Abstract
Objective:
Intracranial pressure (ICP) monitors have been used in some patients with spontaneous intracranial hemorrhage (ICH) to provide information to guide treatment and prevent uncontrolled intracranial hypertension. However, the evidence for use of ICP monitors in patients with ICH is largely drawn from the literature on patients with traumatic brain injury. We sought to assess the impact of ICP monitor placement, including external ventricular drains and intraparenchymal monitors, on neurologic outcome among patients with spontaneous nontraumatic ICH.
Methods:
In this secondary analysis of the Minimally Invasive Surgery Plus Alteplase for Intracerebral Hemorrhage Evacuation III (MISTIE III) trial, the primary outcome was poor outcome (modified Rankin Scale score 4–6) and the secondary outcome was death, at 1 year from onset. We compared outcomes in the subgroups of patients with or without an ICP monitor using unadjusted and adjusted logistic regression models. We repeated the analyses in a balanced cohort created with propensity score matching.
Results:
Out of a total of 494 patients with complete data, 70 underwent placement of an ICP monitor and 424 did not. Poor outcome was seen in 77.1% of patients in the ICP monitor subgroup compared with 53.8% in the subgroup without ICP monitors (p<0.001). Of patients in the ICP monitor subgroup, 31.4% died, compared with 21.0% in the subgroup without ICP monitors (p=0.053). In multivariate models, ICP monitor placement was associated with a more than twofold increased risk of poor outcome (odds ratio 2.76, 95% CI 1.30–5.85, p=0.008), but not with death (p=0.652). Our findings remained consistent in the propensity score matched cohort.
Conclusion:
These results question whether the placement of ICP monitors in patients with spontaneous nontraumatic ICH improves outcome. Further work is required to define the causal pathway in these preliminary findings and improve identification of patients that might benefit from invasive ICP monitoring.
Keywords: intracranial pressure monitoring, spontaneous intracranial hemorrhage, neurologic outcome
Introduction
Spontaneous intracerebral hemorrhage (ICH) affects more than 2 million adults worldwide every year. Forty percent of these patients die within a month, and there is substantial morbidity among survivors.1–4 The Minimally Invasive Surgery Plus Alteplase for Intracerebral Hemorrhage Evacuation III (MISTIE III) trial studied the potential benefit of surgical clot evacuation.5 It has been established that intracranial hypertension in patients with spontaneous ICH is a negative prognostic factor;6 however, prior research has not fully evaluated whether the placement of intracranial pressure (ICP) monitors, including external ventricular drains (EVD) and intraparenchymal monitors (IPM), and treatment of elevated ICPs lead to improved outcomes in this patient population. Current guidelines for placement of ICP monitors in patients with spontaneous ICH are derived from the traumatic brain injury literature and data supporting whether these two patient populations are similar enough to share recommendations is lacking.7,8 In MISTIE III, the decision to place an ICP monitor was at the discretion of the treating physician. Our objective was to examine the impact of ICP monitoring on neurologic outcome among the well characterized MISTIE III trial cohort.
Methods
Cohort
This is a secondary analysis of MISTIE III (clinicaltrials.gov identifier NCT01827046) using a deidentified publicly available dataset obtained from the National Institute of Neurologic Disorders and Stroke at https://www.ninds.nih.gov/Current-Research/Research-Funded-NINDS/Clinical-Research/Archived-Clinical-Research-Datasets. MISTIE III was an open-label, blinded-endpoint, phase III trial that enrolled 506 patients with nontraumatic supratentorial ICH of ≥30 mL who were randomized to either minimally invasive catheter evacuation with thrombolysis or standard of care. Additional MISTIE III inclusion criteria included a Glasgow Coma Scale (GCS) score ≤14 or a National Institutes of Health Stroke Scale (NIHSS) score ≥6, symptoms for <24 hours before the diagnostic computed tomography (CT) scan, a second “stability” CT scan at least 6 hours after the diagnostic CT that showed <5 mL of ICH growth, and ability to randomize between 12 and 72 hours after diagnostic CT. Key exclusion criteria were intraventricular hemorrhage (IVH) requiring treatment (although external ventricular drains were allowed for ICP monitoring), supratentorial ICH with midbrain extension, inability to maintain an international normalized ratio <1.4, and GCS score ≤4. We included patients from the study population with outcome data at 1 year from randomization.
