- Email: [email protected]
Secular trends in seasonal variation in birth weight
Early Human Development 91 (2015) 361–365
Contents lists available at ScienceDirect
Early Human Development journal homepage: www.elsevier.com/locate/earlhumdev
Secular trends in seasonal variation in birth weight Camilla B. Jensen a,b,⁎, Michael Gamborg a, Kyle Raymond a, John McGrath d,e, Thorkild I.A. Sørensen a,c,f, Berit L. Heitmann a,g,h a
Institute of Preventive Medicine, Bispebjerg and Frederiksberg Hospital, The Capital Region, Copenhagen, Denmark Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Denmark d Queensland Brain Institute, The University of Queensland, Brisbane, QLD, Australia e Queensland Centre for Mental Health Research, Brisbane, QLD, Australia f MRC Integrative Epidemiology Unit, University of Bristol, Bristol, UK g The Boden Institute of Obesity, Nutrition, Exercise & Eating Disorders, University of Sydney, Australia h National Institute of Public Health, University of Southern Denmark, Denmark b c
a r t i c l e
i n f o
Article history: Received 4 February 2015 Received in revised form 24 March 2015 Accepted 26 March 2015 Available online xxxx Keywords: Birth weight Denmark Fetal Development Pregnancy Seasons Registries Vitamin D
a b s t r a c t Background: Many environmental factors have been shown to influence birth weight (BW) and one of these are season of birth. Aim: The aim of the present study was to investigate the seasonal variation in BW in Denmark during 1936–1989, and to see if the variation could be explained by sunshine exposure during pregnancy. Methods: The study population was selected from the Copenhagen School Health Records Register and included 276 339 children born between 1936 and 1989. Seasonal variation was modeled using a non-stationary sinusoidal model that allowed the underlying trend in BW and the amplitude and phase of the yearly cycles to change. Results: There was a clear seasonal pattern in BW which, however, changed gradually across the study period. The highest BWs were seen during fall (September – October) from 1936 to 1963, but a new peak gradually grew from the early 1940s during early summer (May – June) and became the highest from 1964 to 1989. The amplitude of the fall peak started at 25.5 (95%CI 24.6; 25.9) grams and gradually disappeared. The amplitude of the early summer peak gradually arose from nothing to a peak of 18.6 (95%CI 17.7; 19.6) grams in the mid 1980s where it started to decrease again. Sunshine did not explain the seasonal variation in BW. Conclusion: There was a clear seasonal pattern in BW in Denmark 1936–1989, which however changed across the study period. Throughout the study period we observed a peak in BW during the fall, but gradually, starting in the early 1940s, an additional early summer peak emerged and became the highest from 1964 and onwards. © 2015 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Both low and high birth weights (BW) are characteristics that besides being the major determinant of fetal mortality and morbidity has been shown to be associated with a wide range of adult traits and health outcomes [1–3]. Both genetic and environmental factors have been shown to influence BW [4]. Gestational age, parity, maternal age, gestational weight gain, maternal smoking, diet and life style during pregnancy, parental body size (BW, adult height and BMI) and occupation has been
Abbreviations:BW, Birth weight; BMI,BodyMass Index; CSHRR, TheCopenhagenSchool Health Records Register; UVB, Ultraviolet B. ⁎ Corresponding author at: Institute of Preventive Medicine, Frederiksberg Hospital, Nordre Fasanvej 57, Hovedvejen, Entrance 5, 2000 Frederiksberg, Denmark. Tel.: +45 38163101. E-mail address: [email protected] (C.B. Jensen).
http://dx.doi.org/10.1016/j.earlhumdev.2015.03.010 0378-3782/© 2015 Elsevier Ireland Ltd. All rights reserved.
associated with BW [4]. A better understanding of the determinants may help enabling identification of high-risk groups for both high and low BW. Season of birth has also been associated with BW, and numerous studies have shown that mean BW varies across seasons [5–12]. Identified seasonal patterns include both 1 and 2 annual peaks, and season of BW peaks and lows (nadirs) is inconsistent between sites [13]. It has been speculated that the heterogeneity in the seasonal patterns is a reflection of the difference in latitudes between the study sites [14]. Also, differences in analytical strategy might be responsible. No obvious interpretation of the seasonal variation in BW has been recognized, but several mechanisms have been proposed [6,10,13]. One of the proposed mechanisms behind seasonal variation in BW is that BW follows the seasonal fluctuations in vitamin D [9]. Vitamin D is synthesized in the skin when exposed to ultraviolet B (UVB) radiation from the sun, and the amount of synthesized vitamin D therefore depends on the season [15]. The hypothesis is that exposure to vitamin D during pregnancy ensures proper fetal development and growth,
362
C.B. Jensen et al. / Early Human Development 91 (2015) 361–365
and that vitamin D deficiency during pregnancy may lead to reduced BW as a function of alterations in the fetal development. If the majority of the pregnancy has been experience during the colder months (October to April in Denmark) the synthesis of vitamin D would have been minimal leaving the mother and the child at an increased risk of vitamin D insufficiency (assuming that they did not take vitamin D supplements). If low vitamin D is associated with lower BW, then children born during the early spring would be expected to have the lowest BWs and children born during the fall would have the highest BWs. Two previous studies have investigated seasonal variation in BW in Danish children [10,16]. None of these have looked at the seasonal pattern before birth year 1973, or across a longer time period (maximum time period 30 years (10)), and they have not investigated the potential determinants of the varying BW. The aim of the present study was to investigate the seasonal variation in BW in Denmark during a 53 year period (1936–1989). An additional aim was to see if this variation could be explained by the fluctuations in sunshine hours. 2. Methods Information on BW was collected from the Copenhagen School Health Records Register (CSHRR) which contains information on BW of all children who went to school in the Copenhagen Municipality born from 1936 to 1989. Validation of BW information in CSHRR against medical birth records has shown a high validity of the information (in preparation). At school entry the mother or father reported the BW of the child. The cohort contains 372 636 records and it is described in details elsewhere [17]. Seasonal variation was modeled using a non-stationary sinusoidal model that allowed the underlying trend in BW and the amplitude and phase of the yearly cycles to change. This type of model has previously been employed in similar analyses and is well described elsewhere [6,10,18]. In short the model is a function were BW is predicted by Birthweight ¼ AmplitudeðtimeÞ cosineðperiod time þ phaseðtimeÞÞ The phase determines where the BW peaks are during the year and the amplitude determines how big the peak is. The amplitude is the difference in mean BW between the extreme month and the yearly mean BW (mean difference (95% confidence intervals)). The phase is reported in months evaluated from R outputs of date including 95% confidence intervals. The period is set to both 6 and 12 months which allows the model to have either 1 or 2 annual peaks based on results from previous studies [10,16]. The model is non-stationary because all terms are time dependent and therefore are changeable with calendar years. Individual BWs were used to calculate monthly mean BW for each year, and data were arranged as a continuous time series of 648 consecutive months. The primary analysis consisted of a model including only BW and month and year of birth. The model was created for boys and girls separately and combined; however only the combined model is reported here since there was no difference between them when divided by gender. In secondary analyses we wanted to investigate if maternal exposure to sunshine during pregnancy was causally related to the seasonal variation in BW. From the Danish Meteorological institute information was retrieved on hours of sunshine on daily basis covering the entire time period [19]. To include sunshine hours in the model we calculated cumulative sunshine hours during pregnancy and 1st, 2nd and 3rd trimester by calculating backwards from the date of birth. We adjusted BW for sunshine hours by a linear regression model with BW as the dependent variable and sunshine hours as the predicting variable. The residuals were entered in the non-stationary seasonal model similarly as BW was in the initial analyses. We
