Short-term effects of noisy pressure support ventilation in patients with acute hypoxemic respiratory failure

This monocenter, randomized crossover study was approved by the institutional ethics review board of the University Hospital Dresden, Germany (EK 276112007) and registered at ClinicialTrials.gov (NCT00786292). Thirteen patients from the anesthesiological and surgical intensive care units of the University Hospital Dresden who were diagnosed with acute hypoxemic respiratory failure were enrolled in this study. Written informed consent was obtained from patient’s legal representative. Patients were included in the study if all of the following criteria were fulfilled: a) age between 18 and 75 years; b) ratio of arterial partial pressure of oxygen and inspired oxygen fraction (PaO₂/F_IO₂) between 150 and 300 mmHg; total duration of mechanical less than 14 days; mechanical ventilation with biphasic positive airway pressure ventilation with at least 20% of minute ventilation corresponding to spontaneous breathing, or assisted mechanical ventilation with pressure support ventilation, at time of inclusion. Patients were excluded from the study if at least one of the following criteria was present: i) body mass index > 35; ii) esophageal disease; iii) neuromuscular disease; iv) instable thorax; v) pneumothorax; brain injury/increased intracranial pressure; vi) brain tumor; vii) agitation; viii) increased need for vasopressors or hemodynamic drug support, defined as dopamine or dobutamine dosage > 5 μg/kg/min, noradrenaline > 2 μg/kg/min, or use of vasopressin or milrinone; ix) chronic pulmonary disease; x) acute coronary insufficiency; xi) participation in other clinical trials in the preceding four weeks.

The study intervention was mechanical ventilation with conventional or variable PSV for one hour each in all patients, in a randomized fashion. One of two envelopes containing either conventional or variable PSV was drawn to determine the sequence of mechanical ventilation modes.

After preparing the measurement devices, patients were switched to the study ventilator and a stabilization period of 20 minutes was allowed prior to baseline measurements. Following baseline measurements, the sequence of conventional and noisy PSV was randomized and the patients were ventilated for one hour with each mode. The study was interrupted if one of the following criteria was fulfilled: 1) acute change of mental status; 2) diaphoresis; 3) exacerbated movement of intercostal muscles; 4) dyspnea; 5) paradoxical abdominal breathing; 6) respiratory rate >30/min or <6/min; g) arterial pH <7.30; 7) heart rate increase >20% compared to baseline, or <50/min, or >110/min (absolute); 8) mean arterial pressure (MAP) increase >20% compared to baseline, or <70 mmHg, or >110 mmHg (absolute).

After finishing the study protocol, patients were switched back to their initial ventilator and therapy was continued at the discretion of the treating physician.

All patients were monitored with electrocardiography, pulse oxymetry and invasive arterial blood pressure measurement according to the current standard of the intensive care units. Additionally, an esophageal balloon catheter (Erich Jäger, Höchberg, Germany) and electrical impedance tomography (EIT Evaluation Kit II, Dräger Medical, Lübeck, Germany) were used for study specific measurements in case of the absence of contraindications to these procedures. Epidemiological data, current therapy and diagnosis were reviewed and captured using the intensive care units’ electronic patient data management system (ICM, Dräger Medical).

All patients were already on assisted spontaneous breathing prior to the start of the study, and laid in the supine and semirecumbent 45° position. A commercial intensive care ventilator (Dräger Evita XL, Dräger Medical) was used for both PSV and noisy PSV. To perform noisy PSV, the ventilator was remote controlled by an external computer using the Dräger MediBus protocol as previously described [10]. During noisy PSV, pressure support values were generated randomly and followed a Gaussian (normal) distribution, whereby the coefficient of variation (100 × standard deviation (SD)/mean value) was 30%, as described in detail in a previous study from our group [12]. The ventilator settings and alarm limits are defined in Table 1. Briefly, pressure support was titrated to reach tidal volumes (V_T) of approximately 8 ml/kg ideal body weight. We chose that level since it represents the upper limit of V_T still compatible with protective ventilation. The level of positive end-expiratory pressure (PEEP) and fraction of inspired oxygen (F_IO₂) were kept unchanged according the current therapy. The flow trigger was set at 3 L/min and cycling-off criteria at 25% of peak inspiratory flow in all patients, for both conventional and noisy PSV.

Gas exchange variables (PaO₂/F_IO₂; arterial partial pressure of carbon dioxide (PaCO₂)) were measured by point of care blood gas analysis (ABL800, Radiometer, Copenhagen, Denmark). Hemodynamic variables (heart rate (HR), MAP) were recorded from the hemodynamic monitoring system. Airway flow signals were continuously recorded from the mechanical ventilator at a sample frequency of 125 Hz using the MediBus interface to calculate V_T by means of numerical integration of the flow signal over time. Airway (peak and mean airway pressures (P_aw peak, P_aw mean)) and esophageal pressures were measured, digitized and recorded using previously described routines [15]. EIT was used to measure relative changes in thoracic impedance as a surrogate for the spatial distribution of ventilation. EIT data was recorded and stored in the device and processed offline using a routine developed by our group. The transpulmonary pressure (P_L) was calculated as the difference between airway and esophageal pressure. Pressure time product (PTP) was calculated by numerical integration of esophageal pressure and inspiratory time. The coefficient of variation of V_T (CV V_T) was calculated as the ratio of the standard deviation and the mean of V_T. Patient-ventilator asynchrony was assessed from respiratory signals and quantified as described elsewhere [16]. Briefly, we computed the occurrence of ineffective triggering (single patient effort fails to trigger the ventilator and hence no pressure support is delivered), double triggering (ventilator triggered twice without an expiration between cycles), as well as double inspiratory effort (patient had two inspiratory efforts, just one effort triggered the ventilator), and values were added to obtain the total number of asynchrony events.

Values are given as mean and SD unless stated otherwise. Differences between noisy and conventional PSV were tested using paired t tests, or Wilcoxon signed-rank test, as appropriate. EIT data were compared using repeated measures two-way analysis of variance (ANOVA) (factors: group and region) with Bonferroni adjustment. All tests were performed using GraphPad Prism (Version 5a, GraphPad Software, La Jolla, CA, USA). Statistical significance was accepted at P <0.05.

This study has several limitations. First, it was designed as a randomized crossover study aimed at testing the safety and feasibility, as well as physiological effects, of the newly developed ventilator mode noisy PSV. Therefore, we did not obtain long-term information or clinically relevant outcome variables. Second, since most of our patients were classified as moderate acute hypoxemic respiratory failure, the effects of noisy PSV in severe or mild as well as lung-healthy patients remain to be determined. Third, we investigated a selected population of patients, mainly in the postoperative period, and our results cannot be extrapolated to other situations. Fourth, we did not compare noisy PSV with other modes capable of increasing the variability of the breathing pattern, namely NAVA and PAV. Fifth, noisy PSV was performed with a coefficient of variation of 30% in pressure support in all patients, which resulted in a similar coefficient of variation in V_T. Thus, possible individual needs may have been overlooked. However, such coefficient of variation has been shown to deliver the best compromise between gas exchange and lung mechanics in a model of ARDS [12]. Sixth, patient-ventilator asynchrony events have been registered from respiratory signals, including esophageal pressure. Thus, we were not able to quantify the asynchrony between the electric activation of the diaphragm and pressure support cycling-in and cycling-off. Furthermore, the counts of events could not be made in a blinded fashion, since tracings from noisy PSV and PSV differ considerably, and the mechanisms for such findings could not be addressed. Thus, our findings are indicative of an association.