Establishing International Blood Pressure References Among Non-Overweight Children and Adolescents Aged 6–17 Years

Abstract

Background

Several distributions of country-specific blood pressure (BP) percentiles by sex, age and height for children and adolescents have been established worldwide. However, there are no globally unified BP references for defining elevated BP in children and adolescents, which limit international comparisons of prevalence of pediatric elevated BP. We aimed to establish international BP references for children and adolescents using seven nationally representative data (China, India, Iran, Korea, Poland, Tunisia and USA).

Methods and Results

Data on BP for 52,636 non-overweight children and adolescents aged 6–19 years were obtained from seven large nationally representative cross-sectional surveys in China, India, Iran, Korea, Poland, Tunisia, and USA. BP values were obtained with certified mercury sphygmomanometers in all seven countries, using standard procedures for BP measurement. Smoothed BP percentiles (50^th, 90^th, 95^th and 99^th) by age and height were estimated using the Generalized Additive Model for Location Scale and Shape (GAMLSS) model. BP values were similar between males and females until the age of 13 years and were higher in males than females thereafter. Compared to BP level of the 90^th and 95^th percentiles of the U.S. Fourth Report at median height, systolic BP of the corresponding percentiles of these international references was lower while diastolic BP was similar.

Conclusions

These international BP references will be a useful tool for international comparison of the prevalence of elevated BP in children and adolescents and may help identify hypertensive youths in diverse populations.

Introduction

Elevated blood pressure (BP) in children and adolescents is a public health concern worldwide,¹ notably because of obesity epidemic and high salt intake.² Elevated BP in children and adolescents is associated with target organ damage³. In addition, elevated BP in childhood moderately tracks into adulthood⁴ which in turn increases the risk of subclinical atherosclerosis among adults.⁵ Hence, decreasing the level of BP in children and adolescents could help reduce the risk of elevated BP related diseases in both childhood and adulthood. BP measurement in children above 3 years of age in the course of routine healthcare is currently recommended by several prominent medical organizations in Europe and the U.S.^6–8 Screening programs require that elevated BP is well defined in children and adolescents. In adult populations, the definition of hypertension is based on the relation between BP and subsequent development of CVD, renal or other related events. However, since it is difficult to assess events related to elevated BP in children and adolescents because of generally long time lag between exposure and outcomes,⁹ elevated BP in children and adolescents is usually defined based on the distribution of BP in a reference population. It is generally considered that children and adolescents have elevated BP if their BP values are in the upper range of the distribution (e.g., hypertension if BP above the 95th percentile). ^{6, 10} In 2004, the US Fourth Report recommended that pre-hypertension and hypertension should be defined based on the age-, sex- and height- specific 90^th and 95^th BP percentiles in the US youths, using data collected between 1973 and 2000 ⁶. These US references are widely used in US and European countries.^{2, 5, 7} However, there are several limitations for the establishment of BP percentiles from the Fourth Report, such as included overweight youths, data collection spanned a large time period, and used data with the first BP reading only. Of note, several studies confirmed that the inclusion of overweight/obese children would, as expected, raise the cut-off points for elevated BP.^11–15 Although several country-specific BP percentiles for children and adolescents have been established worldwide,^11–17 there is no globally unified BP reference for defining elevated BP in children and adolescents, which limits international comparisons of the prevalence of pediatric elevated BP between countries and regions.

In this study, we aimed to develop international BP percentiles by sex-, age-, and height among 52,636 non-overweight children and adolescents aged 6–19 years who had standardized BP measurements with a mercury sphygmomanometer, using nationally representative datasets from seven countries (China¹⁸, India¹⁹, Iran^{20, 21}, Korea²², Poland²³, Tunisia²⁴ and USA²). In addition, we compared BP percentiles in our study with values from the U.S. Fourth Report and the updated US references by sex and age at median height (according to the WHO growth charts ²⁵).