performed analyses adjusted for cumulative sunshine hours in pregnancy and in 1st, 2nd and 3rd trimester, respectively. The seasonal pattern, amplitudes and peak and low months from the adjusted models were compared to the ones from the unadjusted model manually. Analyses were performed using R [20] (Season, 2014) [21] and STATA [22]. 3. Results We identified 326 520 children born during 1936 to 1989. The total number of children included in the study was 276 339 (50.77 % boys) after exclusion of 48 313 children with missing BW information, 1 007 children with BWs below 1.5 kg, and 860 children with BWs above 5.5 kg. The number of children born per year ranged from 2 243 in 1983 to 10 697 in 1946 (Fig. 1). Mean BW in the study population was 3353.4 (SD 568) grams, and mean BW was not stable across the period. There was a decreasing trend in mean BW from 1936 until approximately 1955, where an increasing trend in mean BW took over and continued until approximately 1965. From 1965 until the beginning of the 1980s the mean BW was stable and during the 1980s the mean BW was increasing steeply. Monthly mean BW per year is presented in the top panel of Fig. 2. We observed seasonal patterns in BW with two seasonal cycles one of 12 months and one of 6 months corresponding to either one or two annual peaks (panel 2 and 3 in Fig. 2). The amplitude of the seasonal patterns varied over time for both these seasonal patterns, which indicated that neither of the seasonal patterns was stable across time. To interpret the seasonal patterns in terms of annual peak and low BW months the two seasonal cycles (6 and 12 month periodicities) were combined. A presentation of peak and low months can be seen in Fig. 3 and Table 1. From 1936–1963 BWs peaked during fall (September – October), but from 1964–1989 BWs peaked during early summer (May – June) (Table 1). BWs were the lowest during early summer (May – June) from 1936–1946, but from 1947–1989 the lowest BWs were during winter (January – March) (Table 1). The amplitude decreased across the time period from a difference of 25.5 (95%CI 24.6; 25.9) grams to one of 10.3 (95%CI 8.8; 11.8) grams in the months with the highest BW compared to the mean BW (Table 1). The difference between the months with the lowest BW and the mean BW also changed across the time period but there was no clear pattern in the change (Table 1). In the initial regression analyses sunshine hours in pregnancy, 1st, 2nd or 3rd trimester explained very little of the variation in BW (R2 0.001 – 0.007). The seasonal patterns in BW were essentially the same before and after adjusting BW for sunshine exposure during pregnancy, and results were essentially similar after adjusting the analyses for cumulative sunshine hours during pregnancy, 1st, 2nd, or 3rd trimester (Fig. 4). According to our hypothesis inclusion of sunshine hours in the model would decrease the seasonal variation in BW; however, this was not the case (Fig. 4). The seasonal variation in the
10000
7500
5000
2500
0 1940
1950
1960 1970 Year of birth
1980
1990
Fig. 1. Number of observations per year included in the study.
C.B. Jensen et al. / Early Human Development 91 (2015) 361–365
363
Fig. 2. Output from the non-stationary model of seasonal variation in BW. Top panel presents the trend in mean BW (grams) per month per year in the period 1936–1989. Second and third panel present 6 month and 12 month seasonal cycles in BW (grams), respectively.
residuals from these regression models showed essentially the same pattern as the seasonal variation in BW from the analyses without sunshine (Fig. 4).
4. Discussion In the present study we identified a seasonal variation in BW that was not stable across the 53 years of study. In the first 20 years of study we observed a fall peak in BW, but gradually over time this peak diminished and was taken over by an early summer peak. The lowest BW month also changed over time from an early summer low in the first 10 years of study to a winter low in the remaining study period. Our results clearly showed that the seasonal pattern in BW was nonstationary since both the phase and amplitude changed across time. The major strength of the model that we used in this study is that it allows the seasonal fluctuations to be non-stationary, which other simpler models do not e.g. regression models with month of birth as a categorical term or a sinusoidal term. Allowing this flexibility in the modelling enables a better fit of data compared to a more simple model [10]. Our initial hypothesis was that vitamin D status during pregnancy expressed by sunshine hours was the mechanism responsible for the variation in BW. However, we found no support for this hypothesis. Two studies have previously investigated seasonal variation in BW in Danish children, however, in other populations and time periods. Wohlfart et al. 1990 [16] looked at all Danish births from 1973–1994 and found a spring peak in April. McGrath et al. 2007 [10] looked at all Danish births from 1973 to 2003 and found a spring peak in BW before 1997, just as Wohlfart et al. 1990, and a fall peak after 1997. The results from the overlapping period between these studies and our study (1973–1989) agree that the peaks are in spring or early summer, however, the previous studies reported peak month to be April while we reported May and June. We included children from the Copenhagen Municipality, only, while the two other studies included children from
all over Denmark. This might explain the minor difference that we observe. Our measure of sunshine exposure was based on calculations of gestational length from date of birth and via extractions of daily number of sunshine hours in Denmark for the study period. We had no information on the length of gestation, and also do not know how much time the women spend outdoor, how they dressed, whether they used sunscreen or if they had overseas holidays, hence we do not know if they in fact were exposed to the amount of sunshine reported by the Danish Meteorological Institute. A few other studies have looked at the association between sunshine hours during pregnancy and size at birth. Sunshine during first trimester was positively associated with birth weight in one study (p b 0.05) [23]. Another study reported a significant positive correlation between birth length and sunlight (p b 0.001) [8]. In our model of sunshine hours and BW we assumed a linear association between the two. The lack of an association might be the consequence of this assumption being erroneous. Hours of sunshine were treated equally in our dataset even though sunshine hours during the winter months cannot drive the vitamin D synthesis. The exposure definition might suffer from the lack of quality distinction between sunshine with and without UVB-radiation. We do not have any obvious explanation for the change in seasonal pattern in BW across the study period, but changes in behavior in relation to exposure to UVB light, hence in cutaneous vitamin D synthesis, might hold some of the explanation of the emergence and increase of the early peak and the later decline in amplitude of both peaks. Thus, the use of sunlamps and solaria began during the study period [24], and they were presumably used the most during the winter period, changing the seasonal variation in vitamin D synthesis. Travels to sunny destinations became available to the general population via the introduction of the charter concept and the popularity increased during the study period [24], which also could have had an impact on the seasonal variation in vitamin D synthesis and thereby on the BW. Later during the study period awareness of the damaging effect of sunshine in
364
C.B. Jensen et al. / Early Human Development 91 (2015) 361–365
40 20 0 -20 -40
40 20 0 -20 -40
1936
1936 1937 1938 1939 1940 40 20 0 -20 -40 1940 40 20 0 -20 -40 1950
40 20 0 -20 -40 1960 40 20 0 -20 -40 1970 40 20 0 -20 -40 1980
1937
1938
1939
1940
40 20 0 -20 -40
1942
1944
1946
1948
1950
1940
1942
1944
1946
1948
1950
1952
1954
1956
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
40 20 0 -20 -40
1952
1954
1956
1958
1960
1950
40 20 0 -20 -40 1960
1962
1964
1966
1968
1970 40 20 0 -20 -40 1970
1972
1974
1976
1978
1980 40 20 0 -20 -40 1980
1982
1984
1986
1988
1990
Fig. 3. Fitted monthly mean BW (grams) for each decade from 1936–1989.
