Skip to main content
Download PDF

Prevalence of childhood hypertension and associated factors in Zhejiang Province: a cross-sectional analysis based on random forest model and logistic regression

BMC Public Health volume 24, Article number: 2101 (2024) Cite this article

Abstract

With childhood hypertension emerging as a global public health concern, understanding its associated factors is crucial. This study investigated the prevalence and associated factors of hypertension among Chinese children. This cross-sectional investigation was conducted in Pinghu, Zhejiang province, involving 2,373 children aged 8–14 years from 12 schools. Anthropometric measurements were taken by trained staff. Blood pressure (BP) was measured in three separate occasions, with an interval of at least two weeks. Childhood hypertension was defined as systolic blood pressure (SBP) and/or diastolic blood pressure (DBP) ≥ age-, sex-, and height-specific 95th percentile, across all three visits. A self-administered questionnaire was utilized to collect demographic, socioeconomic, health behavioral, and parental information at the first visit of BP measurement. Random forest (RF) and multivariable logistic regression model were used collectively to identify associated factors. Additionally, population attributable fractions (PAFs) were calculated. The prevalence of childhood hypertension was 5.0% (95% confidence interval [CI]: 4.1–5.9%). Children with body mass index (BMI) ≥ 85th percentile were grouped into abnormal weight, and those with waist circumference (WC) > 90th percentile were sorted into central obesity. Normal weight with central obesity (NWCO, adjusted odds ratio [aOR] = 5.04, 95% CI: 1.96–12.98), abnormal weight with no central obesity (AWNCO, aOR = 4.60, 95% CI: 2.57–8.21), and abnormal weight with central obesity (AWCO, aOR = 9.94, 95% CI: 6.06–16.32) were associated with an increased risk of childhood hypertension. Childhood hypertension was attributable to AWCO mostly (PAF: 0.64, 95% CI: 0.50–0.75), followed by AWNCO (PAF: 0.34, 95% CI: 0.19–0.51), and NWCO (PAF: 0.13, 95% CI: 0.03–0.30). Our results indicated that obesity phenotype is associated with childhood hypertension, and the role of weight management could serve as potential target for intervention.

Peer Review reports

Introduction

Hypertension, characterized by persistently high blood pressure (BP), used to be considered as a condition prevalent among adults but is now increasingly diagnosed in children [1, 2]. Unlike adulthood hypertension, several guidelines have recommended childhood hypertension should be defined as elevated BP over three separate occasions [3, 4]. Globally, the prevalence of childhood hypertension was estimated to be 4%, with an increasing rate of 75–79% between 2000 and 2015 [5]. In China, this trend is compounded by the rapid economic development and rising incidence of childhood obesity [6, 7]. Although symptoms of childhood hypertension are often mild, it can lead to organ damage [8,9,10] and later progress into adulthood toward premature cardiovascular diseases [9, 11,12,13,14,15]. Therefore, the early detection, prevention, and management of hypertension from childhood are critical to mitigate these long-term risks and to improve cardiovascular health across the lifespan.

Currently, since the diagnosis criterion of childhood hypertension presents a methodological challenge in early and accurate identification, previous studies have predominantly focused on elevated BP rather than hypertension diagnosed over three separate occasions [16,17,18]. Lacking of three separate occasions diagnosis would result in an overestimation of the prevalence of childhood hypertension. There are many factors influencing the prevalence of childhood hypertension. Previous studies have predominantly focused on biological and individual behavioral factors, such as obesity, family history of hypertension, low birth weight, and an unhealthy diet [19,20,21,22,23,24,25], while paying less attention to the macro- and meso-level factors. Social-ecological perspective is an essential framework to better understand and effectively improve children’s health issues [26, 27], which conceptualizes the interactions and influences on individual health outcomes at the individual, interpersonal, family, community and policy levels. Therefore, the social-ecological perspective serves as a framework for exploring the multidimensional associated factors of childhood hypertension.

Among previous cross-sectional studies exploring the associations between factors with childhood hypertension, most of them relied on multivariable logistic regression models to quantify the associations and use odds ratios (ORs) for clear interpretation, which often analyzes factors in isolation, without assessing their relative importance [21, 24, 28]. Random Forest (RF) is a machine learning algorithm characterized by its capacity to assess the importance of numerous variables without overfitting, which offers a solution to this issue [29]. Yet, RF’s complexity makes its findings less straightforward than ORs provided by logistic regression. Combining RF with logistic regression could enhance our understanding by providing a detailed assessment and comparison of the effects of various factors on childhood hypertension [30, 31]. Furthermore, previous studies paid less attention on the attributable fractions of each associated factors. Population attributable fractions (PAFs), is a counterfactual estimate method, which has been widely employed to reveal the population-level impact of preventive interventions on diseases [32,33,34,35].

To address these research gaps, our study aimed to: (1) offer an overview of childhood hypertension prevalence, based on three separate visits, in Zhejiang province, China; (2) assess and compare the effects of various factors associated with childhood hypertension using an integrated approach that combines RF and logistic regression; and (3) calculate the PAFs for significant associated factors and evaluate the theoretical prevalence of childhood hypertension if these specific factors were addressed.

Methods

Study design and participants

This study is based on the project titled “Effect of Intensive Lifestyle Intervention on Overweight and Obesity in Children and Adolescents”, a school-based project since 2022. Our study used data from the baseline information of this project. Stratified cluster sampling was employed to enroll students aged 8–14 years in Pinghu, Zhejiang Province. Subjects were cluster sampled from grade three to four in primary schools and grade seven in middle schools from randomly chosen 12 schools (six primary schools and six middle schools). Sample size formula for cross-sectional studies was used, \(\:n={z}_{\alpha\:}^{2}*p*q/{\sigma\:}^{2}\), where \(\:\alpha\:=0.05\), \(\:{z}_{\alpha\:}=1.96\), \(\:p=4\%\), \(\:q=1-p=96\%\), \(\:\sigma\:=0.2,p=0.08\). Non-response rate was estimated to be 10% and the total sample size for the study was at least 2507 children.

