Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Apr 14.
Published in final edited form as: J Psychiatr Res. 2022 Oct 12;156:284–290. doi: 10.1016/j.jpsychires.2022.10.026

Sons and parental cognition in mid-life and older adulthood

Katrin Wolfova a,b,c, Di Wu d, Jordan Weiss e, Pavla Cermakova b,f, Hans-Peter Kohler g, Vegard Fykse Skirbekk h, Yaakov Stern c, Alison Gemmill i, Sarah E Tom c,j,*
PMCID: PMC10103684  NIHMSID: NIHMS1885032  PMID: 36279678

Abstract

Prior research suggests a relationship between number of sons and maternal long-term health outcomes, including dementia. We assessed the relationship between having sons and parental cognitive aging. Specifically, we investigated the relationship between having at least 1 son and parental baseline cognition level and rate of cognitive decline, accounting for life course sociodemographic characteristics in a cohort of 13 222 adults aged ≥50 years from the US Health and Retirement Study. We included only participants with at least one child. We further explored whether this relationship varies by parental sex and whether the magnitude of the relationship increases with each additional son. Cognition was assessed biennially for a maximum of nine times as a sum of scores from immediate and delayed 10-noun free recall tests, a serial 7s subtraction test, and a backwards counting test. Associations were evaluated using linear mixed-effects models, stepwise adjusting for sociodemographic and health-related factors. In our analytic sample of parents, a total of 82.3% of respondents had at least 1 son and 61.6% of respondents were female. Parents of at least 1 son had a faster rate of cognitive decline in comparison to parents without any son. Our results also suggest that cognitive decline was faster among parents of multiple sons, compared to parents with only daughters. Thus, the results support the theory that having sons might have a long-term negative effect on parental cognition.

Keywords: Offspring sex, Cognitive aging, Health and retirement study

1. Introduction

A growing body of evidence shows a relationship between parenthood and long-term parental health. In previous studies of number of children and outcomes in mortality, cardiovascular events, and cognitive health, mothers and fathers of two children were more advantaged than other groups (Li et al., 2019; Ning et al., 2020; Read and Grundy, 2017; Zeng et al., 2016b). Pregnancy-related changes may contribute to long-term maternal health. For example, women experience an increase in sex steroid hormone levels during pregnancy and normotensive pregnancies are associated with a long-term decrease in maternal blood pressure (Haug et al., 2018; Peterson and Tom, 2021). At moderate doses, both changes may benefit cognition, contributing to the relative advantage in mothers of two children. A greater number of pregnancies, especially ≥ five, may result in a poor health-risk profile, such as increased weight or insulin resistance (Kim and Lee, 2017). These changes may contribute to long-term risk for poor cognition (Ogunmoroti et al., 2019). Studies that also include fathers suggest that apart from pregnancy-related pathways, social conditions may relate to number of children and health in both parents. For example, having large families was associated with an increased risk of obesity in both men and women (Lawlor et al., 2003). In addition, previous research shows that the association between number of children and parental cognition was attenuated after adjustment for socioeconomic indicators or health risk behaviours in both sexes (Gemmill and Weiss, 2021).

Recent studies suggest that not only number of children, but also composition of offspring sex might have effect on parental health outcomes. For example, researchers have found increased mortality in parents of sons (Næss et al., 2017; Zeng et al., 2016b). Several social pathways might contribute to these differential associations in health outcomes of both mothers and fathers. First, having daughters might benefit health through better management of chronic diseases as they are more likely to provide informal caregiving than sons (Cascella Carbó and García-Orellán, 2020). In addition, daughters provide more emotional support than sons to both parents and might reduce parents’ depressive symptoms (Buber and Engelhardt, 2008; Zeng et al., 2016a). The benefits of having a daughter on physical comorbidities and depression may contribute to lower risk of dementia (Livingston et al., 2020). On the other hand, parents of sons are less likely to divorce, and being married has been linked to better cognitive functioning in later life (Kabátek and Ribar, 2020; Liu et al., 2019; Mascaro et al., 2017). In addition, gendered behavior of parents towards their sons and daughters can affect lifestyle of both parents and their offspring. Higher physical activity in mothers have been associated with higher physical activity in daughters, while fathers who were more active had more active sons (Brouwer et al., 2018).

Having sons versus daughters might further influence parental cognition through sex-specific pathways. Specifically, mothers of sons have been observed to have increased cardiovascular mortality (Cesarini et al., 2007; Jasienska et al., 2006; Næss et al., 2017). Biological changes during pregnancy might contribute to this gap. Maternal placenta reacts in a sexually dimorphic manner, causing greater maternal microvascular vasodilatation in pregnancies with male fetuses (Gabory et al., 2013). Whether the differences in cardiovascular reactivity between mothers of sons and daughters persist until later life is not known. Another pathway specific to maternal health relates to microchimerism of foetal origin, where a small number of foetal cells, genetically distinct from maternal cells, subsequently persist in maternal tissues, including brain (Johnson et al., 2020). Exposure to antigens from foetal male cells may be associated with longer maternal survival, protective role against cancer development, and reduced rate of maternal ischemic heart disease (Cirello and Fugazzola, 2014; Gadi and Nelson, 2007; Hallum et al., 2020; Kamper-Jørgensen et al., 2014). The presence of male microchimerism in maternal brain may protect from age-related cognitive decline (Chan et al., 2012). Although the potential biological mechanism is not fully elucidated, microchimeric cells may benefit maternal immune system regulation (Cogle et al., 2004; Kallenbach et al., 2011). Additionally, women pregnant with male foetuses have been found to outperform women pregnant with female foetuses in tasks focused on working memory and spatial ability (Vanston and Watson, 2005). Higher testosterone and lower serum human chorionic gonadotropin levels during pregnancy with boys might contribute to these differences (Vanston and Watson, 2005). However, whether this effect persists until older ages remains unknown.

