The results are shown according to age at diagnosis among women and men. There were 3529 ischemic strokes among 151 955 women and 5370 ischemic strokes among 155 722 men.
Shown are early (A) vs late (B) ischemic stroke among women and men. All analyses are stratified by birth cohort. BMI indicates body mass index; HR, hazard ratio. The dashed line indicates an HR of 1. The dark blue lines indicate the HR for the association between child BMI and adult ischemic stroke. The light blue lines indicate the upper and lower bounds of the 95% CIs for the associations.
eFigure 1. Flowchart Describing the Eligibility Criteria in the Overall Study (n = 307 677), Based on Data from the Copenhagen School Health Records Register
eFigure 2. Childhood BMI at 8-12 Years of Age and Early (Age at Diagnosis ≤55 Years) vs Late (Age at Diagnosis >55 Years) Ischemic Stroke Among Women
eFigure 3. Childhood BMI at 8-12 Years of Age and Early (Age at Diagnosis ≤55 Years) vs Late (Age at Diagnosis >55 Years) Ischemic Stroke Among Men
eFigure 4. Childhood BMI and Early (Age at Diagnosis ≤55 Years) vs Late (Age at Diagnosis >55 Years) Ischemic and Unspecified Stroke Among Women
eFigure 5. Childhood BMI and Early (Age at Diagnosis ≤55 Years) vs Late (Age at Diagnosis >55 Years) Ischemic and Unspecified Stroke Among Men
eTable 1. Birth Weight and Early (Age at Diagnosis ≤55 Years) vs Late (Age at Diagnosis >55 Years) Ischemic Stroke Among Women and Men
eTable 2. Hazard Ratios and 95% CIs for Early (Age at Diagnosis ≤55 Years) vs Late (Age at Diagnosis >55 Years) Ischemic Stroke According to a Given BMI z Score in Childhood With and Without Adjustment for Birth Weight
Customize your JAMA Network experience by selecting one or more topics from the list below.
Gjærde LK, Gamborg M, Ängquist L, Truelsen TC, Sørensen TIA, Baker JL. Association of Childhood Body Mass Index and Change in Body Mass Index With First Adult Ischemic Stroke. JAMA Neurol. 2017;74(11):1312–1318. doi:https://doi.org/10.1001/jamaneurol.2017.1627
Is childhood body mass index and change in body mass index associated with adult ischemic stroke?
In this population-based cohort study of 307 677 Danish individuals (8899 ischemic stroke cases), above-average body mass indexes at ages 7 to 13 years and increases in body mass index across these ages heightened the risk of ischemic stroke at ages 25 to 55 years but not at older ages. The associations were not influenced by birth weight.
To avoid early ischemic strokes associated with childhood overweight and obesity from occurring, these results suggest that, in addition to weight reduction and maintenance in obese children, other individualized measures may be necessary.
The incidence of ischemic stroke among young adults is rising and is potentially due to an increase in stroke risk factors occurring at younger ages, such as obesity.
To investigate whether childhood body mass index (BMI) and change in BMI are associated with adult ischemic stroke and to assess whether the associations are age dependent or influenced by birth weight.
Design, Setting, and Participants
This investigation was a population-based cohort study of schoolchildren born from 1930 to 1987, with follow-up through national health registers from 1977 to 2012 in Denmark. Participants were 307 677 individuals (8899 ischemic stroke cases) with measured weight and height at ages 7 to 13 years. The dates of the analysis were September 1, 2015, to May 27, 2016.
Main Outcomes and Measures
Childhood BMI, change in BMI, and birth weight. Ischemic stroke events were divided into early (≤55 years) or late (>55 years) age at diagnosis.
