OUP user menu

Tracking of blood pressure from childhood to adolescence in a Greek cohort

Anastasios Kollias , Kyratsoula Pantsiotou , Nikolaos Karpettas , Leonidas Roussias , George S. Stergiou
DOI: http://dx.doi.org/10.1093/eurpub/ckr082 389-393 First published online: 25 June 2011


Background: Studies have reported tracking of blood pressure (BP) from childhood to adulthood but with inconsistent results mainly due to methodological and ethnic differences. We aimed to examine BP tracking during a 7-year period in a Greek cohort. Methods: This is a longitudinal school-based study conducted during 1990–96 in Athens, Greece. Children underwent BP and anthropometric measurements on two to three visits annually (averaged to annual values) for 7 years. Results: A total of 166 children with complete yearly follow-up data for the examined period were included (mean baseline age 9 ± 1.7 years, range: 5–12 years, 89 boys). At baseline, the prevalence of pre- and hypertension was 22.9 and 24.1% respectively and at the end of the follow-up 24.1% (P = NS vs. baseline) and 13.3% (P = 0.02 vs. baseline) respectively. Systolic/diastolic BP tracking correlation coefficients between 1990 and 1996 were 0.38 (P < 0.001)/0.20 (P = 0.06) for boys and 0.30 (P = 0.007)/0.22 (P = 0.06) for girls. Among children with baseline BP ≥90th centile (systolic and/or diastolic), 44% remained in the same BP range after 7 years. In stepwise multiple regression analysis, baseline systolic BP, male gender, baseline body mass index (BMI) and change in BMI from baseline to the end of the follow-up (ΔBMI) were significant predictors of systolic BP levels at the end of the follow-up. Baseline diastolic BP, baseline BMI and ΔBMI were significant predictors of diastolic BP at the end of the follow-up. Conclusions: These data suggest that the risk of developing high BP during adolescence can be predicted by BP and BMI at childhood.


In the last two decades there is increasing interest in the field of hypertension in children and adolescents.1,2 It is now recognized that hypertension in children is not as uncommon as previously believed and in most cases represents early onset of essential rather than secondary hypertension.1,2 School-based screening studies have reported a prevalence of high blood pressure (BP) in children and adolescents ranging from 2% to 24%.3–8 More importantly, recent data suggest increasing trends of average BP in children and adolescents which appear to be mainly attributed to the increasing prevalence of obesity.7,9

Although there are many cross-sectional studies on the relationship of BP with age and body mass index (BMI), only few have investigated these associations longitudinally during childhood and adolescence. Moreover, the results are rather inconsistent partly due to differences in protocols and methodology used, as well as differences in the ethnic background.10–13 Despite the inconsistency in these results, it seems that increased BP in childhood is predictive of high BP and hypertension in early adulthood.10–13 This fact may have important implications in terms of cardiovascular health since it allows the early identification of individuals at high-risk for developing sustained hypertension.

In Greece, a country with a predominant place regarding childhood obesity and hypertension, there are no longitudinal data on BP levels in children and adolescents. The aim of this study was to examine BP tracking during a 7-year period in a Greek cohort.


Subjects and measurements

This is a longitudinal study conducted from 1990 to 1996 in Athens, Greece. An open invitation was sent to six randomly selected public schools of Athens districts inhabited mostly by children of middle-to-upper class. Inclusion criteria were Greek nationality, age from 5 to 12 years and willingness to participate in the study. Students with persistently high levels of systolic/diastolic BP suspected for secondary hypertension were excluded from the study and referred for investigation. All the 1021 children that accepted to participate in the study were enrolled. The study protocol was approved by the Ministry of Education and written consent was obtained by the participants’ guardians.

