Identification of hypertension in hospitalized children prescribed as‐needed antihypertensive medication

Abstract

Imperfect measurement conditions, transient blood pressure (BP) elevation due to pain or anxiety, and heavy clinical demands complicate hypertension (HTN) diagnosis in hospitalized children, and may prevent recognition of hypertensive episodes for children prescribed as‐needed (PRN) antihypertensive medication. The authors sought to describe the incidence and predictors of missed BP elevation among hospitalized children prescribed PRN antihypertensive medication at our hospital. BP data were retrospectively audited for children age 2‐18 admitted in 2018 to the general ward, and prescribed PRN oral nifedipine, intravenous [IV] or oral hydralazine, or IV labetalol. Appropriate recognition of BP elevation (exceeding the parameters in the medication order) was defined as administering medication within the ordered dosing interval, administering the medication earlier than planned, or physician documentation of why HTN treatment was withheld. Mixed‐effects logistic regression was used to identify factors associated with recognition of BP elevation. Fifty‐six hospitalizations including 616 BP measurements were analyzed. BP elevation was appropriately recognized in 230 (37%) instances, in most of which (n = 190) the antihypertensive medication was given after excessive BP was noted. On multivariable analysis, higher systolic BP and BP elevation occurring at night were associated with increased likelihood of appropriate recognition. BP elevations are frequently missed in hospitalized children prescribed PRN antihypertensive medication. Particularly, there was low recognition of diastolic BP elevation and of systolic BP elevation close to but still exceeding the ordered threshold. Further staff education may be needed to raise awareness of lower BP thresholds for HTN in younger and smaller children.

1. INTRODUCTION

Hypertension (HTN) in children is defined in relation to the normal blood pressure (BP) in healthy children and affects approximately 2%‐4% of US children aged 3‐18 years. ¹ , ² For patients age 1‐12 years, stage 1 HTN is defined as BP above the 95th percentile for a given age, sex, and height; and stage 2 HTN is defined as BP more than 12 mm Hg in excess of the 95th percentile. While these guidelines have been used to diagnose HTN in outpatient settings, documentation and diagnosis of HTN among hospitalized children are fraught with challenges. For example, among patients in a pediatric emergency department, 55% of children had elevated BP at triage, but only 1% of these cases were appropriately documented. ³ In a similar study of children admitted to a tertiary children’s hospital, HTN was often under‐recognized, especially for stage 1 as compared to stage 2 HTN (5% vs 26%). ⁴ Transient BP elevation due to pain or anxiety and imperfect measurement conditions (eg, patients lying supine while measurements are obtained in the lower extremities due to venous access devices in the upper extremities) complicate diagnosis of HTN in the hospital environment. As further evidence of HTN under‐recognition, administrative data relying on diagnosis codes suggest that HTN occurs in only 1% of all pediatric inpatients, ⁵ , ⁶ , ⁷ though studies analyzing BP measurement data have found that 19% of BP readings in pediatric critical care patients exceed the 95th percentile, ⁸ and up to 25% of children may experience elevated BP during a hospital stay. ²

To address HTN among hospitalized children, “as‐needed” (PRN) antihypertensives are commonplace in inpatient care. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ However, recognition of abnormal BP in hospitalized children can be hindered by the need to consult age, sex, and height‐based BP percentiles for children <13 years of age; distractions and heavy clinical demands; the asymptomatic nature of hypertensive urgency; and non‐specific symptoms of hypertensive emergency in younger, especially non‐verbal, children. For example, systolic blood pressure (SBP) as low as 116 mm Hg could be hypertensive in a short 12‐year‐old boy, while 127 mm Hg would not be considered hypertensive in a taller male child of the same age. In turn, failure to recognize hypertensive crises in hospitalized children can result in end‐organ damage or progression of certain chronic diseases. ¹ , ¹¹ , ¹⁶ , ¹⁷ At our institution, we observed that PRN antihypertensive medications were often held without documented explanation, despite ordered administration parameters. To quantify clinically significant BP elevations that were missed in hospitalized pediatric patients at our institution, we retrospectively reviewed charts of children prescribed PRN antihypertensive medication and evaluated whether medication was given when BP exceeded the ordered threshold. Our secondary aim was to identify which, if any, patient factors were associated with poorer recognition of hypertensive episodes.