Exposure and Outcomes
The exposure of interest was the placement of an ICP monitor, including EVDs and IPMs, which we used to create two subgroups: patients with an ICP monitor and those without an ICP monitor. In MISTIE III, the decision to place an ICP monitor was at the discretion of the treating physician. The trial protocol stated “placement of an ICP monitor is recommended for subjects demonstrating obtundation, which we define as GCS ≤8 on a minimum of two observations over eight hours. The goals of ICP management are to sustain intracranial pressure below 20 mmHg and to improve the patient’s level of consciousness.”5 The protocol specified that placement of an ICP monitor had to be followed by a new CT scan of the brain to monitor for ICH stability and any new areas of hemorrhage.5 Patients with obstructive hydrocephalus were ineligible for enrollment. We also compared patients with an EVD to patients without an ICP monitor and patients with an IPM to patients without an ICP monitor.
The primary outcome for our analysis was an mRS score of 4–6 at 1 year from enrollment, which we termed “poor outcome.” The adjudication of mRS in MISTIE III had particular rigor, with both local and central reviewers and the use of archival video recordings of individual patients to allow additional reviewers in cases of disagreement. We utilized the adjudicated mRS in this analysis. The secondary outcome was death within 1 year.
Statistical Analysis
Subject characteristics at randomization are summarized by the full cohort and stratified by ICP monitor versus no ICP monitor. Categorical data are presented as proportions, normally distributed continuous data as mean with standard deviation, and nonnormally distributed continuous data as median with interquartile range. We tested for intergroup differences with Student’s t-test or ANOVA for continuous variables, the chi-squared test for categorical variables, and the Wilcoxon Rank Sum or Kruskal-Wallis test for ordinal variables. We fit logistic regression models to our outcomes with the primary predictor of our exposure of interest: ICP monitoring. Model 1 was unadjusted; Model 2 was adjusted a priori for patient age, baseline GCS score, sex, white race, current smoking, premorbid mRS score, history of diabetes, prerandomization systolic blood pressure, mechanical ventilation, end-of-treatment (EoT) ICH and IVH volume, deep vs. lobar, ICH location (basal ganglia, frontal, occipital, parietal, temporal, or thalamic), MISTIE randomization arm, and the following adverse events during the trial: ischemic stroke, brain infection, worsened or new intracranial hemorrhage (including intraparenchymal, subarachnoid, subdural, epidural, and ventricular hemorrhages), seizure, and subarachnoid hemorrhage; and Model 3 was adjusted for all baseline covariates from Table 1 with p<0.05 after stepwise backwards selection. To limit multicollinearity, we confirmed that all models had a variance inflation factor that was less than 10.
Table 1.
Baseline demographics of the MISTIE III trial patients stratified by ICP monitor versus no ICP monitor placement.
| Variable | Full cohort (n=494) |
ICP monitor (n=70) |
No ICP monitor (n=424) |
p value* |
|---|---|---|---|---|
| Age (years) | 61.3±12.4 | 57.4±11.0 | 61.9±12.5 | 0.005 |
| Male | 303 (61.3) | 47 (67.1) | 256 (60.4) | 0.282 |
| White | 372 (75.3) | 48 (68.6) | 324 (76.4) | 0.046 |
| Hispanic | 68 (13.8) | 13 (18.6) | 55 (13.0) | 0.208 |
| Diabetes | 141 (28.5) | 15 (21.4) | 126 (29.7) | 0.155 |
| Hypertension | 478 (96.8) | 68 (97.1) | 410 (96.7) | 0.846 |
| Hyperlipidemia | 194 (39.3) | 24 (34.3) | 170 (40.1) | 0.357 |
| Cardiovascular disease | 74 (15.0) | 12 (17.1) | 62 (14.6) | 0.584 |
| Current smoker | 87 (17.6) | 12 (17.1) | 75 (17.7) | 0.912 |
| Cocaine abuse | 19 (3.9) | 2 (2.9) | 17 (4.0) | 0.642 |
| Alcohol abuse | 68 (13.8) | 8 (11.4) | 60 (14.2) | 0.540 |
| Antiplatelet medication | 146 (29.6) | 19 (27.1) | 127 (30.0) | 0.633 |
| Anticoagulant medication | 36 (7.1) | 7 (10.0) | 28 (6.6) | 0.305 |
| Premorbid mRS score = 1 | 36 (7.3) | 4 (5.7) | 32 (7.6) | 0.585 |
| Surgical arm of MISTIE III | 253 (51.2) | 33 (47.1) | 220 (51.9) | 0.462 |
| Admission GCS score | 10, 8–13 | 8, 7–10 | 11, 9–13 | <0.001 |
