Cough Peak Flow Assessment Without Disconnection From the ICU Ventilator in Mechanically Ventilated Patients

Abstract

BACKGROUND:

Because ICU ventilators incorporate flow velocity measurement, cough peak expiratory flow (CPF) can be assessed without disconnecting the patient from the ICU ventilator. Our goal was to estimate the correlation between CPF obtained with the built-in ventilator flow meter (ventilator CPF) and CPF obtained with an electronic portable handheld peak flow meter connected to the endotracheal tube.

METHODS:

Cooperative mechanically ventilated patients who entered the weaning process and who were ventilated with pressure support < 15 cm H₂O and PEEP < 9 cm H₂O were eligible for the study. Their CPF measurements obtained on the extubation day were kept for analysis.

RESULTS:

We analyzed CPF obtained in 61 subjects. The mean ± SD value of ventilator CPF and peak flow meter CPF were 72.6 ± 27.5 L/min and 31.1 ± 13.4 L/min. The Pearson correlation coefficient was 0.63 (95% CI 0.45–0.76), P < .001. The ventilator CPF had an area under the receiver operating characteristic curve of 0.84 (95% CI 0.75–0.93) to predict a peak flow meter CPF < 35 L/min. Neither ventilator CPF nor peak flow meter CPF differed significantly between subjects who were or were not re-intubated within 72 h (n = 5) and failed to predict re-intubation at 72 h (area under the receiver operating characteristic curve of 0.64 [95% CI 0.46–0.82] and 0.47 [95% CI 0.22–0.74]).

CONCLUSIONS:

CPF measurements using a built-in ventilator flow meter were feasible in routine practice with cooperative ICU subjects who were intubated and correlated with CPF assessed by an electronic portable peak flow meter.

Introduction

Weaning from invasive mechanical ventilation is a crucial step in the course of the ICU patient. One of the causes of weaning failure is secretion retention, which increases the respiratory load, ^1,2 and may be partly explained by an ineffective cough. ³ Assessing cough strength in ICU patients may help predict weaning success or failure. ⁴ In addition, insufficient cough strength seems to correlate with in-hospital mortality. ⁵ That is why assessing the cough strength of patients who are intubated is a tool that is gradually becoming part of ICU extubation algorithms. ⁶

To assess cough strength in patients who are intubated, semi-quantitative scales (eg, graded from 0, meaning no cough on demand, to 5, meaning strong and repeated cough) can be used but require trained personnel. ⁷ However, a number of methods to measure objectively obtain the cough peak expiratory flow (CPF) in patients who are intubated are available. ⁸ When being assessed by using an electronic spirometer connected to the endotracheal tube (ETT), a CPF ≥ 60 L/min is usually considered as reflecting an “effective cough.” ³ Electronic peak flow meters such as those used in patients with asthma can also be connected to the tracheal tube (Fig. 1) and used for assessing CPF in patients who are intubated or tracheostomized. ^9,10 With the use of these devices, a CPF < 35 L/min was shown to predict extubation failure. ⁹ Lastly, CPF can be measured by using the expiratory flow meter incorporated in ICU ventilators. ^11,12

Fig. 1.

The electronic peak flow meter is connected to the proximal tip of the endotracheal tube via a bacterial/viral filter.

Using either an electronic spirometer or an electronic flow meter to measure CPF requires additional equipment (dedicated expensive devices, filters, and connectors) and requires disconnecting the patient from the ventilator, which may expose caregivers to aerosol contamination. Moreover, the coughing effort produced by the patient must be perfectly synchronous with the activation of the electronic flow meter. Conversely, the measurement of CPF through the expiratory flow meter embedded in ICU ventilators does not require additional equipment nor does it require disconnection from the ventilator. CPF values are easily visualized on flow curves displayed on the ventilator screen, ^11,12 thereby minimizing risks of contaminating caregivers with aerosolized bacterial and viral particles.

Only 2 studies compared the CPF measured through the built-in ventilator flow meter (ventilator CPF), with the CPF measured by using an electronic spirometer connected to the tracheal tube. ^11,12 To the best our knowledge, CPF measured with an electronic peak flow meter connected to the tracheal tube (peak flow meter CPF), which is less expensive and more easily available in the ICU, has never been compared with ventilator CPF. We investigated whether the correlation between peak flow meter CPF and ventilator CPF, both measured on the day of extubation in ICU patients, would be sufficient for the ventilator CPF to be used as a predictive tool for successful weaning.

