Abstract
Objective. To determine the prevalence of left ventricular hypertrophy (LVH) in a multiethnic group of children and adolescents with hypertension.
Design/Methods. Pooled data from 1998 to 2001 from 3 sites belonging to the International Pediatric Hypertension Association were reviewed. Patients undergoing echocardiography to detect LVH as part of the evaluation for hypertension were included for analysis. Left ventricular mass was calculated from 2-dimensional guided M-mode echocardiographic measurements of the left ventricle. Left ventricular mass index (LVMI) was calculated as left ventricular mass/height2.7. LVH by adult criteria was defined as LVMI > 51 g/m2.7 and by pediatric criteria as LVMI > 38.6 g/m2.7. Left ventricle geometry was classified as concentric, concentric remodeling, eccentric, or normal.
Results. Data on 129 patients with a mean age of 13.6 ± 3.6 years were analyzed. The population was 67% male, 46.5% white, 38.0% African American, and 15.5% Hispanic. The prevalence of LVH was 15.5% using adult criteria and 41.1% using pediatric criteria. Increasing body mass index (BMI) was associated with a higher LVMI. Using either pediatric or adult criteria LVH was associated with BMI ≥95th percentile for age and gender. LVH and concentric hypertrophy were identified most frequently in Hispanic children.
Conclusions. LVH occurs commonly in children with hypertension and is associated with an increased BMI. LVH may be more prevalent in Hispanic children than in other ethnic groups. Prevention and treatment of obesity is important in reducing the cardiovascular risk for children with hypertension. Further evaluation of the frequency of LVH in multiethnic populations is needed.
Studies in adults with hypertension have shown that individuals with left ventricular hypertrophy (LVH) are at greater risk than those without LVH for complications such as myocardial infarction, cerebrovascular events, and congestive heart failure.1–3 With the paucity of such end points in children, documentation of LVH has been used as a marker to identify hypertensive children at risk for complications later in life. In adults, LVH has also been implicated as a predisposing factor for cardiac events.2,3
Left ventricular mass (LVM) increases during growth and thus normal must be defined in the context of body size. Pediatric studies have reported frequencies of LVH ranging from 8% to 38% of hypertensive children.4–9 The method of correcting LVM for body size and the criteria used to define LVH have varied between studies. LVM has been adjusted for height, body surface area, weight, and height raised to various powers. Because of the rise in the prevalence of obesity, indexing of LVM to weight or body surface area may allow an increased LVM to be interpreted erroneously as normal. Height2.7 (in meters) has been validated as an indicator of lean body mass and has been recommended for indexing LVM. Use of height2.7 to index LVM also minimizes the effect of age, gender, and race.10,11
The purpose of this study was to examine the prevalence of LVH in children and adolescents with hypertension. Factors that may be associated with LVH were also analyzed. Unlike most previous pediatric studies that have focused only on white and African American subjects, this multicenter study included a large number of Hispanic children and adolescents.
METHODS
In this study, a retrospective review of pooled data from 1998 to 2001 from 3 sites belonging to the International Pediatric Hypertension Association were reviewed. Data were available from the University of Texas in Houston (UT), Children’s Hospital Medical Center of Cincinnati (CHMC), and the Medical College of Georgia (MCG). Patients receiving antihypertensive medications as well as those children not on medication were included in the analysis. The following information was compiled: gender, ethnic group, age at the time of echocardiogram, weight, height, calculated body mass index (BMI), and concurrent treatment with antihypertensive medications. BMI was calculated as weight in kg/height2 in meters and was categorized as ≥95th percentile or <95th percentile according to the 2000 Centers for Disease Control and Prevention growth charts for the United States.12 The diagnosis of hypertension was made by the treating physician using the published 95th percentiles for age, gender, and height.13 For a patient to be diagnosed with hypertension, he or she must have had ≥3 elevated systolic or diastolic blood pressure readings that were >95th percentile and obtained at least 1 week apart. All children were being evaluated for hypertension in a tertiary setting.
Echocardiograms had been performed on all patients to evaluate for LVH. These studies were performed as part of an initial evaluation of hypertension or as part of ongoing care of known hypertensive patients. The timing of echocardiography and the criteria for treating the hypertension were determined at the individual center. Using 2-dimensional guided M-mode echocardiograms measurements of the left ventricular internal dimension, interventricular septal thickness and posterior wall thickness were made during diastole according to established methods by the American Society of Echocardiography.14 These measurements were performed at the individual site, and the data then were combined for analysis.
