Search

International audience; Background: Preoperative COVID-19 has been associated with excess postoperative morbi-mortality. Consequently, guidelines were developed that recommended the postponement of surgery for at least 7 weeks after the infection. We hypothesised that vaccination against the SARS-CoV-2 and the large predominance of the Omicron variant attenuated the effect of a preoperative COVID-19 on the occurrence of postoperative respiratory morbidity.Methods: We conducted a prospective cohort study in 41 French centres between 15 March and 30 May 2022 (ClinicalTrials NCT05336110), aimed at comparing the postoperative respiratory morbidity between patients with and without preoperative COVID-19 within 8 weeks prior to surgery. The primary outcome was a composite outcome combining the occurrence of pneumonia, acute respiratory failure, unexpected mechanical ventilation, and pulmonary embolism within the first 30 postoperative days. Secondary outcomes were 30-day mortality, hospital length-of-stay, readmissions, and non-respiratory infections. The sample size was determined to have 90% power to identify a doubling of the primary outcome rate. Adjusted analyses were performed using propensity score modelling and inverse probability weighting.Findings: Of the 4928 patients assessed for the primary outcome, of whom 92.4% were vaccinated against the SARS-CoV-2, 705 had preoperative COVID-19. The primary outcome was reported in 140 (2.8%) patients. An 8-week preoperative COVID-19 was not associated with increased postoperative respiratory morbidity (odds ratio 1.08 [95% CI 0.48-2.13]; p = 0.83). None of the secondary outcomes differed between the two groups. Sensitivity analyses concerning the timing between COVID-19 and surgery, and the clinical presentations of preoperative COVID-19 did not show any association with the primary outcome, except for COVID-19 patients with ongoing symptoms the day of surgery (OR 4.29 [1.02-15.8]; p = 0.04).Interpretation: In our Omicron-predominant, highly immunised population undergoing general surgery, a preoperative COVID-19 was not associated with increased postoperative respiratory morbidity.

Background Hyperoxia is associated with increased morbidity and mortality in the intensive care unit. Classical noninvasive measurements of oxygen saturation with pulse oximeters are unable to detect hyperoxia. The Oxygen Reserve Index (ORI) is a continuous noninvasive parameter provided by a multi-wave pulse oximeter that can detect hyperoxia. Primary objective was to evaluate the diagnostic accuracy of the ORI for detecting arterial oxygen tension (PaO2) > 100 mmHg in neurocritical care patients. Secondary objectives were to test the ability of ORI to detect PaO2 > 120 mmHg and the ability of pulse oximetry (SpO2) to detect PaO2 > 100 mmHg and PaO2 > 120 mmHg. Methods In this single-center study, we collected ORI and arterial blood samples every 6 h for 3 consecutive days. Diagnostic performance was estimated using the area under the receiver operating characteristic curve (AUROC). Results There were 696 simultaneous measurements of ORI and PaO2 in 62 patients. Considering the repeated measurements, the correlation between ORI and PaO2 was r = 0.13. The area under the receiver operating characteristic curve (AUROC), obtained to test the ability of ORI to detect PaO2 > 100 mmHg, was 0.567 (95% confidence interval = 0.566–0.569) with a sensitivity of 0.233 (95%CI = 0.230–0.235) and a specificity of 0.909 (95%CI = 0.907–0.910). The AUROC value obtained to test the ability of SpO2 to detect a PaO2 > 100 mmHg was 0.771 (95%CI = 0.770–0.773) with a sensitivity of 0.715 (95%CI = 0.712–0.718) and a specificity of 0.700 (95%CI = 0.697–0.703). The diagnostic performance of ORI and SpO2 for detecting PaO2 > 120 mmHg was AUROC = 0.584 (95%CI = 0.582–0.586) and 0.764 (95%CI = 0.762–0.766), respectively. The AUROC obtained for SpO2 was significantly higher than that for ORI (p Conclusion In a specific population of brain-injured patients hospitalized in a neurointensive care unit, our results suggest that the ability of ORI to diagnose hyperoxia is relatively low and that SpO2 provides better detection.

