Introduction

Despite significant progress in the management of patients with stroke in Russia and other countries, the problem of severe cerebrovascular accident is still extremely important due to high mortality and disability [1—4]. Intensive care protocols for patients with stroke have been widely implemented in recent years [5—9]. At the same time, the issues of respiratory support in severe stroke are still unresolved [10]. Specialists discuss the safest and most effective strategies of ventilation in patients with acute cerebral failure. Currently, there are only single-center cohort studies devoted to this issue [11—16].

The peculiarities of respiratory support in intensive care of patients with stroke are high incidence and duration of mechanical ventilation [13]. Mechanical ventilation mode in patients with stroke, safe and effective positive end-expiratory pressure (PEEP) and timing of tracheostomy are unresolved questions. Complications of prolonged mechanical ventilation (ventilator-associated pneumonia and tracheobronchitis) are of special attention [17]. There is an opinion that volume-controlled ventilation and PEEP increase intracranial pressure (ICP) [18]. However, PEEP is necessary within the concept of protective ventilation, as well as for correction of hypoxia [19—21]. There are important data on the absence of significant ICP increase following PEEP increment up to 15 cm H₂O [21, 22]. At present, effectiveness of various methods of inspiration control during mechanical ventilation is unclear [22, 23]. Hypocapnia following hyperventilation is an important problem in the management of patients with acute cerebral insufficiency [24, 25]. On the one hand, hypocapnia can be considered as effective approach to decrease ICP via cerebral vasoconstriction [26]. On the other hand, hyperventilation can impair cerebral perfusion that is highly undesirable [27].

Restoration of adequate spontaneous breathing and tracheobronchial tree drainage in case of impaired cerebral autoregulation are ones of the most difficult stages in the treatment of patients with stroke [28]. Effectiveness of early tracheostomy for prevention of bronchopulmonary inflammation was revealed in various trials [2, 29–31]. However, there are not enough studies devoted to this approach in the management of patients with stroke.

Despite multiple studies devoted to the course and outcomes of severe stroke, there are no multiple-center clinical trials which would be aimed at comprehensive analysis of respiratory support in these patients.

The purpose of the study was to analyze the features of respiratory support and identify predictors of mortality in patients with severe stroke.

Material and methods

A multiple-center observational clinical study involved 14 hospitals. This trial was carried out under support of the Russian Federation of Anesthesiologists and Reanimatologists (FAR). The FAR committee on clinical guidelines and multipla-center trials approved the study protocol. The study was also approved by local ethics committees. Inclusion criteria: patients with hemorrhagic (HS) and ischemic (IS) stroke, age 18—90 years, need for mechanical ventilation. Exclusion criteria: pregnancy, histologically confirmed malignancies, cardiovascular diseases (NYHA class III—IV), liver cirrhosis (terminal), chronic kidney disease stage V (hemodialysis).

Consecutive patients have been enrolled for the period from November 1, 2017 to November 1, 2019 according to inclusion criteria. Database was formed using e-questionnaires (State registration certificate for a computer program No. 2019619217 dated May 21, 2019) [32]. Study design was observational. We analyzed the following parameters and criteria:

— indications for mechanical ventilation;

— baseline parameters of ventilation (Vt, ml/kg; MV, l/min; PIP, cm H₂O; PEEP, cm H₂O; Pplat, cm H₂O; Pdr, cm H₂O; FiO₂,%) and maximum PEEP;

— baseline ventilation modes — volume-controlled ventilation (VC), pressure-controlled (PC) and dual-control ventilation (DC), adaptive ventilation modes (ASV/AutoMVG), SIMV, ASV, PSV, SBT test, CPAP and their influence on mortality;

— period until tracheostomy;

— duration of mechanical ventilation and weaning from ventilator;

— influence of hyperventilation on correction of intracranial hypertension;

— GOS (Glasgow Outcome Scale) outcomes.

