Early growth, stress, and socioeconomic factors as predictors of the rate of multimorbidity accumulation across the life course : a longitudinal birth cohort study

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ Early growth, stress, and socioeconomic factors as predictors of the rate of multimorbidity accumulation across the life course : a longitudinal birth cohort study © 2023 The Author(s). Published by Elsevier Ltd. Published version Haapanen, Markus J.; Vetrano, Davide L.; Mikkola, Tuija M.; Calderón-Larrañaga, Amaia; Dekhtyar, Serhiy; Kajantie, Eero; Eriksson, Johan G.; von Bonsdorff, Mikaela B. Haapanen, M. J., Vetrano, D. L., Mikkola, T. M., Calderón-Larrañaga, A., Dekhtyar, S., Kajantie, E., Eriksson, J. G., & von Bonsdorff, M. B. (2024). Early growth, stress, and socioeconomic factors as predictors of the rate of multimorbidity accumulation across the life course : a longitudinal birth cohort study. Lancet Healthy Longevity, 5(1), e56-e65. https://doi.org/10.1016/S26667568(23)00231-3 2024 www.thelancet.com/healthy-longevity Published online December 13, 2023 https://doi.org/10.1016/S2666-7568(23)00231-3 1 Articles Introduction The risk of developing chronic diseases is not only influenced by genetics and lifestyle, but also by individual and environmental factors during the periconceptional, fetal, and infant phases of life.1,2 The Developmental Origins of Health and Disease (DOHaD) hypothesis suggests that early-life factors (eg, undernutrition, growth patterns, or stress) could shape future health outcomes through suboptimal organ development, altered metabolic and hormonal pathways, and epigenetic changes.1,2 This theory is well established for individual diseases, including diabetes and ischaemic heart disease,2–4 but the evidence remains scarce regarding the association between early life factors and multimorbidity (ie, the co-occurrence of multiple chronic diseases in the same individual) in later life.5–12 Multimorbidity is associated with premature mortality, poorer functioning, and higher use of health-care services.13 The rate of multimorbidity accumulation over time reflects the speed at which an individual is ageing,14 and consequently predicts the future development of frailty15 and use of health-care services.13 Previous studies on early life factors and multimorbidity are characterised by a small number of studies and conflicting findings. A previous study reported an association between lower birthweight and Early growth, stress, and socioeconomic factors as predictors of the rate of multimorbidity accumulation across the life course: a longitudinal birth cohort study Markus J Haapanen, Davide L Vetrano, Tuija M Mikkola, Amaia Calderón-Larrañaga, Serhiy Dekhtyar, Eero Kajantie, Johan G Eriksson, Mikaela B von Bonsdorff Summary Background Early growth, stress, and socioeconomic factors are associated with future risk of individual chronic diseases. It is uncertain whether they also affect the rate of multimorbidity accumulation later in life. This study aimed to explore whether early life factors are associated with the rate at which chronic diseases are accumulated across older age. Methods In this national birth cohort study, we studied people born at Helsinki University Central Hospital, Helsinki, Finland between Jan 1, 1934, and Dec 31, 1944, who attended child welfare clinics in the city, and were living in Finland in 1971. Individuals who had died or emigrated from Finland before 1987 were excluded, alongside participants without any registry data and who died before the end of the registry follow-up on Dec 31, 2017. Early anthropometry, growth, wartime parental separation, and socioeconomic factors were recorded from birth, child welfare clinic, or school health-care records, and Finnish National Archives. International Classification of Diseases codes of diagnoses for chronic diseases were obtained from the Care Register for Health Care starting from 1987 (when participants were aged 42–53 years) until 2017. Linear mixed models were used to study the association between early-life factors and the rate of change in the number of chronic diseases over 10-year periods. Findings From Jan 1, 1934, to Dec 31, 2017, 11 689 people (6064 [51·9%] men and 5625 [48·1%] women) were included in the study. Individuals born to mothers younger than 25 years (β 0·09; 95% CI 0·06–0·12), mothers with a BMI of 25–30 kg/m² (0·08; 0·05–0·10), and mothers with a BMI more than 30 kg/m² (0·26; 0·21–0·31) in late pregnancy accumulated chronic diseases faster than those born to older mothers (25–30 years) and those with a BMI of less than 25 kg/m². Individuals with a birthweight less than 2·5 kg (0·17; 0·10–0·25) and those with a rapid growth in height and weight from birth until age 11 years accumulated chronic diseases faster during their life course. Additionally, paternal occupational class (manual workers vs upper-middle class 0·27; 0·23–0·30) and wartime parental separation (0·24; 0·19–0·29 for boys; 0·31; 0·25–0·36 for girls) were associated with a faster rate of chronic disease accumulation. Interpretation Our findings suggest that the foundation for accumulating chronic diseases is established early in life. Early interventions