Reference Values of Pulse Wave Velocity in Healthy Children and Teenagers
Carotid-femoral pulse wave velocity is an established method for characterizing aortic stiffness, an individual predictor of cardiovascular mortality in adults. Normal pulse wave velocity values for the pediatric population derived from a large data collection have yet to be available. The aim of this study was to create a reference database and to characterize the factors determining pulse wave velocity in children and teenagers. Carotid-femoral pulse wave velocity was measured by applanation tonometry. Reference tables from pulse wave velocities obtained in 1008 healthy subjects (aged between 6 and 20 years; 495 males) were generated using a maximum-likelihood curve-fitting technique for calculating SD scores in accordance with the skewed distribution of the raw data. Effects of sex, age, height, weight, blood pressure, and heart rate on pulse wave velocity were assessed. Sex-specific reference tables and curves for age and height are presented. Pulse wave velocity correlated positively (P<0.001) with age, height, weight, and blood pressure while correlating negatively with heart rate. After multiple regression analysis, age, height, and blood pressure remained major predictors of pulse wave velocity. This study, involving >1000 children, is the first to provide reference values for pulse wave velocity in children and teenagers, thereby constituting a suitable tool for longitudinal clinical studies assessing subgroups of children who are at long-term risk of cardiovascular disease.
Cardiovascular disease is the leading cause of death in Western societies.1 Although there is ample evidence that arteriosclerosis begins in childhood,2–4⇓⇓ hard end points, such as stroke and ischemic heart disease, are rare or virtually lacking in the pediatric population. There is, thus, an increasing need to establish validated noninvasive tools to forecast early arterial disease and to be able to characterize elevated cardiovascular risk in youngsters.5
It has been widely recognized that aortic pulse wave velocity (PWV) is a sensitive marker of arterial stiffness and, consequently, of cardiovascular outcome.6–9⇓⇓⇓ However, large multicenter clinical studies aimed at generating normal PWV values and assessing the influence of anthropometric factors on PWV in healthy children and teenagers are still lacking.
In our previous investigations, we showed that, in specific patient populations with growth retardation, the use of age-matched controls failed to reflect the true impact of cardiovascular disease on PWV. Rather, controls matched for both age and height or normalized PWV data such as PWV Z score or PWV/height should be used in such instances.10–12⇓⇓
The aim of the present study was to create a reference database and to characterize the factors determining PWV in children and teenagers. To achieve this, the distribution mode of PWV was evaluated in a large group of healthy children and adolescents using a novel statistical approach, the LMS method (where L indicates skewness, M indicates median, and S indicates coefficient of variation), to calculate SD score (SDS) values in accordance with the skewed distribution of the raw data. In addition, the relationship among anthropometric factors, blood pressure (BP), heart rate (HR), and age- and height-normalized PWV was also analyzed. Finally, an analysis assessing the correlation between novel PWV Z scores for age and height was also performed.
Subjects and Methods
A total of 1008 healthy children and teenagers (mean age: 15.2 years [range: 6.5 to 19.9 years]; 495 males) was assessed between 2006 and 2009. Only children with no history of diseases affecting BP and without current antihypertensive medication or other BP-affecting drugs were included in the study.
A total of 450 Hungarian (12.4 years [range: 6.5 to 19.9 years]; 214 males), 455 Italian (17.3 years [15.8 to 19.9 years]; 178 males), and 103 Algerian (17.8 years [16.4 to 19.8 years]; all males) healthy school children and adolescents participated in the study. The study was conducted at 3 primary and 2 high schools in Hungary and in 1 high school in both Italy and Algeria, respectively.
Anthropometric measurements were performed by trained staff from the school. Weight and height were measured with precision electronic scales and fixed stadiometers.
BP was measured by oscillometric devices (Omron M742E and Omron 705IT, Omron Co). The technical validity of these monitors has been confirmed previously in adults, children, and adolescents.13–15⇓⇓ The midarm circumference was measured and the cuff width adapted accordingly (pediatric, standard adult, or oversized cuff for midarm circumference of 17 to 22, 22 to 32, or 32 to 42 cm, respectively). Systolic BP (SBP), diastolic BP (DBP), mean arterial pressure (MAP), and heart rate (HR) were calculated as the mean of the 3 measurements. MAP was calculated as (SBP+2×DBP)/3. To allow for comparison between different age groups, SDSs were calculated.
