Abstract
Background:
Several early-life factors have been associated with higher risk of developing dementia. It is unclear whether season of birth (SOB) can affect cognitive aging in older adults or not.
Objective:
We aimed to study the association of SOB with the level of cognitive performance as well as with the rate of cognitive decline.
Methods:
We studied 70,203 individuals who participated in the Survey of Health, Aging and Retirement in Europe. Cognition was measured with tests on verbal fluency and immediate and delayed recall. We assessed the association of SOB with the level of cognitive performance using multiple linear regression and with the rate of cognitive decline using linear mixed-effects models.
Results:
When compared to individuals born in winter and adjusted for sociodemographic and health-related characteristics, being born in summer was associated with a higher level of delayed recall (B 0.05; 95%CI 0.01 to 0.09) and verbal fluency (B 0.15; 95%CI 0.00 to 0.29) and being born in fall with a higher level of immediate recall (B 0.04; 95%CI 0.01 to 0.08) and verbal fluency (B 0.15; 95%CI 0.01 to 0.29). Individuals born in summer had a higher yearly decline in delayed recall (B –0.005; 95%CI –0.009 to 0.000), while the scores in delayed recall in participants born in spring showed an inverse trend (B 0.005; 95%CI 0.000 to 0.010).
Conclusion:
Individuals born in winter seem to carry a life-long disadvantage in a lower level of cognitive performance; however, being born in winter does not seem to affect the rate of cognitive decline.
INTRODUCTION
In the quest to prevent cognitive disorders, there is a substantial interest to investigate the effects of early-life risk factors for low cognitive performance, greater rate of cognitive decline, or risk of dementia [1]. The emerging concept of life-course approach in dementia research, that considers factors that act during development and might influence disease onset, provides opportunities to understand the underlying mechanisms behind the brain and cognitive reserve as the timing of exposure to such risk factors may crucially alter brain development, causing vulnerability to dementia later in life [2]. As introduced in theoretical framework by Whalley, multiple causal concepts probably interact to determine life-course pathways toward dementia [3]. One of these concepts is the fetal origins of adult disease (FOAD) hypothesis, that proposes associations between a disadvantageous early fetal environment indicated by low birthweight and dementia development. FOAD can directly affect brain in embryogenesis phase, acts indirectly on brain via its connection with cardiovascular diseases or it increases rates of aging causing age-dependent disorders [3]. Further nature and timing of environmental contributions together with sensitive periods, during which an individual is at greatest risk of damage, if exposed, can set off a chain of adverse events leading to dementias [3].
There is already sufficient evidence indicating that some early-life risk factors, such as socioeconomic hardship [4], low education, head trauma, and family-related factors, such as number of siblings or parental death, have long-lasting effect on cognitive functions in later life [1]. Other early-life factors, including season of birth (SOB), rural residence, or parental re-marriage [1], have been suggested as potential risk factors for psychiatric and neurological disorders [5], but have not been sufficiently investigated in the context of cognitive aging. However, previous studies show a link between SOB and multiple aspects of human health, ranging from cancer [6] to life expectancy [7]. Within brain disorders, the most explored topic is the effect of SOB on schizophrenia [8], where those born in winter are at a 10%higher risk of its development [9]. An influence of SOB on other mental disorders, such as depression, personality disorder [10, 11], bipolar disorder [12], panic disorder [13], eating disorder [14], autism [15, 16], and risk of suicide [17, 18] has also been suggested.
One of the proposed explanations for the connection between SOB and psychiatric disorders is that babies born in winter are exposed to lower levels of vitamin D during sensitive periods of brain development, which may apply also for cognitive aging [19]. Previous studies have demonstrated that lower levels of vitamin D alter the development of brain areas related to cognitive functions, such as limbic areas or prefrontal cortex, and also the production of neurotransmitters, especially dopamine and serotonin [20–24]. Other biologically plausible hypotheses that would explain the link between SOB and cognitive functions are maternal nutrition, lower placental blood flow [25] or respiratory virus infections [8, 26–28] with influenza viruses in particular [8]. SOB therefore represents a surrogate measure for multiple biological pathways and can be more strongly associated with cognition than each individual risk.