might be needed for vulnerable populations, including war evacuee children and children with lower socioeconomic status. Funding Finska Läkaresällskapet, Liv och Hälsa rf, the Finnish Pediatric Research Foundation, and Folkhälsan Research Center. Copyright © 2023 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Lancet Healthy Longev 2023 Published Online December 13, 2023 https://doi.org/10.1016/ S2666-7568(23)00231-3 See Online/Comment https://doi.org/10.1016/ S2666-7568(23)00242-8 For the Finnish translation of the abstract see Online for appendix 1 For the Swedish translation of the abstract see Online for appendix 2 Public Health Research Program, Folkhälsan Research Center, Helsinki, Finland (M J Haapanen DMedSc, T M Mikkola PhD, Prof J G Eriksson DMedSc, M B von Bonsdorff PhD); Department of General Practice and Primary Health Care (M J Haapanen, Prof J G Eriksson), Clinicum, Faculty of Medicine (T M Mikkola), University of Helsinki, Helsinki, Finland; Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden (M J Haapanen); Aging Research Center, Department of Neurobiology, Care Sciences and Society, Karolinska Institutet and Stockholm University, Stockholm, Sweden (D L Vetrano PhD, A Calderón-Larrañaga PhD, S Dekhtyar PhD); Stockholm Gerontology Research Center, Stockholm, Sweden (D L Vetrano, A Calderón-Larrañaga); Population Health Unit, Finnish Institute for Health and Welfare, Helsinki, Finland (T M Mikkola, Prof E Kajantie DMedSc); Clinical Medicine Research Unit, Medical Research Center Oulu, Oulu University Hospital, University of Oulu, Oulu, Finland (Prof E Kajantie); Children’s Hospital, Helsinki University Hospital, University of Helsinki, Helsinki, Finland Articles 2 www.thelancet.com/healthy-longevity Published online December 13, 2023 https://doi.org/10.1016/S2666-7568(23)00231-3 (Prof E Kajantie); Department of Clinical and Molecular Medicine, Norwegian University of Science and Technology, Trondheim, Norway (Prof E Kajantie); Yong Loo Lin School of Medicine, Department of Obstetrics and Gynaecology and Human Potential Translational Research Programme, National University Singapore, Singapore (Prof J G Eriksson); Singapore Institute for Clinical Sciences, Agency for Science, Technology and Research, Singapore (Prof J G Eriksson); Gerontology Research Center and Faculty of Sport and Health Sciences, University of Jyväskylä, Jyväskylä, Finland (M B von Bonsdorff) Correspondence to: Dr Markus J Haapanen, Public Health Research Program, Folkhälsan Research Center, 00250 Helsinki, Finland [email protected] multimorbidity at 46–48 years,5 but no such association has been observed in old age.6 Scarce findings on early growth (birth to 1 year) reported no association with multimorbidity risk.6 Parental socioeconomic status and childhood health have shown varying degrees of association with multimorbidity risk,6–9 but not with multimorbidity accumulation.7,9,10 Still, adequacy of childhood nutrition has been associated with slower multimorbidity accumulation.9 Childhood adverse experiences have been associated with multimorbidity more consistently,11,12 particularly parental physical abuse, which was associated with faster disease accumulation.9 Yet, evidence remains scarce regarding the effect of severe early stress resulting from war, conflicts, and migration movements on multimorbidity. The prevalence of multimorbidity has increased over the past two decades.16 Understanding factors connected with this trend is crucial, especially modifiable early risk factors, because chronic disease accumulation poses a substantial challenge to individuals, their caregivers, and health-care systems. To address previous gaps in research, we aimed to provide a comprehensive overview of early factors covering physical, psychosocial, and socioeconomic aspects from developmentally important life stages. Recent evidence from this cohort highlighted the importance of earlier life phases on healthy ageing.17 With the DOHaD hypothesis1,2 as our theoretical framework, our study investigated whether factors during gestation, birth, infancy, and childhood predict the rate of chronic disease accumulation over a 30-year period. We hypothesised that adverse early life circumstances would accelerate the depletion of physiological reserves and the ageing process, leading to faster chronic disease accumulation. Methods Study design and participants This national birth cohort study used the Helsinki Birth Cohort, which includes 6975 men and 6370 women who were born at Helsinki University Central Hospital, Helsinki, Finland between Jan 1, 1934, and Dec 31, 1944, attended child welfare clinics in the city, and were living in Finland in 1971 when unique identification numbers were assigned to all Finnish residents (appendix 3 p 25).3 Early life characteristics, spanning from birth to 11 years old, were extracted from records held at the hospitals, child welfare and school health-care clinics, and the Finnish National Archives.3,4,18 After excluding individuals who had died (n=404) or emigrated from Finland (n=895) before 1987, we used the participants’ unique identification number to integrate clinical data from the Care Register for Health Care19 from Jan 