relation to skin cancer risk became apparent, and people generally started using sunscreen and reducing direct sun exposure [24,25]. Overall the sun seeking behavior has not been constant during the 53 years of study and the changing seasonal pattern in BW might be a reflection of this. Other possible mechanisms behind the seasonal variation in BW than vitamin D could be temperature, humidity or rainfall [7,8,12,13], changes in food availability and nutrient deficiencies [13], or maybe infection rates, outdoor activity, or seasonal variation in gestational length. Further exploration of the seasonal variation in BW should include analyses of these factors. Some studies have suggested that there might also be a seasonal variation in twinning rate [26,27]. In general, twins have lower BWs and if a bigger proportion of children are
Fig. 4. Fitted monthly mean BW (grams) for each decade from 1936–1989 for the unadjusted model (blue) and models adjusted for cumulated sunshine hours during pregnancy (red), during 1st trimester (green), second trimester (yellow) and third trimester (grey).
born as twins during specific months, this could cause the mean BW to drop. Unfortunately, it was not possible to exclude twins from our study, but we encourage others to investigate this in future studies of seasonal variation in BW. During the study period various societal changes happened in Denmark, for example in occupational status of women [28] and in maternity leave regulations, and we speculate if these changes might have had an impact on gestational health in general and in that way on the seasonal variation in BW. Introduction of import of food from southern parts of the world was likely to have reduced the variation in food availability across the year. Obesity prevalence among younger women increased in Denmark during parts of the study period and so
Table 1 Table of peak and low BW months and estimates of the amplitude (grams). Period Fall Peak 1936-1942 1943-1963 Early summer peak 1964-1984 1985-1989
Peak month
Amplitude (95%CI)
Period
Low month
Amplitude (95%CI)
September October
25.5 (24.6; 25.9) 22.0 (20.9; 23.2)
Early summer Low 1936-1943 1944-1946
May June
−14.2 (−14.7; −13.8) −13.0 (−14.2; −11.9)
May June
18.6 (17.7; 19.6) 10.3 (8.8; 11.8)
Winter Low 1947-1969 1970-1977 1978-1988 1989
February January February March
−19.1 (−20.9; −17.4) −16.6 (−17.4; −15.9) −13.9 (−14.8; −13.0) −7.6 (−)
C.B. Jensen et al. / Early Human Development 91 (2015) 361–365
did prevalence of gestational diabetes, which might also have had an influence on the susceptibility of BW to seasonal changes [29,30]. Our study population consisted of children who went to school in Copenhagen Municipality, and therefore did not include infants and toddlers who died before reaching school age. Potential changes in the seasonality in infant mortality over the years could have influenced the seasonal variation in BW. During the 53 years of study many societal changes presumably have happened that could have had an impact on the seasonal variation in BW. Using a non-stationary model allowed the seasonal variation in BW to vary across the years and thereby to reflect changes for example caused by societal changes. The number of children included in the study varied greatly across the study period. The differing size of birth cohorts should not influence the output of the model since data was included as a time series of monthly means which implies that the larger birth cohorts are not given more weight statistically than the smaller ones. In summary, we identified a clear seasonal pattern in BW that changed across the study period. The highest BWs were seen during fall from 1936 to 1963, but during early summer from 1964 to 1989. Sunshine exposure during pregnancy did not seem to explain the variations in BW, and hence, we do not have the evidence to support our hypothesis that sunshine-related exposures for example vitamin D underpinned the seasonal variation in BW.
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
Acknowledgements [16]
We would like to acknowledge Adrian Barnett for his support regarding the use of the R package Season. The present study was funded by the Danish Agency for Science Technology and Innovation, the Ministry of Science, Innovation and Higher Education, under the instruments “Strategic Research Projects” and by a research grant from the Danish PhD School of Molecular Metabolism funded by the Novo Nordisk Foundation. The supporting bodies had no role in the design, implementation, analysis and interpretation of the data presented herein. The authors contributions are as follows: BLH, TIAS, CBJ conceived the research idea; All authors designed the research and wrote the paper; CBJ performed the statistical analyses and had the primary responsibility for the final content. The CSHRR was built in collaboration between the Institute of Preventive Medicine and the Copenhagen City Archives. Access and linkage permission was obtained from the Danish Data Protection Agency (J. No. 2012-41-1156). This type of research based on routinely collected data do not require ethical permission in Denmark. The authors declare that they have no conflicts of interest.
[17]
[18]
[19]
[20]
[21] [22] [23]
[24]
[25]
References [1] Black SE, Devereux PJ, Salvanes KG. From the cradle to the labor market? The effect of birth weight on adult outcomes. Q J Econ 2007;122:409–39 [February]. [2] Risnes KR, Vatten LJ, Baker JL, Jameson K, Sovio U, Kajantie E, et al. Birthweight and mortality in adulthood: a systematic review and meta-analysis. Int J Epidemiol Jun 2011;40(3):647–61 [Internet, cited 2014 May 6, Available from: http://www.ncbi. nlm.nih.gov/pubmed/21324938]. [3] Rogers I. The influence of birthweight and intrauterine environment on adiposity and fat distribution in later life. Int J Obes Relat Metab Disord Jul 18 2003;27(7): 755–77. http://dx.doi.org/10.1038/sj.ijo.0802316 [Internet,cited 2014 May 11]. [4] Xue F, Willett WC, Rosner B a, Forman MR, Michels KB. Parental characteristics as predictors of birthweight. Hum Reprod Jan 2008;23(1):168–77 [Internet, Available from: http://www.ncbi.nlm.nih.gov/pubmed/17934185]. [5] Torche F, Corvalan A. Seasonality of birth weight in Chile: environmental and socioeconomic factors. Ann Epidemiol Dec 2010;20(11):818–26 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/20933189]. [6] McGrath JJ, Barnett AG, Eyles DW. The association between birth weight, season of birth and latitude. Ann Hum Biol 2005;32(5):547–59 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/16316912]. [7] Flouris AD, Spiropoulos Y, Sakellariou GJ, Koutedakis Y. Effect of seasonal programming on fetal development and longevity: links with environmental temperature.