Methods

Study population

Data on BP for 52,636 non-overweight children and adolescents aged 6–19 years were obtained from seven large nationally representative cross-sectional surveys in China, India, Iran, Korea, Poland, Tunisia and USA (Table 1). The characteristics of the participants in each survey have been described in detail elsewhere.^{2, 18–24} The prevalence of overweight and obesity in the whole sample of each survey is shown in Supplemental Table 1. Overweight and obese status was assessed based on age- and sex-specific BMI percentiles as recommended by the International Obesity Task Force.²⁶ Since the inclusion of overweight or obese subjects would raise the threshold for normal BP, we excluded overweight/obese children from the reference population. After exclusion of 13,598 overweight/obese children, the final data included a total of 52,636 children and adolescents aged 6–19 years, with the sample sizes of the seven surveys ranging from 2,416 to 14,844. Three (India, Poland and Tunisia) of the seven national datasets were based on a single cross-sectional survey while data from China, Iran, Korea and USA included pooled samples from several continuous cross-sectional surveys. All participants were healthy with no genetic diseases and acute or serious chronic diseases. Written informed consent had been obtained from parents and/or children and adolescents in each national survey. Each survey had been approved by their respective Institutional Ethics Review Board.

Table 1.

Description of the Seven Nationally Representative Surveys of Blood Pressure among Non-overweight Children and Adolescents Aged 6–19 Years.

Country	Surveyed year	Description	Ethnicity	No. of participants	Age range, y	Males, %	Weight, kg	Height, cm	BMI, kg/m2	SBP, mmHg	DBP, mmHg
China 18	1997–2011	Data pooled from six cycles of the China Health and Nutrition Survey	East Asian	9334	6–19	52.2
				2113	6–9	52.2	24.3 (4.3)	126.0 (8.2)	15.2 (1.5)	91.0 (11.1)	60.0 (9.0)
				4378	10–14	50.9	37.7 (8.9)	147.4 (11.4)	17.1 (2.1)	98.3 (11.5)	64.6 (8.7)
				2843	15–19	54.3	52.6 (8.0)	163.1 (8.6)	19.7 (2.0)	107.7 (10.8)	70.6 (8.2)
India 19	2010–2012	The national growth survey in India	South Asian	5732	6–18	55.7
				1921	6–9	57.0	24.1 (5.1)	125.9 (8.7)	15.0 (1.7)	95.2 (9.9)	64.9 (8.0)
				2750	10–14	55.5	40.7 (8.9)	151.0 (10.3)	17.6 (2.3)	105.7 (9.6)	70.5 (7.7)
				1061	15–18	53.7	52.3 (8.9)	163.5 (9.0)	19.5 (2.5)	111.8 (8.6)	73.8 (6.4)
Iran 20, 21	2009–2012	Data pooled from two cycles of the national survey of CASPIAN in Iran	West Asian	14844	6–18	50.6
				3551	6–9	53.4	24.5 (4.4)	126.6 (8.1)	15.2 (1.5)	94.5 (12.0)	60.9 (10.6)
				6556	10–14	49.3	38.9 (9.5)	148.6 (11.3)	17.4 (2.4)	100.0 (12.0)	63.8 (10.5)
				4737	15–18	50.4	53.6 (9.3)	164.4 (9.5)	19.7 (2.4)	106.2 (11.9)	67.7 (10.3)
Korea 22	2005–2013	Data pooled from four cycles of the Korea National Health and Nutrition Examination Survey	East Asian	6210	10–19	48.2
				3505	10–14	48.3	43.5 (9.0)	154.1 (10.4)	18.2 (2.1)	102.3 (9.4)	63.0 (8.8)
				2705	15–19	48.1	56.9 (8.5)	167.0 (8.2)	20.4 (2.2)	105.8 (10.3)	67.8 (8.4)
Poland 23	2002–2005	The national representative survey of children and adolescents in Poland	Caucasian	5329	6–18	48.4
				1308	6–9	48.9	26.5 (4.5)	129.3 (7.4)	15.7 (1.5)	100.8 (8.6)	62.7 (7.4)
				2182	10–14	48.5	41.9 (9.7)	152.7 (11.2)	17.7 (2.2)	107.9 (9.8)	66.2 (7.5)
				1839	15–18	47.9	58.9 (8.6)	170.3 (8.8)	20.2 (2.0)	113.5 (10.7)	69.1 (7.8)
Tunisia 24	2005	The national representative survey of adolescents in Tunisia	African	2416	15–19	46.4	56.0 (7.9)	165.4 (9.0)	20.4 (2.2)	111.5 (9.9)	66.5 (8.2)
USA 2	1999–2012	Data pooled from six cycles of the U.S. National Health and Nutrition Examination Survey	Mixed *	8771	8–19	49.9
				1127	8–9	49.4	29.1 (4.4)	133.3 (6.8)	16.3 (1.4)	98.3 (8.0)	53.3 (10.5)
				3668	10–14	48.7	45.0 (9.8)	154.8 (11.0)	18.6 (2.2)	103.4 (9.1)	58.4 (10.5)
				3976	15–19	51.3	59.5 (9.1)	168.0 (9.6)	21.0 (2.1)	109.1 (9.7)	62.9 (10.1)

^*Including Hispanic, Caucasian, Black and other ethnic groups.

