Social, demographic and health characteristics of men fathering children at different ages

The aim of this study was to examine associations between selected sociodemographic, socioeconomic, and health characteristics and the rates of fatherhood in different age groups. We investigated rates between 2011 and 2015 in a population-based register study including all men born from 1945 to 1995 residing in Denmark in 2011. The study population consisted of 1,867,108 men who fathered 268,612 children during the follow-up. The associations were quantified as incidence rate ratios using Poisson regression. Young men had higher rates of fathering a child if they lived outside the Capital Region, had a relatively high income, were previously diagnosed with cardiovascular disease, psychoactive substance abuse, personality disorders, schizophrenia or behavioural and emotional disorders. Men of advanced age had higher rates of fathering a child when born outside Denmark, living in the Capital Region, were in the lower or upper 10th percentile income group, were self-employed or unemployed or previously diagnosed with depression. Men of advanced age had lower rates of fathering a child if previously diagnosed with somatic diseases, psychoactive substance abuse or mental retardation. The findings highlight the importance of consideration of various sociodemographic, socioeconomic, and health characteristics when studying associations between paternal age and offspring health.


Materials and methods
Study population. This study was performed as a population-based register study including all men born from January 1, 1945 to December 31, 1995 and who were residing in Denmark on January 1, 2011. We investigated the rate of fathering a child in the period from January 1, 2011 to December 31, 2015, a total of five years. We included all men on January 1, 2011, regardless of parity, and followed them until December 31, 2015, regardless of whether they fathered a child or not (or whether they would father a child after December 31, 2015). In total, 1,867,108 men constituted the study population and 218,959 of these fathered 268,612 children during the five-year follow-up period.
The study population was divided into nine age groups according to their age on January Data. Data were obtained from several Danish health and administrative registers. Since 1968, anyone living in Denmark is given a unique personal registration number, the so-called CPR (Central Personal Register) number 28 . This number is used to identify individuals in all official registers. Statistics Denmark extracted information from the various registers using the CPR number, pseudo-anonymised the CPR numbers to PNR (personal number) numbers and stored the data on a secure server for analysis. With the PNR, it was possible to link information about individuals from different registers.
We used data from registers held by the Danish Health Data Authority and Statistics Denmark, including the Civil Registration System, the Employment Classification Module, the Medical Birth Registry, the Danish National Patient Register (DNPR) and the Danish Psychiatric Central Research Register [28][29][30][31][32][33][34] . The processing and linkage of data were approved by the Danish Data Protection Agency (UCHP reference number: 514-0230/18-3000). Ethical approval or informed consent was not required for register-based studies according to Danish legislation.
Information about country of birth, region of residency, educational level, employment status, and income was obtained from the Civil Registration System, the Population Education Register, the Employment Classification Module, and the Income Statistics Register in Statistics Denmark [35][36][37] .
Place of birth was categorised as Denmark and other than Denmark. Region of residency (on January 1, 2011) was defined as one of the five administrative regions of Denmark: the Capital region, Other Zealand, Southern Denmark, Central Jutland, and Northern Jutland. The highest educational level attained (as of January 1, 2011) was categorised according to the Danish International Standard Classification of Education 15, D-ISCED-15, and was divided into primary and lower secondary school (ISCED 1 and 2, approximately 9-10 years of education), upper secondary school (ISCED 3, approximately 12-13 years of education), bachelor's or equivalent (ISCED 5 and 6, approximately 15-16 years of education), and master's and doctoral level (ISCED 7 and 8, often at least 17-18 years of education) 38 . Employment status was defined using the primary source of income in 2010 and categorised as employed, self-employed, student, retired (including disability pension), unemployed (persons registered as job-ready who received unemployment benefits or cash benefits more than half of the year in 2010), or unattached to the labour market (persons not registered as job-ready with no or a very low income from work, and no or very low income from unemployment benefits and cash benefits, and who are not in training/ education). The individual's disposable income in the period from January 1, 2010 to December 31, 2010 was divided into percentiles: lower 10th percentile, 10-25th percentile, 25-75th percentile, 75-90th percentile, and upper 90th percentile.