[26]
[27]
[28]
[29]
[30]
365
Am J Hum Biol [Internet]. [cited 2014 Mar 31];21(2):214–6. Available from: http:// www.ncbi.nlm.nih.gov/pubmed/19194861. Siniarska A, Kozieł S. Association of birth weight and length with air temperature, sunlight, humidity and rainfall in the city of Warsaw, Poland. Homo Oct 2010; 61(5):373–80 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm. nih.gov/pubmed/20813365]. McGrath JJ, Keeping D, Saha S, Chant DC, Lieberman DE, O’Callaghan MJ. Seasonal fluctuations in birth weight and neonatal limb length; does prenatal vitamin D influence neonatal size and shape? Early Hum Dev Jul 2005;81(7):609–18 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/15972254]. McGrath J, Barnett A, Eyles D, Burne T, Pedersen CB, Mortensen PB. The impact of nonlinear exposure-risk relationships on seasonal time-series data: modelling Danish neonatal birth anthropometric data. BMC Med Res Methodol Jan 2007;7:45 [Internet, cited 2014 Mar 31, Available from: http://www.pubmedcentral.nih.gov/ articlerender.fcgi?artid=2151954&tool=pmcentrez&rendertype=abstract]. Chodick G, Shalev V, Goren I, Inskip PD. Seasonality in birth weight in Israel: new evidence suggests several global patterns and different etiologies. Ann Epidemiol Jun 2007;17(6):440–6 [Internet, cited 2014 Mar 27, Available from: http://www. ncbi.nlm.nih.gov/pubmed/17300954]. Murray LJ, O’Reilly DP, Betts N, Patterson CC, Davey Smith G, Evans AE. Season and outdoor ambient temperature: effects on birth weight. Obstet Gynecol Dec 2000; 96(5 Pt 1):689–95 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi. nlm.nih.gov/pubmed/11042302]. Chodick G, Flash S, Deoitch Y, Shalev V. Seasonality in birth weight: review of global patterns and potential causes. Hum Biol Aug 2009;81(4):463–77 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/20067370]. Matsuda S, Hiroshige Y, Furuta M, Doi T, Sone T, Kahyo H. Geographic differences in seasonal variation of mean birth weight in Japan. Hum Biol Aug 1995;67(4):641–56 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/ 7649536]. Holick MF. Vitamin D: a d-lightful solution for health. J Invest Med Aug 2011;59(6): 872–80 [Internet, cited 2014 Mar 20, Available from: http://www.pubmedcentral. nih.gov/articlerender.fcgi?artid=3738435&tool=pmcentrez&rendertype= abstract]. Wohlfahrt J, Melbye M, Christens P, Andersen AM, Hjalgrim H. Secular and seasonal variation of length and weight at birth. Lancet 1990;352(9145) [Internet, Jan, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/9872258]. Baker JL, Olsen LW, Andersen I, Pearson S, Hansen B, Sørensen TI. Cohort profile: the Copenhagen School Health Records Register. Int J Epidemiol Jun 2009;38(3):656–62 [Internet, cited 2014 Mar 20, Available from: http://www.pubmedcentral.nih.gov/ articlerender.fcgi?artid=2722813&tool=pmcentrez&rendertype=abstract]. Barnett AG, Dobson AJ. Estimating trends and seasonality in coronary heart disease. Stat Med Dec 30 2004;23(22):3505–23 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/15505891]. Cappelen J. Denmark – DMI Historical Data Collection 1768–2011 – with Danish Abstracts. DMI Technical Report 12–02. Copenhagen: Danish Meteorological Institute; 2012. R Development Core Team. R: A language and environment for statistical computing [Internet]. Vienna, Austria: R Foundation for Statistical Computing; 2008 [Available from: http://www.r-project.org]. Barnett A, Baker P. Package “season” documentation [Internet]. [cited 2014 Oct 16]. Available from http://cran.r-project.org/web/packages/season/season.pdf; 2014. StataCorp. Stata Statistical Software: Release 12. College Station, TX: StataCorp LP; 2011. Tustin K, Gross J, Hayne H. Maternal exposure to first-trimester sunshine is associated with increased birth weight in human infants. Dev Psychobiol Dec 2004;45(4): 221–30 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih. gov/pubmed/15549686]. De Vries E, Coebergh JW. Cutaneous malignant melanoma in Europe. Eur J Cancer Nov 2004;40(16):2355–66 [Internet, cited 2015 Feb 2, Available from: http:// www.ncbi.nlm.nih.gov/pubmed/15519506]. Norval M, Wulf HC. Does chronic sunscreen use reduce vitamin D production to insufficient levels? Br J Dermatol Oct 2009;161(4):732–6 [Internet, cited 2015 Feb 12, Available from: http://www.ncbi.nlm.nih.gov/pubmed/19663879]. Bonnelykke B, Søgaard J, Nielsen J. Seasonality in twin birth rates, Denmark, 1936–84. J Epidemiol Community Health Dec 1987;41(4):338–43 [Internet, cited 2014 Oct 16, Available from: http://www.pubmedcentral.nih.gov/articlerender. fcgi?artid=1052657&tool=pmcentrez&rendertype=abstract]. Fellman J, Eriksson AW. Statistical analysis of the seasonal variation in the twinning rate. Twin Res Mar 1999;2(1):22–9 [Internet, cited 2014 Oct 16, Available from: http://www.ncbi.nlm.nih.gov/pubmed/10392799]. The Equality Council. Ligestillingsrådets årsberetning, 1992 [Gender Equality Council’s Annual Report, 1992]; 1992 149–50 [Internet, Available from: http://www. kvinfo.dk/kilde.php?kilde=336#1]. Heitmann BL, Strøger U, Mikkelsen KL, Holst C, Sørensen TIA. Large heterogeneity of the obesity epidemic in Danish adults. Public Health Nutr May 2004;7(3):453–60 [Internet, cited 2015 Jan 26, Available from: http://www.ncbi.nlm.nih.gov/ pubmed/15153276]. Hunt KJ, Schuller KL. The increasing prevalence of diabetes in pregnancy. Obstet Gynecol Clin North Am Jun 2007;34(2):173–99 [Internet, cited 2015 Feb 2, vii. Available from: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid= 2043158&tool=pmcentrez&rendertype=abstract].
