Introduction

Bilevel positive airway pressure (BLPAP) ventilation was introduced in 1990 [1, 2]. BLPAP ventilators are relatively inexpensive and much simpler than conventional ventilators [2]. They offer good performance [3], are currently used to deliver noninvasive ventilation (NIV) in various settings at home or in intensive care units (ICU) [49], and have occasionally also been used in intubated or tracheostomized patients [10].

The BLPAP circuit consists of single-limb tubing for inspiration and expiration, including a passive intentional leak port located either at its distal end or on the mask [11]. During expiration, exhaled gases leave the circuit through the leak port but may also more or less fill the circuit beyond the port and then become part of the following delivered tidal volume (VT). This defines an additional rebreathing process specific to BLPAP ventilation.

Rebreathing risk has already been studied and only negatively related to the level of expiratory positive airway pressure (EPAP) [1214]. Since CO2 rebreathing increases the drive to ventilate and the work of breathing [12, 15, 16], it is important to clarify this risk.

Our aim was to assess the actual risk of additional rebreathing and its predictive factors in patients receiving BLPAP ventilation in ICU.

Materials and methods

This observational study was performed during routine care in a 15-bed adult ICU. Ethical approval was obtained from the local institutional review board. Patients gave informed consent and the study was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki.

Patients

Eighteen consecutive patients who required mechanical ventilation were studied during their weaning period. All patients were stable and conscious. They were either still intubated or under NIV using an inflatable air cushion oronasal Tyco™ mask (Mallinckrodt DAR, Mirandola, Italy).

Ventilator and settings

The BLPAP ventilator (Vision®; Respironics Inc., Murrysville, PA, USA) was used in pressure support mode. This ventilator is routinely utilized in our ICU either for NIV or in intubated patients during the weaning trial when the necessity for postextubation NIV is anticipated. It allows one to set the inspiratory fraction of oxygen (FiO2), displays flow, pressure, and volume curves, and is alarm-equipped. The circuit consists of a single limb for inspiration and expiration with an exhalation port at its distal end (Respironics Whisper Swivel device, Murrysville, PA, USA). Ventilator settings were those routinely used. Inspiratory positive airway pressures (IPAP) were chosen to deliver VT of 8–12 ml/kg. In case of intrinsic positive end-expiratory pressure (PEEPi), detected on the ventilator screen, EPAP was adjusted to obtain zero end-expiratory flow without exceeding 9 cmH2O. For patients with no PEEPi, 4 cmH2O EPAP (the lowest possible with that ventilator) was applied. FiO2 was adjusted to maintain oxygen saturation from 90% to 95%.

Measurements

One recording, lasting from 20 to 45 min, was obtained from each patient using the Biopac® MP 100 apparatus (Biopac Systems Inc., Santa Barbara, CA, USA). Calibrated pressure, airflow, and CO2 (aspirating capnograph device) were sampled at a position 2 cm away from the port, between the exhalation port and the patient. The delay representing transit time from the sample point to the analyzer was measured beforehand. Signals were recorded at a frequency of 200 Hz.

Analyses were made breath by breath, after phasing CO2 curves with flow and pressure, using Biopac® AcqKnowledge® 3.7 software. The beginning and end of each inspiration were defined at the points where flow crossed the zero line. For each breath, inspiratory positive airway pressure (IPAP), EPAP, inspiration time (T i), expiration time (T e), respiratory rate (RR), inspiratory tidal volume (VTi) as the integral of the surface under the inspiratory flow, and end-tidal CO2 concentration (ETCO2) were measured. The ratio of T i/(T i + T e), called T i/T TOT, was calculated. The volume of CO2 delivered from the circuit at each breath was calculated by integrating the CO2 flow curve, defined as the product of inspiratory airflow and CO2 concentration curves [17]. Mean inspiratory fraction of CO2 for each tidal volume (tidal FiCO2), expressed as a percentage, was defined as: inspired volume of CO2/VTi × 100. An estimation of the additional dead space volume (VDadd) from each previous expiration accumulated in the circuit between the port and the ventilator was assessed as: VTi × tidal FiCO2/ETCO2.