Continuous variables are expressed as mean (SD)

SBP: systolic blood pressure; DBP: diastolic blood pressure.

Measurements

In all seven counties, BP values were obtained with certified mercury sphygmomanometers by trained examiners following the standard protocol recommended by the AHA.²⁷ In brief, after at least 5 minutes of rest, BP was obtained on the right arm of the seated children with the elbow at the level of the right atrium, using an appropriately sized cuff. The feet of children were on a platform during BP measurement. SBP was determined by the onset of the first Korotkoff sound (i.e., appearance of tones) and DBP was determined by the fifth Korotkoff sound (i.e., total disappearance of tones). BP was measured up to three times on one occasion, at several min intervals. Children with DBP values equal to zero mmHg were excluded in all datasets before the data analysis. For five countries (China, India, Korea, Poland and USA), participants had three separate BP values were included, and the mean of the last two readings was used for our analysis; for the other two countries (Iran and Tunisia), only two readings were available and the averaged BP value was used for our analysis. Weight and height were recorded for each individual in light clothing without shoes. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters.

Statistical analysis

The datasets from each country were weighted according to population size (when analyzing the individual countries, we weighted the included studies by study size; then the data from countries were combined by weighted population size) ²⁸. We estimated SBP and DBP centile curves for non-overweight children and adolescents by age, sex and height using the Generalized Additive Model for Location Scale and Shape (GAMLSS) model for the Box-Cox power exponential (BCPE) or Box-Cox t (BCT) distribution with cubic spline smoothing.²⁹ Both distributions have four parameters including μ, σ, ν, and τ which represent location (median), scale (approximate coefficient of variation), skewness (power transformation to symmetry) and kurtosis (degrees of freedom or power exponential parameter), respectively. Data analyses were performed using the GAMLSS 4.3-1 library running under R 3.1.2.³⁰ Goodness of fit of the models was assessed by the Bayesian Information Criterion and by Q-Q plots³¹. The final best models were based on the BCT distribution for SBP in males and females and on the BCPE distribution for DBP in males and females, in consideration of the additive effects of the functions of age and height as well as their multiplicative effects on SBP and DBP under two distributions. With regard to SBP, log(μ) for males and females were modeled as a smooth function of age (df=0.8) and a linear function of height for males and as a linear function of age and height for females. With regard to DBP, log(μ) for males and females were modeled as a linear function of age and height for males and as a linear function of age and a smooth function of height (df=0.2) for females. For both SBP and DBP, log(σ) was considered as a linear function of age and ν and log(τ) as a constant for both sexes. Smoothed country-specific median SBP and DBP curves for age and height were modeled simultaneously to examine the average difference in SBP/DBP between seven countries at median height derived from seven pooled data.²⁸ The reference values of SBP and DBP percentiles (50^th, 90^th, 95^th and 99^th) were computed by the age and height percentiles recommended by the WHO²⁵ (5^th, 10^th, 25^th, 50^th, 75^th, 90th and 95^th) for boys and girls, separately. The SBP and DBP percentile values (50^th, 90^th, 95^th and 99^th) in our study were compared with the percentile values reported in the U.S. Fourth Report⁶ as well as with the updated U.S. BP reference values after excluding overweight/obese children (https://sites.google.com/a/channing.harvard.edu/bernardrosner/pediatric-blood-press) ¹¹ at median height (as recommended by the WHO²⁵).

Results

A total of 52,636 non-overweight children and adolescents aged 6–19 years were included in the final data analysis. As shown in Table 1, children aged 6–9 years in China, India, Iran and Poland had similar mean values of weight, height and BMI; children aged 10–14 years in USA and Korea had higher values than in China, India, Iran and Poland; adolescents aged 15–19 years had similar BMI values in all seven countries but adolescents in USA, Poland and Korea had higher height than in the other four countries. However, because each country collected data in different years, comparisons should be made with caution.