Our study enrolled non-random sample who had voluntarily completed the questionnaire. Those with intellectual disabilities, severe mental disorders, and other conditions that prevented them from participating in questionnaire survey and physical examinations were excluded. A total of 1366 children in grade three to four were enrolled from six primary schools, while 1007 children in grade seven were enrolled from six middle schools. Of these, 235 were excluded for unplausible information on anthropometrics and parental information filled by caregivers other than parents. The ultimate analytical sample consisted of 2373 children aged 8–14 years (Fig. 1).

Fig. 1
figure 1

Flowchart for selecting subjects in this study

Note: the exclusion criteria were in an order as listed in the box, some subjects might meet several exclusion criteria simultaneously

Full size image

This study was approved by the Ethical Committee of Zhejiang University School of Medicine, Zhejiang, China (2021–029). Written informed consents were provided by the guardians of the included children.

Data collection

At the first visit of BP measurement, a self-administered questionnaire was designed to collect demographic and socioeconomic data, as well as health behaviors information for all included children from primary and middle schools (Questionnaire S1). Before administering the questionnaire, researchers introduced its content and allowed sufficient time for children to fully understand it. The questionnaires were completed under the supervision of pre-trained teachers, with researchers on hand to provide necessary assistance and guidance. Meanwhile, parental information (e.g., family income, education level, vocation) was provided by their caregivers through questionnaires (Questionnaire S2).

All anthropometric measurements were performed by trained staff from the local Centers for Disease Control and Prevention, following the national standard protocol. While standing upright, height and body weight of the children were measured under standard protocols, and in increments of 0.1 cm and 0.1 kg, respectively. Waist circumference (WC) was measured in the natural standing position at the umbilical level to the nearest 0.1 cm using a non-flexible tape. Body mass index (BMI) was calculated by dividing body weight (kg) by the square of the height (m). After at least 5 min of seated rest, BP was measured three consecutive times with 1 min intervals in a seated position, with the OMRON HBP-1300 automated device (OMRON Corporation, Kyoto, Japan) and medium cuff size, and the averages of three readings were taken. According to the Chinese BP reference standards for children and adolescents based on the levels of systolic blood pressure (SBP) and diastolic blood pressure (DBP) in 2017 [36], children with SBP and/or DBP ≥ age-, sex-, and height-specific 95th percentile underwent a second round of BP measurements at least two weeks later, and then a third round of measurements if BP was more than 95th percentile at the second screening.

Definitions

Based on previous evidence for childhood hypertension, we focused on associated factors that are routinely available or easily ascertained through anthropometric measurements or questionnaires. Finally, a total of 30 independent variables in six categories were included as potential associated factors for childhood hypertension, and were put into the social-ecological framework for better understanding of their levels (Fig S1).

  1. (1)

    Demographics (residence, sex, and age groups): residence was divided into urban and rural; sex was categorized into two groups: boy and girl; age groups included primary school and middle school.

  2. (2)

    Socioeconomics (monthly household income/expense, maternal occupation, and maternal education): monthly household income/expense was categorized into three groups: < CNY 10,000, CNY 10,000–20,000, and > CNY 20,000; maternal occupation was divided into four groups: administrator or manager or professional and technical, sales or service workers or self-employed, agriculture or worker, and unemployed or other; maternal education was categorized into three groups: middle school and below, high school (including secondary), and tertiary and undergraduate and above.

  3. (3)

    Family (family structure, family history of hypertension, maternal BMI status, mother’s smoke/drink use, mother’s moderate-to-vigorous physical activity [MVPA] adherence, and parental migration): family structure was categorized into three groups: nuclear families, large families, and single or reconstituted families; family history of hypertension could be coded as no and yes; maternal BMI status was assessed through self-reported status, and could be sorted into: normal, overweight, and obesity; mother’s smoke/drink use could be coded as no and yes; mother’s MVPA adherence could be categorized into two groups: insufficient MVPA (the total time of MVPA < 150 min per/week) and sufficient MVPA (the total time of MVPA ≥ 150 min per/week); parental migration could be coded as no and yes.

  4. (4)

    Birth (maternal reproductive age, natural delivery, breastfeeding, term birth, birthweight category, and one-child): maternal reproductive age was grouped into < 35 years and ≥ 35 years; natural delivery, breastfeeding, term birth, and one-child could be coded as no and yes; birthweight was recorded as continuous values, and could be categorized into normal (birthweight ≥ 2.5 kg and ≤ 4.0 kg) and abnormal (birthweight < 2.5 kg or > 4.0 kg).

  5. (5)

    Anthropometrics (obesity phenotype): obesity phenotypes were defined as the combinations of general and central obesity statuses. Children were grouped into normal weight (BMI < 85th percentile for age and sex), overweight (85th percentile for age and sex ≤ BMI < 95th percentile for age and sex), and general obesity (BMI ≥ 95th percentile for age and sex), according to the Chinese National Standard (WS/T 586–2018) [37]”. Overweight and general obesity were merged into abnormal weight. Besides, children with WC > 90th percentile for age and sex were sorted into central obesity [38]. Then all children were further grouped into four mutually exclusive groups: normal weight with no central obesity (NWNCO), normal weight with central obesity (NWCO), abnormal weight with no central obesity (AWNCO), and abnormal weight with central obesity (AWCO).

