OBJECTIVE: The goal was to determine patient-, provider-, and clinic-level predictors of unrecognized elevated blood pressure (BP) in children. We hypothesized that being of healthy weight, having a BP of <120/80 mmHg, and being seen by a less experienced provider would result in decreased recognition.
METHODS: A cross-sectional study of clinic visits for children 3 to 20 years of age at an urban, pediatric primary care practice between January 1, 2006, and June 30, 2006, was performed. Children with elevated systolic or diastolic BP (≥90th percentile or ≥120/80 mmHg) were included. Recognition was defined as having any of the following documented: repeat BP measurement, elevated-BP/hypertension diagnosis, plan to recheck BP, or initiation of hypertension evaluation. Multivariate logistic regression analysis was used to identify characteristics associated with underrecognition.
RESULTS: Elevated BP occurred in 779 (39%) of 2000 visits. Of 726 cases included in the analysis, 87% were not recognized by providers. Patient-level predictors of underrecognition included systolic BP of <120 mmHg (odds ratio: 7.7 [95% confidence interval: 3.2–18.6]), diastolic BP of <80 mmHg (odds ratio: 2.4 [95% confidence interval: 1.1–5.0]), decreasing BMI z score, male gender, older age, lack of family history of cardiovascular disease, and negative medical history findings. Being seen by a nurse practitioner and being seen by a less-experienced provider also were significant predictors.
CONCLUSIONS: Most BP elevations were not recognized by providers. Poor recognition was most influenced by the absence of obviously elevated BP, obesity, and family history of cardiovascular disease.
WHAT'S KNOWN ON THIS SUBJECT:
Hypertension and prehypertension in childhood are underdiagnosed, likely because of decreased recognition of elevated BP.
WHAT THIS STUDY ADDS:
Pediatric providers often do not measure BP during scheduled visits, despite current guidelines. Providers also often do not recognize BP elevations, and this lack of recognition is most influenced by the absence of CVD risk factors.
The prevalence of hypertension in children has increased dramatically in the past several decades.1,–,4 Studies have shown that 3.2% to 4.5% of all children have hypertension, an increase from 0.3% to 1.2% of all children only 25 to 35 years ago. Unfortunately, the burden of pediatric hypertension is projected to increase further, bringing with it not only increases in known hypertensive sequelae in childhood but also increases in the numbers of adults with hypertension. To prevent the long-term complications of untreated hypertension, it is important for pediatric health care providers to recognize children with elevated blood pressure (BP) as early as possible. Once elevated BP is recognized, confirmation, evaluation, and treatment can commence.
Despite current recommendations that all children ≥3 years of age have their BP measured at each health care visit5 and increased awareness and efforts aimed at primary prevention, providers have been shown to underdiagnose hypertension in children.6 This is likely attributable to diminished recognition of elevated BP. To our knowledge, no study has investigated how frequently elevated BP is recognized during scheduled clinic visits or what factors may be associated with lack of recognition. Therefore, we sought to identify patient-, provider-, and clinic-level predictors of underrecognition of elevated BP in children seen during scheduled clinic visits at a large, urban, academic, pediatric primary care clinic.
Study Design and Population
The study consisted of a nonconcurrent, cross-sectional review of all scheduled visits of children 3 to 20 years of age from January 1, 2006, through June 30, 2006. The large, academic, primary care practice has 10 half-day clinic sessions (“clinic types”) with (1) nurse practitioners, (2) pediatric residents supervised by an attending pediatrician or general pediatrics fellows (postgraduate year [PGY] 4–6), or (3) a pediatric resident or adolescent medicine fellow supervised by an attending adolescent medicine specialist. Each resident clinic session includes 1 or 2 supervisors, or preceptors, and up to 7 residents at different training levels. The initial triage BP is measured by a nurse or nursing assistant through oscillometry. Clinic visit information is maintained in an electronic medical record (EMR) with all information for the visit, including BP measurements, documented and transcribed there by the primary care provider.
All visits in which a child had elevated systolic BP (SBP) or diastolic BP (DBP) measurements were included. Elevated BP was defined as SBP or DBP measurements of ≥90th percentile for age, gender, and height percentile5 or SBP of ≥120 mmHg and/or DBP of ≥80 mmHg, if these values were in <90th percentile. Children who had a history of hypertension or were being seen because of an acute illness were excluded, as were children with missing height information who also had SBP of <120 mmHg and DBP of <80 mmHg.
