Introduction

Arterial blood pressure (BP) is a basic vital parameter that is measured at all ages1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48 (http://www.mercuryfreehealthcare.org). In adults, elevated BP is defined as a risk-related threshold, that is, as that level of BP above which clinical trials have shown (and international guidelines have agreed upon) that lowering BP reduces the likelihood of CV diseases and death.47, 48 This threshold is usually 140/90 mm Hg, and although it may be lower in higher risk conditions (for example, in diabetes), it is not considered to be dependent on other factors such as age, sex, body mass, ethnicity and so on.3, 4 However, an essential aspect of BP in the pediatric and adolescent populations is that it changes (increases) with growth, making BP measurements meaningful only if related to some indices of growth or body size.4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46 As a result, BP values in children, as well as any other parameter,49, 50, 51, 52, 53 need to be normalized according to age and somatic growth.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46 Thus, the availability of robust nomograms is essential for the correct evaluation of BP values in children.49, 50, 51, 52, 53

For the determination of reference values, of paramount importance is the selection of reference individuals based on extensively documented inclusion and exclusion criteria and the use of quality-controlled procedures to collect data.49, 50, 51, 52, 53 When only small numbers of values are available, reference intervals (RIs) may be highly imprecise.49, 50, 51, 52, 53

From a clinical point of view, when multiple nomograms are available and no clear recommendations exist regarding the use of one nomogram over another, a clinician makes an arbitrary choice and relies on it in the decision making process.49, 50, 51 As a result, if various nomograms present significant differences in RIs, this may generate confusion in the estimation of disease severity,49, 50, 51 thus affecting the management of children with hypo/hypertension.

The aim of this study was to critically review the strengths, accuracy and limitations of available pediatric blood pressure nomograms. New perspectives for research in the field are also described to convey the current knowledge gaps.

Materials and methods

Search strategy

Candidate studies for inclusion were identified by a systematic search of the National Library of Medicine (PubMed access to MEDLINE citations, http://www.ncbi.nlm.nih.gov/PubMed/) conducted in March 2013. The period searched covered 1985–2013. The search strategy included a combination of Medical Subject Headings (MeSH) and free text terms for the key concepts starting with BP, pediatric and reference values/nomograms. The search was further refined by adding the keywords neonates/infants/adolescent, range and intervals. In addition, we identified other potentially relevant publications using a manual search of the references from all eligible studies and review articles as well as from the Science Citation Index, expanded on Web of Science.

Titles and abstracts of all articles identified by the search strategy were evaluated and assessed independently by two reviewers; articles were excluded (1) if the studies included populations other than normal subjects or included adults with children (N=6), and (2) if the reports were published in languages other than English (N=16).

Results

A total of 67 publications were identified for potential inclusion in the study. Twenty-two studies were excluded on the basis of the criteria listed above, leaving 45 publications for analysis; 36 surveys2, 3, 4, 6, 8, 9, 10, 11, 12, 13, 16, 19, 20, 21, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 39, 40, 44, 45, 54, 55 were performed with the specific aim of providing reference BP values for children in a definite population, whereas the other surveys1, 5, 18, 23, 24, 36, 37, 42, 43, 46 were for other purposes, such as comparison with reference data, identification of time trends, comparison among different ethnicities and testing of the methodological issues related to BP normalization in the pediatric age group.

Numerical issues

Nomograms divided according to geographic area

In Table 1, major pediatric BP nomograms, divided according to the major region of origin, are detailed. For some countries/continents (North America, Europe, Middle East, part of Asia), there are one or more nomograms calculated over a wide sample of healthy children (from 2876 to 63 227 healthy children).6, 9, 39, 45, 54, 55 In contrast, some geographic areas, particularly Africa, East Europe, South America and wide areas of Asia (particularly China and Russia) are poorly/not represented.