PWV was measured by applanation tonometry using the PulsePen device (DiaTecne s.r.l.)16 interfaced with a computer, as described previously.12,16⇓ Briefly, the probe was connected to a hand-held ECG unit while pressure and electrocardiographic signals were transmitted to a computer. Aortic PWV was measured by sequential recordings of the arterial pressure wave at the carotid and femoral arteries and defined as the distance of the sampling sites divided by the time difference between the rise delay of the distal and proximal pulse according to the R wave belonging to the ECG qRs complex and calculated by the software. The pulse wave was calibrated by measuring BP immediately after each recording. To assess pulse wave travel distance, surface tape measurements were performed between the carotid site and the jugular notch and between the jugular notch and the femoral site. The difference between these 2 distances was considered as the pulse travel distance.17 The measurement of transit time was discarded and repeated if BP and HR varied by >10% in the carotid and femoral sites. Recordings were also discarded when the variability between consecutive systolic or diastolic waveforms was >10% or when the amplitude of the pulse wave signal was <80 mV.
To avoid possible methodological bias (center effect), a single senior investigator (P.S.) trained and supervised the staff performing all of the PWV measurements. All of the measurements were performed twice to confirm reproducibility, with the resulting PWV value consisting of the mean of both measurements. The intraobserver and interobserver coefficients of variation of distance measurements were 2.3% and 4.0%, respectively. The intraobserver and interobserver coefficients of variation of PWV measurements were 5.7% and 6.1%, respectively.
Evaluation of Sex Differences
To assess sex differences in PWV, pairs composed of males and females matched for age and height were formed and compared. The maximum interindividual difference allowed within 1 pair was <5 cm for height and <1 year for age.
Normalization of PWV Data: The LMS Method
Age- and height-specific reference values for PWV were generated by the LMS method,18 which characterizes the distribution of a variable by its median (M), the coefficient of variation (S, ie, the ratio of the SD and mean), and skewness (L) required to transform the data to normality. Evaluation for these parameters is obtained by a maximum-likelihood curve-fitting algorithm to the original data plotted over the independent variable. In this study, 2 sets of tables were created, one using age and the other height as independent variables.
L, M, and S can be used to create percentiles (Cα) according to the following equation (1):
where M(t), L(t), and S(t) or Cα(t) indicate the corresponding values of each parameter at a given age or height (t). zα is the normal equivalent deviate corresponding with the centile (eg, α=50, zα=0; α=75, zα=0.674; α=90, zα=1.282; α=95, zα=1.645; and α=97, zα=1.881).
where Y is the child’s individual parameter (PWV), and L(t), M(t), and S(t) are the specific values of L, M, and S interpolated for the child’s age or height.
Data Analysis and Statistics
The data analysis was performed using the STATISTICA 8.0 (Stat Soft Inc). The Shapiro-Wilks test and normal probability plot analysis were used to characterize the normality of data distribution. Data with nonnormal distribution are presented as median (range), whereas normally distributed data are expressed as mean and 95% CI. SDSs for height, weight, body mass index, SBP, and DBP were calculated using standard reference charts.19
Univariate correlation and regression models were performed to assess the relationship between anthropometric data and PWV. Major determinants of PWV were established in accordance with the backward multiple regression model.
Anthropometric and clinical data were compared by Student t test or by ANOVA where appropriate. The Mann-Whitney U test was used to compare data with nonnormal distribution. A P<0.05 was considered as statistically significant.
The study protocol was approved by the local ethics committees. Participants provided informed consent, and for underage subjects, a parental written informed consent was obtained.
The mean values of anthropometric and BP data of subject groups from the 3 geographic regions were within the normal limits of the 25th to 75th percentile range for both age and sex. Furthermore, there was no significant difference between PWV data obtained from the different regions within the respective age quartiles. Thus, the data were subsequently pooled for further analysis and summarized in Table 1.
The PWV of boys and girls was similar in the first 2 age quartiles. In contrast, in the third and fourth quartiles, boys had significantly higher PWV, as well as height and BP data, compared with the girls. Sixty-six pairs of boys and girls matched appropriately for age and height were ultimately formed to evaluate the effect of sex on PWV. Their comparative analysis revealed no differences in anthropometric, BP, or PWV data. However, age- and sex-specific SDSs for height and weight of girls were significantly higher compared with their matched counterparts (height SDS: −0.081 [range: −0.249 to 0.086] versus 0.962 [range: 0.762 to 1.161], P<0.05; weight SDS: 0.034 [−0.140 to 0.208] versus 0.353 [0.188 to 0.518] P<0.05 for males and females, respectively; please see also Table S1 in the online Data Supplement at http://hyper.ahajournal.org). Thus, because of sex differences in growth and height, the standardized values of PWV are presented separately for sexes.
