Documente Academic
Documente Profesional
Documente Cultură
doi: 10.1093/ije/dyw073
Advance Access Publication Date: 10 May 2016
Original article
Abstract
Background: Previous studies have suggested that young mothers more often have chil-
dren with ADHD. We used sibling comparisons to examine the nature of this association
and to investigate if this association is explained by early environment or genetic and
socioeconomic factors.
Methods: A large population-based cohort including all singletons born in Denmark
from 1 January 1991 through 31 December 2005 was followed from birth until 30 April
2011. Data were available for 94% (N ¼ 943 785) of the population. Offspring ADHD was
identified by an ICD-10 diagnosis of Hyperkinetic Disorder (HKD). We used sibling-
matched Cox regression to control for genetic and socioeconomic factors.
Results: In the population cohort we found that children born by parents aged 20 years
or younger had more than twice the risk of being diagnosed with ADHD compared with
children with parents between 26 and 30 years of age. When comparing full siblings the
associations were attenuated, but we found a trend of increased risk of ADHD with
decreasing maternal age, which was not seen for paternal age.
Conclusions: Sibling comparisons suggested that the associations between both mater-
nal and paternal age and ADHD are partly explained by common genetic and socioeco-
nomic factors. The trend of increased risk of ADHD with decreasing maternal age, but
not with paternal age, may be linked to pregnancy or early-life environmental factors.
Even though only a smaller part of the association can be attributed to environmental
factors, there is a public health interest to support young parents through their first years
of parenthood.
Key words: Maternal age, paternal age, parental age, hyperkinetic disorder, attention-deficit/hyperactivity
disorder, sibling design
C The Author 2016; all rights reserved. Published by Oxford University Press on behalf of the International Epidemiological Association
V 409
410 International Journal of Epidemiology, 2017, Vol. 46, No. 2
Key messages
• Children diagnosed with ADHD were more likely to have younger parents than children without ADHD.
• The highest risk of ADHD was found when both parents were very young.
• Sibling comparisons suggested that the association between both young maternal and paternal ages and ADHD
young parents and their children may not be fully effective in reducing offspring ADHD. Public policy initiatives
should be targeted at attention to young parents and individual support of at-risk parents and their children.
association is to be expected between young paternal age Registry (DMBR) and the Danish Psychiatric Central
and ADHD and, further, an additional risk is to be ex- Research Register (DPCR) with psychiatric outcome.
pected when both parents are very young. Few studies We defined a cohort study with follow-up of singletons
have reported on associations between paternal age and born in Denmark from 1 January 1991 to 31 December
ADHD, and results have been conflicting.23,27,39,40 A study 2005. All births by women with permanent residence in
from Sweden reported advanced paternal age to be associ- Denmark (N ¼ 999 713) were identified in the Danish
ated with an increased risk of ADHD.40 In contrast, an- Medical Birth Registry, DMBR. We excluded multiple
other study from Israel found that the paternal age births (n ¼ 37 406) and, after linkage with the DCPR, we
distribution among children with ADHD was similar to excluded additionally 18 522 children with insufficient or
what we see in the general population.39 missing data for the biological father; this left 943 785
We studied offspring ADHD according to maternal as children in the cohort (Figure 1). The study was approved
Figure 1. Flowchart.
412 International Journal of Epidemiology, 2017, Vol. 46, No. 2
and outpatient visits at psychiatric hospitals and clinics splines with five knots, which are the default knot values
since 1995. based on Harrell’s recommended percentiles.41 The adjusted
HR was in addition adjusted for parity, maternal smoking
during pregnancy and gender. As maternal and paternal
Statistical analyses
ages are highly correlated, we adjusted for the other parent’s
Maternal and paternal ages were studied separately. We age by including the age difference between parents in the
used maternal or paternal age between 26 and 30 years as model; this covered both the parental age difference and the