LVM was calculated from measurement of the left ventricle (LV) by using the equation LVM (g) = 0.81 [1.04 (interventricular septal thickness + posterior wall thickness + LV end diastolic internal dimension)3 − (LV end diastolic internal dimension)3] + 0.06.15 LVM index (LVMI) was calculated as LVM/height in meters2.7. Correcting LVM for height2.7 minimizes the effect of gender, race, age, and obesity.10, 11 One adult criterion for LVH is LVMI > 51 g/m2.7.16 As reported by de Simone et al,16 adult patients with hypertension and LVMI > 51 g/m2.7 have been found to be at a fourfold greater risk of cardiovascular morbid outcomes. The 95th percentile for LVMI in pediatric patients (based on normative pediatric LVMI data) is 36.88 g/m2.7 for females and 39.36 g/m2.7 for males.17 LVH by pediatric criteria was defined as LVMI >95th percentile for gender. LV geometry was determined after calculation of the relative wall thickness (RWT) by using the formula (interventricular septal thickness + posterior wall thickness)/LV end diastolic internal dimension.18 RWT was considered abnormal if it was ≥0.41.18 LV geometry was defined as concentric (elevated LVMI and RWT), concentric remodeling (normal LVMI and elevated RWT), eccentric (elevated LVMI and normal RWT), and normal.
χ2 tests, or Fisher’s exact tests when assumptions were violated, were conducted to examine whether ethnic group, gender, BMI category (<95th or ≥95th percentile for age and gender), and antihypertensive medication treatment distributions were different between centers. Analysis of covariance was used to examine differences in mean LVMI among ethnic group, gender, and BMI category, controlling for center and antihypertensive medication treatment. Posthoc multiple comparison tests were performed by using a Tukey-Kramer multiple comparison procedure on the adjusted least-square means, which allows for different sample sizes between groups.19 Logistic regression was used to examine associations among ethnic group, gender, and BMI status and the presence of LVH using adult criteria or pediatric criteria. To control for differences between the sites, center was coded as a dummy variable in these analyses. The use of antihypertensive medication treatment was also controlled for in these analyses. Finally, Cochran-Mantel-Haenszel χ2 tests were used to examine whether the geometry of LVH differed by ethnic group or BMI category, controlling for center. Because of the small number of individuals in the “other” ethnicity group (n = 4), these individuals were excluded from all analyses. All analyses were repeated excluding individuals with secondary causes of hypertension. No differences in the results were seen when eliminating those with secondary hypertension, and thus these results are not presented. All analyses were performed by using SAS 8.2, and statistical significance was assessed by using an α level of .05.
RESULTS
Data from 129 patients were reviewed from 3 centers. Overall, there were 86 males and 43 females. The mean age was 13.6 ± 3.6 years with a range of 4.2 to 22 years. The mean weight was 72 ± 27 kg, and the mean BMI was 27.5 ± 7.4 kg/m2. Seventy-one patients had a BMI >95th percentile for their age. There were 60 white patients, 49 African American patients, and 20 Hispanic patients. There were 55 patients receiving antihypertensive medication and 74 patients who were not on medication. A secondary cause of hypertension was identified in 25 subjects at MCG. The causes included chronic renal failure (9 children), renal transplantation (5 children), autoimmune diseases (3 children), glomerulopathies (4 children), reflux nephropathy (1 child), renovascular disease (1 child), hydronephrosis (1 child), and polycystic kidney disease (1 child). A diagnosis of primary hypertension was made in the remaining children after evaluation excluded secondary causes.
The prevalence of LVH was 15.5% (20 of 129) using adult criteria and 41.1% (53 of 129) using pediatric criteria. The mean LVMI was 37.7 ± 11.7 g/m2.7. Using adult criteria for LVH, concentric hypertrophy was identified in 14 of 129 (11%) children, eccentric hypertrophy was found in 6 of 129 (5%) children, and concentric remodeling was present in 23 of 129 (18%) children. Using pediatric criteria for LVH, concentric hypertrophy was identified in 25 of 129 (19%) children, eccentric hypertrophy was found in 28 of 129 (22%) children, and concentric remodeling was present in 12 of 129 (9%) children. In children identified as having LVH, concentric hypertrophy was present in 14 of 20 (70%) using adult criteria and 25 of 53 (47%) using pediatric criteria.
