Article Text

Download PDFPDF

Tracking of cardiovascular risk factors across generations: family linkage within the population-based HUNT study, Norway
  1. Kirsti L Vik1,2,
  2. Pål Romundstad3,
  3. Tom IL Nilsen1
  1. 1Department of Human Movement Science, Norwegian University of Science and Technology, Trondheim, Norway
  2. 2Liaison Committee between the Central Norway Regional Health Authority (RHA), Stjørdal, Norway and the Norwegian University of Science and Technology (NTNU), Trondheim, Norway
  3. 3Department of Public Health, Norwegian University of Science and Technology, Trondheim, Norway
  1. Correspondence to Kirsti L Vik, Department of Human Movement Science, Norwegian University of Science and Technology, BEV, SVT, NO-7491 Trondheim Norway; kirsti.lund.vik{at}svt.ntnu.no

Abstract

Background Parent-offspring studies have shown that cardiovascular risk factors cluster within families. However, most studies have assessed the offspring cardiovascular risk factor level at a young age, and whether an association persists into the offspring's adult life is less clear. This study linked information between parents and their adult offspring to investigate the intergenerational association of anthropometric measures, blood pressure, blood lipid levels and physical activity.

Methods The study population consisted of parent and adult offspring pairs (11 931 fathers–sons, 12 563 fathers–daughters, 15 626 mothers–sons and 16 449 mothers–daughters) who participated in the second and third cross-sectional waves of the Nord-Trøndelag Health Study (HUNT 2, 1995–1997 and HUNT 3, 2006–2008). A general linear model and logistic regression were used to estimate the association between the parent and offspring risk factor levels.

Results All continuously measured cardiovascular risk factors under study showed a statistically significant positive association between parents and offspring, except the waist-hip ratio. Adjusted coefficients from linear regression ranged from 0.09 (95% CI 0.07 to 0.11) for waist circumference to 0.29 (95% CI 0.27 to 0.32) for body weight. Moreover, offspring were two to three times more likely to be obese, have a high cholesterol level, or hypertension when comparing extreme categories of the corresponding parental risk factor level. Physically active parents had a lower risk of having physically inactive offspring.

Conclusions The results suggested that cardiovascular risk factors track across generations and persist into the offspring's adult life.

  • Epidemiology
  • Obesity
  • Lipids
  • Blood Pressure
  • Exercise

Statistics from Altmetric.com

Introduction

Cardiovascular risk factors tend to cluster within families,1–3 and several studies have shown that parental obesity and weight gain are associated with a high body mass index (BMI) and increased risk of obesity in offspring.4 ,5 It is likely that both genetic and environmental factors influence cardiovascular risk,6 although a recent study suggests that genetics plays an increasingly important role in explaining the variation in BMI.7 The general rise in energy balance in the population suggests that variance between individuals is genetically based; however, the genetic factors make a minimal contribution to the overall burden of obesity on their own.8

Although previous studies have mainly focused on BMI, a few smaller studies have showed associations between parents and offspring for factors such as blood lipids9 and blood pressure.10 Parent–offspring associations of physical activity have also been examined, but with conflicting results.11 ,12 A recent study showed that objectively measured activity in childhood was correlated with maternal, but not paternal, activity.13 Most family studies have used offspring data obtained in childhood and adolescence, and the evidence that intergenerational associations persist into adulthood is sparse. It has been found that parental BMI is associated with offspring BMI trajectories from childhood to adulthood,14 ,15 and that mid-parental BMI is positively associated with adiposity in adult offspring.16 However, the association between parent and adult offspring levels of other cardiovascular risk factors has not been previously studied. It is important to explore the intergenerational association between parents and their adult offspring because it maximises the potential for predispositions to cardiovascular risk factors to be fully expressed in offspring. It is also interesting to see whether the familial aggregation found when offspring were living at home persists into adulthood, when less of the same environment is shared.

In the present study, we linked information on 56 569 adult parent–offspring pairs who participated in a large population-based health study in Norway. The main objective was to examine the extent to which cardiovascular risk factors such as body mass, blood lipids, blood pressure, and physical activity track across generations, from parents to their adult sons or daughters.

