Archives of Sexual Behavior

, Volume 39, Issue 1, pp 117–123

Body Size at Birth and Same-Sex Marriage in Young Adulthood


    • Department of Epidemiology ResearchStatens Serum Institut
  • Slobodan Zdravkovic
    • Department of Epidemiology ResearchStatens Serum Institut
Original Paper

DOI: 10.1007/s10508-008-9408-z

Cite this article as:
Frisch, M. & Zdravkovic, S. Arch Sex Behav (2010) 39: 117. doi:10.1007/s10508-008-9408-z


An unexplained excess of overweight has been reported among lesbians. In contrast, reports suggest that gay men may be, on average, slightly lighter and shorter than heterosexual men. We studied associations between weight, length, and body mass index (BMI) at birth and same-sex marriage in young adulthood among 818,671 Danes. We used linear regression to calculate differences in mean body measures at birth and Poisson regression analysis to calculate confounder-adjusted incidence rate ratios (IRR) of same-sex marriage according to body measures at birth. Overall, 739 persons entered same-sex marriage at age 18–32 years during 5.6 million person-years of follow-up. Birth year-adjusted mean body measures at birth were similar for same-sex married and other women. However, same-sex marriage rates were 65% higher among women of heavy birth weight (IRR = 1.65; 95% CI = 1.18–2.31, for ≥4000 vs. 3000–3499 g, p = .02), and rates were inversely associated with birth length (ptrend = .04). For same-sex married men, birth year-adjusted mean weight (−72 g, p = .03), length (−0.3 cm, p = .04), and BMI (−0.1 kg/m2, p = .09) at birth were lower than for other Danish men. Same-sex marriage rates were increased in men of short birth length (IRR = 1.45; 95% CI = 1.01–2.08, for ≤50 vs. 51–52 cm), although not uniformly so (ptrend = .16). Our population-based findings suggest that overweight in lesbians may be partly rooted in constitutional factors. Novel findings of smaller average body measures at birth in same-sex marrying men need replication. Factors affecting intrauterine growth may somehow influence sexual and partner-related choices in adulthood.


EpidemiologyCohort studiesBirth weightBody sizeObesityHomosexuality


Studies have shown that overweight and obesity and their associated health complications are more frequent among lesbian than heterosexual women (Case et al., 2004; Valanis et al., 2000). Recently, a population-based study of 5979 U.S. women aged 20–44 years from the 6th cycle of the National Survey of Family Growth showed that lesbians were approximately 250% more likely to be overweight (body mass index [BMI] 25–29 kg/m2) or obese (BMI 30+ kg/m2) compared with heterosexual women, after adjustment for a number of potential confounding factors known to affect obesity risk (Boehmer, Bowen, & Bauer, 2007). Boehmer et al. called for weight-reduction initiatives that target lesbian women and suggested that future studies should address whether it is the lesbian identity per se or the choice of a female partner that puts women at risk for overweight and obesity.

No similar association with overweight has been reported for homosexual men. Indeed, available evidence suggests that adult homosexual men may, on average, be somewhat lighter and smaller than heterosexual men. In an analysis of 4948 white postpubertal males who participated in surveys conducted by the Kinsey Institute for Research in Sex, Gender, and Reproduction between 1938 and 1963, it was reported that homosexual men have, on average, significantly lower body weight than heterosexual men (Blanchard & Bogaert, 1996). Other studies showed that homosexual men (Bogaert, 2003) or homosexual men with gender dysphoria (Blanchard, Dickey, & Jones, 1995) may be, on average, somewhat shorter than the general male population. In a recent survey of 18–24 year-old U.S. men who have sex with men (Kipke et al., 2007), participants were significantly less likely to be overweight or obese (19% of 513) compared with 18 year-old men in a national sample (30% of 1001). However, a study based on a large, national probability sample in Britain failed to identify any significant differences in height or weight between homosexual and heterosexual men, whereas bisexual men were slightly taller than heterosexual men (Bogaert & Friesen, 2002).

