Many studies have shown a positive association between obesity and hypertension in children of various populations [5,6]. The definition of hypertension in children is based on percentile cutoffs from normative distribution of BP in children presumed healthy [7]. There have been concerns that these statistically derived thresholds may underestimate children at risk [8]. BP is a continuous variable that is positively correlated with cardiovascular risk across the entire BP range [9,10]. Race/ethnicity is another factor associated with differences in the prevalence of hypertension and the degree of arterial stiffness associated with high BP in children [11,12]. Thus, it is prudent to have population-specific evaluation of childhood BP and its associated factors for the full range of BP levels.

Health behaviours established in childhood tend to continue into adulthood. Most studies support the beneficial effect of physical activity (PA) in lowering BP in children [13,14], although unfavourable effects of moderate to vigorous intensity PA on BP have also been reported [15,16]. Family history of hypertension (FHH) is a particularly important factor associated with high BP risk and its influence on BP occurs well before adolescence [17]. Although FHH cannot be modified, we do not yet know if children genetically predisposed to hypertension respond differently to excess body weight or other factors with regard to their BP. Furthermore, while socioeconomic inequalities are linked to childhood obesity in western developed countries [18], there are no reports on their impact on BP among children.

The CHMS cycle 1 included data for 1878 participants aged 6 to 17years. For our purposes, we excluded 7 respondents missing BP measures, 2 missing BMI measures, and one being treated with hypertension medication. All study respondents were cross-tabulated for conditions that are the primary causes of BP elevation; one reported having diabetes mellitus type 1 and one other diabetes mellitus type 2; 11 reported congenital heart disease; and 4 reported kidney dysfunction or disease. We also excluded them from our analyses. Two participants who reported having hypertension but were not being treated with medications were retained. After an initial exploration of the data, we also excluded one 9-year-old boy whose BP was 173/127mmHg, for which an underlying condition was suspected. Thus, the final analyses included data for 1850 participants.

The survey consisted of a personal household interview followed by a physical examination at a mobile examination center within 2days to 6weeks of the interview. The household interview included a questionnaire about general demographic information and an in-depth health questionnaire. Parents/guardians answered all the applicable household questions for participants aged 6 to 13years and, with assistance from children responded to questions about, for example, participation in physical activity during schools hours. Participants aged between 14 and 17years answered all questions on their own.

Standing height and weight were measured and used to calculate BMI (weight [kilograms] divided by height [meters] squared). Age- and sex-adjusted BMI z-scores (BMIZ) were calculated based on World Health Organization (WHO) child growth charts [21]. According to WHO, obesity is defined as a BMIZ that is more than two standard deviations (SDs) above the mean (≈ 97.7^th percentile) and overweight as a BMIZ between one and two SDs above the mean (≈ 84^th to 97.7^th percentile).

Participants’ resting BP was measured according to the new protocol for standard BP measurements in surveys [19]. Participants were invited to wait in a quiet and comfortable environment for at least 5 minutes before the measurement. BP was measured using an automated BP cuff (BpTRU) on the right arm with a cuff size suitable for the arm circumference. Six BP readings were taken with a clinically validated oscillometric device (BpTRU^TM BP-300, BpTRU^TM Medical Devices Ltd., Coquitlam, British Columbia) at 1-minute intervals, with BP determined as the average of the last 5 measurements. Heart rate was measured in beats per minute (bpm) with each BP measurement and averaged using the last 5 measurements. BP was measured manually only when inconsistency was related to high variability or too many errors, which occurred in fewer than 1% of cases. More detailed information on BP measurement in this survey can be found elsewhere [22].

We calculated age-, sex- and height-adjusted SBP z-scores (SBPZ) and DBP z-scores (DBPZ) using the US reference data [7]. We defined systolic or diastolic hypertension as SBPZ or DBPZ at or above the 95^th percentile. Pre-hypertension was defined as SBPZ or DBPZ between the 90^th and 95^th percentile or as for adults (≥120/80mmHg).