| Admission NIHSS score | 19, 15–23 | 23, 18–27 | 19, 15–23 | <0.001 |
| Prerandomization SBP (mm Hg) | 139.6±12.1 | 139.2±12.1 | 139.7±13.1 | 0.740 |
| Prerandomization DBP (mm Hg) | 70.0±11.3 | 71.7±12.1 | 69.7±11.1 | 0.171 |
| Deep (vs. lobar) location of ICH | 304 (61.5) | 55 (78.6) | 249 (58.7) | 0.002 |
| Right hemisphere location | 243 (49.2) | 37 (52.9) | 206 (48.6) | 0.508 |
| Mechanical ventilation | 208 (42.1) | 53 (75.7) | 155 (36.6) | <0.001 |
| Enrollment in US | 389 (78.7) | 57 (81.4) | 107 (78.7) | 0.554 |
| Ictus to site arrival (hours) | 7.4±8.0 | 8.5±8.3 | 6.4±7.7 | 0.036 |
| Prerandomization ICH volume | 49.0±18.0 | 53.3±21.1 | 48.3±17.3 | 0.29 |
| Prerandomization IVH volume | 2.7±5.7 | 6.6±9.2 | 2.1±4.6 | <0.001 |
| EoT ICH volume (mL) | 31.2±22.0 | 38.2±25.3 | 30.0±21.2 | 0.004 |
| EoT IVH volume (mL) | 1.7±3.8 | 3.7±5.5 | 1.4±3.3 | <0.001 |
| Percent change of ICH volume at EoT (%) | −36.5±36.8 | −29.2±34.4 | −37.7±37.0 | 0.071 |
| Percent change of IVH volume at EoT (%) | −4.1±116.7 | −30.2±65.5 | 0.2±122.6 | 0.044 |
| Days hospitalized in ICU and acute care | 22.1±20.0 | 31.5±33.4 | 20.5±16.4 | <0.001 |
| Days from onset to disposition home (n=321, 33 vs 288 patients) | 81.1±69.8 | 107.2±70.1 | 78.1±69.3 | 0.023 |
| Poor outcome (mRS score 4–6) | 282 (57.1) | 54 (77.1) | 228 (53.8) | <0.001 |
| Dead | 111 (22.5) | 22 (31.4) | 89 (21.0) | 0.053 |
Categorical variables presented as number (percentage); ordinal variables presented as median, IQR; continuous variables presented as mean±SD.
P values calculated with the Chi-squared test for categorical variables, the Wilcoxon Rank Sum test for ordinal variables, and Student’s t-test for continuous variables.
ICP, intracerebral pressure; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; SBP, systolic blood pressure; DBP, diastolic blood pressure; GCS; Glasgow Coma Scale; ICH, intracerebral hemorrhage; IVH, intraventricular hemorrhage; EoT, end of treatment
To further account for potential selection bias in patients who received an ICP monitor, we created a propensity score to match patients with an ICP monitor to those without. We matched on the following variables: mechanical ventilation, stability scan and EoT IVH volume, EoT ICH volume, male sex, prior history of diabetes, ICH location, current smoking, prerandomization systolic blood pressure, premorbid mRS, MISTIE III randomization arm, white race, age, deep vs. lobar, and GCS. The resulting decrease in bias after propensity score matching is shown in Figure 1. The logit of the propensity score for the cohort was taken, and a 0.2 standard deviation of the logit propensity scores was used as the caliper distance for propensity score estimation with a radius matching option. The final propensity-matched models took into account the propensity score weights and modelled the outcomes with the exposure of interest. The average effect of the treatment on the treated (ATT) and its odds ratio for our outcomes was calculated. The ATT measures the difference in mean (average) outcomes between patients assigned to the treatment (ICP monitor) and patients assigned to the control (no ICP monitor).
Figure 1.
Bias present in the cohorts of intracranial pressure (ICP) monitor vs. no ICP monitor shown as standardized percentage before propensity score matching (circle). The change in bias is shown by an x. IVH, intraventricular hemorrhage; EoT, end-of-treatment; ICH, intracerebral hemorrhage; GCS, Glasgow Coma Scale.
We also evaluated the rate of neurologic adverse events in our subgroups and, because they were not balanced, we performed mediation analysis to determine whether this could account for differences in our outcomes. To test the possibility of a variable being a mediator, we performed mediation tests proposed by Baron and Kenny9 for continuous and ordinal variables and by Iacobucci10 for categorical variables. Through a series of linear and logistic regressions, depending on the level of the variables, model coefficients were used to calculate a test-statistic that was tested against a standard normal at the alpha 0.05 level. If a test-statistic was significant, we estimated the total, direct, and indirect effects of the variable.