Information about medical history was obtained from the DNPR, which has registered all diagnoses given at hospitals since 1977. The International Classification of Diseases 8th Revision (ICD-8) and 10th Revision (ICD-10) were used for diagnoses. In Denmark, the ICD-8 was used from 1969 to 1993, and the ICD-10 from 1994 onwards. Men were classified as having a mental illness or a somatic disease if they had been admitted as an inpatient or had been in outpatient care before January 1, 2011. We included somatic and mental illnesses where a paternal age-related increased risk had previously been suggested 8,9 or as rough proxies for lifestyle diseases. The following somatic diseases were included: malignant and benign tumours, cardiovascular diseases, endocrine diseases and congenital anomalies. The following mental illnesses were included: mental and behavioural disorders due to psychoactive substance uses (from here denoted psychoactive substance abuse), schizophrenia and related disorders, depression, bipolar disorders, personality disorders, mental retardation, pervasive developmental disorders, and behavioural and emotional disorders. The diseases can be seen in Supplementary Table I,  together with the corresponding ICD-10 and ICD-8 codes for the included diagnoses.
Statistical analyses. We investigated characteristics of men fathering a child at different ages. Older men will differ from younger men simply because they are older. As we were not interested in this difference, we www.nature.com/scientificreports/ investigated how selected sociodemographic, socioeconomic and health characteristics were associated with the rate of fatherhood in the predefined age groups, estimating incidence rate ratios (IRRs) using Poisson regression. Crude IRRs were estimated in each of the nine age groups according to the men's age on January 1, 2011 and presented graphically. For all sociodemographic characteristics, the reference group is all men in that age group who are in a specific stratum of the variable, e.g., all men in regular employment. For mental illnesses and somatic diseases, the reference group is all men in that age group who had not received the specific diagnosis. The IRRs were not adjusted for other factors, as the analysis performed is a description of the relationship between the selected characteristics and fatherhood in different age groups-and not an attempt to establish any causal relationships. To evaluate if the IRRs for fatherhood relating to different characteristics differed between the various age groups, we performed a Wald test of interaction between age group and the specific characteristic. The men in the study population entered the study on January 1, 2011 and were followed until death, emigration, or December 31, 2015, whichever came first. If a man fathered a child several times during the follow-up period, all births were counted in the rate.
All statistical analyses was carried out using STATA (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC). Table 1 shows the distribution of selected sociodemographic, socioeconomic, and health characteristics within each age group of Danish men aged 15-65 years on January 1, 2011. More than 75% of the children were born to fathers 25-39 years of age. Figure 1 shows the incidence rates (IRs) of fathering a child in each age group for the entire study period. The highest rates of fathering a child were seen in the age groups 25-29 years and 30-34 years and the lowest rates of fathering a child were seen in the age groups 15-19 years and 40-65 years. Figures 2, 3, 4, 5, 6, 7 and 8 show the IRRs of fathering a child in each age group according to a specific sociodemographic characteristic, somatic disease or mental illness compared with the reference group. The IRRs and the corresponding confidence intervals can be seen in the Supplementary Table II. Socioeconomic characteristics. Men born in countries outside Denmark had the highest rates of fathering a child in the youngest-and oldest age groups compared with men born in Denmark (Fig. 2); from age 35 years, the IRRs were higher with higher age, reaching a sevenfold higher rate (IRR = 6.97, 95% CI 5.75-8.43) of fathering a child in the oldest age group.

Results
Compared with men living in the Capital Region, men in all other regions had higher rates of fathering a child at young ages (15-29 years) and lower rates for fathering a child at older ages (35 + years) ( Fig. 3A-D).