Data Collection and Variables
Patient and Provider
Each patient's clinic note was reviewed in its entirety. Data collected through manual review of the EMR included patient age, gender, and self-reported race/ethnicity (white, black, Hispanic, Asian, Native American, or other), medical history (defined as any medical problem listed in the clinic note or patient's problem list for which information was updated by providers at each clinic visit), specific medication use, family history of cardiovascular disease (CVD) (defined as having any of the following listed in the family history: hypertension, early myocardial infarction, hyperlipidemia, or stroke), and presence of hypertensive symptoms at the time of the visit (defined as having any of the following recorded for the visit: headache, dizziness, blurry vision, chest pain, or palpitations). In addition, height, weight, and initial BP data were collected. BMI, BMI percentile, and BMI z score values were calculated on the basis of the 2000 Centers for Disease Control and Prevention growth charts.7 Children were classified according to degree of obesity (healthy weight, BMI of <85th percentile for age and gender; overweight, BMI of 85th–94th percentile; obese, BMI of ≥95th percentile).8
Children were categorized as having a comorbid condition if they had any of the following listed in their medical history: obesity, metabolic syndrome, insulin resistance, diabetes mellitus, prematurity, or kidney disease. Children were considered to be taking a medication known to increase BP if they were taking ≥1 of the following medications: attention-deficit/hyperactivity disorder medications, asthma/allergy medications, estrogen-containing contraceptives, or cough/cold medications.
Provider characteristics included level of training (intern, resident, senior resident [PGY 1–3], fellow [PGY 4 or 5], or nurse practitioner), self-reported race (white, white Hispanic, black, Asian, or from Indian subcontinent), and gender. Because of the small number of patients in race categories other than black, patient race and provider race were categorized as black or not black.
Clinic variables included clinic volume (measured as both total number of patients seen by a single provider per session and total number of patients seen by all providers during a clinic session) and clinic complexity (measured as number of coexisting medical problems per patient and total number of medical problems per clinic session). Clinic type also was included.
Recognition of elevated BP was defined as the presence of ≥1 of the following in the EMR for that visit: (1) BP measurement repeated; (2) assessment of abnormal BP, elevated BP, or hypertension; (3) plan to recheck BP; or (4) plan for evaluation of elevated BP. Data were abstracted from the EMR by 2 trained abstractors and were reconciled to ensure data validity. This study was approved by the institutional review board of Johns Hopkins University School of Medicine.
To allow for comparisons of degree of BP elevation among children of different ages, genders, and heights, each child's 90th and 95th percentile BP indices were calculated by dividing his or her initial BP by the 90th and 95th percentile BP values, respectively.5 Normative values for 17-year-old children were used for all children of ≥17 years. A BP index of ≥1 indicates elevated BP (above either the 90th or 95th percentile BP, depending on the threshold used). In addition, each child's SBP and DBP were categorized as being above or below 120 mmHg and 80 mmHg, respectively.
The χ2 test and Student's t test were used to compare patient, provider, and clinic characteristics for patients with recognized and unrecognized elevated BP. Univariate and multivariate logistic regression analyses were used to determine the odds of underrecognition on the basis of these characteristics. Because providers in each clinic group might practice similarly, generalized estimating equations were used in the multivariate logistic regression analysis to account for clustering according to clinic session. In the building of the multivariate logistic regression model, the following variables were considered a priori for inclusion in the regression analyses and ultimately were included in the final model: age, gender, race, BMI z score, family history of CVD, SBP of <120 mmHg, DBP of <80 mmHg, presence of hypertensive symptoms, no medical history, taking any medication known to increase BP, provider training level, clinic volume, clinic complexity, and clinic type. Of the BP variables, we chose SBP of <120 mmHg and DBP of <80 mmHg because they were the most significant BP variables in the univariate analyses and were considered a priori to be the most likely predictors of underrecognition, given that these cutoff values are often used for adults. Of the factors that included obesity or weight measurements, only BMI z score was retained in the final model, to avoid overadjustment.
The predicted probabilities of recognition according to degree of obesity and age were calculated on the basis of median values for continuous variables and reference group values for categorical variables in the regression equation. Stratified analyses were performed to check for effect modification, and diagnostic checks for colinearity using variance inflation factors and tolerance were conducted; these checks revealed no statistically significant interactions or signs of colinearity. Missing weight and height data limited BMI calculations for 36 children (5% missing). For the multivariate logistic regression analysis, missing data were imputed on the basis of age, height, and weight data. To be most conservative in the analysis, additional missing information for 5 participants was handled as follows: family history of CVD and presence of hypertensive symptoms were coded as not present; total number of patients seen by the provider was replaced with the median value of 4; provider training level was replaced with PGY 3 as the median; and missing DBP variables were coded as <80 mmHg in both the univariate and multivariate logistic regression analyses. P values of <.05 were considered significant. All analyses were conducted by using Stata 9.2 (StataCorp, College Station, TX).
