Does hypertension at initial presentation adversely affect outcomes in pediatric traumatic brain injury?^,

Abstract

Background:

Adults with traumatic brain injury (TBI) who present hypertensive suffer worse outcomes and increased mortality compared to normotensive patients. The purpose of this study is to determine if age-adjusted hypertension on presentation is associated with worsened outcomes in pediatric TBI.

Methods:

A retrospective chart review was conducted on pediatric patients with severe TBI admitted to a single system pediatric tertiary care center. The primary outcome was mortality. Secondary outcomes included length of stay, need for neurosurgical intervention, duration of mechanical ventilation, and the need for inpatient rehabilitation.

Results:

Of 150 patients, 70% were hypertensive and 30% were normotensive on presentation. Comparing both groups, no statistically significant differences were noted in mortality (13.3% for both groups), need for neurosurgical intervention (51.4% vs 48.8%, p = 0.776), length of stay (6 vs 8 days, p = 0.732), duration of mechanical ventilation (2 vs 3 days, p = 0.912), or inpatient rehabilitation rates (48.6% vs 48.9%, p = 0.972). In comparing just the hypertensive patients, there was a trend toward increased mortality in the 95th and 99th percentile groups at 15.8% and 14.1%, versus the 90th percentile group at 6.7% but the difference was not statistically significant (p = 0.701).

Conclusions:

Contrary to the adult literature, pediatric patients with severe TBI and hypertension on presentation do not appear to have worsened outcomes compared to those who are normotensive. However, a trend toward increased mortality did exist at extremes of age adjusted hypertension. Larger scale studies are needed to validate these findings.

Traumatic brain injury (TBI) is a significant health burden in the pediatric population. In 2013, 640,000 children age 14 years and under had emergency department visits related to traumatic brain injury and 1,500 associated deaths. TBI is classified as mild with a Glasgow Coma Scale (GCS) of 15–13, moderate a GCS 12–9, and severe a GCS of less than or equal to 8. Of children with moderate to severe TBI, greater than 61% experience an associated disability [1].

Management of TBI largely consists of preventing secondary injury. Hypotension following TBI increases mortality and worsens outcomes [2–5]. Adult TBI management guidelines recommend prevention of secondary insult by avoiding hypotension, defined as a systolic blood pressure (SBP) of ≪90 mmHg [6,7]. The deleterious effects of hypotension in pediatric TBI are also well documented [8–11]. The Guidelines for the Acute Medical Management of Severe Traumatic Brain Injury in Infants, Children, and Adolescents-Third Edition, propose maintaining a cerebral perfusion pressure of ≫40 as level III evidence, without defining a specific SBP goal [12]. Different from adult guidelines, defining hypotension as a SBP ≪5 percentile for age better predicts poor outcomes as compared to a set systolic threshold of ≪90 mmHg [10].

Conversely, less is known about the associations of hypertension with outcomes in TBI. Hypertension following traumatic brain injury can be a normal physiologic response to increased intracranial pressure, as the mean arterial pressure increases to preserve cerebral perfusion pressure as part of Cushing reflex. However, emerging evidence from adults with TBI suggests an association between hypertension, either pre-hospital or on hospital presentation, and increased mortality and worsened outcomes [2–4,13–15]. These effects are attributed to breakdown of the blood brain barrier, loss of cerebral blood flow autoregulation, and development of vasogenic edema in TBI patients [16–19]. Limited data are available regarding pediatric TBI and the effects of hypertension on outcome. While some studies show an increase in morbidity and mortality associated with hypertension in pediatric TBI [20–22], other studies show no difference [11] or demonstrate a protective effect of hypertension [23].

This current study aims to determine associations between age adjusted systolic hypertension and outcomes in pediatric TBI. We hypothesized that pediatric patients with severe TBI and hypertension on initial presentation would have both increased mortality and worsened outcomes compared to those who were normotensive on presentation.

1. Materials and Methods

We conducted a retrospective observational cohort study on pediatric patients less than or equal to 18 years of age presenting to a single institution pediatric tertiary care center, from January 2010 to December 2015, with a diagnosis of severe TBI. We queried our institution’s trauma registry for patients with a diagnosis TBI and a Glasgow Coma Scale (GCS) of 8 or less, defining severe TBI, on presentation to our facility. The Institutional Review Board approved our study (IRB# 16–193).