  6. (6)

    Lifestyle (children’s sleep quality, children’s sleep/screen time/MVPA adherence, children’s commuting modes, children’s habit of breakfast/midnight snack, and home snack): children’s sleep quality were grouped into bad and well; if the sleep duration of children aged 5–13 years was in the range of 9–11 h and for children aged 14–17 years was in the range of 8–10 h, then they could be grouped as having adequate sleep duration, otherwise could be categorized into having inadequate sleep duration [39]; children’s screen time adherence could be categorized into two groups: unlimited screen time (> 2 h per day spent in total watching TV, playing video games, using the computer, or using portable electronic devices) and limited screen time (≤ 2 h per day of screen time) [39]; children’s MVPA adherence could be categorized into two groups: insufficient MVPA (daily MVPA for < 60 min) and sufficient MVPA (daily MVPA for ≥ 60 min) [39]; children’s commuting modes could be categorized into two groups: walk or bike, and car or others; children’s habit of breakfast was assessed by asking respondents “do you eat breakfast every day”, and responses were categorized into no and yes; children’s habit of midnight snack was assessed by asking respondents “do you eat midnight snack every day”, and responses were categorized into no and yes; home snack was evaluated by question “are there any snacks at home”, and responses were categorized: no and yes.

Hypertension status was considered as an outcome variable and was defined as SBP and/or DBP ≥ age-, sex-, and height-specific 95th percentile from visit one to visit three, according to the Chinese BP reference standards for children and adolescents [36].

Statistical analysis

Continuous variables were presented as median with interquartile range (IQR) if the parameter exhibited skewness in normality tests and as mean ± SD when the parameter was normally distributed. Categorical variables were reported as number (n) with percent (%). Differences in basic characteristics by sex were evaluated using T-tests for normally-distributed continuous variables and Mann-Whitney test for skewed-distributed continuous variables, respectively. In terms of categorical variables, Chi-square (χ²) tests were conducted.

RF model and multivariable logistic regression model were used in combination to rank the importance of various associated factors and then provide intuitive explanations through ORs and 95% confidence intervals (CIs) [30, 31]. RF is a non-parametric machine learning model that enables establishing a series of decision trees and outputs the classes of a single tree, which could avoid overfitting and rank the importance of variables so that the key variables can be focused [31]. The original data were divided into training set and test set, respectively. Then, RF generated bootstraps through randomly resampling from the training set, and ultimately aggregating the results [31]. Out-of-bag (OOB) refers to a dataset that is not included in resampling during the construction of a decision tree, which serves as a validation set to assess the performance of the decision tree and estimate RF’s goodness of fit [40]. Given the averages of the predictions made by all decision trees, the model predicts the output as follows:

$$\:-\stackrel{-}{f}=\frac{1}{m}{\sum\:}_{i=1}^{m}{f}_{i}\left({x}^{{\prime\:}}\right)$$

where \(\:m\) is the total number of decision trees, \(\:{f}_{i}\left({x}^{{\prime\:}}\right)\) is the prediction of i-th decision tree given input \(\:{x}^{{\prime\:}}\).

Mean decrease accuracy (MDA) was used to evaluate the importance of associated factors [41]. A higher MDA value indicates that the corresponding associated factor is more significant [41]. Based on the results of MDA values, the 10-fold cross-validation was used to obtain the number of variables that displayed the best performance. Therefore, in our multivariable logistic regression, the top seven variables were selected as independent variables, and childhood hypertension was regarded as the outcome variable. Additionally, sex, age groups, and residence were incorporated into the multivariable logistic regression and were used for stratified analyses. Model 1 was the crude model without any adjustment. Model 2 was further adjusted for sex, age groups, residence, children’s habit of breakfast, children’s sleep adherence, children’s screen time adherence, children’s midnight snack habit, obesity phenotypes, family history of hypertension, one-child, home snacks, type of delivery, and children’s MVPA adherence. Subsequently, the multicollinearity of the factors in the adjusted model was assessed by using generalized Variance Inflation Factor (gVIF). To evaluate the fitness of model and detect any potential issues with the residuals, a quantile residuals plot was generated. The PAF (%) and its corresponding 95% confidence intervals (CI) were calculated employing adjusted odds ratios [aOR] derived from those mentioned above in the multivariable logistic regression model, and the detailed formula was shown in Supplementary Eq. S1. After adjusting for age, sex, residence, and other factors mentioned in model 2, the PAFs and their 95% CI were shown by age, sex, and residence groups. The theoretical prevalence of childhood hypertension under the hypothesis of elimination of certain risk factor was calculated utilizing the formula of the actual prevalence of hypertension multiplied by (1-PAF).

Data were analyzed in Stata version 17.0 (StataCorp) and R version 4.3.0 (R Foundation for Statistical Computing). A two-sided P of less than 0.05 indicated statistical significance.

Results

Characteristics of study participants

Table S1 displayed the characteristics of the included children by hypertension status. There were 1181 boys (49.8%), 1192 girls (50.2%), and a total of 119 (5.0%) children who ultimately met the diagnostic criteria for childhood hypertension based on BP measurements in three separate visits. The median age of the children was 9.5 (IQR: 9.0, 12.5) years. The proportions of urban location and primary school students were 76.3% and 57.6%, respectively.