The register included 1,289 patients; 1,144 patients met inclusion criteria. There were 609 (53.23%) men and 535 (46.77%) women.

Statistical analysis was carried out in accordance with the generally accepted statistical methods using the STATISTICA 10.0 software package. Categorical data are described as absolute values and percentages. We used the Pearson χ² test for between-group comparison of mortality. Logistic regression was applied to analyze the effect of certain factor on mortality. These data are presented as odds ratio (OR) with a 95% confidence interval (95% CI). Statistical comparability of groups by age, sex, severity of stroke at admission and ventilation onset was essential for decision-making. Analyzing between-group comparability and numerical data, we described data as median (Me), upper and lower quartiles (Q1; Q3) in case of non-parametric distribution. The Mann-Whitney test was used to assess significance of between-group differences in severity of stroke. Between-group differences were considered significant at p <0.05.

Results

Hypoxia/hypoxemia was the most common indication for respiratory support in patients with stroke (30.7%) (Table 1). This indication was more common in IS patients compared to those with HS. Isolated impairment of consciousness (GCS score ≤ 9) was observed in 19.9% of patients with stroke. Incidence of this indication for respiratory support was similar in patients with IS and HS. Hypocapnia (hyperventilation) as an indication for respiratory support was found in 2.7% of patients. Combination of several indications for respiratory support was observed in 21.2% of patients and only in those with subarachnoid hemorrhage (SAH) and HS.

Table 1. Distribution of patients with stroke depending on indications for ventilation

Note. IS — ischemic stroke; HS — hemorrhagic stroke; SAH — subarachnoid hemorrhage; GCS — Glasgow Coma Scale.

Patients with HS required more aggressive baseline parameters of ventilation compared to respiratory support in patients with IS (Table 2). In particular, median of minute ventilation was 8 [6.5; 9] and 7 [6; 8] l/min in patients with HS and IS, respectively (p <0.001). Median of peak inspiratory pressure was 18 [12; 22] and 15 [0; 16] cm H₂O in both groups, respectively (p <0.001). Median of positive end-expiratory pressure was 5 [5; 7] and 5 [4; 5] cm H₂O in patients with HS and IS, respectively (p <0.001).

Table 2. Baseline parameters of respiratory support in patients with stroke

Note. IS — ischemic stroke; HS — hemorrhagic stroke; SAH — subarachnoid hemorrhage.

Baseline VC ventilation was the most common (65.56%). PC ventilation was initially used in 21.42% of cases, DC ventilation — in 9.88% of patients. Auxiliary ventilation was mainly carried out at the beginning of respiratory support. Adaptive ventilation (ASV/AutoMVG, etc.) was applied in 2.1% of cases, CPAP — in 1.05% of patients. Patients with basic PC ventilation were characterized by significantly lower mortality compared to the group of VC ventilation (p <0.001). Both groups were not completely comparable regarding stroke severity at admission (NIHSS score 18 [13; 25] in VC ventilation group (n = 750) and 16 [12; 20] in PC ventilation group (n = 245), p <0.001). We also analyzed subgroups of patients with NIHSS score > 20 at admission. There were significant differences in mortality between comparable groups (OR 0.36, 95% CI 0.21 — 0.60, p <0.001) (Fig. 1). NIHSS score at admission in patients on VC ventilation (n = 343) was 26 [23; 30] points, in patients on PC ventilation (n = 71) — 26 [22; 30] points (p = 0.409). We analyzed daily survival of patients with IS and obtained the following data. Survival in patients on VC ventilation was significantly lower compared to those on PC ventilation (plog-rank <0.05) (Fig. 2). It should be noted that 50% of patients on PC ventilation died after 17 days, 50% of patients on VC ventilation — after 9 days. In acute period of stroke, the 3^rd — 4^th days were characterized by the highest risk of mortality in patients on VC ventilation (0.14). Maximum probability of mortality was observed on the 17^th—18^th days (0.16). In patients on PC ventilation, the highest risk of mortality in acute period of stroke was observed between the 4^th and the 5^th days (0.06), maximum probability of mortality — between the 30^th and the 31^st days (0.09).