Because adult body composition is positively correlated with body measures at birth (Ong, 2006; Rogers, 2003; Whitaker & Dietz, 1998), we hypothesized that the reported higher BMI in lesbians might be reflected in a heavier average body composition at birth compared with heterosexual women. Likewise, we hypothesized that the lighter and smaller body composition of adult homosexual men reported in some studies might be detectable already at birth. We utilized national Danish registry data on birth characteristics and same-sex marriage in the age interval 18–32 years to examine these expectations.


Study Cohort

Using data from the Civil Registration System, a national database with continuously updated demographic information for all Danish citizens (Danish Ministry of Interior Affairs and Health, 2008), we established a national cohort of 399,486 women and 419,185 men who (1) were born between 1973 and 1987 to Danish-born mothers who were themselves born after 1935, (2) had available information about birth weight, birth length, or both in the Danish Medical Birth Registry (Knudsen & Olsen, 1998), and (3) lived in Denmark on their 18th birthday. From the Civil Registration System, we obtained information about each cohort member’s marital status, including same-sex marriage which has been a legal option in Denmark since 1989 (Anonymous, 1989), birth place, mother’s and father’s age at birth, and older and younger siblings, factors that were recently shown to influence the likelihood of same-sex marriage (Frisch & Hviid, 2006, 2007).

Data Analysis

Initially, we compared average body measures at birth for women and men who married a same-sex partner on or before July 14, 2005, with corresponding measures for all other women and men in the cohort. Specifically, using linear regression (GLM procedure in SAS v. 9.1), we tested for crude and birth year-adjusted differences in average weight, length, and BMI at birth between same-sex marrying and other women and men in the cohort. Birth year adjustment was applied to take the marked secular increase in neonatal body size into account, which has been reported in Denmark (Schack-Nielsen, Mølgaard, Sørensen, Greisen, & Michaelsen, 2006), other Nordic countries (Odlind, Haglund, Pakkanen, & Otterblad Olausson, 2003; Skjaerven, Gjessing, & Bakketeig, 2000), and Canada (Wen et al., 2003).

Secondly, we performed Poisson regression analysis to compare rates of first same-sex marriage in strata of body measures at birth. Specifically, we counted person-years of observation from each cohort member’s 18th birthday and until the date of her or his first same-sex marriage, censoring periods after the date of first heterosexual marriage, emigration, disappearance, death, or end of study (July 14, 2005). We counted numbers of first same-sex marriage and observation years in strata of birth weight (<3000, 3000–3499, 3500–3999, ≥4000 g), birth length (≤50, 51–52, ≥53 cm), BMI at birth (in quartiles, birth weight in kg divided by the squared birth length in m), age (18, 19, 20,…,32 years), calendar year (1991–1992, 1993, 1994,…,2005), mother’s age (<20, 20–24, 25–29,…,≥40 years) and father’s age (<20, 20–24, 25–29,…,≥40 years) at the time of birth, number of older (0, 1, 2, ≥3) and younger (0, 1, 2, ≥3) siblings born to the same mother, and urbanicity of the birth place (five categories ranging from the capital of Copenhagen to small towns/rural areas with <10,000 inhabitants). By means of log-linear Poisson regression analysis (GENMOD procedure in SAS v. 9.1), we calculated ratios of incidence rates (IRR) of first same-sex marriage with accompanying 95% confidence intervals (CI) between cohort members with different body measures at birth, using persons with birth weight 3000–3499 g, birth length 51–52 cm, and BMI in the 2nd quartile as reference categories. IRRs thus provide estimates of the relative rate of subsequent same-sex marriage among persons with different body measures at birth with adjustment for age and period (IRR-1), or with adjustment for age, period, and all other stratification variables mentioned above (IRR-2). Likelihood ratio tests for homogeneity were performed to assess possible differences in rates of same-sex marriage between categories of women and men with different body measures at birth. We also performed likelihood ratio tests to examine if associations with weight, length, and BMI at birth could be adequately described by linear associations and, when applicable, we report p-values for such linear trends. Throughout, p-values <.05 and 95% CIs excluding unity were considered statistically significant.