During the study period, there were a total of 2501 scheduled clinic visits for children ≥3 years of age. Of those visits, 2000 (80%) had BP measurements recorded in the EMR. BP was elevated in 779 cases (39%), but elevation was recognized in only 133 of those cases (17%). Thirty-six of the 133 recognized cases and 7 of the 646 unrecognized cases had previous diagnoses of hypertension and were excluded from further analyses. In addition, 10 patients were seen outside the predefined clinic sessions and were excluded from further analyses. The final study population consisted of 726 total visits, including 630 (87%) with unrecognized BP elevations.
Table 1 compares the characteristics of children with unrecognized versus recognized BP elevations, and the initial BP measurements are compared in Table 2. As might be expected, children with more-modest BP elevations were less likely to have their BP elevations recognized.
One-half of the children with elevated BP were either obese or overweight (BMI of ≥85th percentile for age and gender) (Table 3). There was decreased recognition of elevated BP among children who were of healthy weight (7%), compared with obese children (22%; P < .001).
Logistic regression analyses revealed several patient-, provider-, and clinic-level predictors of underrecognition (Table 4). Older age, male gender, decreasing BMI z score, absence of CVD in the family, lack of medical history, SBP of <120 mm Hg, and DBP of <80 mm Hg were patient characteristics associated with underrecognition. Of the clinic- and provider-level characteristics, being seen by a nurse practitioner and being seen by a provider with less experience, measured as training level, also were associated with decreased recognition. Being seen by a provider who was being precepted by an attending physician (compared with being precepted by a fellow) did not affect recognition, as evaluated in both χ2 (rate of underrecognition by attending preceptor versus fellow preceptor: 85% vs 87%; P = .6) and multivariate logistic regression (odds ratio: 1.0 [95% confidence interval: 0.5–1.9]) analyses.
To investigate further how childhood obesity may affect recognition, we calculated the predicted probability of recognition according to degree of obesity. The predicted probability of having elevated BP recognized decreased as the SBP decreased (measured as either millimeters of mercury or SBP index); the probability was lowest for children of healthy weight, with obese children demonstrating the highest predicted probability of recognition (Fig 1). Consistent with the multivariate logistic regression analysis results, with stratification according to age, the predicted probability of having elevated BP recognized was 30% less among older children (≥14 years of age), compared with the youngest children (<5 years of age), across all levels of SBP (Fig 2).
In this study of children 3 to 20 years of age who were seen for a scheduled visit in a large, urban, academic, primary care practice, we found that the majority of children without a previous diagnosis of hypertension who had elevated BP measurements were not recognized as such by their providers. The most notable predictors of underrecognition were related to the absence of obvious risk factors for hypertension, such as obesity, medical history or family history of CVD, and not “obviously” elevated BP (≥120/80 mm Hg). The association of these clinical characteristics with decreased recognition has several possible explanations. Providers caring for patients who lack obvious CVD risk factors, such as those described above, may be falsely reassured by BPs that are not obviously elevated and, as a result, may be less likely to refer to tables of normative values. Providers may be more likely to justify elevations as measurement errors and not attribute the appropriate significance to these readings. Alternatively, providers may be less likely to pay attention to BP measurements for children without obvious CVD risk factors.
It may not be that patient characteristics lead to underrecognition but that certain patient characteristics are more likely to trigger recognition. Obesity, which is epidemic among the pediatric population and has received widespread attention, is a known risk factor for hypertension and CVD. Increasing BMI z scores were associated with less underrecognition in our study; obese and overweight children had higher rates of recognition, compared with children of healthy weight. Increased provider awareness of the association of obesity with hypertension and CVD might have increased providers' attentiveness to patients' BP, thus leading to improved recognition.
Although the cause of this differential recognition is not completely clear, it should be noted that the overall rate of recognition among obese and overweight children was still poor at 20%. This highlights the need for greater efforts aimed at educating providers about this important pediatric health risk, given the growing number of children with hypertension. In addition, providers need to pay attention to a potentially vulnerable group of children, namely, those of healthy weight, who experienced the lowest rate of recognition among providers in our study (7%). Similar findings from a large-scale, school-based, screening study found that children who were referred for evaluation of hypertension by their primary care providers were more overweight than those who were referred through school screening,9 which indicates that providers may be more attuned to the cardiovascular health of children at most risk.