In general, compared with men with a bachelor's degree or equivalent, men in all age groups had a lower rate of fathering a child if their educational level was primary/lower secondary school or upper secondary school (Fig. 4A,B). Men with a master's or doctoral level education had a higher rate of fathering a child from age 25 years and older; at age 55-65 years, the rate was almost doubled (IRR = 1.97, 95% CI 1.49-2.61) (Fig. 4C).
Men at an advanced age (≥ 35 years) that were self-employed had higher rates of fathering a child compared with men in regular employment; the highest IRR was seen for men aged 55-65 years (IRR = 1.81, 95% CI 1.39-2.34) (Fig. 5A). For men who were unemployed or unattached to the labour market, lower rates of fathering a child at a younger age (≤ 39 years) and higher rates of fathering a child at an older age (≥ 40 years) were seen ( Fig. 5C,E). Retired men had, in general, lower rates of fathering a child (Fig. 5D). Students had lower rates of fathering a child at a young age (≤ 34 years) and higher rates of fathering a child at an advanced age (≥ 35 years) (Fig. 5B). Wide confidence intervals were seen for students of advanced age, because there are few students and very few men fathered a child in these age groups.
Compared to men with a yearly disposable income corresponding to the 25-75th percentile, men with a lower income had lower rates of fathering a child at a young age (15-34 years) and higher rates of fathering a child at older ages (40 + years) (Fig. 6A,B). In general, men with a higher income had higher rates of fathering a child at all ages (Fig. 6C,D).

Health characteristics. Somatic diseases.
For men previously diagnosed with a somatic disease, the rate of fathering a child was, in general, lower than or similar to the rate for men who had not previously been diagnosed with the diagnosis in question ( Fig. 7A-D). This was, however, not the case for young men (15-24 years) who were previously diagnosed with a cardiovascular disease; these men had a slightly higher rate of fathering a child compared with men who had not received a diagnosis for a cardiovascular disease (Fig. 7B).
Mental illnesses. Men previously diagnosed with a psychoactive substance abuse had higher rates of fathering a child at young ages compared with men who were not previously diagnosed with a psychoactive substance abuse (15-19 years IRR = 2.77, 95% CI 2.48-3.09; 20-24 years IRR = 1.20, 95% CI 1.18-1.29). Among men above 24 years, rates of fathering a child were lower in all age groups (Fig. 8A).
Concerning schizophrenia and other related disorders, men who had received such diagnoses had a higher rate of fathering a child at age 15-19 years (IRR = 1.66, 95% CI 1.27-2.15) and a similar rate at age 45-49 years compared with men without such diagnoses. In all other age groups, lower rates of fathering a child were seen. Among the oldest fathers (≥ 50 years), the IRRs were estimated with a large uncertainty, as few men diagnosed with schizophrenia and related disorders fathered a child in these age groups (Fig. 8B).
Men previously diagnosed with depression had lower rates of fathering a child at the age 20-44 years compared with men who had not earlier been diagnosed with depression. For men ≤ 19 years and ≥ 45 years, the www.nature.com/scientificreports/ rates were higher. The incidence rate was 62% higher (IRR = 1.62, 95% CI 1.05-2.51) for men aged 55-65 years (Fig. 8C).
Higher rates of fathering a child at age 15-19 years were seen for men previously diagnosed with personality disorders (Fig. 8E) or behavioural and emotional disorders (Fig. 8H) compared with men who had not previously received the diagnoses in question (IRR = 2.31, 95% CI 1.62-3.29 and IRR = 1.73, 95% CI 1.60-1.88). For men aged 25-44 years, the rates were lower, and, for men aged ≥ 45 years, the rates were similar to the rates seen in men who were not previously diagnosed with a personality disorder or behavioural and emotional disorders.
