Variation of anthropometric traits with birth season

In models adjusting for confounders, nine of 11 traits were associated with birth season. Traits that showed the most prominent variation between birth seasons were hip width (0.11 SD difference), sitting height and weight (0.09 SD), BMI (0.08 SD), shoulder width and height (0.07 SD) and thorax circumference (0.06 SD). These traits can be thus considered somewhat more sensitive to seasonal exposures during pregnancy than other studied variables, particularly leg length and handgrip strength.

The relevant exposures include variation in daylength, ambient temperatures, infectious disease burden, availability of sources of micronutrients such as fresh fruits and vegetables and outdoor activities, including workload (e.g., ^{Reference Sohn48} ). The most strictly cyclical of these is exposure to sunlight, which directly relates to the synthesis of vitamin D (e.g., ^{Reference Gaml-Sørensen, Brix and Ernst1,Reference Cashman49} ) and melatonin ^{Reference Cipolla-Neto50,Reference Banegas, Rodríguez-Artalejo, Graciani, De La Cruz and Gutiérrez-Fisac51} in expecting mothers and, in the northern temperate zone, also seasonally peaking food (including folate) availability during summer and autumn (e.g., ^{Reference Vaiserman6} ). Assuming that vitamin D availability for the developing fetus has long-term positive effects on offspring growth, our results would suggest that such effects do not occur during the first half of pregnancy. This is because children born from May to August were heavier and taller and had higher sitting height, BMI, and wider shoulders and hips than children born in all other months (Figs. 1 and 2). Mothers of such children had spent the first half of their pregnancy under conditions of short daylength where the intensity of UV radiation is too low for vitamin D-synthesis.

Instead, our findings suggest that maternal exposure to sunlight towards the end of pregnancy positively associates with offspring growth. Such interpretation is consistent with the hypothesis that vitamin D production in the third trimester of human pregnancy is critical for the development of bone mass after birth (reviewed by ^{Reference Sanguesa23,Reference Krenz-Niedbała, Puch and Kościński29,Reference Sayers and Tobias30} ). An alternative hypothesis states that because the first trimester of pregnancy is a period of rapid tissue differentiation, during which the fetus is most sensitive to stressors, maternal exposure to a relatively high amount of ultraviolet light during this period exerts a long-lasting positive influence on the organism (reviewed by ^{Reference Chmielewski and Borysławski3,Reference Krenz-Niedbała, Puch and Kościński29} ). This hypothesis was not supported because it predicts a growth penalty for children born in spring or summer because their mothers’ first trimester of pregnancy coincides with low UV exposure.

Importantly, however, there is a caveat in deriving the causal inferences from the covariation between anthropometric traits and birth season. At least in high latitudes, limited exposure to sunlight during pregnancy in autumn and winter overlaps largely with an increased rate of upper respiratory infections, particularly influenza, which has been associated with reduced birthweight and later childhood obesity ^{Reference Song52} . Additionally, it is possible that children born in spring and summer benefitted from the seasonally increasing high-quality food to their nursing mothers during the first post-natal months in autumn. Also, sufficient folate availability is considered to have the most substantial impact on fetus during the first months of pregnancy (e.g., ^{Reference Lucock27} ). Folate intake from natural sources in northern latitudes has been reported to peak in late summer and autumn (e.g., ^{Reference Hao, Ma and Stampfer53,Reference Krajcovicová-Kudlácková, Valachovicová and Blazícek54} ), so if this factor has a positive effect on children’s growth, one might expect that this effect emerges among spring and summer borns. Further, we cannot exclude the possible roles of other factors such as physical activity, maternal mental health, and overall nutrition. The summer season, with its conducive environment for outdoor activities and potential for improved mental well-being, can positively influence these factors, thereby supporting healthier fetal development and newborn size. ^{Reference Mitku, Zewotir, North, Jeena and Naidoo55,Reference Voellmin, Entringer, Moog, Wadhwa and Buss56}

Comparison with previous large-scale studies

Our results about weight/BMI are partly consistent with conclusions derived from the review of 50 studies showing that in most cases, spring births were associated with overweight/obesity in children. ^{Reference Hemati, Keikha, Riahi, Daniali, Goudarzi and Kelishadi12} Our findings align with a study of ∼500,000 Chinese men and women, showing that subjects born between March and July had the highest BMI (Lv, et al., Reference Lv, Yu and Guo2015). Interestingly, however, another large-scale study of ∼450,000 participants of the UK Biobank ^{Reference Day, Forouhi, Ong and Perry8} found no associations between birth season and adult BMI.

