, and one study using fixed effects does not support that conclusion (Rostila et al., 2014). Although studies addressing the relationship between birth order and mortality are generally identifying long-term effects on adult health, as most mortality occurs at advanced ages, the effect on health in late adolescence is much less known. Later born men in Sweden have been shown to be shorter than € et al., 2013). The inverse association their older siblings (Myrskyla between birth order and height suggests that at least part of the adverse effects of birth order on later outcomes arise relatively early in life. We therefore hypothesize that birth order may have a similar association with physical fitness at young adult ages to that which has been found for the relationship between birth order and height: that is, later-born siblings will have worse physical fitness at young adult ages than their earlier-born siblings. This study will investigate the relationship between birth order and physical fitness in late adolescence by using unprecedentedly large and high quality Swedish military conscription data on 218,873 Swedish males born between 1965 and 1977. We measure physical fitness using a stationary bicycle test to identify maximal working capacity (MWC), a measure of cardio-respiratory fitness closely related to other measures of maximal aerobic capacity, including V02max (Patton et al., 1982). Measures of cardio-respiratory fitness reflect the ability of the heart and lungs to provide oxygen to working muscles, which in turn reflects the interaction of multiple parameters, including heart rate, cardiac output, and maximal oxygen consumption. Our study has two unique contributions both to the birth order literature and to the literature on physical fitness. First, this is the first study to analyse the association between birth order and physical fitness in late adolescence. Lower physical fitness at young adult ages is associated with continued poor health into the future, as attitudes and habits become ingrained over time (Biddle et al., 2010). Furthermore, lower physical fitness is associated with increased mortality risk even in early adulthood (Blair et al., 1995; Myers et al., 2002). Understanding the potential impact of birth order on physical fitness is important both in terms of shedding new light on the mechanisms through which birth order influences later life outcomes, and in terms of understanding where in the life course the potentially adverse effects of being of high birth order emerge. Second, while prior work has studied the relationship between birth order and cardiovascular system related mortality (Barclay and Kolk, 2013; Rostila et al., 2014), no prior work has considered birth order as a potentially important determinant of cardiovascular fitness. 2. Potential mechanisms linking birth order to later-life outcomes Prior research has proposed both physiological and social explanations for the finding that adult outcomes are patterned by birth order. One physiological theory that has been offered for why later born children have worse outcomes is depletion of maternal nutritional reserves during pregnancy, which is particularly

relevant when the birth interval is short (Gunawardana et al., 2011). However, recent research using data on families where an infant has died (Kristensen and Bjerkedal, 2007), and fully adopted sibling groups (Barclay, 2014), suggest that it is social order within the family rather than biological birth order that matters. The results by € et al. (2013) also suggest that the birth order effect may be Myrskyla attributable to post-natal factors, as the inverse association between adult height and birth order was robust to controls for birth weight and length. Research on military conscripts in Switzerland shows that both alcohol consumption and cigarette smoking have a detrimental effect on cardiovascular fitness even in early adulthood (Marti et al., 1988). Empirical research has also shown that later born siblings are more likely to begin developmentally inappropriate activities, such as a drinking and smoking, at a younger age than older siblings (Blane and Barry, 1973; Harakeha et al., 2007). This is likely to be due to both a combination of decreasing parental control as the size of sibling group increases, as well social learning based on the behaviour of older siblings, who may also facilitate access to alcohol or cigarettes for their younger siblings (Bard and Rodgers, 2003). The initiation of smoking, in particular, at a younger age is associated with a higher probability of continued addiction over the life course (Chen and Millar, 1998; Khuder et al., 1999). Given that previous research has shown that health behaviours such as cigarette and alcohol consumption are key predictors of physical fitness (Marti et al., 1988), these behaviours could be important mediators for the relationship between birth order and physical fitness in early adulthood. An alternative theory that may be important for understanding the relationship between birth order and physical health in late adolescence is the hygiene hypothesis (Strachan, 1989). The hygiene hypothesis describes how atopic diseases, meaning hyperallergenic diseases such as hay fever, are less common in larger families (Strachan, 1989). Research suggests that infection early in life is protective against the development of atopic diseases (Strachan, 2000). Households with more children are likely to have higher rates of infections, largely due to sibling interaction. In application to birth order, the theory suggests that earlier born children may grow up in a cleaner environment, and therefore be more likely to develop atopic diseases, than later born children (Butland et al., 1997). If birth order is associated with the development of diseases that affect the respiratory system, such as hay fever, it would not be surprising to find that higher birth order children perform better on this test of physical fitness than their earlier born siblings. It should be noted, however, that a relationship between birth order and another atopic disease, allergic asthma, has not been consistently documented (Strachan, 2000). This interpretation of the hygiene hypothesis predicts a positive relationship between birth order and physical health. The potential importance of the hygiene hypothesis is, however, complicated by an alternative interpretation of the theory that predicts that later born children should in fact do worse than earlier borns. This interpretation is based upon the assumption that children bring communicable diseases into the household, and as the number of children increases, this increases the disease load within the household. Later born children may be more susceptible to these diseases, as they are younger and more frail, and this might be associated with lower physical fitness in late adolescence. Another theory that can applied to understand the relationship between birth order and physical fitness is the resource dilution hypothesis (Blake, 1981). The resource dilution hypothesis describes how early born children are cumulatively advantaged over their later born siblings because they have exclusive, or greater, access to parental resources. Although later born children may have greater access to parental resources later in life as their older siblings leave