Table 1 Major pediatric blood pressure nomograms dividedPl according to geographic area
Full size table

Neonates and infants

The data on neonates and infants are extremely limited (0–12 months) with only nine studies2, 3, 14, 17, 29, 30, 31, 32, 33, 34 exhibiting a relatively good sample size of healthy subjects (over 400 subjects) (Table 2). Most of the above studies evaluated neonates at birth,2, 17, 29, 30, 31, 32, 34 and the majority contained measurements repeated at 4 days of life.2, 30, 32, 33, 34 Few data are available for days 2–3 of life,17, 30, 32, 34 and up to 6 months;2, 30, 33 data at 6 and 12 months2, 31, 33 are also limited. Some ethnicities, in particular black, are almost absent, and some geographic areas are not represented at all (East Europe, Asia, South America). Premature and full-term neonates have been considered separately only by a few authors,29, 34 and various age intervals have been used in different studies.

Table 2 Major blood pressure nomograms for neonates and infants
Full size table

On the basis of available data, a progressive increase in systolic BP (SBP) and diastolic BP (DBP) seems to occur during the first week of life2, 3, 17, 34 with the highest variations during the first days2, 17, 32, 34 and with little variation thereafter up to 6–12 months of life.2, 17, 31, 33 Weak but significant positive correlations of SPB and DBP with birth weight,2, 17, 29, 32, 33, 34 birth length,29, 32 head circumference29 and gestational age17, 29, 34 were also found.

Naturally delivered neonates had higher mean BP values than neonates delivered by cesarean section29 and a more rapid rate of increase was found in preterm compared with full-term infants.17, 34 As expected, BP was lower when measured during the sleep state than when the child was in a quietly awake state.34

Ambulatory blood pressure and response to stress test

Ambulatory blood pressure monitoring is of great relevance for the diagnosis of hypertension in children because of the white coat and masked hypertension phenomena.56 However, only a few studies have tried to establish reference values for ambulatory BP in normal healthy children57, 58, 59, 60, 61, 62, 63, 64, 65 and in disease states.66 The only normative data generated in a relatively large number of healthy children and adolescents (1141 healthy children and adolescents) are from Germany,62 and there are very limited data for infants and toddlers.67, 68, 69, 70 Special considerations are required for neonates and infants where important variations in BP values may occur depending on sleeping/awake state. In particular, infants frequently exhibit sudden hypotensive or hypertensive events during sleep because the autonomic nervous system is immature or may be altered in preterm babies.69, 70 Finally, data on BP response to exercise in children are extremely limited,71, 72, 73 and no clear BP cut-off values exist to distinguish between the healthy and pathological states; in addition, there are no criteria for the interruption of stress tests in cases of excessive BP increases.73

Methodological issues

Selection criteria

Inclusion/exclusion criteria have been reported in many but not all of the nomograms. In particular, studies involving neonates generally described the criteria employed to select the healthy subjects in some detail3, 14, 17, 29, 30, 31, 32, 34 with a few exceptions.2, 33 In contrast, in articles involving older children, the inclusion/exclusion criteria were often not reported.6, 9, 10, 12, 19, 24, 26, 28 Furthermore, data from the Fourth Report of the National Health and Nutrition Examination Surveys and from other US studies have been utilized by various authors with different selection criteria.5, 24 The National High Blood Pressure Education Program Working Group on High BP in Children and Adolescents decided to use height (rather than weight or body mass index (BMI)) in constructing the nomograms representing 63 227 American children to discourage the misinterpretation of relatively high BP as normal just because a child is overweight/obese.9 In fact, overweight children tend to exhibit19, 21, 24, 36 higher BP values compared with normal weight children of the same age and height. However, overweight children were still included in the normative database.7, 9 To overcome this limitation, Rosner and colleagues, using the same data, rebuilt the nomograms excluding overweight children,21 which produced only slightly lower nomograms.

Methods of BP measurement

Various methodological issues affected the evaluation of the method of BP measurement, including the use of different recorders, the cuff sizes, operator skill and the number of measurements (Table 3). Consistent with both US,9 European44 and Asian74 guidelines, most of the studies used the auscultatory manual method for BP measurement in older children, whereas the oscillometric method was employed in neonates and infants. Regarding recorders however, aneroid sphygmomanometers should be considered the method of choice, because mercury sphygmomanometers are being progressively banned worldwide (http://www.mercuryfreehealthcare.org) as a result of being poorly employed. Furthermore, only a few oscillometric devices have been validated in children.1 Finally, regarding auscultatory measurements, in almost all recent surveys, the disappearance of Korotkoff sounds (K5) was established as the definition of DBP, whereas some older studies used K4.9, 26, 45 In the majority of the surveys,2, 6, 8, 11, 26, 27, 28, 30, 31, 32, 33, 34, 35, 39 multiple cuff sizes were used according to arm size, although the criteria for selection of cuff length and width differed among the various studies.