the reference group. In a subanalysis, maternal and pater- age of the other parent as a confounder.43
nal ages were further divided into three age groups— < 26, Gestational age and birthweight are assumed to be po-
26–30, > 30 years—to study parental age combining both tential intermediate factors and were therefore not ad-
maternal and paternal age. We used maternal and paternal justed for in the final models,44 but in subanalyses we
respectively) (Table 1). We found that both decreasing ma- children with mothers aged 29 years or fathers aged 32
ternal and paternal ages were associated with higher risk of years, a higher risk of ADHD was found with decreasing
being diagnosed with ADHD; the associations were margin- maternal or paternal age (Figures 2 and 3) For children
ally attenuated after adjustment for the measured covariates. with an older father (> 30 years), the younger the mother
Children with mothers or fathers 20 years of age or younger the higher risk of ADHD. The risk increased by 25%
had more than twice the risk of ADHD (HR ¼ 2.24, 95% CI (HR ¼ 1.25, 95% CI 1.18–1.33) among children born to
2.07–2.42 and HR ¼ 2.28, 95% CI 2.03–2.57, respectively) mothers aged 26–30 years and by 98% (HR ¼ 1.98, 95%
(Table 2) compared with children with parents between 26 CI 1.81-2.15) among children born to mothers aged below
and 30 years of age. The associations between the catego- 26 years (Table 3). Among children born to mothers older
rized maternal as well as paternal age and ADHD followed a than 30 years, the risk of ADHD was also affected by pa-
‘dose-response’ relation with increasing risk of ADHD with ternal age (Table 3). The combination of both a young
Variable All births No. (%) ADHD-diagnose Maternal age, mean Paternal age, mean
No. (%) (95% CI) (95% CI)
Total no. (%) 943 785 (100) 12 294 (1.30) 29.43 (29.42–29.44) 32.32 (32.31–32.34)
Gender no. (%)
Female 458 749 (48.61) 2647 (0.58) 29.63 (29.61–29.64) 32.32 (32.30–32.34)
Male 485 036 (51.39) 9647 (1.99) 29.64 (29.62–29.65) 32.33 (32.31–32.34)
Birthweight, no. (%), grams
2000 13 224 (1.40) 281 (2.12) 29.74 (29.65–29.83) 32.50 (32.39–32.61)
2001–2500 23 953 (2.54) 467 (1.95) 29.36 (29.29–29.42) 32.03 (31.95–32.11)
2501–3000 111 746 (11.84) 1723 (1.54) 29.09 (29.06–29.12) 31.91 (31.87–31.94)
3001–3500 307 588 (32.59) 3827 (1.24) 29.35 (29.33–29.37) 32.10 (32.08–32.12)
3501–4000 318 112 (33.71) 3845 (1.21) 29.78 (29.76–29.80) 32.43 (32.41–32.45)
>4000 158 909 (16.84) 2013 (1.27) 30.27 (30.25–30.29) 32.81 (32.78–32.83)
Missing 10 253 (1.09) 138 (1.35) 30.08 (29.99–30.17) 32.99 (32.87–33.10)
Gestational age, no. (%), full weeks
< 32 6581 (0.70) 144 (2.19) 29.86 (29.74–29.99) 32.58 (32.43–32.73)
32–36 38 441 (4.07) 708 (1.84) 29.52 (29.47–29.57) 32.18 (32.12–32.24)
37–40 634 759 (67.26) 8218 (1.29) 29.66 (29.65–29.67) 32.35 (32.34–32.37)
> 40 250 395 (26.53) 3015 (1.20) 29.59 (29–57.29.60) 32.27 (32.25–32.29)
Missing 13 609 (1.44) 209 (1.54) 29.27 (29.19–29.35) 32.22 (32.11–32.32)
Apgarscore 5 min
9 67 232 (7.12) 1070 (1.59) 29.59 (29.56–29.63) 32.16 (32.12–32.21)
10 86 3277 (91.47) 11 031 (1.28) 29.63 (29.62–29.64) 32.33 (32.32–32.34)
Missing 13 276 (1.41) 193 (1.45) 29.70 (29.61–29.78) 32.69 (32.59–32.80)
Parity
1 406 476 (43.07) 5565 (1.37) 27.76 (27.75–27.78) 30.54 (30.53–30.56)
2 342 850 (36.33) 4245 (1.24) 30.22 (30.21–30.23) 32.85(32.84–32.87)
>2 177 717 (18.83) 2255 (1.27) 32.82 (32.80–32.84) 35.40 (35.38–35.43)
Missing 16 742 (1.77) 229 (1.37) 29.10 (29.03–29.18) 32.02 (31.93–32.11)
Smoking during pregnancy
Non smoker 676 117 (71.64) 6661 (0.99) 29.89 (29.88–29.90) 32.56 (32.55–32.57)
Smoker 219 876 (23.30) 4957 (2.25) 28.85 (28.83–28.87) 31.59 (31.56–31.61)
Missing 47 792 (5.06) 676 (1.41) 29.54 (29.50–29.58) 32.37 (32.32–32.42)
Birthyear
1991–95 318 085 (33.70) 4839 (1.52) 28.87 (28.86–28.89) 31.63 (31.61–31.65)
1996–2000 318 521 (33.75) 5025 (1.58) 29.67 (29.65–29.68) 32.36 (32.34–32.38)
2001–2005 307 179 (32.55) 2430 (0.79) 30.38 (30.37–30.40) 33.01 (32.99–33.03)
414 International Journal of Epidemiology, 2017, Vol. 46, No. 2
Table 2. Population cohort;crude and adjusted associations between maternal and paternal age and ADHD
Age groups, years Crude HR*1 Adjusted HR*2 Crude HR*1 Adjusted HR*2
No. ¼ 943 785 No. ¼ 883 933 No. ¼ 943 785 No. ¼ 883 933
Figure 2. Population cohort; crude and adjusted association between maternal age as a continuous variable expressed as cubic spline and ADHD.
