To determine if elevated blood pressure (EBP) in hospitalized children accurately predicts EBP outpatient.
A multicenter retrospective chart review was conducted at a large hospital system in Northeastern United States. Mean blood pressures during hospitalizations were classified as elevated or not elevated, by using the American Academy of Pediatrics (AAP) 2017 parameters. Mean blood pressure was then compared with each patient’s mean blood pressure measured 3 times postdischarge. The data were analyzed to determine if inpatient EBP is an accurate predictor of outpatient EBP.
Of 5367 hospitalized children, 656 (12.2%) had EBP inpatient. Inpatient EBP was highly predictive of outpatient EBP, with a positive predictive value of 96% and negative predictive value of 98%.
Diagnosing hospitalized children with EBP, as defined by the AAP 2017 guidelines, accurately predicts true EBP outpatient.
In 2017, American Academy of Pediatrics (AAP) published, “Clinical Practice Guideline for Screening and Management of High Blood Pressure in Children and Adolescents”1 Classification and management of elevated blood pressure (EBP) and hypertension were reviewed. The new guidelines removed weight and height as a factor in determining blood pressure norms,2 hence simplifying the process (Table 1). This has led to increased diagnosis of pediatric EBP and, thus, improved early identification of children with cardiovascular risk factors.3
Screening BP Values Requiring Further Evaluation, Age, y . | BP, mmcHg . | |||
---|---|---|---|---|
Boys . | Girls . | |||
Systolic . | Diastolic . | Systolic . | Diastolic . | |
1 | 98 | 52 | 98 | 54 |
2 | 100 | 55 | 101 | 58 |
3 | 101 | 58 | 102 | 60 |
4 | 102 | 60 | 103 | 62 |
5 | 103 | 63 | 104 | 64 |
6 | 105 | 66 | 105 | 67 |
7 | 106 | 68 | 106 | 68 |
8 | 107 | 69 | 107 | 69 |
9 | 107 | 70 | 108 | 71 |
10 | 108 | 72 | 109 | 72 |
11 | 110 | 74 | 111 | 74 |
12 | 113 | 75 | 114 | 75 |
≥13 | 120 | 80 | 120 | 80 |
Screening BP Values Requiring Further Evaluation, Age, y . | BP, mmcHg . | |||
---|---|---|---|---|
Boys . | Girls . | |||
Systolic . | Diastolic . | Systolic . | Diastolic . | |
1 | 98 | 52 | 98 | 54 |
2 | 100 | 55 | 101 | 58 |
3 | 101 | 58 | 102 | 60 |
4 | 102 | 60 | 103 | 62 |
5 | 103 | 63 | 104 | 64 |
6 | 105 | 66 | 105 | 67 |
7 | 106 | 68 | 106 | 68 |
8 | 107 | 69 | 107 | 69 |
9 | 107 | 70 | 108 | 71 |
10 | 108 | 72 | 109 | 72 |
11 | 110 | 74 | 111 | 74 |
12 | 113 | 75 | 114 | 75 |
≥13 | 120 | 80 | 120 | 80 |
While providing extensive guidance for screening and management in the outpatient setting, the guideline does not address EBP observed during an inpatient stay, in the context of an acute illness, surgery, or chronic illness flare.
Studies on adults have revealed that inpatient hypertension is often a predictor of true hypertension. Axon et al4 found a 50% to 72% prevalence of hypertension in hospitalized adults, with a majority remaining hypertensive after discharge. Conen et al5 showed that adult patients with no history of hypertension can be accurately diagnosed with hypertension in an inpatient setting.
Prevalence of hypertension and EBPs in children has increased in recent years, currently at 14% and 3% to 4% respectively. It remains unknown whether EBP found in children admitted to the hospital is predictive of true EBP that persists after discharge.
The objective of this study was to explore whether using the 2017 guidelines in the inpatient setting would accurately predict EBP after hospitalization. We hypothesize that inpatient EBP is a positive indicator of true EBP (Table 1).