Significant differences were observed between hypertension and normotension in terms of anthropometric indicators, family disease history, and type of delivery. Children with hypertension had higher height (Median: 150.4 [IQR: 135.7, 159.2] cm vs. Median: 143.1 [IQR: 134.5, 156.0] cm, P = 0.024), weight (Median: 50.0 [IQR: 37.8, 65.7] kg vs. Median: 38.5 [IQR: 30.0, 47.8] kg, P < 0.001), SBP (Median: 126.2 [IQR: 120.5, 133.0] mmHg vs. Median: 104.8 [IQR: 98.7, 111.3] mmHg, P < 0.001), and DBP (Median: 76.3 [IQR: 71.0, 79.0] mmHg vs. Median: 65.7 [IQR: 61.3, 70.0] mmHg, P < 0.001) than those with normotension. The proportions of family history of hypertension, mother being overweight, mother being obese, and non-natural labor were higher in children with hypertension than their normotensive counterparts (8.4% vs. 4.1%, 26.1% vs.17.3%, 9.2% vs.3.5%, and 66.4% vs. 53.0%, respectively). Meanwhile, compared with those with normotension, children with hypertension had higher rates of NWCO, AWNCO, and AWCO (5.0% vs. 3.5%, 21.0% vs.14.2%, and 53.8% vs. 17.7%, respectively).

Associated factors of childhood hypertension

The importance of associated factors assessed by RF was displayed in Figure S2. The top ten associated factors based on the MDA from high to low are age groups, habit of breakfast, children’s sleep adherence, children’s screen time adherence, habit of midnight snack, residence, obesity phenotype, family history of hypertension, one-child, and parental migration. According to the results of 10-fold cross-validation, we included the top seven variables, together with sex, age groups, and residence into the logistic regression to derive ORs and 95% CIs. As shown in Table 1, NWCO (OR = 4.61, 95% CI: 1.83–11.59), AWNCO (OR = 4.74, 95% CI:2.67–8.41), AWCO (OR = 9.73, 95% CI: 6.01–15.76), having family history of hypertension (OR = 2.16, 95% CI: 1.09–4.26), and natural labour (OR = 1.75, 95% CI: 1.19–2.58) were associated with a higher odd of childhood hypertension based on univariate model. Meanwhile, after adjusting for age, sex, residence, and all other factors, NWCO (aOR = 5.04, 95% CI: 1.96–12.98), AWNCO (aOR = 4.60, 95% CI:2.57–8.21), and AWCO (aOR = 9.94, 95% CI: 6.06–16.32) were associated with higher odds of childhood hypertension. Table S2 showed the results of the multicollinearity test and the mean adjusted gVIF was 1.05, indicating low collinearity. The quantile-quantile (Q-Q) plot indicated well fitness of model, with no significant deviations (KS test p = 0.878, dispersion test p = 0.976), while significant outliers were detected (outlier test p = 0.002) (Figure S3). The residual vs. predicted values plot showed a generally uniform distribution of residuals across the range of predicted values (Figure S3).

Table 1 Associations between associated factors and childhood hypertension
Full size table

Population attributable fractions of associated factors and the theoretical prevalence of childhood hypertension

At the population level, AWCO had the largest PAF (PAF: 0.64, 95% CI: 0.50–0.75), followed by AWNCO (PAF: 0.34, 95% CI: 0.19–0.51), and NWCO (PAF: 0.13, 95% CI: 0.03–0.30) (Fig. 2). Among various age, sex, and residence groups, the magnitudes of PAFs were not dramatically changed. AWCO demonstrated the largest PAF in nearly all subgroups.

Fig. 2
figure 2

Adjusted population attributable fractions of associated factors and predicted prevalence of childhood hypertension if the factor was removed

Note: NWNCO, normal weight with no central obesity. NWCO, normal weight with central obesity. AWNCO, abnormal weight with no central obesity. AWCO, abnormal weight with central obesity

Full size image

Moreover, we estimated the theoretical prevalence of childhood hypertension on the condition that each associated factor was eliminated (Fig. 2). Removing AWCO (0.02, 95% CI: 0.01–0.03), AWNCO (0.03, 95% CI: 0.02–0.04), and NWCO (0.04, 95% CI: 0.04–0.05) were correlated with a large decrease in the theoretical prevalence of childhood hypertension. Among different age, sex, and residence groups, the magnitudes of reduction in the theoretical prevalence were not dramatically varied.

Discussion

This comprehensive study seeks not only to identify significant associated factors of childhood hypertension but also to prioritize interventions based on their potential population impacts among students aged 8–14 years. To the best of our knowledge, this study is one of the first, based on a framework namely social-ecological theory, to explore the effects of multidimensional factors associated with childhood hypertension based on three separate visits using an integrated approach that combines RF and logistic regression. Our findings indicated a childhood hypertension prevalence of 5.0%, ascertained through consistent elevated BP across three separate visits. In the crude model, obesity phenotype, having family history of hypertension, and non-natural delivery were associated with increased odds of childhood hypertension, while in the adjusted model, obesity phenotype was associated with it. At the population level, AWCO, AWNCO, and NWCO accounted for about 64%, 34%, and 13% of childhood hypertension cases, respectively.

The revealed childhood hypertension prevalence was markedly lower than prior estimates from single-visit BP measurements in China, which have reported rates ranging from 2.0 to 26.0% [16,17,18, 42,43,44,45]. According to the fourth report from the National High Blood Pressure Education Program Working Group in the United States, the assessment of childhood hypertension should be based on three separate visits [46]. This multi-visit approach is crucial in obtaining an accurate estimate of hypertension prevalence, which reduces the variability of diagnosis from a single-visit measurement. Building upon a meta-analysis conducted to determine a summary prevalence of elevated BP over three visits, the prevalence of elevated BP decreased by 53.7% during visit two and by 77.7% during visit three, compared to visit one [47]. Our results align closely with the global (4.0%) and national (4.0%) prevalence of data derived from three BP occasions [5, 48], suggesting a reliable estimation for informed public health interventions.