1, 1987. We excluded an additional 357 participants without any registry data and those who died before the end of the registry follow-up on Dec 31, 2017, resulting in an analytical sample of 11 689 individuals. Follow-up ended with participant death, emigration, or on Dec 31, 2017, whichever occurred first. The study was approved by the Ethics Committees of the Hospital District of Helsinki and Uusimaa and that of the National Public Health Institute, Helsinki, Finland. Procedures Multimorbidity was quantified as the sum of chronic diseases, serving as a proxy measure of accelerated ageing.14 Chronic diseases were defined as conditions with prolonged duration, resulting in residual disability or requiring extended care, treatment, or rehabilitation.20 Each chronic disease was categorised into one of the Research in context Evidence before this study We searched PubMed for studies published from the inception of the database to Dec 1, 2022, using the search terms “multimorbidity” AND (“birth weight” OR “early growth” OR “childhood adversity” OR “childhood socioeconomic status”) along with all the synonyms (appendix 3 p 28). We only included studies published in English. We specifically looked for studies that defined multimorbidity in late adulthood or old age. Although survival rates have improved, the globally ageing population and unhealthy lifestyle factors have been identified as contributing factors to the rising prevalence of multimorbidity. Our comprehensive understanding of how earlier life phases affect this phenomenon remains poor. Early growth, stress, and socioeconomic factors have been shown to influence the future risk of individual chronic diseases in adults. However, when it comes to multimorbidity, the evidence is both scarce and characterised by conflicting findings. It is uncertain whether factors in the periconceptional, fetal, and infant phases of life affect the rate of multimorbidity during the life course. Added value of this study Our study suggest that younger maternal age and higher BMI, small body size at birth, rapid childhood growth in height and weight, adverse socioeconomic circumstances, and wartime parental separation influence the speed of accumulation of chronic diseases in later life. Implications of all the available evidence The influence of early life characteristics on multimorbidity accumulation is broader and more consistent than what previous studies suggest. To improve the long-term health of future generations, interventions should focus on preventing maternal obesity, optimising maternal health during pregnancy, and preventing childhood obesity on a global scale. Groups at a high risk of accumulating chronic diseases during adulthood, such as war evacuee children and those with lower socioeconomic status, might require targeted early interventions. See Online for appendix 3 Articles www.thelancet.com/healthy-longevity Published online December 13, 2023 https://doi.org/10.1016/S2666-7568(23)00231-3 3 60 chronic disease categories proposed by CalderónLarrañaga and colleagues.20 To this end, all hospital inpatient and outpatient health-care records from Jan 1, 1987, until Dec 31, 2017, were extracted from the Care Register for Health Care,19 including visit or admission dates and diagnostic codes, transitioning from International Classification of Diseases (ICD)-9 to ICD-10 in 1995 (appendix 3 pp 2–20). Age at visit or admission indicated the onset of each chronic disease, and when a disease was present, it persisted in all subsequent data points. The primary outcome variable for this study was the changing sum of chronic disease groups a person had over time. In accordance with the DOHaD hypothesis,1,2 we chose physical, psychosocial, and socioeconomic factors from different developmental stages to comprehensively capture the participants’ early life circumstances. Birth records from the Helsinki University Central Hospital provided data on maternal age, parity, height, weight at giving birth, and last menstrual period. Gestational age was calculated as the time in weeks between the last menstrual period and date of birth. Parity information was used to dichotomise the participants as first born or not. The birth records for newborns included information on birthweight and length. The mother’s BMI (in kg/m²) and the BMI (in kg/m²) and ponderal index (in kg/m³) of the newborn were calculated. The children had been regularly measured in child welfare clinics and school health care, archived by the Helsinki City archives. For each child, we estimated their height, weight, and BMI at each birthday from 1 to 11 years. Measurements within 2 years of the specific age were considered. These values were converted into Z scores to indicate the deviation from the cohort mean. Fewer measurements were made between age of 2 years and enrolment at school than the number measured before 2 years of age. Conditional growth (0–2 years for infancy; 2–7 years for early childhood; and 7–11 years for childhood) was determined by examining the residuals from linear regression to indicate how body size at each age differed from the predicted size based on an earlier age. Childhood socioeconomic status (SES) was coded as