Detection of hypertension and prehypertension in paediatric patients with type 1 diabetes using a simple blood pressure table

Abstract

BACKGROUND:

The diagnosis of hypertension is often unrecognized in the general paediatric and type 1 diabetes populations. Reference to a simple blood pressure table has been proposed as a screening tool for the identification of abnormal paediatric blood pressure. This simple table lists one cut-off blood pressure value for each sex and year of age, compared with the standard blood pressure tables, which list 476 abnormal and normal blood pressure values.

OBJECTIVE:

To determine the sensitivity and specificity of the simple blood pressure table in identifying hypertensive and abnormal (hypertensive or prehypertensive) blood pressure values in children with type 1 diabetes.

METHODS:

A retrospective cohort study was conducted. Data regarding baseline characteristics and blood pressure measurements were collected. Blood pressure values were categorized as nonhypertensive, prehypertensive or hypertensive using the standard and simple blood pressure tables. Sensitivity and specificity of the simple blood pressure table was determined for the identification of hypertensive and abnormal (hypertensive or prehypertensive) blood pressure values using a generalized estimating equation approach.

RESULTS:

The simple blood pressure table had sensitivities of 100% (95% CI 95.7% to 100%) and 100% (95% CI 97.6% to 100%), and specificities of 61.1% (95% CI 54.0% to 67.8%) and 81.3% (95% CI 74.4% to 86.6%), for the identification of hypertensive and abnormal blood pressure values, respectively.

CONCLUSION:

The simple blood pressure table is an effective screening tool that may enable earlier detection and timely treatment of abnormal blood pressure.

Hypertension (HTN) in childhood has been associated with early markers of cardiovascular disease (1–9) and is an independent risk factor for HTN in adulthood. Furthermore, in patients with type 1 diabetes (T1D), HTN is a significant contributor to the development of both micro- and macrovascular complications (10–13). It is estimated that up to 75% of the cardiovascular disease risk in individuals with diabetes may be attributable to HTN (14).

Although early identification and management of HTN is essential to prevent and slow the progression of HTN-related complications, studies involving both the general paediatric and diabetes populations suggest that 65% to 89% of patients with pre-HTN and HTN are not identified (15–17).

The diagnoses of pre-HTN and HTN are based on standard paediatric blood pressure (BP) tables outlined in The Fourth Report on the Diagnosis, Evaluation, and Treatment of High Blood Pressure in Children and Adolescents (18); we will refer to these tables as the ‘standard BP tables’. Briefly, the standard BP tables list 476 normal and abnormal BP values based on sex, age and height, providing a precise classification of BP according to body size (18). To identify abnormally elevated BP values, the clinician must know a patient’s height percentile and use the standard BP tables to determine the BP percentile, which may not always be practical in a busy clinical setting. Hansen et al (15) and Brady et al (17) have both suggested that high rates of unrecognized HTN in children may be due to a lack of knowledge of normal paediatric BP ranges and to the time required to use the standard BP tables. Recognizing the challenges associated with the use of the standard BP tables, Kaelber and Pickett (19) proposed a simplified BP table as a screening tool for identifying abnormal BP values in the paediatric population. This table lists one systolic and one diastolic BP cut-off value according to sex and age, corresponding to the fifth percentile for height and the 90th percentile for BP (maximum 120/80 mmHg) on the standard BP tables (19). The cut-off values in this table, therefore, represent the lowest abnormal BP values in the prehypertensive range, regardless of height percentile (19). This table is referred to as the ‘simple BP table’.

In the present study, we aimed to determine the sensitivity and specificity of the simple BP table compared with the standard BP tables for the identification of hypertensive and abnormal (pre-HTN or HTN) BP values in a paediatric T1D population.

METHODS

A retrospective cohort study involving paediatric T1D patients, who were randomly selected (using a computer-generated list of random numbers) from the Children’s Hospital of Eastern Ontario (Ottawa, Ontario) Diabetes Database, was conducted. This database contains approximately 900 T1D patients. Patients who were between three and 18 years of age who had been diagnosed with T1D for at least one year were included. Patient charts were reviewed for all diabetes clinic encounters over a two-year period between January 1, 2008 and December 31, 2009. Data were abstracted for age, sex, height and BP values. As per the Canadian Diabetes Association guidelines, BPs are measured in the authors’ clinic at least two times per year, or approximately once every six months, using an oscillometric BP device (Dinamap V100, GE Medical Systems Information Technologies Inc, USA) (20). If more than one BP reading was taken during a clinic encounter, the lowest BP value was used. Height percentiles were calculated based on the Centres for Disease Control and Prevention (Georgia, USA) growth curves published in 2000 (21,22).