Univariate regression analysis revealed a strong positive correlation among age, height, weight, SBP, DBP, MAP, and PWV (r=0.47, 0.44, 0.39, 0.43, 0.33, and 0.43; P<0.00001). Conversely, there was a negative correlation between HR and PWV (r=−0.08; P=0.02). By multiple regression analysis, age, height, and MAP (ß=0.21, 0.16, and 0.24 and SE of ß=0.04, 0.039, and 0.029, respectively; P<0.001) proved to be major determinants of PWV in the healthy population (see also Table S2). Accordingly, PWV of healthy children may be expressed as follows:equation
Standardized PWV Data
PWV of the 1008 subjects showed a nonnormal distribution (P<0.001). The fitting procedure used for the transformation of the raw PWV data to Z scores resulted in a PWV zageof 0.0008 (0.995; mean [SD]) and a PWV zheightof 0.0901 (1.024), indicating effective normalization of the skewed original data. LMS values were determined and percentile boundaries were calculated and plotted according to sex, age, and height (Table 2⇓ and Figure).
A close correlation was found between PWV Z for age and for height (r=0.95; P=0.00001; see Figure S1). An individual example for the practical use of tables and data transformation is provided in the online Data Supplement.
This study is the first report to provide novel, distribution-independent age- and height-specific reference values for PWV in children and teenagers, making them suitable for the evaluation and follow-up of subgroups in the pediatric population who are at risk for long-term cardiovascular disease. PWV is a powerful parameter of arterial stiffness. It is a surrogate marker for cardiovascular events in adult hypertension and end-stage renal disease20–22⇓⇓; however, such data in children are sparse.5,10–12,23⇓⇓⇓⇓ The difficulty of using PWV in children is its dependence on age and body dimensions.10–12⇓⇓ To evaluate the data obtained in different pediatric patient groups, the current practice is to use either controls matched for age or to use age-specific normal values established in large healthy populations. Unfortunately, the problem of growth deficit is not solved by this approach.
In a study of children on hemodialysis, Covic et al23 found an increase in PWV compared with healthy controls. However, in their study, age- and height-matched controls were used for comparison purposes. In our previous study aimed at confirming the increase in PWV in children on dialysis, to our surprise we only noted a weak (not significant) tendency toward an increase in PWV in dialysis patients. Because of the growth deficit caused by end-stage renal disease, we were, therefore, compelled to use a control group matched for age and height to demonstrate the true magnitude of PWV differences.11 In a subsequent step, normative PWV values were established with the use of arithmetic mean and SD of a database of 188 healthy children.12
In the present work, using an extended database of 1008 healthy children and young adults, we applied the LMS method, which takes into account asymmetrical data distribution to create reference tables for sex-, age-, and height-specific percentiles of PWV in healthy children and teenagers. This type of approach is widely accepted in pediatrics. The universally used reference tables for 24-hour BP monitoring, as well as normative values for intima-media thickness in healthy adolescents, are based on this very principle.24,25⇓
Herein, analysis of the raw PWV data revealed a non-Gaussian distribution across the entire pediatric age range. The skewed distribution invalidates the calculation of conventional SD (Z) scores based on arithmetic means and SDs. On the other hand, the LMS method is a technique accounting for any degree and direction of distribution skewness of the sample. It has the advantage of not requiring grouping of the covariate for the calculation of percentiles, thereby avoiding the distorting effects of variable cell sizes. Hence, the L, M, and S reference values published herein are readily applicable for defining PWV abnormalities independent of age and of body size in diseased pediatric populations, and the statistical basis of the PWV percentiles provided here is superior to our previous estimates.12
The growth pattern and final height of the sexes differ significantly. Thus, although the PWV of age- and height-matched boys and girls is similar, as shown herein and in our previous studies,11,12⇓ the PWVs of boys and girls did differ significantly in the third and fourth age quartiles, that is, after puberty. This is because the girls matched to the boys are taller than the average of their age group; in other words, they have a higher height SDS. Hence, because of the difference in height development between sexes, the normalized data obtained for boys and girls are presented separately.