Background We determined whether an audit on the adherence to guidelines for hospital-acquired pneumonia (HAP) can improve the outcomes of patients in intensive care units (ICUs). Methods This study was conducted at 35 ICUs in 30 hospitals. We included consecutive, adult patients hospitalized in ICUs for 3 days or more. After a 3-month baseline period followed by the dissemination of recommendations, an audit on the compliance to recommendations (audit period) was followed by a 3-month cluster-randomized trial. We randomly assigned ICUs to either receive audit and feedback (intervention group) or participate in a national registry (control group). The primary outcome was the duration of ICU stay. Results Among 1856 patients enrolled, 602, 669, and 585 were recruited in the baseline, audit, and intervention periods, respectively. The composite measures of compliance were 47% (interquartile range [IQR], 38–56%) in the intervention group and 42% (IQR, 25–53%) in the control group (P = .001). As compared to the baseline period, the ICU lengths of stay were reduced by 3.2 days in the intervention period (P = .07) and by 2.8 days in the control period (P = .02). The durations of ICU stay were 7 days (IQR, 5–14 days) in the control group and 9 days (IQR, 5–20 days) in the intervention group (P = .10). After adjustment for unbalanced baseline characteristics, the hazard ratio for being discharged alive from the ICU in the control group was 1.17 (95% confidence interval, .69–2.01; P = .10). Conclusions The publication of French guidelines for HAP was associated with a reduction of the ICU length of stay. However, the realization of an audit to improve their application did not further improve outcomes. Clinical Trials Registration NCT03348579.

Background From a physiological viewpoint, changes in end-tidal carbon dioxide (EtCO2) could be a simple, noninvasive, and inexpensive way to monitor changes in cardiac index. This study aimed to assess the utility of changes in EtCO2 as a marker of fluid responsiveness after volume expansion in the operating room. Methods A prospective observational study was conducted in a tertiary university teaching hospital, from August 2018 to February 2019. A total of 109 non-consecutive, mechanically ventilated adults undergoing neurosurgery in the supine position with cardiac output monitors were included. Patients with major respiratory disease, arrhythmia, or heart failure were excluded. Volume expansion with 250 ml of saline 0.9% was performed over 10 min to maximise cardiac output during surgery, according to current guidelines. A positive fluid challenge was defined as an increase in stroke volume index of more than 10% from baseline. Changes in stroke volume index (monitored using pulse contour analysis) and EtCO2 were recorded before and after infusion. Results A total of 242 fluid challenges in 114 patients were performed, of which 26.9% were positive. Changes in EtCO2 > 1.1% induced by infusions had utility for identifying fluid responsiveness, with a sensitivity of 62.9% (95% confidence interval [CI], 62.5–63.3%) and a specificity of 77.8% (95% CI, 77.6–78.1%). The area under the receiver operating characteristic curve for changes in EtCO2 after volume expansion was 0.683 (95% CI, 0.680–0.686). Conclusions Changes in EtCO2 induced by rapid infusion of 250 ml saline 0.9% lacked accuracy for identifying fluid responsiveness in mechanically ventilated patients in the operating room. Clinical trial registration NCT03635307.

International audience; Changes in stroke volume (deltaSV) induced by a lung recruitment manoeuvre (LRM) have been shown to accurately predict fluid responsiveness during protective mechanical ventilation. Cardiac output monitors are used in a limited number of surgical patients. In contrast, all patients are monitored with a pulse oximeter, that may enable the continuous monitoring of a peripheral perfusion index (PI). We postulated that changes in PI (deltaPI) may reflect deltaSV during brief modifications of cardiac preload. We studied 47 patients undergoing neurosurgery and ventilated with a tidal volume of 6-8 ml/kg. All patients were monitored with a pulse contour system enabling the continuous monitoring of SV and with a pulse oximeter enabling the continuous monitoring of PI. LRMs were performed by increasing airway pressure up to 30 cmH20 for 30 s. Fluid loads (250 ml of saline 0.9% in 10 min) were performed only in patients who experienced a deltaSV > 30% during LRMs (potential fluid responders). LRMs induced a 26% decrease in SV (p 30% with a sensitivity of 83% and a specificity of 78% (AUC = 0.84, 95%CI 0.71-0.93). 24 patients experienced a deltaSV > 30% and subsequently received fluid. Fluid loads induced a 16% increase in SV and a 17% increase in PI, but fluid-induced deltaPI and deltaSV were weakly correlated (r2 = 0.19). In neurosurgical patients, we conclude that deltaPI may be used as a surrogate for deltaSV during LRMs but not during fluid loading.