Table 3 Methodologies of major nomograms
Full size table

A standard protocol9 (right arm, sitting position and rest period preceding the measurement) was usually observed, with few exceptions.10 In particular, the position was not always indicated2, 8, 30, 31 or it was supine,10, 32, 34 especially in neonates (where there are obvious limitations in obtaining a fixed position). Multiple readings taken at a single visit were common in most studies, whereas in a few studies, the data were based on single readings.9, 21 Only a few reports indicated the time of the day (mostly in the morning)2, 8, 16, 54 and the seasons in which the data were collected.6, 17, 40, 54 Table 3. Finally, intra-observer and inter-observer variability were generally ignored, with only a few exceptions.27

Statistical issues

Data normalization and expression

In children, BP measurements must be adjusted for body size, age and/or some other parameter to compare the values with those from a healthy pediatric population.9 Most of the studies had estimated values of BP according to age2, 3, 4, 5, 6, 8, 11, 12, 14, 16, 17, 19, 28, 31, 32, 33, 39, 40, 45, 54 or height35 or both.9, 10, 13, 20, 21, 26, 27, 35 In the majority of the studies, significant linear correlations with age were observed both for systolic and DBP with widely ranging ‘r’ values3, 4, 6, 8, 9, 11, 13, 19, 21, 27, 28, 35, 45, 54 (SBP r: 0.26–0.64; DPB r: 0.018–0.56); in a few instances, statistically significant correlations were found for systolic but not DBP.11 An acceleration of the increase in BP values during puberty was also reported.39

In some studies, linear correlations with height9, 11, 13, 19, 21, 27, 28, 33, 35, 39, 40, 45 were demonstrated more strongly than with age (SBP r: 0.11–0.64; DBP r: 0.13–0.65).14, 35, 45 The use of age and height to normalize data was confronted with a number of potential problems. First, the age range and the height intervals used, as well as the number of subjects enrolled for various age/height subgroups differed widely (Table 1). In particular, the lower ages (that is, < 5–6 years) were generally less well represented, and for some age groups (especially neonates and infants), the data were very limited. Furthermore, the choice to rank BP according to just age and height implies, by definition, no consideration of body weight. BP, however, was found to strongly correlate with body weight,2, 28, 33, 39, 40, 45 BM,8, 12, 25, 36, 45 body surface area (BSA)28, 31 and waist circumference.39, 75

Data expression also varied among the publications. The data were generally expressed by percentile table2, 6, 8, 9, 11, 13, 26, 27, 29, 35, 39, 45 and/or percentile charts3, 10, 11, 12, 28, 29, 35, 39 and/or mean values plus s.d.,2, 3, 4, 6, 8, 11, 12, 16, 19, 25, 27, 28, 29, 32, 33, 34, 35, 37, 39, 42, 45, 54, 55, 74 without parametric normalization. Nonparametric methods do not assume that the response variable adopts a given distribution and are therefore less prone to bias.49, 50, 51 However, because BP reference ranges change as the body grows, if one wants to define precise reference ranges across growth using non-parametric methods, one must compute the percentiles for several growth strata (from birth to adulthood) of the studied population.49 Parametric normalization (where the dependent variables were SBP and DBP) has only been performed by a few authors8, 12, 25, 26, 27, 35 by using different types (and numbers) of independent variables. The regressions were the same only in two studies,26, 27 but the age groups evaluated were different, making comparisons among the studies difficult. Furthermore, quantile regressions, which offer a greater flexibility and the best fit21 compared with polynomial/spline models, have been seldom employed. Lastly, the issue of heteroscedasticity, a statistical term used to describe the behavior of variance of the residuals, which may create important bias especially when using parametric normalization, has never been addressed.49, 50, 51

Confounders

Other methodological limitations are represented by the different attention to some relevant confounders including gender, ethnicity and geographic regions and socio-economic factors.2, 3, 8, 11, 27, 28, 30, 31, 34, 35, 39, 75, 76, 77, 78, 79

Gender

The data have usually been presented separately regarding gender with limited exceptions.27, 30, 31, 34 Gender differences in BP were found in some2, 3, 35, 39 but not all studies.28 Generally higher BP values have been reported in boys,8, 11, 33 although a few authors29, 35 reported higher values in girls.