Regarding measures of height, our findings comply with large-scale Polish ^{Reference Chmielewski and Borysławski3} and British ^{Reference Day, Forouhi, Ong and Perry8} studies showing that individuals born in autumn and winter months were significantly shorter than their peers born around the middle of the year. A study of ∼507,000 Austrian conscripts measured at age 18 ^{Reference Weber, Prossinger and Seidler13} found that the tallest ones were born between March and June and the shortest ones between September and December, which deviates slightly from the patterns found in the current study, but still indicates that birth at sunny period of year associates with tallness advantage. A similar pattern emerged in the study of ∼48,000 Swedish conscripts. ^{Reference Kihlbom and Johansson14} Contrary to the above-cited studies, in the sample of ∼16,000 Spanish conscripts, birth at the end of summer and during the autumn was favourable for height. ^{Reference Candela-Martínez, Ramallo-Ros, Cañabate and Martínez-Carrión15} Among the ∼22,000 seven-year-old US children, those born during winter and spring months were significantly taller (and heavier) than summer/autumn-born infants. ^{Reference McGrath, Saha, Lieberman and Buka16}

Focussing on components of height, our results clearly demonstrate that the seasonal pattern was driven by the growth of the torso but not legs, as sitting height showed the highest seasonality of height measures and leg length showed none. The latter finding differs from the large-scale Chinese study, where in four of 10 geographically diverse areas, individuals born in February–August had shorter legs than those born in the other parts of the year. ^{Reference Lv, Yu and Guo2} The authors suggest that the heterogeneity of their findings over survey sites indicates that the effect of sunlight exposure (and hence D prenatal vitamin synthesis) is a site-specific phenomenon. Legs grow faster than the trunk before puberty. Therefore, leg length is considered an anthropometric marker of environmental quality in early life, which has been linked to maternal smoking during pregnancy, birth weight, breastfeeding and socioeconomic adversity in childhood (reviewed by ^{Reference Lv, Yu and Guo2,Reference Pomeroy20,Reference Hõrak and Valge41} ). In our study population, leg length was insensitive to the pathogenicity of the growth environment ^{Reference Hõrak and Valge41} but was the anthropometric trait most strongly associated with psychosocial deprivation. ^{Reference Valge, Meitern and Hõrak57} Thus, more substantial aseasonal exposures in our study likely overrode any possible effects of seasonal factors on leg length.

Head size, proxied as cranial volume or circumference, is sensitive to nutrient limitations ^{Reference Ivanovic36} and parental stimulation during growth. ^{Reference Kok37} In a US sample of ∼22,000 infants, those born in summer/autumn had larger head circumference at age seven than those born in winter/spring. ^{Reference McGrath, Saha, Lieberman and Buka16} This contrasts with our finding that spring-born children had larger crania than those born in autumn. Both studies were conducted in the northern hemisphere and involved roughly similar sample sizes; however, the children involved in the current study were, on average, six years older than those involved in the US study. One possible explanation for the contrasting findings of these studies thus is that possible birth-season effects on cranial growth vanish after the seventh year of age. Alternatively, other seasonal processes than daylength cycle might have affected the study populations in the US and Estonia differently.

Our findings on variation in cranial volume are also relevant concerning the ‘thrifty phenotype’ and ‘fetal origins of adult disease’ hypotheses, which assume that intra-uterine exposures affect health outcomes in later life. ^{Reference Fall10,Reference Hales and Barker11} Extensions to this hypothesis have proposed that at resource limitation, essential organs such as the brain, heart and lungs are protected at a cost to other tissues/body parts such as limb length and skeletal muscle (brain- and trunk-sparing concept, ^{Reference Pomeroy31,Reference Wells and Stock58} ). In a small sample of Peruvian children, ^{Reference Pomeroy20} found that in harsh conditions of rural highland, birth month associations were significant most strongly for tibia length, followed by total lower limb length and stature. The authors concluded that prenatal or early post-natal environmental effects on growth are more marked in limb lengths than trunk length or head size, consistent with the brain- and trunk-sparing concept. Results of our study (performed under milder environmental conditions) do not support this view, as cranial volume (and particularly strongly, trunk length) varied with birth season while leg length did not.