Men previously diagnosed with bipolar disorder (Fig. 8D), mental retardation (Fig. 8F), or pervasive developmental disorders (Fig. 8G) had lower rates of fathering a child in all age groups compared with men who had not earlier received the diagnosis in question (Fig. 8D,F,G). Very few men with these diagnoses fathered a child when above the age of 40 years (for pervasive developmental disorders none in the age group 40-65 years and for mental retardation none in the age group 55-65 years) and, therefore, the IRRs in these age groups were either not estimated or estimated with great uncertainty.

Discussion
This study showed that the rate of fathering a child was highest for men aged 25-34 years and that the rate at different ages varied according to sociodemographic-and socioeconomic factors and health characteristics among men living in Denmark in 2011-2015.
Men in this age group, 25-34 years, had higher rates of fathering a child if they had a bachelor or higher educational level and if they were working or studying and had an income within the 25-75th percentile or higher. Concerning mental health, for all diagnosis included, the rates of fathering a child in this age group were lower if previously diagnosed with a mental illness, than if not previously diagnosed with a mental illness.
Since our study is a descriptive study, the results cannot explain mechanisms behind the differences in the rates of fathering a child. It is, however, worth highlighting that men diagnosed with a mental illness in general have a lower rate of fathering a child in the ages where most men become fathers. We can only speculate on the  www.nature.com/scientificreports/ reasons for this, e.g. whether it is an active choice or involuntary, but knowledge about these mechanisms may form the basis for initiatives directed to improve health and quality of life for men suffering from mental illness and might alter the rates of fatherhood for this group of men. We found that young men (≤ 24 years) who lived outside the capital region and young men who had a relatively high income had higher rates of fathering a child than those living in the capital region and those who had a lower income, respectively. In the youngest age group (≤ 19 years) those born outside of Denmark had higher rates of fathering a child than those born in Denmark. Moreover, we found that young men previously diagnosed with cardiovascular disease had a higher rate of fathering a child, which was not seen among men aged 25-34 years. This group of young men likely consisted mainly of men with congenital heart disease, who may have had to deal with difficulties in life and many hospitalizations due to their disease. This may contribute to an earlier maturation, but other mechanisms behind may be present. No studies investigating family planning in young men with congenital heart disease were found to confirm this hypothesis.
We also found that men were more likely to father a child in the very young age group (≤ 19 years) if they had previously been diagnosed with psychoactive substance abuse, personality disorders, schizophrenia, or behavioural and emotional disorders. A previous study investigating mental health in offspring find that offspring of young mothers, and to a lesser degree also young fathers, have a higher risk of developing mental illness, e.g. psychoactive substance abuse, hyperkinetic disorders, mental retardation, lower mean intelligence scores, and psychological disorders 8 . It is well known that mental illness in parents has negative health impact in children [39][40][41][42] . Various factors may contribute to this association, including factors related to age. Studies investigating the circumstances and mechanisms underlying this association could potentially lead to a better prevention of adverse outcomes in the offspring.
A higher rate of fathering a child at an advanced age (≥ 35 years) was seen among men who lived in the Capital Region, men who had the lowest and highest income, and men who were either self-employed or unemployed. In correspondence with the results of our study, a Norwegian study from 2013 showed that men who became fathers for the first time at an advanced age formed a heterogeneous group, in which the majority had a stable income and social situation and a minority were characterised by a low educational level, unemployment or single status 26 . This suggests that there might be two groups of men, in terms of sociodemographic and socioeconomic factors, who father a child at an advanced age; men with a high educational level and a relatively high income, and men with a low educational level, with no or little attachment to the labour market and a relatively low income.
In our study, we did not only include first time fathers because negative health effects in offspring have been found in studies including children fathered by men of advanced age regardless of whether the child was the first, second or third or more child. However, we were not able to include details about the number of previous children. Future studies examining the above mentioned two groups of men in relation to the number of previous www.nature.com/scientificreports/ children and how adverse outcomes in offspring are distributed between these two groups could contribute to the discussion about underlying mechanisms, including potential selection forces related to late fatherhood. In terms of health characteristics, a higher rate of fathering a child at an advanced age was also seen among men who were previously diagnosed with depression. This trend may reflect the postponement of parenthood in men previously diagnosed with a depression; however, as we investigated the rate of men fathering a child regardless of whether the child was the first, second or later child, this study cannot shed light on this aspect. Postponement of parenthood is a rising tendency and further studies investigating differences in the rate of fathering a child at an advanced age and the underlying mechanisms could contribute with important knowledge.