Results
Table 1 shows baseline demographics of the full cohort (n=494). Overall, 424 (85.8%) patients did not have an ICP monitor and 70 (14.2%) had an ICP monitor, of which 49/70 (70.0%) had an EVD and 21/70 (30.0%) had an IPM. Patients who had ICP monitors placed were younger (57.4±11.0 vs. 61.9±12.5 years, p=0.005), had a lower median admission GCS score (8 vs. 11, p<0.001) and higher NIHSS score (23 vs. 19, p<0.001), were more likely to require mechanical ventilation (75.7% vs. 36.6%, p<0.001), and had hemorrhages in deep as opposed to superficial (lobar) locations (78.6% vs. 58.7%, p=0.002). Other premorbid factors, such as race, sex, diabetes, hypertension, illicit drug use, and anticoagulation and antiplatelet use, were not significantly different between the two groups, nor was randomization to the surgical arm of the MISTIE trial (47.1% vs. 51.9%, p=0.462).
Prior to randomization, mean IVH volume was higher in the ICP monitor cohort (6.6±9.2 vs. 2.1±4.6 mL, p<0.001). At the EoT, both the residual ICH and IVH volumes remained higher in the ICP monitor cohort as compared with patients without ICP monitors (EoT ICH volume 38.2±25.3 vs. 30.0±21.2 mL, p=0.004; EoT IVH volume 3.7±5.5 vs. 1.4±3.3 mL, p<0.001). On average, patients with ICP monitors spent 11 more days in the ICU and acute care than patients without ICP monitors (31.5±33.4 vs. 20.5±16.4, p<0.001) and required 29 more days from the onset of hemorrhage to disposition home (n=321, 107.2±70.1 vs. 78.1±69.3, p=0.023).
Of the 70 patients with an ICP monitor, 54/70 (77.1%) had a poor outcome (mRS score 4–6) at the 1-year follow-up compared with 228 (53.8%) of the 424 patients in the cohort that did not have ICP monitors (p<0.001). In the ICP monitor cohort, 22/70 patients (31.4%) were dead at 1 year compared with 89/424 (21.0%) in the cohort that did not have ICP monitors placed (p=0.053). The rate of death at 180 days was significantly higher in patients with ICP monitors (28.6% vs. 17.9%, p=0.039), but by one year the difference was not significant.
Table 2 shows the results of logistic regression models fit to the primary and secondary outcomes. In the unadjusted Model 1, patients who had ICP monitors were more likely to have poor outcome [odds ratio (OR) 2.90, 95% confidence interval (CI) 1.61–5.23], and there was a trend towards a higher rate of the secondary outcome of death (OR 1.73, 95% CI 0.99–3.01) compared with patients without ICP monitors. In the a priori adjusted Model 2, patients with ICP monitors were still more likely to have a poor outcome (OR 2.76, 95% CI 1.30–5.85), but the difference in the rate of death at 1 year lost significance (OR 1.19, 95% CI 0.56–2.55). In Model 3, adjusted for covariates identified with stepwise backwards selection, patients who had ICP monitors placed were more likely to have a poor outcome (OR 2.37, 95% CI 1.15–4.86), but were not more likely to die (OR 1.32, 95% CI 0.68–2.60).
Table 2.
Logistic regression models fit to primary and secondary outcomes showing odds ratios for having an ICP monitor versus not having an ICP monitor.
| Poor outcome | Death | |||||
|---|---|---|---|---|---|---|
| Odds ratio | 95% CI | p value | Odds ratio | 95% CI | p value | |
| Model 1* | 2.90 | 1.61–5.23 | 0.002 | 1.73 | 0.99–3.01 | 0.055 |
| Model 2† | 2.76 | 1.30–5.85 | 0.008 | 1.19 | 0.56–2.55 | 0.652 |
| Model 3‡ | 2.37 | 1.15–4.86 | 0.019 | 1.33 | 0.68–2.60 | 0.411 |
Model 1 unadjusted.
Model 2 a priori adjusted for patient age, baseline GCS score, sex, white race, current smoking, premorbid mRS score, history of diabetes, prerandomization systolic blood pressure, mechanical ventilation, EoT ICH and IVH volume, deep vs. lobar, ICH location, MISTIE randomization arm, and the following adverse events during the trial: ischemic stroke, brain infection, worsened or new intracranial hemorrhage, seizure, and subarachnoid hemorrhage (n=487).
Model 3 when fit to good outcome is adjusted for baseline covariates selected with backwards stepwise selection set to p<0.01, which included patient age, baseline NIHSS score, deep vs. lobar, history of diabetes, MISTIE randomization arm, and EoT ICH volume. Model 3 when fit to death is adjusted for baseline covariates selected with backwards stepwise selection set to p<0.01, which included patient age, baseline GCS score, current smoking, and EoT ICH volume (n=487).
