Article Text

This article has a correction. Please see:

Download PDFPDF

Physical activity and cardiovascular risk factors in children
  1. Lars Bo Andersen1,2,
  2. Chris Riddoch3,
  3. Susi Kriemler4,
  4. Andrew Hills5
  1. 1Department of Exercise Epidemiology, Center for Research in Childhood Health, University of Southern Denmark, Odense, Denmark
  2. 2Department of Sports Medicine, Norwegian School of Sport Sciences, Oslo, Norway
  3. 3Department for Health, University of Bath, Bath, UK
  4. 4Swiss Tropical and Public Health Institute, University of Basel, Basel, Switzerland
  5. 5Griffith University and Mater Medical Research Institute, Queensland, Australia
  1. Correspondence to Lars Bo Andersen, Department of Exercise Epidemiology, Center for Research in Childhood Health, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark; lboandersen{at}


Background A number of recent systematic reviews have resulted in changes in international recommendations for children's participation in physical activity (PA) for health. The World Health Authority (WHO) has recently released new recommendations. The WHO still recommends 60 min of moderate to vigorous physical activity (MVPA), but also emphasises that these minutes should be on top of everyday physical activities. Everyday physical activities total around 30 min of MVPA in the quintile of the least active children, which means that the new recommendations constitute more activity in total compared with earlier recommendations.

Objective To summarise evidence justifying new PA recommendation for cardiovascular health in children.

Methods The results of recent systematic reviews are discussed and supplemented with relevant literature not included in these reviews. PubMed was searched for the years 2006–2011 for additional topics not sufficiently covered by the reviews.

Results PA was associated with lower blood pressure and a healthier lipid blood profile in children. The association was stronger when a composite risk factor score was analysed, and the associations between physical fitness and cardiovascular disease (CVD) risk factors were even stronger. Muscle strength and endurance exercise each had an effect on blood lipids and insulin sensitivity even if the effect was smaller for muscle strength than for aerobic exercise. New evidence suggests possible effects of PA on C-reactive protein.

Conclusion There is accumulating evidence that PA can have beneficial effects on the risk factors of CVD in children. Public health policy to promote PA in children, especially the most sedentary children, may be a key element to prevent the onset of CVD later in the children's lives.

Statistics from

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.


This overview includes a consideration of literature published since 2005 summarising associations between PA and cardiovascular disease (CVD) risk factors in children. In 2005, Strong et al1 published a comprehensive systematic review of the effects of physical activity (PA) on seven different health outcomes. The review identified 850 articles, which included CVD risk factors such as blood lipids, blood pressure and clustered CVD risk factors. A number of studies reporting associations with haemostatic factors and markers of inflammation were also included. Based on the review, Strong et al recommended that children should accumulate 60 min of moderate to vigorous (MVPA) every day. This recommendation was primarily based on a judgment of the amount of PA included in different intervention studies. It is important to note that none of the intervention studies reported how active children were beside the intervention, and despite that recommendations included information regarding total MVPA. Moreover, few studies at that time had collected objective measures of PA.

Since this review, studies have been published where PA has been assessed objectively. Another improvement has been to analyse a composite score of CVD risk factors. These studies generally report stronger associations because CVD risk factors tend to cluster in sedentary and obese children. A further area of development has been studies analysing the association between PA and inflammatory markers.

Previous recommendations for PA in children have suggested a total of 60 min of MVPA per day. However, our knowledge of total PA levels of children has been limited, primarily because activity has been assessed by self report – a method known to carry unacceptable levels of error in terms of quantifying PA. The emergence of more precise, objective methods of assessing activity has greatly enhanced our understanding in this field. A further issue is that previous recommendations have been based primarily on the results of generally small intervention studies. However, interventions generally do not take the PA of daily living into account. They only quantify the activity added to everyday living. This may have caused an underestimation of the total PA necessary to maintain cardiovascular health.

A further review by Janssen and LeBlanc2 updated the literature on PA and health. However, due to the large number of studies in the area, they limited their search to studies that reported outcomes as dichotomous variables. Limiting studies in this way effectively excludes most of the more recent (and larger) studies that have used objective measures of PA, because these studies have often treated the outcome variables as continuous variables. We believe this to be a major limitation of this review for two reasons. First, the studies utilising the more precise measurements of PA have been omitted. Yet, their greater measurement precision probably gives their results greater validity and hence importance. Second, the process of dichotomising outcome variables reduces power and hence associations will tend to be weaker. Furthermore, most of the CVD risk factors show linear relationships with PA in adults.