2. METHODS

BP data were audited for the general and intermediate level pediatric inpatients at a tertiary care children’s hospital (James and Connie Maynard Children’s Hospital) serving 29 counties in eastern North Carolina. The study included children aged 2‐18 years who were admitted to the general inpatient pediatric units in 2018 (excluding surgical admissions), and prescribed PRN antihypertensive medication (oral nifedipine, intravenous [IV] or oral [PO] hydralazine, or IV labetalol). Other medications were excluded as they are not routinely used at our institution as a rescue medication for hypertension in hospitalized children. Medication type (including PO vs IV) was primarily determined by physician preference. For some patients, PO medication may be prescribed if there is no IV access already in place; and for other patients, IV medication may be prescribed if the patient could not tolerate PO intake. BP measures obtained in the pediatric intensive care unit (PICU), in association with an anesthesia/sedation event, and prior to ordering the PRN hypertensive or after discontinuation of this medication were excluded from the analysis.

At our institution, all children admitted to the general ward have BP checked every 4 hours regardless of the reason for admission. As per our standard clinical practice, BP measurements were obtained by the oscillometric method and were not confirmed with manual readings. Measures were preferentially obtained from the arm, if possible, and otherwise from the leg. On the units included in our analysis, the standard practice is for a care partner (nurse assistant) to track BPs in the patient’s chart and communicate abnormal BPs to a registered nurse (RN). RNs may also review documented vital signs in the electronic medical record (EMR) and communicate values outside the call parameters ordered by the physician to the resident on call. Since there are only 2 care partners assigned to these units, many RNs take vitals on their patients personally.

The authors analyzed all BP measurements meeting inclusion criteria for eligible patients hospitalized during the review period. Thresholds for PRN medication dosing were determined by each physician and differed from patient to patient. In most cases, thresholds specified both SBP and DBP values (eg, BP> 140/90), but in other cases, thresholds specified SBP values only (eg, SBP > 125) or MAP values only (eg, MAP > 90). Because our study focused on adherence to thresholds specified by the physician for medication administration, BP readings below ordered thresholds for PRN medication dosing were excluded from further analysis. The authors also excluded duplicate BP readings and BP readings taken while it was too soon to administer the next medication dose.

Among BP readings meeting or exceeding the ordered threshold, our primary outcome was appropriate recognition of elevated BP. This was defined as administering PRN antihypertensive medication within the ordered PRN dosing interval after BP was evaluated. Alternatively, recognition of elevated BP was deemed appropriate if the scheduled BP medication was administered earlier than planned, or if there was a contemporaneous physician note justifying withholding the treatment of HTN. In 2 cases where medication was refused by the family, we coded elevated BP as having been appropriately recognized. In addition to quantifying the proportion of BP readings where elevated BP was appropriately recognized, we sought to identify patient factors associated with poorer recognition of HTN. Patient characteristics collected from the EMR included age (dichotomized as 2‐12 vs 13‐18 years, reflecting the more complex determination of BP thresholds in the younger age‐group), sex, race/ethnicity, body mass index (BMI) percentile for age (classified as normal weight or underweight; overweight; or obese), and primary reason for hospitalization.

