OBJECTIVES: To present oscillometric blood pressure (BP) references from German nonoverweight children and compare them with US references.
METHODS: From children and adolescents, aged 3 to 17 years, from the German Health Interview and Examination Survey for Children and Adolescents (KiGGS 2003–2006), we obtained standardized BP measurements by using an oscillometric device validated in children. Gender-specific systolic (SBP) and diastolic (DBP) BP percentiles, which simultaneously accounted for age and height by use of advanced statistical methods, were derived from nonoverweight children to avoid overweight prevalence in the reference population influencing BP references.
RESULTS: The age- and gender-specific 95th percentiles from nonoverweight children (n = 12 199) were lower by up to 3 mm Hg for SBP and up to 2 mm Hg for DBP compared with the total sample (N = 14 349). KiGGS percentiles from nonoverweight children accounting simultaneously for age and height were mostly lower than in the US reference sample but higher for SBP in boys aged 14 years or older. At median height, the age-specific differences in 95th percentiles of SBP ranged from −4 to 4 mm Hg in boys and −2 to 1 mm Hg in girls and, for DBP, from −6 to 2 mm Hg in boys and −5 to 2 mm Hg in girls.
CONCLUSIONS: Compared with current US references, the proposed German BP reference values are not influenced by the prevalence of overweight children in the reference population, they are based on a validated oscillometric device, and they take advantage of improved statistical methods.
WHAT'S KNOWN ON THIS SUBJECT:
Current pediatric US blood pressure references are widely used internationally because of scarce international data and rare percentile derivation by age and height simultaneously. However, the US references may not fit other populations, and improved statistical methods have become available.
WHAT THIS STUDY ADDS:
The German blood pressure references by age and height from nonoverweight children and adolescents aged 3 to 17 years use a national sample, oscillometric measurements validated in children, and improved statistical methods. These references were not influenced by the increasing prevalence of overweight children in the sample.
Elevated blood pressure (BP) in childhood and adolescence is increasingly gaining attention because of its association with subclinical organ damage1,–,5 and to its tracking into adulthood hypertension.6 By widespread convention, the 95th BP percentile of a reference population defines hypertension and the 90th BP percentile defines prehypertension in children.
Older European reference values7 are based on heterogeneous pooled data from 6 regional European studies and were dismissed in the 2009 Guidelines for Management of High Blood Pressure in Children and Adolescents: Recommendations of the European Society of Hypertension (2009 European guidelines)8 for not accounting for age and height simultaneously. None of the few more recent European reference values9,–,11 fulfill the desirable criteria, foremost standardized BP measurement either using the auscultatory method or an oscillometric method with good validation results (preferably in children), population sampling, a sufficiently large age range, and reference value presentation by age and height percentile simultaneously. Therefore, the 2009 European guidelines recommend use of the US reference values presented in the “Fourth Report on the Diagnosis, Evaluation, and Treatment of High Blood Pressure in Children and Adolescents” (Fourth Report).12
The aim of this study was to present BP reference values by age and height from children and adolescents aged 3 to 17 years on the basis of a nationally representative sample of 12 199 nonoverweight children and adolescents who participated in the German Health Interview and Examination Survey on Children and Adolescents (KiGGS) 2003–2006 and had standardized BP measurements with an oscillometric device (Datascope Accutorr Plus [Datascope Corporation, Mahwah, NJ]) validated in children. We compared these reference values with the Fourth-Report references. In addition, our goal was to illustrate how excluding overweight children affects BP percentiles.