Data analyzed included: initial systolic blood pressure available in the patient chart, location of initial presentation, GCS on presentation to our facility, injury severity score (ISS) [24], trauma and injury severity score (TRISS) [25,26], hospital length of stay, need for neurosurgical intervention, need for and duration of mechanical ventilation, need for in-patient rehabilitation, and mortality. We defined neurosurgical intervention as placement of an intracranial pressure monitoring device (bolt or external ventricular drain (EVD)), decompressive craniectomy, evacuation of intracranial hemorrhage, or other surgical intervention. The primary outcome of interest was mortality. Secondary outcomes included hospital length of stay, need for neurosurgical intervention, need for and duration of mechanical ventilation, need for inpatient rehabilitation.

Patients were excluded from analysis if their initial SBP on presentation was less than the 5th percentile for age based on Pediatric Acute Life Support and American Heart Association guidelines or if they received cardiopulmonary resuscitation prior (CPR) to admission or at presentation to our facility [27]. If any of the trauma team physicians noted the patient’s GCS to be over 8 on arrival, they were excluded.

For patients who met inclusion criteria, we obtained and classified the initial SBP recorded in the patient chart as either normotensive or hypertensive. SBP measurements were used as the blood pressure metric for this study, in keeping with the available adult and pediatric literature regarding effects of blood pressure on outcomes in TBI which use SBP [11,13,20–23]. We obtained blood pressure recordings from either the emergency medical service (EMS) records, referral hospital emergency department records, or on presentation to our institution. We only included patients transferred from a referring facility if their blood pressure at the referring hospital or from emergency medical services in route to the referring hospital was available in the medical record, as too many interventions affecting blood pressure may have occurred in transport from the referring facility to ours. According to National Heart, Lung, Blood Institute guidelines for pediatric blood pressure measurements, a SBP ≪90th percentile for age, gender and height is normotensive. SBP in the 90th to 95th percentile is pre-hypertensive and requires follow-up, and ≫95th percentile is hypertensive [28]. For the purposes of our study, we included pre-hypertensive (age adjusted SBP 90th to 95th percentile) and hypertensive (age adjusted SBP ≫ 95th percentile) patients into the hypertensive category. We further subclassified hypertension as greater than the 90th, 95th or 99th percentile by the same guidelines. If the patient’s height was not available in the chart, they were only classified as hypertensive if their blood pressure would be hypertensive for an age and gender matched peer of any height. For children less than 1 year of age, hypertension is not clearly defined in the literature; therefore, based on available data we used a SBP greater than or equal to 100 as our threshold [29,30]. However, if a patient under 1 year of age had a SBP greater than or equal to the 95th or 99th percentile for a gender matched 1 year old, their blood pressure was classified by that corresponding percentile.

We performed a subgroup analysis exclusively on patients who initially presented to our institution, to account for any effects on outcome that may have occurred as a result of interventions performed in transport between a referral facility and arrival to our institution.

The primary analysis compared hypertensive to normotensive patients with severe TBI for primary and secondary outcomes. Descriptive statistics were calculated for all variables of interest and included counts and percentages for categorical variables and medians and interquartile ranges for continuous variables. Demographic and clinical characteristics and outcomes were compared across subgroups (e.g., normal vs. hypertensive) using chi-square tests for categorical variables and Wilcoxon rank sum tests for continuous variables. For categorical comparisons with small event counts, Fisher’s exact test was used in substitute of the chi-square tests. For continuous measures, where more than two groups were being compared (e.g., hypertensive range groups), a Kruskal-Wallis test was used instead of a Wilcoxon rank sum test. Analysis were conducted using SAS v. 9.4 (Cary, NC) and statistical significance was assessed at the 0.05 level.

2. Results

Of the 312 patients with TBI identified, 150 met inclusion criteria. Eighty-two patients were excluded from analysis based on hypotension or receiving CPR, 53 patients were excluded because of incomplete data (referral hospital or referral EMS BP not available or missing other data), and 27 patients were excluded secondary to a GCS of ≫8 documented by a member of the trauma team (Fig. 1).

Fig. 1.

Flow diagram of patient selection for analysis.

Of the 150 patients analyzed, 45 (30%) patients were normotensive for age and 105 (70%) were hypertensive for age, based on age adjusted systolic blood pressures (AASBP). Of the normotensive group, 66.6% (n = 30) had a systolic blood pressure greater than or equal to the 50th percentile for age. None of the hypertensive patients were on vasopressors at the time of their initial blood pressure measurement. The baseline characteristics of the normotensive and hypertensive groups were similar. Both groups had a median GCS of 3 at time of presentation to our facility. The two groups had statistically similar injury severity scores, need for and duration of mechanical ventilation, hospital length of stay, and need for inpatient rehab. The rate of mortality was identical in both groups at 13.3% (Table 1).