Anthropometric and BP measurements were carried out every 4–6 months for 7 years. All measurements were performed in a quiet and tempered room at school by a single physician (K.P.). At each examination standing height was measured by portable Harpenden stadiometer to the nearest mm, and body weight with a Soehnle electronic weighing device to the nearest 100 g, with the child dressed only in light clothing and without shoes. BMI was calculated as the weight in kilograms divided by the square of height in meters. For the definitions of overweight and obesity the age- and sex-specific BMI criteria of the International Obesity Task Force were used in the analysis.14

Single BP measurements were taken at each visit after 5 min sitting rest using a standard mercury sphygmomanometer and appropriate cuff size according to the individual’s arm circumference. Systolic BP was determined as the first perception of sound (Korotkoff sound K1) and diastolic BP as the disappearance of sounds (Korotkoff sound K5). The normative tables of the US 4th Report on the Diagnosis, Evaluation and Treatment of High BP in Children and Adolescents according to age, gender and height were used for the evaluation of the BP in the analysis.1

Statistical analysis

Only children with at least one assessment per year for all seven consecutive years were included in the analysis. Average annual BP and anthropometric values per participant were used in the analysis. Results are expressed as mean values with SD. Student’s t-tests were used for the comparison of quantitative values (paired tests for comparison of baseline versus end of the follow-up BP and BMI measurements, and unpaired tests for gender comparisons). Chi-square test was performed for the comparison of percentages. Pearson correlations were performed to examine the tracking of BP and BMI between baseline and follow-up. A stepwise multiple regression analysis was performed to identify variables that predict BP levels at the end of the follow-up. Independent variables included baseline age, gender, baseline BP (systolic or diastolic), baseline BMI and change in BMI from baseline to the end of the follow-up (ΔBMI). A probability value P < 0.05 was considered statistically significant. Statistical analysis was performed using the Statistical Package for Social Sciences software (SPSS release 13.0; SPSS Inc., Chicago, Illinois, USA).


Participants’ characteristics

Two children with especially high levels of BP were excluded from the study. From the initially recruited 1021 children (mean age 7.8 ± 2.8 years, BMI 17.5 ± 2.6 kg m−2, systolic/diastolic BP 104.4 ± 13.1/68.2 ± 10.9 mmHg, 520 boys), 166 (16.3%) had complete yearly data throughout the 7-year follow-up and were included in the analysis (mean baseline age 9 ± 1.7 years, range 5–12 years, 89 boys). The main characteristics of the study participants at baseline and at the end of the follow-up are presented in table 1. Participants compared with non-participants had higher age (9 ± 1.7 vs. 7.6 ± 2.9 years, P < 0.001), higher systolic/diastolic BP (107.3 ± 10.8/71.1 ± 9.2 vs. 103.9 ± 13.4/67.7 ± 11.1 mmHg, P < 0.001), but similar BMI (17.8 ± 2.7 vs. 17.4 ± 2.5 kg m−2, P = 0.13) and gender distribution (males: 53.6 vs. 50.4%, P = 0.50).

View this table:
Table 1

Participants’ characteristics at baseline and at the end of the follow-up (mean ± SD)

CharacteristicsBaseline (1990)P-valueEnd of follow-up (1996)P-value
Boys (n = 89)Girls (n = 77)Boys (n = 89)Girls (n = 77)
Age: mean, range (years)8.9 ± 1.7, 5–129.2 ± 1.7, 5–120.3014.9 ± 1.7*, 11–1815.2 ± 1.7*, 11–180.26
Height (cm)136.2 ± 11.5137.1 ± 11.70.62170.6 ± 10.6*163.4 ± 6.7*<0.001
Weight (kg)34.5 ± 10.232.6 ± 7.50.1865.5 ± 14.7*56.7 ± 8.3*<0.001
BMI (kg m−2)18.2 ± 2.817.2 ± 20.00822.3 ± 3.5*21.3 ± 2.6*0.04
Systolic BP (mmHg)107.8 ± 11.8106.7 ± 90.50118.2 ± 13.2*110.4 ± 10.4*<0.001
Diastolic BP (mmHg)71.6 ± 9.870.5 ± 8.10.4467.7 ± 11.4*64.7 ± 9.5*0.07
  • *P < 0.05 vs. baseline values