Studies of the relationship between offspring sex and parental health are limited in inclusion of confounding variables and study population. First, studies of maternal brain health and microchimerism do not account for socioeconomic variables. The Trivers-Willard hypothesis states that females in good conditions are more likely to have sons, whereas females in poor conditions are more likely to have daughters (Trivers and Willard, 1973). This hypothesis assumes a positive correlation between mother’s and her offspring’s condition at the time of reproduction and that “good” condition increases reproductive success more in sons than in daughters. Mothers in good condition are more likely to have sons, compared to mothers in poor condition, due to this fitness advantage. In contemporary society, socioeconomic status represents condition, which supports previous findings of more sons among parents of high socioeconomic status (Cameron and Dalerum, 2009; Luo et al., 2017). Second, brain health in mothers of boys might be advantaged due to social benefits of parenting boys, such as higher likelihood of life in marriage, which may similarly influence paternal brain health. It may be particularly important to study effects of offspring sex in males, given that previous research shows that fathers of boys spend more time parenting and have higher wages than fathers of girls (Choi et al., 2008; Lundberg and Rose, 2002; Mammen, 2011). However, previous studies have not considered the role of offspring sex on paternal brain health.

As one of the first studies to synthesize offspring sex composition and later life parental cognition, we conducted an exploratory analysis of the relationship between having sons and subsequent parental cognitive aging, accounting for life course sociodemographic characteristics. We utilized data from the Health and Retirement Study (HRS), a prospective study of middle-aged and older adults in the United States. We further determined whether this relationship varies by parental sex. Our primary exposure of interest was having at least one son. We secondarily explored whether the magnitude of the relationship increases with each additional son and whether the relationship differs by cognitive domain.

2. Methods

2.1. Source of data

We utilized data from the HRS, an ongoing prospective study in the United States of more than 30 000 adults aged 50+ years and their spouses of any age. The study began in 1992 and since then has included biennial surveys. To maintain a representative sample, the study enrolls refresher samples every 6 years. All participants signed a written informed consent. The HRS was approved by the Ethics Committee of the University of Michigan and the National Institute on Aging.

2.2. Offspring sex

Participants self-reported number of sons and number of daughters, which included biological, adopted, and step-children. Disentangling the effect of offspring sex from number of children is analytically challenging as these concepts are strongly correlated. Therefore, we used two definitions of offspring sex. In primary analysis, we classified offspring sex using a binary indicator defined as no son vs. at least 1 son. In secondary analysis, we used a categorical variable number of sons (0 sons, 1 son, 2 sons, 3 or more sons).

2.3. Cognition

Cognition was assessed in all waves using a sum of scores from immediate and delayed 10-noun free recall tests, a serial 7s subtraction test, and a backwards counting test. Immediate recall scores (range: 0–10) were measured as the number of recalled words after the interviewer read a list of 10 nouns. After approximately 5 minutes, interviewers asked participants to recall any of the previously presented nouns (range: 0–10). The serial 7s subtraction scores (range: 0–5) denotes the number of times over 5 trials the participant was able to consecutively subtract 7 from 100. The backward counting scores (range: 0–2) were measured as the number of successful trials when asked to count backwards for 10 continuous numbers beginning with the number 20 and then 86. A total cognition score is a sum of individual scores (range: 0–27). In primary analysis, we utilized the total score as the outcome. Secondary analyses separately used each cognitive test as the alternative outcome.

2.4. Covariates

Participants reported their baseline ages (which were centered at the median by subtracting median of 62 years form baseline age for each participant); race and ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, other); birth cohort (pre–World War I, World War I, and Spanish influenza (1899–1920); pre–Great Depression (1921–1928); Great Depression (1929–1939); World War II and postwar (1940–1950)); number of children (1 child, 2 children, 3 children, 4 children, ≥ 5 or children), regardless of whether biological, adopted and/or step-children; education (less than high school, General Educational Development program or high school, college and above); father’s and mother’s education (in years); marital status (married or partnered, other); age at the first birth (<20 years old, ≥20 years old); place of birth (Southern states, other US states, abroad); prevalent heart disease (yes, no), stroke (yes, no), and diabetes (yes, no); and smoking status (ever smoked, never smoked). Body mass index (BMI) was calculated from self-reported baseline data as weight in kilograms divided by height in meters squared. Depressive symptoms were measured at baseline by the 8-item Center for Epidemiological Studies-Depression Scale (CES-D; score range: 0–8). For secondary analysis, we created a categorical variable number of daughters (0 daughters, 1 daughter, 2 daughters, 3 or more daughters).

2.5. Sample selection

Only participants with cognition assessment in 2000 (cognitive functions were measured consistently from 2000 onwards) were considered for the present analysis. As our exposure of interest is offspring sex and the probability of having a son or a daughter is zero in people without children, we included only participants with ≥1 child (n = 17 516). We restricted the analytic sample to participants who had at least two assessments of cognition (n = 15 784) and had no missing data on number of children and offspring sex (n = 13 563). We further restricted the sample to participants who were ≥50 years at baseline (n = 13 222) (Fig. 1).

Fig. 1.

Fig. 1.

Open in a new tab

Selection of the analytical sample.

Note: HRS=Health and Retirement Study.

2.6. Statistical analysis

2.6.1. Descriptive statistics

Descriptive statistics are presented as weighted frequency (n (%)), mean ± standard deviation (SD), or median and interquartile range (IQR), where appropriate.