The KiGGS 2003–2006 study is a population-based cross-sectional study used to collect representative information on the health of children and adolescents aged 0 to 17 years living in Germany. The 2-stage sampling procedure involved the selection of 167 study locations from strata formed according to federal state, community type, and population size.13 In a second step, an equal number of children per birth year from each location were identified through local population registries and invited to participate in the study. The response rate was 66.6%, and 17 641 children and adolescents aged 0 to 17 years participated in the study (8656 girls and 8985 boys). The study was approved by the ethical committee of Charité–University Medicine, Berlin, and by the Federal Commissioner for Data Protection and Freedom of Information. Informed written consent and assent were obtained from all parents and from adolescents aged 14 years or older. The KiGGS includes 17% of children with a 2-sided migration background; migrants from Turkey and the former Soviet Union were the 2 largest groups.13 A computer-assisted personal interview by a study physician covered current and past medical conditions and medication within the 7 days preceding the interview. The girls' median age at menarche was 12.8 years, and the boys' voices started breaking/reached full adult pitch at a median of 13.5/15.1 years.14
In children aged 3 to 17 years, 2 readings of systolic blood pressure (SBP), diastolic blood pressure (DBP), and mean arterial BP and heart rate were obtained by using an automated oscillometric device (Datascope Accutorr Plus) at 2-minute intervals after a nonstrenuous part of the examination and an additional 5-minute rest.15 The measurements were taken using the right arm, in the sitting position with the elbow at the level of the right atrium, using 1 of 4 cuff sizes (6 × 12, 9 × 18, 12 × 23, or 17 × 38.6 cm), which had to cover at least two-thirds of the upper arm length (from the axilla to the antecubital fossa). The mean of the 2 measurements was used for analysis.
Body height was measured by trained staff according to a standardized protocol to the nearest 0.1 cm by using portable devices (Harpenden Stadiometer; Holtain Ltd, Crymych, United Kingdom). Body weight was measured with the child wearing only underwear to the nearest 0.1 kg with a calibrated scale (Seca, Birmingham, United Kingdom).13 BMI was calculated as the ratio of weight (in kg) by height2 (in m2) and rounded to 3 digits. A BMI at >90th percentile for gender and age was categorized as overweight according to the current German reference system by Kromeyer-Hauschild et al.16 KiGGS BMI percentiles were published recently, but for individual diagnosis of overweight, the references of Kromeyer-Hauschild et al17 are still recommended.
Inclusion and Exclusion Criteria for the Sample on Which the Percentiles Are Based
Of 14 836 KiGGS participants aged 3 to 17 years, we excluded 149 children with missing or invalid BP or height data, participants with chronic conditions possibly influencing growth (n = 302)18 or BP (n = 30 [ie, chronic renal diseases, aortic coarctation, hyperthyroidism, congenital adrenal hyperplasia, porphyria]) and children taking antihypertensive agents (n = 25 [Anatomical Therapeutic Chemical Classification System codes C01-03 and C07-09 or any medication given with the indication hypertension]). Seventeen children had >1 exclusion criterion, leaving 7038 girls and 7311 boys for analysis. For the BP percentiles by age and height from nonoverweight children, an additional 2150 overweight children were excluded (Table 1).
BP percentiles as a function of either age or height were modeled using the LMS method19 with the program LMSChartMaker Pro 2.2 (Medical Research Council, London, United Kingdom). The reference curves from nonoverweight children by age and height simultaneously were fitted by using an extension of the LMS method for 2 covariates, namely the generalized additive models for location scale and shape (GAMLLS) with the Box-Cox-Cole-Green distribution family,20,–,22 fitted with gamlss 1.9-4 in the free statistical software R 2.8.0 (www.cran.r-project.org). Both models require neither the assumption of a normal distribution nor of a constant variance of BP values with age and/or height. The skewness parameter L, the median M, and the coefficient of variation S have been modeled as a function of age and/or height either as polynomials or nonparametrically by cubic splines. Height was entered in centimeters or as z scores (based on the KiGGS population18), and the version with the better fit was used. Goodness of fit was examined via the generalized Akaike information criterion with k = 8,20 Q tests,23,24 and wurmplots25 and by examining the percentage of data outside the smoothed percentiles.
Models were fit for boys and girls separately. DBP was normally distributed but with a coefficient of variation S that varied with age. Median DBP was given by a function quadratic in age and linear in height z score. SBP was log-normally distributed in girls; in boys, the distribution was even more skewed. The model for the median included a linear and quadratic term in age, a linear term for height and the interaction terms height × age and height × age.2 In boys, S varied with age, whereas it was constant in girls. Only selected percentiles were tabulated in this study, but any (100α) percentile Pα can be calculated as or with zα the α quantile of a standard normal distribution. For example, the 10th SBP percentile for a boy at 10 years and 140 cm is given by and for a girl at this age and height by both of which equal 94 mm Hg. Age in years was used as a continuous variable and calculated to 3 decimal places. Fourth-Report values were calculated from the published formula12 by using age as 3.5, 4.5, 5.5 and so in years. Calculations apart from the fitting of the percentile curves were conducted using SAS 9.2 (SAS Institute Inc, Cary, NC).