Moreover, the prevalence of childhood hypertension was higher in boys than in girls in our study, which was parallel to observed outcomes in other study populations. For instance, a previous study employing data from the China Child and Adolescent Cardiovascular Health demonstrated the prevalence of childhood hypertension was 5% in boys and 3% in girls [48]. Numerous studies suggested that variations in sex and age could exert impacts on the status of childhood hypertension. While before puberty the BP levels tend to be consistent in boys and girls, and after puberty the levels in boys surpassed those in girls [6]. During puberty, children were prone to experiencing transient insulin resistance, which leads to a more rapid elevation of BP levels. Simultaneously, estrogen played a crucial role in enhancing endothelium-dependent vasodilation and regulating smooth muscle cells [49, 50], protecting girls from hypertension. Besides, unhealthy lifestyles were more prevalent among Chinese adolescent boys than girls, which might be the potential explanations for sex difference [51, 52].

Consistent with existing literature, obesity phenotype, having family history of hypertension, and non-natural delivery were associated with an increased odd of childhood hypertension [42, 53, 54]. The pathophysiological link between obesity and hypertension is complex and multifaceted. Firstly, obesity could induce changes in the nervous system and insulin resistance, which together disrupt the function of blood vessels and alter sodium balance, contributing to increased BP [55, 56]. Secondly, excessive visceral adipose tissue accumulation triggers immune responses, leading to elevated free fatty acids (FFAs) levels [57]. These FFAs not only interfere with insulin’s ability to reduce blood glucose but also promote the accumulation of angiotensinogen, a precursor to angiotensin, which is known to increase BP. Thirdly, the heightened levels of angiotensinogen contribute to both the constriction of blood vessels and the enhancement of sodium retention, resulting in elevated BP [57]. Furthermore, a family history of hypertension may predispose children to the condition due to genetic factors. A previous study involving 2638 high school adolescent students in Italy demonstrated that certain gene variants, previously linked to hypertension in adults, also affect BP and sodium excretion among adolescents [53]. Additionally, these genetic variants might interact with pro-inflammatory pathways, which are often activated in hypertensive individuals, potentially leading to early cardiovascular damage [53]. In line with previous studies [54, 58], our findings indicated that the type of delivery may impact offspring health outcomes. A Dutch birth cohort showed that children born by cesarean section were 1.52 times more likely to be overweight than those born via natural delivery [54], and another birth cohort study in southern Brazil displayed that those born by cesarean section have higher BP values [58].

Though the decreased urban-rural disparity in BP between 1985 and 2010 [24], and the escalating burden of childhood obesity among urban and rural between 1985 and 2019 [59] were observed, the urban-rural differences in childhood hypertension attributable to obesity phenotype was still unclear. Despite existing studies had figured out the distribution of childhood hypertension in different sex and age groups [5], as well as explored its association with obesity phenotype [42], there were still few studies to investigate the attributable fraction of obesity phenotype to hypertension in children across different sex and age groups. Besides the identification of associated factors of childhood hypertension, we quantified the population impact using the PAFs of each factor to pinpoint potential intervention targets. In accordance with previous evidence, our study indicated that alleviating the problem of general and central obesity (NWCO, AWNCO, and AWCO) is beneficial for managing children’s hypertension [42]. Moreover, our study suggested the ranked associated factors for childhood hypertension varied by sex, age, and residence, which warranted population-specific strategies for hypertension prevention and control.

Strengths and limitations of the study

Our study had several strengths. First, the assessment of childhood hypertension was based on BP measurements across three separate occasions, minimizing the influence of white-coat hypertension (an elevated BP in the medical setting, but a normal BP out of the medical setting) and masked hypertension (a normal BP in the medical setting, but an elevated BP out of the medical setting). Besides, the school-based study ensured a high rate of participation, thereby reducing selection bias. Notably, we did not immediately inform children detected with elevated BP. Instead, we provided lifestyle advice only after completing the questionnaire and BP measurements, thereby minimizing the bias introduced by changes in their lifestyle. Additionally, the application of a social-ecological framework and an integrated approach that combines RF and logistic regression could enhance our understanding by ranking the importance of multidimensional associated factors and also provide clear explanations for these associations. Furthermore, the calculation of PAFs was instrumental in identifying and prioritizing targeted interventions, thereby contributing to more effective management of childhood hypertension.

Nevertheless, several limitations warrant mention when interpreting the present findings. First, the sampling was confined to Pinghu, Jiaxing City within Zhejiang Province, which potentially limited the generalizability and representativeness of our findings to the entire province or beyond. Second, the cross-sectional nature of the study also restricted our ability to assess causality between the associated factors and childhood hypertension. Third, involving children those volunteered to participate in might incur selection bias and social desirability bias. Finally, information on associated factors (e.g. family income) at the first visit was gathered by self-administered questionnaires, thus information bias and reporting bias ought to be considered and future studies should be undertaken that incorporate more objective measures.

Conclusions

In summary, among children aged 8–14 years in Pinghu, Zhejiang province, China, positive associations with hypertension were observed in those with NWCO, AWNCO, and AWCO. Our results suggested that targeted modifications to these factors could lead to a decreased prevalence of childhood hypertension. Furthermore, the risk attributed to these factors differed across age, sex, and residence, highlighting the need for tailored approaches in hypertension prevention initiatives.

Data availability

The datasets used and/or analyzed during the current study were available from the corresponding authors on reasonable request.

References

  1. Bromfield S, Muntner P. High blood pressure: the leading global burden of disease risk factor and the need for worldwide prevention programs. Curr Hypertens Rep. 2013;15(3):134–6.

    Article  PubMed  PubMed Central  Google Scholar 

  2. Forouzanfar MH, Alexander L, Anderson HR, et al. Global, regional, and national comparative risk assessment of 79 behavioural, environmental and occupational, and metabolic risks or clusters of risks in 188 countries, 1990–2013: a systematic analysis for the global burden of Disease Study 2013. Lancet. 2015;386(10010):2287–323.

