Study approval was obtained from the ethical commission of the University Hospital Basel Switzerland. Patients gave written informed consent prior to the study. The inclusion criteria were chronic hypercapnic respiratory failure (i.e., hypoventilation syndrome), age>18 years, patients receiving NPPV for at least six months and compliance with NPPV (>5 hours/night). All patients had documented improvement of daytime blood gases, nocturnal oximetry and symptoms of hypoventilation syndrome under current treatment with NPPV. Prior to the titration study, overnight PcCO₂ measurements were undertaken in another six patients to assess the extent of signal drift in the PcCO₂ measurement during the recorded period. The mean signal drift for PcCO₂ was –0.7mm Hg (±0.3) per hour. Therefore, a correction factor (0.7 x the hours of recording) was derived to estimate the total signal drift and was added to the awake PcCO₂ value to obtain the drift corrected awake PcCO₂ value.

Complete polysomnography (Poly-MEPAL, MAP AG, Germany) was performed in all patients and in addition cutaneous carbon dioxide ear sensor (Sentec AG, Therwil, Switzerland) was placed on the ear lobe as described elsewhere [7,8]. A baseline arterial blood gas was performed, once equilibration of the PcCO₂ was reached (typically 5 minutes). Similarly, another blood gas was obtained in the morning at end of the study. Data from pre-study and post study calibrations on individual patients were used to calculate the actual drift of the PcCO₂ sensor during the study period. The study night was split in two parts, the first part during which the patients would try to sleep without NPPV for three hours and in the second part using the PcCO₂ as a guide to titrate NPPV pressures [11]. If the patient had difficulties in sleeping without NPPV, it was offered before the end of the run in period [11]. NPPV was started with an Inspiratory Positive Airway Pressure (IPAP) of 7cmH₂O, and an Expiratory Positive Airway Pressure (EPAP) of 4cmH₂O. The pressures were adjusted at an interval of approximately 30 minutes with the goal of reducing the PcCO₂ by 10 to 15mmHg from baseline measurement during sleep without NPPV or achieve a PcCO₂ between 45 and 50mmHg, whichever occurred first [11]. This target was reached by increasing IPAP (increasing IPAP EPAP difference). EPAP was increased to prevent obstructive apneas and hypopneas as assisted by the polysomnography [12]. The physician deciding about the changes in ventilation pressures was blinded to the pre-existing ventilator settings. Epworth sleepiness score was recorded prior to the study and at follow-up. All patients charted overall satisfaction with home ventilation with regards to symptoms and use of the NPPV on a 10-centimeter visual analogue scale prior to the study and at a follow-up clinic visit six to eight weeks after the study.

Simple descriptive statistics were used [13]. Data were presented as mean (± standard deviation). Pearson’s correlation coefficient was used to assess the correlation between the PcCO₂ and PaCO₂.

The mean age of the 12 patients was 62 years (range 42 to 73 years) and eight were males. The cause of chronic respiratory failure included different etiologies {(e.g., obesity hypoventilation syndrome, neuromuscular disease, chest wall deformities, Chronic Obstructive Pulmonary Disease (COPD)} [Table/Fig-1]. The baseline clinical characteristics of the patients prior to initiation of NPPV pressures are presented in [Table/Fig-1]. The mean baseline PcCO₂ value at start of the study was 45.4 mm Hg (±6.5) and drift corrected for awake PcCO₂ value was 45.1 mm Hg (±8.3). The correlation between the baseline PcCO₂ at start of the study and parallel PaCO₂ was R = 0.91 (95% confidence intervals 0.68 to 0.98). The correlation between the predicted awake PcCO₂ and the parallel awake PaCO₂ was R = 0.85 (95% confidence intervals 0.48 to 0.96). The mean sensor drift during the study period was –0.8 (± 0.8) mmHg/hour. There were no local complications such as skin redness or burns related to the use of the combined sensor.

As a result of the titration study IPAP pressures were changed in nine patients (increased in five; decreased in four) and EPAP in eight patients (increased in four; decreased in four). The mean change in IPAP pressure after the study was 2.3 ± 2.1 cm H₂0 [Table/Fig-2] and EPAP pressure was 1.5 ± 1.4 cm H₂0 [Table/Fig-3]. At follow-up, no changes were made to the EPAP pressures; however, minor changes were made to IPAP pressures (not to the pre-study levels) on request for three patients. Patient FC felt that the IPAP pressure of 12cm H₂0 was too low as he was used to a pressure of 18 cm H₂0. However, at an IPAP of 12cm H₂0 nocturnal oximetry and day time blood gases were normal and not different from the results with IPAP 18cm H₂0. Patients CC and RS felt that the new pressures were too high for them to tolerate. The mean Epworth sleepiness score prior to the study was 8.8 (±3.8) and post study was 8.2 (±3.5) [Table/Fig-4]. The mean visual analogue score prior to the study was 7.0 (±1.5) and post study was 7.4 (±1.4) [Table/Fig-5]. The arterial blood gas values during wakefulness on NPPV treatment prior to the study and at follow-up clinic visit after the study are presented in [Table/Fig-6].