Contents lists available at ScienceDirect
Early Human Development journal homepage: www.elsevier.com/locate/earlhumdev
Secular trends in seasonal variation in birth weight Camilla B. Jensen a,b,⁎, Michael Gamborg a, Kyle Raymond a, John McGrath d,e, Thorkild I.A. Sørensen a,c,f, Berit L. Heitmann a,g,h a
Institute of Preventive Medicine, Bispebjerg and Frederiksberg Hospital, The Capital Region, Copenhagen, Denmark Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Denmark d Queensland Brain Institute, The University of Queensland, Brisbane, QLD, Australia e Queensland Centre for Mental Health Research, Brisbane, QLD, Australia f MRC Integrative Epidemiology Unit, University of Bristol, Bristol, UK g The Boden Institute of Obesity, Nutrition, Exercise & Eating Disorders, University of Sydney, Australia h National Institute of Public Health, University of Southern Denmark, Denmark b c
a r t i c l e
i n f o
Article history: Received 4 February 2015 Received in revised form 24 March 2015 Accepted 26 March 2015 Available online xxxx Keywords: Birth weight Denmark Fetal Development Pregnancy Seasons Registries Vitamin D
a b s t r a c t Background: Many environmental factors have been shown to influence birth weight (BW) and one of these are season of birth. Aim: The aim of the present study was to investigate the seasonal variation in BW in Denmark during 1936–1989, and to see if the variation could be explained by sunshine exposure during pregnancy. Methods: The study population was selected from the Copenhagen School Health Records Register and included 276 339 children born between 1936 and 1989. Seasonal variation was modeled using a non-stationary sinusoidal model that allowed the underlying trend in BW and the amplitude and phase of the yearly cycles to change. Results: There was a clear seasonal pattern in BW which, however, changed gradually across the study period. The highest BWs were seen during fall (September – October) from 1936 to 1963, but a new peak gradually grew from the early 1940s during early summer (May – June) and became the highest from 1964 to 1989. The amplitude of the fall peak started at 25.5 (95%CI 24.6; 25.9) grams and gradually disappeared. The amplitude of the early summer peak gradually arose from nothing to a peak of 18.6 (95%CI 17.7; 19.6) grams in the mid 1980s where it started to decrease again. Sunshine did not explain the seasonal variation in BW. Conclusion: There was a clear seasonal pattern in BW in Denmark 1936–1989, which however changed across the study period. Throughout the study period we observed a peak in BW during the fall, but gradually, starting in the early 1940s, an additional early summer peak emerged and became the highest from 1964 and onwards. © 2015 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Both low and high birth weights (BW) are characteristics that besides being the major determinant of fetal mortality and morbidity has been shown to be associated with a wide range of adult traits and health outcomes [1–3]. Both genetic and environmental factors have been shown to influence BW [4]. Gestational age, parity, maternal age, gestational weight gain, maternal smoking, diet and life style during pregnancy, parental body size (BW, adult height and BMI) and occupation has been
Abbreviations:BW, Birth weight; BMI,BodyMass Index; CSHRR, TheCopenhagenSchool Health Records Register; UVB, Ultraviolet B. ⁎ Corresponding author at: Institute of Preventive Medicine, Frederiksberg Hospital, Nordre Fasanvej 57, Hovedvejen, Entrance 5, 2000 Frederiksberg, Denmark. Tel.: +45 38163101. E-mail address: [email protected] (C.B. Jensen).
http://dx.doi.org/10.1016/j.earlhumdev.2015.03.010 0378-3782/© 2015 Elsevier Ireland Ltd. All rights reserved.
associated with BW [4]. A better understanding of the determinants may help enabling identification of high-risk groups for both high and low BW. Season of birth has also been associated with BW, and numerous studies have shown that mean BW varies across seasons [5–12]. Identified seasonal patterns include both 1 and 2 annual peaks, and season of BW peaks and lows (nadirs) is inconsistent between sites [13]. It has been speculated that the heterogeneity in the seasonal patterns is a reflection of the difference in latitudes between the study sites [14]. Also, differences in analytical strategy might be responsible. No obvious interpretation of the seasonal variation in BW has been recognized, but several mechanisms have been proposed [6,10,13]. One of the proposed mechanisms behind seasonal variation in BW is that BW follows the seasonal fluctuations in vitamin D [9]. Vitamin D is synthesized in the skin when exposed to ultraviolet B (UVB) radiation from the sun, and the amount of synthesized vitamin D therefore depends on the season [15]. The hypothesis is that exposure to vitamin D during pregnancy ensures proper fetal development and growth,
362
C.B. Jensen et al. / Early Human Development 91 (2015) 361–365
and that vitamin D deficiency during pregnancy may lead to reduced BW as a function of alterations in the fetal development. If the majority of the pregnancy has been experience during the colder months (October to April in Denmark) the synthesis of vitamin D would have been minimal leaving the mother and the child at an increased risk of vitamin D insufficiency (assuming that they did not take vitamin D supplements). If low vitamin D is associated with lower BW, then children born during the early spring would be expected to have the lowest BWs and children born during the fall would have the highest BWs. Two previous studies have investigated seasonal variation in BW in Danish children [10,16]. None of these have looked at the seasonal pattern before birth year 1973, or across a longer time period (maximum time period 30 years (10)), and they have not investigated the potential determinants of the varying BW. The aim of the present study was to investigate the seasonal variation in BW in Denmark during a 53 year period (1936–1989). An additional aim was to see if this variation could be explained by the fluctuations in sunshine hours. 2. Methods Information on BW was collected from the Copenhagen School Health Records Register (CSHRR) which contains information on BW of all children who went to school in the Copenhagen Municipality born from 1936 to 1989. Validation of BW information in CSHRR against medical birth records has shown a high validity of the information (in preparation). At school entry the mother or father reported the BW of the child. The cohort contains 372 636 records and it is described in details elsewhere [17]. Seasonal variation was modeled using a non-stationary sinusoidal model that allowed the underlying trend in BW and the amplitude and phase of the yearly cycles to change. This type of model has previously been employed in similar analyses and is well described elsewhere [6,10,18]. In short the model is a function were BW is predicted by Birthweight ¼ AmplitudeðtimeÞ cosineðperiod time þ phaseðtimeÞÞ The phase determines where the BW peaks are during the year and the amplitude determines how big the peak is. The amplitude is the difference in mean BW between the extreme month and the yearly mean BW (mean difference (95% confidence intervals)). The phase is reported in months evaluated from R outputs of date including 95% confidence intervals. The period is set to both 6 and 12 months which allows the model to have either 1 or 2 annual peaks based on results from previous studies [10,16]. The model is non-stationary because all terms are time dependent and therefore are changeable with calendar years. Individual BWs were used to calculate monthly mean BW for each year, and data were arranged as a continuous time series of 648 consecutive months. The primary analysis consisted of a model including only BW and month and year of birth. The model was created for boys and girls separately and combined; however only the combined model is reported here since there was no difference between them when divided by gender. In secondary analyses we wanted to investigate if maternal exposure to sunshine during pregnancy was causally related to the seasonal variation in BW. From the Danish Meteorological institute information was retrieved on hours of sunshine on daily basis covering the entire time period [19]. To include sunshine hours in the model we calculated cumulative sunshine hours during pregnancy and 1st, 2nd and 3rd trimester by calculating backwards from the date of birth. We adjusted BW for sunshine hours by a linear regression model with BW as the dependent variable and sunshine hours as the predicting variable. The residuals were entered in the non-stationary seasonal model similarly as BW was in the initial analyses. We