In the cohort of 818,671 persons born between 1973 and 1987, 452 women and 287 men married a same-sex partner in the age interval 18–32 years during a total of 5.6 million person-years between 1991 and 2005.


Of the 452 women who married a same-sex partner, 450 (99.6%) had information about both birth weight and birth length (Table 1). The crude mean birth weight was 110 g lower for women who subsequently married a same-sex partner than for 398,949 other women in the cohort, a statistically significant difference (3122 vs. 3232 g, p < .001). However, when taking secular changes in birth weight into account in the linear regression model, this difference in birth weight disappeared entirely. Indeed, the birth year-adjusted mean birth weight was 6 g higher among women who eventually married a same-sex partner during follow-up than among other women in the cohort, but this difference was not statistically significant. Mean birth length was similar in the two groups (51.16 cm in same-sex marrying vs. 51.24 cm in other women), whereas mean BMI at birth was 0.40 kg/m2 lower among same-sex marrying women than among other women in the cohort (11.84 vs. 12.24 kg/m2). None of the mean differences in body measures at birth were statistically significant after adjustment for birth year (Table 1).
Table 1

Body measures at birth among 739 same-sex married persons and 817,932 other persons born in Denmark during the period 1973–1987




Same-sex married women

Other women

Same-sex married men

Other men

Birth weight (grams)

(n = 451)

(n = 398,949)

(n = 284)

(n = 418,828)

Mean (SD)

3122 (593)

3232 (563)

3163 (599)

3351 (588)

Mean difference, crude* (p-value)

–110 (p < .001)

−188 (p < .001)

Mean difference, adjusted# (p-value)

+6 (p = .82)

−72 (p = .03)

Birth length (centimeters)

(n = 451)

(n = 396,738)

(n = 283)

(n = 416,267)

Mean (SD)

51.16 (2.45)

51.24 (2.45)

51.73 (2.60)

51.98 (2.59)

Mean difference, crude* (p-value)

−0.08 (p = .54)

−0.25 (p = .11)

Mean difference, adjusted# (p-value)

−0.14 (p = .23)

−0.32 (p = .04)

BMI (kg/m2)

(n = 450)

(n = 396,653)

(n = 283)

(n = 416,194)

Mean (SD)

11.84 (1.57)

12.24 (1.49)

11.72 (1.52)

12.33 (1.48)

Mean difference, crude* (p-value)

−0.40 (p < .001)

−0.61 (p < .001)

Mean difference, adjusted# (p-value)

+0.08 (p = .21)

−0.14 (p = .09)

* p-value obtained in linear regression analysis

#Adjusted mean difference and p-value obtained in linear regression analysis with adjustment for birth year

In the age- and period-adjusted Poisson analysis (Table 2), rates of same-sex marriage were significantly heterogeneous across birth weight categories, with highest rates for women in the ≥4000 g birth weight category. Upon additional adjustment for birth length and other potential confounders in the multivariate Poisson regression analysis, significant heterogeneity remained (p = .02), and rates became even relatively higher in the heaviest birth weight category (IRR = 1.65, 95% CI = 1.18–2.31, for ≥4000 vs. 3000–3499 g). There was a statistically significant inverse linear association between birth length and same-sex marriage (ptrend = .04), with women who were ≤50 cm long at birth being 26% more likely to marry a same-sex partner than women of birth length 51–52 cm (IRR = 1.26, 95% CI = 0.98–1.64). BMI at birth was weakly positively associated with the propensity to marry a same-sex partner (ptrend = .11).
Table 2