2.6.2. Primary analysis

To assess the association between offspring sex with baseline cognition and cognitive decline, we employed linear mixed-effects models with subject-specific random intercept and slope effects with an unstructured covariance using the R package “nlme”. The primary outcome was total cognition. Time was treated as a continuous variable specified as years since baseline. The primary exposure variables were offspring sex (no son vs. ≥ one son), and the interaction of time with offspring sex, which allows for rate of change over time to differ by offspring sex. Models were stepwise adjusted for covariates: 1) baseline age (in years, centered at median), participant’s sex and race and ethnicity in Model 1; 2) all variables in previous model plus number of children in Model 2; 3) all variables in previous model plus potential confounders (birth cohort, education, father’s education, mother’s education, age at the first birth, place of birth) in Model 3. Age and time in our model reflect changes related to the effect of aging. We have included birth cohort as it captures the effect of being born at a specific point in time, which is independent of the process of aging (Holford, 1991). Finally, we stratified all models by participant’s sex and by the number of children (one child vs. ≥ two children). When sample included only parents of one child, models did not adjust for number of children. The primary analysis equation is given in Supplement.

2.6.3. Secondary analysis

We conducted two sets of secondary analyses with 1) an alternative exposure and 2) alternative outcomes.

In the alternative exposure models with total cognition as the outcome, the primary exposure variables were the number of sons, and the interaction of time with the number of sons (time × number of sons). Models were stepwise adjusted for covariates: 1) baseline age (in years, centered at median), the number of daughters (0 daughters, 1 daughter, 2 daughters, 3 or more daughters), the interaction of time with the number of daughters (time × number of daughters), participant’s sex and race and ethnicity in Model 1; 2) all variables in previous model plus potential confounders (birth cohort, education, father’s education, mother’s education, age at the first birth, place of birth) in Model 2.

The alternative outcomes models were constructed the same way as models in the primary analysis with having at least 1 son as the exposure and separately for each cognitive test (immediate recall, delayed recall, serial 7s subtraction, and backwards counting).

2.6.4. Sensitivity analysis

To check the robustness and consistency of our results, we conducted six sets of sensitivity analyses using the same methodological approach as in the primary analysis. First, to account for potential mediation, the model adjusted for potential confounders (Model 4) was adjusted additionally for potential mediators (baseline marital status; smoking status; BMI; depressive symptoms; and prevalent diabetes, heart disease, and stroke). Second, to assess the impact of missing data on covariates, we repeated the analysis using complete cases who had all variables used in Model 4 (n = 10 827). Third, to examine the impact of clustering effect, we constructed a multilevel model with subject-specific and household-specific random intercept and slope effects. Fourth, to correct for selection of specific observations with unequal probabilities, we applied HRS-provided person-level sampling weights which reflect the complex survey design. Survey weights account for differential probability of inclusion in the sample and nonresponse, thus allowing for the estimation of nationally representative estimates. HRS intentionally oversamples Black participants, Hispanic participants, and Florida residents. Standard error estimation accounts for geographic stratification and clustering. Full description of sampling weights can be found at https://hrs.isr.umich.edu/documentation/survey-design. Sampling weights were first rescaled to frequency weights and weighted models were then fitted using the R package “lme4”. Fifth, we excluded from the sample the participants who were identified by algorithm based on cognitive scores as having developed dementia during the course of the study (n = 2 311). We created a variable representing estimated de mentia status using Langa-Weir approach (Crimmins et al., 2011). Finally, we used joint modelling using the R package “JM” to account for selection bias due to death as previously done in other studies (Rouanet et al., 2022). The joint model combines a linear mixed-effects model (cognition as a continuous outcome) and a Cox model (death as a time-to-event process) while incorporating their shared random effects.

Analyses were conducted using R statistical programming language (version 4.0.5) and Stata (version 16.0).

3. Results

Among 13 222 participants in the unweighted analytic sample (median age, 65; IQR, 59–73; 61.6% females), a total of 10 872 (82.3%) participants had at least one son. Of those, 4 862 (44.7%) participants had one son, 3 523 (32.4%) had two sons, and 2 487 (22.9%) had three or more sons. Of participants without any sons, 891 (37.9%) had one daughter, 905 (38.5%) had two daughters, and 554 (23.6%) had three or more daughters (Supplementary Table 1). The baseline primary outcome measure is normally distributed (Supplementary Fig. 1). Participants in the final analytical sample had a total of 86 901 cognitive assessments, with a median follow-up period of 14 years (IQR, 8–16 years). A total of 97.7% of follow-up assessments occurred in two-year intervals. During follow up, 5 809 (43.9%) participants died. The sociodemographic and health-related characteristics of participants with at least one son and with no sons were similar (Table 1).

Table 1.

Characteristics of the sample.