The study cohort comprised 307 677 participants (approximately 49% female and 51% male). During the study period, 3529 women and 5370 men experienced an ischemic stroke. At all ages from 7 to 13 years, an above-average BMI z score was positively associated with early ischemic stroke. At age 13 years, a BMI z score of 1 was associated with hazard ratios (HRs) of 1.26 (95% CI, 1.11-1.43) in women and 1.21 (95% CI, 1.10-1.33) in men. No significant associations were found for below-average BMI z scores. Among children with above-average BMI z scores at age 7 years, a score increase of 0.5 from ages 7 to 13 years was positively associated with early ischemic stroke in women (HR, 1.10; 95% CI, 1.01-1.20) and in men (HR, 1.08; 95% CI, 1.00-1.16). Similarly, among children with below-average BMI z scores at age 7 years, a score increase of 0.5 from ages 7 to 13 years was positively associated with early ischemic stroke in women (HR, 1.14; 95% CI, 1.06-1.23) and in men (HR, 1.10; 95% CI, 1.04-1.18). Adjusting for birth weight minimally affected the associations.
Conclusions and Relevance
Independent of birth weight, above-average childhood BMI and increases in BMI during childhood are positively associated with early adult ischemic stroke. To avoid the occurrence of early ischemic stroke associated with childhood overweight and obesity, these results suggest that all children should be helped to attain and maintain healthy weights.
Ischemic stroke is a leading cause of death worldwide and is a major cause of acquired disability and loss of productive life-years, especially in younger individuals.1,2 Women are particularly affected because at young ages they have poorer functional outcomes than men, and at older ages they have a poorer prognosis.3,4 While declining incidence rates of overall stroke have been reported in many high-income countries,5Quiz Ref ID the incidence of ischemic stroke among young adults has increased in some populations, including Denmark.6 This occurrence may be related to earlier exposure, even in childhood, to risk factors for ischemic stroke, such as hypertension, type 2 diabetes, and obesity.1,7
Studies8-13 investigating the associations between childhood body mass index (BMI) and adult ischemic stroke have shown inconsistent results. In 1 study,10 there were indications of a nonlinear association between childhood obesity and later stroke, but the other studies did not investigate the shapes of the associations. Furthermore, only 2 studies8,9 distinguished between major stroke subtypes, only 1 study9 investigated the potential importance of birth weight, and another sole study8 assessed the age at stroke occurrence.
Because the prevalence of childhood obesity is high in many countries,14 it is important to clarify the association between childhood BMI and risk of adult ischemic stroke, especially at younger adult ages, to justify earlier targeted prevention. We investigated if BMI in childhood and changes in BMI during childhood are associated with the risk of first ischemic stroke in women and men. Furthermore, we assessed whether the associations are dependent on the age at stroke occurrence and birth weight.
Quiz Ref IDThe study was based on data from the Copenhagen School Health Records Register,15 which includes computerized information on 372 636 children born from 1930 to 1989 and measured from 1936 to 2002, with follow-up through national health registers from 1977 to 2012 in Denmark. The dates of the analysis were September 1, 2015, to May 27, 2016. As part of a school-based health care program, each child who attended public or private schools in Copenhagen, Denmark, before 1983 underwent an annual examination and thereafter was examined at school entry and exit. Weight and height were measured by trained school physicians and nurses throughout the entire period. This information was recorded on a health card along with the child’s name, date of birth, identification number, and birth weight, as reported by parents, for individuals born from 1936 onward. The analyses were conducted on anonymous data, and the study was approved by the Danish Data Protection Agency. According to Danish law, informed consent and ethical approval are not required for register-based analyses.
