Influence of Tracheal Obstruction on the Efficacy of Superimposed High-frequency Jet Ventilation and Single-frequency Jet Ventilation

THE new interventional bronchoscopic techniques emerging during the last 2 decades have significantly improved the outcome after treatment of tracheobronchial obstruction, particularly the patients’ exercise capacity, resulting in a higher quality of life.^1,2 

However, these interventions are high-risk procedures, especially in patients with high-grade stenosis, multiple stenoses, and/or other preexisting pulmonary conditions.³ 

The procedures are mainly performed using rigid tracheoscopy/bronchoscopy and jet ventilation (JV). There is a large body of clinical experience supporting the use of superimposed high-frequency JV (SHFJV) in patients undergoing airway interventions and/or laser tumor debulking procedures.⁴  Even in cases with severe airway obstruction, SHFJV has been shown to maintain adequate p_ao₂ and carbon dioxide removal.^5–7 

Superimposed high-frequency jet ventilation is a technique that combines a high-frequency (HF) (f_HF; commonly >500 min⁻¹) with a normal-frequency (NF) jet (f_NF; 12 to 20 min⁻¹). The rationale for using a normofrequent ventilation component is to increase minute ventilation and thereby facilitate carbon dioxide removal.⁸  It may also provide lung recruitment and improve oxygenation.⁹  In a recent investigation of JV at different frequencies in unobstructed airways, we found that SHFJV effectively increased end-expiratory chest wall volume (EEV_CW) and tidal volume (V_T) providing adequate arterial oxygen (p_ao₂) and carbon dioxide tension (p_aco₂) levels for a wide range of frequencies.¹⁰ 

In this study, we hypothesized that SHFJV would also achieve better oxygenation and gas exchange than single-frequency (high-frequency) JV (HFJV) at severe airway obstruction. The aim was to systematically compare the efficacy of SHFJV with HFJV over a wide range of JV frequencies and with varying degrees of airway obstruction. Using the obtained data, we developed a descriptive model to predict the influence of variable airway obstruction and JV frequency on ventilatory parameters during SHFJV.

We studied 10 healthy pigs (Swedish mixed country breed) at 3 months of age (weight 25 to 31.5 kg) at the Hedenstierna Laboratory. Uppsala University Animal Research Ethics Committee approved the study (reference number: C 140/11), and the National Institutes of Health guidelines for animal research were followed.

On arrival to the laboratory, the animal was premedicated with an intramuscular injection of xylazine 2.2 mg/kg, tiletamine 3 mg/kg, and zolazepam 3 mg/kg. When the premedication showed an effect (after approximately 10 min), the animal was placed supine on the operating table and an ear vein was cannulated with a peripheral venous catheter. A bolus of 100 to 500 μg fentanyl was injected IV and the trachea was intubated orally (Hi-Contour Tracheal Tube, ID 8.0 mm; Mallinckrodt Medical, Ireland). Volume control ventilation was started (Servo-i^®, Sweden) with V_T of 10 ml/kg, and the respiratory rate was adjusted to achieve normocarbia (37.5 to 45 mmHg).

General anesthesia was induced and maintained intravenously using pentobarbital (7 to 9 mg kg⁻¹ h⁻¹) and morphine (420 to 540 μg kg⁻¹ h⁻¹). After ascertaining adequate depth of anesthesia by painful stimulation between the front toes (absence of withdrawal reaction, awakening, or autonomous response), neuromuscular block was established and maintained by IV infusion of pancuronium (280 to 360 μg kg⁻¹ h⁻¹). To ensure adequate anesthesia during neuromuscular block, heart rate and blood pressure were monitored, and signs of responsiveness to manipulations were observed.

An arterial line for invasive blood pressure measurement and blood gas sampling was surgically placed in the left carotid artery. A Swan-Ganz catheter (pressure verified position in the pulmonary arterial bed) and a central venous line were inserted via the ipsilateral external jugular vein.

A urinary catheter was inserted via a minilaparotomy. Throughout the experiment, electrocardiogram was continuously monitored and oxygen saturation measured by pulse oximetry was acquired by photoplethysmography on the tail base.

At completion of the experiment, the animal was euthanized by an IV injection of potassium chloride under deep anesthesia.