Ethnicity

Controversies remain regarding the specific influence of ethnicity on BP values in children. A major issue is whether the higher prevalence of hypertension found in black compared with white adults in the US9 and the UK42 is reflected in BP differences in the pediatric age group. In some series, only slight differences between black and white subjects have been shown, with a prevalence of BP elevation in Hispanic and black male youth compared with white male youth in the USA.25 It is supposed, however, that ethnic differences in the prevalence rates of pediatric BP elevation may be not entirely explained by obesity; rather, they may be because of genetic or environmental factors.77

Geographic areas: continents, countries and regions

Geographic influence may be even more relevant than ethnicity. In fact, several studies highlighted region specific childhood BP distributions. There were important differences among continents, countries and even within same the country among various regions.20, 45 Therefore, for instance, studies indicate that BP may be higher in European and Asian children than in their American counterparts.6, 16, 40 Sanchez et al.6 found that SBP and DBP values were higher in 34 946 Spanish children than in their US counterparts. Furthermore, Tobagonian adolescents had SBP and DBP that were 10 mm Hg lower and higher, respectively, than their UK counterparts, whereas Jamaican adolescents had DBP consistently lower than Tobagonians.19 Additionally, it has been demonstrated that in Lebanese children, normal US values were too high and neither the European normal standards nor the Italian standards were suitable for this population.45 Heterogeneity within the same country between diverse regional areas has also been reported.20, 41

Other confounders

Many other factors have been reported to variably influence childhood BP; they include school attendance,39 physical activity,39 breastfeeding,39 dietary habits25, 39 and familiarity.25, 39 In particular, in American children, a significant and independent association between high sodium intake and elevated BP has been recently demonstrated.78 Sleep duration (that is, >8.5 h) was also positively associated with high BP in female adolescents.79 However, it is most likely doubtful that all these factors should be taken into account in the construction of BP nomograms or reference tables.

Discussion

In the past 30 years, several surveys2, 5, 6, 8, 11, 12, 13, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 39, 40 and a few reviews4, 9, 10 that provide normative data of childhood BP have been published. Nevertheless, many numerical and methodological limitations still remain. For some geographic areas (North America, Europe, India, Japan), there are one or more robust nomograms providing normative percentiles based on the large size of healthy children, especially for children >5 years of age. The data for many other geographic areas (South America, Africa, East Europe, wide areas of Asia) and for some ethnicities however are extremely limited or even absent. Furthermore, globally, there is a paucity of data for neonates and infants. Finally, data on ambulatory BP and BP response to exercise in children are extremely limited,57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73 and no clear BP cut-off values exist to distinguish between healthy and pathological states. Additionally, the criteria for interruption of stress tests in the presence of excessive BP increases are lacking, which could because for concern.73

There are also various methodological limitations including the lack of clear inclusion/exclusion criteria with a few exceptions,9, 13, 30, 31, 32, 39, 45 differences in the method employed for measuring BP (for example, manual vs. automated devices), heterogeneity in the data normalization (according to age and/or height) and units of expression (percentile/mean value). The choice to express data according to age and/or height and not according to BMI was made to avoid the assumption that being overweight may be a justification for high BP at a given age/height.9, 21 The inclusion of overweight/obese in the database however, implies a bias in the nomograms that some authors have tried to overcome by excluding the lowest and highest BMI percentiles.21 The inclusion of data collected over multiple decades (in reviews), also represents another potential bias. As observed by Munter and colleagues7 and more recently by Rosner et al.78 an increase in childhood BP over the past decade has occurred, which may be partially attributable to an increased prevalence of overweight children.