Background The aim of the present study was to explore the relationship between creatinine clearance (ClCr), cardiac index (CI) and renal vascular index (RVI) in order to assess the potential mechanisms driving ARC in critically ill trauma patient. The secondary objective was to assess the performance of RVI for prediction of ARC. Methods Every trauma patient who underwent cardiac and renal ultrasound measurements during their initial ICU management was retrospectively reviewed over a 3-month period. ARC was defined by a 24-hr measured ClCr ≥ 130 mL/min/1.73m2. A mixed effect model was constructed to explore covariates associated with ClCr over time. The performance of RVI for prediction of ARC was assessed by receiver operating characteristic (ROC) curve and compared to the ARCTIC (ARC in trauma intensive care) predictive scoring model. Results Thirty patients, contributing for 121 coupled physiologic data, were retrospectively analysed. There was a significant correlation between ClCr values and RVI (r = −0.495; P = 0.005) but not between ClCr and CI values (r = 0.023; P = 0.967) at day 1. Using a mixed effect model, only age remained associated with ClCr variations over time. The area under the ROC curve of RVI for predicting ARC was 0.742 (95% CI: 0.649–0.834; P Conclusion Ultrasonic evaluation of CI and RVI did not allow approaching the haemodynamic mechanisms responsible for ARC in patients. RVI was inaccurate and not better than clinical score for predicting ARC.

International audience; BACKGROUND:In mechanically ventilated patients, an increase in cardiac index during an end-expiratory-occlusion test predicts fluid responsiveness. To identify this rapid increase in cardiac index, continuous and instantaneous cardiac index monitoring is necessary, decreasing its feasibility at the bedside. Our study was designed to investigate whether changes in velocity time integral and in peak velocity obtained using transthoracic echocardiography during an end-expiratory-occlusion maneuver could predict fluid responsiveness.METHODS:This single-center, prospective study included 50 mechanically ventilated critically ill patients. Velocity time integral and peak velocity were assessed using transthoracic echocardiography before and at the end of a 12-sec end-expiratory-occlusion maneuver. A third set of measurements was performed after volume expansion (500 mL of saline 0.9% given over 15 minutes). Patients were considered as responders if cardiac output increased by 15% or more after volume expansion.RESULTS:Twenty-eight patients were responders. At baseline, heart rate, mean arterial pressure, cardiac output, velocity time integral and peak velocity were similar between responders and non-responders. End-expiratory-occlusion maneuver induced a significant increase in velocity time integral both in responders and non-responders, and a significant increase in peak velocity only in responders. A 9% increase in velocity time integral induced by the end-expiratory-occlusion maneuver predicted fluid responsiveness with sensitivity of 89% (95% CI 72% to 98%) and specificity of 95% (95% CI 77% to 100%). An 8.5% increase in peak velocity induced by the end-expiratory-occlusion maneuver predicted fluid responsiveness with sensitivity of 64% (95% CI 44% to 81%) and specificity of 77% (95% CI 55% to 92%). The area under the receiver operating curve generated for changes in velocity time integral was significantly higher than the one generated for changes in peak velocity (0.96 ± 0.03 versus 0.70 ± 0.07, respectively, P = 0.0004 for both). The gray zone ranged between 6 and 10% (20% of the patients) for changes in velocity time integral and between 1 and 13% (62% of the patients) for changes in peak velocity.CONCLUSIONS:In mechanically ventilated and sedated patients in the neuro Intensive Care Unit, changes in velocity time integral during a 12-sec end-expiratory-occlusion maneuver were able to predict fluid responsiveness and perform better than changes in peak velocity.