manual workers, lower-middle class, and upper-middle class based on the father’s highest occupational status reported in birth, child welfare clinic, or school health-care records. Information on separations during World War 2, when participants were sent abroad without their families to protect them from the war, was obtained from the Finnish National Archives and included details on the age at and duration of the separation.18 Statistical analysis Our analysis aimed to explore and quantify the association between early life factors and the change in chronic diseases with age. As disease accumulation was linear in this sample (appendix 3 p 26), we used separate linear mixed models for each early life factor. Hypothesised relationships between early life factors and chronic disease accumulation are reported in appendix 3 (p 27). We centred age at 42 years, the youngest age in our dataset, and other continuous variables at their means. Model comparisons, including assessments of model fit, residual plots, and plotted model predictions, showed no evidence of quadratic time effects or interactions with early-life factors. Furthermore, overall assessment of the comparisons supported a Gaussian distribution over a Poisson distribution for the number of chronic diseases. We determined fixed effects based on the literature, including covariate–age interactions due to their significant associations with chronic disease accumulation. To account for individual variability, we included random intercepts for each individual, determining the compound symmetry within-individual variance–covariance structure. Age-adjusted and sex-adjusted models were controlled for sex and its interaction with age, with age serving as the underlying time scale. Fully adjusted models were additionally controlled for childhood SES and its interaction with age (appendix 3 p 21). Inter-relationships between early life factors (appendix 3 p 22) were considered, such as including gestational age in models of weight, length, and BMI at birth, and birthweight in models including maternal BMI. We incorporated quadratic terms for early life factors to test for non-linear relationships and included them if significant. Complete case results are presented for the entire sample. For wartime separation models, we present separate results for boys and girls due to significant sex interactions (p<0·001). 95% CIs were calculated using parametric bootstrapping. We set statistical significance as a p value of less than 0·05 and used Bonferroni correction for multiple comparisons. We performed all analyses using R21 packages lme422 and lmerTest.23 Role of the funding source The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. Results From Jan 1, 1934, to Dec 31, 2017, 11 689 participants (6064 [51·9%] men and 5625 [48·1%] women) were included in the study. At birth, boys had a mean weight of 3·47 kg (SD 0·49) and measured 50·6 cm (SD 2·0), whereas girls had a mean weight of 3·34 kg (SD 0·46) and measured 49·9 cm (SD 1·8; table 1). Socioeconomically, most participants came from manual working back grounds. During World War 2, 1437 (12·5%) participants were evacuated. At the start of the follow-up on Jan 1, 1987, the mean age of participants was 45·8 years (SD 2·8). The proportion of participants with chronic diseases, and the number of chronic diseases individuals had, increased over time (figure 1). Participants were followed up for a median of 31 years (IQR 28–31), with the oldest participant reaching 84 years Articles 4 www.thelancet.com/healthy-longevity Published online December 13, 2023 https://doi.org/10.1016/S2666-7568(23)00231-3 old. During follow-up, the entire cohort accumulated 1·9 more diseases every 10 years (table 1). The incidence rate of two or more chronic diseases was 29·8 (95% CI 29·2–30·4) per 1000 person-years and the incidence rate of three or more chronic diseases was 25·4 (24·8–25·9) per 1000 person-years (table 1). Between Jan 1, 1987, and Dec 31, 2017, 2887 (24·7%) participants of the original cohort died (table 1). However, birth and maternal factors All (n=11 689) Men (n=6064) Women (n=5625) Birth characteristics Weight, kg 11 689 (100%); 3·40 (0·48) 6064 (100%);3·47 (0·49) 5625 (100%); 3·34 (0·46) Length, cm 11 579 (99·1%); 50·2 (1·9) 6011 (99·1%); 50·6 (2·0) 5568 (99·0%); 49·9 (1·8) BMI, kg/m211 579 (99·1%); 13·4 (1·2) 6011 (99·1%); 13·5 (1·2) 5568 (99·0%); 13·4 (1·2) Ponderal index, kg/m311 579 (99·1%); 26·7 (2·2) 6011 (99·1%); 26·7 (2·3) 5568 (99·0%); 26·8 (2·2) Gestational age, weeks 11 294 (96·6%); 39·9 (1·9) 5841 (96·3%); 39·9 (1·9) 5453 (96·9%); 40·0 (1·9) First born 5705/11 686; (48·8%) 2994/6063 (49·4%) 2711/5623 (48·2%) Second born or later 5981/11 686 (51·2%) 3069/6063 (50·6%) 2912/5623 (51·8%) Maternal characteristics Age, years 11 682 (99·9%); 28·4 (5·4) 6060 (99·9%); 28·3 (5·4) 5622 (99·9%); 28·5 (5·5) BMI, kg/m210 529 (90·1%); 26·2 (2·9) 5461 (90·1%); 26·2 (2·9) 5068 (90·1%); 26·2 (2·9) Childhood socioeconomic status Manual worker 6648/11 392 (58·4%) 3412/5925 (57·6%) 3236/5467 (59·2%) Lower-middle class 2763/11 392 (24·2%) 1433/5925 (24·2%) 1330/5467 (24·3%) Upper-middle class 1981/11 392 (17·4%) 1080/5925 (18·2%) 901/5467 (16·5%) Wartime separation