The analyses where we separated out the ICP monitors into EVDs and IPMs are shown in Table 3. The comparison between patients with an EVD (n=49) versus those without an ICP monitor (n=424), showed that EVD placement was associated with poor outcome (ORs 3.82, 3.92, and 3.63 for Models 1, 2, and 3; p<0.01 for all), but not death (Table 3). The comparison between patients with an IPM (n=21) versus those without an ICP monitor (n=424), showed the same direction of effect for the association with poor outcome (ORs 1.72, 1.61, 1.10; p>0.05 for all), but it failed to achieve significance. However, there was a stronger trend towards an association between having an IPM and death (ORs 2.32, 3.04, 2.47; p=0.071, 0.069, 0.105).
Table 3.
Logistic regression models fit to primary and secondary outcomes showing odds ratios for having an EVD or an IPM versus not having an ICP monitor.
| Predictor | Model | Poor outcome | Death | ||||
|---|---|---|---|---|---|---|---|
| Odds ratio | 95% CI | p value | Odds ratio | 95% CI | p value | ||
| EVD (n=49) (vs. not having an ICP monitor, n=424) | Model 1* | 3.82 | 1.81-8.07 | 0.002 | 1.51 | 0.78-2.92 | 0.226 |
| Model 2† | 3.92 | 1.50-10.3 | 0.005 | 0.81 | 0.33-1.97 | 0.641 | |
| Model 3‡ | 3.63 | 1.47-9.00 | 0.005 | 1.03 | 0.47-2.24 | 0.940 | |
| IPM (n=21) (vs. not having an ICP monitor, n=424) | Model 1* | 1.72 | 0.68-4.35 | 0.252 | 2.32 | 0.93-5.76 | 0.071 |
| Model 2† | 1.61 | 0.53-4.89 | 0.402 | 3.05 | 0.92-10.2 | 0.069 | |
| Model 3‡ | 1.10 | 0.37-3.32 | 0.865 | 2.47 | 0.83-7.33 | 0.105 | |
Model 1 unadjusted.
Model 2 a priori adjusted for patient age, baseline GCS score, sex, white race, current smoking, premorbid mRS score, history of diabetes, prerandomization systolic blood pressure, mechanical ventilation, EoT ICH and IVH volume, deep vs. lobar, ICH location, MISTIE randomization arm, and the following adverse events during the trial: ischemic stroke, brain infection, worsened or new intracranial hemorrhage, seizure, and subarachnoid hemorrhage (n=466 for EVD, n=438 for IPM).
Model 3 when fit to good outcome is adjusted for baseline covariates selected with backwards stepwise selection set to p<0.01, which included patient age, baseline NIHSS score, deep vs. lobar, history of diabetes, MISTIE randomization arm, and EoT ICH volume. Model 3 when fit to death is adjusted for baseline covariates selected with backwards stepwise selection set to p<0.01, which included patient age, baseline GCS score, current smoking, and EoT ICH volume (n=466 for EVD, n=438 for IPM).
We explored major adverse neurological events for the entire cohort and for patients with and without an ICP monitor (Table 4). Overall, patients who had an ICP monitor had a higher incidence of new or worsened intracranial hemorrhage (41.4% vs. 22.9%, p=0.001) along with a higher likelihood of an intracranial infection (7.1% vs. 1.7%, p=0.006). Comparing the subgroups of ICP monitor types to those without an ICP monitor, patients with an EVD had a higher rate of infection (10.2% vs. 1.7%, p<0.001) and patients with an IPM had a higher rate of new or worsened intracranial hemorrhage (57.1% vs. 22.9%, p<0.001). The rates of seizures, ischemic stroke, and subarachnoid hemorrhage were not statistically different between patients with ICP monitor versus without one (Table 4) and in the EVD versus IPM comparisons (data not shown).
Table 4.
Major adverse neurological events
| Variable | Full cohort (n=494) |
ICP Monitor (n=70) |
No ICP Monitor (n=424) |
p value* |
|---|---|---|---|---|
| Worsened or new intracranial hemorrhage | 126 (25.5) | 29 (41.4) | 97 (22.9) | 0.001 |
| Brain infection | 12 (2.4) | 5 (7.1) | 7 (1.7) | 0.006 |
| Ischemic stroke | 101 (20.5) | 18 (25.7) | 83 (19.6) | 0.238 |
| Seizure | 70 (14.2) | 7 (10.0) | 63 (14.9) | 0.280 |
| Subarachnoid hemorrhage | 26 (5.3) | 3 (4.3) | 23 (5.4) | 0.693 |
Categorical variables presented as number (percentage)
P values calculated with the Chi-squared test.