BP percentiles were used to classify BP values based on the standard BP tables. BP readings were converted to percentiles using the standard BP tables (18). Based on the definitions of The Fourth Report on the Diagnosis, Evaluation, and Treatment of High Blood Pressure in Children and Adolescents, BP values were categorized using the standard BP tables as either ‘nonhypertensive’ if they were ≤90th percentile, as ‘prehypertensive’ (pre-HTN) if they were between the 90th and 95th percentile (or ≥120/80 mmHg), or as ‘hypertensive’ (HTN) if they were ≥95th percentile for sex, age and height percentile. The all-inclusive term ‘abnormal BP’ was also used to refer to BP values classified by the standard BP tables as HTN or pre-HTN.

Absolute BP values were used to classify BP values based on the simple BP table (19). The simple BP table lists one systolic and one diastolic BP cut-off value for each sex and year of age. BP values were categorized as ‘nonhypertensive’ if they were less than the cut-off and as ‘abnormal’ if they were equal to or greater than the cut-off for sex and age.

The present study was approved by the Research Ethics Board at the Children’s Hospital of Eastern Ontario.

Statistical analyses

Statistical analyses of the data were performed using GraphPad version 5.0 (GraphPad Software Inc, USA) for Windows (Microsoft Corporation, USA) and R software, version 2.15.1 (R Foundation for Statistical Computing, Austria). Descriptive statistics were used to describe baseline characteristics. The variables were tested for normal distribution using the D’Agostino and Pearson omnibus test. All variables were found to be non-normally distributed and were, therefore, presented as medians and interquartile ranges (IQRs).

Specificity estimates and CIs adjusted for correlation within patients were computed using a generalized estimating equation approach (23).

Given the design of the simple BP table, sensitivity was 100%. The generalized estimating equation approach could not be used to measure sensitivity because the simple BP table was always positive for patients who were hypertensive or had abnormal blood pressure; therefore, the estimation algorithm could not converge. Instead, bounds on the exact binomial lower 95% confidence limit were considered, either ignoring the fact that there were multiple abnormal or HTN BPs in some patients (ie, considering all abnormal values as independent), or using a single abnormal or HTN BP from each patient with one or more abnormal or HTN BP values.

The kappa statistic was used to assess agreement between the simple BP and standard BP tables. The following qualitative descriptors grading the level of agreement according to kappa values were used: <0, no agreement; 0 to 0.20, slight agreement; 0.21 to 0.40, fair agreement; 0.41 to 0.60, moderate agreement; 0.61 to 0.80, substantial (very good) agreement; and 0.81 to 1.0, almost perfect agreement (24).

RESULTS

The study population consisted of 62 females and 45 males, with a median age of 13.2 years (IQR 10.2 to 15.8 years). In total, 378 BP values were documented in 107 patients. The median number of BP values per patient was three (IQR three to four) over a two-year period. Overall, there were 154 abnormal BP values identified using the standard BP tables (83 HTN and 71 pre-HTN), and 194 abnormal BP values identified using the simple BP table.

For identifying HTN BP values, the simple BP table had a specificity of 61.1% (95% CI 54.0% to 67.8%) and a sensitivity of 100% (95% CI 95.7% to 100%) considering 83 independent HTN BP values, or 95% CI 92.0% to 100% considering 44 patients with at least one HTN BP value (Table 1). For identifying abnormal BP values (HTN or pre-HTN), the simple BP table had a specificity of 81.3% (95% CI 74.4% to 86.6%) and a sensitivity of 100% (95% CI 97.6% to 100%) considering 154 independent abnormal BP values, or 95% CI 94.6% to 100% considering 66 patients with at least one abnormal BP value (Table 2).

TABLE 1.