At least 1 son No son Overall
Female, n (%) 6 697 (61.6%) 1 447 (61.6%) 8 144 (61.6%)
Cognition, mean (SD) 15.6 (4.51) 15.5 (4.58) 15.6 (4.52)
Age, mean (SD) 66.5 (9.18) 66.6 (10.0) 66.5 (9.34)
Number of children, n (%)
 1 child 909 (8.4%) 891 (37.9%) 1 800 (13.6%)
 2 children 3 150 (29.0%) 905 (38.5%) 4 055 (30.7%)
 3 children 2 740 (25.2%) 347 (14.8%) 3 087 (23.3%)
 4 children 1 808 (16.6%) 139 (5.9%) 1 947 (14.7%)
 5 or more children 2 265 (20.8%) 68.0 (2.9%) 2 333 (17.6%)
Age at first birth < 20, n (%) 2 269 (20.9%) 349 (14.9%) 2 618 (19.8%)
Birth cohort, n (%)
 1899–1920 1 194 (11.0%) 342 (14.6%) 1 536 (11.6%)
 1921–1928 2 134 (19.6%) 447 (19.0%) 2 581 (19.5%)
 1929–1939 4 585 (42.2%) 828 (35.2%) 5 413 (40.9%)
 1940–1950 2 959 (27.2%) 733 (31.2%) 3 692 (27.9%)
Race and ethnicity, n (%)
 Non-Hispanic White 8 385 (77.1%) 1 859 (79.1%) 10 244 (77.5%)
 Hispanic 1 428 (13.1%) 313 (13.3%) 1 741 (13.2%)
 Non-Hispanic Black 861 (7.9%) 142 (6.0%) 1 003 (7.6%)
 Others 198 (1.8%) 36.0 (1.5%) 234 (1.8%)
Place of birth, n (%)
 Southern states, n (%) 3 699 (34.0%) 833 (35.4%) 4 532 (34.3%)
 Other US states, n (%) 6 163 (56.7%) 1 334 (56.8%) 7 497 (56.7%)
 Abroad, n (%) 1 010 (9.3%) 183 (7.8%) 1 193 (9.0%)
Education, n (%)
 Less than high school or GED 3 165 (29.1%) 616 (26.2%) 3 781 (28.6%)
 High school or some college 5 739 (52.8%) 1 280 (54.5%) 7 019 (53.1%)
 College and above 1 968 (18.1%) 454 (19.3%) 2 422 (18.3%)
Father’s education, mean (SD)a 8.88 (3.58) 8.97 (3.52) 8.90 (3.57)
Mother’s education, mean (SD)b 9.18 (3.33) 9.35 (3.20) 9.21 (3.31)
Married/partnered, n (%) 7 473 (68.7%) 1 542 (65.6%) 9 015 (68.2%)
Diabetes, n (%) 1 516 (13.9%) 308 (13.1%) 1 824 (13.8%)
Heart disease, n (%) 2 101 (19.3%) 432 (18.4%) 2 533 (19.2%)
Stroke, n (%) 610 (5.6%) 145 (6.2%) 755 (5.7%)
Ever smoked, n (%) 6 247 (57.5%) 1 385 (58.9%) 7 632 (57.7%)
BMI, mean (SD)c 27.4 (5.23) 27.2 (5.36) 27.4 (5.25)
Depressive symptoms, mean (SD)d 1.50 (1.88) 1.49 (1.90) 1.50 (1.89)
Died, n (%) 4 767 (43.8%) 1 042 (44.3%) 5 809 (43.9%)
Open in a new tab

Abbreviations: SD = standard deviation; BMI = body mass index; GED = General Educational Development program.

Note: Percentage and mean and SD presented.

a

Does not include 14.1% of participants who did not provide information on father’s education.

b

Does not include 10.0% of participants who did not provide information on mother’s education.

c

Does not include 1.3% of participants who did not provide information on BMI.

d

Does not include 0.04% of participants who did not provide information on depressive symptoms.

In the primary analysis, parents of at least one son had a faster rate of cognitive decline (B = −0.015; 95% CI −0.029, −0.002; Model 1; Supplementary Table 2, Fig. 2) in comparison to those with no sons. Further adjustment of sociodemographic and health-related variables did not change the results. When stratified by parental sex, the estimates of the association of having at least one son with the rate of cognitive decline was similar in both men (B = −0.016; 95% CI −0.036, 0.005; Model 1; Supplementary Table 3) and women (B = −0.014; 95% CI −0.032, 0.004; Model 1; Supplementary Table 3, p value for a three-way interaction of time, parental sex and offspring sex in the model including all participants = 0.956). Further adjustment for potential confounders did not change the results and the rate of cognitive decline was similar across sexes (Supplementary Table 3, Fig. 3). When modeled separately for parents of one child, having a son was not associated with cognition (Supplementary Table 4). When modeled separately for parents of 2 or more children, the models produced similar results as the models in the main analysis (Supplementary Table 4, p value for a three-way interaction of time, parental sex and number of children in the model including all participants = 0.348). Having at least one son was not associated with the level of baseline cognition in any model.

Fig. 2.

Fig. 2.

Open in a new tab

Relationship between having at least 1 son and cognition.

Model 1: baseline age (in years, centered at median), sex and race and ethnicity.

Model 2: + number of children.

Model 3: + birth cohort, education, father’s education, mother’s education, age at the first birth, place of birth.

Model 4 (sensitivity analysis): + marital status, smoking status, body mass index, depressive symptoms, diabetes, heart disease, stroke.

Fig. 3.

Fig. 3.

Open in a new tab

Relationship between having at least 1 son and cognition (primary analysis, Model 3).

Note: Estimates from linear mixed-effects models in an unweighted sample. Models are adjusted for baseline age, race and ethnicity, number of children, birth cohort, education, father’s education, mother’s education, age at the first birth, place of birth.

In the secondary analysis using an alternative primary exposure, having more sons was associated with a faster rate of cognitive decline, with a consistent association in parents of three or more sons across all three models (Table 2, Supplementary Fig. 2). Having more sons was not associated with the level of baseline cognition. In secondary analysis using alternative outcomes, having at least one son was associated with a faster rate of cognitive decline in immediate and delayed recall (Supplementary Table 2). We did not find any association with cognitive decline in serial 7s and backwards counting (Supplementary Table 6).

Table 2.

Relationship between number of sons and cognition (secondary analysis).