Sampling weights were used to account for unequal sampling probabilities and to reflect the distribution of the population in Germany.13
The reference population of nonoverweight children and adolescents aged 3 to 17 years consisted of 6210 boys and 5989 girls. Baseline characteristics of the nonoverweight reference population are shown in Table 1. Smoothed BP percentiles from nonoverweight children by age and height are shown in Tables 2 and 3. These BP references do not require consultation of additional height reference tables because the height percentiles are given in cm.
BP increased in children and adolescents aged 3 to 17 years by both age and height percentile. SBP and DBP were very similar in boys and girls until age 13 years. The pubertal rise was more pronounced in boys than in girls, resulting in BP differences between boys and girls up to 17 mm Hg for SBP 95th percentile and 2 mm Hg for DBP 95th percentile (at age 17.99 years, both for the 90th height percentile). At a given age and gender, BP percentiles vary by height. Figure 1 illustrates how the 95th BP percentile, which is commonly taken for the definition of hypertension in children and adolescents, differs between the 5th and the 95th height percentile: SBP in boys by 2 to 8 mm Hg and in girls by 2 to 6 mm Hg and DBP by 2 mm Hg in both boys and girls.
To illustrate the impact of excluding overweight children from the reference population, we compared BP percentiles from nonoverweight children with those calculated without excluding overweight children (all other exclusion criteria being the same). The exclusion of overweight children resulted in slightly lower SBP and DBP for all percentiles both by age (Fig 2) and by height (data not shown but similar). This difference was more pronounced with increasing age and height and was generally more pronounced for the higher percentiles and in boys. For example, the difference between age-specific 95th percentiles for SBP from the samples with and without overweight children in boys aged 14 to 17 years was 2.6 to 2.7 mm Hg and in girls it was 1.3 to 1.6 mm Hg. For DBP, the difference between age-specific 95th percentiles from the samples with and without overweight children aged 14 to 17 years was 1.1 to 1.7 mm Hg in boys and 0.1 to 0.6 mm Hg in girls.
We compared KiGGS and Fourth-Report BP percentiles, both modeled by age and height simultaneously. Of note, KiGGS references are based on a nonoverweight reference population whereas overweight children were not excluded from the Fourth-Report data set. Fig 3 shows KiGGS and Fourth-Report BP percentiles by age, both for median height according to Centers for Disease Control and Prevention (CDC) growth charts (www.cdc.gov/growthcharts). Fourth-Report percentiles, which are based on a model that differs statistically from the KiGGS model, show a higher spread between the median and the outer percentiles. The 90th and 95th percentiles were mostly lower in the KiGGS, but they were higher for SBP in boys aged >14 years. At median CDC height, the difference for the 95th percentile (KiGGS minus Fourth-Report values) ranged by age for SBP from −4 to 4 mm Hg in boys and −2 to 1 mm Hg in girls; and for DBP from −6 to 2 mm Hg in boys and −5 to 2 mm Hg in girls.
This study presents BP references by age and height simultaneously for children and adolescents aged 3 to 17 years. These findings are based on standardized BP measurements using a validated oscillometric device (Datascope Accutorr Plus) on a nationally representative sample of 12 199 nonoverweight children from the KiGGS 2003–2006 study in Germany. These references use both a database with high validity as a normative population and improved statistical methods for percentile derivation. However, KiGGS BP data are influenced by the BP epidemic, which has touched Germany.17 Because of the strong relationship between BP and overweight and obesity,26 the inclusion of this high proportion of overweight subjects would raise the threshold for normal BP (because it would include those with obesity-induced BP elevations in the normative population). As a result, obesity-related BP elevations would be more difficult to detect (ie, the references would be less sensitive to obesity-related hypertension than references based on older data). To avoid this, we excluded overweight children from the reference population.