Quick look.

Current Knowledge

Assessment of cough capability in the mechanically ventilated patient is a component of the ventilator weaning process because it reflects cough strength once the patient is extubated. Subjective or quasi-objective measurements display high variability and require trained personnel. Currently, the reference standard technique for getting a reliable numerical value of cough peak flow requires disconnecting the ventilator to connect a flow meter in series to the endotracheal tube.

What this paper adds to our knowledge

Cough assessment can be easily performed while the patient is still connected to the ventilator. Cough peak flow with the built-in ventilator flow sensor correlated with that obtained with a portable electronic peak flow meter that requires disconnection from the ventilator. Larger studies are needed to confirm the routine clinical application of this technique during weaning.

Methods

This was a single-center prospective observational study, with randomized order of cough strength assessments, conducted in the medical ICU of the Center Hospitalier Régional d’Orléans between November 2017 and January 2019. All patients who were treated with invasive mechanical ventilation for > 24 h and who were entering the weaning process with a PEEP < 9 cm H₂O and pressure support < 15 cm H₂O, who were hemodynamically stable, meaning with (1) heart rate ≤ 140 beats/min and (2) systolic blood pressure between 90 and 160 mm Hg, with no or minimal vasopressors ¹ and with a state of alertness strictly between –1 and +1 (allowing response to a simple order) ¹³ according to the Richmond Agitation-Sedation Scale, were eligible for the study.

The exclusion criteria were as follows: patients < 18 years old, bronchospasm on auscultation, ${\text{F}}_{{\text{IO}}_{\text{2}}}$ > 0.7, abdominal or thoracic surgery or thoracic trauma with rib fractures in the past 7 d, and pneumothorax diagnosed within the 24 h preceding the screening. The protocol for mechanical ventilation weaning used in our ICU follows the recommendations published in 2007. ¹ The study protocol was approved by an independent ethics board (CPP Sud-Est VI, Clermont-Ferrand, France) in October 2017 (2017-A010882-51). Thorough information was provided to the subjects, and verbal consent was obtained by the physiotherapist in charge (ED, LC, CF, and GF).

CPF Measurements

Once consent was obtained, we assessed cough strength (peak flow meter CPF and ventilator CPF) once a day until the day of extubation. The order of measurements was randomly assigned (beginning with the ventilator or with the peak flow meter), with block sizes unknown to the investigators. For the ventilator CPF assessment, the subjects remained connected to the ventilator (Servo-i, Maquet, Solna, Sweden or V500, Dräger, Lübeck, Germany), with pressure support set at 7 cm H₂O and a PEEP of 0 cm H₂O. The subjects were then asked to take a deep breath and cough as hard as possible. The evaluator (GF) froze the ventilator screen and measured the maximal expiratory flow generated in L/min on the flow curve (Fig. 2).

Fig. 2.

On this ventilator screen, the central curve displays the inspiratory and expiratory flow. The red arrow shows the peak expiratory flow.

For the assessment of peak flow CPF, we used an electronic peak flow meter (Piko-1, Ferraris Respiratory, Hertford, United Kingdom). This is a device that uses a proprietary technology that has previously been shown to have good reproducibility and agreement with reference measurements when using a pneumotachograph in healthy volunteers and subjects with asthma who were stable. ¹⁴ We disconnected the subjects from the ventilator and connected the peak flow meter at the proximal tip of the ETT, with a bacterial/viral heat and moisture exchange filter (Anest-Guard, Teleflex Medical Europe, Westmeath, Ireland) that allows inhalation (Fig. 1). We then turned on the peak flow meter and asked the subjects to take a deep breath and cough as hard as possible. The value displayed represented the peak flow meter CPF. The 2 CPF measurement (peak flow meter CPF and ventilator CPF) modalities were spaced 30 min apart to give the subjects time to rest.

Each modality (peak flow meter CPF and ventilator CPF) was performed 3 times at 30-s intervals. We collected the 3 values to determine the maximum value and the mean value in each subject. The CPF values that were saved for the main analysis were the ones measured on the day of extubation. After obtaining these measurements, we measured the subjects’ Medical Research Council sum score (MRCs) 30 min after the last CPF assessment to estimate his or her global muscle strength. ¹⁵ The decision to continue weaning until extubation was left to the attending intensivist, who was not aware of the CPF and MRC values. After extubation, the subjects were evaluated for 72 h to assess the success of weaning from mechanical ventilation. The criteria for extubation failure used in our ICU were those described in the 2007 weaning recommendations. ¹ If the subjects were re-intubated at any time during their stay in ICU, even after 72 h, they would no longer be eligible to participate in the study.