B (95% CI)
Model 1 Model 2 Model 3
Intercept 16.355 *** 12.787 *** 13.306 ***
(16.126; 16.584) (12.307–13.268) (12.736–13.875)
1 son 0.169 (−0.012; 0.351) 0.114 (−0.065 −0.293) 0.089 (−0.088 −0.267)
2 sons 0.097 (−0.097; 0.292) 0.077 (−0.116 −0.271) 0.059 (−0.133 −0.250)
3 or more sons − 0.099 (−0.309; 0.111) 0.177 (−0.035 −0.390) 0.167 (−0.043 −0.378)
Time −0.194 *** (−0.213; − 0.176) − 0.193 *** (−0.213 to − 0.173) −0.183 *** (−0.203 to − 0.163)
1 son × time − 0.014 (−0.029; 0.002) −0.016 (−0.033 −0.000) −0.018 * (−0.034 to − 0.001)
2 sons × time − 0.020 * (− 0.036; − 0.003) −0.017 (−0.034 −0.001) −0.017 (−0.035 −0.001)
3 or more sons × time − 0.023 * (−0.041; − 0.005) −0.025 * (−0.044 to −0.006) −0.024 * (−0.043 to −0.005)
N 13 222 10 967 10 827
AIC 449 012.055 373 549.494 367 824.328
Open in a new tab
*

P < 0.05

**

P < 0.01

***

P < 0.001.

Abbreviations: B = unstandardized beta, CI = confidence interval; N = number of participants; AIC = Akaike information criterion.

Note: Estimates from linear mixed-effects models in an unweighted sample.

Model 1: baseline age (in years, centered at median) and race and ethnicity.

Model 2: + birth cohort, education, father’s education, mother’s education, age at the first birth, place of birth.

Model 3 (sensitivity analysis): + marital status, smoking status, body mass index, depressive symptoms, diabetes, heart disease, stroke.

In sensitivity analysis, the model that included potential mediators produced similar results (Table 2, Supplementary Table 2). We did not find any substantial differences in results from the analysis of complete cases (Supplementary Table 7). Multilevel models with participants nested within households produced results that were nearly identical to the main analysis (results not shown). In analysis of the weighted sample and the sample restricted to participants without estimated dementia status, the relationships between having sons and cognition were attenuated, but the point estimates were similar to the estimates from the main analysis (Supplementary Table 7, Supplementary Table 8). Results from the joint model were similar to results from linear mixed-effects models (Supplementary Table 7).

4. Discussion

In the HRS, a nationally-representative, population-based sample of middle-aged and older adults in the United States, parents with ≥ one son had a faster rate of cognitive decline. This difference was modest compared to the overall cognitive decline over time, but persisted when accounting for sociodemographic and health factors. Our results also suggest that cognitive decline was faster among parents of multiple sons, compared to parents with only daughters. Having sons was not related to the level of baseline cognition. We found similar relationships between having sons with cognition for mothers and fathers, though the strength of the relationship was more modest compared to the full sample.

Our study challenges previous findings that show a link between having male offspring and maternal health advantage. Some of the studies suggest a protective effect of having boys on maternal health through microchimerism (Hallum et al., 2020; Kamper-Jørgensen et al., 2014). These studies have several limitations, including lack of key social variables related to parenting boys, such as lower likelihood of divorce or higher socioeconomic position. In such cases, the effect of having sons would be present also in fathers, or would diminish after adjusting for sociodemographic factors. Although our study did not include a direct measure of male microchimerism, we accounted for several sociodemographic and health-related variables in a sample of both men and women. We found that cognition deteriorated faster in mothers and fathers of sons than in those without any sons. This association was attenuated in weighted analysis that accounts for clustering effect, which suggests that aspects of parenting sons shared by both parents might play role in cognitive aging. For example, daughters are more likely to show positive affect and sociability, while sons have higher activity levels (Olino et al., 2013).

On the other hand, our findings are in line with the evidence from non-human studies, in which raising male offspring has been linked to accelerated maternal aging (Douhard et al., 2020; Froy and Gamelon, 2020). These findings have been explained by higher demands for maternal resources during pregnancy with male offspring as well as greater energy investment after birth (Douhard et al., 2020). Similar pathways have been proposed in humans. Male newborns have higher average birthweight (Loos et al., 2001), and birth of a son is followed by a longer birth interval (Mace and Sear, 1997). In addition, some, but not all, previous studies suggest higher long-term mortality in mothers of sons (Cesarini et al., 2007; Jasienska et al., 2006; Næss et al., 2017). Given the potential difference in mortality between parents of sons, we accounted for selective attrition by joint modelling, which produced similar results to the linear mixed-effects models.

Studies on the relationship between the number of children and parental cognition imply that inclusion of social aspects of parenting is of importance in human research. Previously, researchers have found the same relationship favouring parents of two children in comparison to individuals without children or parents of four or more children in mothers as well as in fathers (Ning et al., 2020). While in more traditional societies having sons would be expected to advance family’s socioeconomic status, this might not be the case in the contemporary United States (Hurt et al., 2006). On the contrary, it is possible that parents of sons are more likely to be disadvantaged later in life as daughters provide more social support than sons and more often become informal caregivers (Cascella Carbó and García-Orellán, 2020; Friedman et al., 2015). In addition, offspring sex might influence parental health behaviors. For example, mothers of first-born daughters weigh less than mothers of first-born sons a long time after pregnancy (Pham-Kanter, 2010). Also, parents of daughters are less likely to smoke, drink alcohol or take drugs (N Powdthavee, 2010). Our results show that having sons is related to more rapid cognitive decline rather than to pre-existing level of cognition, suggesting that later-life factors might play a role. Future studies should examine the relationship between the gendered parent-offspring interactions in later life and parental cognitive aging.