The airway obstruction was achieved by inserting a stent (fig. 1A) into the trachea with the following method: The upper part (approximately 5 cm) of the trachea and the larynx were dissected circumferentially and a tracheostomy was performed at the proximal end of the trachea. For stent insertion/exchange, the previously inserted oral endotracheal tube was uncuffed and withdrawn as much as possible without exiting the larynx. Through the tracheostomy stoma, obstructive stents (length 2 cm, outer diameter 10.8 mm, IDs 2, 4, 6, and 8 mm) were alternately introduced into the tracheal lumen. Airflow bypassing the stent was prevented by tightening a circumferential cotton band around the trachea (fig. 1B). The stent was secured with a suture at the proximal end to restrict any caudal dislocation. After stent insertion, the oral endotracheal tube was advanced until the cuff appeared flush with the tracheostomy stoma and cuff inflation prevented air leakage through the stoma. With this setup, we were repeatedly able to manipulate the degree of tracheal obstruction by simply exchanging the stents. A schematic diagram of the experimental protocol is illustrated in figure 2.

A Twinstream^® jet ventilator (Carl Reiner GmbH, Austria) was attached to the oral endotracheal tube by means of a Veres T-adaptor (Carl Reiner GmbH) and a bias flow of 55 l/min of warmed and humidified respiratory gas with Fio₂ = 0.5 (HumiCare^® 200 breathing gas humidifier; Gründler medical GmbH, Germany) was entrained through the proximal opening of the T-adaptor.

The Twinstream^® jet ventilator consists of two independently adjustable modules for JV: an NF unit and an HF unit. For each module, operation (on/off), frequency (f_NF or f_HF), working pressure, and inspiratory/expiratory (I/E) ratio can be adjusted separately. Thus, the ventilator can be operated in either HFJV or NFJV mode, with only the respective module activated, or in SHFJV mode, with both modules operating simultaneously.

Throughout the experiment, the NF unit had the following settings: f_NF = 16 min⁻¹, I/E = 1:1, and a working pressure of 1.6 bar. The HF unit was adjusted as follows: different f_HF were used (f_HF = 50, 100, 150, 200, 300, 400, and 600 min⁻¹), I/E = 1:1, and a constant working pressure of 0.8 bar. Fio₂ was globally set to 0.5, including bias flow.

Whenever a condition led to a pulmonary artery pressure greater than 60 mmHg or an occurrence of arrhythmias during a hypoxic period, the protocol was interrupted and rescue ventilation (volume control ventilation with Servo-i^® ventilator, V_T of 10 ml/kg and Fio₂ = 1.0) was started. After normalization of pulmonary artery pressure, acid–base status, and oxygenation, the protocol was continued.

Primary outcome variables were ΔEEV_CW and chest wall V_T. Secondary outcome variables were p_ao₂, p_aco₂, and prestenotic (P_UPPER) and poststenotic airway pressure (P_LOWER).

To preclude possible influences of preceding settings on outcome, we chose a crossover design. Using a computer-generated randomization method (Excel; Microsoft, USA), we randomized the order of application for all interventions (stent ID and HF component of JV). A diagram of the study protocol is illustrated in figure 2.

We measured chest wall volume with optoelectronic plethysmography (OEP).^11,12  The volumes obtained by this method correspond well to lung volume changes in a variety of conditions,^13–16  and the method has recently been used for studies of different modes of JV.^10,17 

Optoelectronic plethysmography provides spatial (SD for repeated volume measurements 0.17%)¹⁵  and temporal resolution (60 Hz)¹⁸  adequate for the study of JV.

An OEP System^® (BTS Bioengineering, Italy) with six infrared cameras acquired the position and movement of an array of 57 adhesive retro-reflective markers on the animal’s trunk. The cameras were arranged on supports below the laboratory ceiling (three on each side of the animal) in a manner that each of the reflective markers was viewed by at least two cameras.

From the camera registrations, the bundled OEP Capture^® motion analysis software (BTS Bioengineering) calculated the position of each marker in a three-dimensional coordinate system for each time point. The volume enclosed by the surface formed by triangulation of the markers represents the chest wall volume.

Arterial blood gas samples were obtained after 5 min of each ventilation condition, and p_ao₂ and p_aco₂ were immediately measured with a blood gas analyzer (ABL 500; Radiometer, Denmark).

Prestenotic airway pressure (P_UPPER) was measured via a 200-mm long 16-gauge Secalon Seldy catheter (BD Medical Surgical Systems, Sweden) at the tip of the endotracheal tube. For measurement of poststenotic airway pressure (P_LOWER), another Secalon Seldy catheter was introduced into the trachea below the stent by the Seldinger technique. Identical analog pressure transducers (RCEM250DU; Sensortechnics GmbH, Germany) were used to acquire pressure signals that were synchronized with the OEP volume registrations and continuously recorded by the OEP Capture^® software.