Objectives

The primary objective of the study was to quantify the correlation between the 2 modalities of CPF measurements (peak flow meter CPF and ventilator CPF) in subjects undergoing mechanical ventilation. The secondary objectives were (1) to estimate the agreement between the 2 CPF assessment methods when considering the peak flow meter CPF as the reference value and (2) to evaluate the ability of CPF in predicting the success of mechanical ventilation weaning, defined as the non-necessity to re-intubate the subject within 72 h after extubation.

Statistical Analysis

We assumed that the Pearson correlation coefficient r between the 2 CPF measurements would be at least 0.90. For the range of the 95% CI of this correlation coefficient to be between 0.85 and 0.95, 62 subjects were needed. ¹⁶ Continuous variables are shown as mean ± SD or median (25th–75th percentile), depending on their distribution. Categorical data are expressed as numbers and percentages. To estimate the correlation between the 2 CPF measurements, the Pearson correlation coefficient r was calculated when assuming the normal distribution of the CPF values. Normality was checked by inspecting density plots and quantile-quantile plots.

The bias and agreement intervals between the 2 CPF measurement methods were calculated according to a modified Bland-Altman method when considering the multiple measurements per subject as well as the potential variability due to the subject’s age, the history of chronic respiratory failure, the order of measurements (peak flow meter CPF or ventilator CPF first), and the type of ventilator used. ^17,18 The ability of ventilator CPF to predict a peak flow meter CPF value below or above certain thresholds and the ability of CPF (with either measurement method) to predict the success of weaning were assessed by calculating the area under the receiver operating characteristic curve, the best threshold (Youden method), and related sensitivity and specificity. The correlation between CPF measurements and MRCs was estimated by using the Spearman correlation coefficient ρ. The analyses were conducted by using the R 4.2 software (R Foundation for Statistical Computing, Vienna, Austria). A 2-sided P value of <.05 indicated statistical significance. P values were not adjusted for multiple testing.

Results

During the study period, 589 patients were admitted to the ICU. Among them, 192 were eligible for the study, 10 refused to participate, 40 patients were not offered to participate in the study because the research team was not available, and 80 were not able (lack of awareness) to undergo CPF measurements. A total of 62 subjects were included in the study, and 1 subject could not be extubated (Fig. 3). This allowed us to obtain peak flow meter CPF and ventilator CPF in 61 subjects on the day of extubation. Subject characteristics, shown in Table 1, were similar between the 2 randomization groups.

Fig. 3.

Flow chart.

Table 1.

Subject Characteristics

Correspondence Between Peak Flow Meter CPF and Ventilator CPF

The mean ± SD values of the 3 measurements of ventilator CPF and peak flow meter CPF were 72.6 ± 27.5 L/min and 31.1 ± 13.4 L/min. The Pearson correlation coefficient between the 2 means was 0.63 (95% CI 0.45–0.76) (Fig. 4). The maximum mean ± SD values of ventilator CPF and peak flow meter CPF were 81.8 ± 32.8 L/min and 35.4 ± 15.6 L/min. The Pearson correlation coefficient between these 2 maximum values was 0.57 (95% CI 0.38–0.72). The coefficient of determination was not significantly different whether ventilator CPF (r² = 0.55) or peak flow meter CPF (r² = 0.21) was measured first (P = .09). With regard to the type of ventilator used to obtain the CPF measures, we found an r² of 0.57 (n = 12) with the Dräger V500 and r² of 0.32 (n = 49) with the Maquet Servo-i. The difference was not statistically significant (P = .19) (Fig. S1 [see the supplementary materials at http://www.rcjournal.com]).

Fig. 4.

The correlation between cough peak flow measured by the ventilator and by an external peak flow meter. The Pearson correlation coefficient (r) and the coefficient of determination (r²) are given with their 95% confidence intervals.