The positive correlation of PWV with age, body dimension, and BP, as well as the negative correlation with HR, was confirmed in this large population of healthy children.11,12⇓ However, according to the results of the multivariate regression analysis, only age, height, and MAP were independent predictors of PWV. Based on the multivariate model, a general equation can be generated allowing for the prediction of the PWV value, as shown in the Results section. The use of MAP herein was justified by previous data showing that, aside from age, MAP (but not SBP or DBP) proved to be an independent predictor of carotid-femoral PWV.26
The reference tables generated herein expressing PWV according to height are presented for use in pediatric populations with growth deficit. The close relationship between PWV Z for age and PWV Z for height values indicates that smaller children have lower PWV values, whereas taller children have higher PWV values in any given age group. Evidence that the dimensions and elastic properties of the arterial tree are intimately related to height was initially provided by morphological studies.27,28⇓ The physiological basis of the influence of body size on arterial wall properties was further demonstrated by Senzaki et al,29 who established reference ranges for age-associated changes in arterial pulsatile properties in 112 pediatric patients after cardiac catherization. Their results indicated a progressive increase in arterial compliance despite a decrease in arterial wall elasticity, leading to the conclusion that the increase in arterial size accompanying increased body size outweighs the effects of age on intrinsic elastic properties of arterial walls. Such influence of body dimensions on the elastic properties of arteries was also demonstrated in a more recent study by Jourdan et al,25 in which both intima-media thickness and arterial stiffness were shown to change with age and body size. Accordingly, the authors concluded that morphological and functional measurements of large arteries should be normalized to take into account changes during adolescence.
The reference tables provided in this multicenter study involving a cohort of >1000 children and teenagers are based on the LMS method, which enables the calculation of appropriate SDS values of PWV in children and teenagers. Age, height, and MAP were found to be the major determinants of PWV, hence emphasizing the need to consider height in patients with growth deficits.
This study is the first report to provide distribution-independent age and height-specific reference values for PWV in children and teenagers. Because the traditional end points such as stroke, myocardial infarction, and mortality used in adult studies are unsuitable to evaluate the risks or benefit of a given intervention or treatment in pediatric clinical trials, surrogate markers are needed in pediatric studies. Examples for such studies are the use of angiotensin-converting enzyme inhibitors or angiotensin receptor blockers in hypertensive and/or diabetic children and the use of calcium-free phosphate binders in uremic children. PWV is a sensitive marker of arterial stiffness and, consequently, of cardiovascular outcome in adults.1,2,4⇓⇓
Our reference values in children represent a suitable tool for use in such longitudinal interventional studies, as suggested by the recent scientific statement of the American Heart Association.5 However the final evaluation and validation of the method to establish the cutoff points for predicting hard cardiovascular end points could only be provided by the follow-up of children growing into adulthood.
A complete analysis of all of the risk factors potentially influencing PWV in this healthy, young population was beyond the scope of the original goals of this study. We did perform a subgroup analysis in 300 healthy children and could not establish any correlation among (normal) glucose and lipid values and PWV. Because the analysis was not complete, we did not include all of these results into the article. For follow-up studies in populations with increased cardiovascular risk, it will be necessary to include these data in the analysis.
Sexual steroids could influence the elastic properties of the arteries in women. We did not address this question in the present study; however, in their work, Robb et al30 could show that PWV is not significantly affected by the periodical physiological changes of steroid hormones in women.
We thank Prof Dea Campana, the headmistress of the “Liceo Scientifico Augusto Righi” high school in Cesena; Prof Marino Mengozzi, the assistant headmaster, and Veronika Hámori and Dr Mária Papp, the directresses of the “Fazekas” and “Raoul Wallenberg” high schools, respectively; Kolos Kovács, the directress of the “János Arany Primary and High School”; and László Bodó, the director of the “Bakáts Square Primary School” in Budapest, as well as the staff and all of the students for their invaluable cooperation. We are particularly grateful to Prof Giovanna Amaduzzi, Filomena Racioppa, and Maurizio Marchetti, who made the research in Italy possible. We also thank the nursing staff of the Department of Internal Medicine of the Bufalini Hospital in Cesena and of the First Department of Pediatrics at the Semmelweis University in Budapest for contributing to the realization of this study. In addition, we thank Dr Ali Belkheir, public health director of Ghardaia; Bachir Bahaz, director of the Metlili Hospital; the medical team of Metlili and Ghardaia; the staff and the students of Metlili High School; and Fatma Chehima for secretarial service.