Supplemental Figure 1 shows the median SBP and DBP at median height (derived from mixed data of the seven countries by GAMLSS) by age and sex for non-overweight children and adolescents aged 6–19 years in the seven nationally representative datasets. The median BP curves show similar shapes, although some differences exist, for both sexes, between the studies especially for median DBP curves. Overall, for median SBP, Poland had the highest values for both sexes at a given age, while Korea, Iran and China had the lowest values. For median DBP, India had the highest values for both sexes, while USA had the lowest values. Finally, the international BP percentiles were obtained after weighting the fitted country-specific distributions according to population size since the trends of BP curves by age were similar in all the datasets.

Tables 2 and 3 show the smoothed BP percentiles (50^th, 90^th, 95^th, 99^th) for non-overweight children and adolescents by age and height. Height in centimeters for each age represents the 5^th, 10^th, 25^th, 50^th, 75^th, 90^th, 95^th percentiles as recommended by the WHO.²⁵ SBP and DBP values were similar in males and females until 13 years of age, while males had higher BP values than females at age 14 years and older. In addition, BP values increased with height percentiles (from 5^th percentile to 95^th percentile) at a given age for both sexes.

Table 2.

Percentiles of Blood Pressure Levels for Non-overweight Males by Age and Height.

Age, y	Height, cm	SBP, mmHg				DBP, mmHg
		50th	90th	95th	99th	50th	90th	95th	99th
6	111	90	104	108	115	59	71	74	80
	112	90	104	108	116	59	71	74	81
	115	91	105	109	117	59	71	75	81
	119	92	106	110	118	60	72	75	82
	122	93	107	112	120	60	72	76	82
	125	94	108	113	121	61	73	76	83
	127	95	109	113	122	61	73	76	83
7	116	91	105	109	117	60	71	75	81
	118	92	106	110	117	60	72	75	81
	121	93	107	111	118	60	72	75	82
	125	94	108	112	120	61	73	76	82
	128	95	109	113	121	61	73	76	83
	132	96	110	114	122	61	73	77	83
	134	97	111	115	123	62	74	77	84
8	120	93	106	110	118	60	72	75	82
	123	94	107	111	119	60	72	76	82
	126	95	108	112	120	61	73	76	83
	130	96	109	113	121	61	73	77	83
	134	97	111	115	123	62	74	77	84
	137	98	112	116	124	62	74	78	84
	140	99	113	117	125	62	75	78	85
9	125	94	108	111	119	61	73	76	83
	127	95	108	112	120	61	73	76	83
	131	96	110	114	121	62	74	77	83
	135	97	111	115	123	62	74	78	84
	139	99	113	117	124	62	75	78	85
	143	100	114	118	126	63	75	79	85
	145	101	115	119	127	63	75	79	86
10	130	96	109	113	120	62	73	77	83
	132	97	110	114	121	62	74	77	84
	136	98	111	115	123	62	74	78	84
	140	99	113	117	124	63	75	78	85
	145	101	114	118	126	63	75	79	86
	149	102	116	120	128	64	76	80	86
	151	103	117	121	129	64	76	80	87
11	135	98	111	114	122	62	74	78	84
	137	98	111	115	123	63	75	78	85
	141	100	113	117	124	63	75	79	85
	146	101	115	118	126	64	76	79	86
	151	103	116	120	128	64	76	80	87
	155	104	118	122	130	65	77	81	87
	157	105	119	123	131	65	77	81	88
12	141	100	113	116	124	63	75	79	85
	143	100	113	117	125	63	76	79	86
	148	102	115	119	127	64	76	80	86
	152	103	117	121	128	65	77	80	87
	157	105	119	123	130	65	77	81	88
	162	106	120	124	132	66	78	82	88
	164	107	121	125	133	66	78	82	89
13	147	102	115	119	126	64	76	80	86
	150	103	116	120	127	65	77	80	87
	155	104	118	122	129	65	77	81	87
	160	106	120	124	131	66	78	82	88
	165	108	122	126	133	66	79	82	89
	169	109	123	127	135	67	79	83	90
	172	110	124	129	137	67	80	83	90
14	154	104	117	121	129	65	77	81	87
	156	105	119	122	130	66	78	81	88
	161	107	120	124	132	66	78	82	89
	166	109	122	126	134	67	79	83	89
	172	110	124	128	136	67	80	83	90
	176	112	126	130	138	68	81	84	91
	179	113	127	131	140	68	81	85	91
15	158	106	120	123	131	66	78	82	88
	161	107	121	124	132	66	79	82	89
	166	109	122	126	134	67	79	83	89
	171	111	124	128	136	68	80	84	90
	176	112	126	131	138	68	81	84	91
	181	114	128	132	141	69	81	85	92
	184	115	129	134	142	69	82	86	92
16	162	108	121	125	132	67	79	82	89
	164	109	122	126	134	67	79	83	89
	169	110	124	128	135	68	80	83	90
	174	112	126	130	138	68	81	84	91
	179	114	128	132	140	69	81	85	92
	184	116	130	134	142	70	82	86	93
	187	117	131	135	143	70	83	86	93
17	163	109	122	126	133	67	79	83	89
	166	110	123	127	134	68	80	83	90
	171	111	125	129	136	68	81	84	91
	176	113	127	131	139	69	81	85	91
	181	115	129	133	141	70	82	86	92
	186	117	131	135	143	70	83	86	93
	188	118	132	136	144	71	83	87	93