Longitudinal changes in BP

The prevalence of BP between 90th and 95th centile (range of prehypertension1 or high-normal BP2) was similar at baseline and at the end of the follow-up (22.9 vs. 24.1% respectively, P = 0.89), while the prevalence of BP ≥ 95th centile (range of hypertension)1,2 was higher at baseline (24.1 vs. 13.3% respectively, P = 0.02). Annual average systolic and diastolic BP levels are shown in figure 1. Strong positive correlations were observed for systolic BP levels (average annual BP values per individual) with age for both boys and girls, although this was stronger for boys (figure 2). On the contrary, diastolic BP was not correlated with age in boys, while an inverse correlation with age was observed in girls (figure 2). The tracking coefficients for systolic and diastolic BP for the examined period are shown in table 2. In particular, coefficients were higher for systolic compared with diastolic BP and higher for boys compared with girls (table 2). From the children with baseline BP levels ≥90th centile, 43.6% retained these BP levels at the end of the follow-up. This percentage was higher in boys compared with girls (58.7 vs. 21.9%, P = 0.06). Boys exhibited a higher increase in systolic BP levels over time compared with girls (10.4 vs. 3.9%, P = 0.001).

Figure 1

Average systolic and diastolic BP per year in boys and girls (error bars represent standard deviation)

Figure 2

Association of systolic and diastolic BP levels with age in boys and girls. Filled triangle represent annual average values

View this table:
Table 2

Tracking coefficients for systolic/diastolic BP and body mass index

Examined periodSystolic BPDiastolic BPBMI
1990–920.50 **0.32**0.28**0.200.94**0.88**
1990–930.49 **0.35**0.35**0.220.91**0.88**
  • *P < 0.05; **P < 0.01

Longitudinal changes in BMI

The prevalence of overweight and obesity was 21.1 and 2.4% respectively at baseline and 24.1 and 2.4% at the end of the follow-up (P = NS vs. baseline for both comparisons). The tracking coefficients for BMI in both genders are shown in table 2. These were all significant and slightly higher for boys. Among children with baseline BMI corresponding to overweight or obese state, 66.7% retained these levels at the end of the follow-up. In particular, from the 35 overweight children at baseline, 19 retained this BMI status and three became obese at the end of the follow-up. On the other hand, all the four obese children at baseline were classified as overweight at the end of the study period.

Predictors of follow-up BP

In stepwise multiple regression analysis [independent variables including baseline age, gender (1: boys, 2: girls), baseline BP (systolic or diastolic), baseline BMI and ΔBMI], systolic BP at the end of the follow-up was mainly determined (R2 = 0.28) by baseline systolic BP (β ± SE 0.34 ± 0.08, P < 0.001, 95% CI 0.18, 0.50), male gender (–6.68 ± 1.72, P < 0.001, 95% CI –10.07,–3.29), baseline BMI (1.1 ± 0.37, P < 0.01, 95% CI 0.36, 1.83) and ΔBMI (0.24 ± 0.07, P = 0.001, 95% CI 0.10, 0.38). Likewise, baseline diastolic BP (0.21 ± 0.09, P < 0.05, 95% CI 0.04, 0.39), baseline BMI (0.90 ± 0.34, P < 0.01, 95% CI 0.23, 1.56) and ΔBMI (0.16 ± 0.07, P < 0.05, 95% CI 0.03, 0.29) were significant predictors of diastolic BP at the end of the follow-up (R2 = 0.10). When BMI z-scores were used instead of BMI values, systolic BP at the end of the follow-up was predicted (R2 = 0.21) by baseline systolic BP and male gender, whereas diastolic BP was predicted (R2 = 0.07) by baseline diastolic BP and baseline BMI z-score.


This study provides data on BP tracking in a Greek cohort of children. The main findings are: (i) the prevalence of high BP as well as that of overweight/obesity was relatively high, both at baseline and at the end of the follow-up period; (ii) BP tracked at moderate levels; (iii) tracking was higher for systolic compared with diastolic BP and for boys compared with girls; and (iv) BP and BMI in childhood, as well as changes in BMI during growth predict BP levels at adolescence.