We performed mediation analysis to determine whether the higher rate of new or worsened intracranial hemorrhage, intracranial infection, or number of days spent in acute care occurred contemporaneously with increased likelihood of poor outcome in patients with an ICP monitor. New or worsened intracranial hemorrhage and intracranial infection were not significant mediators (test statistic: p=0.92 and 0.73, respectively), but the longer time spent in acute care, either intensive care or inpatient hospitalized care, for patients with an ICP monitor was a significant mediator in the pathway between ICP monitor and poor outcome (OR=1.17, 95% CI 1.04–1.53, 21.66% mediated, p=0.027).
We used propensity score matching to minimize the presence of bias in the two cohorts (Figure 1). After matching, the balanced cohort had 413 individuals, 70 with ICP monitors and 343 without ICP monitors. When we examined the propensity score–matched cohort ATT results for the primary endpoint of poor outcome (Table 5), the ATT log odds estimate was −0.24, with an OR of 0.79 (95% CI 0.70–0.88, p<0.001), indicating a significantly higher propensity for poor outcome in patients with ICP monitors. For the secondary outcome of death at 1 year, the ATT log odds estimate was 0.09 with a non-significant ATT OR of 1.10 (95% CI 0.84–1.17, p=0.124), consistent with the primary analysis results.
Table 5.
Propensity score-matched cohort ATT results for patients with and without an ICP monitor
| Outcome | ATT log odds estimate | ATT OR | 95% CI of OR | p value |
|---|---|---|---|---|
| Poor outcome | 0.236 | 1.5 | (1.40-1.75) | < 0.001 |
| Dead | 0.096 | 1.10 | (0.84-1.17) | 0.124 |
Discussion
The placement of ICP monitors for patients with spontaneous, nontraumatic supratentorial ICH was associated with a negative effect on functional outcome at one year, although we did not find that it was associated with an increase in the risk of death. Of note, intracranial hypertension (defined as an ICP ≥20 mm Hg) has been observed in up to 70% of patients in whom ICP monitors are placed for spontaneous ICH.11 Increased ICP is thought to be more common in younger patients and in patients with supratentorial ICH.11 However, we found that even in age and ICH location adjusted models there was no functional outcome benefit for ICP monitor placement. In the MISTIE III trial, ICP elevation was less common in the MISTIE intervention cohort than in the cohort assigned to standard medical care.5 ICP elevations ≥20 mm Hg occurred in 26% of MISTIE patients and 58% of the cohort assigned to standard medical care (p=0.007). We do not comment on this aspect of ICP monitoring, because the elevated ICPs were not analyzed for association with poor outcome in our analysis or the MISTIE III primary outcome publication.5 Nonetheless, the hypothesis-generating result of our analysis questions the utility of ICP monitors in this patient population, irrespective of the ICP values.
Guidelines for the management of spontaneous ICH have addressed the use of ICP monitors.12 The current recommendations for placement of an ICP monitor represent Class IIb evidence and define the patient population as those with a GCS score ≤8, evidence of transtentorial herniation, or significant IVH or hydrocephalus.7,8 Because there are limited data regarding monitoring and treatment of elevated ICP in ICH, these recommendations are largely borrowed from the existing body of literature on patients with severe traumatic brain injury. The question remains whether these two patient populations are similar enough to share management guidelines.
Our results are in accordance with other recently published literature on ICP monitoring in patients with spontaneous ICH. Chen et al.13 recently demonstrated higher infection rates and longer hospital stays in their ICP monitoring cohort. The primary outcome of 90-day mortality was similar between patients with and without an ICP monitor. When patients with IVH were excluded from their analysis, the cohort with ICP monitors had a lower rate of mortality at 90 days: 23% vs. 36% (OR 0.54, 95% CI 0.302–0.975, p=0.041), but patients with ICP monitors also had a lower rate of good outcome (defined as functional independence) at 90 days: 14% vs. 29% (OR 0.416, 95% CI 0.203–0.813, p=0.010). Our finding that patients with spontaneous ICH who underwent ICP monitoring placement experienced higher rates of poor neurologic outcome at the 1-year mark are similar, but we were able to further demonstrate that even after propensity score matching, patients who had ICP monitors placed in the context of spontaneous ICH were more likely to have a poor outcome at 1 year.
We found that patients who had an ICP monitor placed had significantly longer time spent in ICU and acute care, nearly double the rate of new or worsened intracranial hemorrhage, and a fourfold higher rate of intracranial infection. However, our mediation analysis showed that new or worsened intracranial hemorrhage and intracranial infection did not account for their higher rate of poor outcome, while the longer time spent in ICU and acute care was a significant mediator. This suggests either that ICP monitor placement leads to additional complications that we were not able to measure or that patients with ICP monitors had higher disease severity, which could independently create additional complications and longer length of ICU and acute care. With the current data, we cannot measure if adverse events were the reason for or cause of ICP monitor placement. Without this information, the causal influence of the adverse events on ICP monitor placement cannot be known. Nonetheless, mitigation of complications with stringent ICP monitor placement protocols including attention to sterile technique, review of coagulation parameters, and administration of antibiotic prophylaxis prior to placement may represent an important strategy for achieving a meaningful improvement in morbidity and mortality among patients with spontaneous ICH undergoing ICP monitor placement.