The bias between CPF measurements (ventilator minus peak flow meter) was 41.6 L/min (95% CI 39.2–44.0) and the limits of the agreement interval were −8.2 L/min (95% CI –12.9 to –3.3) to 91.4 L/min (95% CI 83.9–99.0) (Fig. S2 [see the supplementary materials at http://www.rcjournal.com]). The ventilator CPF had an area under the receiver operating characteristic curve for predicting a peak flow meter CPF < 35 L/min of 0.84, 95% CI 0.75–0.93. The best threshold of 76 L/min for ventilator CPF had a sensitivity and a specificity of 0.77 (95% CI 0.51–0.91) and 0.94 (95% CI 0.72–1.00), respectively, in predicting a peak flow meter CPF < 35 L/min. A ventilator CPF ≤ 60 L/min had a sensitivity of 0.58 (95% CI 0.35–0.65) and a specificity of 1.00 (95% CI 1.00–1.00) in predicting a peak flow meter CPF < 35 L/min (Fig. 5).

Fig. 5.

Ventilator CPF measurement for predicting a peak flow meter CPF < 35 L/min. A: Ventilator CPF values according to the peak flow meter CPF (<35 L/min or ≥35 L/min). The dashed line at ventilator CPF = 76 L/min indicates the best threshold (Youden method) to predict peak flow meter CPF <35 L/min or ≥35 L/min. The dashed line at ventilator CPF = 60 L/min indicates the threshold at which ventilator CPF had a 100% specificity to predict peak flow meter CPF < 35 L/min. B: AUC (with 95% CI) of ventilator CPF for predicting a peak flow meter CPF < 35 L/min. Ventilator CPF = Cough peak flow measured by the flow meter embedded into the ventilator; peak flow meter CPF = cough peak flow measured by the electronic, external peak flow meter; AUC = area under the receiver operating characteristic curve.

CPF and Risk of Re-Intubation

Seven subjects (11.5%) were re-intubated during their stay in the ICU, including 5 subjects (8.2%) who were re-intubated within 72 h after extubation. Both the averages of ventilator CPF and of peak flow meter CPF values per subject did not differ significantly between the subjects who needed and those who did not need re-intubation within 72 h (Fig. S3 [see the supplementary materials at http://www.rcjournal.com]) and failed to predict re-intubation at 72 h (area under the receiver operating characteristic curve of 0.64 [95% CI 0.46–0.82] and 0.47 [95% CI 0.22–0.74]). The correlation between ventilator CPF and subjects’ MRCs assessed on the day of extubation was not significant (rho² = 0.046, P = .10). The correlation was significant but weak between peak flow meter CPF and MRCs (rho² = 0.079, P = .030).

Discussion

We showed that the measurement of subjects’ CPF through the flow meter embedded in ICU ventilators (ventilator CPF), which required no ventilator disconnection, was feasible and significantly correlated with that obtained with a portable electronic expiratory peak flow meter (peak flow meter CPF). The ventilator CPF that was < 76 L/min had a good ability to predict a peak flow meter CPF < 35 L/min, a threshold that reflects poor cough strength that has been associated with a propensity to a failure of mechanical ventilation weaning. ⁹

So far, only 2 studies compared ventilator CPF with the CPF obtained by using a high-fidelity spirometer, an expensive device that is not available in all ICUs. ^11,12 Both studies showed that built-in ventilator flow meters were as accurate as reference methods to measure expiratory flow in ICU subjects who were intubated. In a study by Gobert et al ¹¹ that used the built-in ventilator flow meter (Evita XL, Dräger) at different inflation pressures of a pneumatic lung model and different sizes of ETTs provides CPF values identical to those measured by using a linear pneumotachograph (TSD 137 G, Hans Rudolph, Kansas City, Kansas) connected to the ETT. The clinical part of the study showed that a ventilator CPF that is < 60 L/min may predict extubation failure, a threshold value identical to that reported in clinical studies by using a high-precision spirometer. ^5,7

Bai and Duan ¹² compared ventilator CPF (PB840, Covidien, Mansfield, Massachusetts) with CPF obtained by using a spirometer connected to the ETT (ChestGraph HI-101, Chest MI, Tokyo, Japan). The subjects who successfully completed a spontaneous breathing trial underwent CPF measurements by using both methods, always completing the ventilator measurements first. In the 126 subjects included, the correlation between the CPF values obtained with both methods (r = 0.82) was higher than in our study. ¹² Both measurement methods identified the value of 56 L/min as the best threshold value to predict extubation success or failure. This value was once again close to the 60 L/min threshold value commonly found by other researchers. ^5,7 In our study, most of the ventilator CPF values were in the 20–150 L/min range, such as those observed in the above-cited studies that used different ICU ventilators. ^11,12

