Abstract
OBJECTIVE: To investigate the predictive value of the maximal inspiratory pressure obtained by a digital vacuometer using a unidirectional valve (PImaxUV) as to weaning outcome, and to compare its performance with the respiratory drive using airway occlusion pressure at 0.1 second (P0.1), and P0.1/PImaxUV.
METHODS: Patients on mechanical ventilation for > 24 hours who fulfilled the weaning criteria were prospectively enrolled. Measurements of PImaxUV and P0.1 were accomplished with a digital vacuometer with a unidirectional valve that allows only exhalation. Measured values were electronically recorded and stored on the digital vacuometer measurement device. Cutoff points for the used parameters were: absolute values of PImaxUV > 30 cm H2O, P0.1 < 2.3 cm H2O, and P0.1/PImaxUV < 0.10. Receiver operating characteristic curves were calculated to compare the predictive values of the indexes.
RESULTS: One hundred three subjects completed the test. The areas under the receiver operating characteristic curve were 0.79 ± 0.04, 0.65 ± 0.05, and 0.74 ± 0.04 for PImaxUV, P0.1, and P0.1/PImaxUV, respectively. The area under the receiver operating characteristic curve for PImaxUV was higher than for P0.1 and P0.1/PImaxUV, but statistical significance was only found against P0.1 (P = .007).
CONCLUSIONS: Every studied index had only a modest performance regarding prediction of weaning outcome. Of note, PImaxUV values obtained by digital technology using a unidirectional valve performed better than historically reported using a conventional techniques, surpassing P0.1 and P0.1/PImaxUV in this regard.
Introduction
Maximal inspiratory pressure (PImax) can be defined as the maximum pressure generated during inspiration against an occluded airway. Measurement of PImax has been used in the clinical setting to globally assess the inspiratory muscle strength for a long time.1,2 Although some ventilator-dependent patients may display poor cooperation when executing voluntary maneuvers, high absolute values of PImax have been associated with successful weaning in some studies.3,4
The PImax performance has been reported in general as having low predictive value to assess the weaning success. In this setting, low performance does not seem to be an exclusivity of PImax, considering that most of studied indexes have demonstrated not more than modest accuracy to predict the weaning outcome.2,5–8
Marini and co-workers suggested an approach for standardization of the operational procedure to measure PImax in such setting. They used a unidirectional valve to permit exhalation while inhalation was blocked, thereby allowing patients to perform the maximal inspiratory effort at a lung volume approaching residual volume.9 PImax values using a unidirectional valve (PImaxUV) are expected to be maximal.
However, even using such a standardized approach, the reproducibility of PImax values in ventilator-dependent patients has been reported to be poor.6,10–12 A clear disadvantage of this approach, which employs an analogic vacuometer, is the direct visual inspection and the manual computation of the PImaxUV values, which lacks accuracy due to the inherent investigator dependence.10–12
Despite such limitations, some parameters have been found to be useful as to prediction of weaning outcome.2,5,7,8,13 Of special interest for the present study, low absolute values of both airway-occlusion pressure 0.1 s after the start of inspiratory flow (P0.1)14,15 and P0.1/PImaxUV have been associated with successful weaning.8
In view of that, we resort to an approach in which the acquisition of PImax and P0.1 measurements are obtained with a digital device that automatically records consecutive inspiratory muscle activity. Our main objective was to investigate the value of the PImaxUV obtained with a digital vacuometer as a weaning outcome predictor, and to compare its performance with P0.1 and P0.1/PImaxUV in this regard.
QUICK LOOK
Current knowledge
Maximal inspiratory pressure (PImax) is defined as the maximum pressure generated during inspiration against an occluded airway. Measurement of PImax is clinically used to globally assess inspiratory muscle strength and predict the success/failure of discontinuation of mechanical ventilation.
What this paper contributes to our knowledge
PImax values obtained by digital technology with unidirectional valve performed better than historically reported values using a conventional technique without a unidirectional valve.