during World War 2 Separated 1437/11 500 (12·5%) 774/5995 (12·9%) 663/5505 (12·0%) Age at separation, years 1297/1437 (90·3%); 4·6 (2·4) 704/774 (91·0%); 4·5 (2·4) 593/663 (89·4%); 4·8 (2·4) Duration of separation, years 1266/1435 (88·2%); 1·7 (1·0) 685/774 (88·5%); 1·8 (1·1) 581/663 (87·6%); 1·7 (1·0) Serial measurements of body size Height (cm) at age 1 year 11 673 (99·9%); 75·7 (2·7) 6054 (99·8%); 76·5 (2·6) 5619 (99·9%); 74·8 (2·6) 2 years 11 676 (99·9%); 86·0 (3·2) 6056 (99·9%); 86·6 (3·2) 5620 (99·9%); 85·5 (3·2) 7 years 8763 (75·0%); 120·3 (4·8) 4554 (75·1%); 120·7 (4·9) 4209 (74·8%); 119·8 (4·7) 11 years 8582 (73·4%); 141·4 (6·2) 4476 (73·8%); 141·4 (6·0) 4106 (73·0%); 141·4 (6·4) Weight (kg) at age 1 year 11 683 (99·9%); 10·2 (1·1) 6059 (99·9%); 10·5 (1·1) 5624 (100%); 9·8 (1·0) 2 years 11 686 (100%); 12·1 (1·2) 6062 (100%); 12·4 (1·2) 5624 (100%); 11·9 (1·2) 7 years 8773 (75·1%); 22·3 (2·8) 4560 (75·2%); 22·5 (2·7) 4213 (74·9%); 22·1 (2·9) 11 years 8581 (73·4%); 34·0 (5·2) 4477 (73·8%); 33·7 (4·6) 4104 (73·0%); 34·3 (5·7) BMI (kg/m2) at age 1 year 11 676 (99·9%); 17·7 (1·4) 6055 (99·9%); 17·9 (1·4) 5621 (99·9%); 17·5 (1·4) 2 years 11 680 (99·9%); 16·5 (1·2) 6058 (99·9%); 16·7 (1·2) 5622 (99·9%); 16·4 (1·2) 7 years 8748 (74·8%); 15·5 (1·2) 4548 (75·0%); 15·5 (1·1) 4200 (75·7%); 15·5 (1·3) 11 years 8571 (73·3%); 17·0 (1·7) 4471 (73·7%); 16·8 (1·5) 4100 (72·9%); 17·1 (1·9) Follow-up information Age at the start of the follow-up, years 11 689 (100%); 45·8 (2·8) 6064 (100%); 45·8 (2·8) 5625 (100%); 45·9 (2·8) Follow-up, years 11 689 (100%); 31 (28–31) 6064 (100%); 31 (25–31) 5625 (100%); 31 (31–31) Multimorbidity accumulation,* diseases per 10 years 11 679 (99·9%); 1·9 (1·9) 6057 (99·9%); 2·0 (2·2) 5622 (99·9%); 1·8 (1·6) Incidence of two or more diseases, incidence rate per 1000 person-years (95% CI) † 11 679 (99·9%); 29·8 (29·2–30·4) 6057 (99·9%); 29·0 (28·1–29·8) 5622 (99·9%); 30·7 (29·9–31·6) Incidence of three or more diseases, incidence rate per 1000 person-years (95% CI)† 11 679 (99·9%); 25·4 (24·8–25·9) 6057 (99·9%); 24·4 (23·6–25·2) 5622 (99·9%); 26·3 (25·5–27·1) Died between 1987 and 2017 2887/11689 (24·7%) 1845/6064 (30·4%) 1042/5625 (18·5%) Data are n (%), n/N (%), mean (SD), or median (IQR), unless otherwise stated. *Calculated by dividing the difference between the number of chronic disease groups each participant had at death or the end of the follow-up and at baseline by the individual follow-up time of each participant in years. †Confidence intervals are exact 95% Poisson confidence intervals. Table 1: Characteristics of the study population Articles www.thelancet.com/healthy-longevity Published online December 13, 2023 https://doi.org/10.1016/S2666-7568(23)00231-3 5 were not significant predictors of all-cause mortality (appendix 3 p 23). Smaller body size at birth and later birth order were linked to faster multimorbidity accumulation from the age of 42 years onwards, even after adjusting for sex, childhood SES, and their interactions with age (table 2). One-unit higher birthweight in kg was associated with 0·05 (95% CI 0·02 to 0·08) fewer chronic disease groups per 10 years and one-unit longer birth length in cm was associated with 0·02 (0·01 to 0·02) fewer chronic disease groups per 10 years. Additionally, the children of younger mothers (<25 years vs 25–30 years; β 0·09; 95% CI 0·06 to 0·12; table 2), mothers with BMI in the overweight range (25–30 kg/m² vs <25 kg/m²; 0·08; 0·05 to 0·10), or mothers with BMI in the obese range (>30 kg/m² vs <25 kg/m²; 0·26; 0·21 to 0·31) in late pregnancy had a faster rate of chronic disease accumulation (table 2). We standardised birth and maternal factors for better comparability and found that an increase of one standard deviation in maternal BMI (β 0·07; 95% CI 0·06 to 0·08) and maternal age (–0·06; –0·07 to –0.04) exhibited the strongest associations with an accelerated rate of chronic disease accumulation, followed by birth length (–0·03; –0·04 to –0·02) and birthweight (–0·02; –0·04 to –0·01; appendix 3 p 24). Participants from manual worker and lower-middle class backgrounds experienced faster rates of chronic disease accumulation (table 2). Compared with participants from upper-middle class backgrounds, those from a manual worker background were associated with 0·27 (95% CI 0·23–0·30) more disease groups per 10 years and those from lower-middle class backgrounds were associated with 0·15 (0·12–0·19) more disease groups per 10 years. Boys and girls who grew more rapidly in height, weight, and BMI from birth until 11 years had faster chronic disease accumulation (table 2). In particular, those with a faster weight (per 1 SD β 0·08; 95% CI 0·07–0·09) and BMI (per 1 SD 0·08; 0·07–0·10) growth from age 7 to 11 years were associated with faster chronic disease accumulation. Boys and girls who as adults were in the highest third of chronic disease accumulation grew more rapidly in height, weight, and BMI during infancy, which then accelerated in childhood (figure 2). Boys and girls evacuated during World War 2 accumulated diseases faster during their life course than those who had not been evacuated (table 3). Separated boys presented with 0·24 (95% CI 0·19–0·29) and separated