Table 1.

Comparison of normotensive vs hypertensive patients.

	Normotensive	Hypertensive	p-Value
	n = 45	n = 105
Sex, n (%)			0.853
Male	29 (66.4)	66 (62.9)
Initial presentation to our institution, n (%)	26 (57.8)	66 (62.9)	0.558
Admit Age - yearsa	6.1 (1.6–12.9)	4.5 (2.2–11.4)	0.862
GCS at presentation to our institutiona	3.0 (3.0–6.0)	3.0 (3.0–6.0)	0.560
ISS scorea	18.0 (13.0–25.0)	20.0 (10.0–25.0)	0.998
TRISS scorea,b	0.7 (0.5–0.9)	0.8 (0.5–0.9)	0.521
Required neurosurgical intervention, n (%)	22 (48.9)	54 (51.4)	0.776
Length of stay (days)a	8.0 (4.0–13.0)	6.0 (3.0–10.0)	0.732
Required mechanical ventilation, n (%)	45 (100)	103 (98.1)	0.351
Duration of mechanical ventilation (days)a,c	3.0 (1.0–7.0)	2.0 (1.0–7.5)	0.912
Discharge disposition, n (%)
Home	17 (37.8)	40 (38.1)	0.971
Rehabilitation	22 (48.9)	51 (48.6)	0.972
Death	6 (13.3)	14 (13.3)	1.000

^a, Median (IQR 25th–75th);

^b, n = 149 because one patient missing TRISS score;

^c, n = 149 excludes patient who received tracheostomy.

Table 2 demonstrates the hypertensive patients subdivided into 90th, 95th and 99th percentiles for age adjusted hypertension. Demographics were similar between the three groups. However, comparing the need for neurosurgical intervention, the 99th percentile group required intervention 59.2% of the time, 95th percentile group 26.3% of the time, and 90th percentile group 46.7% (p = 0.036). Also, there was a trend toward increased mortality in the 95th and 99th percentile groups at 15.8% and 14.1%, versus the 90th percentile group at 6.7% but the difference was not statistically significant (p = 0.701). Otherwise, outcomes were similar between the three groups.

Table 2.

Comparison of hypertensive patients by age adjusted systolic blood pressure percentiles.

	90th Percentile	95th Percentile	99th Percentile	p-Value
	n = 15	n = 19	n = 71
Sex, n (%)				0.325
Male	12 (80.0)	11 (57.9)	43 (60.6)
Admit age - yearsa	5.0 (0.4–10.6)	3.0 (2.2–7.9)	4.6 (2.4–12.8)	0.217
GCS at presentation to our institutiona	5.0 (3.0–6.0)	5.0 (3.0–6.0)	3.0 (3.0–7.0)	0.748
ISS scorea	20.0 (17.0–25.0)	16.0 (10.0–25.0)	20.0 (9.0–25.0)	0.433
TRISS scorea, b	0.7 (0.6–0.8)	0.8 (0.6–0.9)	0.7 (0.5–0.9)	0.139
Required neurosurgical intervention, n (%)	7 (46.7)	5 (26.3)	42 (59.2)	0.036
Length of stay (days)a	7.0 (5.0–10.0)	6.0 (4.0–10.0)	6.0 (3.0–12.0)	0.900
Required mechanical ventilation, n (%)	14 (93.3)	18 (94.7)	71 (100)	0.114
Duration of mechanical ventilation (days)a, c	1.5 (0.0–5.0)	2.0 (1.0–8.0)	3.0 (1.0–9.0)	0.111
Discharge disposition, n (%)
Home	8 (53.3)	8 (42.1)	24 (33.8)	0.339
Rehabilitation	6 (40.0)	8 (42.1)	37 (52.1)	0.573
Death	1 (6.7)	3 (15.8)	10 (14.1)	0.701

^a, Median (IQR 25th–75th);

^b, n = 104 because one patient missing TRISS score;

^c, n = 104 excludes patient who received tracheostomy.

As a means of controlling for interventions done in transfer from a referral facility, we then analyzed only the patients who initially presented to our institution. These patients were also classified as normotensive or hypertensive based on their initial blood pressure. We did not detect a difference in demographics or in outcomes between the normotensive and hypertensive groups (Table 3).

Table 3.

Comparison of normotensive and hypertensive patients who initially presented to our institution.