Patient characteristics at the time of admission were summarized using medians with interquartile range (IQR) or counts with percentages, as appropriate. Appropriate recognition of elevated BP was analyzed as a percentage of all BP readings meeting inclusion criteria, potentially including multiple readings per patient. Since the number of BP elevations could have been different across patients, we used mixed‐effects regression models to account for the clustering of BP elevations within the same patient, as well as the unequal number of such events across patients. These models were used to evaluate the association between patient characteristics and the likelihood of appropriate recognition of each elevated BP reading. ¹⁸ The model included a random intercept for hospitalization, to account for non‐independence of BP readings taken during the same hospital stay. The multivariable analysis excluded cases with missing data on study covariates. In a secondary analysis, we evaluated whether older age in the 2‐ to 12‐year‐old group was associated with increased likelihood of appropriate recognition of elevated BP, treating age as a continuous variable. The authors also evaluated whether HTN recognition varied according to factors specific to each BP measurement, that is, SBP, DBP, current type of PRN medication, and whether the measurement was taken at nighttime (after 7 pm and before 7 am). Data analysis was performed in Stata/SE 15.1 (College Station, TX), and P < .05 was considered statistically significant.

3. RESULTS

The authors identified 83 hospitalizations during which a PRN antihypertensive medication was ordered while patients were on the general pediatric ward, and screened 1063 BP measurements collected from these hospitalizations where BP exceeded the order thresholds. After removing duplicates, BP measurements in the PICU, perioperative BP measurements, and measurements taken while it was too early to administer the next dose of antihypertensive medication, we retained 616 BP measurements from 56 hospitalizations for analysis. Patient characteristics for the 56 hospitalizations are summarized in Table 1, and descriptive statistics for the 616 BP measurements are presented in Table 2.

Table 1.

Patient characteristics for hospitalizations included in the study (N = 56 hospitalizations)

Variable	Median (IQR) or N (%)
Age (y)	12 (7, 15)
Sex
Male	38 (68%)
Female	18 (32%)
Race/ethnicity
Non‐Hispanic black	25 (45%)
Non‐Hispanic white	17 (30%)
Other	14 (25%)
BMI percentile for age a
Normal weight/ underweight	18 (34%)
Overweight	11 (21%)
Obese	24 (45%)
Antihypertensive medications prescribed while on general pediatric ward b
Nifedipine	41 (73%)
Hydralazine	26 (46%)
Labetalol	4 (7%)
Primary reason for hospitalization
Renal disease	24 (43%)
Hypertension	8 (14%)
Malignancy	8 (14%)
Infection or fever	7 (13%)
Other	9 (16%)

Abbreviations: BMI, body mass index; IQR, interquartile range.

^aData missing for 3 hospitalizations.

^bCategories are not mutually exclusive.

Table 2.

Blood pressure measurements included in the study (N = 616 observations)

Variable	Median (IQR) or N (%)
BP (mm Hg)
SBP	129 (122, 141)
DBP	82 (73, 89)
MAP	170 (160, 182)
HTN recognition	230 (37%)
Medication given after BP measurement	190 (31%)
Medication given early	32 (5%)
Documented why medication was not given	8 (1%)
Time of BP measurement
Daytime (7 am‐7 pm)	449 (73%)
Nighttime (7 pm‐7 am)	167 (27%)
Antihypertensive medication
Nifedipine	420 (68%)
Hydralazine	127 (21%)
Labetalol	40 (7%)
Hydralazine + labetalol	28 (5%)

Abbreviations: BP, blood pressure; DBP, diastolic blood pressure; HTN, hypertension; IQR, interquartile range; MAP, median arterial pressure; and SBP, systolic blood pressure.

The most common reason for hospitalization were related to renal disease (43%), with 14% of hospitalizations related to hypertension, 14% related to malignancy, 13% related to infection or fever, and 16% due to other reasons. Patients were primarily prescribed nifedipine (73%), followed by hydralazine (46%) and labetalol (7%). Overall, HTN was appropriately recognized in 230 (37%) instances, where the antihypertensive medication was given after the excessive BP was noted (n = 190), the antihypertensive medication had already been given ahead of schedule (n = 32), or the physician documented in a note that HTN was recognized but medication was not given (n = 8).