Methods

Study population

The Nord-Trøndelag Health Study (HUNT) is a large population-based longitudinal study conducted in the county of Nord-Trøndelag in Norway, where the total adult population of 20 years and older was invited to participate. The study consists of three cross-sectional waves conducted in 1984–1989 (HUNT 1), 1995–1997 (HUNT 2) and 2006–2008 (HUNT 3). Since the first wave did not obtain information on blood lipid values, the present study includes data from the latter two waves. At HUNT 2, 94 194 people were invited and 66 140 (71.2%) accepted the invitation, whereas at HUNT 3, 94 194 inhabitants were invited, and 50 839 (54%) chose to participate. A non-responder study after HUNT 217 showed that slightly fewer men than women participated, and that the participation was largest in the age group 40–70 years. Another study reported that non-participants had a higher prevalence of cardiovascular disease (CVD), diabetes mellitus and psychiatric disorders, a lower socioeconomic status (SES) and a higher mortality than participants.18

Record linkage

Each participant's record in the HUNT database has the unique 11-digit personal identification number of Norwegian citizens attached. This was used to establish a family linkage at Statistics Norway between parents and their biological offspring who had participated in either HUNT 2 or HUNT 3.

The parent-offspring linkages were constructed separately for fathers and mothers, and were based on three possible parent–offspring combinations of participants in the surveys. In order to reduce the possibility that shared family events, such as accidents and traumas, could influence the results, we first linked parents participating in HUNT 2 with offspring participating in HUNT 3, and then parents participating in HUNT 3 with offspring participating in HUNT 2. This accounted for approximately 85% of the linkages. In the remaining parents and offspring linkages, both generations had participated in the same study wave. To be eligible for participation, information on height and weight had to be recorded. This procedure resulted in 56 569 parent–offspring pairs: 11 931 fathers–sons, 12 563 fathers–daughters, 15 626 mothers–sons and 16 449 mothers–daughters.

Study variables

All participants completed a comprehensive questionnaire on lifestyle and health-related factors, including current CVD, physical activity and smoking. At a clinical examination, standardised measures of anthropometry and blood pressure were obtained by trained personnel. A random (non-fasting) venous blood sample was drawn from all participants. A more detailed description of procedures and methods can be found at http://www.ntnu.edu/hunt.

Anthropometric factors were measured with participants wearing light clothes without shoes. Height was measured to the nearest centimetre and weight to the nearest half kilogram. Waist and hip circumference was measured with a steel band to the nearest centimetre at the height of the umbilicus and at the thickest part of the hip, respectively. Waist-to-hip ratio (WHR) was then calculated. BMI was calculated as weight divided by the squared value of height, and categorised into underweight (<18.5 kg/m²), normal weight (18.5–24.9 kg/m²), overweight (25.0–29.9 kg/m²) and obesity (≥30 kg/m²).

Systolic and diastolic blood pressure was measured three times at 1 min intervals using a Dinamap 845XT (Citricon, Tampa, Florida, USA). The mean of the second and third measures was used in the analyses to avoid an artificially high reading at the first measure. Systolic blood pressure (SBP) was then categorised into normal blood pressure (<120 mm Hg), prehypertension (120–139 mm Hg) and hypertension (≥140 mm Hg).

Total cholesterol, high-density lipoprotein (HDL) cholesterol and triglycerides were measured using enzymatic colorimetric methods. Total cholesterol were categorised into low (<5.2 mmol/l), moderate (5.2–6.2 mmol/l) and high (≥6.2 mmol/l). Blood glucose was measured using an enzymatic hexokinase method.

Leisure time physical activity was assessed from the question: ‘How much of your leisure time have you been physically active during the last year? (Think of a weekly average for the year. Your commute to work counts as leisure time.)’ The participants were asked to report the number of hours of light (no sweat/not being out of breath) and hard (sweat/out of breath) activity with the following response options for each type of activity: None, <1, 1–2 and ≥3 h. Based on this, we constructed four categories: inactive (no or <1 h light, and no hard), low (at least 1–2 h light and/or <1 h hard), moderate (at least ≥3 h light and/or 1–2 h hard) and high (any light and ≥3 h hard) activity per week. The questionnaire did not incorporate the type of activity, or fluctuations in activity levels throughout the year, such as seasonal variations.

Statistical methods

All the analyses described below were performed without weighting, and were conducted separately on pairs of fathers–sons, fathers–daughters, mothers–sons and mothers–daughters. To avoid violation of independence assumptions when one parent was represented with multiple children, all SEs in the regression analyses were adjusted for clusters using the vce (cluster) option in Stata to allow for intragroup correlation.

We calculated adjusted coefficients from linear regression between the parent and offspring levels of continuously measured anthropometric factors (ie, weight, BMI, waist circumference and WHR), blood pressure (ie, diastolic and systolic), blood lipids (ie, total cholesterol, HDL cholesterol and triglycerides) and blood glucose.