performed analyses adjusted for cumulative sunshine hours in pregnancy and in 1st, 2nd and 3rd trimester, respectively. The seasonal pattern, amplitudes and peak and low months from the adjusted models were compared to the ones from the unadjusted model manually. Analyses were performed using R [20] (Season, 2014) [21] and STATA [22]. 3. Results We identified 326 520 children born during 1936 to 1989. The total number of children included in the study was 276 339 (50.77 % boys) after exclusion of 48 313 children with missing BW information, 1 007 children with BWs below 1.5 kg, and 860 children with BWs above 5.5 kg. The number of children born per year ranged from 2 243 in 1983 to 10 697 in 1946 (Fig. 1). Mean BW in the study population was 3353.4 (SD 568) grams, and mean BW was not stable across the period. There was a decreasing trend in mean BW from 1936 until approximately 1955, where an increasing trend in mean BW took over and continued until approximately 1965. From 1965 until the beginning of the 1980s the mean BW was stable and during the 1980s the mean BW was increasing steeply. Monthly mean BW per year is presented in the top panel of Fig. 2. We observed seasonal patterns in BW with two seasonal cycles one of 12 months and one of 6 months corresponding to either one or two annual peaks (panel 2 and 3 in Fig. 2). The amplitude of the seasonal patterns varied over time for both these seasonal patterns, which indicated that neither of the seasonal patterns was stable across time. To interpret the seasonal patterns in terms of annual peak and low BW months the two seasonal cycles (6 and 12 month periodicities) were combined. A presentation of peak and low months can be seen in Fig. 3 and Table 1. From 1936–1963 BWs peaked during fall (September – October), but from 1964–1989 BWs peaked during early summer (May – June) (Table 1). BWs were the lowest during early summer (May – June) from 1936–1946, but from 1947–1989 the lowest BWs were during winter (January – March) (Table 1). The amplitude decreased across the time period from a difference of 25.5 (95%CI 24.6; 25.9) grams to one of 10.3 (95%CI 8.8; 11.8) grams in the months with the highest BW compared to the mean BW (Table 1). The difference between the months with the lowest BW and the mean BW also changed across the time period but there was no clear pattern in the change (Table 1). In the initial regression analyses sunshine hours in pregnancy, 1st, 2nd or 3rd trimester explained very little of the variation in BW (R2 0.001 – 0.007). The seasonal patterns in BW were essentially the same before and after adjusting BW for sunshine exposure during pregnancy, and results were essentially similar after adjusting the analyses for cumulative sunshine hours during pregnancy, 1st, 2nd, or 3rd trimester (Fig. 4). According to our hypothesis inclusion of sunshine hours in the model would decrease the seasonal variation in BW; however, this was not the case (Fig. 4). The seasonal variation in the
10000
7500
5000
2500
0 1940
1950
1960 1970 Year of birth
1980
1990
Fig. 1. Number of observations per year included in the study.
C.B. Jensen et al. / Early Human Development 91 (2015) 361–365
363
Fig. 2. Output from the non-stationary model of seasonal variation in BW. Top panel presents the trend in mean BW (grams) per month per year in the period 1936–1989. Second and third panel present 6 month and 12 month seasonal cycles in BW (grams), respectively.
residuals from these regression models showed essentially the same pattern as the seasonal variation in BW from the analyses without sunshine (Fig. 4).
4. Discussion In the present study we identified a seasonal variation in BW that was not stable across the 53 years of study. In the first 20 years of study we observed a fall peak in BW, but gradually over time this peak diminished and was taken over by an early summer peak. The lowest BW month also changed over time from an early summer low in the first 10 years of study to a winter low in the remaining study period. Our results clearly showed that the seasonal pattern in BW was nonstationary since both the phase and amplitude changed across time. The major strength of the model that we used in this study is that it allows the seasonal fluctuations to be non-stationary, which other simpler models do not e.g. regression models with month of birth as a categorical term or a sinusoidal term. Allowing this flexibility in the modelling enables a better fit of data compared to a more simple model [10]. Our initial hypothesis was that vitamin D status during pregnancy expressed by sunshine hours was the mechanism responsible for the variation in BW. However, we found no support for this hypothesis. Two studies have previously investigated seasonal variation in BW in Danish children, however, in other populations and time periods. Wohlfart et al. 1990 [16] looked at all Danish births from 1973–1994 and found a spring peak in April. McGrath et al. 2007 [10] looked at all Danish births from 1973 to 2003 and found a spring peak in BW before 1997, just as Wohlfart et al. 1990, and a fall peak after 1997. The results from the overlapping period between these studies and our study (1973–1989) agree that the peaks are in spring or early summer, however, the previous studies reported peak month to be April while we reported May and June. We included children from the Copenhagen Municipality, only, while the two other studies included children from
all over Denmark. This might explain the minor difference that we observe. Our measure of sunshine exposure was based on calculations of gestational length from date of birth and via extractions of daily number of sunshine hours in Denmark for the study period. We had no information on the length of gestation, and also do not know how much time the women spend outdoor, how they dressed, whether they used sunscreen or if they had overseas holidays, hence we do not know if they in fact were exposed to the amount of sunshine reported by the Danish Meteorological Institute. A few other studies have looked at the association between sunshine hours during pregnancy and size at birth. Sunshine during first trimester was positively associated with birth weight in one study (p b 0.05) [23]. Another study reported a significant positive correlation between birth length and sunlight (p b 0.001) [8]. In our model of sunshine hours and BW we assumed a linear association between the two. The lack of an association might be the consequence of this assumption being erroneous. Hours of sunshine were treated equally in our dataset even though sunshine hours during the winter months cannot drive the vitamin D synthesis. The exposure definition might suffer from the lack of quality distinction between sunshine with and without UVB-radiation. We do not have any obvious explanation for the change in seasonal pattern in BW across the study period, but changes in behavior in relation to exposure to UVB light, hence in cutaneous vitamin D synthesis, might hold some of the explanation of the emergence and increase of the early peak and the later decline in amplitude of both peaks. Thus, the use of sunlamps and solaria began during the study period [24], and they were presumably used the most during the winter period, changing the seasonal variation in vitamin D synthesis. Travels to sunny destinations became available to the general population via the introduction of the charter concept and the popularity increased during the study period [24], which also could have had an impact on the seasonal variation in vitamin D synthesis and thereby on the BW. Later during the study period awareness of the damaging effect of sunshine in
364
C.B. Jensen et al. / Early Human Development 91 (2015) 361–365
40 20 0 -20 -40
40 20 0 -20 -40
1936
1936 1937 1938 1939 1940 40 20 0 -20 -40 1940 40 20 0 -20 -40 1950
40 20 0 -20 -40 1960 40 20 0 -20 -40 1970 40 20 0 -20 -40 1980
1937
1938
1939
1940
40 20 0 -20 -40
1942
1944
1946
1948
1950
1940
1942
1944
1946
1948
1950
1952
1954
1956
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
40 20 0 -20 -40
1952
1954
1956
1958
1960
1950
40 20 0 -20 -40 1960
1962
1964
1966
1968
1970 40 20 0 -20 -40 1970
1972
1974
1976
1978
1980 40 20 0 -20 -40 1980
1982
1984
1986
1988
1990
Fig. 3. Fitted monthly mean BW (grams) for each decade from 1936–1989.