The aim of this review is to summarise results from the previous reviews of Strong et al1 and Janssen et al,2 and to extend their findings with (1) studies relating objective measures of PA to CVD risk factors,3 (2) recent studies including inflammatory markers as outcomes and (3) studies analysing CVD risk using composite – or ‘clustered’ – CVD risk scores.4


A PubMed search (2005–January 2011) was undertaken for publications in English related to PA and individual biological CVD risk factors in children (excluding obesity, which is reported in a separate article). PubMed was then searched to identify review articles. Additional articles were added from reviews and reference lists from other articles with special emphasis given to studies relating objective measures of PA to cardiovascular risk factors. Finally, a search was conducted to identify studies analysing composite risk factor scores for CVD risk factors.


Blood pressure

A meta-analysis from 2003 indicates no clear association between PA and blood pressure in normotensive children,5 while there are some indications that prolonged programs can lower blood pressure in hypertensive children. Some studies of children with systemic hypertension show a beneficial effect of aerobic activity programs of 12–32 week duration on blood pressure,6,,8 but an 8-week strength training program by itself had no influence on blood pressure in hypertensive children.9 10 The lack of effect of exercise on blood pressure in this study may be related to the short duration of the training intervention, because interventions with longer duration do find a gradual increase in effect.8 Strength training after an aerobic activity intervention has been shown to prevent the return of blood pressure to preintervention levels in hypertensive adolescents.11 12

Observational studies have found a dose–response association between aerobic fitness and blood pressure,13 14 but these studies found no association between self-reported sports participation and blood pressure. The inverse association between aerobic fitness and blood pressure was stronger in overweight children.14 Associations between aerobic fitness and hypertension were moderate in magnitude with ORs for hypertension of 1.5–3.0 for the least fit. In the study of Nielsen and Andersen,14 risk was only elevated in the lowest quintile of fitness, but in the study of Andersen,13 an inverse, graded association between fitness and blood pressure was observed from a fitness level below 50 and 45 ml/min/kg in boys and girls, respectively.

Experimental studies have almost entirely focused on children with hypertension6 7 11 12 or obesity,8 15 16 and most studies had small sample sizes. The two largest studies included 996 and 67 children, respectively.8 Hansen et al studied both normotensive and hypertensive overweight children. They observed a reduction in systolic and diastolic blood pressures in the training subgroups of 6.5 and 4.1 mm Hg, respectively, in the normotensive group, and 4.9 and 3.8 mm Hg, respectively, in the hypertensive group, after 8 months of training. Most intervention studies have included between 60 and 180 min/week of prescribed exercise. This equates to 9–30 min/day when averaged over a week. Overall, the results from these intervention studies were positive with reports of significant reductions in systolic blood pressure in response to aerobic exercise training, with effect sizes in excess of 0.80.6,,8 11 12 15,,17 Three of the interventions including aerobic training also reported significant reductions (∼6–11%) in diastolic blood pressure.8 11 15 Studies that used training modalities other than aerobic exercise, such as muscular resistance exercise, were less conclusive with small to modest effect sizes being observed.2

In conclusion, these data suggest that a PA/exercise intervention with a duration of at least 30 min, a frequency of 3 times/week and intensity sufficient to improve aerobic fitness can be effective in reducing blood pressure in children with hypertension.

Blood lipids

In observational studies, associations between PA and total cholesterol, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C) and triglyceride levels are generally weak.18 Nevertheless, associations suggest an overall beneficial effect of PA on HDL-C and triglyceride levels, but no consistent effect on total cholesterol or LDL-C levels.1

Dobbins et al19 published a Cochrane review in 2009 including seven school-based randomised controlled trials (RCTs) that included a measure of lipids as outcome measure in response to a PA intervention. This review identified seven studies.20,,26 Two further studies have been published since this review.27 28 Overall, studies including clinical or school-based trials (randomised and non-randomised) show a weak beneficial effect on HDL-C and triglyceride levels, but no effect on total cholesterol or LDL-C levels.29,,31 School-based interventions have generally not been effective in improving lipid and lipoprotein levels, but many of these interventions have also failed to increase PA or fitness.24 32 In a recent 2-year school-based intervention, which did increase physical fitness, beneficial effects were also found on triglyceride (13% decrease) and the ratio of total:HDL cholesterol (6% decrease).27 Similar beneficial changes in blood lipids were observed in a Swiss school-based intervention.28 Beneficial effects are usually seen when the intervention is substantive enough to change aerobic fitness.33