€ / Social Science & Medicine 123 (2014) 141e148 K. Barclay, M. Myrskyla

the parental home, it has been suggested that access to more resources early in life is particularly important for later development (Campbell and Ramey,1994). Research also shows that many parents in Sweden continue to support their children even after they move € rnberg and Latta, 2007). It may be that out of the parental home (Bjo children who have parents with more resources may be ferried to more extracurricular activities, such as sports training, be fed more nutritious food by their parents, and generally receive a greater amount of attention that could transmit into knowledge about a healthy lifestyle and greater physical fitness. To test the resource dilution hypothesis, we conduct analyses to test whether the association between birth order and physical health varies by sibship size, parental socioeconomic status, and the length of the birth interval, as a larger family size, lower parental socioeconomic status, and shorter birth intervals would all imply a disadvantage in terms of access to the pool of parental resources.

143

relative to their brothers and sisters due to the multigenerational register described above, and we are not restricted to calculating birth order from the individuals included in the military conscription data. Although we construct the variable for birth order using information from the entire population, we restrict the analysis to individuals born 1965 to 1977. This means that some individuals must also be excluded from the within-family comparison analyses even though they were not only children. The reason for this is that if the second last born child was born before 1965, and the last born child was born after 1965, then this child would come from a sibling group with more than 2 children, but would be the only one included in the analysis, for example. This is what the term ‘cohort cut’ in Table 1 refers to. We have also estimated models based upon fully observed sibling groups, meaning sibling groups where all individuals were born between 1965 and 1977, and all were males. The results based upon this study population are shown in the supplementary online material.

3. Data and methods 3.2. Outcome variable 3.1. Data In this study we used Swedish administrative register data, and our population group consists of men born between 1965 and 1977. We are able to construct a variable for birth order by using the Swedish multigenerational register, which provides linking information for the mother and father of any given individual, and therefore also enables us to identify their siblings. For this study we classify siblings as those who share the same biological mother and father. We focus on families with between 2 and 5 children, as there are relatively few families with more than five children in Sweden, The registers contain information on 1,426,689 individuals born in Sweden in these years, and the steps by which we reach the final study population are listed in Table 1. As we will describe below, the main statistical analyses that we use for this study utilise sibling fixed effects, meaning a within-family comparison. This means that we also exclude only children as there is no variance in the outcome in a one-child sibling group. We also exclude sibling groups with multiple births as the meaning of birth order is less clear in these sibling groups. A substantial proportion of children in Sweden experience family complexity in one form or another as they grow up. Amongst those borns in the 1960s, 23% of individuals have at least one halfsibling, and for those born in the 1970s the corresponding figure is 25% (Thomson, 2014). We have estimated additional models based upon the population who have no half-siblings. The results based upon this study population are available in the supplementary online material. In Sweden reporting to the military conscription tests was only required of males, and thus we cannot study women in this analysis as there is no information for them on the outcome variable. However, we are able to calculate the birth order of individuals Table 1 Sample exclusion process. Exclusion criteria