Background Mini-fluid challenge of 100 ml colloids is thought to predict the effects of larger amounts of fluid (500 ml) in intensive care units. This study sought to determine whether a low quantity of crystalloid (50 and 100 ml) could predict the effects of 250 ml crystalloid in mechanically ventilated patients in the operating room. Methods A total of 44 mechanically ventilated patients undergoing neurosurgery were included. Volume expansion (250 ml saline 0.9%) was given to maximize cardiac output during surgery. Stroke volume index (monitored using pulse contour analysis) and pulse pressure variations were recorded before and after 50 ml infusion (given for 1 min), after another 50 ml infusion (given for 1 min), and finally after 150 ml infusion (total = 250 ml). Changes in stroke volume index induced by 50, 100, and 250 ml were recorded. Positive fluid challenges were defined as an increase in stroke volume index of 10% or more from baseline after 250 ml. Results A total of 88 fluid challenges were performed (32% of positive fluid challenges). Changes in stroke volume index induced by 100 ml greater than 6% (gray zone between 4 and 7%, including 19% of patients) predicted fluid responsiveness with a sensitivity of 93% (95% CI, 77 to 99%) and a specificity of 85% (95% CI, 73 to 93%). The area under the receiver operating curve of changes in stroke volume index induced by 100 ml was 0.95 (95% CI, 0.90 to 0.99) and was higher than those of changes in stroke volume index induced by 50 ml (0.83 [95% CI, 0.75 to 0.92]; P = 0.01) and pulse pressure variations (0.65 [95% CI, 0.53 to 0.78]; P < 0.005). Conclusions Changes in stroke volume index induced by rapid infusion of 100 ml crystalloid predicted the effects of 250 ml crystalloid in patients ventilated mechanically in the operating room.

Objective: continuous monitoring of cerebral oxygenation is part of the therapeutic tools for brain damaged patients. Direct and invasive measurement of cerebral oxygenation using a probe (PbtO2) is promising but invasive. A non-invasive evaluation of regional transcranial oxygen saturation using near-infrared spectroscopy (NIRS) may be possible. The aim of this study was to assess the interchangeability between PbtO2 and NIRS over time in brain damaged patients. Methods: this is a retrospective study conducted in the neuro-intensive care unit of Bordeaux University Hospital. Fifty-eight severely brain-injured patients were equipped with a PbtO2 probe and a homolaterally NIRS sensor for simultaneous hourly measurements of PbtO2 and NIRS values for 72 consecutive hours. Measurements and Main Results: a total of 3988 simultaneous measurements of NIRS and PbtO2 was obtained. Point-wise correlation ranged from -0.25 and 0.25. Global correlation over time was 0.013 (p= 0.457). The bivariate linear-mixed model confirmed no significant correlation between the Pbt02 and NIRS follow-up. The concordance rate between rSO2 and PbtO2 was 41.8%. NIRS was not able to detect PbtO2 values less than 20 mmHg (Area Under Receiver Operating curve = 0.537 ([0.534 - 0.539]; p= 0.664) and percentage changes in NIRS was not able to detect a decrease in PbtO2 10% (Area Under Receiver Operating curve = 0.615 [0.614 - 0.616], p < 0.001). Conclusion: PbtO2 and NIRS measurements were not correlated. There is no evidence that NIRS could constitute a substitute for monitoring of PbtO2 at bedside in neurocritical care unit patients.; Introduction : chez les patients cérébrolésés, le monitorage multimodal, dont le monitorage de l’oxygénation cérébrale à travers la mesure de la pression tissulaire en oxygène (PtiO2) est maintenant admis comme améliorant le pronostic. Néanmoins, le caractère invasif de la PtiO2 nous a incité à étudier sa corrélation avec un dispositif non invasif : la saturation cérébrale en oxygène (rSO2) mesurée par NIRS, et ainsi déterminer si elles pouvaient être interchangeables. Matériel & méthodes : il s’agissait d’une étude observationnelle rétrospective menée dans l’unité de neuroréanimation du centre hospitalier universitaire de Bordeaux. Les sujets ont bénéficié de la mesure horaire simultanée de la PtiO2 et de la rSO2 pendant 72 heures. Résultats : un total de 3988 mesures simultanées de NIRS et de PtiO2 a été obtenu. La corrélation ponctuelle était comprise entre -0,25 et 0,25. La corrélation globale dans le temps était de 0,013 (p= 0,457). Le modèle bivarié linéaire-mixte a confirmé l'absence de corrélation significative entre la PtiO2 et la rSO2. Le taux de concordance entre la rSO2 et la PtiO2 était de 41,8 %. La NIRS n'était pas capable de détecter des valeurs de PtiO2 inférieures à 20 mmHg (Area Under Receiver Operating curve = 0,537 ([0,534 - 0,539] ; p= 0,664) et les variations en pourcentage de rSO2 ne détectaient pas une diminution de 10 % de la PtiO2 (Area Under Receiver Operating curve = 0,615 [0,614 - 0,616], p < 0,001). Conclusion : la PtiO2 et la rSO2 ne sont pas corrélées au cours du temps, probablement parce qu’elles ne mesurent pas le même processus biologique. Il n’y a aucune preuve que la NIRS pourrait constituer un substitut à la surveillance de PtiO2 au chevet des patients en neuro-réanimation.