relation to skin cancer risk became apparent, and people generally started using sunscreen and reducing direct sun exposure [24,25]. Overall the sun seeking behavior has not been constant during the 53 years of study and the changing seasonal pattern in BW might be a reflection of this. Other possible mechanisms behind the seasonal variation in BW than vitamin D could be temperature, humidity or rainfall [7,8,12,13], changes in food availability and nutrient deficiencies [13], or maybe infection rates, outdoor activity, or seasonal variation in gestational length. Further exploration of the seasonal variation in BW should include analyses of these factors. Some studies have suggested that there might also be a seasonal variation in twinning rate [26,27]. In general, twins have lower BWs and if a bigger proportion of children are
Fig. 4. Fitted monthly mean BW (grams) for each decade from 1936–1989 for the unadjusted model (blue) and models adjusted for cumulated sunshine hours during pregnancy (red), during 1st trimester (green), second trimester (yellow) and third trimester (grey).
born as twins during specific months, this could cause the mean BW to drop. Unfortunately, it was not possible to exclude twins from our study, but we encourage others to investigate this in future studies of seasonal variation in BW. During the study period various societal changes happened in Denmark, for example in occupational status of women [28] and in maternity leave regulations, and we speculate if these changes might have had an impact on gestational health in general and in that way on the seasonal variation in BW. Introduction of import of food from southern parts of the world was likely to have reduced the variation in food availability across the year. Obesity prevalence among younger women increased in Denmark during parts of the study period and so
Table 1 Table of peak and low BW months and estimates of the amplitude (grams). Period Fall Peak 1936-1942 1943-1963 Early summer peak 1964-1984 1985-1989
Peak month
Amplitude (95%CI)
Period
Low month
Amplitude (95%CI)
September October
25.5 (24.6; 25.9) 22.0 (20.9; 23.2)
Early summer Low 1936-1943 1944-1946
May June
−14.2 (−14.7; −13.8) −13.0 (−14.2; −11.9)
May June
18.6 (17.7; 19.6) 10.3 (8.8; 11.8)
Winter Low 1947-1969 1970-1977 1978-1988 1989
February January February March
−19.1 (−20.9; −17.4) −16.6 (−17.4; −15.9) −13.9 (−14.8; −13.0) −7.6 (−)
C.B. Jensen et al. / Early Human Development 91 (2015) 361–365
did prevalence of gestational diabetes, which might also have had an influence on the susceptibility of BW to seasonal changes [29,30]. Our study population consisted of children who went to school in Copenhagen Municipality, and therefore did not include infants and toddlers who died before reaching school age. Potential changes in the seasonality in infant mortality over the years could have influenced the seasonal variation in BW. During the 53 years of study many societal changes presumably have happened that could have had an impact on the seasonal variation in BW. Using a non-stationary model allowed the seasonal variation in BW to vary across the years and thereby to reflect changes for example caused by societal changes. The number of children included in the study varied greatly across the study period. The differing size of birth cohorts should not influence the output of the model since data was included as a time series of monthly means which implies that the larger birth cohorts are not given more weight statistically than the smaller ones. In summary, we identified a clear seasonal pattern in BW that changed across the study period. The highest BWs were seen during fall from 1936 to 1963, but during early summer from 1964 to 1989. Sunshine exposure during pregnancy did not seem to explain the variations in BW, and hence, we do not have the evidence to support our hypothesis that sunshine-related exposures for example vitamin D underpinned the seasonal variation in BW.
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
Acknowledgements [16]
We would like to acknowledge Adrian Barnett for his support regarding the use of the R package Season. The present study was funded by the Danish Agency for Science Technology and Innovation, the Ministry of Science, Innovation and Higher Education, under the instruments “Strategic Research Projects” and by a research grant from the Danish PhD School of Molecular Metabolism funded by the Novo Nordisk Foundation. The supporting bodies had no role in the design, implementation, analysis and interpretation of the data presented herein. The authors contributions are as follows: BLH, TIAS, CBJ conceived the research idea; All authors designed the research and wrote the paper; CBJ performed the statistical analyses and had the primary responsibility for the final content. The CSHRR was built in collaboration between the Institute of Preventive Medicine and the Copenhagen City Archives. Access and linkage permission was obtained from the Danish Data Protection Agency (J. No. 2012-41-1156). This type of research based on routinely collected data do not require ethical permission in Denmark. The authors declare that they have no conflicts of interest.
[17]
[18]
[19]
[20]
[21] [22] [23]
[24]
[25]
References [1] Black SE, Devereux PJ, Salvanes KG. From the cradle to the labor market? The effect of birth weight on adult outcomes. Q J Econ 2007;122:409–39 [February]. [2] Risnes KR, Vatten LJ, Baker JL, Jameson K, Sovio U, Kajantie E, et al. Birthweight and mortality in adulthood: a systematic review and meta-analysis. Int J Epidemiol Jun 2011;40(3):647–61 [Internet, cited 2014 May 6, Available from: http://www.ncbi. nlm.nih.gov/pubmed/21324938]. [3] Rogers I. The influence of birthweight and intrauterine environment on adiposity and fat distribution in later life. Int J Obes Relat Metab Disord Jul 18 2003;27(7): 755–77. http://dx.doi.org/10.1038/sj.ijo.0802316 [Internet,cited 2014 May 11]. [4] Xue F, Willett WC, Rosner B a, Forman MR, Michels KB. Parental characteristics as predictors of birthweight. Hum Reprod Jan 2008;23(1):168–77 [Internet, Available from: http://www.ncbi.nlm.nih.gov/pubmed/17934185]. [5] Torche F, Corvalan A. Seasonality of birth weight in Chile: environmental and socioeconomic factors. Ann Epidemiol Dec 2010;20(11):818–26 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/20933189]. [6] McGrath JJ, Barnett AG, Eyles DW. The association between birth weight, season of birth and latitude. Ann Hum Biol 2005;32(5):547–59 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/16316912]. [7] Flouris AD, Spiropoulos Y, Sakellariou GJ, Koutedakis Y. Effect of seasonal programming on fetal development and longevity: links with environmental temperature.