    Article  PubMed  Google Scholar 

  3. Harris KC, Benoit G, Dionne J, et al. Hypertension Canada’s 2016 Canadian Hypertension Education Program Guidelines for blood pressure measurement, diagnosis, and Assessment of Risk of Pediatric Hypertension. Can J Cardiol. 2016;32(5):589–97.

    Article  PubMed  Google Scholar 

  4. Flynn JT, Kaelber DC, Baker-Smith CM et al. Subcommittee on Screening and Management of High Blood Pressure in Children. Clinical Practice Guideline for Screening and Management of High Blood Pressure in Children and Adolescents. Pediatrics. 2017;140(3):e20171904. Pediatrics 2018; 142(3).

  5. Song P, Zhang Y, Yu J, et al. Global prevalence of hypertension in children: a systematic review and Meta-analysis. JAMA Pediatr. 2019;173(12):1154–63.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Ye X, Yi Q, Shao J, et al. Trends in Prevalence of Hypertension and Hypertension Phenotypes among Chinese Children and adolescents over two decades (1991–2015). Front Cardiovasc Med. 2021;8:627741.

    Article  PubMed  PubMed Central  Google Scholar 

  7. Qiao J, Wang Y, Li X, et al. A Lancet Commission on 70 years of women’s reproductive, maternal, newborn, child, and adolescent health in China. Lancet. 2021;397(10293):2497–536.

    Article  CAS  PubMed  Google Scholar 

  8. Balagopal PB, de Ferranti SD, Cook S, et al. Nontraditional risk factors and biomarkers for cardiovascular disease: mechanistic, research, and clinical considerations for youth: a scientific statement from the American Heart Association. Circulation. 2011;123(23):2749–69.

    Article  PubMed  Google Scholar 

  9. Chen X, Wang Y. Tracking of blood pressure from childhood to adulthood: a systematic review and meta-regression analysis. Circulation. 2008;117(25):3171–80.

    Article  PubMed  PubMed Central  Google Scholar 

  10. Sun SS, Grave GD, Siervogel RM, Pickoff AA, Arslanian SS, Daniels SR. Systolic blood pressure in childhood predicts hypertension and metabolic syndrome later in life. Pediatrics. 2007;119(2):237–46.

    Article  PubMed  Google Scholar 

  11. Franks PW, Hanson RL, Knowler WC, Sievers ML, Bennett PH, Looker HC. Childhood obesity, other cardiovascular risk factors, and premature death. N Engl J Med. 2010;362(6):485–93.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  12. Lubrano R, Travasso E, Raggi C, Guido G, Masciangelo R, Elli M. Blood pressure load, proteinuria and renal function in pre-hypertensive children. Pediatr Nephrol. 2009;24(4):823–31.

    Article  PubMed  Google Scholar 

  13. Lucini D, Mela GS, Malliani A, Pagani M. Impairment in cardiac autonomic regulation preceding arterial hypertension in humans: insights from spectral analysis of beat-by-beat cardiovascular variability. Circulation. 2002;106(21):2673–9.

    Article  PubMed  Google Scholar 

  14. Sundström J, Neovius M, Tynelius P, Rasmussen F. Association of blood pressure in late adolescence with subsequent mortality: cohort study of Swedish male conscripts. BMJ. 2011;342:d643.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Urbina EM, Khoury PR, McCoy C, Daniels SR, Kimball TR, Dolan LM. Cardiac and vascular consequences of pre-hypertension in youth. J Clin Hypertens (Greenwich). 2011;13(5):332–42.

    Article  PubMed  Google Scholar 

  16. Yan W, Li X, Zhang Y, et al. Reevaluate secular trends of body size measurements and prevalence of hypertension among Chinese children and adolescents in past two decades. J Hypertens. 2016;34(12):2337–43.

    Article  CAS  PubMed  Google Scholar 

  17. Yang Y, Dong B, Wang S, et al. Prevalence of high blood pressure subtypes and its associations with BMI in Chinese children: a national cross-sectional survey. BMC Public Health. 2017;17(1):598.

    Article  PubMed  PubMed Central  Google Scholar 

  18. Zhao Y, Wang L, Xue B, Wang Y. Associations between general and central obesity and hypertension among children: the childhood obesity study in China Mega-cities. Sci Rep. 2017;7(1):16895.

    Article  PubMed  PubMed Central  Google Scholar 

  19. Hamoen M, Welten M, Nieboer D, et al. Development of a prediction model to target screening for high blood pressure in children. Prev Med. 2020;132:105997.

    Article  PubMed  Google Scholar 

  20. Liang X, Xiao L, Luo Y, Xu J. Prevalence and risk factors of childhood hypertension from birth through childhood: a retrospective cohort study. J Hum Hypertens. 2020;34(2):151–64.

    Article  PubMed  Google Scholar 

  21. Manios Y, Karatzi K, Moschonis G, et al. Lifestyle, anthropometric, socio-demographic and perinatal correlates of early adolescence hypertension: the healthy growth study. Nutr Metab Cardiovasc Dis. 2019;29(2):159–69.

    Article  CAS  PubMed  Google Scholar 

  22. Kaczmarek M, Stawińska-Witoszyńska B, Krzyżaniak A, Krzywińska-Wiewiorowska M, Siwińska A. Who is at higher risk of hypertension? Socioeconomic status differences in blood pressure among Polish adolescents: a population-based ADOPOLNOR study. Eur J Pediatr. 2015;174(11):1461–73.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Zhang Z, Dong B, Li S, et al. Exposure to ambient particulate matter air pollution, blood pressure and hypertension in children and adolescents: a national cross-sectional study in China. Environ Int. 2019;128:103–8.