Background: many maneuvers assessing fluid responsiveness (minifluid challenge, lung recruitment maneuver, endexpiratory occlusion test, passive leg raising) are considered as positive when small variations in cardiac index, stroke volume index, stroke volume variation or pulse pressure variation occur. Pulse contour analysis allows continuous and real-time cardiac index, stroke volume, stroke volume variation and pulse pressure variation estimations. To use these maneuvers with pulse contour analysis, the knowledge of the minimal change that needs to be measured by a device to recognize a real change (least significant change) has to be studied. The aim of this study was to evaluate the least significant change of cardiac index, stroke volume index, stroke volume variation and pulse pressure variation obtained using pulse contour analysis (ProAQT ®, Pulsion Medical System, Germany). Methods: in this observational study, we included 50 mechanically ventilated patients undergoing neurosurgery in the operating room. Cardiac index, stroke volume index, pulse pressure variation and stroke volume variation obtained using ProAQT ® (Pulsion Medical System, Germany) were recorded every 12 s during 15-min steady-state periods. Least significant changes were calculated every minute. Results: least significant changes statistically differed over time for cardiac index, stroke volume index, pulse pressure variation and stroke volume variation (p < 0.001). Least significant changes ranged from 1.3 to 0.7% for cardiac index, from 1.3 to 0.8% for stroke volume index, from 10 to 4.9% for pulse pressure variation and from 10.8 to 4.3% for stroke volume variation. Conclusion: to conclude, the present study suggests that pulse contour analysis is able to detect rapid and small changes in cardiac index and stroke volume index, but the interpretation of rapid and small changes of pulse pressure variation and stroke volume variation must be done with caution.; Introduction : plusieurs procédures utilisées pour déterminer la précharge-dépendance (« Minifluid » challenge, manœuvre de recrutement alvéolaire, pause télé-expiratoire, lever de jambes passif) sont considérées positives quand de petites variations dans l’index cardiaque (IC), le volume d’éjection systolique indexé (VEI), ou la variation respiratoire du volume d’éjection systolique (VVE) ou de la pression pulsée (VPP) sont notées. L’analyse du contour de l’onde de pouls permet une estimation en continu et en temps-réel de ces paramètres. Pour utiliser ces procédures avec l’analyse du contour de l’onde de pouls, la connaissance du changement minimal devant être mesuré par un appareil pour reconnaître un changement réel d’un paramètre (plus petit changement significatif ou least significant change, LSC) doit être étudiée. Le but de cette étude est d’évaluer le LSC de l’IC, du VEI, de la VVE et de la VPP en utilisant l’analyse du contour de l’onde de pouls (ProAQT ®, Pulsion Medical System, Germany). Méthodes : dans cette étude observationnelle, nous avons inclus 50 patients sous ventilation mécanique au bloc opératoire subissant une intervention de neurochirurgie. L’IC, le VEI, la VVE et la VPP obtenus en utilisant le ProAQT ® (Pulsion Medical System, Germany) ont été enregistrés toutes les 12 s pendant des périodes de stabilité hémodynamique de 15 min. Les LSC ont été calculés chaque minute. Résultats : les LSC différent de manière statistiquement significative chaque minute pour l’IC, le VEI, la VVE et la VPP (p < 0.001). Les LSC varient de 1,3 à 0,7% pour l’IC, de 1,3 à 0,8% pour le VEI, de 10 à 4,9% pour la VPP et de 10,8 à 4,3% pour la VVE. Conclusion : cette étude suggère que l’analyse du contour de l’onde de pouls est capable de détecter des variations rapides et de faible intensité pour l’IC et le VEI, mais l’interprétation des petits changements rapides des VVE et VPP doivent être prudentes.