Methods
This was a cohort study, by using a retrospective multicenter observational chart review from November 2015 to October 2017. Automated data mining was used to collect data from Allscripts, an electronic medical record within a large health care system in the Northeastern United States. Patients 3 to 18 years of age who were discharged from the health system’s pediatric units and had at least 3 subsequent visits at the system’s outpatient clinics were included. Prestudy biostatistical analysis concluded that 200 participants with EBP would be required to power this analysis. Patients with any record of chronic kidney disease by International Classification of Diseases, 10th Revision, code were excluded. To maximize the size of our cohort, we did not limit the follow-up time to a specific period. We collected demographic data that has been previously found to influence risk of hypertension, including age, diagnosis, ethnicity and race (as per patient self-identification), sex, and length of stay.6–49
Systolic and diastolic blood pressures for the entire hospital stay were collected, and the mean of each (systolic and diastolic) was calculated separately. The initial triage blood pressure was excluded from the analysis, on the basis of previous studies revealing that this measurement, taken in children in the emergency department, is poorly predictive of true hypertension.50 The calculated mean systolic and mean diastolic pressures were then used to categorize each subject as having EBP or no EBP, as per the parameters set by the AAP 2017 guidelines.
The systolic and diastolic blood pressures recorded in the electronic medical record during the first 3 postdischarge ambulatory measurements were then collected, and means were calculated in the same manner. Each patient was classified as having EBP or no EBP, in the ambulatory setting. The authors manually reviewed the charts of the subset that were classified as EBP, to exclude those deemed to be implausible or erroneous, (for example, a systolic BP of 1100, assumed to be a typographical error). Statistical analysis was then conducted to determine if inpatient EBP was predictive of outpatient EBP.
Statistical Analysis
The data were described by using frequencies and percentages for categorical variables and means and SDs for quantitative variables. The McNemar χ2 test was used to compare patients classified as having EBP inpatient with their classification outpatient (EBP or no EBP). Agreement was measured for each subject. Additionally, validity indices such as sensitivity and specificity of inpatient EBP in comparison with outpatient EBP were computed. All statistical tests were 2-sided. P values <0.05 were considered significant. All statistical analyses were performed by using SAS software (SAS Institute, Inc, Cary, NC).
Results
A total of 11 699 patients were eligible for the study, of which 61 were excluded because of chronic kidney disease and 6213 were excluded because of having <3 follow-up visits after discharge. Another 58 were excluded manually because of missing or erroneous data, leaving 5367 subjects.
The average age was 11 years, with an SD of 5 years, ranging 3 to 18 years. Sex was 49% female and 51% male. A total of 40% of participants were White, 20% were Black or African American, and 40% were other or multiracial. The median length of stay was 2 days, with an interquartile range of 3 days.
In this cohort, 656 children (12.2%) had EBP according to the AAP criteria. On follow-up outpatient, 628 of the 656 (96%) continued to have EBP after discharge (Table 2). Only 28 (5.2%) had resolved their EBP on follow-up. Very few (80 [1.5%]) had EBP outpatient but normal BPs during their previous hospitalization.
. | With EBP Outpatient, n (%) . | Without EBP Outpatient, n (%) . | Total, n (%) . |
---|---|---|---|
With EBP inpatient | 628 (11.7) | 28 (5.2) | 656 (12.2) |
Without EBP inpatient | 80 (14.9) | 4631 (86.3) | 4711 (87.8) |
Total | 708 (13.2) | 4659 (86.8) | 5367 |
. | With EBP Outpatient, n (%) . | Without EBP Outpatient, n (%) . | Total, n (%) . |
---|---|---|---|
With EBP inpatient | 628 (11.7) | 28 (5.2) | 656 (12.2) |
Without EBP inpatient | 80 (14.9) | 4631 (86.3) | 4711 (87.8) |
Total | 708 (13.2) | 4659 (86.8) | 5367 |
Table 2 reveals the breakdown of patients with EBP inpatient that were found to have EBP on follow-up outpatient visits.