Sibling cohort in the half sibling cohort (P-value 0.21 and 0.56,
In the full sibling cohort we found a weaker ‘dose- respectively).
response’ relation with higher risk of ADHD with decreas- Further, results from the full sibling cohort suggested
ing maternal age (P-value < 0.05). Compared with children that a sibling born when both parents were < 26 years old
born to mothers aged between 26 and 30 years, the risk of had 61% (HR ¼ 1.61, 95% CI 1.13–2.30) higher risk of
ADHD was increased by 28% (HR ¼ 1.28, 95% CI 0.94– ADHD compared with a later-born sibling with both par-
1.73) among children born to mothers aged 20 years or ents being older than 30 years old. Compared with a sib-
younger (Table 4). Maternal age above 30 years remained ling with both parents above 30 years, a sibling born
associated with lower risk of ADHD in the offspring earlier in the parents’ lifespan with only a father above 30
(Table 4). In the half sibling cohort, the same tendency of a years had a 32% (HR ¼ 1.32, 95% CI 1.12–1.56) and a
‘dose-response’ relation was seen(P-value 0.30). We 70% (HR ¼ 1.70, 95% CI 1.26–2.29) higher risk of
found no indication of a ‘dose-response’ relation be- ADHD if the mother was 26–30 years old or younger than
tween paternal age and ADHD either in the full sibling or 26 years, respectively (Table 5). Conversely, in full siblings,
International Journal of Epidemiology, 2017, Vol. 46, No. 2 415
Table 3. Population cohort; crude and adjusted combined effect of parental age on risk of ADHD
born to mothers aged above 30 years, decreasing paternal including comparisons between children with different de-
age was no longer associated with an additional risk for grees of genetic similarity support the hypothesis that
ADHD. (Table 5). part of the population-wide association between young
maternal age as well as young paternal age and offspring
ADHD may be explained by family-related factors. The as-
Discussion sociation may partly be caused by genetic risk factors, as
In the population-based cohort we found that young ma- we observed the weakest association in the full sibling
ternal age as well as young paternal age were associated comparison, which provides a stronger adjustment for
with offspring ADHD, with the highest risk found when shared genetics than the half sibling comparison. Initiating
both parents were younger than 26 years old. In the sibling parenting at a young age may be a proxy for behavioural
comparison, the association between maternal age and genetic risk factors shared among members of the same
ADHD was weaker, with the weakest association observed family and transmitted from parents to their offspring,
in full siblings. We found no association between contributing to young parental age as well as offspring
paternal age and ADHD in full or half siblings. The results ADHD.
416 International Journal of Epidemiology, 2017, Vol. 46, No. 2
Table 4. Sibling cohorts; crude and adjusted associations between parental age and ADHD
Age groups, years Crude HR*1 Adjusted HR*2 Crude HR*1 Adjusted HR*2
Maternal age No. ¼ 12 659 No. ¼ 11 853 No. ¼ 3549 No. ¼ 3281
20 1.16 (0.90–1.51) 1.28 (0.94–1.73) 1.33 (0.81–2.20) 1.38 (0.78–2.45)
21–25 1.10 (0.96–1.26) 1.15 (0.98–1.34) 1.16 (0.87–1.55) 1.24 (0.90–1.73)
26–30 1.00 (ref.) 1.00 (ref.) 1.00 (ref.) 1.00 (ref.)
31–35 0.77 (0.67–0.89) 0.73 (0.62–0.85) 0.80 (0.58–1.10) 0.89 (0.62–1.27)
> 35 0.65 (0.49–0.87) 0.61 (0.44–0.85) 0.71 (0.39–1.29) 0.79 (0.41–1.54)
Table 5. Sibling cohorts; crude and adjustedcombined effects of parental age on risk of ADHD
Full siblingcohort
Young parents represent a selected group of young peo- history of hyperactivity problems32 and they may transmit
ple, and many of their characteristics may increase the risk the genetic susceptibility to ADHD in the offspring even
of ADHD in offspring. Compared with those who become though the severity of their own symptoms would not
parents at an older age, young parents more often have a reach the diagnostic threshold. This hypothesis is also
International Journal of Epidemiology, 2017, Vol. 46, No. 2 417
supported by research reporting that young people with child especially in the first years, which may explain this
ADHD behaviour are found to have earlier sexual debut finding.