Sources of Funding
The study was supported by grants OTKA-071730 (Országos Támogatási Kutatási Alap [National Scientific Research Fund]; to G.S.R.), TÁMOP-4.2.2-08/1/KMR-2008-0004 (Társadalmi Megújulás Operativ Program [Social Renewal Operational Program]; to A.J.S.), and ETT 06-123/2009 (Egészségügyi Tudományos Tanács [Academic Health Council Grant]; to G.S.R.).
P.S. is a consultant for PulsePen (DiaTecne srl, Milan, Italy).
G.S.R. and O.C. contributed equally to this article.
- Received March 4, 2010.
- Revision received April 3, 2010.
- Accepted May 19, 2010.
- ↵Najjar SS, Scuteri A, Lakatta EG. Arterial aging: is it an immutable cardiovascular risk factor? Hypertension. 2005; 46: 454–462.
- ↵Juonala M, Järvisalo J, Mäki-Torkko N, Kähönen M, Viiakari JSA, Raitakari OT. Risk factors identified in childhood and decreased carotid artery elasticity in adulthood: the Cardiovascular Risk in Young Finns Study. Circulation. 2005; 112: 1486–1493.
- ↵Li S, Chen W, Shrinivasan SR, Berenson GS. Childhood blood pressure as a predictor of arterial stiffness in young adults: the Bogalusa Heart Study. Hypertension. 2004; 43: 514–546.
- ↵Urbina EM, Williams RV, Alpert BS, Collins RT, Daniels SR, Hayman L, Jacobson M, Mahoney L, Mietus-Snyder M, Rocchini A, Steinberger J, McCrindle B, for the American Heart Association Atherosclerosis, Hypertension and Obesity in Youth Committee of the Council on Cardiovascular Disease in the Young. Noninvasive assessment of subclinical atherosclerosis in children and adolescents: recommendations for standard assessment for clinical research: a scientific statement from the American Heart Association. Hypertension. 2009; 54: 919–950.
- ↵Bramwell LC, Hill AV. Velocity of transmission of the pulse wave. Lancet. 1922; 197: 891–892.
- ↵Khosdel AR, Carney SL, Nair BR, Gillies A. Better management of cardiovascular diseases by pulse wave velocity: combining clinical practice with clinical research using evidenced based medicine. Clin Med Res. 2007; 5: 45–52.
- ↵Pannier B, Guerin AP, Marchais SJ, Safar ME. London GM. Stiffness of capacitive and conduit arteries: prognostic significance for end-stage renal disease patients. Hypertension. 2005; 45: 592–596.
- ↵Laurent S, Cockcroft J, Van Bortel L, Boutouyrie P, Giannattasio C, Hayoz D, Pannier B, Vlachopoulos C, Wilkinson I, Struijker-Boudier H, for the European Network for Non-invasive Investigation of Large Arteries. Expert consensus document on arterial stiffness: methodological issues and clinical applications. Eur Heart J. 2006; 27: 2588–2605.
- ↵Cseprekal O, Kis E, Schaffer P, Othmane Tel H, Fekete BC, Vannay A, Szabo AJ, Remport A, Szabo A, Tulassay T, Reusz GS. Pulse wave velocity in children following renal transplantation. Nephrol Dial Transplant. 2009; 24: 309–315.
- ↵Weber T, Ammer M, Rammer M, Adji A, O'Rourke MF, Wassertheurer S, Rosenkranz S, Eber B. Noninvasive determination of carotid: femoral pulse wave velocity depends critically on assessment of travel distance–a comparison with invasive measurement. J Hypertens. 2009; 27: 1624–1630.
- ↵Laurent S, Boutouyrie P, Asmar R, Gautier I, Laloux B, Guize L, Ducimetiere P, Benetos A. Aortic stiffness is an independent predictor of all-cause and cardiovascular mortality in hypertensive patients. Hypertension. 2001; 37: 1236–1241.
- ↵Covic A, Mardare N, Gusbeth-Tatomir P, Brumaru O, Gavrilovici C, Munteanu M, Prisada O, Goldsmith DJ. Increased arterial stiffness in children on haemodialysis. Nephrol Dial Transplant. 2006; 21: 729–735.
- ↵Robb AO, Mills NL, Din JN, Smith IBJ, Paterson F, Newby DE, Denison FC. Influence of the menstrual cycle, cregnancy, and preeclampsia on arterial stiffness. Hypertension. 2009; 53: 952–958.