To date, surgical intervention has not been shown to improve outcome in patients with spontaneous ICH.13,14 This remained true for patients in the MISTIE III trial, although there was a significant benefit in patients who achieved a reduction in their ICH volume to ≤15 mL via the MISTIE procedure. The literature shows that patients with spontaneous ICH who undergo ICP monitor placement are more likely to undergo surgical evacuation of their ICH.15,16 These data raise the question of whether patients with spontaneous ICH with concomitant ICP monitor placement are being “overtreated” without Class I evidence of significant benefit. Although there is some evidence that elevations in ICP, even transient, in patients with ICH may lead to worse short-term outcomes, associations with long-term poor outcomes are more variable.17–19 Furthermore, a randomized controlled trial in patients with severe TBI done by Chesnut et al, comparing outcomes between a protocol involving empiric placement of ICP monitors versus treatment based on imaging and clinical examination, did not show a benefit from the empiric use of ICP monitors.18 This implies that, even in patients with severe TBI, placement of ICP monitors to guide acute management may not result in improved outcomes and therefore cannot be extrapolated to patients with spontaneous ICH.
Limitations
The main limitation of our analysis is that this is a secondary analysis of a clinical trial that was not designed to answer this question and, thus, the subgroups are not balanced. Although we tried to reduce that bias with adjusted models and a propensity score matched cohort, it remains a major limitation. In addition, the number of patients in MISTIE III that underwent placement of an ICP monitor and had complete data was small (n=70). It is possible that a benefit associated with ICP monitor placement would be shown with a larger number of patients. Because MISTIE III was a protocolized clinical trial, we were unable to explore the question of whether ICP monitors guide decisions about, or timing of, surgical interventions in this patient population. We did not have full data on the adverse events, such as timing and exact location of new or worsened intracranial hemorrhage. In future research on this topic, adverse events should be fully adjudicated and classified as “related to ICP monitor placement” or “indication for ICP monitor placement.” We also did not have data on the indication for ICP monitor, placement location, or the duration of the monitoring, which could have produced analyses pointing to specific subgroups or protocols that have benefit.
We were also underpowered to study the different types of ICP monitors. While the direction of effect was consistent for patients with an EVD or IPM, we could not fully explore differences between them. Finally, we were unable to explore the effect of ICP or cerebral perfusion pressure (CPP) in this cohort and whether patients benefited from ICP and CPP directed care. Ziai18 performed a randomized trial of intraventricular thrombolysis in 100 patients with IVH and ICH <30 mm3 and demonstrated that, although initial ICPs were >20 mm Hg upon insertion of a ventricular catheter, ICP was not frequently elevated during the monitoring period. However, it is also important to note that elevated ICP, which can lead to inadequate CPP, was predictive of higher short- and long-term mortality in patients with hypertensive intraventricular hemorrhage that were enrolled in the Clot Lysis: Evaluating Accelerated Resolution of Intraventricular Hemorrhage (CLEAR III) trial.19
Conclusion
To date, there is no medical or surgical intervention shown to benefit patients with spontaneous ICH on primary outcome analysis. Additionally, there is no published evidence showing that monitoring of ICP or management of elevated ICP in patients with spontaneous ICH leads to improved neurologic outcomes. Our study demonstrates that in patients with spontaneous ICH who have ICP monitors placed there is an association with poor neurologic outcome at one year that remains significant despite multiple strategies to minimize bias. This is an association study and is hypothesis generating. It cannot imply causality or any treatment recommendations regarding placement of ICP monitors in this population. Current management principles for the placement of an ICP monitor in ICH are largely extrapolated from the literature on patients with severe traumatic brain injury, but these patient populations are distinct and should be treated as such, particularly from the standpoint of ICP monitor placement. Further studies are needed to confirm our preliminary findings and determine which patients with spontaneous ICH require ICP monitoring.