Background: from a physiological viewpoint, changes in end-tidal carbon dioxide (EtCO2) could be a simple, noninvasive, and inexpensive way to monitor changes in cardiac index. This study aimed to assess the utility of changes in EtCO2 as a marker of fluid responsiveness after volume expansion in the operating room. Methods: a prospective observational study was conducted in a tertiary university teaching hospital, from August 2018 to February 2019. A total of 109 non-consecutive, mechanically ventilated adults undergoing neurosurgery in the supine position and equipped with cardiac output monitors were included. Patients with major respiratory disease, arrhythmia, or heart failure were excluded. Volume expansion with 250 ml of saline 0.9% was performed for 10 min to maximise cardiac output during surgery, according to current guidelines. A positive fluid challenge was defined as an increase in stroke volume index of more than 10% from baseline. Changes in stroke volume index (monitored using pulse contour analysis) and EtCO2 were recorded before and after infusion. Results: a total of 242 fluid challenges were performed (26.9% positive challenges). Changes in EtCO2 greater than 1.1% induced by infusions had utility for identifying fluid responsiveness, with a sensitivity of 62.9% (95% confidence interval, 62.5-63.3%) and a specificity of 77.8% (95% CI, 77.6-78.1%). The area under the receiver operating curve for changes in EtCO2 after volume expansion was 0.683 (95% CI, 0.680-0.686). Conclusions: changes in EtCO2 induced by rapid infusion of 250 ml saline 0.9% lacked accuracy for identifying fluid responsiveness in mechanically ventilated patients in the operating room., Contexte : physiologiquement, les variations de la fraction expirée en CO2 (EtCO2) après un remplissage vasculaire pourraient permettre, de manière simple, non invasive et peu coûteuse, de monitorer les variations de l'index cardiaque (IC). Cette étude évaluait l'utilité des variations de l'EtCO2 en tant que marqueur de la réponse au remplissage vasculaire à la suite d'une expansion volémique au bloc opératoire. Méthodes : une étude observationnelle prospective, menée au CHU de Bordeaux d'août 2018 à février 2019 a inclus des patients ventilés mécaniquement, opérés d’une neurochirurgie en décubitus dorsal et équipés d’un monitorage du débit cardiaque. Les patients souffrant de maladies respiratoires graves, d'arythmie ou d'insuffisance cardiaque ont été exclus. Une expansion volémique de 250mL de sérum salé à 0,9% a été effectuée sur 10 minutes, afin de maximiser le débit cardiaque pendant l'opération, conformément aux recommandations actuelles. Une épreuve positive de remplissage a été définie comme une augmentation du volume d’éjection systolique (VES) de plus de 10% par rapport au niveau de référence. Les variations de VES (monitorées par analyse du contour de l’onde de pouls) et de l'EtCO2 ont été enregistrées avant et après la perfusion. Résultats : un total de 242 épreuves de remplissage vasculaire ont été réalisées chez 109 patients non consécutifs (26,9% d'épreuves positives). Les variations de l'EtCO2 supérieures à 1,1% induites par ces perfusions ont permis d'identifier la réponse au remplissage vasculaire, avec une sensibilité de 62,9% (IC 95% : 62.5-63.3%) et une spécificité de 77,8% (IC 95% : 77.6-78.1%). L'aire sous la courbe ROC des variations de l'EtCO2 après expansion volémique était de 0,683 (IC 95% : 0,680-0,686). Conclusions : Les variations de l'EtCO2 induites par la perfusion rapide de 250mL de sérum salé à 0,9% manquaient de précision pour identifier la réponse au remplissage vasculaire chez les patients sous ventilation mécanique au bloc opératoire. Contexte : physiologiquement, les variations de la fraction expirée en CO2 (EtCO2) après un remplissage vasculaire pourraient permettre, de manière simple, non invasive et peu coûteuse, de monitorer les variations de l'index cardiaque (IC). Cette étude évaluait l'utilité des variations de l'EtCO2 en tant que marqueur de la réponse au remplissage vasculaire à la suite d'une expansion volémique au bloc opératoire.