Strengths of our study include longitudinal nature of the analysis, large and population-based sample, and methodological approach that accounts for the risk of death. The key explanatory variables are less likely to vary over time, as a reproductive period of 50+ individuals is usually completed. Our sample includes a diverse population, and we were able to control for detailed demographic characteristics. First, we adjusted for race and ethnicity or whether the participant was born in a Southern state. Literature suggests enduring racial cognitive disparities among Southern-born Black individuals may reflect historical patterns of impacts of historical trauma, deprivation and segregation beginning in early childhood and lasting throughout the life course (Liu et al., 2015). Second, we adjusted for whether the participant was born abroad. Prior research documented a healthy immigrant effect, which means that foreign-born individuals are more likely to have better health in comparison to individuals born in the host country (Corlin et al., 2014; Hill et al., 2012). In addition, preference of sons has been documented among populations of immigrants to the US (Howell et al., 2018). Finally, inclusion of diverse family structures in our analysis reflects contemporary families (Nakonezny et al., 2003). Divorces and formation of stepfamilies increased following the no-fault divorce laws implemented in the US beginning in 1953, which allow for divorce without proving wrongdoing by either party.

Nevertheless, our study has several limitations. We were not able to distinguish whether participants’ offspring were biological, adopted, or step-children. Thus, if biological pathways of pregnancy contribute to the link between having a son and cognition, our inclusion of all children may bias the results in mothers. Next, if the association between offspring sex and cognition is driven by microchimerism, data on pregnancies and sex of the fetus rather than offspring sex would produce less biased results, as the foeto-maternal transfer begins as early as four weeks and five days after conception (Dawe et al., 2007). Participants in the HRS survey report the number of sons and daughters rather than number of pregnancies and sex of the fetus. However, even data on pregnancies may not accurately capture all aspects of gravidity, as the incidence of clinically unrecognized miscarriage is estimated to be around 20% and the sex of the fetus is unlikely to be reported (Apgar and Churgay, 1993). Also, despite the breadth of available data in HRS, we were not able to control for some important factors, such as shared genetic factors between cognition and fertility, or sex and gender differences in parenting, in caregiving over the life course, or in the social and emotional aspects of the relationship with children and in risk taking behaviors of children (Hallum et al., 2020; Liao and Scholes, 2017; Woodley of Menie et al., 2016). Future studies should consider inclusion of these variables to improve our understanding of the possible pathways. Finally, our findings might not apply to other populations that differ in social, cultural and economic characteristics, compared to this nationally representative sample of the United States population of adults aged ≥50 years, born in 1899–1950.

Our study supports the hypothesis that having sons is associated with long-term negative parental costs. Although the specific mechanisms linking this relationship remain to be elucidated, our results suggests that the pathways are rather of a social than biological character, as we observed similar trends in mothers as well as fathers.

Supplementary Material

Supplementary Material

Acknowledgments

We gratefully acknowledge the HRS pariticpants and the HRS team. KW received travel grant support from the Fulbright Comission and Charles University. ST received funding from the United States National Institutes of Health (K01AG050723)

Footnotes

Declaration of competing interest

None.

Ethics considerations

The HRS was approved by the Ethics Committee of the University of Michigan and the National Institute on Aging. Approval from the Charles University is not needed as the Health and Retirement Study has been approved by the US committees.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.jpsychires.2022.10.026.

Data availability

Data from the HRS are openly available after registration at the corresponding website [http://hrsonline.isr.umich.edu/index.php?p=data].