N

N Excluded

All born 1965e1977 No missing ID for parents All siblings born in Sweden No multiple births No only children Sibling group size 40

300.8 301.1 301.9 298.3 293.7 289.9 301.9 294.3 296.0 303.8 313.6 295.8 303.3 299.6 302.0 296.9 302.4 299.4 281.6 290.6 299.7 304.3 302.8 299.0 293.0

51.1 50.9 51.1 51.5 51.3 51.4 51.0 51.4 50.6 51.2 50.1 50.6 51.1 51.1 51.0 50.8 50.9 53.9 53.7 50.0 50.8 51.2 51.1 51.2 54.2

218,873 88,486 92,961 29,982 6262 1182 186,943 31,930 84,422 134,451 25,056 106,428 87,389 108,078 110,795 53,885 155,406 8345 1237 21,892 82,713 78,327 28,883 6297 761

example, for first born mean MWC is 301.1, for second-born 301.9 and for third-borns 298.3. For the fifth-born children, MWC is 289.9, or 21% of a standard deviation below the mean and 22% of a standard deviation below the first-born. It is illustrative to compare this pattern of MWC by birth order to how MWC varies by age. Previous research on MWC in Sweden has shown that the mean value for men aged 20e29 is 303W (Wohlfart and Farazdaghi, 2003), which is very similar to that seen for first borns in our data. However, MWC is a marker for overall fitness and declines rapidly with age. Wohlfart and Farazdaghi (2003) report that the mean score amongst 30e39 year old men was 288W; this corresponds closely to the mean MWC observed for fifth borns in our data (Wohlfart and Farazdaghi, 2003). Thus in terms of physical fitness, the fifth-borns perform at a level consistent with being ten years older, relative to the first-borns. Looking at set size we can see that MWC decreases with increasing set size. This is expected given the pattern by birth order, as larger set sizes include more higher birth order children that on average have lower MWC. It is important to note, however, that our fixed effect analyses fully adjust for set size (through the family fixed effects), and thus set size cannot confound our results. For parental socioeconomic status, birth density, and birth year the gradients are largely in the expected direction. Compared to lower SES families, those that have higher parental SES have on average higher MWC (313.6 vs. 295.8). Higher birth density is associated with lower physical health: MWC is 299.6 in the sibling sets that belong to the top 50% of birth density distribution vs. 302.0 in the bottom 50%. For the later-born cohorts (1970e77 vs. 1965e69) MWC is higher, 303.8 vs. 296.0, potentially reflecting improvements in public health conditions over time. Conscription age and age of the mother when the child was born have an inverted U-shape association with MWC. These summary statistics show that 18 year-olds have the highest mean, and 20 year-olds the lowest. The variable for mother's age at the time of birth shows that the highest mean is found amongst those whose mothers were aged 26e30, and the mean decreases for children born to mothers both younger and older than that.

€ / Social Science & Medicine 123 (2014) 141e148 K. Barclay, M. Myrskyla

145

4.2. Multivariate analyses The results from the full multivariate analyses, pooling all sibling set sizes from 2 to 5, can be seen in Table 3. Table 3 includes the results from the fixed effects linear regression, the random effects linear regression, and the standard OLS regression. All three models adjust for age at conscription, birth year, and the mother's age at the time of birth. The random effects and standard OLS regressions also adjust for birth density in the sibling group, socioeconomic status, and the size of the sibling group. In the fixed effects models it can be seen that MWC decreases by birth order, and relative to first borns, the difference in MWC for all later born children is statistically significant. The results in the standard OLS model and random effects model show a sharper decline in MWC with birth order than in the fixed effects models. This is likely to be attributable to residual confounding. Nevertheless, the results for birth order are qualitatively similar across the three different models, as each model documents that MWC declines with birth order. The fixed effects model cannot estimate the relative importance of the birth order effects in comparison to factors that vary across families so we use the OLS and random effects models for such comparisons. The random effects and OLS models show that in sibling sets of size 5, MWC is 6W lower than in sibling sets of 2. This difference is the same that we estimate to be the effect of being fifth-born relative to first born (6.2W) in the fixed effect model. Further, the random effects and OLS models document that individuals from a high SES background have an MWC 14e15W higher than that of individuals from a low SES background. Thus, the difference in MWC between a first and fifth born child in the fixed effect model results is a little less than half the difference in MWC between individuals from a high and a low SES background. Fig. 1 shows the results from analyses stratifying by the size of the sibling group using fixed effects. It can be seen that the negative relationship between birth order and MWC is very similar in families with 2e3 children, and 4e5 children, with all later born