	Normotensive	Hypertensive	p-value
	n = 26	n = 66
Sex			0.211
Male, n (%)	19 (73.1)	39 (59.1)
Admit Age - yearsa	5.3 (2.3–10.4)	6.1 (2.5–12.0)	0.862
GCS at presentation to our institutiona	5.5 (3.0–6.0)	5.0 (3.0–7.0)	0.560
ISS scorea	18.5 (13.0–25.0)	20.0 (10.0–25.0)	0.998
TRISS scorea	0.7 (0.6–0.9)	0.8 (0.6–0.9)	0.521
Required neurosurgical Intervention, n	15 (57.7)	36 (54.6)	0.785
(%)
Length of stay (days)a	7.5 (4.0–13.0)	6.0 (4.0–10.0)	0.732
Required mechanical ventilation, n (%)	26 (100)	64 (97)	0.370
Duration of mechanical ventilation (days)a, b	3.5 (1.0–8.0)	3.0 (1.0–8.0)	0.914
Discharge disposition, n (%)
Home	8 (30.8)	24 (36.4)	0.612
Rehabilitation	15 (57.7)	36 (54.6)	0.785
Death	3 (11.5)	6 (9.1)	0.720

^a, Median (IQR 25th–75th);

^b, n = 91 excludes patient who received tracheostomy.

3. Discussion

Traumatic brain injury remains a major health problem in both pediatric and adult populations. Unfortunately, very little literature relates initial patient characteristics to prognosis in pediatric TBI. Our data indicates hypertension on initial presentation is not associated with worsened morbidity or mortality compared to normotension. However, at extremes of hypertension, ≫95th percentile for AASBP, we observed a trend toward increased mortality.

Adult data shows a U-shaped relationship between blood pressure and morbidity and mortality, with both hypotension and hypertension being risk factors for worsened outcomes [2–4,13–15]. Krishnamoorthy et al. performed a review of current literature regarding initial hypertension and the effect on outcomes in TBI, citing six adult studies where TBI patients with pre-hospital hypertension or hypertension on admission had worsened morbidity and mortality [13]. One study reviewed 315,242 patients from the National Trauma Data Bank and looked at outcomes based on emergency medical service systolic blood pressure measurements. They found that an EMS SBP of 150 mmHg or greater was associated with increased odds of mortality [14]. Sellman et al., also found prehospital hypertension greater than 160 mmHg to be associated with increased mortality [15]. Still, no guidelines exist for management of hypertension in TBI or SBP thresholds for harm in either the adult or pediatric populations.

The association between systolic blood pressure and outcomes in pediatric TBI is not well described. Similar to adults, hypotension in pediatric TBI is associated with increased morbidity and mortality [9,10]. However, a clear relationship between hypertension and outcomes does not exist. Three pediatric studies have shown a trend toward worsened outcomes in pediatric TBI patients with hypertension [20–22]. Vavilala et al. showed a trend toward increased mortality in pediatric TBI patients with a SBP greater than the 95th percentile for age [20]. Kanter et al. also demonstrated poor outcomes, defined as mortality or severe impairment of independent functioning, in pediatric acute brain injury patients whose blood pressure was more than 20 Torr above the 95th percentile for age at any point during their ICU course [21]. Johnson et al. showed pediatric military trauma patients with AASBP ≫99th percentile had increased 24 h mortality compared to those who were normotensive [22]. Conversely, White et al. showed, in a retrospective study of 136 children with TBI, increased odds of survival in children who had a maximum SBP ≥135 mmHg [23]. More recently, a study with a larger sample size found no difference in mortality between hypertensive and normotensive patients [11]. The study used the National Trauma Data Bank to analyze 10,473 children with severe TBI. They found that mortality was highest in patients with an initial blood pressure less than the fifth percentile for age with an adjusted relative risk of mortality of 3.2, and the relative risk of mortality remained increased in patients with a SBP ≪75 percentile. However, a blood pressure greater or equal to the 95th percentile did not have an increased relative risk of mortality, but rather a similar relative risk to the normotensive group (relative risk of 1.1 vs 1.0).