More than one-third of the studied population had BP above normal, both at baseline and at the end of the follow-up. Furthermore, almost one in four children had increased BMI. Other cross-sectional studies in children and adolescents in Greece have suggested an especially high prevalence of elevated BP—with rates often exceeding 30%—as well as of overweight/obesity in the studied populations.7,15–18 These results should be interpreted with caution, since the diagnosis of hypertension was based on BP measurements obtained on a single occasion, and therefore the prevalence of hypertension was probably overestimated in these studies.1,19 Moreover, whether the American BP percentiles can be adopted in European populations is a matter of concern, since data from European studies suggest higher levels for BP percentiles compared with the American data.20,21

This study examined BP tracking by estimating the correlation coefficients between BP levels at different time points and by quantifying the percentage of children with elevated BP who retained these levels at the end of the follow-up period. More specifically, tracking coefficients for systolic BP ranged from 0.28 to 0.52 and for diastolic from 0.09 to 0.37 (table 2). It should be noted that the coefficients were higher for shorter time periods, which is in line with those of other studies.11,12 The tracking coefficients reported from previous studies vary considerably from –0.17 to 0.81 for systolic and from –0.22 to 0.80 for diastolic BP.12,22–24 The results on BP tracking seem to be inconsistent in different ethnic groups,10 however in recent meta-analyses only the follow-up time and mean baseline age were significantly associated with the tracking coefficients.11,12 In addition, measurements protocols (oscillometry vs. mercury, Korotkoff K4 or K5, multiple vs. single BP measurements) have been found to have an impact on BP tracking.13 Another finding of this study was that 43.6% of the children with baseline BP levels ≥90th centile retained these BP levels after 7 years. This finding is in accord with the Bogalusa Heart Study, where 40% of individuals with systolic and 37% with diastolic BP >80th percentile, retained these levels 15 years later.25 It should be noted that a decrease in the prevalence of BP ≥95th centile was observed during the study period. This is in line with the findings of other studies which have shown that repeated measurements in multiple visits lead to a decrease in office BP levels and thereby decrease in the estimated prevalence of childhood hypertension.3 Possible reasons include familiarization with the presence of the observer and the BP measurement procedure as well as the regression to the mean phenomenon.

Another interesting finding was that systolic BP tracking was higher than that of diastolic BP. Previous studies and meta-analyses also showed the average systolic BP tracking coefficient to be higher than of diastolic.11,12 It should be noted that the measurement of diastolic BP in children can be difficult in terms of discrimination and use of K4 and K5 sounds. The official recommendation by the Fourth Report of the US National High BP Education Program is the use of K5, and only if the very low K5 persists, K4 should be used.1 Moreover, a recent metaregression analysis revealed that the choice of different BP measurements protocols in children might affect diastolic BP tracking more than that of systolic.13 More specifically, it was shown that the use of K4 instead of K5 predicts higher diastolic BP tracking.13 In addition, the average of multiple BP measurements is related to higher diastolic BP tracking (first or single readings are higher than the average of multiple readings).13 These findings could explain, at least in part, the lower diastolic compared with systolic BP tracking coefficients, as well as the lack of (in boys) or even the inverse (in girls) association of diastolic BP with age. Other factors, such as the potential effect of hormonal factors (menarche), limitations of the protocol applied (single measurements, observer bias), etc, might also have contributed to the observed unexpected inverse association of diastolic BP with age in girls.

In the present study, boys had higher BP tracking coefficients and exhibited a higher increase in systolic BP levels over time compared with girls. Previous studies have reported conflicting findings regarding this issue. In the Dormont High School Follow-Up Study26 systolic BP tracking was higher for males compared with females (0.27 vs. 0.24 respectively), whereas in the Muscatine Study27 females had higher diastolic BP tracking than males in some age groups. A recent metaregression analysis showed marginally higher systolic BP tracking in males compared with females (0.39 vs. 0.38 respectively) and higher diastolic BP tracking (0.29 vs. 0.26).11 Dasgupta et al showed that boys are more likely than girls to develop high systolic BP as they approach adulthood.28 These findings attest to the conclusion that boys with high systolic BP during childhood/adolescence are at increased risk of developing hypertension in the adult life.