The review of Janssen and LeBlanc2 was limited to studies with dichotomous outcomes. As discussed earlier, most observational studies report linear associations rather than risk of hypercholesterolaemia. This is in contrast to blood pressure, where the existence of hypertension as a dichotomised variable is commonly used. Some of the observational studies contradict findings of the intervention studies that commonly report that HDL rather than total cholesterol is associated with PA or fitness. Data from the European Youth Heart Study suggest weak but highly significant associations between objectively measured PA and total cholesterol and triglyceride, but no association with HDL.18 In the same project, triglyceride and HDL have also been reported to be associated with indicators of muscle fitness such as handgrip strength, situps, muscular endurance and a composite score of these tests.34

Intervention studies conducted outside school tend to only include children with hypercholesterolaemia or obesity. Janssen and LeBlanc2 included six RCTs and two non-randomised interventions. Results from these studies are mixed. The five studies that were based on aerobic exercise alone observed significant improvements in at least one lipid/lipoprotein variable. The interventions that were based on resistance training35 and circuit training15 reported small and/or insignificant changes for all of the lipid/lipoprotein variables examined, and the effect sizes within these studies tended to be quite small.

It appears that a minimum of 40 min of moderate activity per day, 5 days per week for at least 4 months is required to achieve improvement in lipid and lipoprotein levels, primarily increased HDL-C and decreased triglyceride levels.1 This implies the need for a sustained amount of MVPA on a regular basis in order to induce and maintain a beneficial effect. Observational studies indicate a graded association between amount of PA and blood lipid levels.

Clustered CVD risk

Metabolic syndrome (MetS) was first described in adults, but retrospective evaluation of paediatric data suggests that MetS characteristics exist in 3–14% of the general population of children and in 13–37% of obese children.36 Clustering of CVD risk factors is based on the fact that CVD risk factors are not independently distributed in the population but cluster in some individuals. Although there appears to be consensus regarding the risk factors for MetS in adults, there is no consensus for the definition in children and adolescents. The reason for the lack of consensus lies in the fact that children do not routinely exhibit CVD; thus, it is difficult to relate the criteria to a health outcome. The condition is therefore defined as clustered CVD risk by some authors, while others still use MetS, even if definitions differ between authors. In addition, in early stages of insulin resistance, fasting blood glucose is not elevated, because the resistance is compensated by a large increase in insulin production. Fasting glucose is therefore a problematic component of the MetS in children. A number of suggestions for defining cutoff points for MetS in children have been published.37,,39 Alternatively, composite scores based on the sum of percentile ranking and sum of sex- and age-specific z-scores have also been used.18 40,,42 The different CVD risk factors usually included in the definitions are waist circumference (or BMI), triglycerides, blood pressure, fasting glucose and reduced HDL-C level. It would make sense to include physical fitness as part of the score, as it is difficult to find a rationale for including measures of fatness but not fitness, considering that both these conditions are strongly associated with clustering of other CVD risk factors such as insulin resistance, blood lipids and blood pressure.

Studies relating self-reported PA to MetS are inconclusive. Pan and Pratt43 used the PA section of the NHANES 1999–2002 questionnaire and found no significant relationship between PA and MetS in 4450 adolescents aged 12–19 years. Andersen et al44 also reported no association between self-reported PA and MetS in 305 children participating in the Danish Youth and Sport Study. Conversely, Moore et al,45 using the Youth Risk Behavior Surveillance, found that those children reporting low PA had three times the risk of MetS compared with children with high PA levels. McMurray et al46, also using a validated questionnaire, found that children who developed MetS as adolescents had 22% lower PA scores than those children who did not develop MetS. Kelishadi et al,47 also using a PA questionnaire in 4811 adolescents aged 6–18 years, found a small difference in the overall rate of MetS between tertiles of PA. Children with low PA levels were 1.6–1.8 times more likely to have MetS. In the review of Janssen and LeBlanc,2 the summary OR for the least active group relative to the most active group was 1.68 for self-reported PA (95% CI: 1.22 to 2.31).

The findings of a relationship between PA and MetS have been more consistent when accelerometry was used to estimate PA. Brage et al48 determined the relationship between accelerometer-measured PA and MetS z-score in 389 Danish children. They found a graded negative association between PA levels and MetS z-score. A study of 529 Swedish children aged 9 and 15 years also found an inverse relationship between PA and MetS, particularly in 15-year-old girls.49 They suggested that the relationship was strongest because the 15-year-old girls had the lowest levels of PA. Other studies from the European Youth Heart Study measured PA levels with accelerometry in 1730–2800 children aged 9 and 15 years and used a z-score classification for MetS.18 41 They found a graded association in MetS z-score through all PA percentiles. Of note, accelerometry data suggested that MVPA had to be about 90 min/day to effectively reduce the risk of MetS.