We recorded BMI (calculated as weight in kilograms divided by height in meters squared) and transformed it to a BMI z score based on an internal age-specific and sex-specific reference chosen from a period when the prevalence of obesity was low and stable.16 We obtained information on stroke events from the Danish National Cause of Death Register,17 which has been computerized since 1970, and the Danish National Hospital Register,18 which was established in 1977; follow-up was initiated from 1977 onward. In both registers, the International Classification of Diseases (ICD) Eighth Revision before 1994 and 10th Revision thereafter was used to classify ischemic stroke (ICD-8 codes 433-434 and ICD-10 code I63) and unspecified stroke (ICD-8 code 436 and ICD-10 code I64). Diagnoses of ischemic stroke were based on imaging except for the few occurring before the 1990s (522 cases [5.9%]), which were based on clinical assessment. Vital status was obtained by linkage to the computerized Danish Civil Registration System.19
We included women and men with a personal identification number who were at least age 25 years at study entry (corresponding to birth years of 1930 to 1987) and had a minimum of 1 measurement of both weight and height (eFigure 1 in the Supplement). Quiz Ref IDFollow-up ended on the date of a first-ever ischemic stroke diagnosis, the date of death from another cause, loss to follow-up, emigration, or December 31, 2012, whichever occurred first.
We conducted all analyses separately for women and men.4 We used Cox proportional hazards regression models, with age as the underlying time axis, to estimate hazard ratios (HRs) and 95% CIs for ischemic stroke according to BMI z scores at ages 7 to 13 years. Because the patterns of the associations were similar at all ages, we present the results only for age 13 years (other ages are shown in eFigure 2 and eFigure 3 in the Supplement).
Indications of nonlinear associations between childhood BMI z scores and ischemic stroke were detected by testing against a restricted cubic spline model using likelihood ratio tests (P < .05 for deviations from linearity at each age from 7 to 13 years). Therefore, we present the results using a restricted cubic spline based on 4 knot points of childhood BMI z scores. We tested the proportional hazards assumptions by including a time-varying effect of BMI in the model and through assessment using double cumulative hazard plots.20 Because we found violations of proportional hazards indicating that the associations were different for women and men who experienced an ischemic stroke at 55 years or younger vs after 55 years, we stratified the analyses by age. This stratification fulfilled the assumptions of proportional hazards. All analyses were stratified by birth cohort (1930-1934, 1935-1939, 1940-1944, 1945-1949, 1950-1954, 1955-1959, 1960-1964, 1965-1969, 1970-1974, 1975-1979, and 1980-1987), allowing the baseline hazard to differ for each birth cohort. Moreover, we found no consistent differences in the associations by birth cohort when testing for potential interactions.
We also analyzed the associations using Cox proportional hazards regressions by categories based on percentiles of the BMI z score distributions as follows: less than fifth, fifth to less than 25th, 25th to 75th (reference), greater than 75th to 95th, and greater than 95th. We estimated population attributable fractions using the US Centers for Disease Control and Prevention classifications of overweight and obesity for 13-year-old children to illustrate potential contributions of contemporary levels of childhood overweight and obesity on early adult ischemic stroke.21,22
We examined the association of change in BMI z score from ages 7 to 13 years with ischemic stroke by stratifying the analyses by BMI because of the nonlinear associations for level of BMI. Therefore, change in BMI z score, conditioned on the baseline BMI at age 7 years or the attained BMI at age 13 years (both as continuous variables), was investigated among those with a BMI below or above average (BMI z score ≤0 or >0) at ages 7 and 13 years. Likelihood ratio tests demonstrated no potential interactions between change in BMI z score and either baseline or attained BMI (P > .05) within the 2 BMI strata.
Given the reported opposing associations of birth weight and childhood BMI with ischemic stroke,23,24 we performed an analysis in the subcohort with information on birth weight (n = 237 501) to investigate the influence of birth weight, potential interactions between birth weight and BMI, and the association of adjusting for birth weight with ischemic stroke. We used Cox proportional hazards regression models and the likelihood ratio test to investigate potential interactions. All analyses were stratified by birth cohort.
To estimate the HRs for ischemic stroke in the presence of the likely competing risk of death, we conducted a subdistribution hazard regressions model using the approach by Fine and Gray.25 In a sensitivity analysis, a combined group of ischemic and unspecified strokes was examined as the outcome because a Danish study26 showed that strokes coded as unspecified are actually ischemic in approximately 60% of cases, with the remainder being a combination of hemorrhagic stroke and other diagnoses.