SBP: systolic blood pressure; DBP: diastolic blood pressure

Height in centimeters for each age represents the 5th, 10th, 25th, 50th, 75th, 90th, 95th percentiles for males.

BP percentiles apply exactly for the midpoint of each age group (e.g., 6 years 6 months old) and can be applied to all children and adolescents of that age.

For research purposes, one can run the SAS codes in the Supplemental File to get prevalence of pre high BP, stages 1 and 2 high BP.

Table 3.

Percentiles of Blood Pressure Levels for Non-overweight Females by Age and Height

Age, y	Height, cm	SBP, mmHg				DBP, mmHg
		50th	90th	95th	99th	50th	90th	95th	99th
6	109	91	104	108	115	59	71	75	82
	111	91	105	108	116	59	72	75	83
	114	92	105	109	117	60	72	76	83
	118	93	106	110	118	60	72	76	83
	122	93	107	111	119	60	73	77	84
	125	94	108	112	119	60	73	77	84
	127	95	108	112	120	61	73	77	84
7	114	92	106	109	117	60	72	76	83
	116	93	106	110	117	60	72	76	83
	120	93	107	111	118	60	73	76	84
	124	94	108	112	119	61	73	77	84
	128	95	109	113	120	61	73	77	84
	131	96	110	114	121	61	74	78	85
	133	96	110	114	122	61	74	78	85
8	120	94	107	111	118	60	73	76	83
	122	94	108	111	119	61	73	77	84
	126	95	108	112	120	61	73	77	84
	130	96	109	113	121	61	74	77	85
	134	97	110	114	122	62	74	78	85
	137	98	111	115	123	62	75	78	86
	139	98	112	116	124	62	75	79	86
9	125	95	109	113	120	61	73	77	84
	128	96	109	113	121	61	74	77	84
	131	97	110	114	122	62	74	78	85
	136	98	111	115	123	62	74	78	85
	140	99	112	116	124	62	75	79	86
	144	99	113	117	125	63	75	79	86
	146	100	114	118	126	63	76	79	87
10	131	97	110	114	122	62	74	78	85
	133	98	111	115	122	62	74	78	85
	137	98	112	116	123	63	75	78	86
	142	99	113	117	125	63	75	79	86
	146	100	114	118	126	63	76	80	87
	150	101	115	119	127	64	76	80	87
	153	102	116	120	128	64	77	80	88
11	137	99	112	116	124	63	75	78	85
	140	99	113	117	124	63	75	79	86
	144	100	114	118	125	63	76	79	86
	148	101	115	119	127	64	76	80	87
	153	102	116	120	128	64	77	80	88
	157	103	117	121	129	65	77	81	88
	159	104	118	122	130	65	78	81	89
12	143	100	114	118	125	64	76	79	86
	145	101	115	118	126	64	76	79	86
	149	102	116	120	127	64	76	80	87
	154	103	117	121	128	65	77	81	88
	159	104	118	122	130	65	78	81	88
	163	105	119	123	131	66	78	82	89
	165	106	120	124	132	66	78	82	89
13	147	102	115	119	127	64	76	80	87
	149	102	116	120	127	65	76	80	87
	154	103	117	121	129	65	77	81	88
	158	105	118	122	130	65	78	81	88
	163	106	120	124	131	66	78	82	89
	167	107	121	125	133	66	79	82	90
	170	107	121	125	133	67	79	83	90
14	150	103	116	120	128	65	77	80	87
	152	104	117	121	128	65	77	80	87
	156	105	118	122	130	66	77	81	88
	161	106	119	123	131	66	78	82	89
	166	107	121	125	132	67	79	82	89
	170	108	122	126	134	67	79	83	90
	172	108	122	127	134	67	80	83	90
15	151	104	117	121	128	65	77	80	87
	153	104	118	122	129	66	77	81	87
	158	105	119	123	130	66	78	81	88
	162	106	120	124	132	67	78	82	89
	167	108	121	125	133	67	79	83	89
	171	109	122	126	134	68	80	83	90
	173	109	123	127	135	68	80	83	90
16	152	104	118	121	129	66	77	81	87
	154	105	118	122	129	66	77	81	87
	158	106	119	123	131	66	78	81	88
	163	107	121	124	132	67	79	82	89
	167	108	122	126	133	67	79	83	89
	171	109	123	127	135	68	80	83	90
	174	110	124	128	135	68	80	84	90
17	152	105	118	122	129	66	77	81	87
	154	106	119	122	130	66	78	81	87
	159	107	120	124	131	67	78	81	88
	163	108	121	125	132	67	79	82	89
	168	109	122	126	134	68	79	83	89
	172	110	123	127	135	68	80	83	90
	174	110	124	128	136	69	80	84	90