References
- 1.Sacco S, Marini C, Toni D, et al. Incidence and 10-year survival of intracerebral hemorrhage in a population-based registry. Stroke. 2009;40(2):394–399. [DOI] [PubMed] [Google Scholar]
- 2.van Asch CJ, Luitse MJ, Rinkel GJ, et al. Incidence, case fatality, and functional outcome of intracerebral haemorrhage over time, according to age, sex, and ethnic origin: a systematic review and meta-analysis. Lancet Neurol. 2010;9(2):167–176. [DOI] [PubMed] [Google Scholar]
- 3.Dennis MS, Burn JP, Sandercock PA, et al. Long-term survival after first-ever stroke: the Oxfordshire Community Stroke Project. Stroke. 1993;24(6):796–800. [DOI] [PubMed] [Google Scholar]
- 4.Broderick JP, Brott T, Tomsick T, et al. The risk of subarachnoid and intracerebral hemorrhages in blacks as compared with whites. N Engl J Med. 1992;326(11):733–736. [DOI] [PubMed] [Google Scholar]
- 5.Hanley DF, Thompson RE, Rosenblum M, et al. Efficacy and safety of minimally invasive surgery with thrombolysis in intracerebral haemorrhage evacuation (MISTIE III): a randomised, controlled, open-label, blinded endpoint phase 3 trial. Lancet. 2019;393(10175):1021–1032. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Godoy DA, Nunez-Patino RA, Zorrilla-Vaca A, et al. Intracranial hypertension after spontaneous intracerebral hemorrhage: a systematic review and meta-analysis of prevalence and mortality rate. Neurocrit Care. 2019;31(1):176–187. [DOI] [PubMed] [Google Scholar]
- 7.Brain Trauma Foundation, American Association of Neurological Surgeons, Congress of Neurological Surgeons, et al. Guidelines for the management of severe traumatic brain injury. VIII. Intracranial pressure thresholds. J Neurotrauma. 2007;24 Suppl 1:S55–58. [DOI] [PubMed] [Google Scholar]
- 8.Brain Trauma Foundation, American Association of Neurological Surgeons, Congress of Neurological Surgeons, et al. Guidelines for the management of severe traumatic brain injury. VI. Indications for intracranial pressure monitoring. J Neurotrauma. 2007;24 Suppl 1:S37–44. [DOI] [PubMed] [Google Scholar]
- 9.Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol. 1986;51(6):1173–1182. [DOI] [PubMed] [Google Scholar]
- 10.Iacobucci D Mediation analysis and categorical variables: The final frontier. Journal of Consumer Psychology. 2012;22(4):582–594. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kamel H, Hemphill JC, 3rd. Characteristics and sequelae of intracranial hypertension after intracerebral hemorrhage. Neurocrit Care. 2012;17(2):172–176. [DOI] [PubMed] [Google Scholar]
- 12.Hemphill JC 3rd, Greenberg SM, Anderson CS, et al. Guidelines for the management of spontaneous intracerebral hemorrhage: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2015;46(7):2032–2060. [DOI] [PubMed] [Google Scholar]
- 13.Chen CJ, Ding D, Ironside N, et al. Intracranial pressure monitoring in patients with spontaneous intracerebral hemorrhage. J Neurosurg. 2019:1–11. [DOI] [PubMed] [Google Scholar]
- 14.Mendelow AD, Gregson BA, Rowan EN, et al. Early surgery versus initial conservative treatment in patients with spontaneous supratentorial lobar intracerebral haematomas (STICH II): a randomised trial. Lancet. 2013;382(9890):397–408. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Keep RF, Hua Y, Xi G. Intracerebral haemorrhage: mechanisms of injury and therapeutic targets. Lancet Neurol. 2012;11(8):720–731. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Shah M, Birnbaum L, Rasmussen J, et al. Effect of hyperosmolar therapy on outcome following spontaneous intracerebral hemorrhage: ethnic/racial variations of intracerebral hemorrhage (ERICH) study. J Stroke Cerebrovasc Dis. 2018;27(4):1061–1067. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Tian Y, Wang Z, Jia Y, et al. Intracranial pressure variability predicts short-term outcome after intracerebral hemorrhage: a retrospective study. J Neurol Sci. 2013;330(1-2):38–44. [DOI] [PubMed] [Google Scholar]
- 18.Ziai WC, Melnychuk E, Thompson CB, et al. Occurrence and impact of intracranial pressure elevation during treatment of severe intraventricular hemorrhage. Crit Care Med. 2012;40(5):1601–1608. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ziai WC, Thompson CB, Mayo S, et al. Intracranial hypertension and cerebral perfusion pressure insults in adult hypertensive intraventricular hemorrhage: occurrence and associations with outcome. Crit Care Med. 2019;47(8):1125–1134. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Chesnut RM, Temkin N, Carney N, et al. A trial of intracranial-pressure monitoring in traumatic brain injury. N Engl J Med. 2012;367(26):2471–2481. [DOI] [PMC free article] [PubMed] [Google Scholar]