Fig. 1. Within-family analyses: birth order and physical capacity by the size of the sibling group. Error bars are 95% confidence intervals.

children performing worse than the first born. In sibling groups with 2e3 children third borns have an MWC approximately 9% of a standard deviation worse than first borns, while fifth borns in sibling groups with 4e5 children have an MWC approximately 12% of a standard deviation worse than first borns. Fig. 2 shows the results from models where we split the pooled sample into two groups using fixed effects: those born from 1965 to 1969, and those born from 1970 to 1977. As can be seen, the negative birth order pattern is similar in both cohort groups. Apart from fifth born children born in the 1960s, all later born children have a statistically significantly lower MWC than first borns. Fifth born children born in the 1970s have an MWC approximately 14% of a standard deviation lower than first borns. The results from a fixed effects model testing how the relationship between birth order and MWC varies by socioeconomic status can be seen in Fig. 3. As can be seen, there is a negative and statistically significant relationship between birth order and MWC in low socioeconomic status sibling

Table 3 Results from multivariate analysis using fixed effects linear regression, random effects linear regression, and OLS: maximal working capacity in watts by birth order, Swedish men born 1965e1977. Variable

Category

Fixed effects b

Birth order

Age in year of conscription

Mother's age at time of birth

Birth interval density Sibling group size

SES

r N

1 2 3 4 5 17 18 19 20 40 Low High 2 3 4 5 Low High Other

0.00 1.89 4.63 5.03 6.21 2.42 0.00 0.52 5.78 0.84 0.14 0.00 0.04 0.29 2.57

0.51 218,873

se

Random effects 95% CI

b

0.41 0.81 1.34 2.30 0.31

2.69, 6.22, 7.66, 10.71, 3.03,

1.09 3.04 2.39 1.71 1.80

0.77 2.14 0.90 0.49

0.99, 2.02 9.97, 1.60 2.62, 0.93 1.10, 0.81

0.59 1.27 2.95

1.20, 1.11 2.21, 2.79 8.35, 3.22

0.00 3.93 8.14 9.37 10.73 2.45 0.00 2.55 17.97 9.76 3.57 0.00 0.40 2.57 8.21 0.00 0.15 0.00 1.10 1.90 5.54 0.00 14.56 6.51 0.30 218,873

se

OLS 95% CI

0.24 0.42 0.79 1.67 0.25

4.40, 3.45 8.96, 7.31 10.91, 7.82 14.01, 7.46 2.94, 1.96

0.61 1.57 0.44 0.27

3.74, 1.36 21.06, 14.89 10.62, 8.91 4.10, 3.05

0.36 0.70 2.05

1.10, 0.30 3.95, 1.19 12.23, 4.19

0.27

0.68, 0.38

0.30 0.48 0.84

0.51, 1.70 2.83, 0.96 7.17, 3.90

0.41 0.27

13.76, 15.37 5.99, 7.03

b 0.00 3.66 7.80 9.05 10.39 2.41 0.00 2.05 15.89 8.16 3.01 0.00 0.02 1.46 6.65 0.00 0.12 0.00 1.02 2.12 5.85 0.00 14.69 6.57 218,873

se

95% CI

0.24 0.41 0.76 1.59 0.24

4.12, 3.19 8.60, 7.00 10.54, 7.55 13.51, 7.27 2.88, 1.94

0.58 1.52 0.42 0.26

3.18, 0.91 18.87, 12.92 8.98, 7.33 3.52, 2.51

0.34 0.68 1.95

0.65, 0.70 2.79, 0.13 10.48, 2.82

0.27

0.65, 0.40

0.30 0.47 0.82

0.44, 1.61 3.04, 1.20 7.46, 4.25

0.41 0.26

13.89, 15.49 6.05, 7.09

146

€ / Social Science & Medicine 123 (2014) 141e148 K. Barclay, M. Myrskyla

Fig. 2. Within-family analyses: birth order and physical capacity by cohort group. Error bars are 95% confidence intervals.

