COMMENTARY Assessing secular trends in blood pressure in children and adolescents A Chiolero 1,2 , P Bovet 2 and G Paradis 1,3,4 1 Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, Canada; 2 Institute of Social and Preventive Medicine (IUMSP), University Hospital Center and University of Lausanne, Lausanne, Vaud, Switzerland; 3 McGill University Health Center Research Institute, Montreal, Canada and 4 Quebec Public Health Institute, Montreal, Canada Journal of Human Hypertension (2009) 23, 426–427; doi:10.1038/jhh.2008.165; published online 15 January 2009 Until recently, hypertension was rarely a matter of concern among children and adolescents. Per- ception of this condition has changed because of evidence showing that elevated blood pressure (BP) during childhood is not benign. Elevated BP in children is associated with left ventricular hyper- trophy and a thickening of carotid intima–media thickness. 1 Furthermore, BP tracks over time and children with elevated BP are at high risk of developing hypertension in adulthood. 2 These find- ings support the potential for primordial prevention, that is, the prevention of elevated BP during child- hood and adolescence. 3 The new interest shown to paediatric hyperten- sion is also related to the worldwide ‘epidemic’ of obesity. On the basis of the strong association between obesity and elevated BP, 4,5 it is often believed that mean population BP levels have increased in the past decades because of upward trends in the prevalence of obesity in children. 6,7 However, the few population-based studies of BP trends have not consistently shown that BP increased in parallel with the obesity trends. 8,9 Some studies show upward BP trends (for example, in the USA in the last decade), 7 whereas others show downward BP trends. 10 Furthermore, trends in systolic and diastolic BP diverged in some studies. 9 There are major challenges in assessing BP trends. Measurement methods (setting, number of BP read- ings, device used to measure BP, position of the subjects) and populations sampled (ethnicity com- position, age distribution, stage of puberty and study place) need to be identical between surveys. 8,9 For instance, if BP is based on two measurements in a survey and on three measurements in subsequent surveys, BP will be higher initially because BP decreases with repeated measurements. 11 As BP varies in children and adolescents according to age, sex and height, the prevalence of elevated BP should be examined across age, sex and height categories and large sample sizes are essential to have the power to identify trends with confidence and to obtain precise estimates of BP. 9 Secular trends in height also have to be considered because BP increases with height throughout childhood and adolescence. 11 In the current issue of the Journal of Human Hypertension, Kollias et al. compared the mean BP in 446 and 557 adolescents aged 12–17 years, surveyed in 2004 and 2007, respectively, in the Island of Samos, Greece. 12 A large difference in the mean age- and sex-adjusted BP was reported between the surveys: þ 4.3 mm Hg for systolic BP and þ 10.6 mm Hg for diastolic BP in 2007 com- pared with 2004. The prevalence of high BP (X95th percentile 11 ) was 16.1% and 22.9% in 2004 and 2007, respectively. The authors concluded that ‘rising trends seem to characterize the prevalence of high BP’. These findings are surprising. How could such a large change in BP take place within such a short period of time (3 years)? First, the usual suspect—obesity—cannot be blamed. The mean body mass index (BMI) was slightly higher in 2004 than in 2007, and prevalence of overweight and obesity was fairly identical in both surveys. With or without adjustment for BMI, the difference in the age- and sex-adjusted BP was similar. 12 This shows that the higher BP in 2007 compared with 2004 could not be accounted for by an increase in obesity. Factors related to changes in lifestyles (for example, physical activity) may be responsible but were not measured in both surveys. Second, the comparability of the two samples is disputable. Although the average BMI was similar between surveys, the differences in height (3 cm), weight (3.8 kg) and age (0.6 year) between partici- pants in 2004 and 2007 suggest that different types of adolescents were sampled. At that age, these Correspondence: Dr A Chiolero, Community Prevention Unit, Institute of Social and Preventive Medicine (IUMSP), University of Lausanne, 17 Rue du Bugnon, 1005 Lausanne, Vaud, Switzer- land. E-mail: arnaud.chiolero@chuv.ch Journal of Human Hypertension (2009) 23, 426–427 & 2009 Macmillan Publishers Limited All rights reserved 0950-9240/09 $32.00 www.nature.com/jhh