Identification of hypertension (HTN) using the simple blood pressure (BP) table versus the standard BP tables

		Standard BP tables		Total
		HTN	Non-HTN
Simple BP table	Abnormal	83	111	194
	Non-HTN	0	184	184
Total		83	295	378

Data presented as n

TABLE 2.

Identification of abnormal blood pressure (BP) (prehypertension and hypertension) using the simple BP table versus the standard BP tables

		Standard BP tables		Total
		Abnormal	Non-HTN
Simple BP table	Abnormal	154	40	194
	Non-HTN	0	184	184
Total		154	224	378

Data presented as n. HTN Hypertension

The agreement between the simple BP table and the standard BP tables was found to be moderate for the detection of HTN BP values (kappa=0.42 [95% CI 0.35 to 0.50]) and very good for the detection of abnormal BP values (kappa=0.79 [95% CI 0.73 to 0.85]).

DISCUSSION

In our study, we found the simple BP table to be a valid screening tool. The present study was the first to validate the use of the simple BP table in a paediatric T1D population.

The simple BP table is a screening tool designed to have 100% sensitivity for identifying abnormal paediatric BP values. In developing this tool, Kaelber and Pickett (19) acknowledged that a certain number of false positives would occur given the choice of cut-offs, but concluded that the benefits of increased identification of true positives outweighed the risk of false positives given the high rate of undiagnosed HTN in the paediatric population. Their original article did not report the specificity of this screening tool. In our study, we confirm that the simple BP table has a sensitivity of 100% for the detection of HTN and abnormal BP (HTN or pre-HTN) values, as predicted based on its design. We also report specificities of 61.1% (95% CI 54.0% to 67.8%) and 81.3% (95% CI 74.4% to 86.6%) for the detection of HTN and abnormal (HTN and pre-HTN) BP values, respectively.

Due to its high sensitivity, the simple BP table is a powerful tool for screening T1D patients for abnormal BP values, a finding that may be applicable to the general paediatric population. Any loss of specificity is due to the high sensitivity of this tool, which sets the cut-off point at the lowest abnormal BP value according to age and sex. False-positive identification of abnormal BP values may occur in children whose height is above the fifth percentile. As with any screening tool, a positive screen is simply indicative of the need for further evaluation, which must include consultation of the standard BP tables.

The clear alternative to the use of the standard BP tables for identification of abnormal BP values is to integrate computerized programs into the clinical setting that are capable of calculating BP percentiles for physicians and alerting them to abnormal values (15,17). However, the availability of these programs is not universal, particularly in many low-resource countries around the world. The simple BP table is free, easy to use and can be immediately integrated into any clinical practice.

We acknowledge that our study had limitations. Given the retrospective nature of our study, it was impossible to control for the accuracy of the BP measurements, including nursing technique and/or choice of cuff size. Also, in our clinic, BP is primarily measured using an oscillometric device (unless repeated by manual BP cuff), which could lead to slightly higher SBP measurements (by 2 mmHg to 5 mmHg) compared with measurements using a manual cuff, on which the standard BP tables are based. However, oscillometric devices are widely used and accepted for BP screening measurements, which is consistent with the purposes of our study. Furthermore, the purpose of our study was to compare the classification of BP levels as abnormal or normal, using two BP tables, which are based on the same reference dataset; therefore, the fact that BP values were measured using an oscillometric device does not impact the study question. Our findings suggest that the simple BP table may be used to detect patients at risk for abnormal BP; patients identified as being at risk for abnormal BP then require further evaluation, including consultation of the standard BP tables and assessment of BP by auscultation before a BP reading can be classified as abnormal.

CONCLUSION

The simple BP table proposed by Kaelber and Pickett (19) is an effective screening tool for the identification of abnormal BP in children with T1D. Although it will identify false positives, the simple BP table is easy to use and ensures that no abnormal BP values are missed. We propose that the simple BP table can be used to screen for abnormal BP values in the paediatric T1D population. If a BP value is found to be abnormal based on the simple BP table, it then requires further evaluation, including consultation of the standard BP tables (Figure 1). This approach may ultimately reduce the risk of HTN-related complications in T1D through the timely recognition and treatment of paediatric HTN.

Figure 1).

Algorithm for use of the simple blood pressure (BP) table to screen for abnormal BP values