Optoelectronic plethysmography measures total chest wall volume, that is, not only the lung volume. Thus, it was necessary to compensate for changes in volume that were not attributable to ventilation, for example, fluid administration or gas production in the intestines. For that purpose, each f_HF step (with SHFJV and HFJV) was preceded by another 5 min of ventilation with NFJV (f_NF = 16 min⁻¹, f_HF = 0), followed by a disconnection from the ventilator. At f_NF = 16 min⁻¹, the duration of the expiratory phase is 1.9 s resulting in a decrease of P_LOWER to 0 at a low degree of obstruction. With more severe stenosis, it was necessary to disconnect the ventilator for some seconds to allow P_LOWER to decrease to 0. We chose chest wall volume at P_LOWER = 0 cm H₂O as the reference volume as we assumed that it was approximately equal to the functional residual capacity (FRC) at apnea.

For each condition, we derived chest wall volumes and airway pressures as averages of at least five consecutive steady-state breaths.

Changes in end-expiratory chest wall volume above apneic FRC (ΔEEV_CW) are hence defined as follows:

and

Tidal volume was calculated by subtracting EEV_CW from end-inspiratory chest wall volume (EIV_CW):

At higher rates, there is a phase offset between the volume and the pressure signals. Hence, defining positive end-expiratory pressure (PEEP) as airway pressure at end-expiration would result in erroneous values. Therefore, we defined PEEP as the lowest observed pressure during the respiratory cycle.

The length of the experiments made it necessary to investigate whether there were changes of respiratory mechanics over time that could influence the outcome. For that purpose, we assessed static compliance at the beginning and at the end of the protocol, without any degree of airway obstruction. In pressure control ventilation (Servo-i^®), an incremental-decremental PEEP maneuver was performed. Starting from zero end-expiratory pressure (ZEEP), PEEP was increased by 2 cm H₂O every five breaths until a maximum of 20 cm H₂O. In the same manner, PEEP was then decreased to ZEEP.

For each animal, static compliance was computed using triplicate measurements of end-expiratory pressure and end-expiratory volume at ZEEP and at 20 cm H₂O, respectively. From all available pigs, mean and 95% CI for C_STAT were calculated.

All statistical computations were performed with R (R Foundation for Statistical Computing, Austria). Previous experience has shown that a number of 8 to 10 animals would be adequate to detect outcome differences.

In the first part of this article, we tested the alternative hypothesis that the use of SHFJV would result in an increase of ΔEEV_CW and V_T compared with HFJV at the same frequency. We further hypothesized that SHFJV would maintain p_ao₂ at higher and p_aco₂ at lower levels than HFJV. Statistical analyses of ΔEEV_CW, V_T, p_ao₂ and p_aco₂, and airway pressure were performed using linear mixed model analysis. Details on the definition of fixed and random factors can be found in table 1.

The second part of the article focuses on finding a descriptive model for the influence of variable stent ID and the HF component of ventilation on the outcomes stated in the previous paragraph. Based on our observations of lung volumes, gas exchange, and airway pressures, we used linear mixed model analysis (for details, see table 1) to predict the dependency of these outcome variables on stent ID and on the HF component of JV.

Nonlinear frequency dependency¹⁰  or obstruction dependency¹⁹  of most of our outcome variables have been reported previously. Therefore, the transformation of parameters was necessary to fit the linear mixed models to our observations (table 1).

All previously log-transformed data are presented as geometric mean (V_T, p_ao₂, and p_aco₂) and 95% CI, and untransformed data as arithmetic mean and 95% CI. The mixed model was used to characterize the properties of our observations; therefore, significant results should be interpreted as descriptive rather than confirmatory.

Equations for the mixed models used for the analyses are provided as Supplemental Digital Content 1, https://links.lww.com/ALN/B184.

Testing for statistical differences in C_STAT between the start and the end of the protocol was performed using the paired t test.

One animal was excluded from the study because of bilateral pneumothorax that occurred during a respiratory recovery phase using conventional ventilation. The results of the mixed model analyses for the comparison of SHFJV and HFJV are summarized in table 2; those for the description of frequency and stent ID dependency of SHFJV are given in table 3. In Supplemental Digital Content 2, https://links.lww.com/ALN/B185, we provide observed means and 95% CI for each variable for the SHFJV–HFJV comparison (table 1) and for the description of SHFJV dependency on frequency and stent ID (table 2).