Twenty‐nine (52%) patients were age <13 years at admission, contributing 395 observations to the descriptive analysis (2 patients in this age‐group had missing BMI data, and 4 BP observations from these patients were excluded from multivariable analysis). On descriptive analysis, 30% of elevated BP measurements were recognized when patients were <13 years of age, compared to 50% of elevated BP measurements when patients were 13 years of age or older. However, on multivariable analysis (Table 3), there was no statistically significant association between patient age ≥13 years and the odds of elevated BP being appropriately recognized in the overall cohort (odds ratio [OR] = 1.4; 95% confidence interval [CI]: 0.5, 4.2; P = .523). Continuous age was also not associated with the odds of HTN recognition when analyzing the subset of patients <13 years of age (OR per each year of age = 1.3; 95% CI: 0.9, 1.9; P = .127).

Table 3.

Mixed‐effects logistic regression of hypertension recognition in hospitalized children, according to patient characteristics at admission

Variable	Children aged 2‐18 y (N = 603 observations)		Children aged 2‐12 y (N = 391 observations)
	OR (95% CI)	P	OR (95% CI)	P
Age < 13 y	1.4 (0.5, 4.2)	.523
Continuous age (y)			1.3 (0.9, 1.9)	.127
Sex
Male	Ref.		Ref.
Female	0.6 (0.2, 1.7)	.366	0.5 (0.1, 2.8)	.432
Race/ethnicity
Non‐Hispanic black	Ref.		Ref.
Non‐Hispanic white	0.4 (0.1, 1.5)	.181	0.1 (0.01, 0.4)	.003
Other	0.6 (0.2, 1.8)	.328	0.3 (0.05, 1.4)	.116
BMI percentile for age
Normal weight/underweight	Ref.		Ref.
Overweight	0.4 (0.1, 1.3)	.121	0.9 (0.3, 3.2)	.880
Obese	0.8 (0.3, 2.0)	.581	0.3 (0.1, 1.7)	.185
Primary reason for hospitalization
Renal disease	Ref.		Ref.
Hypertension	2.0 (0.6, 6.8)	.288	9.2 (1.6, 54.1)	.014
Malignancy	3.8 (0.9, 16.7)	.078	2.6 (0.2, 31.1)	.446
Infection or fever	1.4 (0.4, 5.6)	.596	3.6 (1.0, 13.4)	.056
Other	1.7 (0.6, 4.8)	.329	4.5 (1.0, 20.9)	.056

Abbreviations: BMI, body mass index; CI, confidence interval; OR, odds ratio.

No patient characteristics at the time of admission were associated with HTN recognition in the overall cohort. Among patients age <13 years, HTN recognition was more likely among patients admitted for HTN than patients admitted due to renal disease (OR = 9.2; 95% CI: 1.6, 54.1; P = .014). Additionally, HTN recognition was less likely among non‐Hispanic white as compared to non‐Hispanic black patients (OR = 0.1; 95% CI: 0.01, 0.4; P = .003). Other patient characteristics, including sex and BMI group, were not associated with HTN recognition in this subgroup.

In further multivariable analysis, we examined the association between HTN recognition and time‐varying characteristics. In this model (Table 4), higher SBP was associated with increased likelihood of HTN recognition (OR per 1 mm Hg increase in SBP = 1.1; 95% CI: 1.1, 1.2; P < .001), but higher DBP was not associated with this outcome. (MAP was not included in the model since SBP and DBP determine the MAP.) When BP exceeded order thresholds at nighttime, this elevation was almost twice as likely to be recognized as HTN measured during the day (OR = 1.9; 95% CI: 1.2, 3.0; P = .008). This analysis revealed no differences in likelihood of HTN recognition according to the type of PRN antihypertensive medication that was prescribed at the time of the BP measurement.

Table 4.