[Table/Fig-2]:

Comparison of inspiratory positive airway pressure before and after study.

[Table/Fig-3]:

Comparison of expiratory positive airway pressure before and after study.

[Table/Fig-4]:

Comparison of Epworth sleepiness score before and after study.

[Table/Fig-5]:

Comparison of visual analogue scale for patient satisfaction with home ventilation before and after study.

The mean duration of polysomnography recording on NPPV was 322 minutes (± 58 minutes). As most patients could not sleep well without NPPV, further polysomnography data are presented for the time spent on the mask receiving NPPV. The total sleep time was 216 minutes (± 42 minutes) and sleep efficacy was 70% (±12%) respectively. The mean duration of rapid eye movement sleep was 16% (±10%). The mean oxygen saturation was 91% (±3%) and the mean lowest oxygen saturation was 82% (± 4%). A temporary acute reversible rise (two patients) as well as an acute fall (one patient) in the PcCO₂ measurements were noted when the patient lay on the side on which the ear clip was placed.

One of the major limiting factors to the continuous long term measurement of PcCO₂ has been the need to change the site of measurement at three to four hour intervals to prevent skin burns due to high temperature of the sensor probes (up to 45 degrees Celsius) [14,15]. The recommended maximum duration for which a sensor can be placed cutaneously at 42 degrees Celsius is 8 hours. None of the patients in this study manifested a skin burn.

PcCO₂ measurement is a safe and sensitive tool to detect alveolar ventilation in patients receiving nocturnal NPPV for respiratory failure [16,17]. PcCO₂ is superior to intermittent measurement of PaCO₂ in patients receiving nocturnal NPPV, as it is non invasive and gives a more complete picture about alveolar ventilation [18]. PcCO₂ measurement however may underestimate PaCO₂ measurement in acute hypercapnic respiratory failure, however, their agreement is reported to be acceptable otherwise [19].

The use of volume targeted bi-level ventilation in obesity hypoventilation syndrome has shown to reduce nocturnal hypercapnia [20]. The findings of our study show the utility of continuous PcCO₂ measurements during nocturnal sleep studies in the titration of NPPV pressures in patients with chronic hypercapnic respiratory failure due to hypoventilation. We also introduced a correction factor for the signal drift which is known to occur with such PcCO₂ measurement devices. There was a good correlation between the baseline PcCO₂ and the baseline arterial carbon dioxide tension (R=0.92) as well as the morning corrected PcCO₂ and the morning arterial carbon dioxide tension (R=0.85). To the authors current knowledge this is the first study in which continuous cutaneous capnography could be successfully used to titrate NPPV pressures in patients with hypoventilation. However, titration has to include drift correction. Drift correction of PcCO₂ improves the efficacy of PcCO₂ monitoring when compared to PaO₂ [17]. It is possible that in the future such drift correction measures might already be incorporated in the device.

Most patients could not sleep properly in the first part of the study as they were accustomed to the overnight non invasive home ventilation. However, they could sleep better once they were initiated on NPPV. The sleep efficacy when the patients received NPPV was 70%. This relatively low sleep efficacy compared to the normal population is attributed to the study setting, the effect of sleeping in the sleep laboratory and to intermittent changes in NPPV settings. In two patients there was an acute rise in the PcCO₂ and in one an acute fall in PcCO₂ as the patient laid on the side where the ear sensor was placed. Sleeping on the side where the ear sensor is placed may lead to compression of the vessels thereby slowing the blood flow in that area. This can lead to stagnation of blood at the site of measurement and thus to an increase in the PcCO₂ as the newly generated PcCO₂ is not easily flown away by the freely circulating blood. On the other hand, sleeping on the same side where the sensor is placed can also loosen the contact of the sensor with the skin and thus lead to an acute fall in PcCO₂. These findings indicate that a protection device around the sensor or the ear lobe could be a useful solution to prevent pressure on the earlobe or avoid slipping of the ear sensor when the patient sleeps or turns on to the side where the sensor is placed. Also, other sites where PcCO₂ can be measured are forehead, infraclavicular, chest and anterior abdominal wall [21,22].

The findings of our study show that cutaneous capnography can be used to titrate ventilatory settings in patients with chronic respiratory failure due to hypoventilation. We have chosen to study patients who were already receiving home ventilation so that we could compare the NPPV settings determined in this study to the existing settings of the patients.

Several patients included in the study were either being “over” or “under” ventilated based on the study pressures compared to their existing pressures. A small change in the pressure settings can significantly affect tolerability of the patient to bi-level NPPV. Whether quality of life can be substantially improved with adoption of PcCO₂ in the titration studies will need a larger randomized trial. Studies examining the utility of this device in patients undergoing first time titration of bi-level ventilation in the sleep laboratory are needed. Patients with additional oxygen supply would be of special interest, since the oximetry in these patients is not helpful for the ventilator settings.

In conclusion, cutaneous capnography using a digital ear lobe sensor is safe and feasible during overnight polysomnography and might serve as a useful additional tool to optimize NPPV settings in patients with chronic hypercapnic respiratory failure due to hypoventilation.