Fig. 4. Within-family analyses: birth order and physical capacity by birth interval density. Error bars are 95% confidence intervals.

groups. In these low socioeconomic status groups second borns have an MWC approximately 4% of a standard deviation lower than first borns, while fifth borns have an MWC approximately 15% lower than first borns. The pattern in high socioeconomic status groups is less clear; later borns have a lower estimated MWC than first borns, except for fifth borns, but the difference is only statistically significant for third borns. The negative birth order pattern can also be seen in families that were not classified as being of a high or a lower socioeconomic status. The results in Fig. 4 show the results from models stratifying by the birth density of the entire sibling group, using fixed effects. Birth density is calculated as the number of months between the first born and the last born child in the sibling group, divided by the total number of children. The results in Fig. 4 show the relationship between birth order and MWC for individuals in sibling groups above and below the 50th percentile of the density distribution. As can be seen, there is a negative relationship between birth order and MWC in both high and low density families, and there is a statistically significantly lower MWC score for later borns compared to first borns, except for the case of fifth borns in high density families, but no statistically significant difference amongst siblings from the second born to the last. To investigate the importance of birth intervals further, we stratified our analyses by the birth interval in months in sibling groups with only two children, meaning two males in these results. This model also used

fixed effects. These results can be seen in Fig. 5. Please note that the scale on this graph is different from the previous four figures. Each data point on this graph shows the difference in MWC for the second born relative to the first born, for the indicated interval length in months. Fig. 5 indicates that birth intervals may play a role in the relationship between birth order and MWC. While second borns always have a lower MWC score than first borns, the longer the birth interval, the greater is the difference. As mentioned in the Data section, we also estimated additional models based upon the population that have no half-siblings, and a population where all individuals in the sibling group were male and born between 1965 and 1977. Those results, available in the online supplementary materials, are fully consistent with the main results presented here.

Fig. 3. Within-family analyses: birth order and physical capacity by family socioeconomic status. Error bars are 95% confidence intervals.

5. Discussion The results from this study are generally consistent with other research on birth order, showing that relative to first borns, later born children perform worse. As we used a within-family comparison, these results likely show the causal relationship between birth order and maximal working capacity (MWC). These results, looking at MWC, a measure of cardiovascular fitness, show that the relationship between birth order and physical fitness already begins to emerge, at least amongst men, in early adulthood. Given

Fig. 5. Within-family analyses: relationship between birth interval in months and physical capacity in two-child families. Data points indicate physical capacity of second borns relative to first borns. Error bars are 95% confidence intervals.

€ / Social Science & Medicine 123 (2014) 141e148 K. Barclay, M. Myrskyla

that some research indicates a relationship between birth order and mortality (Modin, 2002; Barclay and Kolk, 2013), and physical fitness is also linked to mortality risk at all ages (Myers et al., 2002), physical fitness may be a mediating factor between birth order and mortality. As discussed below, the documented association between birth order and physical fitness is substantially important and is perhaps universal and persistent as it is observed in large and small families as well as in high- and low socioeconomic status families, and shows no signs of attenuation over the 13-year window analysed in this study. First, the magnitude of the birth order effects is sizeable. As our outcome was measured in watts, it is difficult to intuitively grasp what a given score communicates about individual fitness. However, studies have shown that MWC declines monotonically with age (Wohlfart and Farazdaghi, 2003), and thus how the measure of MWC of a given individual corresponds to the mean performance in different age groups gives an idea about how this measure corresponds to physical health. Wohlfart and Farazdaghi (2003) report that the mean score amongst 20e29 year old men in Sweden was 303W and among those aged 30e39, 288W. Our descriptive results found that the first-born scores on average close to the 20e29 year age group (301W), but the fifth-born perform at the level of 30e39 year old men (288W). The descriptive results are likely to overestimate the true birth order effect, and our more conservative main results based on comparison of siblings suggested that the third-borns have 5 W lower and fifth-borns 6 W lower maximal working capacity than the first-borns. These are still quantitatively important effects, as they are a third or more of the difference between 20e29 and 30e39 year old men. We also compared the magnitude of the birth order effects to other potentially important determinants of physical fitness that vary across families, such as sibling group size or parental socioeconomic status. In random effects and ordinary least squares regressions the difference in maximal working capacity between sibling sets of size 5 and size 2 was 6 W, or of the same magnitude that we estimate the birth order effect to be for birth orders 3 and higher. The birth order effect for birth rank five was little less than half the difference between individuals from a high and a low socioeconomic background. Thus the birth-order effects are not small when compared to between-family determinants of maximal working capacity. Second, we observed that the birth order effect on physical fitness persists in large- and small families and in families with high- and low socioeconomic status. These results are important when considering the potential mechanisms. The introductory section of this paper discussed several explanatory theories for the relationship between birth order and physical fitness. The results presented in this study are consistent with the second, but not the first interpretation of the hygiene hypothesis. The second interpretation argues that an increase in the number of siblings in the household increases the disease load within the household, and that this impacts later born children to a greater extent as they are younger and frailer. The resource dilution hypothesis also predicts that later born children should fare less well as earlier born children have a cumulative advantage in terms of access to parental resources. Further support for the resource dilution hypothesis is that later borns have significantly lower physical fitness in low socioeconomic status families, while there was no clear or statistically significant pattern in high socioeconomic status families. We found no evidence that the birth order effect varies by birth cohort. This was surprising as previous research on Swedish men has found that the association between birth order and adult height € et al., 2013). Our test for is weaker in more recent cohorts (Myrskyla cohort differences was based on comparing those born in 1965e1969 to those born 1970e1977. It is possible that we failed to