Statistical significance was set at P < .05, and all tests were 2 sided. Analyses were performed using statistical software (Stata, version 14.1; StataCorp LLC).
Childhood weight and height measurements at ages 7 to 13 years were available for 307 677 of the eligible individuals (eFigure 1 in the Supplement). The mean BMI and variation in BMI increased with childhood age, as expected (Table 1). The children were followed up from ages 25 through 82 years (median follow-up, 26.4 years), corresponding to 8 128 058 person-years of follow-up. During this period, 3529 women and 5370 men were diagnosed as having ischemic stroke, of which 23.0% and 23.4%, respectively, occurred at 55 years or younger. As expected, the incidence rate of ischemic stroke increased with advancing age in both sexes, with a higher rate in men until age 75 years, after which the rate rapidly increased in women (Figure 1).
Quiz Ref IDAlthough we found no significant associations between BMI and risk of late adult ischemic stroke (age >55 years), there was a nonlinear association of BMI at ages 7 to 13 years with early-occurring (age ≤55 years) ischemic stroke in both sexes (Figure 2 and eFigures 2 and 3 in the Supplement). Positive associations with early adult ischemic stroke were observed for those with an above-average childhood BMI (BMI z score >0), and no significant associations were observed for those with a below-average childhood BMI (BMI z score ≤0). Women with a BMI z score of 1 at age 13 years had an HR of 1.26 (95% CI, 1.11-1.43) for early adult ischemic stroke, and the HR increased with higher BMI z scores, reaching an HR of 1.76 (95% CI, 1.38-2.25) in women who had a BMI z score of 2 (Figure 2). In men, a BMI z score of 1 at age 13 years was associated with an HR of 1.21 (95% CI, 1.10-1.33) for early adult ischemic stroke, and a BMI z score of 2 was associated with an HR of 1.58 (95% CI, 1.27-1.98). Put into perspective, moving from a BMI z score of 0 to a BMI z score of 1 equates to an additional 6.8 kg in girls and 5.9 kg in boys based on children with an average weight and height (44.6 kg and 156.7 cm in girls and 42.5 kg and 159.9 cm in boys) in our population. Using the same comparison, moving from a BMI z score of 0 to a BMI z score of 2 equates to an additional 16.4 kg in girls and 14.8 kg in boys. The categorical analyses showed that, compared with the reference category, having a BMI at age 13 years in the greater than 75th to 95th and greater than 95th percentiles was associated with early ischemic stroke. In women, the HRs were 1.24 (95% CI, 1.04-1.48) and 1.71 (95% CI, 1.30-2.26), respectively, and the HRs in men were 1.22 (95% CI, 1.05-1.43) and 1.77 (95% CI, 1.39-2.24), respectively (Table 2).
In 250 437 women and men with BMI measurements available at ages 7 and 13 years, we examined the associations of change in BMI z score with early and late adult ischemic stroke (Table 2). Because we use an internal reference, the mean changes in BMI z score between ages 7 and 13 years were 0.11 in girls and 0.05 in boys (Table 1). Among women, an increase in BMI z score between ages 7 and 13 years was associated with early adult ischemic stroke, regardless of baseline or attained BMI (Table 3). Similar associations were seen among men with a BMI z score greater than 0, but the associations between change in BMI z score and early adult ischemic stroke disappeared when adjusting for BMI at age 13 years.
We found inverse associations between birth weight and early and late adult ischemic stroke in both sexes (eTable 1 in the Supplement). We detected no interactions between birth weight and BMI at any age from 7 to 13 years (P > .05 at each age using likelihood ratio tests) or change in BMI z score and early adult ischemic stroke in women or men (P > .05 for women and men using likelihood ratio tests). At childhood ages, the associations between BMI and ischemic stroke changed little when adjusting for birth weight (eTable 2 in the Supplement).