Apart from BP tracking this study also assessed BMI tracking coefficients. Measurements of BMI in childhood and adulthood have been reported to be correlated with each other, at the same level or closer than other overweight measures.29 Longitudinal studies have shown low/moderate (r = 0.2–0.4) or moderate/high (0.5–0.8) correlations between childhood and adult BMI measures.29 Increased tracking has been reported in older children (>8 years), particularly with sexual maturity, in younger children (aged 6–12 years) who are more overweight, and in children with an obese parent.29 In this study, BMI tracking coefficients were found to be exceptionally high and stronger than those of BP. This finding suggests that overweight/obese children are at increased risk of retaining BMI above normal during adolescence, a finding with significant implications since obesity in children and adolescents is known to be associated with a clustering of cardiovascular risk factors.30 It should be noted that when BMI z-scores were used in the multivariate analyses, the R2-values of the models were lower compared to those including BMI values. Recently, it was shown that while BMI z-score is optimal for the cross-sectional assessment of adiposity, BMI values seem to be superior for the assessment of adiposity changes in growing children.31

The tracking of BP over time is subjected to genetic-environmental interactions and influenced by several factors. In multivariate analyses, the present study showed that baseline BP levels, male gender, baseline BMI and changes in BMI during the follow-up period were significant independent predictors of the BP levels at the end of the follow-up. Although the relationship between BMI and BP is well established,8 it should be mentioned that this issue has not been extensively examined in a longitudinal concept. An important observation is that further to an increased baseline BMI, weight gain during childhood–adolescence is also independently associated with high BP levels. However, the studied factors explained only a part of the variance of the dependent variable (BP levels), suggesting that other factors (such as hormonal factors, lifestyle habits) contribute to the observed BP changes.

The present study should be viewed in light of some limitations. First, only 16% of the initial sample had complete yearly data for the studied period and was finally analysed. This could have introduced a selection bias, since the children included in the analysis had slightly higher baseline age and consequently BP levels compared with non-participants. However, the participants had similar gender distribution, similar BMI and were all of Greek origin and of the same socioeconomic class compared with non-participants. The low percentage of retention could be attributed, at least in part, to several factors. First, children and mainly adolescents usually are less compliant with participation in studies compared with adults. Many children might have felt uncomfortable being subjected to repeated anthropometric measurements and some might have changed school during the follow-up. Guidelines on the diagnosis and clinical significance of childhood obesity and/or hypertension did not exist during the study period and a ‘multilayered’ approach for recruitment and retention including frequent contact and cooperation with school and family infrastructures, emphasis on participant convenience, incentives, etc, was difficult to achieve. Moreover, it should be noted that the lack of pediatric hypertension guidelines at the time of the study period did not allow the implementation of an intervention program in children with elevated BP, apart from those with suspected secondary hypertension who have been excluded. The 2004 normative BP tables, which were published after the study completion, were implemented for a better description of the study data and the BP tracking. Another limitation is that BP values were derived from single BP measurements instead of the average of two or more measurements or visits. Finally, apart from BMI, other factors which may have an effect on BP levels such as dietary and lifestyle habits have not been investigated.

In conclusion, this study demonstrated that BP and BMI track from early childhood to adolescence and that children with high BP and/or BMI levels at early childhood—especially boys—and children who gain weight during this age range are at increased risk of developing high BP during adolescence. These findings reinforce the concept that the early identification of subjects at increased risk will allow the implementation of intervention strategies in order to curtail the obesity/hypertension epidemic.

Conflicts of interest: None declared.

Key points

  • There is growing evidence that high BP is established early in life. Although there are many cross-sectional studies on the relationship of BP with age and body mass index, only few have investigated these associations longitudinally during childhood and adolescence.

  • In this study, a high prevalence of high BP as well as that of overweight/obesity was recorded in a sample of Greek children.

  • During a 7-year period, BP tracked at moderate levels and tracking was higher for systolic compared with diastolic BP and for boys compared with girls.

  • BP and BMI in childhood, as well as changes in body mass index during growth predicted BP levels during adolescence.

  • The early identification of subjects at increased risk will allow the implementation of intervention strategies in order to curtail the obesity/hypertension epidemic.


View Abstract