147

observe cohort differences because of the relatively narrow cohort window, or alternatively because the mechanism linking birth order to physical fitness is different than that for height. For example, if early adoption of smoking or alcohol consumption is part of the mechanism, this could influence physical fitness more than adult height. Previous studies have shown that younger siblings tend to begin smoking and drinking at younger age than older siblings, both because they may model the behaviours of older siblings, and those older siblings can facilitate access to alcohol and cigarettes (Blane and Barry, 1973; Bard and Rodgers, 2003; Harakeha et al., 2007), Given that both these behaviours have a negative impact on physical fitness (Marti et al., 1988), it may be that it is smoking and drinking that mediate the relationship between birth order and physical fitness in early adulthood. Unfortunately we do not have data that would allow us to test this potential mechanism explicitly. The results for the difference in physical fitness by birth interval in sibling groups with two children showed that the longer the birth interval, the greater the difference was between the first and second born child. This might be due to the first child having a greater period of time being the sole focus and sole recipient of the parents' resources, thereby benefitting the physical fitness of the first-born. The longest interval we examined was a ten-year difference between the first and last borns. While raising a ten year old is possibly less demanding of time than raising a one or two year old, they are still fully dependent upon the parents and would be a significant competitor for parental resources against the newborn. Although such a lengthy birth interval would imply that the second born would have a long period of exclusive access to parental resources after the first child leaves the family home in early adulthood, it is very possible that access to a greater quantity of resources has a stronger beneficial effect on development soon after birth in comparison to the teenage years. Along with the results stratifying by parental socioeconomic status, this pattern suggests that the resource dilution hypothesis may explain part of the relationship between birth order and health in early adolescence. However, the support for the resource dilution hypothesis is not entirely conclusive, as the analyses where we investigate the importance of birth density do not show that this is a clear mediating factor between the relationship between birth order and physical fitness. There is also the possibility that the longest birth intervals reflect selection processes where the second birth may not have been planned, or where the parents have lower physical health, as indicated by fecundity. This might also partially account for why second borns could fare worse in these families. Although this study had access to high quality full population data, there do remain some limitations. For one, we did not have access to information from the medical birth register, meaning that we did not have information on birth weight. A greater birth weight has been found to be associated with a range of positive outcomes later in life (Conley and Bennett, 2000). However, since birth weight increases with parity (Magnus et al., 1985), it is likely that not adjusting for this variable means that our estimates actually underestimate the negative gradient between birth order and physical fitness. Another limitation of this study is that our results apply only to men. Because the Swedish state did not require women to perform mandatory military service, they did not have to attend the conscription tests from which we derived the data for this study. Nevertheless, our calculation of birth order is based upon the entire sibling group, including sisters, and is not based only on male siblings. Although this does not introduce bias into the results, it does limit the external validity of these findings. We have also estimated models that test an interaction between the gender composition of the sibling group and birth order, which do not suggest that the birth order effect differs by the gender