The PA model for adults was adopted for adolescents aged 12 to 17years [23]. Survey participants or their parents were asked about the type, frequency and duration of PA in the past 3months during leisure time and at school. Energy expenditure (EE) was calculated using the frequency and duration per session of the PA as well as metabolic equivalents (MET) value of the activity. Calculated total daily EE values were used to develop a PA index: active, moderately active, and inactive. The PA of children aged 6 to 11years was determined from the answers given to the question “Over a typical or usual week, on how many days was he/she physically active for a total of at least 60 minutes per day?” If the answer was “4days or more”, the child was regarded as physically active; otherwise, the child was considered inactive.

We determined FHH from answers to the questions “Has anyone in your immediate family ever had high BP, excluding during pregnancy?” and “What is the youngest age at which a member of your immediate family was first diagnosed with high BP?” Participants who answered “Yes” to the former and reported that age at diagnosis was less than 60years were considered to have an FHH.

We determined the highest education achieved by adult members of the household and dichotomized this based on whether at least one adult had completed post-secondary education. Household income was categorized into two groups (low, middle or high) based on the total family income from all sources and the number of people in the household; a third category was created for those with data missing on this variable. We defined ethnicity as whites and non-whites including Aboriginals living off reserves. Children aged 12years and older were asked about cigarette smoking.

Because BP may increase substantially with the onset of puberty and between sexes [24], we stratified analyses for children (6 to 11years) and adolescents (12 to 17years), and for boys and girls. Associations of SBPZ or DBPZ with overweight and obesity, PA, FHH, education and income were estimated simultaneously using multivariate linear regression models, controlling for covariates - age and heart rate. Variables of ethnicity and cigarette smoking (adolescents only) were initially included in the models, but dropped later because they led no changes to the results. Another consideration in model construction was the 11 degrees of freedom requirement in statistical analysis, which restricts the number of parameters that can be entered into a model to 10. To evaluate the possible interactions between BMIZ and FHH on BPZ, we added an interaction term for BMIZ and FHH to the full regression model. To evaluate the potentially varying effects of BMIZ on BPZ with respect to normal weight, overweight and obesity, we stratified analyses according to BMI category and performed multivariate linear regressions within each stratum. We were not able to carry out subgroup analysis for each stratum because of limited sample size.

Based on the first national physical measures survey, we reported that the prevalence of childhood hypertension and pre-hypertension was at 2-3%, which is significantly lower than other Canadian studies, in which prevalence of hypertension was reported at 7.4% and 14% [2,3]. However, it is difficult to compare across studies because of differences in age, study populations (i.e. rural resident) and variability in BP measurement methods. Our results do seem to be in line with those studies with repeated BP measurements [25,26], suggesting the importance of standardizing BP measures. Our observations that BMI or obesity is consistently associated with increasing SBP are consistent with a series of studies performed on various populations, though some studies also observed the effect of BMI on DBP [2,6]. The mechanism underlying the association between obesity and hypertension is poorly understood. Sorof et al. proposed that obesity-induced hypertension may be mediated in part by sympathetic nervous system hyperactivity, which is partly manifested by increased heart rate and BP variability [6]. This possibility is supported by our data that heart rate is positively associated with BP, especially among children, and that a significant difference in heart rate was observed in the obese compared with nonobese groups (data not shown). Most research has focused on SBP because it is a strong predictor of arterial stiffness and many other cardiovascular risks growing into adulthood [11]. Our results support the close association between SBP and obesity, reinforcing the importance of maintaining a healthy weight for optimal SBP during childhood. In addition to substantial increases in SBP, we found markedly higher DBP in obese adolescent boys. This might be explained by evidence of early atherosclerosis at autopsy and ultrasound findings of the carotid artery, in which severely obese children, especial boys, display arterial stiffness and increased diastolic wall stress, indicating adverse changes in vascular health that could be the culprit for the increases in both SBP and DBP [27].