In our study, handgrip strength was not associated with birth season. Among 8,850 10–16-year-old English children, those born in October were stronger than those born in all other months except November and December. ^{Reference Sandercock21} The authors consider seasonal differences in intrauterine vitamin D concentrations as the most plausible explanation for their results. In line with this explanation, maternal serum vitamin D concentration in the third trimester of pregnancy was positively associated with offspring handgrip strength in English children at four years of age. ^{Reference Harvey24} On the other hand, the superior strength of UK children born from October to December can be explained by the ‘relative age effect’. This means that children born early in the school/sports-selection year, who are relatively older than others, are initially more likely to be involved in various sports activities due to physical and/or maturational advantages over their younger classmates. ^{Reference Sandercock21} We can only speculate that the absence of associations between birth month and handgrip strength in the current study sample might reflect the absence of the ‘relative age effect’ in our study sample.

The magnitude of the associations between anthropometric traits and birth season

Although we found that most of the traits studied varied with birth season, it should also be noted that even among the traits that showed the most prominent annual cyclicity, the variation explained by birth season was very low. For instance, the highest values of partial η²-s (for sitting height and hip width) were 0.0017—0.0012. This means that net covariates, birth season independently explained only 0.12—0.17% of variation in these traits (Supplementary Table S4). In the case of hip width, this variation was about the same magnitude as the variation explained by family SEP (0.15%), while in the case of sitting height, the variation explained by SEP (0.69%) was about six times higher than the variation explained by birth season. In the case of weight, the effect of SEP was three times stronger than the effect of birth season.

Similarly to the current study, SEP has been shown to exert a much stronger effect than a birth season on the variation of anthropometric traits in Swedish conscripts born in 1949–51. ^{Reference Kihlbom and Johansson14} Viewing our findings in the further ecological context suggests that genetic differences and/or post-natal growth conditions can cause about ten times bigger differences in anthropometric trait values than sunshine exposure at different periods of pregnancy. For instance, a previous study in the same population showed that the feet of the children from orphanages were about 0.5 SD shorter than those of children from biparental families, followed by height and cranial volume, which differed by 0.4 SD units. ^{Reference Valge, Meitern and Hõrak57}

More generally, the small birth season-related variation in anthropometric traits is consistent with most previous findings where such magnitudes have been reported. For instance, a study on UK Biobank found the biggest birth month height difference between June vs December was 3.1 mm, ^{Reference Day, Forouhi, Ong and Perry8} which is similar to the finding in Austrian recruits where height varied annually with an amplitude of ±3 mm (Weber, et al., Reference Weber, Prossinger and Seidler1998). In a large-scale Polish study, maximum height differences between birth months were 4.4 mm for men and 2.3 mm for women. ^{Reference Chmielewski and Borysławski3} Swedish recruits born in March–May were 6.5 mm taller than those born in November–December. ^{Reference Kihlbom and Johansson14} Some smaller-scale studies, however, have found more prominent height differences for height and weight. For instance, in a study of 1,241 Polish children, those born from October to March were 0.13 SD units taller and 0.14 SD units heavier than children born from April to September, ^{Reference Kościński, Krenz-Niedbała and Kozłowska-Rajewicz59} which is about twice the magnitude of the between-season difference for these traits found in the current study. In a sample of 810 Spanish men, birth in June/July was associated with a 2 cm taller stature than birth in December/January. ^{Reference Banegas, Rodríguez-Artalejo, Graciani, De La Cruz and Gutiérrez-Fisac51}

Importance of maternal occupation

A clear pattern emerged when comparing the curvatures of monthly variation in offspring traits between mothers in unskilled manual occupations vs mothers in skilled manual and non-manual occupations. We did not detect nonlinearity in associations between any of the offspring traits and birth month among children of mothers who worked in non-manual professions (Supplementary Table S8). However, among children of blue-collar mothers, nonlinear associations with birth month emerged in seven traits of 11.

