We found in a sample of 893 Finnish children that those born lighter in weight, thinner in BMI, shorter in stature and smaller in head circumference were rated by their parents as having higher ADHD symptoms scores at 56 months. Further, we found that a smaller head circumference throughout the period of growth from birth to 56 months was related to higher ADHD symptoms. We also found that a slower gain in BMI from 20 to 56 months and a lower relative BMI at 56 months was associated with higher ADHD symptoms. Controlling for several pre- and neonatal covariates had little effect on the results. The associations were not modified by sex and there were no non-linear associations.
We also found a similar pattern of findings in a subgroup of term (≥ 37 weeks of gestation), normal birth weight (≥ 2,500 g) children not admitted to neonatal ward (n = 293) with three clear exceptions: the association between birth length, growth in BMI from 20 to 56 months and the relative BMI at 56 months and ADHD symptoms scores were turned to non-significant.
There are several strengths in this study. We were able to examine the early childhood growth using multiple anthropometric variables in a prospective study design. Most of the existing studies have been conducted with data derived from clinical practice, which may favor a higher than expected rate of positive findings . Also studies with stimulant-naïve sample are needed to evaluate whether potential growth deficits may indeed be etiologically related to ADHD and not a consequence of stimulant-medication used to treat symptomatology . Further, we had the whole birth weight range available and could control for several variables that may have an impact on growth and/or ADHD symptoms. Controlling for covariates increases the likelihood that the findings are true and not due any confounding factor related to the both independent and dependent variables. We could also exclude children with severe conditions that could have potentially confounded the findings.
There are also some limitations. For practical reasons, several parents did not receive or fill in the ADHD questionnaire. However, those who did and did not receive/fill in the questionnaire did not differ from each other in most of the measured variables. However, compared to those lost to follow-up the participants were taller at five and at 20 months, and their mothers smoked less during pregnancy. Therefore, the results may be more characteristic of children exposed to less adverse pre- and postnatal environments. Further, we had no information on the parental ADHD symptoms and were, thus, unable to take potential genetic susceptibility into account. Neither, did we have information on environmental postnatal factors as non-optimal parenting [27, 28] and family adversity  that have been shown to be related to symptoms of ADHD. However, we have controlled for parental education, which is a crude proxy of several environmental adversities. As we have used multiple measures of body size at four different time points, we have ended up making multiple tests. However, we decided not to use the correction for multiple testing (e.g., Bonferroni correction). First, as the number of Type I errors cannot decrease without increasing type II errors . Second, theoretical assumption behind the correction of multiple testing is that all null hypotheses are true simultaneously, which was not of interest of our study. It is suggested simply describing carefully the used analyses is usually the most rational way of dealing with multiple testing .
We have previously shown with the same data that small for gestational age (SGA) status, rather than prematurity per se, may increase risk for symptoms of ADHD in young children . While current findings showing associations between smaller body size in weight, length, head circumference and BMI at birth and higher ADHD symptoms scores are in accordance with our and majority of others previous findings [5–8, 10, 11], they also extend them in significant ways. Of the few previous studies focusing on children born across the whole gestational age/birth weight range [e.g., [9, 32] to our knowledge none has focused on birth weight, length, head circumference and BMI in a same study. In our cohort, birth weight and head circumference seem to have the strongest associations with parent rated ADHD symptoms.
With regard to postnatal growth, to our knowledge there are only three prior studies focusing on postnatal head circumference and ADHD or its symptoms. Ptcak et al.  suggested that 4-16 years-old boys with diagnosed ADHD have a smaller head circumference than the normal non-clinical population, although the association was only marginally significant among those boys who did not receive medication (n = 52, p = .07). Hack et al.  found among 249 very-low-birth-weight children that those who had a subnormal head size (less than the mean -2 SD for age) at 8 months of corrected age had a higher incidence of hyperactivity at 8-9 years. This association became, however, non-significant after controlling pre- and neonatal covariates. Finally, Stathis et al.  failed to find any significant associations among 87 extremely-low-birth weight born children between subnormal head circumference (comparison to norm z-scores) or head growth velocity from birth to 2 years and parent/teacher rated ADHD diagnoses at school-age. Because of differences in age of the samples and periods of growth, direct comparison of our results showing an association between a smaller head circumference throughout the period of growth from birth to 56 months with higher levels of ADHD symptoms at 56 months, with earlier findings is not possible. The current study is the only one that presents results in a relatively large cohort of children born across the whole birth weight range.
The mechanisms explaining the association between postnatal head circumference and ADHD symptoms remain speculative. Head circumference has been shown to correlate highly with total brain volume (from 1.7 to six years r = .93) . Smaller total brain volume, in turn, has been shown to be related to ADHD in several studies . Further, our results showing that those having the highest levels of parent rated ADHD symptoms were those who had the smaller head circumferences already at birth and had a persistently smallest head circumference across the early childhood pointing, at least partly, to the compromised prenatal period the effects of which the postnatal growth has not been able to correct. The longitudinal growth curves of total brain volumes have been shown to be roughly parallel, although at different levels, between patients with ADHD and healthy controls  suggesting that developmental processes are essentially healthy in ADHD, and that ADHD symptoms appear to reflect fixed earlier neurobiological insults or abnormalities. Another study  provided evidence that the pattern of brain maturation in ADHD was delayed, rather than abnormal. As the brain volume usually reaches its maximum volume by early adolescence  the age range of the participants in the current study does not allow conclusions whether the association between smaller head circumference and higher levels of ADHD symptoms reflect slow maturation that will disappear later.
We also found that lower relative BMI at 56 months and slower gain in BMI from 20 to 56 months was related to higher ADHD symptoms. Earlier studies have shown that ADHD or its symptoms are associated with higher BMI [16, 18, 19] rather than with thinness , or have not found associations with BMI [14, 40, 41]. All previous studies have, however, studied older participants or used diagnoses of ADHD. It should also be noted that our results became non-significant after excluding the participants who were admitted to neonatal ward during first ten days, were born preterm or had a birth weight < 2,500 g suggesting the stronger relationship among those born with compromised prenatal period or early health problems. However, our findings of slower gain in BMI may also reflect the measurement period from 20 to 56 months. At that time children start to move more by themselves. They may burn a lot of energy while hyperactive, but the ADHD-related abnormal eating behaviors (e.g. binge eating) that are present in older individuals may only have a minor role on the choices and amount of food consumed see .
Earlier studies on height and ADHD have been mainly cross-sectional and inconsistent in their findings [13–16, 43]. The one existing longitudinal study focusing on growth in height among 48 ADHD diagnosed boys from two to 17 years of age found a distinct suppression of growth in height from nine to 14 years, earlier onset of adolescent growth spurt and a higher velocity of growth in that moment . That study however differed from the current study in measurement of ADHD and in age range. Our longitudinal findings that parental rated ADHD symptoms at 56-months are not related to relative height at successive time points or to growth trajectories during the first years extend the knowledge of prior studies to years before ADHD-diagnoses are usually made.