148

€ / Social Science & Medicine 123 (2014) 141e148 K. Barclay, M. Myrskyla

composition. The results of this study show not only that birth order has an impact on fitness, but also that this impact is visible at a relatively young age, amongst healthy men, in a highly developed Western society with a comprehensive welfare state. Acknowledgments € was supported by a European Research Council Mikko Myrskyla grant 2013-StG-336475. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.socscimed.2014.11.007. References € nen, K., Pesonen, A.K., Osmond, C., Barker, D.J., Kajantie, E., Alastalo, H., R€ aikko Eriksson, J.G., 2009. Cardiovascular health of Finnish war evacuees 60 years later. Ann. Med. 41 (1), 66e72. Barclay, K.J., 2014. Birth order and educational attainment: evidence from fully adopted sibling groups. Intelligence. http://dx.doi.org/10.1016/j.intell.20 14.10.009 (in press). Barclay, K.J., Kolk, M., 2013. Birth order and mortality: a population-based cohort study. Stock. Res. Rep. Demogr. 14, 1e30. Bard, D.E., Rodgers, J.L., 2003. Sibling influence on smoking behavior: a withinfamily look at explanations for a birth-order effect. J. Appl. Soc. Psychol. 33 (9), 1773e1795. Biddle, S.J., Pearson, N., Ross, G.M., Braithwaite, R., 2010. Tracking of sedentary behaviours of young people: a systematic review. Prev. Med. 51 (5), 345e351. Bjerkedal, T., Kristensen, P., Skjeret, G.A., Brevik, J.I., 2007. Intelligence test scores and birth order among young Norwegian men (conscripts) analyzed within and between families. Intelligence 35, 503e514. € rnberg, U., Latta, M., 2007. The roles of the family and the welfare state: the Bjo relationship between public and private financial support in Sweden. Curr. Sociol. 55 (3), 415e445. Bjørngaard, J.H., Bjerkeset, O., Vatten, L., Janszky, I., Gunnell, D., Romundstad, P., 2013. Maternal age at child birth, birth order, and suicide at a young age: a sibling comparison. Am. J. Epidemiol. 177 (7), 638e644. Black, S.E., Devereux, P.J., Salvanes, K.G., 2005. The more the merrier? The effect of family size and birth order on children's education. Q. J. Econ. 120 (2), 669e700. Blair, S.N., Kohl, H.W., Barlow, C.E., Paffenbarger, R.S., Gibbons, L.W., Macera, C.A., 1995. Changes in physical fitness and all-cause mortality: a prospective study of healthy and unhealthy men. JAMA 273 (14), 1093e1098. Blake, J., 1981. Family size and the quality of children. Demography 18, 421e442. Blane, H.T., Barry, H., 1973. Birth order and alcoholism: a review. Q. J. Stud. Alcohol 34 (3), 837e852. Buckles, K.S., Kolka, S., 2014. Prenatal investments, breastfeeding, and birth order. Soc. Sci. Med. 118, 66e70. Butland, B.K., Strachan, D.P., Lewis, S., Bynner, J., Butler, N., Britton, J., 1997. Investigation into the increase in hay fever and eczema at age 16 observed between the 1958 and 1970 British birth cohorts. Br. Med. J. 315 (7110), 717e721. Campbell, F.A., Ramey, C.T., 1994. Effects of early intervention on intellectual and academic achievement: a follow-up study of children from low-income families. Child Dev. 65 (2), 684e698.