The widely used Fourth-Report BP references12 did not exclude overweight children, but we compared them with the KiGGS references because they are both screening and detection tools for the same problem—elevated BP in children. As expected, we found mostly lower 90th and 95th percentiles in the KiGGS compared with the Fourth Report (by 2–6 mm Hg). In a reanalysis of the Fourth-Report sample excluding overweight children,27 prehypertension thresholds were only 1 to 3 mm Hg lower than in the Fourth Report. Our analyses confirm these rather small differences between percentiles with and without overweight children. A few mm Hg may be less relevant for the individual diagnosis of hypertension, for which measurements on multiple occasions are needed and both intraindividual variation and inaccuracy of the BP device may be larger. However, for population monitoring of BP levels over time and for international comparisons, a shift of the whole BP distribution of a few mm Hg is rather large because it has a substantial effect on the prevalence of hypertension28 and on hypertension-related outcomes. In adults, for example, a 4–mm Hg difference in mean blood pressure level may lead to a 20% difference in stroke death.29
Additional reasons that may contribute to the lower KiGGS nonoverweight BP percentiles compared with the Fourth-Report percentiles are the oscillometric BP measurement in the KiGGS, which according to the Datascope Accutorr validation study in children30 may lead to readings ∼1 mm Hg (mean) less than auscultatory SBP and DBP, and the use of the mean of 2 readings in the KiGGS. However, not all KiGGS percentiles are lower, and it is not possible to merely modify the Fourth-Report percentile cutoff for a definition of prehypertension or hypertension. For example, thresholds for hypertensive SBP in boys aged >14 years are higher than Fourth-Report thresholds (eg, for median CDC height 4 mm Hg higher SBP 95th percentile in boys aged 17 years). Of note, some of the KiGGS differences compared with the Fourth-Report findings, in particular the lower Fourth-Report SBP 95th percentile in adolescent boys, may be a statistical effect because of different statistical models. We applied the Fourth-Report model to KiGGS data and found, for example, for a boy 17.5 years old and 176 cm tall, a SBP 95th percentile of 137 mm Hg, whereas the GAMLSS method yielded 141 mm Hg. It has been shown recently that the Fourth-Report model assumptions do not hold true and that methods which do not require normal distribution of BP or constant variance at all ages (eg, the GAMLSS method we have used) provide a better fit.27,31 Thus, Fourth-Report BP percentiles seem less appropriate for Germany, not only because of geographic, ethnic, and time-related differences in the reference populations, but also for statistical reasons.
The old European references7 are based on heterogeneous pooled data from 1975 to 1984 and are stratified according to age or height alone. However, our data show that at the same height (rounded to 5 cm), boys may differ in age by up to 5 years (5th–95th percentile of age) and girls by up to 7 years, with resulting SBP 95th percentile differences of 12 and 6.5 mm Hg. Conversely, at the same age, height differences in boys and girls of up to 28 and 23 cm are possible (height 95th–5th percentile) with maximum SBP 95th percentile differences of 8 and 6 mm Hg. The old European references represented the best data available at the time but have various limitations such as selection bias and heterogeneous measurement methods. This may explain the rather large differences in systolic BP distribution compared with the KiGGS (Fig 2). These differences are best seen by comparing the older European references with the KiGGS percentiles by age including overweight children: KiGGS SBP 95th percentile by age including overweight children are 6 to 7 mm Hg lower in most age groups in boys and 3 to 7 mm Hg lower in girls compared with the old European references. A secular trend in BP cannot be excluded but should not be affirmed on the basis of a comparison of KiGGS data with old European data, which have too many methodologic caveats.
BP references that include age, gender, and height together are scarce. In addition to the Fourth-Report research12 and the separate analysis of these data after exclusion of overweight children,27 Norwegian oscillometric (Criticare 507N [Criticare Systems Inc., Waukesha, Wisconsin, USA]) BP percentiles by age and height have been reported from nonoverweight adolescents aged 13 to 18 years.11 However, they are considerably higher not only than the KiGGS percentiles from nonoverweight children but also the British and US references that did not exclude overweight children.10,12 Recently, Hong Kong Chinese BP references by age and height based on the same oscillometric BP-measuring device as used in the KiGGS have been published.32 They included overweight children and are mostly higher than KiGGS references from nonoverweight children but are quite similar to the KiGGS overall sample distribution.