Furthermore, our results show that BMI has the strongest effect on BP among obese children and adolescents. Rosner et al. proposed that there may be a different effect of BMI on BP depending on the differing levels of fatness among various ethnic groups [12]. Our result are consistent with a recent study by Tu et al. who found that there is a marked intensification of the influence of adiposity on BP when children reach the categories of overweight and obese [28]. Moreover, the differing effect of BMI on both SBP and DBP may well be explained by some intervention studies. One such study by Rocchini et al. demonstrated that weight loss among obese adolescents changed the BP distribution from right skewed to no difference from the general population; however, they did not examine the effect on BP of weight loss among normal weight adolescents [29]. Further intervention research involving normal weight children would provide more insights into this discussion. While mechanisms underlying these differences are still being researched, recent evidence also indicates that the adipose tissue-derived hormone, leptin, may be a potentially important mediator in linking the effect of adiposity on BP [28]. Our results do suggest that small reductions in weight among obese children may lead to reductions in their BP.

The present study observed an inverse association between physical activity and SBP and DBP in adolescent boys. A number of observational and intervention studies have investigated associations between PA and BP in children and adolescents. Kelley et al. combined the results of 12 randomized trials and concluded that PA leads to a small but statistically significant reduction in BP [30]. Using accelerometer-measured PA data from the NHANES, Mark et al. described a modest dose–response relation of total PA and moderate-to-vigorous intensity PA with hypertension [31]. Leary et al. found a similar inverse association between total PA and SBP; however, when they compared the volume (duration and frequency) versus the intensity (vigorousness) of PA, they found that moderate to vigorous PA did not further reduce BP beyond the reduction achieved by a higher volume of PA, suggesting the volume of PA may be more important than the intensity of PA in reducing BP [15,16]. Since PA levels defined in the CHMS did not distinguish between the volume and intensity of PA, we were unable to assess their individual effects on BP. Nevertheless, our study identified inactive or moderately active adolescent boys as at risk of increased SBP and DBP, suggesting that PA should be intensified to mediate the negative effect on BP of other risk factors, and be promoted particularly for girls as evidence suggests that girls are less likely to participate in moderate to vigorous PA than boys [32]. Current clinical guidelines recommend that vigorous PA such as participation in competitive sports should only be limited in the presence of uncontrolled stage 2 hypertension, which is above 99^th percentile for gender, age and height [7].

Research shows that both genetic and environmental factors significantly influence BP and the development of hypertension during childhood [17]. Our results show that the strength of associations between BMI and BP was not modified by the presence of FHH. What our cross-sectional study was unable to address, however, is whether children with FHH are more susceptible to elevated BP due to increased BMI or decreased PA; more research is needed to explore effects of other risk factors in the presence of FHH so that targeted intervention can be designed for this genetically predisposed population. Interestingly, in future studies, increases in environmentally induced high blood pressure may serve to mask the association of blood pressure risk and genetic factors, for which FHH is a proxy.

Finally, the inverse relationship of post-secondary parental education with SBP in children, especially girls, is similar to previous findings in which BP in girls was related to socioeconomic variables in addition to anthropometric factors [33]. However, living in a low income family situation did not appear to increase risk of high BP, and was even associated with lower BP in children in this study. While some recent epidemiological studies from western developed countries have pointed that SES is inversely associated with childhood obesity, evidence on the relationship of SES with BP in childhood is limited [18]. Howe et al. in a recent study indicated important socioeconomic inequalities in obesity and cardiovascular risk factors including childhood BP in children; however, the SES indicator used in their study was maternal education [34]. In Canada, findings on SES and its relation with childhood obesity are conflicting [35,36]. In fact, we have noticed in our data that prevalence of obesity is lower in the lowest income (9.6%) than middle to high income groups (13.9%). With ongoing data collection for the next cycle of the CHMS, more studies with larger survey samples, detailed information on SES and inclusion of more variables (i.e. dietary, psychological factors) are expected to verify our findings.