Fig. 1. Comparison of mortality among patients with volume- and pressure-controlled ventilation at the beginning of respiratory support (within-group analysis).

Fig. 2. Survival of patients with ischemic stroke ventilated under different principles of inspiration control (Kaplan-Meier analysis).

In the group of hemorrhagic stroke, daily survival was significantly lower in patients on VC ventilation compared to those on PC ventilation (p log-rank <0.05) (Fig. 3). There were between-group differences in 50% survival (18 days in PC ventilation group and 6 days in VC ventilation group). In acute period of stroke, the highest probability of mortality in VC ventilation group was observed between the 4^th and the 5^th days (0.14). Maximum risk of mortality was observed between the 29^th and the 31^st days (0.22). In case of PC ventilation, the highest probability of mortality was found between the 11^th and the 14^th days (0.07).

Fig. 3. Survival of patients with hemorrhagic stroke ventilated under different principles of inspiration control (Kaplan-Meier analysis).

Maximum PEEP within 5 cm H₂O was required in 54.63% of patients, 6—10 cm H₂O — 45.10%, 11—15 cm H₂O — 0.17% of patients. We did not use PEEP > 15 cm H₂O. PEEP in critically ill patients with stroke did not increase mortality (p <0.05).

Period until tracheostomy was similar in patients with IS, HS and SAH. Early tracheostomy (1–3 days) in patients with stroke was associated with less period of ventilation compared to later tracheostomy (0 [0; 2] vs. 2 [0; 4] days, p <0.001). Severity of stroke was similar in both groups at admission (p> 0.05). Early tracheostomy did not affect mortality until the end of acute period of stroke.

Hyperventilation for correction of intracranial hypertension was used in 38.55% of patients with stroke. In this case, PaCO₂ 30—35 mm Hg as a reference value was applied in 26.40% of cases, 30—25 mm Hg — in 12.06% of patients.

Hyperventilation was associated with higher mortality despite certain between-group disparity in clinical severity at the beginning of respiratory support (NIHSS score in patients (n = 441) with hyperventilation was 24 (18; 30) points, in patients (n = 703) without hyperventilation — 22 (16; 28) points (p = 0.001)).

Within-group analysis of patients with NIHSS score ≥ 20 at the beginning of respiratory support revealed that hyperventilation was associated with higher mortality compared to patients without hyperventilation (OR 1.46, 95% CI 1.02 — 2.06, p = 0.0336) (Fig. 4). NIHSS score was similar in groups with (n = 313) and without (n = 422) hyperventilation at the beginning of respiratory support (28 (23; 30) vs. 27 (24; 30) points, p = 0.767). Median of ventilation time in patients with stroke was 5 [3; 10] days. Most (52.36%) patients required respiratory support within 5 days, 18.53% — 6 — 10 days, 7.95% — 11 — 15 days, 6.46% — 16 — 20 days, 8.04% — over 21 days. Duration of respiratory support exceeded 1 month in 2.53% of patients. Ventilation time in survivors is shown in Fig. 5.

Fig. 4. Comparison of mortality in groups with and without hyperventilation (within-group analysis).

Fig. 5. Distribution of survivors depending on duration of mechanical ventilation.

Ventilation within 3 days was performed in 17.49% of cases, 4—7 days — in 33.42% of patients. Prolonged ventilation was required in 49.09% of patients including 10.18% of patients with respiratory support for more than 26 days. There was another duration of ventilation in patients with unfavorable outcome (Fig. 6). Indeed, 45.4% of patients required ventilation for 3 days, 29.78% — 7 days. Prolonged ventilation was performed in 24.82% of patients with subsequent death. Duration of respiratory support was significantly less in patients with IS compared to those with HS (5 [3; 9] vs. 5 [3; 12] days, p = 0.012). SAH required significantly longer ventilation compared to IS (p <0.001) and HS (p = 0.0467). Ventilator-associated pneumonia and tracheobronchitis did not significantly influence duration of weaning of ventilator in patients with stroke.