Statistical analysis

Data are reported in count and proportion, or as mean ± standard deviation (SD), as appropriate. Each breath was classified according to tidal FiCO2 (exceeding or within 0.10%). Univariate comparisons were performed using the unpaired Student t test for continuous variables, and a χ 2 test for categorical variables. Logistic regression analysis, with tidal FiCO2 as the binomial dependent variable and parameters, both of which were significantly different by univariate analysis and clinically relevant as independent factors, were performed to find the best fitted model. Odds ratio (OR) and 95% confidence interval (95% CI) were determined. Differences were considered significant when P < 0.05.

Results

The 18 patients (13 males, mean age 54 ± 19 years) had been mechanically ventilated for 8.2 ± 6.6 days. Eleven patients (61%) were intubated and seven (39%) were under NIV. A total of 11,107 respiratory cycles were analyzed. Of the cycles measured, 8,976 (81%) had FiCO2 below 0.10% and 2,131 (19%) had FiCO2 exceeding 0.10%.

Values and univariate comparisons are presented in Table 1. Results of the logistic regression analysis are presented in Table 2. Four parameters had good predictive value for tidal FiCO2 exceeding 0.10%: high ETCO2, low EPAP, intubation, and high RR.

Table 1 Breath characteristics for all cycles and according to rebreathing
Table 2 Odds ratio to predict tidal FiCO2 (cutoff 0.10%)

Discussion

In this study, we accurately measured the additional CO2 rebreathing from the circuit, specifically induced with BLPAP ventilation. Our results underline the reality of a modest amount of rebreathing under a BLPAP ventilator in the clinical setting. Furthermore, we report for the first time the respective contribution of the different possible factors associated with CO2 rebreathing in ICU.

We have compared a large number of individual breaths, split into two groups according to a low or higher rebreathing with a cutoff of tidal FiCO2 at 0.10%. The comparison of individual breaths, rather than for example patients, is appropriated since the amount of rebreathing may depend on numerous factors, such as VT, EPAP, and timing of the respiratory cycle (RR and T i/T TOT). Most of these exhibit large variations (in the same patient) on a breath-by-breath basis during spontaneous breathing with pressure support. The choice of the cutoff is based on the consideration that such a value corresponds to 0.8 mmHg when the atmospheric pressure is 760 mmHg. Even if apparently low, 1 mmHg is enough to stimulate ventilation during pressure support ventilation [15, 16].

According to logistic regression analysis, we observed four pertinent factors that correlate with the magnitude of rebreathing. First and foremost was high ETCO2, which correlates with increased possibility that tidal FiCO2 will exceed 0.10%. Since ETCO2 approximately corresponds to the concentration of CO2 filling the circuit it is quite normal that, as ETCO2 increases, so does tidal FiCO2. Second, high EPAP correlated with increased likelihood that tidal FiCO2 will be below 0.10%. This confirms previously published data which show that EPAP up to 8 cmH2O may be necessary to eliminate rebreathing [13] and that low EPAP promotes significant rebreathing able to stimulate ventilation and increase the work of breathing [12]. Third, the type of connection is an important predictive factor, since less rebreathing takes place in the course of NIV compared with translaryngeal intubation. Incidental leakage occurring at the skin–mask junction will decrease rebreathing from the circuit dead space, including the volume of the mask. Such a risk of leakage does not exist with intubation, provided the cuff is adequately inflated. In the same way, amplified leakage via the oral cavity during nocturnal nasal mask ventilation explains the uselessness of a non-rebreathing exhalation valve during BLPAP ventilation [14]. Fourthly, high RR was also a predictive factor for rebreathing in this study. Its influence has not been pointed out before. It may be related to the fact that too short a duration of expiration will be insufficient to completely flush expiratory gas from the circuit. In addition, VTi was also statistically significant, with an OR very close to 1 without clinical importance, whereas T i/T TOT had no influence on rebreathing.

The general findings of our study might be applied to all other BLPAP devices, circuits, exhalation ports, and interfaces, with the magnitude of additional rebreathing inversely proportional to the surface of the port and its distance from the patient [11].

We conclude that BLPAP ventilation actually presents a modest specific risk for rebreathing due to circuit leakage particular to the BLPAP ventilator design. That risk applied to about 10% and 20% of breaths, provided EPAP is applied, in NIV and with intubation, respectively. These data also show that BLPAP ventilation may be reasonably used in intubated patients. Recommendations to decrease the amount and incidence of rebreathing are use of EPAP (4–8 cmH2O) and avoiding a rapid RR.