In subsequent analyses, we calculated the adjusted mean difference in offspring BMI, SBP and total cholesterol between categories of the corresponding parental variable. Logistic regression was used to calculate crude and adjusted OR for an unfavourable risk factor level in offspring associated with the parental risk factor level. In these analyses, parents classified as underweight (BMI<18.5 kg/m2) were excluded due to the small numbers. The dependent variable in logistic regression was constructed from the cut-offs described above that define the most adverse risk factor level; obesity was defined as BMI≥30 kg/m²; hypertension was defined as SBP>140 mm Hg and/or diastolic blood pressure >90 mm Hg and/or use of blood pressure medication; high cholesterol was defined as total cholesterol >5.1 mmol/l and physical inactivity as no or <1 h light activity and no hard activity per week.

All persons who used blood pressure medication were excluded from the analysis of blood pressure as a continuous measure (8854 parents and 675 offspring), and since no information on lipid-lowering drugs was available, all persons who reported CVD at baseline were excluded from the analysis of blood lipids (7961 parents and 675 offspring).

Additionally, we reanalysed these associations on the total sample adjusting for blood pressure medication and existing CVD.

Potential confounders were selected after construction of directed acyclic graphs to avoid exclusion of important confounders and incorrect adjustments for non-confounders.19 Estimates were adjusted for parental age (continuous) and smoking status (never, former, current), and also parental physical activity (inactive, low, moderate, high activity) in analyses of BMI, SBP and total cholesterol. In a subsample of 28 840 parent-offspring pairs with information on parental education (only available from HUNT 2), we conducted supplementary analyses to evaluate possible confounding with SES. Since most cardiovascular risk factors are strongly age-dependent, we conducted supplementary analyses adjusting for offspring age, although this was not supported by directed acyclic graphs.

Since studies have shown that the relative effect of cardiovascular risk factors can diminish with increasing age,20 ,21 we conducted stratified analyses to evaluate potential effect modification by parental age (<60 vs ≥60 years). We also conducted analyses stratified by parental physical activity (high vs low activity), to look at a possible compensatory effect of activity on the parent-offspring associations. Statistical interaction was assessed in likelihood ratio tests including a product term in the regression model.

The precision of the estimated associations was assessed by a 95% CI; all statistical tests were two-sided, and all statistical analyses were conducted using Stata/SE V.11 for Windows (StataCorp LP).

Results

Descriptive statistics of the 56 569 parent–offspring pairs are shown in table 1. The average age of parents varied from 60.4 to 63.2 years in different parent–offspring sets, whereas the average age of offspring was between 38.3 and 41.3 years.

Table 1

Descriptive statistics* of the parent–offspring pairs in the Nord-Trøndelag Health study, Norway

Table 2 shows the adjusted coefficients from linear regression for each of the continuously measured risk factors. Apart from WHR, all measures showed a statistically significant association between parents and offspring, with the strongest associations observed for body weight, BMI, total cholesterol and HDL cholesterol (regression coefficients between 0.20 and 0.30 for most comparisons).

Table 2

Linear association between the parent and offspring levels of continuously measured cardiovascular risk factors

Figure 1 displays the mean differences in offspring levels of BMI, SBP and total cholesterol between categories of the corresponding parental risk factor level. All variables show a consistent positive association between parents and offspring. For instance, sons and daughters with an obese father had a BMI that was 2.3 (95% CI 2.0 to 2.6) and 2.5 (95% CI 2.1 to 2.8) kg/m2 higher, respectively, than sons and daughters with normal weight fathers. The corresponding association with maternal BMI was 2.0 (95% CI 1.8 to 2.2) and 3.0 (95% CI 2.7 to 3.3) in sons and daughters. Similar analyses comparing extreme categories of parental values of SBP and total cholesterol gave mean differences at a magnitude of 3–5 mm Hg higher SBP and 0.5–0.6 mmol/l higher total cholesterol. Results for blood pressure and blood lipids are based on a restricted sample excluding persons using blood pressure medication or having prevalent CVD, respectively. Supplementary analyses adjusting for these factors in the regression model, instead of restricting the sample, gave similar results.

Table 3 (father–offspring) and table 4 (mother–offspring) show strong and consistent associations for all cardiovascular risk factors. Sons were nearly three times as likely to be obese if the father (OR 2.85, 95% CI 2.40 to 3.39) or the mother (OR 2.85, 95% CI 2.48 to 3.28) was obese, compared with if they were of normal weight. The association among daughters was largely similar, although the result suggest a somewhat stronger effect of maternal (OR 3.36, 95% CI 2.91 to 3.86) than paternal (OR 2.54, 95% CI 2.15 to 3.01) BMI. Comparing extreme categories of parental levels of total cholesterol and SBP gave largely similar results. Finally, there was a statistically significant inverse association between the father’s physical activity level and the offspring's likelihood of being inactive; sons had an OR of 0.68 (95% CI 0.54 to 0.86) and daughters had an OR of 0.62 (95% CI 0.47 to 0.81) if the father was highly active compared with inactive. Maternal physical activity also showed an inverse association if the mother reported a high activity level; the OR for being inactive was 0.64 (95% CI 0.50 to 0.81) in sons and 0.42 (95% CI 0.32 to 0.56) in daughters.