Fig. 6. Distribution of patients with fatal outcome depending on duration of mechanical ventilation.

Table 3. GOS scores depending on basic ventilation parameters at the beginning of respiratory support

Algorithms of weaning from ventilator are summarized in Table 4. Conventional algorithm with SIMV mode was the most common (89.7%). Adaptive weaning algorithms were used in 38.5% of cases, PS algorithm — in 30.1% of patients. Other algorithms and approaches were applied much less frequently. Duration of weaning from ventilator was similar in different types of stroke. No significant data on advantage of certain algorithm for weaning from ventilator were obtained in comparable groups of patients with stroke.

Table 4. Algorithms (modes) for weaning from respiratory support

Note. IS — ischemic stroke; HS — hemorrhagic stroke; SAH — subarachnoid hemorrhage.

Discussion

In the PubMed database, there are no studies that would substantiate criteria for mechanical ventilation onset in different types of stroke. In our study, we obtained the indications for respiratory support in stroke patients which do not differ from indications described in clinical guidelines for the management of stroke. Combination of impaired consciousness and hypoxemia often indicates a delayed start of respiratory support and/or ineffective therapeutic measures. At the same time, incidence of other indications for ventilation was 25.5% and even 36.4% in ischemic stroke that requires specification considering a large share among all criteria for respiratory support.

It should be noted that despite significantly more aggressive ventilation parameters in hemorrhagic stroke (Table 2), they did not contradict the principles of the concept of "protective" ventilation in the entire sample and in different types of stroke.

We found no data on ventilation strategies in stroke. However, such studies were carried out for other diseases. In particular, there is a meta-analysis of respiratory support strategies in patients without lung damage and/or acute respiratory distress syndrome. This review included data on mechanical ventilation in 575 patients [33]. We excluded 2 (under the letters "E" and "F") out of 6 strategies found in the studies (A — "low" Vt + low PEEP; B — "high" Vt + low PEEP; C — " low " Vt + high PEEP; D — " low " Vt + zero end-expiratory pressure (ZEEP); E — "high" Vt + ZEEP and F — "high" Vt + higher PEEP). PEEP < 10 cm H₂O was defined as low, > 10 cm H₂O — high PEEP. Low tidal volume was defined as ≤ 8 ml/kg, "high" tidal volume — over 8 ml/kg. This meta-analysis showed that strategy C (low Vt + high PEEP) ensured the best PaO₂/FiO₂. Strategy B (“high” Vt + low PEEP) was superior regarding improvement of pulmonary-thoracic compliance. Strategy A ("low" Vt + low PEEP) was associated with less ICU-stay compared to strategies "B", "C", "D". Strategy D (“low” Vt + ZEEP) resulted the lowest PaO₂/FiO₂ and pulmonary-thoracic compliance.

Thus, our data on baseline parameters of respiratory support in patients with stroke are similar to strategy A (maximum PEEP within 5 cm H₂O in 54.63% of patients, 6—10 cm H₂O — 45.1% of patients).

An important conclusion is devoted to no negative effect of moderate PEEP on mortality in patients with stroke that is also consistent with literature data [21, 22].

Respiratory support mode is always discussable in patients with pulmonary injury and intact lungs. Despite certain advantages and disadvantages of VC and PC ventilation, their ratio is approximately 60% / 40%, primarily in case of pulmonary injury and acute respiratory distress syndrome [34—36].

However, the advantages of PC ventilation should be considered in patients with severe acute respiratory distress syndrome and no lung damage [37, 38].