Background: tracheostomy in the intensive care unit is not consensual. Specifically, in the neuro intensive care unit, protecting upper airways is challenging every day, in regards to altered conscience state and deglutition troubles. Tracheostomy could improve ventilatory mechanics in these patients. Factors associated to mechanical ventilation duration are not well defined. Our hypothesis was that tracheostomy could reduce mechanical ventilation duration. Methods: it was a retrospective study at neuro intensive care unit of the university hospital of Bordeaux. Between may 2016 and April 2018, 316 patients receiving mechanical ventilation for at least 72 hours in our neuro intensive care unit were retrospectively included, among them 50 were tracheostomized. Data collection included demographic variables, gravity scores, mechanical ventilation duration, length of stay in the intensive care unit. Multivariable analysis has been realized to study mechanical ventilation duration associated factors. Results: median mechanical ventilation duration in tracheostomized patients was 24,9 days [16,2-30,0] whereas intubated patients have been ventilated 9,5 days [6,0 - 10,9]. On the day they arrived, tracheostomized patients experienced more organ failure in regards to intubated patients (median SOFA score 8,0 [6,0 - 9,0] and 6,0 [5,0 - 8,0], p value = 0,038). Multivariable analysis made the tracheostomy an independent factor augmenting mechanical ventilation duration. We found only one mild complication linked to percutaneous tracheostomy procedure. Conclusion: percutaneous tracheostomy in the neuro intensive care unit make mechanical ventilation duration last longer. Weaning from mechanical ventilation protocol in those patients should be assessed., Contexte : l’indication de la trachéotomie en réanimation n’est pas consensuelle. Dans le cadre de la neuroréanimation, la protection des voies aériennes supérieures est une question de tous les jours, du fait de l’altération de l’état de vigilance et des troubles de la déglutition. La trachéotomie pourrait améliorer la mécanique ventilatoire de ces patients. Les facteurs associés à la durée de ventilation mécanique dans notre population ne sont pas connus. Notre hypothèse était que la trachéotomie permettrait de diminuer la durée de ventilation mécanique invasive. Matériels et Méthodes : il s’agissait d’une étude rétrospective réalisée dans la neuroréanimation au CHU de Bordeaux. Entre mai 2016 et avril 2018, nous avons inclus rétrospectivement 316 patients majeurs ventilés au moins 72 heures, dont 50 patients trachéotomisés. Nous avons recueilli les données démographiques, les scores de gravité, la durée de ventilation mécanique invasive, la durée de séjour pour tous les patients. Une analyse multivariable a été réalisée pour rechercher les facteurs liés à la durée de ventilation mécanique invasive. Résultats : la durée médiane de ventilation mécanique invasive des patients trachéotomisés était de 24,9 jours [16,2-30,0] tandis que celle des patients intubés au long cours était de 9,5 jours [6,0 - 10,9]. Les patients trachéotomisés présentaient plus de défaillances d’organes à l’entrée (score SOFA médian 8,0 [6,0 - 9,0] contre 6,0 [5,0 - 8,0], p valeur = 0,038). L’analyse multivariable ajustée sur les facteurs confondants montrait que la trachéotomie allongeait de manière indépendante la durée de ventilation mécanique. Nous n’avons retrouvé qu’une seule complication non grave parmi les patients trachéotomisés par voie percutanée. Conclusion : la trachéotomie percutanée en neuroréanimation allonge la durée de ventilation mécanique. La mise en place de protocole de sevrage ventilatoire devra être évaluée pour réduire la durée de ventilation des patients trachéotomisés.