girls with 0·31 (0·25–0·36) more chronic disease groups per 10 years compared with those who were not separated. Girls and boys separated at an older age (older than 7 years) were associated with faster chronic disease accumulation (table 3). The duration of the separation was not associated with chronic disease accumulation. Discussion We found that the rate at which chronic diseases were accumulated was influenced by factors during gestation, birth, infancy, and childhood. The children of younger mothers and mothers with BMI in the overweight and obese ranges in late pregnancy had a faster accumulation Figure 1: Distribution of the number of chronic diseases from age 42 to 84 years 0 25 50 75 100 Age (years) Percentage of the Helsinki cohort None One Two Three Four Five Six Seven Eight Nine Ten or more Chronic diseases 40 50 60 70 80 Articles 6 www.thelancet.com/healthy-longevity Published online December 13, 2023 https://doi.org/10.1016/S2666-7568(23)00231-3 Age and sex adjusted* Fully adjusted† Number of participants (n=11 689) β × time (95% CI; p value) Number of participants (n=11 392) β × time (95% CI; p value) Birth factors‡ Birthweight, kg§ 11 689 (100%) –0·08 (–0·10 to –0·05; <0·0001) 11 009 (96·6%) –0·05 (–0·08 to –0·02; 0·0070) <2·5 kg 386/11 689 (3·3%) 0·20 (0·14 to 0·27; <0·0001) 329/11 009 (3·0%) 0·17 (0·10 to 0·25; 0·0002) 2·5 to <3·0 kg 1822/11 689 (15·6%) 0·07 (0·03 to 0·10; 0·0009) 1672/11 009 (15·2%) 0·07 (0·04 to 0·11; 0·0012) 3·0 to <3·5 kg 4725/11 689 (40·4%) 1 (ref) 4483/11 009 (40·7%) 1 (ref) 3·5 to <4·0 kg 3669/11 689 (31·4%) –0·01 (–0·03 to 0·02; 0·99) 3480/11 009 (31·6%) 0·02 (–0·01 to 0·05; 0·99) ≥4·0 kg 1087/11 689 (9·3%) –0·05 (–0·09 to –0·01; 0·79) 1045/11 009 (9·5%) –0·04 (–0·09 to 0·01; 0·99) Birth length, cm§ 11 579 (99·1%) –0·02 (–0·03 to –0·01; 0·0001) 10 913 (95·8%) –0·02 (–0·02 to –0·01; 0·0003) Birth length, cm§¶ 11579 (99·1%) –0·02 (-0·03 to –0·01; <0·0001) 10 913 (95·8%) –0·02 (–0·02 to –0·01; 0·0006) Birth BMI, kg/m²§ 11579 (99·1%) –0·02 (–0·03 to –0·01; 0·0001) 10 913 (95·8%) –0·01 (–0·02 to 0·00; 0·99) Ponderal index, kg/m³§ 11579 (99·1%) –0·01 (–0·01 to –0·00; 0·64) 11 288 (99·1%) –0·01 (–0·01 to –0·00; 0·99) Gestational age, weeks 11294 (96·6%) –0·01 (–0·01 to 0·00; 0·99) 11 009 (96·6%) –0·01 (–0·01 to –0·00; 0·35) Parity First born 5705/11 686 (48·8%) 1 (ref) 5464/11 389 (48·0%) 1 (ref) Second or later 5981/11 686 (51·2%) 0·08 (0·06 to 0·11; <0·0001) 5925/11 389 (52·0%) 0·07 (0·05 to 0·10; <0·0001) Maternal factors‡ Age (continuous; years) 11 682 (99·9%) –0·01 (–0·01 to –0·01; <0·0001) 11 385 (99·9%) –0·01 (–0·01 to –0·01; <0·0001) <25 years 3886/11 682 (33·3%) 0·09 (0·07 to 0·12; <0·0001) 3732/11 385 (32·8%) 0·09 (0·06 to 0·12; <0·0001) 25 to <30 years 3789/11 682 (32·4%) 1 (ref) 3719/11 385 (32·7%) 1 (ref) ≥30 years 4007/11 682 (34·3%) –0·03 (–0·06 to –0·01; 0·80) 3934/11 385 (34·5%) –0·03 (–0·06 to –0·00; 0·73) BMI (continuous; kg/m²)|| 10 529 (90·1%) 0·02 (0·02 to 0·03; <0·0001) 10 262 (90·1%) 0·02 (0·02 to 0·03; <0·0001) BMI (kg/m2)¶|| 10 529 (90·1%) 0·02 (0·02 to 0·03; <0·0001) 10 262 (90·1%) 0·02 (0·02 to 0·03; <0·0001) BMI|| <25 kg/m² 3823/10 529 (36·3%) 1 (ref) 3725/10 262 (36·3%) 1 (ref) 25 to 30 kg/m²§ 5745/10 529 (54·6%) 0·06 (0·03 to 0·09; 0·0002) 5596/10 262 (54·5%) 0·08 (0·05 to 0·10; <0·0001) >30d kg/m²§ 961/10 529 (9·1%) 0·22 (0·17 to 0·26; <0·0001) 941/10 262 (9·2%) 0·26 (0·21 to 0·31; <0·0001) Childhood socioeconomic status‡ Upper-middle 1981/11 392 (17·4%) 1 (ref) ·· ·· Lower-middle 2763/11 392 (24·2%) 0·15 (0·12 to 0·19; <0·0001) ·· ·· Manual worker 6648/11 392 (58·4%) 0·27 (0·23 to 0·30; <0·0001) ·· ·· Conditional growth from birth to age 2 years Weight growth, per SD 8243 (70·5%) 0·04 (0·02 to 0·05; <0·0001) 8133 (71·4%) 0·04 (0·03 to 0·05; <0·0001) Height growth, per SD 8165 (69·9%) 0·03 (0·02 to 0·05; 0·0003) 8057 (70·7%) 0·04 (0·03 to 0·05; <0·0001) BMI growth, per SD 8144 (69·7%) 0·01 (0·01 to 0·03; 0·035) 8036 (70·5%) 0·01 (0·00 to 0·03; 0·99) Conditional growth from 2 years to 7 years Weight growth, per SD 8243 (70·5%) 0·02 (0·01 to 0·04; 0·0004) 8133 (71·4%) 0·03 (0·01 to 0·04; 0·0011) Height growth, per SD 8165 (69·9%) 0·02 (0·01 to 0·03; 0·032) 8057 (70·7%) 0·02 (0·01 to 0·03; 0·062) BMI growth, per SD 8144 (69·7%) 0·03 (0·01 to 0·04; 0·0006) 8036 (70·5%) 0·03 (0·01 to 0·04; 0·0030) Conditional growth from 7 years to 11 years Weight growth, per SD 8243 (70·5%) 0·08 (0·07 to 0·10; <0·0001) 8133 (71·4%) 0·08 (0·07 to 0·09; <0·0001) Height growth, per SD 8165 (69·9%) 0·03 (0·02 to 0·05; 0·0001) 8057 (70·7%) 0·03 (0·02 to 0·04; 0·0002) BMI growth, per SD 8144 (69·7%) 0·09 (0·07 to 0·10; <0·0001) 8036 (70·5%) 0·08 (0·07 to 0·10; <0·0001) p values are Bonferroni adjusted. *Linear mixed model adjusted with sex and its interaction with age. Age was treated as the underlying time scale and was therefore inherently adjusted for. Positive estimates (unstandardised β × time) indicate a faster increase in chronic diseases per 10 years, whereas negative estimates refer to a slower increase. †Linear mixed model adjusted with sex, childhood socioeconomic status, and their interactions with age. Age was treated as the underlying time scale and was therefore inherently adjusted