Hemodynamic response during endotracheal suctioning predicts awakening and functional outcome in subarachnoid hemorrhage patients

The study design was guided by the STROBE statement on observational studies. This is a retrospective analysis of prospectively collected data of patients with non-traumatic SAH that were admitted to the neurological intensive care unit of a tertiary care hospital (Medical University of Innsbruck) between 2010 and 2017. Inclusion criteria encompassed (1) non-traumatic SAH, (2) age ≥ 18 years, and (3) intensive care unit stay for at least 24 h. Out of 324 screened patients, 28 patients were excluded due to (1) lack of informed consent, (2) arteriovenous malformation, and (3) late admission (> 7 days after bleeding onset) resulting in 296 patients included in our registry. Out of these, 105 patients were excluded due to (1) no requirement for aneurysm intervention and intubation due to non-aneurysmal SAH (N = 53) or early withhold of therapy (N = 1), (2) short intubation period without the necessity of ES (early death N = 3; intubation only for intervention N = 42), or (3) intubation/ES after 72 h (N = 6) leaving 191 eligible patients for the current analysis (Fig. 1). All patients without the detection of an aneurysm in repeated angiography fulfilled the clinical and radiographic criteria for spontaneous SAH suggestive of aneurysmal SAH with more blood compared to a typical perimesencephalic SAH. The conduct of the study was approved by the local ethics committee (Medical University of Innsbruck, AM4091-292/4.6). Written informed consent was obtained according to local regulations.

Critical care management adhered to international guidelines [10, 11] with the exception of nimodipine which was prophylactically administered intravenously and not orally in poor-grade patients. Clinical grading was done by using the Hunt and Hess (H&H) score at initial presentation after the bleeding without sedation [12]. Admission computed tomography (CT) scans were scored by a neuroradiologist blinded to clinical data using the modified Fisher (mFisher) score and SEBES (Subarachnoid Hemorrhage Early Brain Edema Score) [13, 14]. Magnetic resonance imaging (MRI) was done when clinically indicated. Ruptured aneurysms were secured by either clipping or coiling after discussion in an interdisciplinary team. Transcranial color-coded duplex sonography (TCD, LOGIQ S8; GE Healthcare, Chicago, IL) was regularly performed in order to follow patients for vasospasm. Vasospasm was defined as elevation of mean velocities greater than 120 cm/s in the middle or anterior cerebral artery or daily change in mean TCD velocities greater than 50 cm/s. Severe vasospasm (> 200 cm/s) was further confirmed by catheter cerebral angiogram. The definition of delayed cerebral ischemia (DCI) was based either on clinical deterioration with a new focal neurologic deficit, a decrease of greater than or equal to 2 points on the Glasgow Coma Scale, or a new infarct on the CT or MRI scan not attributable to other causes [15]. Severe vasospasm or DCI was treated with induced hypertension. When clinical symptoms were refractory to hypertensive treatment or neuromonitoring variables were indicative of further deterioration, cerebral pan-angiography was pursued to evaluate for intra-arterial nimodipine treatment. Hemodynamic augmentation was achieved using vasopressors (most commonly noradrenaline or in the setting of stunned myocardium or low cardiac output dobutamine was used) or volume administration. Sedation and analgesia consisted of continuous IV infusions of midazolam and/or sufentanil at the discretion of the treating physician (Table 1). Generally, a pressure-regulated volume-controlled ventilation mode was used. To wean patients off, assisted spontaneous breathing with pressure support and continuous positive airway pressure (CPAP) was used. Targeted temperature management with the aim of temperatures between 36.5 and 37.5 °C and fever prevention was applied. Endotracheal suctioning was routinely performed during the stay in the ICU to ensure secretion clearing and to maintain airway patency. Preoxygenation before ES was done in all patients.

Patient demographics, hospital complications, and outcomes were prospectively collected and discussed in weekly meetings held by the study team and treating neurointensivists.

Three-minute median values of vital parameters including HR and MAP were automatically saved in the electronic patient data management system (PDMS, CentricityTM Critical Care 8.1 SP7; GE Healthcare Information Technology, Dornstadt, Germany). Due to the inconsistent data entry in the storage system with seconds apart, we chose to calculate 5-min interval data based on the mean as a post hoc analysis. Arterial blood pressure was most often measured through either radial or femoral arterial lines; HR was measured through an ECG or an arterial line. Continuous ECG recordings were performed in all patients.

Only the first 72 h of admission were analyzed in order to avoid a selection bias of poor-grade patients requiring longer intubation and to allow early discrimination of patients. The day of admission was denoted as day 1 (first 24 h of admission). The sedation status was retrospectively assessed by chart review including daily notes of nurses, daily notes of treating neurointensivists, and the medical report using the RASS (Richmond Agitation Sedation Scale) [16, 17]. The 10-point scale was classified as follows: − 5 or − 4 (deep sedation), − 3 to − 1 (moderate to mild sedation), and 0–5 (no sedation). The record of RASS was missing in 12 patients. Moreover, the daily cumulative dose of midazolam in milligram was calculated.

The primary outcome measure was time to arousal defined as the time to RASS ≥ 0. Secondary deterioration was not counted if the patient was conscious (RASS ≥ 0) for at least 5 days. Data were missing in 6 patients due to early repatriation (N = 4) or missing documentation (N = 2). Functional outcome after 3 months was assessed by a study nurse blinded to the clinical course of the patient via telephone interview using the modified Rankin Scale Score (mRS). Poor outcome was defined as a mRS > 2. In 10 patients who were lost to follow-up, the discharge mRS was carried forward to 3 months.