Rizzo et al49 analysed a composite CVD risk factor score against quartiles of fitness and found strong associations in both boys and girls. Further examination of these later studies revealed a clear dose–response relation. A recent follow-up of Copenhagen Schoolchild Intervention Study by Andersen et al showed that the association between fitness and MetS became much stronger in the same children from 6 to 9 years of age. The OR between upper and lower quintiles increased from 2 to 35 in this age span, and metabolic syndrome seemed to develop after the age of 6 years. Janssen and LeBlanc reported four studies that used direct measures of cardiorespiratory fitness.2 42 44 50 51 All reported strong and significant relations with MetS. The summary OR for the least fit group relative to the most fit group in the four studies that measured fitness was 6.79 (95% CI: 5.11 to 9.03). Several studies show improvement in elements of the MetS in association with PA in obese and non-obese children,15 35 52,,55 but the amount of activity necessary to prevent or treat the MetS is not specified.

CVD risk factors and muscle fitness

Few studies have examined the association of the muscular endurance and strength with CVD risk factors among children and adolescents.56,,58 Two of these studies included measurement of maximal strength only, and the associations are only analysed with individual CVD risk factors and not clustered metabolic risk.56 58 García-Artero et al57 measured muscle endurance, explosive strength and maximal strength and included a lipid-metabolic risk score in their analyses. However, they did not directly measure cardiorespiratory fitness, and they could therefore not analyse whether there was an association independent of aerobic fitness. Only one study has evaluated the association of both single and combined muscle strength measurements with MetS risk independent of aerobic fitness.34 In this study, muscle fitness was negatively associated with clustered metabolic risk, independent of cardiorespiratory fitness, but the association was weaker than for cardiorespiratory fitness.

Other cardiovascular variables

Thomas and Williams and Thomas et al have conducted two comprehensive reviews of studies of the effect of exercise on C-reactive protein (CRP), interleukin 6 (IL-6) and fibrinogen.59 60 CRP is an acute-phase reactant that increases significantly in response to trauma and inflammation. CRP is a sensitive marker of inflammation, and there is evidence of its causal role in inflammation.61 Age, sex, body mass index, adiposity, physical inactivity, physical fitness and smoking are associated with CRP levels.62 There is evidence to suggest that regular PA protects against disease associated with chronic low-grade systemic inflammation and decrease the level of CRP.63 The mechanisms responsible for the association between reduced CRP and increased PA are unknown, partly because PA is related to several confounders that are independently associated with CRP concentration. Many researchers claim that PA mainly affects health through its relation to fatness. However, if PA causes changes in fat tissue, which produces cytokines affecting CRP, PA remains the primary cause of the change.

Studies reporting the association between PA and CRP concentration in young people are relatively scant, with much of the research concentrating on overweight and obese individuals.59

Observational studies have shown associations between PA and CRP.64 65 Other studies using self-reported PA have failed to find an association with CRP.66,,69 When physical fitness is analysed, studies find much stronger associations to CRP. Andersen et al found an association between cardiorespiratory fitness and CRP of –0.49 in 9-year-old children. They further found an 11.3 times higher risk for CVD risk factors to cluster in the upper quartile of CRP compared with the lower quartile. The reason why PA shows a weaker association with CRP than fitness could be that only high-intensity aerobic exercise may have an effect. Alternatively, it could be the fact that PA is assessed with higher levels of error.

Most intervention studies only include obese children, and many of them have small sample size. Interventions including PA or PA and diet lasting 4–6 months have found reduction in CRP in obese adolescents.55 70 71 Change in CRP was independent of weight. Conversely, Nassis et al72 found no reduction in CRP in a study of obese girls although cardiorespiratory fitness increased by 18.8%. Lack of effect on CRP was also found in two studies of obese children by Barbeau et al73 and Kelly et al,74 but the latter study only included nine subjects, and the small sample size may have caused a statistical type 2 error.