Mixed‐effects logistic regression of hypertension recognition in hospitalized children, according to time‐varying characteristics at each blood pressure measurement (N = 615 observations)

Variable	OR (95% CI)	P
BP (mm Hg)
SBP	1.1 (1.1, 1.2)	<.001
DBP	1.0 (1.0, 1.0)	.780
Time of BP measurement
Daytime (7 am‐7 pm)	Ref.
Nighttime (7 pm ‐7 am)	1.9 (1.2, 3.0)	.008
Antihypertensive medication
Nifedipine	Ref.
Hydralazine	1.5 (0.7, 3.2)	.291
Labetalol	3.2 (0.7, 14.3)	.134
Hydralazine + labetalol	0.4 (0.1, 2.5)	.314

Abbreviations: BP, blood pressure; DBP, diastolic blood pressure; HTN, hypertension; IQR, interquartile range; MAP, median arterial pressure; SBP, systolic blood pressure.

4. DISCUSSION

Elevated BP in pediatric patients is frequently under‐recognized in the outpatient setting, ⁷ , ¹⁹ and may be similarly under‐recognized in the hospital setting. ³ , ⁴ Our study found that among children and adolescents admitted to the general ward, only 37% of BP elevation episodes were recognized as such among children who had standing orders for PRN antihypertensive medications, with patient‐specific BP thresholds stated in the orders. Since the definitions for HTN in younger children vary based on age, sex, and height, we expected to find that recognition would be improved among adolescents (for whom standards were recently adapted to match adult standards). Nevertheless, we found that neither age nor other patient characteristics affected the likelihood of recognizing BP elevation in the overall sample. While reason for admission and patient race/ethnicity were correlated with BP elevation in children under 13 years of age, the wide confidence intervals around these estimates call into question the reliability of these observations about differences in HTN recognition according to patient characteristics.

Characteristics of each BP measurement were more consistently predictive of appropriately recognizing BP elevation. Similarly to Brady and colleagues’ results from an outpatient cohort, ¹⁹ we found that more modest elevations in BP were less likely to be recognized, especially elevations of diastolic BP. The authors hypothesize that this may be due to poorer familiarity with diastolic thresholds for HTN. In any case, the low rate of documented recognition of BP elevations among children prescribed PRN antihypertensives raises concerns about potential risk of under‐treating HTN and avoidable patient harm that may result. Enhancing staff education and implementing decision support tools in the EMR that link vital signs to medication administration records are two possible avenues to improving documented recognition and appropriate action addressing BP elevations in hospitalized children.

A prior analysis of children seen in outpatient clinics found multiple patient factors to be associated with under‐recognition of elevated BP, including sex and body weight. ¹⁹ In our inpatient cohort, however, we did not find recognition of elevated BP to be different among obese children or to differ based on sex. In our study, patient factors may have had a limited impact on recognition of BP elevation because the index of suspicion for these patients having BP pressure should have been universally high, given the presence of a standing order for a PRN antihypertensive medication. However, our results were consistent with the findings from the outpatient setting insofar as modest elevations of BP were less likely to be recognized, particularly when diastolic BP was elevated. In both inpatient and outpatient settings, familiarity with pediatric BP thresholds among providers may be lacking, and would be an appropriate target for further training and educational efforts. Uniquely to the inpatient setting, we found a statistically significant improvement in RN recognition of elevated BP during the night shift. This could be influenced by several factors not specifically examined in our study, such as fewer distractions during the night shift, higher experience among staff on this shift, or improved communication between nurses and residents at night, since the unit where we conducted our study had only 2 residents responsible for all non‐surgical patients admitted at nighttime. It is also possible that when patients are asleep in the nighttime, staff are less likely to attribute elevated BPs to transient elevations caused by pain or anxiety.