Our study focused specifically on comparing hypertensive and normotensive pediatric patients with severe TBI, and it did not show a difference in mortality between the two groups. These findings support the results published by Suttipongkaset et al. [11], and could potentially be explained by differences in physiology between adult and pediatric populations. The absence of increased morbidity and mortality in our hypertensive pediatric population could be accounted for by the Cushing reflex alone, without breakdown of the blood brain barrier or vasogenic edema and therefore no secondary deleterious effect. Alternatively, healthy adults and children have a similar ability to regulate their cerebral blood flow based on autoregulation index studies [31], and cerebral autoregulation is known to be impaired following TBI [32–34]. One possible difference in physiology to explain this disparity in outcomes could be that pediatric patients experience average mean arterial pressures further away from their upper limit of autoregulation. Pediatric patients are known to have a similar lower limit of autoregulation to adults, but upper limits are not well defined [35]. One study showed the upper limit of autoregulation increased in neonates as post-conceptual age increased [36]. If upper limits of autoregulation in children approximate that of adults, then children could have a higher upper limit reserve compared to adults; meaning their baseline mean arterial pressure is not as close to the upper limit of autoregulation as that of adults. However, when stratifying and comparing the hypertensive groups, there was a trend toward increased mortality in the age adjusted systolic blood pressure groups ≫95% compared to those ≫90%, similar to the findings by Johnson et al. [22] This supports that pediatric patients could have a larger upper limit reserve of cerebral autoregulation and only at the extremes of hypertension or in combination with increased severity of injury do we see an impact on outcomes. Another potential theory to explain differences in mortality would be associated co-morbidities. Most pediatric patients do not have underlying vascular disease or co-morbidities such as hypertension or diabetes mellitus, known to affect vascular integrity and possibly potentiate loss of cerebral autoregulation. These findings are important as they are in contrast to adult literature, and many adult studies are extrapolated to the care of children due to a lack of evidence in the pediatric population.

Majority of the literature regarding the effects of blood pressure on outcomes in pediatric TBI include hypotensive patients in their cohort and analysis [11,20,22,23]. Our study is unique in exclusively comparing outcomes in hypertensive and normotensive pediatric TBI patients. Another study that exclusively compared hypertensive and normotensive patients included non-trauma patients in its cohort [21]. Our study is of modest sample size, and patient blood pressures were stratified by their specific age and gender. Our hospital implemented a protocol to ensure similar treatment of all patients with TBI, in 2011, reducing differences in individual patients’ care. Reisner et al. published this protocol, showing a decrease in mortality of patients with severe TBI following its implementation [37].

The study is limited by its retrospective nature. Additionally, it is a single institution study of modest sample size, limiting the scope of its applicability. A single vital sign parameter was used as a predictor of mortality, while TBI patients are often complex and can have multi-organ systems affected. Initial blood pressure measurements used in the study were the first measurements available in the chart, allowing for differences in location (upper or lower extremity and dexterity) and timing of measurements. We used exclusion criteria of blood pressure less than the 5th percentile for age to define hypotension, as this criterion is shown to confer the greatest risk for morbidity and mortality in regards to the effect of blood pressure on outcomes for pediatric TBI patients [10,11,20]. However, increased morbidity and mortality exists below an age adjusted systolic blood pressure less than the 75th percentile, which was not accounted for in our study or selection criteria [11,20]. Patients’ GCS was assessed on arrival to our institution, but74.7% (n = 112) were intubated on arrival which affects the initial GCS. Patients may have received sedation, analgesia, or paralytic prior to arrival which could have affected their GCS.

Given the limited data available regarding the effects of hypertension on mortality in pediatric TBI, a power analysis was not possible at the beginning of the study (due to the number of assumptions with regards to incidence and outcome). With the preliminary findings uncovered in this analysis, however, we suggest a larger, multi-center analysis could be powered to offer a prediction model for mortality with sufficient power and statistical strength. This would decrease the chances of type II statistical error. In retrospect, if assuming the underlying morality rate in pediatric TBI patients is 10% in those without HTN and 25% in those with HTN, and 60% of patients would present with HTN, our study would have required n = 204 patients to achieve 80% power to detect a 15% difference in mortality. Given our sample size of n = 150, we achieved 64% power to detect a 15% difference in mortality. Furthermore, our observed mortality rate in the HTN group was different than we originally hypothesized. Moving forward, larger multicenter prospective studies are needed with improved power to better predict mortality. Additionally, it would be beneficial to evaluate other vital signs on presentation, such as heart rate, and injury severity scores as they relate to outcomes. Lastly, no multivariate analysis was performed to evaluate confounders.

4. Conclusion

In conclusion, hypertension on presentation was not associated with increased morbidity or mortality. However, a trend toward increased mortality did exist at extremes of age adjusted hypertension. Larger multicenter and prospective studies are needed to clearly delineate the effect of hypertension on outcomes in TBI and to define the upper limit of cerebral blood flow autoregulation.