Several of the characteristics were associated with a lower rate of fatherhood in the older age group, e.g. the somatic diseases and psychoactive substance abuse. A previous diagnosis of schizophrenia or related disorders likewise seemed to be associated with a lower rate of fatherhood at an advanced age, which is interesting as studies show an association between advanced paternal age and schizophrenia in children 9,43 .
Strengths of this study are the nationwide coverage, including all men aged 15-65 years living in Denmark on January 1, 2011, and the large sample size. Another methodological strength is the design of the analysis, enabling us to compare the rate of fatherhood according to specific characteristics across age groups. Because preconditions for many sociodemographic, socioeconomic and health characteristics are closely linked to age, e.g. income and cardiovascular diseases, it is not possible to directly compare younger and older men. To overcome this, we investigated the rate of fathering a child in different age groups according to the included sociodemographic, socioeconomic and health characteristics and thus estimated incidence rate ratios that were comparable across age groups.
The weaknesses of our study include the restriction to variables that Statistics Denmark or other administrative registers routinely collect, including DNPR. DNPR was established in 1976 and we had access to data on diagnoses from 1977 onwards. As a consequence, men receiving a diagnosis before 1977 appeared as if they had never received the concerned diagnosis-unless they had been registered with this diagnosis in relation to later contact with the healthcare system. This misclassification is only possible for men born before 1977 and we expect it to be largest for men with non-chronic diseases, including depression, psychoactive substance abuse and a variety of somatic diseases. It is, however, unlikely that such diseases not registered in the healthcare system for more than 30 years would have a major impact on the results, though this cannot be ruled out. There is also a more general risk of misclassification of the mental illnesses and somatic diseases. Men are in general consulting a primary care physician less frequently than women and are also less likely to seek help for mental health  www.nature.com/scientificreports/ problems 44,45 . As a consequence, a general underdiagnoses might be present, which would dilute the differences between men with and without the diagnosis. The sociodemographic-and socioeconomic characteristics were defined as per 2010, including region of residents, educational level, employment status and disposable income. These characteristics may have changed shortly before 2010 or during the five-year study period, which may also result in misclassification of the characteristics and should be taken into account when interpreting the results.
Men are less likely to seek help for mental health problems even when they are experiencing significant levels of psychological distress; Another limitation related to the diagnosing of the included diseases arises when we compare the IRRs across age groups. We performed a Wald test of interaction between age group and the specific characteristic, but, due to the difference in the number of years where a diagnosis could be registered and due to the fact that the diagnosing of some of the included diseases, e.g. behavioural and emotional disorders, has changed during the period from 1977 to 2010, it is difficult to conclude whether the significant interaction analysis estimates are due to circumstances related to the diagnosing or due to a real difference in the relationship between a diagnosis and fathering a child across age groups.

Conclusion
The results show that the rate of fathering a child in different age groups varied according to sociodemographic, socioeconomic, and health characteristics of men living in Denmark in 2011.
It is beyond the scope of this study to draw conclusions about risk factors for offspring of fathers at different ages or recommendations for men becoming fathers, even though these are both needed and important. Our   www.nature.com/scientificreports/ findings do, however, point to characteristics that are important in relation to the age-related rate of fatherhood and the interpretation of results from studies investigating the association between paternal age and offspring health. Moreover, the results highlight the importance of careful consideration of various sociodemographic, socioeconomic, and health characteristics in future studies. Factors associated with parenthood that may have negative health consequences for offspring are important for the general population and for couples considering parenthood.