References

  1. Apgar BS, Churgay CA, 1993. Spontaneous abortion. Prim Care 20 (3), 621–627. [PubMed] [Google Scholar]
  2. Brouwer SI, Küpers LK, Kors L, Sijtsma A, Sauer PJJ, Renders CM, Corpeleijn E, 2018. Parental physical activity is associated with objectively measured physical activity in young children in a sex-specific manner: the GECKO Drenthe cohort. BMC Publ. Health 18 (1), 1033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Buber I, Engelhardt H, 2008. Children’s impact on the mental health of their older mothers and fathers: findings from the Survey of Health, Ageing and Retirement in Europe. Eur. J. Ageing 5 (1), 31–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cameron EZ, Dalerum F, 2009. A Trivers-Willard effect in contemporary humans: male-biased sex ratios among billionaires. PLoS One 4 (1), e4195. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Cascella Carbó GF, García-Orellán R, 2020. Burden and gender inequalities around informal care. Invest. Educ. Enfermería 38 (1). [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cesarini D, Lindqvist E, Wallace B, 2007. Maternal longevity and the sex of offspring in pre-industrial Sweden. Ann. Hum. Biol. 34 (5), 535–546. [DOI] [PubMed] [Google Scholar]
  7. Chan WFN, Gurnot C, Montine TJ, Sonnen JA, Guthrie KA, Nelson JL, 2012. Male microchimerism in the human female brain. PLoS One 7 (9), e45592. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Choi H-J, Joesch JM, Lundberg S, 2008. Sons, daughters, wives, and the labour market outcomes of West German men. Lab. Econ. 15 (5), 795–811. [Google Scholar]
  9. Cirello V, Fugazzola L, 2014. Positive effect of fetal cell microchimerism on tumor presentation and outcome in papillary thyroid cancer. Chimerism 5 (3–4), 106–108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Cogle CR, Yachnis AT, Laywell ED, Zander DS, Wingard JR, Steindler DA, Scott EW, 2004. Bone marrow transdifferentiation in brain after transplantation: a retrospective study. Lancet 363 (9419), 1432–1437. [DOI] [PubMed] [Google Scholar]
  11. Corlin L, Woodin M, Thanikachalam M, Lowe L, Brugge D, 2014. Evidence for the healthy immigrant effect in older Chinese immigrants: a cross-sectional study. BMC Publ. Health 14, 603. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Crimmins EM, Kim JK, Langa KM, Weir DR, 2011. Assessment of cognition using surveys and neuropsychological assessment: the health and retirement study and the aging, demographics, and memory study. J. Gerontol. B Psychol. Sci. Soc. Sci. 66 (Suppl. 1), i162–i171. Suppl 1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Dawe GS, Tan XW, Xiao ZC, 2007. Cell migration from baby to mother. Cell Adhes. Migrat. 1 (1), 19–27. [PMC free article] [PubMed] [Google Scholar]
  14. Douhard M, Festa-Bianchet M, Pelletier F, 2020. Sons accelerate maternal aging in a wild mammal. Proc. Natl. Acad. Sci. U. S. A. 117 (9), 4850–4857. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Friedman EM, Shih RA, Langa KM, Hurd MD, 2015. US prevalence and predictors of informal caregiving for dementia. Health Aff. 34 (10), 1637–1641. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Froy H, Gamelon M, 2020. Late-life costs of raising sons in bighorn sheep. Proc. Natl. Acad. Sci. U. S. A. 117 (11), 5556–5558. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Gabory A, Roseboom TJ, Moore T, Moore LG, Junien C, 2013. Placental contribution to the origins of sexual dimorphism in health and diseases: sex chromosomes and epigenetics. Biol. Sex Differ. 4 (1), 5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Gadi VK, Nelson JL, 2007. Fetal microchimerism in women with breast cancer. Cancer Res. 67 (19), 9035–9038. [DOI] [PubMed] [Google Scholar]
  19. Gemmill A, Weiss J, 2021. The relationship between fertility history and incident dementia in the US Health and Retirement Study. J. Gerontol. B Psychol. Sci. Soc. Sci. [DOI] [PubMed] [Google Scholar]
  20. Hallum S, Gerds TA, Sehested TSG, Jakobsen MA, Tjønneland A, Kamper-Jørgensen M, 2020. Impact of male origin microchimerism on cardiovascular disease in women: a prospective cohort study. Am. J. Epidemiol. [DOI] [PubMed] [Google Scholar]
  21. Haug EB, Horn J, Markovitz AR, Fraser A, Macdonald-Wallis C, Tilling K, Romundstad PR, Rich-Edwards JW, Åsvold BO, 2018. The impact of parity on life course blood pressure trajectories: the HUNT study in Norway. Eur. J. Epidemiol. 33 (8), 751–761. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hill TD, Angel JL, Balistreri KS, Herrera AP, 2012. Immigrant status and cognitive functioning in late-life: an examination of gender variations in the healthy immigrant effect. Soc. Sci. Med. 75 (12), 2076–2084. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Holford TR, 1991. Understanding the effects of age, period, and cohort on incidence and mortality rates. Annu. Rev. Publ. Health 12, 425–457. [DOI] [PubMed] [Google Scholar]
  24. Howell EM, Zhang H, Poston DL, 2018. Son preference of immigrants to the United States: data from U.S. Birth certificates, 2004–2013. J. Immigr. Minority Health 20 (3), 711–716. [DOI] [PubMed] [Google Scholar]
  25. Hurt LS, Ronsmans C, Quigley M, 2006. Does the number of sons born affect long-term mortality of parents? A cohort study in rural Bangladesh. Proc. Biol. Sci. 273 (1583), 149–155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Jasienska G, Nenko I, Jasienski M, 2006. Daughters increase longevity of fathers, but daughters and sons equally reduce longevity of mothers. Am. J. Hum. Biol. 18 (3), 422–425. [DOI] [PubMed] [Google Scholar]
  27. Johnson BN, Ehli EA, Davies GE, Boomsma DI, 2020. Chimerism in health and potential implications on behavior: a systematic review. Am. J. Med. Genet. 182 (6), 1513–1529. [DOI] [PubMed] [Google Scholar]
  28. Kabátek J, Ribar DC, 2020. Daughters and divorce. Econ. J. 131 (637), 2144–2170. [Google Scholar]
  29. Kallenbach LR, Johnson KL, Bianchi DW, 2011. Fetal cell microchimerism and cancer: a nexus of reproduction, immunology, and tumor biology. Cancer Res. 71 (1), 8–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Kamper-Jørgensen M, Hjalgrim H, Andersen AM, Gadi VK, Tjønneland A, 2014. Male microchimerism and survival among women. Int. J. Epidemiol. 43 (1), 168–173. [DOI] [PubMed] [Google Scholar]
  31. Kim JH, Lee SJ, 2017. Parity and increased risk of insulin resistance in postmenopausal women: the 2010 Korean National Health and Nutrition Examination Survey. Menopause 24 (7), 832–837. [DOI] [PubMed] [Google Scholar]