The main goal of our study is to determine whether SOB is involved in cognitive aging. Specifically, we hypothesize that people born in the winter season would have a lower level of cognitive performance and a higher rate of cognitive decline over time. As opposed to previous studies, we had the opportunity to explore these research questions on a large and diverse population-based sample of European older adults using a broad range of cognitive measures.
In the context of the increasing prevalence of cognitive disorders due to demographic aging in Europe [29], better knowledge on the influence of the SOB could contribute to increased understanding of mechanisms underlying the development of cognitive disorders as well as help to better tailor preventive strategies to those at the highest risk.
METHODS
Source of data
Data for the analysis were used from a prospective, multidisciplinary, and multi-centric study Survey of Health, Ageing and Retirement in Europe (SHARE), that aims to study population aging across Europe [30]. Data on health, social network, and economic conditions are collected from community-dwelling individuals (aged at least 50 + years old) and their partners, irrespective of age, using computer-assisted personal interviewing (CAPI), as previously described in detail [30]. Until now, there have been already 7 waves conducted: wave 1 in 2004, followed by wave 2 in 2006/2007, wave 3 in 2008/2009, wave 4 in 2011/2012, wave 5 in 2013, wave 6 in 2015, and wave 7 in 2017. At present, 27 European countries and Israel participate in SHARE.
Cognitive functions
Cognition was tested in waves 1, 2, 4, 5, 6, and 7, using three measures: verbal fluency, immediate recall and delayed recall. Verbal fluency score, based on animal fluency test and reflecting executive functioning was the sum of acceptable animals named by participants within one minute [31]. Immediate and delayed recall, representing memory performance [32], were extracted from an adapted 10-word delay recall test [33]. The test format used in SHARE is based on the Telephone Interview of Cognitive Status-Modified [34]. Immediate recall score was the number of recalled words after the interviewer read a list of 10 words. Delayed recall score was then assessed at the end of the cognitive testing session, when participants were asked again to recall any of the word.
These tests have been proven to be sensitive measures for the assessment of cognitive function and differentiating cognitively healthy individuals from those with lower cognitive functions [31, 35–37]. In one study based on the same principle of testing immediate and delayed recall, these tests had 96%specificity and 80%sensitivity [38]. These cognitive measures have solid predictable capability as well. Based on The Canadian Study of Health and Aging, these test can accurately predict progression to dementia within 5 years for delayed recall and animal fluency (sensitivity = 75%, specificity = 74%, positive likelihood ratio = 2.90) and within 10 years for delayed recall (sensitivity = 78%, specificity = 72%, positive likelihood ratio = 2.81) [39].
We use data on cognitive functions in two ways: first, we analyze the level of cognitive performance, utilizing information on these three cognitive measures at baseline (as baseline, we consider the first wave the cognitive data was available for an individual). Second, we analyze the rate of cognitive decline for the three cognitive measures by studying their yearly change since baseline.
Season of birth
Information about the date of birth was acquired directly from participants as a part of CAPI. Our sample was then divided into 4 groups based on the month of birth: winter (individuals born in December, January, February), spring (born in March, April, May), summer (born in June, July, August) and fall (born in September, October, November).
Covariates
Covariates were chosen based on the existing literature as sociodemographic and health-related characteristics that are associated with the SOB and cognitive functions [40, 41]. Covariates were used from baseline (the wave when participants had data on cognitive functions for the first time). If a specific covariate was not available at baseline, the data was taken from the closest wave. Sociodemographic factors were age (years), sex (men versus women), birth cohort (categories per 10 years), education (categories based on the International Standard Classification of Education 1997 ISCED-97 [42]), civil status (partner versus alone), children (number), grandchildren (number), current job situation (working versus not working), household size (number), and household net worth (sum of household net financial assets and household real assets).
Health-related characteristics were smoking (ever smoked daily versus never smoked daily), alcohol use (drinking more than 2 glasses of alcohol almost every day versus drinking less), body mass index (BMI), total number of chronic diseases, depressive symptoms (assessed with EURO-D scale [43]), number of limitations in instrumental activities of daily living (IADL), maximal grip strength, and physical inactivity (never vigorous nor moderate physical activity versus ever vigorous or moderate physical activity).