Table 3

OR for an unfavourable risk factor level in offspring associated with the father's risk factor level

Table 4

OR for an unfavourable risk factor level in offspring associated with the mother's risk factor level

Figure 1

Mean differences in offspring levels of body mass index, systolic blood pressure and total cholesterol between categories of the corresponding parental risk factor level. Bars represent 95% CIs. Adjusted for smoking status, age and physical activity level in parents.

Supplementary analysis adjusting for parental education and offspring age did not materially change the above results (data not shown).

Stratified analysis showed that the ORs for offspring obesity, high cholesterol and hypertension were slightly stronger when parents were <60 years at participation compared with ≥60 years (web only file). However, only the father–daughter association of obesity and hypertension showed a statistically significant interaction (both p=0.02). Analyses stratified by the parental physical activity level (inactive vs active) gave no evidence of effect modification (web only file).

Discussion

Main results

In this large, population-based intergenerational study, we found consistent relations between parental and offspring risk factors for CVD. The results suggest that parental BMI, blood pressure and total cholesterol levels were strongly and positively associated with levels of the same risk factors in their adult offspring. Moreover, physically active parents had a lower risk of having physically inactive offspring.

Comparison with the existing literature

Our results supported the previously reported positive associations of parental and offspring cardiovascular risk factors.9 ,22–24 In addition to BMI, weight and waist circumference displayed strong associations; however, no correlation was found for WHR. According to previous studies, waist circumference is more strongly influenced by genetic factors than WHR, possibly because it is more associated with overall obesity, as measured by BMI.25 Stronger associations were found for total cholesterol and HDL-cholesterol than for triglycerides, and systolic and diastolic blood pressure presented similar associations.

Our study differs from previous studies in that the offspring were adults (above the age of 20 years), whereas some previous studies have examined the association between parents and their children. One study found the strength of the association for childhood BMI to be as strong as that for adult BMI.5 Another has indicated that the familial associations of adiposity, blood pressure and total cholesterol levels were substantially higher when comparisons were made when both parents and offspring were children.22 One study of overweight children provided evidence suggesting that parental physical activity could modify the association of cardiovascular risk factor levels,24 but we did not find the parental level of physical activity to be an effective modifier in this relation.

Possible mechanisms

Disentangling the genetic and environmental contributions to cardiovascular risk is difficult, and interactions between the two factors over time can complicate the understanding.5 ,9 ,26 An important genetic influence on BMI, blood pressure and total cholesterol is supported by adoption and twin studies, where a strong correlation has been found between adopted offspring and their biological parents, and between monozygous twins and other pedigrees.23 ,27 ,28 Although several obesity-associated genes (eg, the fat mass-associated and obesity-associated gene and the melanocortin 4 receptor) have been identified, the about 50 genetic loci identified so far can only account for a minor part of the variability in BMI.29 Other cardiovascular risk factors may also be influenced by the underlying correlated genes via pathways mediated by those genes.28 The familial aggregation of hypertension and hyperlipidaemia can be induced by genes, given the appropriate environmental conditions.30

Shared environment and family characteristics such as SES, diet and physical activity may also underlie the associations between the parent and offspring risk factor levels.31 Families with low SES are shown to be more overweight and obese, have higher blood pressure and higher glucose levels.32 However, adjusting for parental education as a measure of SES did not influence the results in the present study. Thus, even though parental education could be an independent risk factor for CVD, it may not be a confounder in the present causal pathway. Alternatively, it is possible that education is not a precise measure for SES in the population under study. Certain behaviours, like an unhealthy diet, inactivity and smoking, can be transmitted from the parent to the offspring when the offspring are living at home, and behavioural patterns can be maintained into adult life.

Strengths and limitations

This study has some obvious strengths, including the large study size and its population-based design, covering a total population within a geographical area with a high participation rate. Other strengths are the standardised measures of anthropometry, blood pressure and blood lipids in both parents and offspring. A unique feature of this study is that the offspring were adults at the time the information was collected, suggesting that familiar risk factor levels persist even though the offspring most likely do not share a household environment with their parents at the time of the study.

