Some key findings: ‘cut-offs for metabolic health and other advances’

A main contribution from EYHS has been the improvement of measures of both exposures and outcomes, which have strengthened associations between PA and health. The objective assessment of PA has opened up new possibilities such as quantifying total PA beneficial to health.2 This could not be performed with self-reported measures. PA recommendation is mainly based on intervention studies, which have only quantified the activity included in the intervention and not the total moderate-to-vigorous PA.3 ,4 The EYHS was the first study to assess activity by accelerometry on population level. A software was developed to handle the increasing amount of data and it has continuously been improved to process data. The field is still developing and there is some way to go before improved algorithms are capable of describing the amount, intensity and possibly the type of PA.

Another progress was the construction of a continuous composite CVD risk factor score, which describes metabolic health status better than single risk factors or composite scores from dichotomised variables as used in metabolic syndrome definitions. A composite score had already been used by Andersen and Haraldsdóttir,5 but Brage6 refined the methods using data from EYHS. To describe the importance of this development, we analysed the association between fitness and health in two ways. First, we used the parameters included in the metabolic syndrome and constructed a composite outcome variable based on dichotomised variables. Second, we constructed the same outcome based on continuous standardised scores of the risk factors and chose a cut-off point selecting the same number of children. In the first analysis, we found a six times increased risk in the least fit participants compared with the most fit. In the second analysis, the risk has increased 15 times in the same low-fit children. This improved association is generalisable and is caused by a variable, which better describes metabolic health. It has been argued that a continuous score makes it impossible to define which children have elevated risk, but this is not true. The CVD risk factors are independently distributed in healthy children, but not in unhealthy children. We calculated that, in about 15% of Danish children, CVD risk factors clustered and in Estonian children this condition was apparent in less than 10%, but clustering occurred at almost the same level of the continuous score in all countries.2 When we plotted children with increasing composite score, the score rose exponentially above this level.

Ried-Larsen et al7 have used 12-year follow-up data from the Danish arm of EYHS to analyse PA in youths as a predictor of clustered CVD risk, arteria carotis intima-media thickness and arterial stiffness. We observed that the association between arterial stiffness and PA was as strong as for the association between clustered risk and PA. In the study of Grontved et al,8 the strength in youths was inversely associated with single risk factors and a composite CVD risk factor score in young adulthood in multivariable-adjusted analyses including cardiorespiratory fitness. A joint association showed that adolescents being in the highest tertile, both muscle strength and cardiorespiratory fitness had the lowest CVD risk score in young adulthood (−1.42 SD), which is a surprisingly strong association. Longitudinal studies, following populations from childhood until hard endpoints occur, are wanted. Older studies suffer from a lack of statistical power, because dropout over decades decreases study sample. Future research in this field may further develop methods and take advantage of pooling data from other ongoing longitudinal studies such as EYHS. Until then, a take home message for BJSM's clinical readers is that PA is strongly related to metabolic health in children, the more PA the better, and both strength training and aerobic training works. Also, it is possible to define an unhealthy metabolic condition and select the children at risk for treatment if blood samples are available. We are currently working on non-invasive composite scores to be able to screen unhealthy children in a simple way with reasonable specificity and sensitivity.