and have more risky sexual behaviour with more sexual This study supports that the association between young
partners, and thus a higher probability of becoming par- maternal age and offspring ADHD is due to risk factors at
ents at an earlier age.2,32,34,45–49 a family level, mainly accounted for by genetic factors, but
If the association between young maternal age and environmental factors or gene-environment interactions
ADHD is explained by a genetic predisposition, the timing might also play a role in the aetiology of ADHD. It has
of the birth in the mothers life may not be the critical factor been suggested that children with some genetic vulnerabil-
in determining offspring ADHD.35,36,50–52 If such a mater- ity to develop ADHD can be more susceptible to environ-
nal disposition to initiate childbearing at a younger age ex- mental risk factors, for example adversities in the perinatal
plained the observed association in the population-based environment, exposure to alcohol and cannabis, and ad-
methodological strength, we controlled for time trends the other siblings within the family. Parental perceptions,
including the increase in ADHD diagnosis during the study experience and behaviour may impact on the associations
period. of interest at the level of an individual child or family, but
Although the Danish register contains complete infor- these factors may be difficult or impossible to measure and
mation for inpatient as well as outpatient contacts since control for. Also, birth order might play an important role
1995 in the public health system, some children are unrec- when comparing the risk of ADHD among siblings. It is
ognized and thus untreated or treated without contact with possible that a diagnosis given to a child will make it more
the hospital system. Accordingly, ADHD is undiagnosed in likely that an undiagnosed sibling with the disorder will be
the hospital system and we have false-negatives, especially diagnosed and probably also lower the threshold for diag-
for the oldest children in the cohort. Among siblings, false- nosing another. In the sibling design, this would generate
negatives might therefore be more likely among the oldest fewer discordant sibling pairs attenuating the estimates.
Our results suggest that young parenthood could be a 14. Lichtenstein P, Carlstrom E, Rastam M, Gillberg C, Anckarsater
marker of genetic traits and, if so, delaying fertility will not H. The genetics of autism spectrum disorders and related neuro-
be fully effective in reducing offspring ADHD. Public pol- psychiatric disorders in childhood. Am J Psychiatry 2010;167:
1357–63.
icy initiatives should still be targeted at attention to young
15. Thapar A, Langley K, Fowler T et al. Catechol O-methyltransfer-
parents to support parenting as well as guidance on healthy
ase gene variant and birth weight predict early-onset antisocial
lifestyle during pregnancy among young mothers. behavior in children with attention-deficit/hyperactivity dis-
order. Arch Gen Psychiatry 2005;62:1275–78.
16. Buss C, Entringer S, Wadhwa PD. Fetal programming of brain
Funding development: intrauterine stress and susceptibility to psycho-
Susanne Hvolgaard Mikkelsen was supported by a full PhD fellow- pathology. Sci Signal 2012;5:7.
ship at Aarhus University, Denmark. 17. Sandman CA, Davis EP, Buss C, Glynn LM. Prenatal program-
normal premenopausal women. J Clin Endocrinol Metab 49. Donahue KL, Lichtenstein P, Lundstrom S et al. Childhood be-
1995;80:1429–30. havior problems and adolescent sexual risk behavior: familial
31. Fraser AM, Brockert JE, Ward RH. Association of young mater- confounding in the child and adolescent twin study in Sweden
nal age with adverse reproductive outcomes. N Engl J Med (CATSS). J Adolesc Health 2013;52:606–12.
1995;332:1113–17. 50. Geronimus AT, Korenman S, Hillemeier MM. Does young ma-
32. Lehti V, Niemela S, Heinze M et al. Childhood predictors of ternal age adversely affect child development? Evidence from
becoming a teenage mother among Finnish girls. Acta Obstet cousin comparisons in the United States. Popul Dev Rev
Gynecol Scand 2012;91:1319–25. 1994;20:585–609.
33. Jaffee S, Caspi A, Moffitt TE, Belsky J, Silva P. Why are children 51. D’Onofrio BM, Goodnight JA, Van Hulle CA et al. Maternal
born to teen mothers at risk for adverse outcomes in young age at childbirth and offspring disruptive behaviors: testing the
adulthood? Results from a 20-year longitudinal study. Dev causal hypothesis. J Child Psychol Psychiatry 2009;50:1018–28.
Psychopathol 2001;13:377–97. 52. Myrskyla M, Fenelon A. Maternal age and offspring adult
34. Trautmann-Villalba P, Gerhold M, Laucht M, Schmidt MH.