If participants had data available from more than one cross-sectional wave, we used information from the earliest point in time, when participants were youngest, in order to limit the possibility that underlying conditions and diseases, such as hypertension, diabetes and hypercholesterolaemia, could have influenced our results. It should be noted that, in addition to adjusting for treatment effects, we conducted supplementary analyses where all persons who reported the use of blood pressure-lowering drugs were excluded from the analysis of blood pressure as a continuous measure, and since existing disease and use of medication might influence risk factor levels, all persons who reported CVD at baseline were excluded from the analysis of blood lipids.

The study has some limitations that should be considered when interpreting the results. Loss to follow-up between the HUNT surveys, and differences between responders and non-responders, could lead to selection bias. If non-responders had more ill-health than responders, the observed associations could be underestimated. Moreover, we cannot exclude the possibility that families in which both parents and offspring have participated could be a selected and conceivably more health-conscious sample than the general population. Information on physical activity was obtained from a self-reported questionnaire, possibly leading to a non-differential misclassification of activity based on seasonal variations, recall bias and subjective interpretation of questions and response options. This may have attenuated the parent-offspring associations in the physical activity analyses. However, validation studies have shown that questionnaires may be useful for classifying people into broad categorisations of physical activity,33 and a recent study showed that the questions used in this study were a good long-term predictor of cardiorespiratory fitness.34

It should also be noted that blood lipids and blood glucose levels were measured in a non-fasting state. However, the potential bias arising from this would most likely be non-differential and attenuate the results. Moreover, it has been shown that lipid levels after normal food intake differ only minimally from levels in the fasting state.35

Given the observational design, we cannot rule out residual confounding due to unknown or unmeasured factors, such as dietary habits.36 However, the family design made it less likely that confounding by such factors have influenced the results. Also, multiple testing in the web only analyses of potential effect modification by parental age and physical activity calls for a cautious interpretation of the results.

The variation in both parental and offspring age at risk factor measurement could have influenced our results. Epidemiological studies suggest that obesity in adults is a stronger risk factor for CVD in younger than in older age, and that fat distribution seems to be more important with increasing age.37 ,38 Some of the participants in the present study may have been too young to have developed cardiovascular risk factors,39 whereas older persons with a low risk factor level may have had a certain genotype that evades premature mortality.40 Analysis stratified on parental age gave largely similar results, and adjusting for offspring age did not change the associations.

The present study compared parents and their biological offspring, but we did not have information on whether the offspring shared environment with one or both of their biological parents, either in adulthood or when growing up. It is possible that the intergenerational association is stronger in offspring who shared environment with their biological parents during childhood.2

Conclusion

This study found consistent associations between parents and their adult offspring for all cardiovascular risk factors studied, except WHR. Offspring were more likely to have unfavourably high levels of BMI, SBP and total cholesterol if their parents had similarly high levels. Physically active parents had a lower risk of having physically inactive offspring. Although the genetic and environmental contributions to these associations should be assessed in future studies, these results contribute to the important understanding that cardiovascular risk factors are clustered within families.

What is already known on this subject

  • Parental obesity is associated with a high body mass index and increased risk of obesity in offspring.

  • Few studies have examined if these associations persists into the offspring's adult life.

  • There is lack of data concerning intergenerational tracking of other cardiovascular risk factors, such as blood lipids, blood pressure, and physical activity.

What this study adds

  • In this large, population-based, intergenerational study we found consistent associations between parents and their adult offspring for all cardiovascular risk factors studied, except waist-hip ratio.

  • Offspring were more likely to have unfavourably high levels of body mass index, systolic blood pressure, and total cholesterol if their parents had similarly high levels, compared to offspring who had parents with lower risk factor levels.

  • Physically active parents had a lower risk of having physically inactive offspring.

Acknowledgments

The Nord-Trøndelag Health Study (The HUNT Study) is a collaboration between the HUNT Research Centre (Faculty of Medicine, Norwegian University of Science and Technology NTNU), Nord-Trøndelag County Council, Central Norway Health Authority and the Norwegian Institute of Public Health.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Contributors TIN initiated the study and acquired the data. KLV was involved in the data preparation and analysis and wrote the first draft of the paper. TIN and PR critically revised the manuscript for important intellectual content. All authors contributed to the final draft.

  • Funding This work was supported by the Liaison Committee between the Central Norway Regional Health Authority (RHA) and the Norwegian University of Science and Technology (NTNU) grant number 2010/2790.

  • Competing interests None.

  • Ethics approval Regional Committee for Medical Research Ethics (project no. 2010/69, REK Midt, Norway).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.