Incidence rate ratios and 95% confidence intervals of same-sex marriage among 818,671 Danes born in the period 1973–1987 according to body measures at birth





Person-years (No. same-sex marriages)

IRR-1 (95% CI)

IRR-2 (95% CI)


Person-years (No. same-sex marriages)

IRR-1 (95% CI)

IRR-2 (95% CI)

Birth weight (grams)



659,047 (122)

1.11 (0.89–1.38)

0.95 (0.73–1.23)


560,766 (67)

1.27 (0.95–1.69)

0.96 (0.66–1.39)



1,468,709 (258)

1 (ref)

1 (ref)


1,588,010 (162)

1 (ref)

1 (ref)



322,396 (20)

0.67 (0.41–1.08)

0.78 (0.47–1.27)


395,314 (20)

1.06 (0.63–1.78)

1.17 (0.69–1.99)



234,954 (51)

1.43 (1.05–1.90)

1.65 (1.18–2.31)


420,415 (35)

0.96 (0.67–1.39)

1.02 (0.68–1.51)

Test for homogeneity


p = .02

p = .02


p = .41

p = .94

Test for trend





p = .15

p = .78

Birth length (centimeters)



932,584 (176)

1.24 (1.00–1.52)

1.26 (0.98–1.64)


724,804 (87)

1.44 (1.06–1.95)

1.45 (1.01–2.08)



941,642 (145)

1 (ref)

1 (ref)


934,481 (79)

1 (ref)

1 (ref)



802,082 (130)

1.04 (0.82–1.31)

0.93 (0.72–1.21)


1,294,882 (117)

1.06 (0.79–1.40)

1.08 (0.79–1.47)

Test for homogeneity


p = .12

p = .11


p = .04

p = .14

Test for trend


p = .07

p = .04


p = .03

p = .16

BMI (quartiles)

1 (lowest)


849,860 (152)

0.90 (0.71–1.15)

0.92 (0.72–1.17)


929,146 (106)

1.02 (0.76–1.37)

1.01 (0.75–1.37)



648,079 (115)

1 (ref)

1 (ref)


728,028 (74)

1 (ref)

1 (ref)



661,244 (111)

1.06 (0.82–1.38)

1.06 (0.81–1.38)


710,243 (64)

1.06 (0.76–1.48)

1.06 (0.76–1.49)

4 (highest)


516,481 (72)

1.11 (0.82–1.50)

1.15 (0.85–1.55)


586,237 (39)

0.90 (0.61–1.34)

0.93 (0.62–1.38)

Test for homogeneity


p = .48

p = .47


p = .88

p = .92

Test for trend


p = .12

p = .11


p = .67

p = .81

IRR, incidence rate ratio; CI, confidence interval, NA, not applicable; IRR-1 adjusted for age and period; IRR-2 adjusted for age, period, number of older siblings, number of younger siblings, mother’s age at birth, father’s age at birth, and birth place. Birth weight additionally adjusted for birth length, and birth length additionally adjusted for birth weight

Values used to define quartiles of BMI (kg/m2) were <11.5, 11.5–<12.3, 12.3–<13.2, and ≥13.2 in women, and <11.5, 11.5–<12.3, 12.3–<13.3, and ≥13.3 in men


Of the 287 men who married a same-sex partner during follow-up, 283 (98.6%) had information about both birth weight and birth length (Table 1). The average birth weight was 188 g lower for men who subsequently married a same-sex partner than for 418,828 other men in the cohort (3163 vs. 3351 g), a crude difference that was statistically significant (p < .001). Same-sex marrying men were also, on average, slightly shorter at birth than other men, with mean birth lengths of 51.73 and 51.98 cm, respectively, in the two groups, and the mean BMI at birth was 0.61 kg/m2 lower among same-sex marrying men than among other men (11.72 vs. 12.33 kg/m2). In the birth year-adjusted linear regression analysis, estimated mean differences in birth weight (−72 g, p = .03) and birth length (−0.32 cm, p = .04) were statistically significant, while the mean difference in BMI at birth lost statistical significance (−0.14 kg/m2, p = .09).