Interestingly, no studies reported BP values normalized for BSA, which is the parameter that is commonly used to normalize some pediatric data (such as cardiac structure dimensions).50 Usually, parametric normalization by percentile has been used, because it is easy to understand and is very familiar to most pediatricians. However, an insufficient number of healthy subjects have been evaluated at times, especially in the case of neonates and infants.47, 48 Furthermore, the use of percentiles in a pediatric population implies a division (from 0 to 18 years) in growth strata,49 but no clear rules exist on how this division should be performed. Conversely, non-parametric methods of normalization (that is, z scores) have been rarely used.21, 47 Z-scores (that is, the number of s.d. from the mean) however, may help in eliminating a number of the sources of variance in raw numbers, especially when there is a significant variation in body size, and they allow for a further simplification of data interpretation, also based on the use of electronic computation.49, 50, 51

Among the confounders, the issue of ethnicity remains debatable. In fact, differences among nomograms from various continents have been highlighted;6, 18, 19, 25, 38 however conversely, in American children, the influence of ethnicity/race did not seem to be relevant for BP values.46 In fact, the higher BP values encountered in black male youths7, 24, 46 seem to be more attributable to the incidence of overweight in this subgroup rather than to ethnicity per se.37

All these numerical and methodological issues are not self-limiting but may have a series of important consequences for clinicians.49, 50, 51 In fact, the lack of clear recommendations regarding the nomogram that should be employed (with the exception of the USA) forces the clinician to arbitrarily chose one (or more nomograms) among those available. Secondly, clinicians interpret blood pressure values and make decisions regarding diagnosis (hyper/hypotensive or not?) and management (to treat or not to treat?) according to the proposed RIs, which at times may have been substantially different. In fact, actual nomograms are heterogeneous, difficult to compare and, when comparable, sometimes show considerably different confidence intervals (Tables 4A and B). For instance, according to different authors, a 3-year-old boy, at the 50th percentile of SBP (mm Hg), may vary from 89 (9) to 105.9 (6), whereas at the 95th percentile, the range is from 100.1 (6) to 109.2 (9). Nomograms from the Fourth Report present various advantages: (i) a solid methodological structure despite a few limitations (that is, the lack of clear inclusion/exclusion criteria, the use of data obtained at more than one age vs. cross-sectional data, the use of single readings for some studies, the employment of mercury manometers); and (ii) a good sample size (with the exception of neonates and infants).21 As a result, data from the Fourth report have been frequently adopted worldwide.20, 45 The simple adoption of nomograms from the USA, however, may not be suitable for other geographic areas/countries/ethnic groups20, 44, 45 where they may introduce an unpredictable bias in the evaluation of BP values resulting in significant over/underestimation of hypo/hypertensive values.20, 44, 45 As a result, calls for the development of country/region specific nomograms have been made by multiple sources.20, 44, 45

Table 4a Example of blood pressure normal values for a male child of 3 y.o. according to different nomograms
Full size table
Table 4b Example of blood pressure normal values for a female child of 9 y.o. according to different nomograms
Full size table

Limitations

This review is limited to results from reports written in English. However, there have been several non-English publications showing pediatric BP nomograms especially in Japan and China where English is not an official language. Review and meta-analysis of the data from publications written in a local language could add helpful information to define the pediatric BP ranges of normality for a given country/geographic area.

Conclusion

Various pediatric BP nomograms exist, but many numerical and methodological limitations still persist. For some geographical areas (particularly the USA), there are nomograms of sufficiently good quality, whereas for others, data are limited/absent. Furthermore, worldwide there is a paucity of data specific to the neonatal age group. Various nomograms are difficult to compare, and when comparable, they sometimes show different RIs that may lead to confusion in the interpretation of the BP values in children and in estimating disease severity. Nomograms from the Fourth Report are of good quality but are not universally applicable.

For every country, the development of new nomograms based on a wide range of healthy children including a sufficient number of neonates and infants is recommended. New nomograms should define detailed exclusion/inclusion criteria, standardized protocols (that is, right arm, sitting position, resting period preceding the measurements and multiple readings), methods (that is, the auscultatory manual method in older children vs. the oscillometric method in neonates and infants) and instruments (that is, aneroid sphygmomanometers and multiple cuff sizes) to measure BP. A rigorous statistical approach is mandatory for the building of solid nomograms as well as the evaluation of a series of confounders. In particular, a special attention to gender, ethnicity and geographic area is required and regional differences among the same countries should be evaluated. Lastly, systems to classify the severity of hyper/hypotension (including clear cut-off values) in pediatric patients need to be implemented as well as nomograms for ambulatory and stress BP values. Table 5.

Table 5 Definition of hyper- and hypotension in children
Full size table