SBP: systolic blood pressure; DBP: diastolic blood pressure

Height in centimeters for each age represents the 5th, 10th, 25th, 50th, 75th, 90th, 95th percentiles for females.

BP percentiles apply exactly for the midpoint of each age group (e.g., 6 years 6 months old) and can be applied to all children and adolescents of that age.

For research purposes, one can run the SAS codes in the Supplemental File to get prevalence of pre high BP, stages 1 and 2 high BP.

We then compared our BP percentiles with the U.S. percentiles from the Fourth Report by sex and age at median height (according to the WHO growth charts ²⁵) (Figure 1).The 90^th and 95^th percentiles of SBP were lower in our data than those in the U.S. while the 90^th and 95^th percentiles of DBP were similar. For example, differences for the 95^th percentile of SBP (our BP values minus the U.S. values) by age ranged from −5 to −2 mmHg in males and −4 to −1 mmHg in females. In addition, we compared the difference in the 90^th and 95^th BP percentile between our references and the updated U.S. references (a re-analysis of the US Fourth Report data after excluding overweight/obese children) at median height (WHO) (Figure 2). The 90^th and 95^th percentile values of SBP were similar until age of 13 years and they were lower in our international values than in the updated U.S. references at age 14 years and older (by age ranging from −3 to −1 mmHg for both percentiles). For the 90^th and 95^th percentiles of DBP, our international values were somewhat higher than values in the updated U.S. references (by age ranging from 1 to 2 mmHg for both percentiles).

Figure 1.

Comparison of selected percentiles (50^th, 90^th, 95^th) of systolic blood pressure (SBP) and diastolic blood pressure (DBP) between the international BP references for non-overweight children and adolescents and the U.S. Fourth Report BP references at median height. A, Males; B, Females. Pn indicates nth percentile.

Figure 2.

Comparison of selected percentiles (50^th, 90^th, 95^th) of systolic blood pressure (SBP) and diastolic blood pressure (DBP) between the international BP references for non-overweight children and adolescents and the updated U.S. BP references for non-overweight participants at median height. A, Males; B, Females. Pn indicates nth percentile.

Discussion

To our knowledge, this is the first study to present BP percentiles by sex, age, and height using mixed data consisting of ≥ 50,000 youths with non-overweight from seven countries. Compared to the percentiles of the U.S. Fourth Report at median height, the 90^th and 95^th percentiles of our international references were lower for SBP but similar for DBP. Consistent with cut-offs for elevated BP categories in children and adolescents in other guidelines,⁶ we suggest that BP percentile values from our study could be used to define pre high BP (SBP/DBP ≥ 90^th percentile and <95^th percentile or SBP/DBP ≥ 120/80 mmHg), and high BP (Stage 1: SBP/DBP ≥ 95^th percentile and <99^th percentile plus 5 mmHg; Stage 2: SBP/DBP ≥ 99^th percentile plus 5 mmHg). For research purposes, the SAS program in the Supplemental File can be run to define the elevated BP.