In the age and period-adjusted Poisson analysis (Table 2), the only body measure at birth that was associated with same-sex marriage rates in adulthood was birth length, which exhibited statistically significant heterogeneity among the three birth length categories (p = .04). However, after further adjustment for confounders in the multivariate analysis, there remained no statistically significant heterogeneity between rates of same-sex marriage among men in different categories of birth weight, birth length, or BMI at birth. The only possible exception was that men of short birth length (≤50 cm) had higher rates of same-sex marriage than men in the 51–52 cm category (IRR = 1.45, 95% CI = 1.01–2.08).


To our knowledge, the present study is the first population-based study to address possible associations between body measures at birth and adult sexual partner preferences. Before us, North American researchers have used other study designs to study possible associations between birth weight and homosexuality in women (Blanchard & Ellis, 2001) and men (Blanchard & Ellis, 2001; Blanchard et al., 2002). Blanchard and Ellis found no overall difference in birth weight between 98 homosexual and 2071 heterosexual women or between 162 homosexual and 898 heterosexual men in their study based on questionnaire data from a convenience sample of 3229 mothers to homosexual and heterosexual probands. However, due to their interest in the possible impact of older brothers, Blanchard and Ellis noted a 169 g lower average birth weight in homosexuals among those 23% of male probands who had one or more older brothers. A subsequent study comprising 250 Canadian prepubertal boys with gender identity disorder (a group considered to be prehomosexual) and 739 control boys without gender identity problems failed to present results for the overall comparison of birth weights in the two groups (Blanchard et al., 2002). However, in a subgroup analysis of those 7% of study participants who had two or more older brothers, the authors noted a 385 g lower average birth weight in prehomosexual boys than in control boys.

As reported in detail elsewhere (Frisch & Hviid, 2006, 2007), older brothers had no measurable impact on the propensity among Danish men to marry a same-sex partner, so to reduce the risk of chance findings we did not search for associations between neonatal body dimensions and same-sex marriage in specific fraternal birth order subgroups. In our population-based cohort study, we found that high birth weight and short birth length, which are correlates of high BMI and overweight in adulthood (Ong, 2006; Rogers, 2003; Whitaker & Dietz, 1998), were positively associated with the propensity among women to marry a same-sex partner. Specifically, the 9% of women with the heaviest birth weight (≥4000 g) were 65% more likely to marry another woman than those in the reference birth weight category 3000–3499 g. Unfortunately, we had no information available about body measures at later points during childhood, adolescence, or adulthood, so we cannot address the possibility that some same-sex attracted women may become overweight or obese due to lifestyle factors that prevail among lesbians. Regardless, our study adds a new angle to prior findings of overweight in this population (Boehmer et al., 2007; Case et al., 2004; Valanis et al., 2000), suggesting that among some lesbians, as among probably any other group of individuals, overweight may be rooted in genetic factors or prenatal factors.

Unlike for women, there was no evidence to suggest an association between above-average body measures in adulthood and male homosexuality. Rather, a limited number of reports have suggested that homosexual men may be, on average, somewhat lighter (Blanchard & Bogaert, 1996) or shorter (Blanchard et al., 1995; Bogaert, 2003) or have lower BMI (Kipke et al., 2007) than heterosexual men, although the literature is not conclusive in this regard (Bogaert & Friesen, 2002). In the present study, we observed that men who later married a same-sex partner were, on average, significantly lighter and shorter at birth than other men. Specifically, the 25% boys with the shortest birth length (≤50 cm) were 45% more likely to marry another man within the time window of observation up to the maximum age of 32 years compared with men of birth length 51–52 cm. Interestingly, researchers from the United Kingdom and Finland examined heterosexual marriage rates in relation to neonatal body measures and showed that men who were small at birth were less likely than their taller and heavier peers to marry heterosexually (Phillips et al., 2001; Vågerö & Modin, 2002). These observations and the findings of the present study are compatible with the idea that factors responsible for intrauterine growth may be related not only to subsequent body dimensions (Ong, 2006) and health-related outcomes (Barker et al., 1993), but possibly also to the complex brain processes that influence sexual and partner-related choices in adulthood.