Methods
Patients who were clinically stable and ready to undergo a weaning trial by their primary physicians were enrolled in this prospective observational study. Patients were selected from 2 critical care units in Niterói, Rio de Janeiro, Brazil. The study was approved by the ethics committee of human research of Universidade Federal Fluminense under the number 259/09. An informed consent was obtained from each subject, whenever possible, or from the subject's next of kin.
To be included, subjects had to be older than 18 years and on mechanical ventilation for at least 24 hours. Patients were accepted for enrollment if they met the guidelines of the American Thoracic Society/European Respiratory Society (2007) to start weaning trials,13 in which the inclusion criteria are: resolution of disease acute phase for which the patient was intubated, adequate cough or absence of excessive tracheobronchial secretion, stable cardiovascular status (heart rate ≤ 140 beats/min, systolic blood pressure 90–160 mm Hg, no or minimal vasopressors), stable metabolic status, adequate oxygenation (SaO2 > 90% on FIO2 ≤ 0.4 or PaO2/FIO2 > 150 mm Hg) with PEEP ≤ 8 cm H2O, adequate pulmonary function (respiratory rate [f] ≤ 35 breaths/min or f/VT < 105 breaths/min/L), no substantial respiratory acidosis (pH > 7.30), adequate mentation or Glasgow coma score > 10 (only for subjects with endotracheal tube).
Exclusion criteria were as follow: tracheal stenosis, intracranial pressure > 20 mm Hg, sedation, post-abdominal surgery with evisceration risk, overt cardiac failure or hemodynamic instability, and signs of infection during weaning process. The ventilators used were the 840 (Nellcor/Covidien, Boulder, Colorado) and Servo S (Maquet, Rastatt, Germany).
Procedures
Before spontaneous breathing trials with a T-piece for 30 min with an FIO2 of 0.40, measurements were performed for a 20-second period. Three parameters were analyzed: PImaxUV, P0.1, and P0.1/PImaxUV. PImax and P0.1 were obtained using a unidirectional valve method and a digital vacuometer (MVD 300, Globalmed, Porto Alegre, Rio Grande do Sul, Brazil). The device has a scale of 300 cm H2O with a 1 cm H2O interval, and was designed to perform measurements at each 0.1 s. Prior to testing, all subjects were on pressure support ventilation mode (12–20 cm H2O) and were not on sedatives. The head of the subject's bed was elevated to 45 degrees and the cuff of the artificial airway was hyperinflated to prevent air leakage during measurement.12 The subject had the airway cleared of secretions by suctioning with standard tracheal suction procedures after pre-oxygenation for 2 minutes with FIO2 of 1.0. After a brief rest period, pre-oxygenation was repeated. The subject was then disconnected from mechanical ventilation and allowed to breathe spontaneously for 10 seconds before connecting the artificial airway to the digital vacuometer when the subject was at functional residual capacity.12,16 The subject was continuously monitored by pulse oximetry and electrocardiogram, and had continual respiratory physiotherapist attention. Measurements were digitally recorded, allowing easy access to values of parameters whenever needed. Figure 1 shows a typical plot of inspiratory pressure against time obtained during a test. For the present study, respiratory drive (P0.1) was calculated as an average of the measured inspiratory pressure 0.1 s after the initiation of the inspiration for the first 3 cycles, and P0.1/PImaxUV as the ratio of absolute values of P0.1 and PImaxUV. The subjects were not coached for the maneuver. Thresholds for the 3 parameters were calculated from their respective receiver operating characteristic (ROC) curves. Weaning was considered successful if spontaneous breathing was sustained for > 48 hours after withdrawal from mechanical ventilation. The decision to return to mechanical ventilation was made by the respiratory physiotherapist and physician in charge (who was blinded to the results of the indexes evaluated), based on the clinical and laboratory signs of poor tolerance, as mentioned below.
Typical register showing that the maximal inspiratory peak pressure (59 cm H2O), along approximately 21 seconds of observation, was achieved at approximately 20 seconds.