End-expiratory chest wall volume was at least 1.6 times higher with SHFJV compared with HFJV (fig. 3A) throughout the frequency range.

Tidal volumes were greater with SHFJV compared with HFJV, the difference increasing with frequency (fig. 3B). This difference relied solely on the V_T changes during HFJV, resulting in a more than 30-fold decrease of V_T when f_HF was increased from 50 to 600 min⁻¹, reaching negligible V_Ts at f_HF greater than 150 min⁻¹.

Superimposed HFJV provided higher levels of p_ao₂ than HFJV, especially at high frequencies, again explained by changes during HFJV (fig. 3C). During SHFJV, p_ao₂ remained practically constant at levels greater than 225 mmHg throughout the frequency range, whereas increasing f_HF during HFJV resulted in a 5.5 times decrease of p_ao₂ with hypoxemic values at f_HF greater than 150 min⁻¹.

Likewise, carbon dioxide elimination was acceptable and largely unaffected by frequency when SHFJV was used (fig. 3D), whereas during HFJV, p_aco₂ almost doubled to a plateau at f_HF greater than 150 min⁻¹.

Superimposed HFJV resulted in higher PIP above and below the stenotic segment (PIP_UPPER and PIP_LOWER, respectively) (Supplemental Digital Content 3, fig. 1, https://links.lww.com/ALN/B186). There was a significant influence of frequency on both pressures for SHFJV and HFJV. End-expiratory pressures above the stenotic segment (PEEP_UPPER) were approximately equal for SHFJV and HFJV. Below the stenosis, this was only true for high frequencies because SHFJV increased to a maximum of 2.4 times higher end-expiratory pressure (PEEP_LOWER) at the lowest frequency.

The degree of obstruction (stent ID) and frequency both affected ΔEEV_CW (fig. 4A) and V_T (fig. 4B). Decreasing the stent diameter from 8 to 2 mm increased ΔEEV_CW and V_T by a factor of three. At low stent ID, increasing frequency resulted in an increase in ΔEEV_CW that reached a plateau at higher frequencies, whereas V_T was independent of frequency. With minor obstruction, EEV_CW and V_T both decreased with increasing frequency to reach a plateau.

Oxygenation during SHFJV was dependent on the stent ID, reduced by 30% at the narrowest stent, but independent of frequency (fig. 4C).

Carbon dioxide removal was strongly influenced not only by stent ID but also by frequency and the interaction of stent ID and frequency (fig. 4D). At the widest stent ID, p_aco₂ was normal for all frequencies but increasing the degree of obstruction resulted in a marked increase of p_aco₂ to levels greater than 75 mmHg.

Peak inspiratory pressure above (PIP _UPPER) and below the stenotic segment (PIP_LOWER) were dependent on stent ID, frequency, and the interaction of both parameters (Supplemental Digital Content 3, fig. 2, https://links.lww.com/ALN/B186).

At the widest stent ID, PIP_UPPER was lowest, and it increased by at least 11% when the stent ID was decreased to 2 mm. With minor obstruction, PIP_UPPER and PIP_LOWER were inversely related.

PIP_LOWER was approximately equal to PIP_UPPER for the widest stent ID and expressed the same frequency dependency. With decreasing stent ID, the pressure fall over the stent increased significantly. There was no influence of the frequency f_HF on PIP_LOWER in the narrowest stent ID. Importantly, PIP_UPPER always exceeded PIP_LOWER.

Positive end-expiratory pressures above (PEEP_UPPER) and below the stenotic segment (PEEP_LOWER) were also dependent on stent ID, frequency, and the interaction of both.

Positive end-expiratory pressures above (PEEP_UPPER) increased with both stent ID and frequency f_HF. At low degrees of stenosis, PEEP above and below the obstruction were almost equal, but with severe stenosis, we observed markedly higher values of PEEP_LOWER compared with PEEP_UPPER.

The current study consists of two parts: (1) a comparison between SHFJV and HFJV in a porcine model of severe airway obstruction and (2) an investigation of the interaction between the degree of tracheal obstruction and the jet frequency on ventilatory efficiency during SHFJV. It is the first in vivo study that systematically investigates the dependence of gas exchange on ventilation frequency for variable airway obstruction.