Because providers were least effective at recognizing elevated BP when a child's BP was less than the adult normative value of 120/80 mm Hg, the greatest barrier to recognition may be the complex steps involved in categorizing a child's BP as normal or abnormal. The increased odds of underrecognition with increasing age illustrate the concept that, the older the child is, the more obviously elevated BP needs to be for recognition. Figure 2 depicts this graphically, showing the higher predicted probability of recognition for younger children at lower BP measurements, compared with older children; a SBP of 120 mm Hg is more obviously elevated for a 4-year-old child than for an 18-year-old youth. The aforementioned characteristics, in addition to male patient gender and being seen by a nurse practitioner or by a provider with less experience, were all shown to contribute to the poor overall rate of recognition by providers in our study. The gender discrepancy is interesting, given that, among adults, women historically are less frequently positively identified as experiencing cardiovascular events, compared with men. Provider inexperience likely speaks to the complex steps involved in recognition and the need for simplification to enhance recognition.
Our finding that 39% of children had elevated BP measurements is higher than rates reported in other studies of hypertension and prehypertension in adolescents.3,4,6 This is expected, because we were reporting not the number of children with hypertension seen in a pediatric clinic but the number of children with a single elevated BP value measured through oscillometry during a scheduled clinic visit. If these children were to undergo repeat BP measurements through manual auscultation, as recommended in current guidelines,5 the majority most likely would be found to be normotensive.4 In our study, among the children with repeated BP measurements performed in the clinic, 51 of 56 BP values decreased in subsequent measurements.
Although our study achieves its stated aims of describing the predictors for underrecognition of elevated BP in children, there are some limitations. Several potential factors, including provider familiarity with the patient, actual visit length, number of clinic support staff members, parental education, and native language, were not measured because of our inability to capture the information through chart review.10 Each of these variables might affect the time available to compare measured BPs against tables of normative values, which would provide additional evidence that the greatest barrier to recognition may be the complex steps involved. In addition, provider familiarity with recommended guidelines and definitions of elevated BP was not assessed. Because this study was conducted in a single, urban, medical center with trainees, the findings may not be generalizable to other providers and practice settings.
Despite these limitations, our study has significant strengths. We were able to study a large cohort of children seen for scheduled visits in a primary care clinic, with detailed patient, provider, and clinic factors not examined previously in a pediatric study. The data obtained were more inclusive and precise, compared with billing or administrative data with International Classification of Diseases, Ninth Revision codes. The only other published study to date that examined provider recognition of hypertension in children used specific data fields within the EMR and found, through manual review of a small subset of 50 charts (of 14 000), that the rate of recognition increased by 6% when specific clinic notations not included in the International Classification of Diseases, Ninth Revision diagnostic codes were included,6 similar to how data were abstracted in our study.
Our study emphasizes that rates of recognition of elevated BP in a busy pediatric clinic are low. Although obvious signs such as obesity, family history of CVD, or BPs significantly elevated above adult values may prompt providers to recognize BP elevation in children, cases lacking these findings are underrecognized. This speaks to the many demands placed on pediatric health care providers, as well as the complexity involved in measuring and recognizing elevated BP in children in a primary care clinic. Innovative strategies are needed to assist providers in detecting elevated BP. Computerized systems have been shown to improve patient care in hospital settings, with modification of drug dosing on the basis of renal function.11 To improve recognition and ultimately to increase early diagnosis and treatment of children with hypertension, it is imperative that systems such as computerized alerts be put into place to help improve provider recognition of elevated BP. In addition, enhanced provider education is needed to ensure that providers are practicing in the context of current guidelines and to improve our ability to detect elevated BP in children who may seem to be at low risk.
This work was supported in part by the American Kidney Fund Clinical Scientist in Nephrology program and the National Kidney Foundation of Maryland Mini-Grant program. Dr Parekh was supported by National Institutes of Health grants 1T32 DK07732, 1UO1 DK57304-01, and R01 DK72367-01A2.
We thank Jessica Levy, BA (Emory University), Nwabundo Nwankwo, MBBS, MPH (Johns Hopkins University), and Diane Koher, MS (Johns Hopkins University) for their contributions toward data collection.
- Accepted January 28, 2010.
- Address correspondence to Tammy M. Brady, MD, MHS, Johns Hopkins University, David M. Rubenstein Child Health Building, 200 North Wolfe St, Room 3057, Baltimore, MD 21287. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- BP =
- blood pressure •
- SBP =
- systolic blood pressure •
- DBP =
- diastolic blood pressure •
- CVD =
- cardiovascular disease •
- PGY =
- postgraduate year •
- EMR =
- electronic medical record
- Fixler DE,
- Laird WP,
- Fitzgerald V,
- Stead S,
- Adams R
- Sorof JM,
- Lai D,
- Turner J,
- Poffenbarger T,
- Portman RJ
- National High Blood Pressure Education Program Working Group on High Blood Pressure in Children and Adolescents
- Centers for Disease Control and Prevention
- Barlow SE
- Sorof JM,
- Turner J,
- Martin DS,
- et al
- Copyright © 2010 by the American Academy of Pediatrics