The trial was stopped if at least one of the following intolerance criteria was present: SaO2 < 90% or PaO2 < 60 mm Hg with FIO2 > 0.4; PaCO2 > 50 mm Hg or increased by 8 mm Hg or more; arterial pH < 7.33 or decreased by 0.07 or more; respiratory rate > 35 breaths/min or increased by 50% for 5 min or longer; heart rate >140 beats/min or a sustained increase or decrease in > 20%; mean blood pressure > 130 mm Hg or < 70 mm Hg; or presence of agitation, diaphoresis, disorientation, or depressed mental status.
The subjects who demonstrated one of these signs during the spontaneous breathing trial or within 48 hours after the discontinuation from mechanical ventilation were considered not weaned and were returned to ventilatory support.13
Statistical Analysis
The performance of the indexes to predict weaning outcome was evaluated by sensitivity, specificity, positive predictive value, negative predictive value, likelihood ratio of positive test, likelihood ratio of negative test, and diagnostic accuracy. The predictive performance of each index was also evaluated through the calculation of the area under the ROC curve (Fig. 2). The area under the curve (AUC) was compared using the Hanley and McNeil method.17 P values < .05 were considered significant. Statistical analysis was performed employing statistics software (SPSS 18.0, SPSS, Chicago, Illinois). Comparison between ROC curves was accomplished using statistics software (MedCalc 11.4.2.0, Mariakerke, Oost-Vlaanderen, Belgium).
Receiver operating characteristic curves for airway occlusion pressure at 0.1 second (P0.1), maximal inspiratory pressure obtained by a digital vacuometer using a unidirectional valve along 20 seconds of airway occlusion (PImaxUV), and P0.1/PImaxUV as predictors of successful weaning. * P = .007 versus P0.1
Results
From 128 initially screened subjects, 103 were selected for the study. Reasons for exclusion were: Glasgow coma score < 10 (n = 4, in subjects on orotracheal tube), reinfection (n = 6), hypoxemia and bradycardia during the test (n = 3), tracheomalacia (n = 5, in subjects with tracheostomy), laryngospasm (n = 2, in subjects on orotracheal tube), and informed consent not obtainable (n = 5). General features of the 103 subjects evaluated are summarized in Table 1. Forty-two (40.8%) subjects had an endotracheal tube and 61 (59,2%) tracheotomy, with internal diameters varied from 7.5 mm to 8.5 mm. The averages of respiratory drive P0.1, PImaxUV, and P0.1/PImaxUV in the 103 subjects were 2.98 ± 1.02 cm H2O, 32.01 ± 18.15 cm H2O, and 0.11 ± 0.05 cm H2O, respectively. From the 103 studied subjects, 60 subjects were weaned. General features are summarized in Table 1.
General Features of Studied Patients
Quality indicators of the 3 indexes used to predict the weaning outcome are in Table 2. Cutoff values (calculated from the ROC curves) for PImaxUV, P0.1, and P0.1/PImaxUV were 30 cm H2O, 2.33 cm H2O, and 0.10, respectively. PImaxUV showed the best diagnostic accuracy, with 0.72, followed by P0.1/PImaxUV, with 0.69, and P0.1, with 0.66.
Quality Indicators of the Indexes Used to Predict the Weaning Outcome
The AUC of the 3 indexes are in Figure 2. The largest AUC was found for PImaxUV, but statistical significance was only found toward P0.1 (0.797 ± 0.044 vs 0.650 ± 0.056, P = .007). The P values for comparisons between PImaxUV and P0.1/PImaxUV, and between P0.1 and P0.1/PImaxUV were 0.16, and 0.18, respectively.
Discussion
Values of PImax have been suggested as useful to predict successful weaning, but its performance varies substantially among studies.2–8 Despite such discrepancy, PImax remains a helpful parameter in the decision for weaning.13 In our study, we reevaluated the predictive performance of PImaxUV in comparison to P0.1 and P0.1/PImaxUV in a new setting: measurements were performed using a digital vacuometer and a unidirectional valve. The employment of digital technology that automatically records the test results allowed us to prevent the observer from influencing the data collection, to revisit the data whenever needed, and to collect PImaxUV and P0.1 values at one step. The use of such technology was found to be very simple with the potential of being easily incorporated in daily ICU clinical practice.
