Involuntary Cough Strength and Extubation Outcomes for Patients in an ICU FREE TO VIEW

Ventilator discontinuation is a two-step process involving the removal of both mechanical support (ie, positive pressure ventilatory assistance) and the artificial airway.^1,2 Tests of mechanics and gas exchange assess the appropriate time for removal of mechanical support, whereas tests for airway protection (cough) and airway patency assess the appropriate timing of extubation. Illustrating the differences in these processes, Khamiees et al³ demonstrated that conventional weaning parameters are not helpful in predicting extubation outcomes of patients who pass a spontaneous breathing trial (SBT) at rapid shallow breathing index (RSBI) < 125. For this reason, we chose to focus on the removal of the artificial airway, rather than the removal of the mechanical support, in our study. We aimed to quantify this second step in the process of artificial airway removal using involuntary cough peak flow (CPFi) as well as other parameters, which include sputum amounts, endotracheal tube size (diameter), and consciousness evaluation. In previous reports, only cough peak flow (CPF) was considered a valuable predictor of extubation success.^4-6

Bach and Saporito⁴ first studied cough peak expiratory flow to evaluate extubation and decannulation outcomes in tracheotomized patients with neuromuscular diseases. Furthermore, Smina et al⁵ showed that a voluntary cough peak flow (CPFv) could predict extubation in patients weaned from mechanical ventilator support in an ICU. They also found that the voluntary cough peak expiratory flow might be influenced by the patient’s own efforts.

Involuntary coughing triggered by a cough reflex is a more natural method that represents actual physiologic potential. It has the added benefit of widening the inclusion criteria to encompass sedated or comatose patients or patients unwilling or unable to produce a voluntary cough.⁴ Therefore, the present study was designed to determine whether a more natural method of estimating cough strength, the CPFi induced by normal saline, can predict different extubation outcomes for patients in an ICU who pass a spontaneous breathing trial.

Three hundred four patients on mechanical ventilation in tertiary intensive care units at Tri-Service General Hospital from January 2003 through December 2003 were evaluated for the study. From this group we recruited 150 patients who had been ventilated for ≥ 3 days, had passed a 2-h SBT, and were judged by their physician to be ready for extubation. The reasons for excluding the other 154 patients are summarized in Figure 1.

The weaning guidelines specified by our weaning and extubation protocol were carried out by nursing staff, respiratory therapists, and resident physicians, who were supervised by an attending physician certified by our Critical Care Medicine Board. The same weaning protocol was used by both medical and surgical ICUs, as attested by the chairman of the Critical Care Medicine Department and Internal Review Board at Tri-Service General Hospital. All patients gave their informed consent. If a patient had an altered level of consciousness, informed consent was obtained from the patient’s nearest relatives.

Weaning criteria were evaluated daily in all ICUs. Criteria to initiate weaning from the ventilator included hemodynamic stability, improvement in underlying medical conditions, withdrawal of pressors or inotropes, and respiratory parameters that included a PaO₂/FIO₂ ratio > 200,⁷positive end-expiratory pressure ≤ 5 cm H₂O, cessation of sedatives for 1 day, and RSBI (respiratory frequency divided by the tidal volume) of < 105 breaths/min/L detected 1 min after mechanical ventilation had been stopped and sputum suctioned. All patients who met these criteria underwent an SBT via a T-piece, pressure support ≤ 7 cm H₂O, or continuous positive airway pressure used with an endotracheal tube for 0.5 to 2 h.⁸ If a patient did not meet these criteria when first tested, he or she was evaluated daily until the criteria were fulfilled or until exclusion criteria were met.

Respiratory frequency, heart rate, systolic BP, and arterial oxygen saturation measured by pulse oximetry were recorded every 15 min during the trial of spontaneous breathing. The primary physician terminated the trial if a patient had any sign of poor tolerance, such as a respiratory frequency of > 35 breaths/min, arterial oxygen saturation < 90%, heart rate > 140 beats/min, or a sustained increase or decrease in the heart rate of > 20%, systolic BP > 200 or < 80 mm Hg, agitation, diaphoresis, or anxiety. If a patient had signs of poor tolerance at any time during the trial, mechanical ventilation was reinstituted. Using our protocol, the SBT was also terminated if there was new-onset arrhythmia, paradoxical breathing, or the use of accessory muscles.

The minute ventilation, respiratory rate, tidal volume, maximal inspiratory pressure (PImax), and maximal expiratory pressure were detected step-by-step using a hand-held respiratory mechanics monitor (VentCheck; Novametrix; Wallingford, CT) during each SBT. Consciousness level was assessed using the Glasgow Coma Scale as well as a mental status assessment (alert and attentive, inattentive or confused, and unresponsive or unable to verbally communicate). Sputum (endotracheal secretions) was suctioned and collected in a suction trap every hour for 4 h before the SBT attempt using a suction trap (mucus extractor) (Symphon Chemical Co; Taipei, Taiwan). The total sputum volume over this 4-h period was recorded.