Distributions of weight, BMI, ethnic group, gender, and number of patients receiving treatment differed among the 3 centers (see Tables 1 and 2). As demonstrated in Table 1, the mean weight and BMI differed among the centers. The CHMC population had the highest, and the MCG group the lowest, mean weight (P = .0211) and BMI (P = .0461). However, as shown in Table 2, the proportion of children with BMI ≥95th percentile for their age and gender did not differ among the centers. CHMC had a greater proportion of white individuals, MCG had a greater proportion of African American individuals, and UT had a greater proportion of Hispanic individuals (P < .0001). Although UT had a larger proportion of males than the other 2 centers, the difference did not reach statistical significance (Table 2). Differences in the number of patients receiving antihypertensive therapy were seen among the 3 centers. At MCG, 56% of patients were on medication, as compared with 36% and 23% of patients at UT and CHMC, respectively (P = .0128). The differences in the prevalence of patients on medication in part reflects differences by center in the number of patients with secondary hypertension as well as differences in the timing of the echocardiographic examination.
Analysis of Variance Comparing Differences Among Centers for Age and Body Measurements
χ2 or Fisher’s Exact Tests for Differences Among Centers for Ethnic Group, Gender, BMI Category, and Antihypertensive Medication Treatment
The effect of center, treatment, BMI category, ethnic group, and gender on LVMI is shown in Table 3. After controlling for center and treatment, ethnicity, gender, and BMI status were significantly associated with LVMI. Children with BMI ≥95th percentile for their age had a significantly higher mean LVMI, as compared with those with BMI <95th percentile for their age. Males had a significantly higher mean LVMI than females, and Hispanics had a significantly higher mean LVMI than whites.
Analysis of Covariance on LVMI (LVM/Height in Meters2.7)
With regard to LVH, logistic regression analysis (Table 4) showed a statistically significant association for BMI status and ethnic group. Other variables including gender, center, and use of antihypertensive treatment were not associated with LVH. Children with LVH were more likely than children without LVH to have a BMI ≥95th percentile when using either the adult (odds ratio [OR]: 4.22) or pediatric (OR: 5.02) criteria. Considering ethnic group, children with LVH were more likely to be Hispanic or African American than those without LVH. Using adult criteria, LVH was found in 8 of 20 (40%) Hispanic children, compared with 8 of 49 (16%) African American and 4 of 60 (7%) white children. Using pediatric criteria, LVH was documented in 14 of 20 (70%) Hispanic children, 19 of 49 (39%) African American children, and 20 of 60 (33%) of white children. However, in the logistic regression analysis, these differences were only statistically significant for the Hispanic children with LVH (OR: 7.56) by adult criteria (Table 4).
Logistic Regression Analysis on LVH
Assessment of the distribution of LV geometry by ethnic group and BMI category is shown in Table 5. When pediatric criteria were used to define LVH, a statistically significant difference was found in LV geometry for ethnic group but not BMI. Concentric hypertrophy was more likely in Hispanics and African Americans, as compared with whites. Eccentric hypertrophy was more likely in whites, as compared with African Americans and Hispanics.
χ2 Tests of Geometry of LVH by Ethnic Group and Obesity
DISCUSSION
In this study we evaluated the frequency of LVH in children identified with hypertension at 3 centers. We found that 41% of the children and adolescents had LVH by pediatric criteria, and 16% had LVH even when using adult criteria. These findings demonstrate that in children LVH is prevalent and can be severe. Other investigators have reported LVH frequencies varying from 8% to 38% depending on the method used to index LVM, criteria for LVH, and technique used to detect LVH.4–9 Using the same pediatric and adult criteria as used in this study, Daniels et al4 found that 46% of hypertensive pediatric patients had an LVMI >95th%, whereas only 8% had an LVMI > 51 g/ht2.7. The frequency of severe LVH was much greater in this study, perhaps related to the greater racial diversity. Although this study included patients with secondary forms of hypertension, they were not more likely to have severe LVH. Children with hypertension and severe hypertrophy such as that identified by the adult cut point would seem to be at significant risk for cardiovascular morbidity in the future. The pediatric threshold of the 95th percentile may be appropriate for identifying hypertensive children at risk of future increase in LVM and in need of lifestyle or pharmacologic intervention.