    Article  CAS  PubMed  Google Scholar 

  24. Dong B, Wang Z, Ma J. Urban-rural disparity in blood pressure among Chinese children: 1985–2010. Eur J Public Health. 2016;26(4):569–75.

    Article  PubMed  Google Scholar 

  25. Menghetti E, Strisciuglio P, Spagnolo A, et al. Hypertension and obesity in Italian school children: the role of diet, lifestyle and family history. Nutr Metab Cardiovasc Dis. 2015;25(6):602–7.

    Article  CAS  PubMed  Google Scholar 

  26. Fu E, Grimm KJ, Berkel C, Smith JD. Parenting and social-ecological correlates with children’s health behaviours: a latent profile analysis. Pediatr Obes. 2020;15(10):e12721.

    Article  PubMed  Google Scholar 

  27. Hu D, Zhou S, Crowley-McHattan ZJ, Liu Z. Factors that influence participation in physical activity in school-aged children and adolescents: a systematic review from the Social Ecological Model Perspective. Int J Environ Res Public Health 2021; 18(6).

  28. Tomás-Rodríguez MI, Palazón-Bru A, Martínez-St John DR, Navarro-Cremades F, Toledo-Marhuenda JV, Gil-Guillén VF. Factors Associated with increased Pain in primary dysmenorrhea: analysis using a Multivariate ordered logistic regression model. J Pediatr Adolesc Gynecol. 2017;30(2):199–202.

    Article  PubMed  Google Scholar 

  29. Wallace ML, Mentch L, Wheeler BJ, et al. Use and misuse of random forest variable importance metrics in medicine: demonstrations through incident stroke prediction. BMC Med Res Methodol. 2023;23(1):144.

    Article  PubMed  PubMed Central  Google Scholar 

  30. Chang W, Liu Y, Xiao Y et al. A machine-learning-based prediction method for hypertension outcomes based on Medical Data. Diagnostics (Basel) 2019; 9(4).

  31. Islam SMS, Talukder A, Awal MA, et al. Machine Learning approaches for Predicting Hypertension and its Associated factors using Population-Level Data from Three South Asian countries. Front Cardiovasc Med. 2022;9:839379.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Hu FF, Cheng GR, Liu D, et al. Population-attributable fractions of risk factors for all-cause dementia in China rural and urban areas: a cross-sectional study. J Neurol. 2022;269(6):3147–58.

    Article  PubMed  Google Scholar 

  33. Lee M, Whitsel E, Avery C, et al. Variation in Population Attributable Fraction of Dementia Associated with potentially modifiable risk factors by race and ethnicity in the US. JAMA Netw Open. 2022;5(7):e2219672.

    Article  PubMed  PubMed Central  Google Scholar 

  34. Mulligan MD, Murphy R, Reddin C, et al. Population attributable fraction of hypertension for dementia: global, regional, and national estimates for 186 countries. EClinicalMedicine. 2023;60:102012.

    Article  PubMed  PubMed Central  Google Scholar 

  35. Marbaniang SP, Lhungdim H, Chauhan S, Srivastava S. Interaction of multiple risk factors and population attributable fraction for type 2 diabetes and hypertension among adults aged 15–49 years in Northeast India. Diabetes Metab Syndr. 2021;15(5):102227.

    Article  CAS  PubMed  Google Scholar 

  36. Fan H, Yan Y, Mi J. [Updating blood pressure references for Chinese children aged 3–17 years]. Chin J Hypertens. 2017;25(05):428–35.

    Google Scholar 

  37. NHC. Screening for overweight and obesity among school-age children and adolescents. http://www.nhc.gov.cn/ewebeditor/uploadfile/2018/03/20180330094031236.pdf (accessed Feb 26 2024).

  38. NHC. High waist circumference screening threshold among children and adolescents aged 7∼18 years. http://www.nhc.gov.cn/ewebeditor/uploadfile/2018/07/20180704145130574.pdf (accessed Feb 26 2024).

  39. Tremblay MS, Carson V, Chaput JP, et al. Canadian 24-Hour Movement Guidelines for Children and Youth: an integration of physical activity, sedentary Behaviour, and Sleep. Appl Physiol Nutr Metab. 2016;41(6 Suppl 3):S311–27.

    Article  PubMed  Google Scholar 

  40. Niu Q, Wang G, Liu B, et al. Selection and prediction of metro station sites based on spatial data and random forest: a study of Lanzhou, China. Sci Rep. 2023;13(1):22542.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. Liu F, Long Q, He H, et al. Combining the fecal immunochemical test with a logistic regression model for screening colorectal neoplasia. Front Pharmacol. 2021;12:635481.

    Article  PubMed  PubMed Central  Google Scholar 

  42. Zhao D, Zhou J, Su D, et al. Combined associations of general obesity and central obesity with hypertension stages and phenotypes among children and adolescents in Zhejiang, China. J Clin Hypertens (Greenwich). 2023;25(11):983–92.

    Article  CAS  PubMed  Google Scholar 

  43. Dong Y, Jan C, Zou Z, et al. Effect of overweight and obesity on high blood pressure in Chinese children and adolescents. Obes (Silver Spring). 2019;27(9):1503–12.

    Article  CAS  Google Scholar 

  44. Dong Y, Song Y, Zou Z, Ma J, Dong B, Prochaska JJ. Updates to pediatric hypertension guidelines: influence on classification of high blood pressure in children and adolescents. J Hypertens. 2019;37(2):297–306.

    Article  CAS  PubMed  Google Scholar 

  45. Zhou Y, Qian Z, Vaughn MG, et al. Epidemiology of elevated blood pressure and associated risk factors in Chinese children: the SNEC study. J Hum Hypertens. 2016;30(4):231–6.

    Article  CAS  PubMed  Google Scholar 

  46. The fourth report on the diagnosis, evaluation, and treatment of high blood pressure in children and adolescents. Pediatrics. 2004; 114(2 Suppl 4th Report): 555 – 76.