Initiatives to improve recognition of HTN in the outpatient setting have included improved clinical decision support built into the EMR, and use of screening tables to provide more concise age‐ and sex‐based norms for pediatric BP, thus limiting the need to consult the more comprehensive tables that include height‐based thresholds. Both approaches have promise based on published studies, but even with the effective use of these tools, elevated BP remains commonly under‐recognized. ²⁰ , ²¹ , ²² , ²³ In the inpatient setting, an EMR‐based intervention, especially one that calculates patient‐specific thresholds for appropriate BP based on admission height, age, and sex, could effectively warn staff of a BP elevation at the moment recognition is needed (when inputting a BP into the patient’s vitals record). BP data could also be automatically cross‐referenced with the medication administration record in the EMR to determine when another dose of PRN antihypertensive medication may be appropriate. Concise reference tables (such as the high negative predictive value table from the 2017 AAP guideline) ¹ could also be effective in improving recognition of elevated BP, but, as seen in our study, even having a single patient‐specific threshold does not prevent under‐recognition of elevated BP . Further work is needed to elucidate the most effective methods for improving BP recognition in both outpatient and inpatient settings.

Although our study presents novel data on recognition of BP elevation among children and adolescents admitted to the general ward with standing orders for PRN antihypertensives, several limitations of the data and analytic approach constrain the conclusions that can be drawn. Despite our best efforts to account for recognition of BP, our study may underestimate recognition if it occurred but was not documented in the chart. There are many reasons for inpatients to have transient elevations of BP that may not be clinically significant, ² and staff members may consciously or unconsciously note these when disregarding an elevated BP measurement, without documenting the reason for withholding PRN antihypertensive medication in the chart. Also, because our study focused on the BP thresholds entered when the antihypertensive medication ordered was placed, we did not analyze recognition specifically based on patient height and height‐specific thresholds. Since subtle BP elevations were less likely to be recognized, recognition may have been particularly limited for elevated BPs in shorter patients, for whom HTN thresholds are lower. Because BP measurements were obtained by nurse assistants in some cases and RNs in other cases, there may have been delayed recognition of BP elevations when less highly trained staff obtained the data, although we did not examine this factor. While outpatient data suggest that increased training and experience enhances recognition of elevated BP, ¹⁹ it is unclear how these data from physicians apply to our study of BP documentation by nursing staff. Lastly, our study was intended to focus on children in whom BP elevations were expected (considering the PRN medication order) and likely to be closely monitored. Our estimates may not reflect hypertension recognition rates at institutions with other practices related to BP monitoring for pediatric inpatients, although this requires confirmation by further studies.

While pediatric HTN is becoming increasingly prevalent, elevated BP in pediatric patients is poorly recognized. Given the universal nature of BP measurement in pediatric inpatients as part of routine vital sign monitoring, the inpatient setting could provide a valuable opportunity to capture patients with previously undiagnosed BP elevations. Furthermore, timely and effective recognition of hypertensive crises among hospitalized children is essential to provide appropriate treatment and avoid patient harm. Our study reveals poor documented recognition of BP elevations among hospitalized children and adolescents with PRN antihypertensive medication orders. The magnitude of BP elevation and time of assessment, rather than patient factors, appear to influence the likelihood of BP recognition in the inpatient setting. Prospective studies and quality improvement efforts are needed to identify the most effective methods for increasing recognition of abnormal BP in pediatric inpatients to facilitate appropriate treatment while inpatient, and to identify patients in need of further evaluation and management during outpatient follow‐up.

CONFLICT OF INTEREST

None declared for all authors.

AUTHOR CONTRIBUTIONS

All authors participated in study conception and design. KRK performed data collection. DT performed data analysis. All authors participated in interpretation of the results. KRK and DT drafted the manuscript. AGL and LMGM critically revised the manuscript. All authors reviewed and approved the final manuscript.

Kocher KR, Tumin D, Lehmann AG, Gomez Mendez LM. Identification of hypertension in hospitalized children prescribed as‐needed antihypertensive medication. J Clin Hypertens. 2020;22:1452–1457. 10.1111/jch.13950