During HFJV, lung volumes, gas exchange, and intrapulmonary pressures were highly dependent on frequency. In our model, the significantly obstructed airway (75% obstruction for the 4 mm stent) resulted in the deterioration of V_T and gas exchange at relatively low frequencies. Already at f_HF greater than 150 min⁻¹, V_T became negligible and resulted in severe hypercapnia. In intact airways, a similar dependence of V_T and p_aco₂ on frequency has been described earlier,^20–22  but in the previous studies, deterioration of tidal ventilation and carbon dioxide accumulation occurred at much higher frequencies.¹⁰ 

We observed an increase of intrapulmonary PEEP (PEEP_LOWER) with increasing f_HF, but we failed to predict an accompanying increase of end-expiratory chest wall volume. In contrast, other investigators found that end-expiratory chest wall volume increased with increasing frequency.²⁰  Ihra et al.²³  made similar observations for intrapulmonary pressure, and theoretically, the increased end-expiratory pressure is caused by an increase of end-expiratory volume (air trapping). In our results, the deviant value of ΔEEV_CW at a frequency of 600 min⁻¹ could be responsible for why the model failed to detect the frequency dependence of ΔEEV_CW. Another explanation is that we used rather low working pressures for HFJV, and that changes in ΔEEV_CW with increasing frequency would have become evident if higher driving pressures had been used.

Interestingly, we found in agreement with some previous studies that oxygenation deteriorated when a certain frequency was exceeded,^10,24  whereas other investigators have found normoxia or hyperoxia even at very high rates.^25,26  When using an Fio₂ of 1.0, oxygen transport at higher frequencies would be expected to mimic apneic oxygenation.²⁷  In the case of our study and that of Lin et al.,²⁴  Fio₂ was set at 0.5 and 0.4, respectively. A possible mechanism for the rapid decrease of p_ao₂ in the latter studies could be alveolar nitrogen accumulation²⁸  at very low V_T with negligible alveolar ventilation. Therefore, for HFJV, sufficiently low frequencies of less than 150 min⁻¹ must be used to ensure adequate oxygenation and carbon dioxide removal in subjects with airway obstruction unless the Fio₂ is set at 1.0. Further studies could focus on or include the influence of working pressure and Fio₂ on JV efficacy versus resulting airway pressure.

In contrast to the observations for HFJV, changes of the f_HF component of SHFJV had less distinctive effect on lung volumes, gas exchange, and airway pressures. These findings support a previous report of the successful use of SHFJV with f_HF ranging from 180 to 900 min⁻¹ in subjects with stenotic airways.²⁹  In our investigation, the greatest effect of f_HF was observed with the widest airway stent in place. The largest alterations were observed for ΔEEV_CW, V_T, airway pressures, and p_aco₂ at the very low-frequency end and a more stable plateau was seen at higher frequencies.

With decreasing stent ID, the stenotic segment increasingly acted as a low-pass filter³⁰  with a lower flow rate across the stenosis.³¹  Accordingly, reduction of stent ID displaced the range of frequencies with significant ventilation alterations toward a lower f_HF spectrum. The passive process of expiration was most affected, as we noted an increase in intrapulmonary PEEP caused by air trapping and a reduction of tidal pressure variations with lower V_T. These findings corroborate previous results from bench studies.^19,32  Increasing frequency further accentuates this effect³¹  and stresses the importance of the use of sufficiently low ventilation frequencies for subjects with airway obstruction.

In analogy to the reduction of V_T, we found a gradual increase in p_aco₂ with decreasing stent ID.³²  Interestingly, p_ao₂ was only mildly reduced for the downsizing step from 8- to 4-mm stent ID (i.e., 75% obstruction). Further reduction of stent ID to 2 mm (94% obstruction) decreased p_ao₂ by approximately 30 to 50%, but notably, still within the clinically acceptable range. It is obvious that the low-frequency component of SHFJV was responsible for the observed differences between HFJV and SHFJV. Apparently, the low-frequency component was low enough to allow for the passage of greater than 100 ml of V_T even at 94% airway obstruction.