The role of exercise on other inflammatory markers is controversial. Hotamisligil et al75 observed that adipose tissue from obese mice was producing tumour necrosis factor α (TNFα), a proinflammatory marker, and this cytokine was responsible for insulin resistance. This observation changed our view on adipose tissue, and it has become apparent that obesity is linked to a state of chronic inflammation. Obesity results not only in the secretion of TNFα, but induces the release of many cytokines including resistin, IL-1 and IL-6. Given this proinflammatory response and the observation that systemic IL-6 concentrations are elevated in obesity, it is generally thought that elevations in IL-6 have a negative effect on metabolism.76 However, it is now proven that muscle cells also act as an endocrine organ, and among other substances produce IL-6 during contraction.77 Myokines, which are cytokines produced by the muscle cells, may be involved in mediating the health beneficial effects of exercise and play important roles in the protection against diseases associated with low-grade inflammation, insulin resistance, hyperlipidaemia such as CVDs, type 2 diabetes and cancer. IL-6 produced by the muscle enhances glucose uptake in the muscle cells, glucose production from the liver and lipolysis in the adipose tissue. Training studies have shown that adaptation to training reduces IL-6 response in plasma. However, even if plasma-IL-6 appears to be downregulated by training, the muscular expression of the IL-6 receptor appears to be upregulated.77 How these observations apply to children is not known. Furthermore, it is difficult to evaluate the role of IL-6 in observational studies, because the acute response to exercise disappears fast, and fasting blood samples may therefore not be suitable.78 Most intervention studies are conducted in obese children. A decrease in IL-6 has been found in obese children after an exercise intervention,70 but it has also been seen in normal weight children.71 Currently, there is no consensus of the effect of PA on inflammation in children.

Cardiovascular fitness (aerobic fitness)

Correlational studies indicate low-to-moderate positive relationships between PA and both maximal and submaximal indicators of aerobic fitness. The strength of the association depends on how accurately PA and to some degree aerobic fitness are assessed. Dencker and Andersen3 reviewed studies using objectively measured PA in children. The objective data from these studies strongly suggest that the amount of PA in childhood is weakly associated with aerobic fitness with r values of 0.25–0.40.

Experimental training studies with children aged 8 years and older indicate improvements in aerobic fitness.24 28 30 32 79,,81 Some intervention studies have used high-intensity aerobic sports, but it is interesting that differences in fitness were also found in relation to everyday activities such as cycling to school, and the differences seem not to be caused by selection.82,,85 Programs involving continuous vigorous activity for >30 min at least 3 days per week result in an approximately 10% increase in fitness (3–4 ml/kg/min). However, cycling to school, which is a PA repeated twice a day, results in 8–9% change in aerobic fitness.


PA recommendations in relation to metabolic health in children have changed since the first recommendation was formulated by the American College of Sports Medicine in 1988.86 PA was focused on aerobic exercise performed at an intensity sufficient to increase aerobic fitness. This has gradually changed, and emphasis is now placed on both aerobic and resistance exercise. The change in recommendations is based on the evidence of an independent association between muscle fitness and metabolic disorders. Another change was related to the quantity of recommended exercise. Since the recommendations published in the late 1990s by Biddle et al,87 most authorities have recommended 60 min of at least moderate-intensity PA accumulated most days as the main goal. WHO's latest Global Guidelines recommend that 60 min of at least moderate intensity should be accumulated on top of activities of daily living.88 Studies using accelerometry have shown that even the most sedentary children accumulate to around 30–40 min of MVPA per day.18 This change is therefore a major increase in recommended PA level in children. The change is logical, because earlier evidence was based primarily on intervention studies and self-reported PA in observational studies. Intervention studies describe the content of the intervention, but there is usually no knowledge of what PA participants participated in outside the intervention. Two factors have increased our understanding in this area. These are a better quantification of PA by accelerometry and a more accurate definition of metabolic outcomes. The introduction of a composite CVD risk factor score based on continuous z-scores has improved associations between exposure and outcome, which makes it possible to better quantify the amount and type of PA children need to improve health.

Development of methods to assess PA objectively can still be improved. There is still no consensus on how to assess PA in children and adolescents reliably regarding the appropriate time period representative for overall PA and the inflation factor that should be used to account for the behavioural variation of PA in youth. Other limitations of accelerometry are related to the lack of ability to quantify cycling and swimming. These limitations may soon be overcome, because much research is focused on combining different measures such as accelerometry, heart rate, cycle computers and global positioning system. Metabolic outcome measures can also be improved. Dichotomisation of metabolic variables has mainly been used to help physicians decide when drug treatment was justified, but treatment of children with metabolic disorders will probably mainly be based on lifestyle changes in the future except for children having specific genetic problems such as familiar hypercholesterolaemia. Therefore, it makes sense to use continuous scores to define metabolic health. Despite all these limitations, we believe that the current 2010 WHO recommendation is based on solid evidence.



  • Competing interests None.

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

Linked Articles

  • Miscellanea
    BMJ Publishing Group Ltd and British Association of Sport and Exercise Medicine