Patients tolerating the SBT and judged to be ready for extubation by their physician underwent cough assessments. The involuntary CPF was induced by removing the T-piece and dripping 2 mL of normal saline into the endotracheal tube at the end-inspiratory point, with the patient in a head-up position at 45° from the horizontal. The end-inspiratory point was chosen to ensure that a full inspiration had occurred. No obvious side effect or complication was found using this method of involuntary cough induction. The T-piece and flow sensor were reconnected to the endotracheal tube as soon as possible. We observed the patients continuously until their breathing became smooth and regular. The maximum expiratory flow value detected by the hand-held respiratory mechanics monitor was recorded as the involuntary CPF.

Following the involuntary CPF assessment, an additional cough strength rating paradigm³ was used and consisted of the following scale: 0 = no cough response, 1 = audible movement of air through the endotracheal tube but no audible cough, 2 = strong cough with phlegm under the end of endotracheal tube, 3 = strong cough with phlegm coming out of the end of endotracheal tube. This cough strength observational rating paradigm showed a moderate correlation (ρ = 0.451, P < .001) with CPFi.

This cough strength assessment was observed by the respiratory therapist. We did not evaluate the interrater reliability. However, the respiratory therapists were trained until the ratings among all respiratory therapists were consistent.

At the completion of the cough assessments, the sputum was suctioned again to clean up the trachea, but it was not added to the total sputum volume collected over the previous 4 h. All patients were then extubated and followed until hospital discharge or death.

Primary diagnosis, the cause of intubation, age, sex, body weight, Acute Physiology and Chronic Health Evaluation (APACHE) II score, and endotracheal tube size were recorded on admission to the ICU. Hemoglobin was measured on the morning of the SBT. The arterial blood gas was measured by arterial blood gas analysis before the SBT and the duration of intubation was recorded when the patient had passed the SBT. Postextubation ICU stay (duration of ICU stay after extubation) and postextubation hospital stay (length of hospital stay after extubation) were also recorded. The numbers of days to reintubation and length of stay in the ICU and the hospital were recorded.

Our main outcome was either extubation failure or success as defined by the need to reinsert the endotracheal tube at any time during the study period. Descriptive statistics, including median and range, were given to the continuous variables. The comparison of continuous variables was performed using the Mann-Whitney U test. The categorical data were tabulated with count number and percentage. The comparisons of categorical data were performed using χ² test or Fisher exact test. The association between variables and extubation failure status were assessed using both univariate and multivariate logistic regression. The variables included in the multivariate logistic regression were determined by the evidence of P < .3 in the univariate analysis. The sensitivity and specificity of using CPFi to predict successful extubation was determined. In addition, a receiver operator characteristic curve was constructed according to the clinical data. Analyses were performed using the SPSS 15.0 software package (SPSS Institute Inc.; Chicago, IL). All statistical assessments were two-sided, using a significance level of 0.05.

Of the 150 patients eligible for this study, 118 (78.7%) were successfully extubated, and 32 (21.3%) patients failed extubation. Table 1 summarizes the demographics and clinical features of these 150 patients. Univariate logistic regression analysis was used to explore the correlation between variables and extubation failure status (Table 1). Patients who failed extubation had lower CPFi (range [minimum, maximum]: 42 [21,112] L/min vs 74 [19,138], P < .001), higher APACHE II scores (χ² = 8.32, degrees of freedom [df] = 1, odds ratio [OR] = 1.12, 95% CI [1.04, 1.20], P = .004), and less negative PImax (χ² = 6.76, df = 1, OR = 1.05, 95% CI [1.01, 1.09], P = .009), and were more likely to have abnormal mental status (χ² = 7.19, df = 1, OR = 3.69, 95% CI [1.42, 9.59], P = .007), more sputum volume (χ² = 3.87, df = 1, OR = 1.25, 95% CI [1.10, 1.57], P = .049) and longer postextubation hospital stays (median of success, median of failure: 15.0, 31.5 days, P < .001) and longer postextubation ICU stays (1.0, 9.5 days, P < .001). Higher involuntary cough peak flow (χ² = 21.71, df = 1, OR = 0.95, 95% CI [0.93, 0.97], P < .001), no audible cough (χ² = 3.86, df = 1, OR = 0.18, 95% CI [0.03, 1.00], P = .049) or strong cough (χ² = 12.20, df = 1, OR = 0.04, 95% CI [0.01, 0.24], P < .001) appear to reduce the risk of extubation failure.

Table 2 shows the results of stepwise multivariate logistic regression. Each one-point increase in APACHE II score was associated with an increased risk of extubation failure (OR = 1.12, 95% CI [1.02, 1.22], P = .014). In contrast, higher CPFi values were associated with better outcomes; the OR of extubation failure in patients with higher CPFi was 0.95 (95% CI [0.93, 0.97], P < .001) when compared with those with lower CPFi values.

A receiver operator characteristic curve was also constructed using various cutoff points for CPFi in order to predict extubation outcomes (Fig 2). We chose the cutoff point of 58.5 L/min, where the sensitivity approximates the specificity, with corresponding values of 0.788 for the sensitivity and 0.781 for the specificity. The sensitivity and specificity were based on predicting extubation success, not failure. The positive predictive value was 0.930 and the negative predictive value was 0.500; the area under curve was 0.802 (95% CI [0.706, 0.898]).