  47. Sun J, Steffen LM, Ma C, Liang Y, Xi B. Definition of pediatric hypertension: are blood pressure measurements on three separate occasions necessary? Hypertens Res. 2017;40(5):496–503.

    Article  PubMed  Google Scholar 

  48. Dong J, Dong H, Yan Y, Cheng H, Zhao X, Mi J. Prevalence of hypertension and hypertension phenotypes after three visits in Chinese urban children. J Hypertens. 2022;40(7):1270–7.

    Article  CAS  PubMed  Google Scholar 

  49. White K, Johansen AK, Nilsen M, et al. Activity of the estrogen-metabolizing enzyme cytochrome P450 1B1 influences the development of pulmonary arterial hypertension. Circulation. 2012;126(9):1087–98.

    Article  CAS  PubMed  Google Scholar 

  50. Zhu Y, Bian Z, Lu P, et al. Abnormal vascular function and hypertension in mice deficient in estrogen receptor beta. Science. 2002;295(5554):505–8.

    Article  CAS  PubMed  Google Scholar 

  51. Gu M, Qi Y, Li M, Niu W. Association of body mass index and alcohol intake with hypertension subtypes among HAN Chinese. Clin Exp Hypertens. 2011;33(8):518–24.

    Article  PubMed  Google Scholar 

  52. Du T, Sun X, Yin P, et al. Secular trends in the prevalence of low risk factor burden for cardiovascular disease according to obesity status among Chinese adults, 1993–2009. BMC Public Health. 2014;14:961.

    Article  PubMed  PubMed Central  Google Scholar 

  53. Bigazzi R, Zagato L, Lanzani C, et al. Hypertension in High School students: genetic and environmental factors: the HYGEF Study. Hypertension. 2020;75(1):71–8.

    Article  CAS  PubMed  Google Scholar 

  54. Pluymen LP, Smit HA, Wijga AH, Gehring U, De Jongste JC, Van Rossem L. Cesarean delivery, overweight throughout Childhood, and blood pressure in adolescence. J Pediatr. 2016;179:111–e73.

    Article  PubMed  Google Scholar 

  55. Anyaegbu EI, Dharnidharka VR. Hypertension in the teenager. Pediatr Clin North Am. 2014;61(1):131–51.

    Article  PubMed  Google Scholar 

  56. Palatini P. Sympathetic overactivity in hypertension: a risk factor for cardiovascular disease. Curr Hypertens Rep. 2001;3(Suppl 1):S3–9.

    Article  PubMed  Google Scholar 

  57. Frayn KN. Visceral fat and insulin resistance–causative or correlative? Br J Nutr. 2000;83(Suppl 1):S71–7.

    Article  CAS  PubMed  Google Scholar 

  58. Horta BL, Gigante DP, Lima RC, Barros FC, Victora CG. Birth by caesarean section and prevalence of risk factors for non-communicable diseases in young adults: a birth cohort study. PLoS ONE. 2013;8(9):e74301.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  59. Dong YH, Chen L, Liu JY, et al. [Epidemiology and prediction of overweight and obesity among children and adolescents aged 7–18 years in China from 1985 to 2019]. Zhonghua Yu Fang Yi Xue Za Zhi. 2023;57:11–9.

    CAS  PubMed  Google Scholar 

Download references

Acknowledgements

We would like to extend our gratitude to all participants for their invaluable contributions to this study. This study was supported by National Natural Science Foundation of China (72104211).

Funding

This study was supported by National Natural Science Foundation of China (72104211).

Author information

Authors and Affiliations

  1. Department of Big Data in Health Science, School of Public Health and The Second Affiliated Hospital, Zhejiang University School of Medicine, Zhejiang University, Hangzhou, Zhejiang, 310051, China

    Jiali Zhou, Weidi Sun, Chenhao Zhang, Leying Hou, Zeyu Luo, Denan Jiang, Boren Tan, Changzheng Yuan & Peige Song

  2. Department of Nutrition and Food Safety, Zhejiang Provincial Center for Disease Control and Prevention, Hangzhou, Zhejiang, 310051, China

    Jiali Zhou, Dong Zhao, Juanjuan Li & Ronghua Zhang

  3. The Fourth Affiliated Hospital of School of Medicine, and International School of Medicine, International Institutes of Medicine, Zhejiang University, Yiwu, Zhejiang, 322000, China

    Denan Jiang

Authors
  1. Jiali Zhou
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Weidi Sun
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Chenhao Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Leying Hou
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Zeyu Luo
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Denan Jiang
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Boren Tan
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Changzheng Yuan
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Dong Zhao
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Juanjuan Li
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Ronghua Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Peige Song
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

PS designed the study. DZ collected the data, JZ analyzed the data. JZ prepared the first draft. PS, RZ, JZ, WS, CZ, LH, ZL, DJ, BT, CY, DZ and JL critically revised the manuscript. All authors were involved in revising the paper and gave final approval of the submitted versions.

Corresponding authors

Correspondence to Ronghua Zhang or Peige Song.

Ethics declarations

Ethics approval and consent to participate

Ethics for conducting the study was approved by the Ethical Committee of Zhejiang University School of Medicine, Zhejiang, China (2021–029). Written informed consents were provided by the guardians of the included children.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it.The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhou, J., Sun, W., Zhang, C. et al. Prevalence of childhood hypertension and associated factors in Zhejiang Province: a cross-sectional analysis based on random forest model and logistic regression. BMC Public Health 24, 2101 (2024). https://doi.org/10.1186/s12889-024-19630-3

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12889-024-19630-3

Keywords

BMC Public Health

ISSN: 1471-2458

Contact us