We showed a significant correlation but a wide agreement interval between both methods. Given the good concordance with reference methods reported for ventilator CPF measurements in ICU patients who are ventilated ¹² and for Piko-1 measurements in subjects who were spontaneously breathing, ¹⁴ we could conceivably expect better agreement and/or concordance between ventilator CPF and peak flow meter CPF than what we finally observed. Measuring either ventilator CPF or peak flow meter CPF first (in random order) might partly explain this discrepancy. The correlation between both measurements tended to be better when ventilator CPF was measured first, as was done by Bai and Duan. ¹² This difference did not reach statistical significance, probably because of insufficient sample size. Moreover, we interposed a filter between the ETT and the Piko-1 device that brought additional resistance (≈1.8 cm H₂O at flow of 60 L/min) to air flow and may have led to this discrepancy.

We observed lower peak flow meter CPF values than Beuret et al ⁹ who also measured peak flow meter CPF by using the Piko-1 device connected to the ETT with a bacterial/viral heat and moisture exchange filter of roughly similar air flow resistance (≈2.5 cm H₂O at flow of 60 L/min). The state of alertness and attention of ICU subjects varied during the day, which may also have induced differences between CPF measurements. The variability in the way instructions are given, changes in motivation, patient heterogeneity, and small sample size may partly explain the inconsistencies of our result when compared with other readings. ^11,12 We assessed the correlation between MRC and CPF on the day of extubation. It was weak, with the peak flow meter CPF (rho² = 0.079, P = .030) and non-existent with the ventilator CPF (rho² = 0.046, P = .10). In a multi-center study by Vivier et al, ¹⁹ the mean ± SD MRCs score was 44 ± 15, whereas MRCs score was 56 (42–60) in our population. It is likely that our subjects did not have sufficient muscle weakness for the relationship between peripheral strength and coughing ability to be sufficiently close.

To summarize, we observed ventilator CPF values comparable with those previously reported, a moderate correlation between ventilator CPF and peak flow meter CPF, and a high specificity of ventilator CPF to predict peak flow meter CPF < 35 L/min. Taken together, analysis of these findings suggests that measuring CPF through the built-in ventilator flow meter is feasible. However, we could not show a satisfactory ability of either ventilator CPF or peak flow meter CPF to differentiate between subjects who would or would not be successfully weaned. Our study was not designed for such an objective. Additionally, our study included a high proportion of subjects (92%) considered to be at risk for extubation failure because of their age (>65 y old) and/or because of the presence of chronic heart or lung disease. ² In these populations, ineffective or weak cough strength is only one among many factors that could lead to mechanical ventilation weaning failure. It should be noted that 54% received prophylactic noninvasive ventilation and/or high-flow nasal oxygen therapy, both of which can avoid weaning failure. ²⁰

Our study had several limitations. First, our sample size, which was calculated to estimate the correlation between ventilator CPF and peak flow meter CPF, was not sufficient to assess the performance of either measurement method in predicting weaning success. Second, we assessed voluntary cough strength, which may not be sufficiently standardized or biased by attentional changes. Voluntary cough seems to be lower than the reflex cough strength triggered by stimuli, for example, tracheal suction. ¹⁰ It remains to be determined whether objective measurements of voluntary cough strength outperform the subjective assessment by using clinical ordinal scales ¹⁹ during the weaning process. Third, we did not analyze the impact of the ETT size used on CPF. However, all the patients who were included had a 7- or 7.5-mm diameter ETT. Fourth, we used 2 types of ventilators, which may have added variability to the ventilator CPF measurements. Fifth, to perform a pragmatic clinical study grounded in daily practice, our patient population was very heterogeneous, and the interpretation of our results must be done with caution. New studies that use CPF on the ventilator must be carried out so that this technique can spread and become the reference in terms of evaluation of the cough capacity of ICU patients under invasive mechanical ventilation.

Conclusions

Our findings add to the existing literature and show that CPF measurement by using built-in ventilator flow meters is feasible in routine practice in cooperative ICU subjects who are intubated and correlate with CPF assessed by an electronic portable peak flow meter.