Extubation failure increases hospital mortality in patients in ICUs^7-11 and prolongs mechanical ventilation, ICU stay, and hospital stay.^6,7,12 We also found longer ICU stays in our extubation failures. In our multivariate analysis, we demonstrated that involuntary CPF and APACHE II were the primary predictors of extubation success in critically ill patients who passed an SBT.

Cough is defined as deep inspiration followed by strong expiration against a closed glottis, which then opens with an expulsive flow of air, followed by a restorative inspiration. Due to intubation, the vocal folds cannot be closed properly. Therefore, it is difficult to quantify the cough strength, which is important in airway protection. CPFi provides a quantitative method to predict a successful extubation.^3,5,13 For our study, the 58.5 L/min CPFi (CPF 58.5) was chosen as the cutoff point (sensitivity: 78.8, specificity: 78.1%) with a positive predictive value of 93.0% and a negative predictive value of 50.0%. We suggest that decisions to extubate subjects with a CPFi < 58.5 be made cautiously. Other assessments of cough effectiveness include the white card test of Khamiees et al,³ which proved to be a powerful predictor of extubation failure. Smina et al,⁵ used a voluntary CPF and found it effective. Salam et al¹⁴ also used a voluntary CPF and found that patients with a CPF of 60 L/min, remarkably similar to our value of 58.8 L/min, had a five times higher rate of extubation failure compared with those with a CPF > 60 L/min (risk ratio = 4.8; 95% CI [1.4, 16.2]). The voluntary CPF requires good coordination and effort. Although voluntary and involuntary methods were not compared in our study, we suggest that both methods are powerful predictors of extubation failure. However, our involuntary CPF has an advantage when used in patients who cannot cough on command, as, for example, in the case of comatose patients.

Peak expiratory flow is related to strength of the expiratory muscles at high lung volumes Thus, we induced involuntary cough at the end of inspiration.

Many studies have shown that older patients have a higher risk of reintubation compared with younger patients.^7,15 Although the mean age of the extubation failures was 68 years in our study and was higher than the mean age of the extubation success group (66 years), these age differences were not significant.

APACHE II scores are suitable for the evaluation of disease severity. Significant differences in APACHE II scores were found between extubation failure and success groups. The duration of both intubation and mechanical ventilation was between 6 and 8 days in our study. Although high extubation failure rates may be associated with longer mechanical ventilation,^8,16 we found no differences between the rates in our study and other studies.^7,17

Khamiees et al³ found that patients with moderate-to-abundant endotracheal secretions were more than eight times as likely to have unsuccessful extubation as those with no secretions or only mild secretions. However, a later study by Smina et al⁵ found that endotracheal secretion amount was not associated with extubation outcome. Using multivariate analysis, our study also found no association between amount of endotracheal secretion and extubation outcome. Perhaps better quantification of suctioned secretions could demonstrate such an association in the future. Our semiobjective method of involuntary cough strength grading (which accounted for both coughing activity and sputum amount), however, did have a predictive value for extubation outcome in the univariate analysis.

We took the RSBI value into consideration when the SBT was performed. However, we found that the RSBI value is more meaningful when used to wean from a ventilator and less important in terms of predicting a successful extubation, as other studies have shown.¹⁰ No significant difference was found between the extubation failure and success groups regarding RSBI scores (a ventilatory support weaning parameter). The RSBI scores were slightly higher in the failure compared with the success group, but usually < 105. This effect may be due to our protocol, which required RSBI to usually be < 105 before SBT could be initiated. However, there were eight patients who were extubated with RSBI values > 105. Because those eight patients passed the SBT test, they were extubated. Among them, six were extubation successes and only two failed extubation.

Using two different methods of mental status assessment, there were 23 patients whose mental status was inattentive or confused and 11 patients who were unresponsive or unable to verbally communicate. The rates of extubation success for these groups was 56.5% (13/23) and 81.8% (9/11), respectively. We believe that our study is limited in that we needed more patients with poor consciousness who failed extubation (greater sample size) in order to modify the extubation criteria for these patients. In conclusion, involuntary CPF is an inexpensive, objective predictor of extubation success or failure and can be used for patients in ICUs who have passed an SBT, especially those patients who cannot cough on command.

Author contributions:Dr W.-L. Su: contributed to designing the study and writing the protocol, managing literature searches and analyses, and writing the first draft of the manuscript.

Dr Y.-H. Chen: contributed to managing the literature searches and analyses.

Dr C.-W. Chen: contributed to designing the study and writing the protocol.

Mr Yang: contributed to statistical analysis.

Ms C.-L. Su: contributed to statistical analysis.

Dr Perng: contributed to managing literature searches and analyses.

Dr Wu: contributed to designing the study and writing the protocol and approving the manuscript.

Dr J.-H. Chen: contributed to writing the protocol and approving the manuscript.

Financial/nonfinancial disclosures: The authors have reported to CHEST that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.

Other contributions: We thank the respiratory therapists who helped determine the weaning profiles of patients before extubation at the Tri-Service General Hospital. This work was performed at the Tri-Service General Hospital, Taipei, Taiwan.