Unreliable systolic (< 40 mmHg or > 250 mmHg; 13 measurements) or diastolic (< 15 mmHg) blood pressure values secondary to transducer flushing, movement artifacts, connection-reconnection artifacts or obstructed arterial lines were deleted. Timing of suctioning episodes was identified in the PDMS. Baseline HR or MAP was calculated by the median HR or MAP between 30 and 10 min before ES. Thirty minutes before and after ES were analyzed. If the gap between two episodes of ES was less than 1 h, the latter was discarded. ∆HR and ∆MAP at each time point were calculated by subtracting HR or MAP at baseline from the HR or MAP during suctioning or the following 30 min. Ten percent increase of HF and MAP at ES was judged significant.

Continuous variables were expressed as mean ± SD or median and interquartile range (IQR). Categorical variables were reported as counts and proportions. Univariate and multivariable analysis was performed with generalized estimating equation models (GEE) with the correlation matrix best fitting the data as proposed by Chan et al. based on the lowest Quasi Likelihood under Independence Model Criterion (QIC) and Corrected Quasi Likelihood under Independence Model Criterion (QICC) in order to account for repeated measurements within one patient [18]. All multivariable models were adjusted for predefined variables including H&H grade (to account for disease severity), age (due to less pronounced autonomic responses in older subjects [19]), RASS or daily cumulative midazolam dose (to account for sedation depth), daily cumulative dose of sufentanil, dobutamine, and noradrenaline and the absolute HR or MAP, and indicated appropriately. Due to the strong influence of vigilance by benzodiazepines and resulting probable collinearity between RASS and midazolam, we built two distinct models including each variable separately. Cases with missing values were included. Adjusted odds ratios (adjORs) with 95% confidence intervals (CI) were calculated. The level of significance was set at α = 0.05. All analyses and graphical representations were performed with IBM-SPSS V24.0 (SPSS Inc., Armonk, NY) and Prism 5 for Windows V5.01 (GraphPad Software, Inc., LA Jolla, CA).

As a response to ES, overall HR increased by mean 2.3 ± 7.1 bpm (p < 0.001) from 75.1 ± 14.8 bpm at baseline which corresponded to an increase of 3.5 ± 10.0%. During 177/1080 (16%) suctioning episodes, a significant increase of > 10% was reached. HR increased in 681/1080 (63%; +5.3 ± 6.8) and decreased in 399/1080 37% (− 2.8 ± 4.4) episodes (Fig. 2). MAP increased by mean 3.2 ± 7.8 mmHg (p < 0.001; 4.1 ± 9.9%) from 80.9 ± 9.8 mmHg at baseline.

Poor-grade patients (H&H grades 4 & 5) showed significantly lower increases in HR during suctioning as compared to good-grade patients (H&H grades 1–3, p < 0.001). However, no difference in MAP (p = 0.435) was observed. Neuroradiographic parameters on admission did not well discriminate between suctioning associated ∆HR (intracerebral hemorrhage on admission, p = 0.192; low-/high-grade SEBES, p = 0.092; low-/high-grade mFisher, p = 0.196). Lower ∆MAP was only associated with high-grade SEBES (p = 0.025) but not with low- or high-grade mFisher (p = 0.101) or ICH present on admission CT scan (p = 0.162; Table 3).

The response of HR (p < 0.001) and MAP (p = 0.004) was significantly lower during controlled ventilation mode as compared to assisted ventilation (Table 3). Importantly, during most ES episodes (829/1080, 83%), patients were deeply sedated (RASS − 5 or − 4). In 46 (4%) episodes, a RASS between − 3 and − 1; in 54 (5%), a RASS between 0 and 1; and in only 11 (1%), a RASS between 2 and 4 were recorded. RASS was missing for 77 (7%) episodes.

Time to arousal was median 13 (IQR, 4–21) days (N = 142). Five patients remained in a vegetative state (mean ∆HR, 0.3 ± 4.0). Out of 42 patients who died, 38 did not regain consciousness before death (mean ∆HR, 1.9 ± 7.7). ∆HR (B = − 0.191, Wald statistic = 25.4, df = 1, p < 0.001) during ES was significantly lower in patients with prolonged time to regain consciousness in a model adjusted for H&H grade, age, and absolute HR. Even when additionally correcting for daily cumulative dose of dobutamine, noradrenaline, sufentanil, and RASS (B = − 0.130, Wald statistic = 12.5, df = 1, p < 0.001) or cumulative midazolam dose (B = − 0.166, Wald statistic = 20.4, df = 1, p < 0.001), this association remained robust. ∆MAP (p = 0.087) was not associated with time to arousal.

∆HR was significantly lower in patients with poor 3-month functional outcome as compared to those with good functional outcome (1.7 ± 6.9 vs. 3.3 ± 7.5; p = 0.001; Fig. 3). In multivariable analysis adjusted for admission H&H grade; age; absolute HR; daily cumulative dose of sufentanil, dobutamine, and noradrenaline; and daily cumulative midazolam dose or RASS, ∆HR was still significantly lower in patients with poor outcome (p < 0.001; Table 4). When separating in increasing and decreasing responses of HR, only a positive ∆HR (p < 0.001) but not negative ∆HR (p = 0.728) was associated with poor outcome.

∆MAP was similar across patients’ outcomes (adjusted for midazolam, p = 0.263; or RASS, p = 0.344; Fig. 4).