From the ROC curves, calculated cutoff values for PImaxUV, P0.1, and P0.1/PImaxUV were 30 cm H2O, 2.3 cm H2O, and 0.10, respectively. It is interesting to mention that the cutoff value found for PImaxUV matches the value cited in most relevant studies.3,4,5 The same did not happen regarding P0.1 and P0.1/PImaxUV, whose calculated cutoff values were lower than traditional ones.7,8,14–16,18–20
When comparing performance of the 3 studied indexes regarding prediction of successful weaning, a higher sensitivity was found for P0.1 (73% vs 63% and 68% for PImaxUV and P0.1/PImaxUV, respectively). However, the best accuracy was found for PImaxUV (72% vs 66%, and 69% for P0.1 and P0.1/PImaxUV, respectively) reflecting the low specificity of P0.1 (56% vs 84% and 70% for PImaxUV and P0.1/PImaxUV, respectively). It should be pointed out that the low specificity of P0.1 in the present study can be partially accounted for by the prolonged time of mechanical ventilation (mean 17.5 d) and the high number of subjects with neuromuscular diseases and brain trauma or stroke in our sample. This can be associated with low values of P0.1 that are not indicative of good prognosis.6,8 Substantiating these findings, the highest value of the AUC of the ROC curves was found for PImaxUV.
Our findings are markedly different from previous studies in which low specificity, low diagnostic accuracy, and low AUC of the ROC curves have been reported for PImax in this regard.2,5,8 The use of new tools in the present study may have accounted for these differences. In contrast to studies that employed total occlusion, and manual registration of visually inspected PImax values on an analogic vacuometer, which typically carries poor precision, we resorted to a unidirectional valve and digital registration of PImaxUV values, which enables measurements at each 0.1 s with a precision of 1 cm H2O.
Many indexes are cited in the literature as useful to predict weaning outcome.2,5,7,8,13,20 The most used and recommended one is f/VT, proposed by Yang and Tobin.5 In a recent study, the integrative weaning index exhibited better performance than f/VT, but these results await confirmation.21 In the present study the performance of the PImaxUV was still below the ones reported for those 2 indexes, but was enough to place this easy to obtain parameter among the top 3 to 4 predictors of weaning outcome.
Some limitations of our study deserve comments. An innovative technique was used to obtain P0.1 values in our study. The gold standard technique to measure this parameter involves insertion of an esophageal catheter.22,23 As an alternative, some commercially available ventilators are equipped with software that can directly provide P0.1 values.8,22,23 Our innovative way of measurement produced values of P0.1 of 2.98 ± 1.02 cm H2O, which were very close to the ones obtained in a previous report (2.97 ± 1.68 cm H2O).8 Further studies are needed to compare values collected by the technique used in the present study with the ones obtained by more traditional ways.
Conclusions
In conclusion, PImaxUV values obtained by digital technology using a unidirectional valve performed better than historically reported conventional techniques, surpassing P0.1 and P0.1/PImaxUV in prediction of weaning outcome. However, it should be acknowledged that the present study revealed modest ROC values for every tested index.
Acknowledgments
The authors thank the respiratory physiotherapists and physicians of the ICU of the Hospital e Clínica São Gonçalo and Hospital Estadual Azevedo Lima, for their collaboration and dedication in our study.
Footnotes
- Correspondence: Leonardo Cordeiro de Souza PT MSc, Estrada da Paciência, 481 Casa 65, CEP 24756–660 Maria Paula, São Gonçalo, Rio de Janeiro, Brazil. E-mail: leonardo.uti{at}gmail.com.
The authors have disclosed no conflicts of interest.
- Copyright © 2012 by Daedalus Enterprises Inc.