We found that PC ventilation ensures significantly higher survival in stroke patients compared to VC ventilation. Moreover, the advantages of PC ventilation took place in patients with IS and HS. These advantages may be explained by certain disadvantages of VC ventilation (no setting the peak inspiratory pressure, excess intrathoracic pressure to ensure a preset tidal volume). Negative effects of VC ventilation increase the risk of ventilator-associated lung damage, systemic and cerebral hemodynamic disorders. These aspects significantly impair treatment outcomes in patients with stroke. Depression of venous return presenting in this approach can lead to ICP increase.

Timing of tracheostomy remains under discussion. A meta-analysis of 9 studies comprising 2,072 patients has been published [39]. The purpose of this analysis was to compare the results of early (within 10 days after ventilation onset) and delayed tracheostomy (after 10 days) or prolonged tracheal intubation in critically ill patients requiring prolonged respiratory support. Compared to delayed tracheostomy and prolonged tracheal intubation, early tracheostomy did not significantly reduce early mortality (relative risk (RR) 0.91; 95% CI 0.81—1.03; p = 0.14) or long-term mortality (RR 0.90; 95% CI 0.76—1.08; p = 0.27). Moreover, early tracheostomy was not significantly associated with less ICU-stay (weighted mean difference (WMD) –4.41 days; 95% CI –13.44–4.63 days; p = 0.34), incidence of ventilator-associated pneumonia (RR 0.88; 95% CI 0.71—1.10; p = 0.27) or duration of mechanical ventilation (WMD –2.91 days; 95% CI –7.21– 1.40 days; p = 0.19). The authors concluded that early tracheostomy was not significantly superior to delayed tracheostomy and prolonged tracheal intubation in patients requiring prolonged respiratory support. However, this study included various categories of critically ill patients.

In our study, early tracheostomy (within 3 days) was significantly effective regarding subsequent weaning from ventilator. Possible causes of earlier weaning from mechanical ventilation are less risk of respiratory tract contamination with oral infectious agents and higher quality of tracheobronchial tree sanitation [2, 29–31]. At the same time, early tracheostomy had no effect on mortality in patients with stroke.

Some authors analyzed possible positive and negative effects of hyperventilation and hypocapnia as factors influencing cerebral blood flow and intracranial pressure in critically ill patients and in those with severe traumatic brain injury [40–42]. It was found that minimum safe level of hyperventilation that can be used for a short time is 30 mm Hg. We also revealed that hyperventilation was associated with a worse prognosis until the end of acute period of stroke. Thus, these data are completely consistent with the modern concept of severe cerebral vasoconstriction following hyperventilation and PCO₂ decrease. Vascular spasm aggravates cerebral ischemia [43–46].

SIMV mode was the most common for weaning from respiratory support in the Russian Federation (87.9%), but ASV or its analogs (38.5%), as well as PS (30.1%) were widely used too. Our findings are inconsistent with some studies. Indeed, Teismann I.K. et al. [46] analyzed 2 options for weaning from ventilator in 39 patients with stroke. This process required 10.7 ± 7.0 and 8.0 ± 4.5 days in case of intermittent weaning (BIPAP and T-tube) and ASV, respectively (p <0.05).

Conclusion

1. Impaired consciousness (GCS score ≤ 9) and hypoxemia, as well as their combination were the most common indications for respiratory support in patients with stroke.

2. Respiratory support in patients with stroke was generally consistent with the concept of “protective” ventilation and strategy of low Vt with low PEEP. In case of hemorrhagic stroke, more aggressive ventilation parameters were required compared to ischemic stroke.

3. Regardless stroke variant, pressure-controlled ventilation ensured significantly higher survival compared to volume-controlled ventilation.

4. Early tracheostomy (within 3 days) is associated with significantly less duration of weaning from respiratory support in patients with acute cerebrovascular accident.

5. SIMV, ASV or its analogs, as well as PS were the most common modes for weaning from ventilator in the Russian Federation. However, adaptive modes of ventilation did not impact on duration of weaning from respiratory support.

Author contribution

All authors equally participated in the study including development of the concept, collection and analysis of data, writing and editing the manuscript, checking and approving the text.

The authors declare no conflicts of interest.