  32. Lawlor DA, Emberson JR, Ebrahim S, Whincup PH, Wannamethee SG, Walker M, Smith GD, 2003. Is the association between parity and coronary heart disease due to biological effects of pregnancy or adverse lifestyle risk factors associated with child-rearing? Findings from the British Women’s Heart and Health Study and the British Regional Heart Study. Circulation 107 (9), 1260–1264. [DOI] [PubMed] [Google Scholar]
  33. Li W, Ruan W, Lu Z, Wang D, 2019. Parity and risk of maternal cardiovascular disease: a dose-response meta-analysis of cohort studies. Eur J Prev Cardiol 26 (6), 592–602. [DOI] [PubMed] [Google Scholar]
  34. Liao J, Scholes S, 2017. Association of social support and cognitive aging modified by sex and relationship type: a prospective investigation in the English longitudinal study of ageing. Am. J. Epidemiol. 186 (7), 787–795. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Liu H, Zhang Y, Burgard SA, Needham BL, 2019. Marital status and cognitive impairment in the United States: evidence from the national health and aging trends study. Ann. Epidemiol. 38, 28–34 e22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Liu SY, Glymour MM, Zahodne LB, Weiss C, Manly JJ, 2015. Role of place in explaining racial heterogeneity in cognitive outcomes among older adults. J. Int. Neuropsychol. Soc. 21 (9), 677–687. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Livingston G, Huntley J, Sommerlad A, Ames D, Ballard C, Banerjee S, Brayne C, Burns A, Cohen-Mansfield J, Cooper C, Costafreda SG, Dias A, Fox N, Gitlin LN, Howard R, Kales HC, Kivimäki M, Larson EB, Ogunniyi A, Orgeta V, Ritchie K, Rockwood K, Sampson EL, Samus Q, Schneider LS, Selbæk G, Teri L, Mukadam N, 2020. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet 396 (10248), 413–446. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Loos RJ, Derom C, Eeckels R, Derom R, Vlietinck R, 2001. Length of gestation and birthweight in dizygotic twins. Lancet 358 (9281), 560–561. [DOI] [PubMed] [Google Scholar]
  39. Lundberg S, Rose E, 2002. The effects of sons and daughters on men’s labor supply and wages. Rev. Econ. Stat. 84 (2), 251–268. [Google Scholar]
  40. Luo L, Ding R, Gao X, Sun J, Zhao W, 2017. Socioeconomic status influences sex ratios in a Chinese rural population. PeerJ 5, e3546. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Mace R, Sear R, 1997. Birth interval and the sex of children in a traditional African population: an evolutionary analysis. J. Biosoc. Sci. 29 (4), 499–507. [DOI] [PubMed] [Google Scholar]
  42. Mammen K, 2011. Fathers’ time investments in children: do sons get more? J. Popul. Econ. 24 (3), 839–871. [Google Scholar]
  43. Mascaro JS, Rentscher KE, Hackett PD, Mehl MR, Rilling JK, 2017. Child gender influences paternal behavior, language, and brain function. Behav. Neurosci. 131 (3), 262–273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. N Powdthavee SW, Oswald A, 2010. The Effects of Daughters on Health Choices and Risk Behaviour. Discussion Papers 10/03.
  45. Næss Ø, Mortensen LH, Vikanes Å, Smith GD, 2017. Offspring sex and parental health and mortality. Sci. Rep. 7 (1), 5285. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Nakonezny PA, Rodgers JL, Shaw K, 2003. Did births decline in the United States after the enactment of no-fault divorce law? Soc. Biol. 50 (3–4), 188–200. [DOI] [PubMed] [Google Scholar]
  47. Ning K, Zhao L, Franklin M, Matloff W, Batta I, Arzouni N, Sun F, Toga AW, 2020. Parity is associated with cognitive function and brain age in both females and males. Sci. Rep. 10 (1), 6100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Ogunmoroti O, Osibogun O, Kolade OB, Ying W, Sharma G, Vaidya D, Michos ED, 2019. Multiparity is associated with poorer cardiovascular health among women from the Multi-Ethnic Study of Atherosclerosis. Am. J. Obstet. Gynecol. 221 (6), 631. e631–631.e616. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Olino TM, Durbin CE, Klein DN, Hayden EP, Dyson MW, 2013. Gender differences in young children’s temperament traits: comparisons across observational and parent-report methods. J. Pers. 81 (2), 119–129. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Peterson A, Tom SE, 2021. A lifecourse perspective on female sex-specific risk factors for later life cognition. Curr. Neurol. Neurosci. Rep. 21 (9), 46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Pham-Kanter G, 2010. The Gender Weight Gap: Sons, Daughters, and Maternal Weight. MPRA Paper 28997. [Google Scholar]
  52. Read SL, Grundy EMD, 2017. Fertility history and cognition in later life. J. Gerontol. B Psychol. Sci. Soc. Sci. 72 (6), 1021–1031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Rouanet A, Avila-Rieger J, Dugravot A, Lespinasse J, Stuckwisch R, Merrick R, Anderson E, Long L, Helmer C, Jacqmin-Gadda H, Dufouil C, Judd S, Manly J, Sabia S, Gross A, Proust-Lima C, 2022. How selection over time contributes to the inconsistency of the association between sex/gender and cognitive decline across cognitive aging cohorts. Am. J. Epidemiol. 191 (3), 441–452. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Trivers RL, Willard DE, 1973. Natural selection of parental ability to vary the sex ratio of offspring. Science 179 (4068), 90–92. [DOI] [PubMed] [Google Scholar]
  55. Vanston CM, Watson NV, 2005. Selective and persistent effect of foetal sex on cognition in pregnant women. Neuroreport 16 (7), 779–782. [DOI] [PubMed] [Google Scholar]
  56. Woodley of Menie MA, Schwartz JA, Beaver KM, 2016. How cognitive genetic factors influence fertility outcomes: a mediational SEM analysis. Twin Res. Hum. Genet. 19 (6), 628–637. [DOI] [PubMed] [Google Scholar]
  57. Zeng Y, Brasher MS, Gu D, Vaupel JW, 2016a. Older parents benefit more in health outcome from daughters’ than sons’ emotional care in China. J. Aging Health 28 (8), 1426–1447. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Zeng Y, Ni ZM, Liu SY, Gu X, Huang Q, Liu JA, Wang Q, 2016b. Parity and all-cause mortality in women and men: a dose-response meta-analysis of cohort studies. Sci. Rep. 6, 19351. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material
NIHMS1885032-supplement-Supplementary_Material.docx (258.1KB, docx)

Data Availability Statement

Data from the HRS are openly available after registration at the corresponding website [http://hrsonline.isr.umich.edu/index.php?p=data].

RESOURCES