Major strengths of the KiGGS BP references are the large and nationally representative sample, coverage of a wide age range, standardized measurements of BP and height, use of a BP device validated in children, averaging of 2 BP measurements per participant, and the modeling by age and height simultaneously with flexible statistical techniques that do not impose normality or constant variance assumptions on the data.
A possible limitation of the KiGGS is the use of an oscillometric BP-measuring device, whereas current guidelines still recommend the auscultatory method as a first choice.8,12 However, these guidelines acknowledge that some oscillometric devices have been successfully validated using established protocols and that because of the banning of mercury devices, oscillometric reference data will be increasingly needed. In addition, oscillometric BP measurements have the advantage of largely eliminating observer error. The Datascope Accutorr Plus device has passed the standards of the Association for the Advancement of Medical Instrumentation33 and of the British Hypertension Society34 in adults and has been also validated in children aged 5 to 15 years against mercury sphygmomanometric measurements according to the international protocol of the European Society of Hypertension.35 In the validation study in children,30 Datascope Accutorr Plus closely matched sphygmomanometric measurements: the mean (SD) of the differences for SBP readings (oscillometric minus auscultatory) was −0.9 (4.3) mm Hg and for DBP it was −1.2 (6.5) mm Hg. Although this is a closer match than previously reported for other oscillometric BP measurement devices, validation protocols often differ.
Another possible limitation of the KiGGS is selection bias. However, the response rate was good (67%), and two-thirds of nonresponders answered a short questionnaire including self-reported height and weight. Self-reported BMI of responders and nonresponders by age and gender was not significantly different,36 indicating that an adverse impact of nonresponder bias is unlikely.
The references presented here are, to our knowledge, the first European BP references by age and height simultaneously based on oscillometric measurements with a device validated in children. Compared with current US BP references, the proposed KiGGS BP reference values are not influenced by the prevalence of overweight children in the reference population and take advantage of improved statistical methods.
Follow-up BP measurements in KiGGS participants will be forthcoming and will allow an analysis of the predictive value of the presented thresholds for reaching adolescent and early adult hypertension thresholds as well as tracking of specific BP percentiles.
The KiGGS was funded by the German Ministry of Health, the Ministry of Education and Research, and the Robert Koch Institute.
We thank Professor Wolfgang Rascher for valuable advice on BP measurement issues, Professor Johannes Peter Haas for discussions on BP plausibility checks, and Dr Karen Atzpodien for advice on exclusion criteria.
- Accepted December 10, 2010.
- Address correspondence to Hannelore K. Neuhauser, MD, MPH, Robert Koch Institute, Department of Epidemiology and Health Reporting, General-Pape-Strasse 62-66, 12101 Berlin, Germany. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- BP =
- blood pressure •
- KiGGS =
- German Health Interview and Examination Survey for Children and Adolescents •
- CDC =
- Centers for Disease Control and Prevention •
- SBP =
- systolic blood pressure •
- DBP =
- diastolic blood pressure •
- GAMLLS =
- generalized additive models for location scale and shape
- Daniels SR,
- Loggie JM,
- Khoury P,
- Kimball TR
- Chen X,
- Wang Y
- Jackson LV,
- Thalange NK,
- Cole TJ
- 12.↵National High Blood Pressure Education Program Working Group on High Blood Pressure in Children and Adolescents. The fourth report on the diagnosis, evaluation, and treatment of high blood pressure in children and adolescents. Pediatrics. 2004;114(2 Suppl4th report):555–576
- Kahl H,
- Schaffrath Rosario A,
- Schlaud M
- Neuhauser H,
- Thamm M
- Rigby RA,
- Stasinopoulos D
- Sorof J,
- Daniels S
- Rosner B,
- Cook N,
- Portman R,
- Daniels S,
- Falkner B
- Rose G,
- Day S
- Kurth BM,
- Schaffrath Rosario A
- Copyright © 2011 by the American Academy of Pediatrics