By using the AAP 2017 EBP definition as the reference, the sensitivity of the inpatient evaluation was 89% and specificity was 99%. The positive predictive value was 96%, and the negative predictive value was 98% (Table 3). By using a simple κ coefficient, the confidence interval was 0.882–0.916, revealing almost complete agreement between the inpatient and outpatient measurements. McNemar’s test also reveals almost complete agreement, with Pr>χ2 of 0.0006. By using binomial proportion, for our total sample of 5367 patients, the accuracy of using elevated inpatient blood pressures to predict outpatient EBP was 97%.
. | Reliability of Inpatient EBP to Predict Ambulatory EBP . | ||
---|---|---|---|
Estimate, % . | SE . | 95% Confidence Limits . | |
Sensitivity | 89 | 0.0119 | 0.8637–0.9103 |
Specificity | 99 | 0.0011 | 0.9918–0.9962 |
Positive predictive value | 96 | 0.0079 | 0.9418–0.9728 |
Negative predictive value | 98 | 0.0019 | 0.9793–0.9867 |
. | Reliability of Inpatient EBP to Predict Ambulatory EBP . | ||
---|---|---|---|
Estimate, % . | SE . | 95% Confidence Limits . | |
Sensitivity | 89 | 0.0119 | 0.8637–0.9103 |
Specificity | 99 | 0.0011 | 0.9918–0.9962 |
Positive predictive value | 96 | 0.0079 | 0.9418–0.9728 |
Negative predictive value | 98 | 0.0019 | 0.9793–0.9867 |
Discussion
In a multicenter large cohort of hospitalized children, our results reveal that diagnosis of EBP is highly accurate in the inpatient setting. This supports using the AAP 2017 guideline to classify hospitalized children as having EBP and guide intervention accordingly.
We considered several options to most accurately reflect the patients’ true inpatient blood pressure. Options included using the mean or median of all recorded BPs, of 3 random BPs, or doing so while excluding the highest and lowest. Given the paucity of clear precedent and after a literature review and consultation with experts, we chose to use the mean of all measurements, excluding the triage BP. We also collected the medians, which were found to be consistently similar to the means and, thus, were not analyzed separately.
The method of using the mean of 3 serial outpatient visits was chosen to increase the validity of the EBP classification. This however led to a limitation, in that the time period in which patients completed 3 follow-up measurements could not be specified and is presumed to be variable. Because serial measurements of BP over time is the standard of practice in the ambulatory setting,1,2 our method better reflected clinical norms.
Strengths of our study include a large cohort of children with EBP, multicenter participation and a diverse population. Use of multiple BP measurements likely provided a stronger screening test than any 1 measurement alone. A limitation of our study was that because EBP increases with age,51 our sample included few younger children with EBP, with only 19 (1.7%) of the subjects with EBP being <13 years of age. Additionally, because of data mining limitations, we were able to exclude patients with chronic kidney disease, but we could not exclude patients who had known previous hypertension. However, the significant positive predictive value of inpatient EBP remains clinically important in terms of its use in guiding follow-up planning and management.
Key Points
Question: Are the 2017 AAP guidelines for diagnosis of EBP in children reliable for use in the inpatient setting?
Findings: The AAP screening parameters for EBP have high sensitivity and specificity when used in the inpatient setting.
Meaning: Children aged 3 to 18 years, who are hospitalized in an acute care setting and found to have EBP according to the AAP 2017 guidelines, should be considered to have true EBP and managed accordingly.
Conclusions
Our study supports using the AAP 2017 guidelines for identifying EBP in children in the hospital setting and planning appropriate intervention and follow-up. Further study is needed to support this finding with a larger number of younger children.
On the basis of our findings and the AAP consensus guidelines,1 we suggest that if hospitalized children are found to have EBP, counseling, follow-up planning, and interventions should be initiated, as per item 4.3b in the guideline.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
Comments