When death was treated as a competing risk, the estimates remained virtually unchanged. In the sensitivity analyses combining 6883 ischemic strokes and 8166 unspecified strokes (6086 women and 8963 men), we found nonlinear associations between childhood BMI at ages 7 and 13 years and ischemic plus unspecified stroke in women and men (eFigures 4 and 5 in the Supplement). The patterns of the associations were similar to those when only ischemic stroke was the outcome.
In this large population-based cohort study of 307 677 individuals (8899 ischemic stroke cases), the largest of its kind among studies identified through our literature search, we show that in women and men a childhood BMI above average and a gain in childhood BMI were associated with an increased risk of early (≤55 years) but not late (>55 years) ischemic stroke. The risk was significantly elevated in the heaviest children. In addition, even though low birth weight was associated with early and late ischemic stroke in both sexes, we found that it did not influence the associations between childhood BMI and early ischemic stroke.
We dichotomized age at outcome after observing that the proportional hazards assumption was violated because it indicated an increased hazard of ischemic stroke due to a high childhood BMI occurring before but not after age 55 years. We believe that this rationale is reasonable and valid because it is likely that other, and probably more dominating, risk factors than childhood BMI, such as hypertension and type 2 diabetes, are more prevalent later in life, thus evening out the increased risk of ischemic stroke caused by a high childhood BMI.
Compared with the few other studies that have investigated the associations between childhood body size and ischemic stroke, we examined a wide range of ages at stroke diagnosis separately in women and men. Compared with our investigation, the next largest on this topic was a Finnish study8 of 12 439 women and men (314 ischemic strokes). It reported indications of a protective effect of BMI at age 7 years on total stroke morbidity and mortality but found no differences in the associations for strokes occurring before or after age 55 years. Other studies10-13 have investigated total stroke only in women and men combined and rarely explicitly examined age effects, thus limiting comparability with our work. Our findings support the hypothesis that the increase in the early-occurring ischemic stroke incidence is due to a rise in common stroke risk factors, such as obesity, at younger ages.
In late adolescence and adulthood, BMI has strong associations with stroke. In a study27 on total stroke in life among 2.3 million Israeli adolescents (ages 16-19 years) and in a meta-analysis28 on ischemic stroke that included prospective studies on 2 million adults, the HRs for individuals classified as obese compared with an average BMI group were 2.1 and 1.68, respectively. Because the classification of obesity used in the studies corresponds to a BMI z score of approximately 2 in our study, the reported HRs are similar to our findings for childhood BMI.
The associations between BMI and early ischemic stroke have been examined in a few adult studies.7,29 A case-control study29 on 1201 cases of ischemic stroke diagnosed at ages 15 to 49 years and 1154 controls showed positive associations between higher BMI and early stroke. The associations persisted after accounting for age, sex, race, and smoking status and were attenuated after adjustment for concurrent risk factors, such as hypertension and type 2 diabetes.29 It was beyond the scope of our study to investigate the influence of adult risk factors. Our finding of an association between above-average childhood BMI and early adult ischemic stroke could reflect tracking of body size from child to adult ages. Previously, our group reported that overweight children are likely to become overweight young adults but found that by middle age the risk decreases because most overweight adults were heavy irrespective of their childhood body size.30Quiz Ref ID In the present study, we found that the risk of early ischemic stroke is not limited to the heaviest children but rather that the risk is already significantly increased from the 75th percentile of our BMI distribution. Therefore, our results herein may indicate early initiation of atherosclerotic processes,31 which potentially renders childhood BMI as a prognosticator of early adult ischemic stroke.