Analytical sample
We restricted the analysis to individuals who had at least two measures of cognition and were older than 50 years. From 206,723 individuals that participated in SHARE, 67,167 did not complete any interview and so were excluded from the analysis. Out of remaining 139,556 participants, we then excluded 66,839 who did not have at least two measures on all cognitive tests and 2,514 individuals younger than 50 years. The final analytical sample consisted of 70,203 people (on average 64 years old, 55%women) from four European regions and Israel (Western Europe: n = 28,947; Southern Europe: n = 13,200; Scandinavia: n = 8,820; Central and Eastern Europe: n = 16,749; Israel: n = 2,487). The flowchart is presented on Supplementary Figure 1.
Statistical analysis
Statistical analysis was performed in two steps: First, we studied the association of SOB with the level of cognitive performance (cross-sectional analysis). Second, we studied the association of SOB with the rate of cognitive decline (longitudinal analysis).
Cross-sectional analysis
We present participants’ baseline characteristics in frequency (n, %), mean±standard deviation (SD) or median and interquartile range (IQR), where appropriate. Participants' characteristics were compared between four groups of SOB (those born in winter, spring, summer and autumn) using χ2 test for binary variables, one-way analysis of variance (ANOVA) for continuous variables with normal distribution, and Kruskall Wallis test for continuous variables with skewed distribution. Effect size has been calculated using Cramér’s V and Eta Squared method, where appropriate.
Next, we applied linear regression to estimate unstandardized B coefficients with 95%confidence intervals (CI) for the associations of SOB (using winter as the reference) with the baseline level of cognitive performance, group-wise adjusting for sociodemographic and health-related covariates. As there were almost no differences in models that were progressively adjusted for different groups of covariates, we present only two models: Model 1, which was adjusted for age, sex, birth cohort and region, and Model 2, which was also adjusted for sociodemographic and health related variables.
Longitudinal analysis
To study the relation between SOB and the rate of cognitive decline, we used linear mixed-effects models as they are appropriate for unbalanced datasets. We set the between-participant variability as a random intercept at the participant level and random slope on time in years since baseline at participant level. SOB (dummy-coded), time in years, time squared, two-way interaction between SOB and time term (SOB×years) and covariates (Model 1 and Model 2, as described above) were included as fixed effects.
In addition, we controlled for practice effect in both models as familiarity with the content of the cognitive tests might lead to an underestimation of the rate of cognitive decline [44]. We tested three sets of models adjusting for practice effect defined as 1) wave 1 set as indicator, 2) number of prior tests, and 3) root square of the number of prior tests in order to select the most suitable indicator of practice effect [45]. The number of prior tests was the best fit to our data and is included in the presented models.
Secondary analyses
We performed several sets of secondary analyses to assess effect modification by sex and region; in other words, to investigate, whether the association of SOB with the level of cognitive performance/the rate of cognitive decline differs for men and women and individuals residing in different regions. We excluded participants from Israel in order to focus on differences across European regions. In cross-sectional analysis, we first included a two-way interaction term SOB×sex, then SOB×region and in the end a three-way interaction term SOB×sex×region. Similarly, in longitudinal analysis, we first tested SOB×years×sex, then SOB×years×region and in the end SOB×years×sex×region. All interactions were tested in Model 1 and likelihood ratio (LR) test assessed the interaction effect. We performed stratified analyses, where appropriate. The analyses were conducted using IBM SPSS Statistics Version 21 and Stata version 16.1. p value < 0.05 was used to indicate statistical significance.
Standard protocols, approvals, and participants’ consent
This paper uses data from SHARE Waves 1, 2, 4, 5, 6, and 7 see elsewhere for details [30] (DOIs: 10.6103/SHARE.w1.600, 10.6103/SHARE.w2.600, 10.6103/SHARE.w4.600, 10.6103/SHARE.w5.600, 10.6103/SHARE.w6.600, 10.6103/SHARE.w7.600). SHARE has been repeatedly reviewed and approved by the Ethics Committee of the University of Mannheim. All participants provided a written consent. Their data were pseudo-anonymized, and they have been informed about the storage and use of the data and their right to withdraw consent.
Data availability statement
Access to the SHARE data is provided free of charge on the basis of a release policy that gives quick and convenient access to all scientific users world-wide after individual registration. All details about the application and registration process can be found on this website: http://www.share-project.org. The study protocol and syntax of the statistical analysis will be shared upon request from the corresponding author of this study.
RESULTS
Description of participants
The analytical sample consisted of 70,203 individuals (on average 64 years old, 55%women) and was divided into four groups based on SOB as follows: winter (n = 17,742; 25.3%), spring (n = 18,616; 26.5%), summer (n = 17,138; 24.4%), and fall (n = 16,707; 23.8%). Characteristics of participants are presented in Table 1. The four groups differed in all three measures of cognitive functions (p = 0.01 for all measures). Individuals born in winter had the lowest scores in all cognitive measures: immediate recall (mean 5.14±SD 1.78), delayed recall (mean 3.70±SD 2.07), and verbal fluency (mean 20.00±SD 7.62). Those born in summer scored the highest in delayed recall (mean 3.77±SD 2.08) and verbal fluency (mean 20.25±SD 7.59) and individuals born in fall scored the highest in immediate recall (mean 5.20 SD±1.76). Except for cognitive functions, almost no differences in sociodemographic and health-related variables were found between the four groups. The only exceptions were slight differences in age (p < 0.001), current job situation (p = 0.01), and BMI (p = 0.04).
Baseline characteristics of participants
IQR, interquartile range; SD, standard deviation; IADL, instrumental activity of daily living. Effect size has been calculated using Cramér’s V and Eta Squared method where appropriate.
Cross-sectional analysis
Multivariable analysis showed that when compared to those born in winter, being born in summer and fall was associated with a higher level of cognitive performance, independently of all covariates (Table 2). Specifically, after full adjustment in Model 2, being born in summer was associated with higher scores in delayed recall (B 0.05; 95%CI 0.01 to 0.09) and verbal fluency (B 0.15; 95%CI 0.00 to 0.29), relative to winter. Being born in fall was related to a higher level of performance in immediate recall (B 0.04; 95%CI 0.01 to 0.08) and verbal fluency (B 0.15; 95%CI 0.01 to 0.29). There were no differences between those born in spring and winter.
Associations of season of birth with the level of cognitive performance
***p < 0.001; **p < 0.01; *p < 0.05; B (95%CI), unstandardized B coefficients with 95%confidence intervals. Model 1: age, sex, region, birth cohort. Model 2: age, sex, region, birth cohort, education, civil status, children, grandchildren, current job situation, household size, household net worth, smoking, alcohol use, BMI, number of chronic diseases, depressive symptoms, IADL, maximal grip strength, physical inactivity.
Longitudinal analysis
Individuals born in summer had a higher yearly decline in delayed recall in both models, as seen in the interaction term summer×years in Table 3 (B –0.005; 95%CI –0.009 to 0.000). On the other hand, the scores in delayed recall in participants born in spring showed an increasing trend (B 0.005; 95%CI 0.000 to 0.010).
Association of season of birth with the rate of cognitive decline
***p < 0.001; **p < 0.01; *p < 0.05; B (95%CI), unstandardized B coefficients with 95%confidence intervals. Model 1: age, sex, region, birth cohort, practice effect. Model 2: age, sex, region, birth cohort, practice effect, education, civil status, children, grandchildren, current job situation, household size, household net worth, smoking, alcohol use, BMI, number of chronic diseases, depressive symptoms, IADL, maximal grip strength, physical inactivity.
Secondary analyses
In cross-sectional analysis, there was no effect modification by sex (immediate recall: p from LR test 0.80; delayed recall: p = 0.55; verbal fluency: p = 0.74) and region alone (immediate recall: p = 0.34; delayed recall: p = 0.06; verbal fluency: p = 0.16). However, the three-way interaction was significant (immediate recall: p < 0.001; delayed recall: p < 0.001; verbal fluency: p < 0.001), therefore, we conducted analyses stratified by sex as well as region (Supplementary Table 1).
In longitudinal analysis, there was no effect modification by sex alone (LR test: verbal fluency p = 0.898, immediate recall p = 0.750, delayed recall p = 0.301). Region alone was found to be a significant effect modifier (LR test: verbal fluency p < 0.001, immediate recall p < 0.001, delayed recall p < 0.001), therefore the analyses were stratified by region (Supplementary Table 2). When we included a four-way interaction term (sex×region×SOB×years), the interaction was significant (p from LR test < 0.001), which allowed us to stratify by sex and region (Supplementary Table 3). However, results of the secondary analyses do no suggest either sex or any region to have a consistently higher or lower level of cognitive performance/rate of cognitive decline associated with any SOB.
DISCUSSION
In the present study, capitalizing on a well-characterized sample of more than 70,000 middle-aged and older individuals, we observed that individuals born in winter have a lower level of cognitive performance later in life, compared to those born in summer and fall. These associations were independent of all sociodemographic and health-related covariates. Findings related to the rate of cognitive decline are less clear. Our study does not suggest that being born in winter is associated with a greater rate of cognitive decline. On the contrary, we observed a greater decline in delayed recall in individuals born in summer.
Although the influence of SOB on the origin and development of diseases is widely studied, there is no clear explanation for the influence of the SOB on cognitive functions in older adults. Other authors have focused mainly on the influence of SOB on cognitive development in childhood and young adulthood. From this point of view, our study is consistent with research of Pintner and Forlan [46], who discovered that those born in winter have lower intellectual abilities. Surprisingly, in contrast to these results, studies focusing on dementia do not suggest an effect of being born in winter at all [47, 48] or have completely opposite results. For example, according to Tolppanen et al., in a sample of more than three hundred thousand people residing in Finland, those born in summer were slightly more likely to develop Alzheimer’s disease than those born in the winter [49]. The same conclusion was reached by Doblhammer based on a study with German citizens [41] and by Ding based on a sample from China [50].
It is important to note that these studies differ in several aspects. First, they vary in sample size, ranging from dozens to thousands of participants. Second, most of the studies were designed as case-control studies. Finally, the outcome measure was a clinical diagnosis of dementia, but different diagnostic methods were used. Taking these contradicting studies into account, we suggest that being born in winter may negatively influence the development of cognitive functions with long-lasting effects during the life-course but does not influence the transition into cognitive impairment in old age itself.
The best documented negative effect of being born in winter on the human body is in the case of schizophrenia [12, 51]. This association is primarily explained by climate influences, either in the first months of life [46], within a sensitive period at birth or within intrauterine development in the first months [52]. Winter months are associated with lower levels of UV radiation, which is directly associated with lower levels of vitamin D [26]. This is linked to multiple disorders, such as rickets, multiple sclerosis, heart disease, autism, or cancer [53]. In the context of brain disorders, low levels of vitamin D have been shown in animal models to have a negative effect on neuronal development, both at the level of brain development and at reduced levels of important signaling molecules [20, 21]. Low levels of vitamin D around the time of birth limit its neuromodulatory and anti-inflammatory functions [54]. Poorer connectivity between neurons caused by insufficient development of neurons and pathological production of neurotransmitters, as well as influence of vitamin D on inflammatory processes in brain tissue may lead to a lower level of cognitive performance, which may persist over the life-course.
Seasonal differences in sunlight can also alter the perinatal metabolism of dopamine and melatonin, which can affect circadian and seasonal rhythms influencing serotonin turnover in the brain [22–24]. Similarly to vitamin D, both dopamine and serotonin affect the development and maturation of brain structures necessary for cognitive development, such as limbic areas, the prefrontal cortex and basal ganglia, possibly manifesting in a lower level of cognitive performance [55].
Changes in seasons can also affect health status due to maternal nutrition (access to fresh fruit), energy output (varying amounts of activity during the year), and perinatal infections- respiratory viruses [27, 56], in particular influenza viruses [8]. Murray et al. also explained the negative effect of low winter temperatures on fetus via reduced blood flow through the placenta [25]. Low temperatures as well as infection increase blood levels of fibrinogen and other inflammatory markers causing vasoconstriction and blood viscosity. This, in combination with worse winter maternal nutrition state, seems to play an important role in the disruption of fragile fetal homeostasis, which could also possibly impair cognitive development.
Despite a clear suggestion that being born in winter carries a life-long disadvantage for the level of cognitive performance, when compared to those born in summer and fall, results related to the rate of cognitive decline are less clear. Surprisingly, we found that individuals born in summer have the highest rate of cognitive decline. However, the magnitude of this association was small, and we detected it only using delayed recall and not with the other two cognitive measures. These counterintuitive findings have been previously described in literature in the context of cognitive reserve. Previous authors suggested that individuals with a higher reserve, as indicated by markers of socioeconomic position, have a higher baseline level of cognitive performance, but do not strongly differ in the rate of cognitive decline, with some evidence that individuals with the highest reserve have even a higher rate of decline [57, 58]. A similar pattern was described in a previous study looking into regional differences in cognition across Europe [59]. Evidence on the effects of cognitive reserve on cognitive trajectories have been inconsistent; however, some studies have reported that with a higher level of cognitive performance at baseline, individuals tend to experience a faster decline, as there may be more neural resources to lose [60]. We suggest that people born in summer, who had a higher level of cognitive functions at baseline, may cope with neuropathology for a longer time using compensatory reserve mechanisms, but once these resources are fully utilized, they deteriorate faster when compared to individuals who started with a lower level of cognition. These findings are also in line with literature indicating that differences in the rates of cognitive decline related to early life risk factors are small and not sufficient to offset larger differences in baseline cognitive scores [4].
Previous studies suggested some evidence of sex and regional specific associations of SOB with aging and several characteristics, such as personality and sleeping habits [24, 61]. Abeliansky et al. found that older men, in particular in Northern European countries, age faster when they were born in spring. On the contrary, SOB did not play a role for health in older age in Southern countries. A different amount of light during the year and good availability of quality fruit was suggested to be the main underlying reason [62]. An imaging study suggested that SOB is detectable with magnetic resonance imaging, implying that SOB exerts effects on the developing brain that persist over the life-course. A significantly greater effect was found for women than men [63]. However, in our study, results stratified by sex and region did not clearly suggest that either sex or any region may be at a particular risk for a lower level of cognitive performance or higher rate of cognitive decline.
This study has several limitations. Analysis of regional effect of SOB could by biased due to migration of participants as just current region of living is recorded. Furthermore, there may be residual confounders, which we did not account for, for example blood pressure, maternal infection, or vitamin D concentrations. Next limitation is that included cognitive measures do not constitute a comprehensive assessment of overall cognitive abilities. Cognitive measures used capture verbal and memory functioning whereas executive and visuospatial abilities were not evaluated. This could possibly lead to underestimation of our results as we were not able to capture the proportion of participant with lower functions in these cognitive areas. On the contrary, our study is a unique contribution to literature due to its extensive sample of well-characterized individuals coming from multiple regions.
This study is important in the context of the increasing burden of age-related cognitive disorders in Europe and the results contribute to ongoing research on early developmental factors influencing postnatal cognitive development with life-long effects. Future randomized trials should explore causal links between factors related to SOB such as vitamin D levels, nutrition status during pregnancy and postnatal period, or vaccination programs and cognitive performance in adulthood.
As we found that SOB explains differences in the level of cognitive performance between older people, but is to a lesser, almost negligible extent, associated to the rate of cognitive decline, we speculate that interventions to enhance cognitive functioning during the life-course may be of particular benefit to individuals born in winter, if applied early in life. We propose that for those people, special attention could be paid to preventable factors of dementia development, such as hypertension, diabetes, dyslipidemia, obesity, depression, or hearing loss. Physical inactivity and alcohol and tobacco consumption could be also taken into account.
Footnotes
ACKNOWLEDGMENTS
The SHARE data collection has been primarily funded by the European Commission through FP5 (QLK6-CT-2001-00360), FP6 (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, SHARELIFE: CIT4-CT-2006-028812) and FP7 (SHARE-PREP: N°211909, SHARE-LEAP: N°227822, SHARE M4: N°261982). Additional funding from the German Ministry of Education and Research, the Max Planck Society for the Advancement of Science, the U.S. National Institute on Aging (U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, OGHA_04-064, HHSN271201300071C) and from various national funding sources is gratefully acknowledged (see 
). The authors were supported by PRIMUS grant (247066) conducted at Charles University and by the Ministry of Health of the Czech Republic (grant NU20J-04-00022). All rights reserved.