This occupation-related pattern has at least two mutually non-exclusive explanations. Assuming that women in manual professions spend, on average, more of their occupational duties outdoors, one might expect that they are exposed to higher seasonal variation in UV exposure than white-collar women working indoors. If this differential pattern in UV exposure translates to higher seasonal variation in vitamin D availability for developing fetuses of mothers in manual professions, it would mean that children of such mothers are more likely to display birth-season-related variation in body dimensions.

Additionally, we consider it likely that children’s growth in high-SEP families is more strongly buffered against any external exposure (including fluctuations in fetal vitamin D and other micronutrient levels) than in low-SEP settings. For instance, children from high SEP families likely have better access to resources during the post-natal period that facilitates compensatory growth even in cases of maternal under-exposure to sunlight or micronutrient deficiency during pregnancy. Consistent with this reasoning, the previous study in our population showed that mothers in unskilled manual professions had 27% lower odds of becoming longevous (reaching the 90^th survival percentile above 92 years) than the rest of mothers. ^{Reference Valge, Meitern and Hõrak60} External factors that reduce life span had thus the greatest impact on blue-collar mothers. If these factors (such as resource limitations and exhausting physical work load) also hinder the ability of mothers to protect their developing offspring from seasonal environmental effects, then a more pronounced influence of birth season on the offspring traits of mothers in unskilled manual jobs is anticipated. Such reasoning relies on the rationale that ample resource availability enables better attainment of genetic growth and development potential than a resource-restricted environment (see ^{Reference Hõrak, Valge, Fischer, Mägi and Kaart61–Reference Turkheimer, Haley, Waldron, D’Onofrio and Gottesman63} ). For instance, Finnish mothers in manual occupations (from 1996—97 and 1997—2014) were more likely experience preterm delivery and give birth to underweight and small-for-gestational-age newborns than non-manual workers. ^{Reference Ahmed and Jaakkola64,Reference Kwegyir-Afful65}

Our result that the associations between birth month and anthropometric traits are more strongly associated in children born to low SEP mothers is in accordance with findings of the Swedish conscript study where birth-season-related patterns in height and underweight emerged only among men whose fathers were in low socioeconomic status. ^{Reference Kihlbom and Johansson14} An alternative hypothesis, namely that fetuses of women in manual professions (e.g., farmers) who are exposed to UV radiation even in winter are not at great risk of vitamin D insufficiency irrespective of their conception/birth season ^{Reference Krenz-Niedbała, Puch and Kościński29} is not supported by our data. Such an outcome is perhaps expected, given that in Estonian latitudes, the period when vitamin D synthesising irradiance is below the threshold level lasts 100–106 days from about the sixth of November to the 19^th February. ^{Reference Eerme, Veismann, Ansko and Lätt66}

Strengths, limitations and conclusions

This is the first study on the associations between birth season/month and multiple anthropometric traits in children. The main limitations are its cross-sectional design and wide age range variation of children. The main strengths are the diversity of anthropometric traits measured, the decent sample size, and the quality of information about children’s family socioeconomic position (SEP) during the measurement period. Due to historical reasons, the participation was not voluntary, which considerably reduces the recruitment selection ^{Reference Stubbs and Achat67} and that all the participants were younger than 20, which almost entirely eliminates the mortality selection in the sample. We found that nine of 11 offspring traits were associated with birth season in models adjusting for family SEP and urban vs rural origin. Seasonal patterns in anthropometric traits were generally weak, particularly compared to the effects of family SEP. Most prominent nonlinear associations between offspring traits and birth month emerged among children of mothers in unskilled manual professions. A possible explanation is that children’s growth in high-SEP families is more strongly buffered against any external exposures (including possible shortage of maternally synthesised vitamin D and essential micronutrient availability) than in low-SEP settings. For most traits, children born in spring/summer were larger than those born in autumn/winter. Hip width, trunk length and weight showed the most distinct seasonal patterns. If these birth-season-related patterns are related to maternal sunlight-induced vitamin D synthesis, our results support the view that UV exposure towards the end of pregnancy benefits offspring growth. It is also possible that children born in spring and summer benefitted from the seasonally increasing nutrient availability during the first post-natal months in autumn. Altogether, our findings support the contention of the developmental origins of health and disease hypothesis that environmental exposures during early life (particularly the in-utero period) can permanently influence adult phenotypes in later life. The biological and practical relevance of our findings, especially as concerns the relative importance of pre-and post-natal environments on offspring growth and development, however, remains to be clarified.