Individuals in this study were born from the 1930s through the 1980s. During this period, Danish society changed, with improvements in living conditions and medical advances in stroke prevention and management. The socioeconomic status of a child’s family is inversely associated with later stroke mortality.32 Stroke is associated with several behavioral risk factors, such as cigarette smoking, excess alcohol consumption, and physical inactivity, which in turn are associated with less advantageous socioeconomic status.1 Such information was unavailable in the present study. However, the associations between BMI and ischemic stroke did not differ herein by birth cohort, suggesting that they are attributable to biological factors rather than socioeconomic differences, which likely have shifted over time.33
Our study has several strengths. We investigated a large population-based cohort that included practically every child who attended school in Copenhagen. The health examinations were performed at public and private schools, eliminating selection bias in the cohort. Measurements obtained by physicians and nurses were used, enhancing validity and reliability compared with self-reported data. We had almost complete follow-up through national registers, and universal health care was provided in Denmark without direct costs to the patients throughout the follow-up period. The coverage of each register is high, with virtually every event recorded.17,18 In the Danish National Hospital Register, ischemic strokes are recorded with a high degree of validity (positive predictive value >97%),26 but cardiovascular disease diagnoses are less accurate in the Danish National Cause of Death Register.17 Moreover, the analyses of the combined end points of ischemic and unspecified strokes resembled the results for ischemic stroke only, which is in accord with the probable shift in diagnoses from unspecified stroke to ischemic stroke, likely due to the increased use of imaging techniques in recent years.34
From a public health perspective, the findings from our study are relevant for children of today because we included individuals born as late as the 1980s, when the obesity epidemic rapidly developed.35 Population attributable fraction estimates of early adult ischemic stroke were 3.1% among women and 2.1% among men, which indicated the low overall prevalence of overweight and obese children among our cohort. However, at levels of overweight and obesity found among contemporary American children (approximately 35% in girls and boys),36 the population attributable fractions of early adult ischemic stroke would be as high as 20.3% among women and 16.2% among men. Although these estimates should be interpreted cautiously because they assume an unverified causal association, they nonetheless suggest that the significance of a high childhood BMI on early adult ischemic stroke may be substantial. Because childhood obesity is difficult to prevent and treatment often fails, these findings suggest that the initiation of early personalized medicine may need to be undertaken to reduce the risk of early stroke among these individuals, in addition to weight reduction or maintenance.
The findings of our study suggest that above-average childhood BMI increases the risk of ischemic stroke occurring up until but not after age 55 years, independent of birth weight. Increases in BMI during childhood are positively associated with ischemic stroke occurring before or at age 55 years. Our results highlight that a high burden of early ischemic stroke may emerge because millions of children are now overweight or obese.
Accepted for Publication: May 30, 2017.
Corresponding Author: Jennifer L. Baker, PhD, Department of Clinical Epidemiology, Bispebjerg and Frederiksberg Hospital, The Capital Region, Nordre Fasanvej 57, 2000 Copenhagen, Denmark (email@example.com).
Published Online: August 21, 2017. doi:10.1001/jamaneurol.2017.1627
Author Contributions: Dr Baker had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gjærde, Sørensen, Baker.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Gjærde, Baker.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Gjærde, Gamborg, Ängquist.
Obtained funding: Gjærde, Baker.
Administrative, technical, or material support: Sørensen, Baker.
Study supervision: Gjærde, Sørensen, Baker.
Conflict of Interest Disclosures: None reported.
Funding/Support: Dr Gjærde received support from the University of Copenhagen for the submitted work. Dr Baker was supported by funding from the European Research Council under the European Union’s Seventh Framework Programme (grant 281419) and from the Horizon 2020 European Union’s Framework Programme for Research and Innovation (grant 633595).
Role of the Funder/Sponsor: The funding sources had no involvement in the study design, data collection, analysis, and interpretation of data; writing of the manuscript; or the decision to submit the manuscript for publication.
Additional Contributions: We thank the school health physicians and nurses from the Copenhagen Municipality School Health Services for their work with the children. We appreciate the opportunity of working with the Copenhagen School Health Records Register, which has been built in collaboration between the Department of Clinical Epidemiology, formerly the Institute of Preventive Medicine (Frederiksberg, Denmark), and the Copenhagen City Archives.
Create a personal account or sign in to: