Childhood Infections, Socioeconomic Status, and Adult Cardiometabolic Risk
BACKGROUND AND OBJECTIVES: Socioeconomic disadvantage throughout the life course is associated with increased risk of cardiometabolic diseases, but traditional risk factors do not fully account for the social gradient. We investigated the interactions between low socioeconomic status (SES) and infection in childhood and adverse cardiometabolic parameters in adulthood.
METHODS: Participants from the Cardiovascular Risk in Young Finns Study, a cohort well phenotyped for childhood and adulthood cardiometabolic risk factors and socioeconomic parameters, were linked to lifetime hospitalization data from birth onward available from the Finnish National Hospital Registry. In those with complete data, we investigated relationships between infection-related hospitalization in childhood, SES, and childhood and adult cardiometabolic parameters.
RESULTS: The study cohort consisted of 1015 participants (age range 3–18 years at baseline and 30–45 years at follow-up). In adults who were raised in below-median income families, childhood infection-related hospitalizations (at age 0–5 years) were significantly associated with higher adult BMI (β ± SE comparing those with 0 vs ≥1 hospitalizations 2.4 ± 0.8 kg/m2, P = .008), waist circumference (7.4 ± 2.3 cm, P = .004), and reduced brachial flow–mediated dilatation (−2.7 ± 0.9%, P = .002). No equivalent associations were observed in participants from higher-SES families.
CONCLUSIONS: Infection was associated with worse cardiovascular risk factor profiles only in those from lower-SES families. Childhood infection may contribute to social gradients observed in adult cardiometabolic disease risk factors. These findings suggest reducing childhood infections, especially in socioeconomic disadvantaged children, may reduce the cardiometabolic disease burden in adults.
- CVD —
- cardiovascular disease
- FIM —
- Finnish markkas
- FMD —
- flow-mediated dilatation
- HDL —
- high-density lipoprotein
- hsCRP —
- high-sensitivity C-reactive protein
- ICD —
- International Classification of Diseases
- IMT —
- intima-media thickness
- IRH —
- infection-related hospitalization
- LDL —
- low-density lipoprotein
- SES —
- socioeconomic status
What’s Known on This Subject:
Cardiometabolic and infectious diseases share similar socioeconomic gradients. Acute and chronic infections may alter long-term host immune responses. Early life events may program a maladaptive immune response to vascular injury and contribute to the socioeconomic inequalities in cardiometabolic disease.
What This Study Adds:
Early life infection worsens adult cardiometabolic risk only in people whose socioeconomic position is below the median. Childhood infection may contribute to social gradients observed in adult cardiometabolic disease risk factors and noncommunicable diseases.
Socioeconomic status (SES) is a strong predictor of cardiovascular disease (CVD, coronary heart, cerebrovascular, and peripheral vascular disease) and of metabolic disease (obesity and type 2 diabetes mellitus).1–3 Lower SES is associated with higher prevalence of traditional risk factors4 and increased cardiometabolic disease prevalence and mortality.5 The mechanisms by which early life and childhood social disadvantage lead to increased adult cardiometabolic diseases are multifactorial and are suggested to include biological, behavioral, psychological, and social factors.5 Overall, traditional risk factors do not fully account for the differences in attributable risk.1,6
Increased inflammation throughout the life course is associated with social disadvantage and adverse childhood experiences.7 Chronic inflammation has a central pathogenic role in cardiometabolic diseases.8 Both chronic infections9 and general markers of inflammation10 show a strong social gradient and may contribute to the effect of SES on CVD11 and type 2 diabetes.12 SES alters innate immune13 and cell-mediated immune responses,14 2 examples out of many possible mechanisms by which infection could lead to chronic disease. To date, there are few longitudinal data from well-phenotyped cohorts on the relationships between standardized definitions of childhood infections, SES, and cardiometabolic status in adulthood.
We previously reported that childhood infection-related hospitalizations (IRHs) are associated with adverse cardiometabolic outcomes in early to middle adulthood.15,16 Here we prospectively investigated the interaction between childhood SES and early life (age 0–5 years) infections on cardiometabolic risk markers in adulthood (age 30–36 years) among 1015 people in the longitudinal Cardiovascular Risk in Young Finns Study.
The Cardiovascular Risk in Young Finns Study is an ongoing prospective study of cardiovascular risk factors from childhood to adulthood. The baseline examination was in 1980, when participants were aged 3 to 18 years, with repeated follow-up assessments in 1983, 1986, 2001, and 2007.17 The current sample included 1015 people with entire lifetime hospitalization data extracted from the Finnish national hospitalization database (which commenced in 1969) and who had participated in the 27-year follow-up study in 2007. Baseline risk factors of those retained in follow-up are largely comparable to those of nonparticipants.18 The study complies with the Declaration of Helsinki and has institutional ethics approval. Written informed consent was obtained from all participants.
In childhood, questionnaires completed by the parents of the participant were used to obtain data on physical activity, birth weight, prematurity, mother’s BMI, family income, parental years of education, fruit and vegetable consumption, and parental smoking. Physical activity was assessed with questions about the frequency and intensity of physical activity, and a physical activity index was calculated based on the variables as previously described.19 There were 2 different kinds of physical activity questionnaires for the younger (3- to 6-year-olds, a parent-completed questionnaire) and older children (9- to 18-year-olds, self-completed questionnaire). The calculated physical activity indices were age standardized to allow comparison across age groups. Annual family income strata at the time of enrollment were determined as follows: category 1, <12 500 Finnish markkas (FIM) (∼5850 euros); category 2, 15 001 to 25 000 FIM; category 3, 25 001 to 35 000 FIM; category 4, 35 001 to 45 000 FIM; category 5, 45 001 to 55 000 FIM; category 6, 55 001 to 75 000 FIM; category 7, 75 001 to 100 000 FIM; category 8, >100 000 FIM. We also analyzed SES by using parental years in education as a measure. In adulthood, questionnaire data were used to gather information on annual income, smoking, diet, and physical activity.
Definition of Infection-Related Hospitalization
IRH was defined as a hospital discharge diagnosis that included ≥1 International Classification of Diseases (ICD) infection-related code as either a primary or secondary code. Hospitalization was defined as an admission that included ≥1 overnight stay. We used both primary and secondary codes to ensure capture of all infections, an approach we and others have used previously.16,20 We selected infection-related ICD codes (ICD versions 9 and 10) a priori, based on a modification of published population-based epidemiologic studies of childhood IRH.15 To investigate possible infection-specific effects, we grouped infection-related codes a priori into clinical diagnostic categories by using a modification of methods described previously.20 Early childhood was defined as birth to 5 years of age, when the infection burden is greatest.21 Data on antibiotic usage in hospital or in the community were not available.
Anthropometric and Clinical Assessment
In all examinations, height and weight, rounded to the nearest 0.5 cm and 0.1 kg, respectively, were measured at all time points via standardized protocols, and BMI was calculated as weight (in kilograms) divided by height (in meters) squared.18 Waist circumference (measured in duplicate at the level of the 12th rib or level with the umbilicus in thin subjects) was measured in adults. Enrollment blood pressure at 3 years of age was measured by ultrasound and at other childhood ages by a mercury sphygmomanometer. A random zero sphygmomanometer was used in adults. The first and fifth Korotkoff sounds were used to define systolic and diastolic blood pressures, which were averaged from 3 measurements. Blood samples were obtained after a 12-hour fast. Standard enzymatic methods were used for serum total cholesterol, triglycerides, high-density lipoprotein (HDL) cholesterol, and plasma glucose. HDL cholesterol was measured after dextran sulfate precipitation, and low-density lipoprotein (LDL) cholesterol was calculated with the Friedewald formula. High-sensitivity C-reactive protein (hsCRP) was measured by an automated analyzer via a latex turbidimetric immunoassay. For hsCRP analyses, childhood serum samples were taken in 1980 and stored at −20°C. These samples were analyzed in 2005. During storage, the samples were not thawed or refrozen.
Common carotid and brachial artery ultrasound studies were performed on Sequoia512 ultrasound mainframes (Acuson, Mountain View, CA) with a 13.0-MHz linear array transducer in 2001 and 2007 follow-ups, as previously described.22 The digitally stored scans were manually analyzed by a single observer blinded to subjects’ details (M.J.). To assess intraindividual reproducibility of ultrasound measurements, 57 subjects were reexamined 3 months after the initial visit (2.5% random sample).
Carotid Intima-Media Thickness
At least 4 measurements of the far wall of the left carotid artery were taken ∼10 mm proximal to the bifurcation to derive mean and maximum carotid intima-media thickness (IMT). The between-visit coefficient of variation of IMT measurements was 6.4%.
Ultrasound loops of the carotid bifurcation and common carotid artery were acquired and stored in digital format, and the best cardiac cycle was selected for subsequent offline analysis. The carotid diameter was measured at least twice (spatial measurements) in end-diastole and end-systole, respectively. Blood pressure was measured during the ultrasound study with an automated sphygmomanometer (Omron M4; Omron Matsusaka Co, Ltd, Japan). Ultrasound and concomitant brachial blood pressure measurements were used to calculate carotid distensibility with the following formula:where Dd is the diastolic diameter; Ds, the systolic diameter; Ps, systolic blood pressure; and Pd, diastolic blood pressure. The between-visit coefficient of variation was 2.7% for diastolic diameter and 16.3% for distensibility index.
Brachial Flow-Mediated Dilatation
The left brachial artery diameter was measured at rest and during reactive hyperemia. Increased flow was induced by inflation of a pneumatic tourniquet placed around the forearm to a pressure of 250 mm Hg for 4.5 minutes, followed by release. Three measurements of arterial diameter at a fixed distance from an anatomic marker were performed at end-diastole at rest and 40, 60, and 80 seconds after cuff release. The vessel diameters in scans after reactive hyperemia were expressed as the percentage relative to resting scan. The average of 3 measurements at each time point was used to derive the maximum flow-mediated dilatation (FMD) (the greatest value between 40 and 80 seconds after cuff release). The between-visit correlation of variation was 3.2% for brachial diameter and 26.0% for FMD.
Group comparisons were performed with t tests and χ2 tests, as appropriate. To examine whether the association of early child IRHs with adult cardiometabolic outcomes differed by SES, we used logistic regression modeling. Family income, early child IRH, and family income × early child IRH interaction terms were used in these models as explanatory variables. Thereafter, the effects of early child IRH on those outcomes with significant interaction were analyzed with linear regression models adjusted for age, gender, and other childhood risk factors (BMI, LDL cholesterol, HDL cholesterol, triglycerides, systolic blood pressure, fruit consumption, physical activity, maternal BMI, and parental smoking) separately among participants with family income below or above the median (Fig 1). In addition, we performed sensitivity analyses by using a lowest quartile as a cutoff point for lower SES and using parental years of education as a socioeconomic measure.
In initial analyses, childhood SES significantly interacted with child IRH before age 5 years in predicting adult BMI. We therefore performed life-course analysis of BMI by using multilevel mixed modeling with maximum likelihood estimation. Although a significant interaction between child IRH and SES was also demonstrated for adult waist circumference and brachial FMD in these analyses, these data were only collected at 2 out of 6 data collection time points and therefore were not considered for life course analyses. BMI trajectories were compared as a function of age for 4 groups: not hospitalized for child infection before 5 years of age and above median family income in childhood, not hospitalized for child infection before 5 years of age and below median family income in childhood, hospitalized for child infection before 5 years of age and above median family income in childhood, and child infection before 5 years of age and below median family income in childhood. All analyses were adjusted for gender and time (a categorical age variable). We fitted interaction terms between the IRH and SES groups and time to compare the trajectory of BMI between groups. These analyses allow the age at which any differences in BMI occur between the groups to be identified. Our models consider correlations between repeated measures on the same participant and allow for missing data. Statistical analyses were performed in SAS 9.3 (SAS Institute, Inc, Cary, NC) or, in the case of the life course models, in Stata 13.1 (Stata Corp, College Station, TX).
Characteristics of the study cohort are shown in Table 1. Rates of early childhood hospitalization with infection did not differ between those of high and low SES (11.6% vs 15.4%, respectively, P = .08). Other childhood comorbidities did not differ significantly between groups (Supplemental Table 3). In childhood, participants with family income below the median had lower HDL cholesterol, less fruit and vegetable consumption, and higher triglycerides levels; their mothers had higher BMI. In adulthood, these participants had lower annual income, vegetable consumption, and physical activity levels and higher BMI, systolic blood pressure, and rates of smoking.
Significant interactions between childhood family income and IRH were observed for adulthood BMI, waist circumference, and brachial FMD (Table 2) but not for carotid IMT or distensibility. In analyses performed separately for participants with family income below or above median level within the cohort, early child IRHs were associated with higher adult BMI (β ± SE comparing those never hospitalized with those with ≥1 hospitalizations, 2.4 ± 0.8 kg/m2, P = .008) (Fig 2) and waist circumference levels (7.4 ± 2.3 cm, P = .004) (Fig 3), independent of age, gender, and other childhood risk factors. This interaction was observed only in participants with lower than median family income. Similarly, there was an inverse association between childhood IRH and reduced brachial FMD only among participants with family income below cohort median level (−2.7 ± 0.9%, P = .002) (Fig 4). Findings were similar in an analysis additionally adjusted for significant possible confounders, including adult BMI in the full cohort, and for birth weight and childhood hsCRP in a subcohort with complete data on these variables (all P values <.005 in analyses in those with family income below median and all P values >.1 in analyses in those with family income above median). In addition, sensitivity analyses using a cutoff point of lowest quartile for family income (Supplemental Figures 6 and 7) or parental years of education as a proxy for SES (Supplemental Figures 8 and 9) gave similar findings to the primary analyses.
We also performed additional analyses taking into account both childhood and adult SES. Among participants with low SES in childhood and high SES in adulthood, those with early child IRHs had significantly higher BMI (28.4 ± 1.2 vs 25.8 ± 0.5 kg/m2, P = .04). In this group there was no difference in FMD (7.9% ± 0.9% vs 8.8% ± 0.4%, P = .64). Among those with low SES in both childhood and adulthood, early child IRHs were associated with decreased brachial FMD (7.0 ± 0.7 vs 10.1 ± 0.3%, P = .002), but a significant difference was not observed in BMI (27.8 ± 1.1 vs 25.4 ± 0.3 kg/m2, P = .12) (Supplemental Table 4)
In Fig 5 life course BMI levels are shown according to early child IRHs and family income in childhood. The most prominent differences became evident at the age of 24 years. At ages 24, 30, and 36 years participants without early child IRHs and high family income had significantly lower BMI than the other 3 groups.
This longitudinal follow-up of Finnish children and adolescents into adulthood suggests that childhood IRHs are a significant predictor of increased BMI, waist circumference, and reduced brachial FMD in adults raised in lower-income families but not in those raised in families with an income above the median. These findings were unchanged in sensitivity analyses using an alternative proxy of SES (duration of parental education) and in comparisons of the lowest SES quartile with the remainder of the population. Differences in cardiometabolic risk become increasingly apparent over the life course; sustained and significant differences in adulthood BMI were evident between groups defined by combined childhood exposures (SES and IRH).
Our data support previous findings related to socioeconomic inequalities in cardiometabolic disease.1,2 The Whitehall II study showed that elevated levels of inflammatory markers account for part of the excess risk of type 2 diabetes associated with (retrospectively assessed) life course socioeconomic disadvantage.12 In the current study, we did not observe an association between childhood IRH, low SES, and adult systemic inflammation.
BMI, waist circumference, and brachial FMD are intermediate cardiometabolic risk phenotypes.23,24 Obesity, particularly in childhood and maintained into adulthood,18 is a well-documented independent risk factor of later CVD.25 Our data show an interaction between childhood infection and intermediate cardiometabolic risk phenotype outcomes only in the context of low childhood SES. Low SES potentiates the effects of cardiovascular stress responses on the progression of carotid atherosclerosis,26 and our results support the view that changes in vascular function, evidenced through brachial FMD, occur before evident vascular structural changes in IMT.15 In additional analyses we demonstrate that vascular changes are unlikely to be mediated by obesity alone, because adjustment for adult BMI did not alter the interaction between childhood infection and brachial FMD significantly.
A plausible explanation for our findings is that there is a significantly better overall risk factor status among high-SES participants, who had better lipid and dietary profiles in childhood, as well as higher vegetable consumption and physical activity levels, lower BMI and systolic blood pressure levels, and lower smoking prevalence in adulthood. We adjusted for these factors and the interaction effect remained, but it is possible that residual confounding factors may still affect the relationship.
The strengths of this study include the completeness of the data and the depth of phenotyping for traditional risk factors throughout the life course. We have standardized and complete statutory data on infection from birth onward for almost 30 years. In the Finnish national database, ICD-based diagnoses are recorded for every hospitalization shortly after discharge by dedicated coders. Consequently, the hospitalization diagnoses will be more reliable and much less prone to bias than retrospective diagnoses from hospital records. Additionally, there are multiple measures of SES during childhood. A significant interactive effect of parental education and IRH suggests that social determinants in addition to relative financial hardship may be involved.27
We have previously shown in an Australian population that childhood hospitalization with infection is associated in a dose–response manner with cardiovascular events in adulthood.28 Although childhood infections, and the associated inflammatory burden, are a potential mechanism through which childhood socioeconomic disadvantage increases cardiometabolic risk, the current study was not designed to investigate causal mechanisms. However, our interpretations can be partially informed by the data. For example, there were no significant differences in the frequency of childhood IRHs between families above and below the median income level. This finding suggests that childhood SES is not a confounding factor in the association between childhood IRH and adult cardiometabolic risk factors. Instead, IRH might lie on a pathophysiological pathway between SES and CVD. Additional investigation is needed to examine the long-term impact of serious childhood infections on clinical cardiovascular events and explore causal pathways for the present findings.
We acknowledge some limitations. Given the young age of the cohort, we can assess only intermediate cardiometabolic phenotypes rather than disease outcomes. However, these phenotypes track from childhood into adulthood, where they are known to be strongly predictive of later disease.29 We are unable to assess whether IRH reflects community or nosocomial infections, although in childhood, infections are largely acquired outside the hospital. Length of hospitalization is also not available from the data. The sample size did not permit meaningful analyses by different clinical groupings of infection. We cannot comment on total infection burden in the cohort; most infections, including those implicated in cardiometabolic diseases, do not usually result in hospitalization. Other data on childhood infections, including primary care, emergency department, and parental data, were not available. In addition, we cannot discount social disadvantage, rather than clinical severity, as a possible contributory factor to the decision to hospitalize some children with infection. Socially disadvantaged children may seek clinical care later, and therefore some infections may be more severe by presentation, necessitating hospitalization.
We are unable to ascribe causation, and it is possible that increased infectious burden represents a poorer overall childhood environment, leading to other adverse exposures that increase future cardiometabolic risk. We should also consider important unmeasured confounders, particularly antibiotic exposures early in life, which may be biologically relevant. There is growing evidence for the role of an altered microbiome in the pathogenesis of obesity in mice and humans.30 Antibiotics can modify the microbiome,31 especially in the context of serious infections that necessitate hospitalization, where use of broad-spectrum antibiotics is common. Detailed prospective studies that capture total infection burden and antibiotic use are needed to address this issue.
We report prospective data showing that early infectious exposures may contribute to social gradients in adult cardiometabolic risk. Replication of these findings in other populations and additional mechanistic studies are warranted to facilitate novel interventions aimed at reducing the growing burden of adult cardiometabolic diseases.
We thank the study participants and their families. We also are grateful to Ville Aalto, MSc, for assistance with data cleaning and initial analysis.
- Accepted March 16, 2016.
- Address correspondence to David P. Burgner, MD, PhD, Murdoch Childrens Research Institute, Melbourne Children’s Campus, 50 Flemington Rd, Parkville, Victoria 3052, Australia. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: This work within the Cardiovascular Risk in Young Finns Study has been financially supported by the Academy of Finland; the Social Insurance Institution of Finland; the Kuopio, Tampere, and Turku University Hospital Medical Funds; the Paulo Foundation; the Juho Vainio Foundation; the Paavo Nurmi Foundation; the Finnish Foundation of Cardiovascular Research; the Finnish Cultural Foundation, the Finnish Medical Foundation, the Sigrid Juselius Foundation; Maud Kuistila Foundation; and the Tampere Tuberculosis Foundation, Emil Aaltonen Foundation, and Yrjö Jahnsson Foundation. Drs Liu, Burgner, Sabin, and Magnussen are supported by National Health and Medical Research Council (Canberra, Australia) fellowships and scholarships. Dr Burgner is an Honorary Future Leader Fellow of the National Heart Foundation of Australia. Dr Kivimäki is supported by the Medical Research Council (grant K013351), the Economic and Social Research Council, and NordForsk, the Nordic Council of Ministers (grant 75021). Drs Juonala, Sabin, and Magnussen are supported by the National Health and Medical Research Council (Canberra, Australia) (grant 1098369). Research at Murdoch Childrens Research Institute is supported by the Victorian Government’s Operational Infrastructure Support Program (Melbourne, Australia). The Heart Research Group at Murdoch Childrens Research Institute is supported by the Royal Children’s Hospital (RCH) Foundation (Melbourne, Australia).
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Lynch JW,
- Kaplan GA,
- Cohen RD,
- Tuomilehto J,
- Salonen JT
- Pulkki-Råback L,
- Elovainio M,
- Hakulinen C, et al
- Libby P,
- Ridker PM,
- Maseri A
- Raitakari OT,
- Juonala M,
- Rönnemaa T, et al
- Murray CJ,
- Vos T,
- Lozano R, et al
- Juonala M,
- Kähönen M,
- Laitinen T, et al
- Dobbelsteyn CJ,
- Joffres MR,
- MacLean DR,
- Flowerdew G
- Yeboah J,
- Crouse JR,
- Hsu F-C,
- Burke GL,
- Herrington DM
- Lu Y,
- Hajifathalian K,
- Ezzati M,
- Woodward M,
- Rimm EB,
- Danaei G; Global Burden of Metabolic Risk Factors for Chronic Diseases Collaboration (BMI Mediated Effects)
- Lorenz MW,
- Markus HS,
- Bots ML,
- Rosvall M,
- Sitzer M
- Copyright © 2016 by the American Academy of Pediatrics