There has been some controversy about the safety of prestenotic JV in obstructed airways. From their results from bench studies, some claim that detrimental intrapulmonary pressures occur during prestenotic JV,^32,33  whereas others report intrapulmonary pressures within acceptable limits.^34,35  In this in vivo investigation, intrapulmonary peak pressure never exceeded 25 cm H₂O. Airway pressure measured at the prestenotic level always overestimated true intrapulmonary pressure, despite a stark discrepancy between the two measurements in severe obstruction. Although we noted a concomitant underestimation of intrapulmonary PEEP, we believe that prestenotic JV with (1) a jet injector position at sufficient distance from the stenotic segment³⁵  and (2) airway pressure monitoring more than 6 cm distal of the jet injector³⁶  help prevent inadvertently high intrapulmonary peak pressure and its complications in subjects with severe airway obstruction.

The amount of PEEP produced by JV depends on the frequency used. At very low rates, there is practically no PEEP, whereas increasing f_HF is known to lead to the development of an “auto-PEEP.”^10,17  In our protocol, we chose to define PEEP not as the pressure at end-expiration, but as the lowest observed pressure during a respiratory cycle. At lower frequencies, this probably reflects the true value of PEEP. At higher rates however, the “PEEP effect” has been suggested to be more closely related to the mean airway pressure.²⁶  Thus, at f_HF greater than 100 min⁻¹, our definition of PEEP may have underestimated the effective PEEP. However, our definition allows us to describe more exactly the influence of frequency on airway pressure and permits direct comparison of the minimum airway pressure at different frequencies.

Furthermore, the interaction of JV delivery mode and various ventilator settings during HFJV is not well elucidated. We exclusively used above-stenosis JV in the current study, and therefore, our observations of pre- and poststenotic pressure and the resulting safety implications apply exclusively to above-stenosis, catheter-free JV. Further systematic research is needed to appraise to what extent our findings on above-stenosis ventilation can be extrapolated to other routes of JV application.

We used OEP to noninvasively monitor chest wall volume changes as a surrogate for lung volume changes. Blood volume shifts, gas compression, and airway distension rather than true recruitment could theoretically be sources of error. However, previous studies have shown a good correlation between chest wall and lung volumes.^11,15 

A statistical limitation of the study is that the mixed model equation is calculated from our observed values. Therefore, the results from the model analysis apply with the stated CIs to the range of frequencies we investigated, which restricts the applicability of the model outside of these boundaries. However, we have previously shown that higher frequencies are unlikely to be useful for JV.¹⁰ 

Finally, we performed our investigations in a porcine model, and the results cannot be directly transferred to the behavior of humans with severe airway obstruction. For instance, the respiratory system compliance is lower in pigs, and airway closure occurs at much lower FRC than in humans. Furthermore, upon positive pressure, stenotic tissue may have a greater expansion capacity than the stents we used in our model in which each stent was fixed in diameter. In patients, especially during spontaneous ventilation, dynamic obstruction may contribute to the deterioration of ventilation. The latter phenomenon is less likely to significantly influence the ventilation efficacy during positive pressure ventilation such as (S)HFJV, but the development of intrinsic PEEP may go undetected when pressure is monitored only above the stent.²³ 

To conclude, HFJV may provide adequate oxygenation at most frequencies when the degree of stenosis is low, but the addition of a low-frequency jet results in improved gas exchange and lung volume, especially when challenged with a high-degree stenosis.

The choice of frequency for the f_HF component has little impact in SHFJV, whereas for HFJV, the chosen frequency has an important impact on carbon dioxide removal and oxygenation, in a reciprocal manner. Thus, the HFJV frequency should not exceed 150 min⁻¹ in subjects with severe airway obstruction.

The skilled technical laboratory assistance of Agneta Roneus, M.Sc., Karin Fagerbrink, M.Sc., and Maria Swälas, M.Sc. (all from Hedenstierna Laboratory, Department of Surgical Sciences, Uppsala University, Uppsala, Sweden), is gratefully acknowledged. The authors also thank Marcus Thuresson, Ph.D., Statisticon AB, Uppsala, Sweden, for assistance with the statistical analyses.

This work was supported by grants from Uppsala University Hospital (Uppsala, Sweden), the Tore Nilsson Research Foundation (Stockholm, Sweden), the Swedish Heart Lung Foundation (Stockholm, Sweden), Swedish Medical Research Council (Stockholm, Sweden, grant no. 5315), ALF funds (Uppsala County, Sweden), and Politecnico di Milano (Milan, Italy).

Dr. Aliverti is one of the inventors of optoelectronic plethysmography. The patents are owned by Politecnico di Milano (Milan, Italy) and licensed to BTS Spa Company (Milan, Italy). The other authors declare no competing interests.