The nature of our epidemiological study does not permit detailed speculations about the underlying mechanisms for the observed associations of body size at birth with adult sexual partner choices. Any interpretation of our observations will depend on one’s preferred general framework for understanding adult sexual orientation. Some tend to believe that sexual orientation is a biologically determined trait explained largely by genetic and/or intrauterine hormonal factors (essentialist view), while others consider sexual orientation more to be the result of a complex interplay of psychosocial influences during prolonged periods of infancy, childhood, and adolescence (constructionist view). To those who favor the essentialist view, our findings may suggest the existence of some underlying biological link between genetic or pregnancy-related hormonal factors on one side and fetal size and later same-sex attraction on the other. Others who favor the constructionist view on sexual orientation may be more inclined to consider body composition at birth as a factor associated with later body composition, which, in turn, may impact on the sexual socialization of boys and girls during formative years of childhood and adolescence. Interestingly, whether biological factors, psychosocial mechanisms, or a combination of the two operate to explain the observed complex associations, sex has to be taken into consideration, because associations differed between women and men.

One limitation of the present investigation was that our outcome, i.e., first same-sex marriage at or before the maximum age of 32 years, captured only a small subset of Danish homosexuals. The mean age at first same-sex marriage among Danish homosexuals who formalized their relationship between 1989 and 2001 was 32 years in both sexes (Frisch & Hviid, 2006), so most cohort members were too young to be identified as homosexuals during the observation window between 1991 and 2005 in the present study. Moreover, only a small proportion of Danish homosexuals will probably ever opt for same-sex marriage, a subgroup previously estimated to comprise no more than a few percent of all homosexuals in Denmark (Frisch, Smith, Grulich, & Johansen, 2003). Consequently, the extent to which the observed significant associations can be assumed to apply to the broader population of individuals with same-sex attraction is unclear. However, in the absence of other data to suggest systematic differences in neonatal body measures between same-sex married and other persons with same-sex attraction, our findings may be viewed as estimates of true differences in neonatal body measures between same-sex attracted and other persons.

Another limitation to consider in our study was that we did not have information about gestational age at the time of birth for our cohort members who were born between 1973 and 1987. Such information has been available in the Danish Medical Birth Registry only since 1978 (Knudsen & Olsen, 1998), implying that only the youngest subset of women and men in our cohort who married their same-sex partner at or before age 27 years would have recorded information about gestational age at the time of their birth in the birth registry. We were, therefore, unable to directly address the impact, if any, of prematurity or postmaturity on sexual partner preferences in adulthood. Nevertheless, it seems reasonable to conclude that our findings do not support the existence of any general association between infant maturity and later sexual partner preference. Boys who eventually married a same-sex partner were, on average, slightly lighter and shorter at birth than other boys, whereas the highest rate of same-sex marriage in women was in the group with the heaviest birth weight.

The current overweight and obesity epidemic in Western societies calls for efficient large-scale and long-term weight-reduction initiatives, possibly including targeted efforts in population segments at increased risk, such as sexual-minority women (Boehmer et al., 2007). However, detailed cost-effectiveness analyses are required before allocating resources in various subgroup-specific projects. General campaigns and continuing efforts to promote healthier lifestyles and prevent and reduce overweight and obesity in childhood and adolescence may appreciably reduce the burden of overweight and obesity-related diseases in all segments of society, including women who self-identify as lesbians.

Copyright information

© Springer Science+Business Media, LLC 2008