for. Positive estimates (unstandardised β × time) indicate a faster increase in chronic diseases per 10 years, whereas negative estimates refer to a slower increase. ‡Analysed in separate models. §Fully adjusted model adjusted additionally with gestational age. ¶Quadratic term included. ||Fully adjusted model adjusted additionally with birthweight. Table 2: The association between early life factors and the rate of multimorbidity accumulation over 31 years Articles www.thelancet.com/healthy-longevity Published online December 13, 2023 https://doi.org/10.1016/S2666-7568(23)00231-3 7 of chronic diseases. Participants who had been born small or who grew faster in size during infancy and childhood accumulated chronic diseases faster during the life course. Childhood socioeconomic factors and wartime parental separation had similar long-lasting associations with chronic disease accumulation. The negative effect of wartime parental separation was more severe overall for girls compared with boys regardless of age at separation, but the largest effect was in boys separated at an older age. The findings persisted when other determinants of health, including sex and childhood socioeconomic status, were accounted for. In accordance with the DOHaD hypothesis,1,2 the findings show that the pace of multimorbidity accumulation could be programmed early by factors during sensitive developmental periods and have long-lasting effects on later health. Our results provide an overview of early life factors and reveal that their influence on chronic disease accumulation is broader and more consistent across biological, psychological, and socioeconomic factors than that suggested in previous studies.5–12 Maternal obesity affects around one in six pregnant mothers,24 and it has been linked to higher levels of proinflammatory and dysmetabolic signalling25 as well as risk of cardiovascular disease in their children.26 Our study suggests that a higher maternal BMI was one of the strongest predictors of morbidity in their children. Those born to obese mothers accumulated 0·26 more diseases per decade than those born to mothers with a BMI of less than 25 kg/m², taking about 40 years to acquire one additional chronic disease. It is unclear whether these associations are solely attributable to maternal BMI or if maternal BMI acts as a surrogate for potential confounding factors, such as maternal education or lifestyle. We also found that the children of younger mothers (younger than 25 years) had faster chronic diseases accumulation. This aligns with previously reported U-shaped associations between maternal age and health of their children27 and might indirectly reflect access to parenting skills and resources. However, advanced maternal age could not be studied due to its rarity in this cohort. Our findings suggest that individuals who were born small are more likely to have a faster pace of chronic disease accumulation. Individuals with a low birthweight (<2·5 kg) accumulated 0·17 more diseases per decade and were estimated to acquire one additional chronic disease after 60 years compared with those weighing 3·0–3·5 kg at birth. According to the DOHaD hypothesis,1,2 disturbances during prenatal life can programme physiological systems and metabolism, leading to long-term changes in the structure and function of organs and tissues. Individuals with lower birthweights have an increased risk of chronic diseases, including cardiovascular disease and diabetes, in adulthood.2–4 The underlying mechanisms have been suggested to involve insulin resistance and other metabolic alterations,3 inflammation,28 and epigenetic changes.29 Boys and girls who had rapid height and weight growth accumulated chronic diseases faster during their life Figure 2: The patterns of infant and childhood growth in individuals stratified by their rate of chronic disease accumulation Height (A), weight (B), and BMI (C) growth curves from infancy to childhood according to thirds of chronic disease accumulation speed. Values above the dashed line indicate faster growth than the cohort average, whereas values below the dashed line indicate slower growth. Chronic disease accumulation speed was calculated by dividing the difference between the number of chronic disease groups each participant had at death or the end of the follow-up and at baseline by the individual follow-up time of each participant in years. Tertile cutoffs were at 0·97 and 1·94 disease groups increase per 10 years. Shaded areas represent 95% CI regions. –0·02 –0·04 0 0·02 0·04 0·06 Height, Z scores –0·050 –0·025 0 0·025 0·050 0·075 Weight, Z scores 0 0·05 –0·05 0·10 01234567891 01 211 Age BMI, Z scores A C B Lowest tertile Middle tertile Highest tertile Articles 8 www.thelancet.com/healthy-longevity Published online December 13, 2023 https://doi.org/10.1016/S2666-7568(23)00231-3 course. Such patterns of growth have been suggested to increase the risk of type 2 diabetes4 and cardiovascular disease.3 Babies who are small have less muscle, and, because of little muscle replication during childhood, the added weight could be disproportionately attributed to adipose tissue, predisposing to insulin resistance.3 Our findings suggest that rapid growth, particularly in BMI, from ages 7 to 11 years could increase chronic disease accumulation. We have previously shown that faster BMI growth in this age range predicts adiposity30 and obesity31 in late midlife. In our study, these factors could affect the causal pathway connecting early growth to multimorbidity development. During World War 2, around 70 000 Finnish children were evacuated without their parents to temporary foster care abroad—primarily to Sweden—to protect them from the war.18 This created a unique natural experiment to study the association between wartime parental separation and later health in this cohort.32,33 Although severe stress probably followed wartime parental separation, a double separation trauma could have occurred when the child returned home from foster care years later. Separated boys (0·24) and girls (0·31) accumulated more diseases each decade, taking 40 years in boys and 35 years in girls for one additional chronic disease. Following the DOHaD hypothesis,1,2 early exposure to severe stress when the hypothalamic-pituitary-adrenal axis is developing can affect or alter subsequent stress responses, predisposing to adult morbidity.32,34 The negative effect of wartime parental separation was more pronounced in boys separated at an older age. We have previously described higher cortisol and adrenocorticotropic hormone concentrations in late adulthood in children who were separated, with separated boys showing higher reactivity in a stress test.18 Similar sex differences have also been shown for other adult chronic diseases32 and frailty in old age.33 Worse childhood socioeconomic status predicted faster chronic disease accumulation in our study. Children in the lowest socioeconomic group accumulated 0·27 more diseases per decade than the highest group, requiring 40 years to acquire one additional chronic disease. Previously, some,8,9 but not all,6,7 studies reported that poorer childhood SES was associated with an increased multimorbidity risk, with little evidence of childhood or birth SES affecting disease accumulation.7,9,10 This heterogeneity could result from how childhood SES (parental education, occupation, or childhood health) and multimorbidity (seven, 14, or 60 conditions) are operationalised. Our results suggest that childhood socioeconomic circumstances, defined with parental occupation, contribute to inequalities in chronic disease accumulation. Socioeconomic factors are structurally embedded in early life factors, suggesting that individuals with a higher SES might have more opportunities for accessing education and income, potentially improving maternal and child health. Age adjusted* Fully adjusted† Number of boys; β × time (95% CI); p value Number of girls; β × time (95% CI); p value Number of boys; β × time (95% CI); p value Number of girls; β × time (95% CI); p value Wartime separation from both parents‡ Not separated 5221/5995 (87·1%); 1 (ref) 4842/5505 (88·0%); 1 (ref) 5104/5857 (87·1%); 1 (ref) 4700/5347 (87·9%); 1 (ref) Separated 774/5995 (12·9%); 0·26 (0·21 to 0·31); <0·0001 663/5505 (12·0%); 0·32 (0·27 to 0·37); <0·0001 753/5857 (12·9%); 0·24 (0·19 to 0·29); <0·0001 647/5347 (12·1%; 0·31 (0·25 to 0·36); <0·0001 Age at separation‡ 1 to <2 years 108/704 (15·3%); 1 (ref) 76/593 (12·8%); 1 (ref) 102/687 (14·8%); 1 (ref) 72/580 (12·4%); 1 (ref) 2 to <4 years 238/704 (33·8%); 0·27 (0·10 to 0·42); 0·0073 185/593 (31·2%); –0·09 (–0·27 to 0·08); 0·99 235/687 (34·2%); 0·28 (0·12 to 0·45); 0·0053 183/580 (31·6%); –0·13 (–0·30 to 0·03); 0·99 4 to <7 years 229/704 (32·6%); 0·54 (0·38 to 0·71); <0·0001 212/593 (35·8%); –0·07 (–0·25 to 0·10); 0·99 226/687 (32·9%); 0·55 (0·38 to 0·71); <0·0001 207/580 (35·7%); –0·13 (–0·30 to 0·05); 0·98 >7 years 129/704 (18·3%); 0·41 (0·21 to 0·59); 0·0001 120/593 (20·2%); 0·28 (0·09 to 0·47); 0·021 124/687 (18·1%); 0·45 (0·27 to 0·62); <0·0001 118/580 (20·3%); 0·22 (0·04 to 0·42); 0·18 Duration of the separation‡ 0 to <1 year 149/685 (21·8%); 1 (ref) 132/593 (22·3%); 1 (ref) 147/668 (22·0%); 1 (ref) 131/568 (23·1%); 1 (ref) 1 to <2 years 328/685 (47·8%); –0·03 (-0·16 to 0·11); 0·99 285/593 (48·1%); 0·02 (-0·12 to 0·15); 0·99 320/668 (47·9%); 0·02 (–0·12 to 0·15); 0·99 277/568 (48·8%); 0·01 (–0·14 to 0·15); 0·99 2 to <3 years 112/685 (16·4%); 0·13 (–0·02 to 0·30); 0·79 102/593 (17·2%); –0·20 (–0·36 to -0·01); 0·99 108/668 (16·2%); 0·19 (–0·05 to 0·37); 0·16 99/568 (17·4%); –0·23 (–0·38 to –0·06); 0·082 >3 years 96/685 (14·0%); 0·14 (–0·03 to 0·30); 0·81 62/593 (10·4%); 0·12 (–0·08 to 0·33); 0·99 93/668 (13·9%); 0·20 (0·01 to 0·38); 0·19 61/568 (10·7%); 0·11 (–0·09 to 0·30); 0·99 p values are Bonferroni-adjusted. *Age was treated as the underlying time scale and was therefore inherently adjusted for. †Linear mixed model adjusted with childhood socioeconomic status and its interactions with age; age was treated as the underlying time scale and was therefore inherently adjusted for; positive estimates (unstandardised β × time) indicate a faster increase in chronic diseases per 10 years, whereas negative estimates refer to a slower increase. ‡Analysed in separate models. Table 3: The association between wartime separation with the rate of multimorbidity accumulation over 31 years