The 90^th and 95^th percentiles of SBP in our study were 1 to 5 mmHg lower than those in the U.S Fourth Report. This may be due, in part, to the fact that we excluded overweight/obese children while the U.S. Fourth Report did not. In addition, different statistical models were used for these two studies, which may partly explain the higher percentile distributions in the U.S. Fourth Report than in our study. The GAMLSS method that we used does not require a normal distribution of BP values or constant variance at all ages, which may provide a better fit,¹² while the statistic model used in the Fourth Report might be less flexible. In addition, we compared the difference in the 90^th and 95^th BP percentile at median height (WHO) between our international references and the updated U.S. references (which excluded overweight/obese children). We found similar SBP values before the age of 13 years. The higher SBP values at the age of 14 years and older in the updated U.S. references, compared to our international values, might partly be due to the different statistical models used in our international data and the US data. Another potentially important reason might be the fact that the US data are based on the first BP reading only while our reference values are based on the mean of several readings. In addition, the weight distribution in our data was lower than the updated US reference population which might further explain some of the differences. However, our DBP values for both sexes were 1–2 mmHg higher than the revised US references, which was not surprising as in our data US children and adolescents had the lowest DBP values as compared to all other six countries by sex and age.

Currently, the recommended method for measurements of BP in children and adolescents remains the auscultation.^6–8 However, because of environmental hazards of mercury, sphygmomanometers based on mercury have been recently banned in most European countries.^{7, 10} Oscillometric devices are being increasingly used because of their convenience and minimization of several biases including observer bias and digit preference.³² However, oscillometric devices should be clinically validated by the auscultatory method before being widely used because of different types using different manufacturer’s proprietary algorithms.^{6, 7} One should note that that the electronic oscillometric monitors can also pollute environment if they are improperly disposed. Aneroid devices have several advantages, including low cost and avoidance of mercury pollution, but it should be regularly checked for accuracy. In practice, values obtained with oscillometric devices are generally interpreted using reference values based on the auscultatory method. By analogy, our established international BP reference can be used for interpretation of BP values obtained with electronic devices.^{33, 34} Our study has several strengths. First, our BP data were from nationally representative samples from seven countries comprising more than 50,000 children and adolescents aged 6–19 years. Second, we excluded overweight/obese subjects and applied advanced statistical methods to construct fitted BP percentiles. Third, all surveys applied rigorous quality control procedures including the use of certified mercury sphygmomanometers and adequately trained examiners. However, several limitations should be noted. First, only seven countries were included in our final data analysis, which may not adequately represent populations from several other parts of the world. Second, our BP reference did not include children younger than 6 years and further studies are required to fill this gap. Third, for China, Korea and USA, data were pooled from several survey cycles, and BP levels have slightly changed over time (Supplemental Table 2). Fourth, our data are based on three BP readings in 5 countries but on two readings in 2 other countries. However, sensitivity analysis did not show differences after exclusion of subjects who had two BP values. Fifth, the GAMLSS model cannot account for sampling weights recommended for both NHANES (USA) and KNHANES (Korea), and this issue should be acknowledged as an unsolved limitation. Sixth, the present statistical method using the distribution of BP percentiles to establish BP norms for children is not optimal. It would be preferable to base BP norms on the relation between BP in childhood and health outcomes in childhood (i.e. target organ damage) or later in adulthood (CVD events and mortality etc).^35–37

In summary, this study provides international BP references by sex, age and height in children and adolescents aged 6–17 years. These BP references will be useful for international comparisons of the prevalence of elevated BP in children and adolescents. In addition, these BP references will also be useful to identify children and adolescents with elevated BP in countries where definition of childhood elevated BP based on national data is not readily available.

Clinical Perspectives.

Elevated blood pressure (BP) in children and adolescents is both a clinical and a public health concern worldwide. Screening and treatment of hypertension require that elevated BP is well defined in children and adolescents. In absence of solid cohort data to define risk associated with BP in children, the distribution of BP are usually used to define BP categories in youths. US BP references from the Fourth Report are widely used worldwide, but there has been no attempt to define BP references based on BP data from multiple countries. We therefore developed international BP percentiles by sex-, age-, and height among 52,636 non-overweight children and adolescents aged 6–19 years using nationally representative datasets from seven countries (China, India Iran, Korea, Poland, Tunisia and USA). These new international BP reference based on the distribution of BP from multiple countries may have advantages for the identification of children and adolescent who have elevated BP in different populations and for the comparison of the prevalence of hypertension between countries.