[26]
[27]
[28]
[29]
[30]
365
Am J Hum Biol [Internet]. [cited 2014 Mar 31];21(2):214–6. Available from: http:// www.ncbi.nlm.nih.gov/pubmed/19194861. Siniarska A, Kozieł S. Association of birth weight and length with air temperature, sunlight, humidity and rainfall in the city of Warsaw, Poland. Homo Oct 2010; 61(5):373–80 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm. nih.gov/pubmed/20813365]. McGrath JJ, Keeping D, Saha S, Chant DC, Lieberman DE, O’Callaghan MJ. Seasonal fluctuations in birth weight and neonatal limb length; does prenatal vitamin D influence neonatal size and shape? Early Hum Dev Jul 2005;81(7):609–18 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/15972254]. McGrath J, Barnett A, Eyles D, Burne T, Pedersen CB, Mortensen PB. The impact of nonlinear exposure-risk relationships on seasonal time-series data: modelling Danish neonatal birth anthropometric data. BMC Med Res Methodol Jan 2007;7:45 [Internet, cited 2014 Mar 31, Available from: http://www.pubmedcentral.nih.gov/ articlerender.fcgi?artid=2151954&tool=pmcentrez&rendertype=abstract]. Chodick G, Shalev V, Goren I, Inskip PD. Seasonality in birth weight in Israel: new evidence suggests several global patterns and different etiologies. Ann Epidemiol Jun 2007;17(6):440–6 [Internet, cited 2014 Mar 27, Available from: http://www. ncbi.nlm.nih.gov/pubmed/17300954]. Murray LJ, O’Reilly DP, Betts N, Patterson CC, Davey Smith G, Evans AE. Season and outdoor ambient temperature: effects on birth weight. Obstet Gynecol Dec 2000; 96(5 Pt 1):689–95 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi. nlm.nih.gov/pubmed/11042302]. Chodick G, Flash S, Deoitch Y, Shalev V. Seasonality in birth weight: review of global patterns and potential causes. Hum Biol Aug 2009;81(4):463–77 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/20067370]. Matsuda S, Hiroshige Y, Furuta M, Doi T, Sone T, Kahyo H. Geographic differences in seasonal variation of mean birth weight in Japan. Hum Biol Aug 1995;67(4):641–56 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/ 7649536]. Holick MF. Vitamin D: a d-lightful solution for health. J Invest Med Aug 2011;59(6): 872–80 [Internet, cited 2014 Mar 20, Available from: http://www.pubmedcentral. nih.gov/articlerender.fcgi?artid=3738435&tool=pmcentrez&rendertype= abstract]. Wohlfahrt J, Melbye M, Christens P, Andersen AM, Hjalgrim H. Secular and seasonal variation of length and weight at birth. Lancet 1990;352(9145) [Internet, Jan, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/9872258]. Baker JL, Olsen LW, Andersen I, Pearson S, Hansen B, Sørensen TI. Cohort profile: the Copenhagen School Health Records Register. Int J Epidemiol Jun 2009;38(3):656–62 [Internet, cited 2014 Mar 20, Available from: http://www.pubmedcentral.nih.gov/ articlerender.fcgi?artid=2722813&tool=pmcentrez&rendertype=abstract]. Barnett AG, Dobson AJ. Estimating trends and seasonality in coronary heart disease. Stat Med Dec 30 2004;23(22):3505–23 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih.gov/pubmed/15505891]. Cappelen J. Denmark – DMI Historical Data Collection 1768–2011 – with Danish Abstracts. DMI Technical Report 12–02. Copenhagen: Danish Meteorological Institute; 2012. R Development Core Team. R: A language and environment for statistical computing [Internet]. Vienna, Austria: R Foundation for Statistical Computing; 2008 [Available from: http://www.r-project.org]. Barnett A, Baker P. Package “season” documentation [Internet]. [cited 2014 Oct 16]. Available from http://cran.r-project.org/web/packages/season/season.pdf; 2014. StataCorp. Stata Statistical Software: Release 12. College Station, TX: StataCorp LP; 2011. Tustin K, Gross J, Hayne H. Maternal exposure to first-trimester sunshine is associated with increased birth weight in human infants. Dev Psychobiol Dec 2004;45(4): 221–30 [Internet, cited 2014 Mar 31, Available from: http://www.ncbi.nlm.nih. gov/pubmed/15549686]. De Vries E, Coebergh JW. Cutaneous malignant melanoma in Europe. Eur J Cancer Nov 2004;40(16):2355–66 [Internet, cited 2015 Feb 2, Available from: http:// www.ncbi.nlm.nih.gov/pubmed/15519506]. Norval M, Wulf HC. Does chronic sunscreen use reduce vitamin D production to insufficient levels? Br J Dermatol Oct 2009;161(4):732–6 [Internet, cited 2015 Feb 12, Available from: http://www.ncbi.nlm.nih.gov/pubmed/19663879]. Bonnelykke B, Søgaard J, Nielsen J. Seasonality in twin birth rates, Denmark, 1936–84. J Epidemiol Community Health Dec 1987;41(4):338–43 [Internet, cited 2014 Oct 16, Available from: http://www.pubmedcentral.nih.gov/articlerender. fcgi?artid=1052657&tool=pmcentrez&rendertype=abstract]. Fellman J, Eriksson AW. Statistical analysis of the seasonal variation in the twinning rate. Twin Res Mar 1999;2(1):22–9 [Internet, cited 2014 Oct 16, Available from: http://www.ncbi.nlm.nih.gov/pubmed/10392799]. The Equality Council. Ligestillingsrådets årsberetning, 1992 [Gender Equality Council’s Annual Report, 1992]; 1992 149–50 [Internet, Available from: http://www. kvinfo.dk/kilde.php?kilde=336#1]. Heitmann BL, Strøger U, Mikkelsen KL, Holst C, Sørensen TIA. Large heterogeneity of the obesity epidemic in Danish adults. Public Health Nutr May 2004;7(3):453–60 [Internet, cited 2015 Jan 26, Available from: http://www.ncbi.nlm.nih.gov/ pubmed/15153276]. Hunt KJ, Schuller KL. The increasing prevalence of diabetes in pregnancy. Obstet Gynecol Clin North Am Jun 2007;34(2):173–99 [Internet, cited 2015 Feb 2, vii. Available from: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid= 2043158&tool=pmcentrez&rendertype=abstract].