Caspersen, C.J., Powell, K.E., Christenson, G.M., 1985. Physical activity, exercise, and physical fitness: definitions and distinctions for health-related research. Public Health Rep. 100 (2), 126e131. Chen, J., Millar, W.J., 1998. Age of smoking initiation: implications for quitting. Health Rep. Stat. Can. 9, 39e48. Conley, D., Bennett, N.G., 2000. Is biology destiny? Birth weight and life chances. Am. Sociol. Rev. 65 (3), 458e467. Erikson, R., Goldthorpe, J.H., 1992. The Constant Flux: a Study of Class Mobility in Industrial Societies. Clarendon Press, Oxford. Ernst, C., Angst, J., 1983. Birth Order: Facts and Fiction Concerning its Influence on Personality: a Survey of the Literature, 1946e1980. Springer-Verlag, Berlin. Gunawardana, L., Smith, G.D., Zammit, S., Whitley, E., Gunnell, D., Lewis, S., Rasmussen, F., 2011. Pre-conception inter-pregnancy interval and risk of schizophrenia. Br. J. Psychiatry 199 (4), 338e339. Harakeha, Z., Engelsa, R.C.M.E., Vermulsta, A.A., De Vriesb, H., Scholtea, R.H.J., 2007. The influence of best friends and siblings on adolescent smoking: a longitudinal study. Psychol. Health 22 (3), 269e289. Khuder, S.A., Dayal, H.H., Mutgi, A.B., 1999. Age at smoking onset and its effect on smoking cessation. Addict. Behav. 24 (5), 673e677. Kristensen, P., Bjerkedal, T., 2007. Explaining the relation between birth order and intelligence. Science 316 (5832), 1717. Magnus, P., Berg, K., Bjerkedal, T., 1985. The association of parity and birth weight: testing the sensitization hypothesis. Early Hum. Dev. 12, 49e54. Marti, B., Theodor, A., Minder, C.E., Vader, J.P., 1988. Smoking, alcohol consumption, and endurance capacity: an analysis of 6,500 19-year-old conscripts and 4,100 joggers. Prev. Med. 17 (1), 79e92. Modin, B., 2002. Birth order and mortality: a life-long follow-up of 14,200 boys and girls born in early 20th century Sweden. Soc. Sci. Med. 54, 1051e1064. Myers, J., Prakash, M., Froelicher, V., Do, D., Partington, S., Atwood, J.E., 2002. Exercise capacity and mortality among men referred for exercise testing. N. Engl. J. Med. 346 (11), 793e801. Myrskyl€ a, M., Fenelon, A., 2012. Maternal age and offspring adult health: evidence from the health and retirement study. Demography 49, 1231e1257. Myrskyl€ a, M., Silventoinen, K., Jelenkovic, A., Tynelius, P., Rasmussen, F., 2013. The association between height and birth order: evidence from 652 518 Swedish men. J. Epidemiol. Community Health 67 (7), 571e577. Patton, J., Vogel, J., Mello, R., 1982. Evaluation of a maximal predictive cycle ergometer test of aerobic power. Eur. J. Appl. Physiol. Occup. Physiol. 49 (1), 131e140. Price, J., 2008. Parent-child quality time does birth order matter? J. Hum. Resour. 43 (1), 240e265. Rostila, M., Saarela, J., Kawachi, I., 2014. Birth order and suicide in adulthood: evidence from Swedish population data. Am. J. Epidemiol. http://dx.doi.org/ 10.1093/aje/kwu090. Sandvik, L., Erikssen, J., Thaulow, E., Erikssen, G., Mundal, R., Rodahl, K., 1993. Physical fitness as a predictor of mortality among healthy, middle-aged Norwegian men. N. Engl. J. Med. 328 (8), 533e537. Shirom, A., Toker, S., Berliner, S., Shapira, I., Melamed, S., 2008. The effects of physical fitness and feeling vigorous on self-rated health. Health Psychol. 27 (5), 567. Strachan, D.P., 1989. Hay fever, hygiene, and household size. Br. Med. J. 299, 1259e1260. Strachan, D.P., 2000. Family size, infection and atopy: the first decade of the “hygiene hypothesis”. Thorax 55, S2eS10. €m, M., Duvander, A.Z.E., 2002. Gender division of childcare and the sharing Sundstro of parental leave among new parents in Sweden. Eur. Sociol. Rev. 18 (4), 433e447. Thomson, E., 2014. Family complexity in Europe. Ann. Am. Acad. Polit. Soc. Sci. 654, 245e258. Wohlfart, B., Farazdaghi, G.R., 2003. Reference values for the physical work capacity on a bicycle ergometer for men e a comparison with a previous study on women. Clin. Physiol. Funct. Imaging 23 (3), 166e170.

Birth order and physical fitness in early adulthood: evidence from Swedish military conscription data.

Physical fitness at young adult ages is an important determinant of physical health, cognitive ability, and mortality. However, few studies have addre...
633KB Sizes 0 Downloads 5 Views