In some studies of adults, LVH has been reported to occur more frequently in African American patients with hypertension than in whites.20–22 Unlike other studies of pediatric populations,5,6 we identified LVH more frequently in Hispanic and African American children, as compared with white children. The frequency of LVH in the Hispanic children was very high at 70% by pediatric criteria and 40% by adult criteria. The higher rate of LVH in the Hispanic children may be related to various factors including the prevalence of obesity, dietary habits, genetic factors, or access to medical care. Further evaluation of LVH in children of various ethnic groups with larger sample sizes is indicated.
We also found an association between ethnic group and concentric geometry. Those individuals with concentric hypertrophy were more likely to be Hispanic or African American, as compared with white. In studies of adult patients it has been noted that individuals with concentric hypertrophy have the highest blood pressure and the worst prognosis for cardiovascular disease, whereas those with eccentric and concentric remodeling patterns have an intermediate prognosis.1,23,24 Some studies in adult and pediatric patients with hypertension have shown a greater frequency of concentric hypertrophy in African Americans as opposed to whites,4,21,22 whereas others have not found such an association.25 Our findings suggest that not only is LVH more prevalent in the Hispanic population, but it also may be more severe, as evidenced by the high prevalence of concentric hypertrophy. The effect of ethnic group on LV geometry has not been studied extensively in pediatric populations.
In this study of children with hypertension, LVH was associated with an elevated BMI. Other studies of hypertensive children and adults4,5,20,26 have shown a relationship between obesity and increased LVMI and LVH. Some investigators have shown an association of obesity itself with an increased LVMI independent of the effect of hypertension.5,20,27–29 In contrast, de Simone et al30 demonstrated that the risk of LVH was significantly higher in children with a high casual blood pressure, compared with children with normal blood pressures, independent of the effect of obesity. Regardless, an aggressive approach to prevention and treatment of obesity in pediatric patients with hypertension is indicated to reduce the future cardiovascular morbidity in these children.
There are limitations of this retrospective study, most notably the lack of specific blood pressure data, the lack of a central laboratory for reading echocardiograms, and the lack of uniform criteria for initiating antihypertensive therapy. Although differences among the 3 centers were considered in our statistical analysis, potential interobserver variability in echocardiography results can not be completely excluded. Also, information on the severity and duration of hypertension, elements that would be expected to affect the evolution of LVH, are not available. Furthermore, the data presented here may overestimate the frequency of LVH in hypertensive children, because the patients in the present study were referred for evaluation and may have had more severe hypertension than would be seen in a population-based study.
Despite these problems, this large collaborative study offers valuable information for those involved in the care of children with hypertension. Unlike previous studies,4–9,31–33 this analysis encompasses a large multiethnic pediatric population from 3 different geographic regions in the United States. The results emphasize that the impact of hypertension may be particularly important in the pediatric Hispanic population. Our findings also highlight the importance of obesity as a risk factor for LVH. In view of the rising prevalence of obesity in children over the last decade, these findings are timely. Many of the previous studies evaluating LVH in children were performed at an earlier point before the prevalence of obesity had increased dramatically.8,9,31–34 In contrast to several earlier pediatric studies,8, 9,32,34 we used a method for indexing LVM that is least affected by obesity. Using LVM indexed to height2.7, we confirmed the previously noted association between obesity and LVH in hypertensive children.4,5 Unfortunately, if current trends continue, the frequency of obesity-related hypertension and LVH can be expected to increase in the future. Hispanic children may be at particularly high risk for these problems. It is crucial that physicians treating children appreciate both hypertension and obesity as risk factors for future cardiovascular morbidity.
OUTSIZED PATIENTS
“The sharply rising number of obese Americans is leading medical equipment manufacturers and ambulance crews to supersize their stretchers. Manufacturers are adding thicker aluminum frames, bulkier connectors and extra spine supports to create stretchers with a capacity of 650 pounds instead of the standard 350 to 500. Ambulance crews are switching to the heavy-duty models to avoid injuries to rescue workers and patients alike.”
Associated Press. Manufacturers beef up stretchers to handle obese. October 18, 2003
Submitted by Student
Footnotes
- Received February 12, 2003.
- Accepted May 29, 2003.
- Address correspondence to Coral Hanevold, MD, Medical College of Georgia, Department of Pediatrics, Section of Pediatric Nephrology, 1120 15th St, Augusta, GA 30912-3795. E-mail: chanevol{at}mail.mcg.edu
REFERENCES
- Copyright © 2004 by the American Academy of Pediatrics