Prognostic Value of Different Dead Space Indices in Mechanically Ventilated Patients With Acute Lung Injury and ARDS*

Umberto Lucangelo, Francesca Bernab, Sara Vatua, Giada Degrassi, Ana Villagr, Rafael Fernandez, Pablo V. Romero, Pilar Saura, Massimo Borelli and Lluis Blanch Chest 2008;133;62-71; Prepublished online November 7, 2007; DOI 10.1378/chest.07-0935 The online version of this article, along with updated information and services can be found online on the World Wide Web at: http://chestjournal.chestpubs.org/content/133/1/62.full.html

Chest is the official journal of the American College of Chest Physicians. It has been published monthly since 1935. Copyright2008by the American College of Chest Physicians, 3300 Dundee Road, Northbrook, IL 60062. All rights reserved. No part of this article or PDF may be reproduced or distributed without the prior written permission of the copyright holder. (http://chestjournal.chestpubs.org/site/misc/reprints.xhtml) ISSN:0012-3692

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

CHEST

Original Research
CRITICAL CARE MEDICINE

Prognostic Value of Different Dead Space Indices in Mechanically Ventilated Patients With Acute Lung Injury and ARDS*
Umberto Lucangelo, MD; Francesca Bernabe, MD; Sara Vatua, MD; ` Giada Degrassi, MD; Ana Villagra, MD; Rafael Fernandez, MD; ` Pablo V. Romero, MD; Pilar Saura, MD; Massimo Borelli, MS; and Lluis Blanch, MD, PhD

Study objective: The aim of this prospective observational study was to evaluate the utility of derived dead space indexes to predict survival in mechanically ventilated patients with acute lung injury (ALI) and ARDS. Study population: Thirty-six patients with ALI (Murray score, >1; PaO2/fraction of inspired oxygen [FIO2] ratio, < 300) in critical care departments at two separate hospitals entered the study. Measurements: At ICU admission, 24 h, and 48 h, we measured the following: simplified acute physiologic score II; PaO2/FIO2 ratio; respiratory system compliance; and capnographic indexes (Bohr dead space) and physiologic dead space (Enghoff dead space [VDphys/VT]), expired normalized CO2 slope, carbon dioxide output, and the alveolar ejection volume (VAE)/tidal volume fraction (VT) ratio. Results: The best predictor was the VAE/VT ratio at ICU admission (VAE/VT-adm) and after 48 h (VAE/VT-48 h) [p 0.013], with a sensitivity of 82% and a specificity of 64%. The difference between VAE/VT-48 h and VAE/VT-adm show a sensitivity of 73% and a specificity of 93% with a likelihood ratio (LR) of 10.2 and an area under the receiver operating characteristic (ROC) curve of 0.83. The interaction between the PaO2/FIO2 ratio and VAE/VT-adm predict survival (p 0.003) with an area under the ROC curve of 0.84, an LR of 2.3, a sensitivity of 100%, and a specificity of 57%. The VDphys/VT after 48 h predicted survival (p 0.02) with an area under the ROC curve of 0.75, an LR of 8.8, a sensitivity of 63%, and a specificity of 93%. Indexes recorded 24 h after ICU admission were not useful in explaining outcome. Conclusions: Noninvasive measures of VAE/VT at ICU admission and after 48 h of mechanical ventilation, associated with PaO2/FIO2 ratio provided useful information on outcome in critically ill patients with ALI. (CHEST 2008; 133:6271)
Key words: acute lung injury; mechanical ventilation; prognosis; pulmonary dead space; volumetric capnography Abbreviations: ALI acute lung injury; CI confidence interval; Crs respiratory system compliance; DSA dead space allowance; Fio2 fraction of inspired oxygen; LS likelihood ratio; OI oxygenation index; PEEP positiveend expiratory pressure; Petco2 end-tidal Pco2; ROC receiver operating characteristic; Rrs tissue resistance; Rrs total respiratory resistance; SAPS simplified acute physiologic score; SlopeN-75% normalized expired CO2 slope; SOFA sequential organ failure assessment; Vae alveolar ejection volume; Vae/Vt difference between alveolar ejection volume/tidal volume ratio at 48 h and at ICU admission; Vae/Vt-adm alveolar ejection volume/tidal volume ratio at ICU admission; Vae/Vt-48 h alveolar ejection volume/tidal volume ratio at 48 h; Vco2 carbon dioxide output; VdBohr/Vt Bohr dead space; Vdphys/Vt Enghoff dead space; Vdphys/Vt-48 h Enghoff dead space after 48 h; Vt tidal volume

62

Original Research

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

end-tidal Pco (Petco B edside measurements ofspace fraction have been) and pulmonary dead
2 2

correlated with outcome in patients undergoing emergency trauma surgery1 and in ARDS patients,2 respectively. However, Petco2 can only be used to approximate the fraction of CO2 in alveolar gas when end-tidal and alveolar CO2 fractions are identical.3,4 In ARDS patients, the degree of respiratory dead space alteration measured in standardized clinical conditions (ie, fixed tidal volume [Vt] and positive end-expiratory pressure [PEEP]), at ICU admission or beyond the first 24 h, is characteristic of nonsurvivors and may have a prognostic value.2,5 The advanced technology combination of airway flow monitoring and mainstream capnography enables noninvasive breath-by-breath bedside calcula tion of carbon dioxide output (Vco2) per breath, the slope of phase III of the expired capnogram plotted as a function of exhaled Vt (ie, CO2-Vt curve),3 Bohr dead space (VdBohr/Vt),6 and the alveolar ejection volume (Vae)/Vt ratio, independent of set ventilatory parameters.7 Variations in Vco2 during rebreathing and the rate of rise in expired CO2 have been used as noninvasive measures of cardiac output8 and ventilation/perfusion inequalities.9 Volumetric capnography has also been used for bedside monitoring during pulmonary embolism.10,11 The Vae/Vt ratio, an index of alveolar dishomogeneity, correlates with the severity of lung injury12 and is not influenced by the ventilatory settings in mechanically ventilated patients with acute lung injury (ALI) and ARDS.7,12 This study aims to evaluate the utility of these noninvasive capnographic indexes to predict
*From the Department of Perioperative Medicine, Intensive Care and Emergency (Drs. Lucangelo, Bernabe, Vatua, and ` Degrassi), Cattinara Hospital, Trieste University School of Medicine, Trieste, Italy; Critical Care Center (Drs. Villagra, Fernan` dez, Saura, and Blanch), Hospital de Sabadell, Institut Universitari Fundacio Parc Taul, Universitat Autonoma de Barcelona, ` Barcelona, Spain; Laboratorio de Neumologa Experimental (Dr. Romero), Hospital Universitario de Bellvitge, LHospitalet de Llobregat, Llobregat, Spain; and the Department of Mathematics and Computer Science (Mr. Borelli), Trieste University, Trieste, Italy. This study was supported by Plan Nacional de Investigacion Cientfica, Desarrollo e Innovacion Tecnologica and Instituto de Salud Carlos III Red GIRA (G03/063), Red Respira (ISCiii RTIC 03/11), and Institut Universitari Fundacio Parc Taul. The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Manuscript received April 17, 2007; revision accepted October 2, 2007. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Correspondence to: Umberto Lucangelo, MD, Ospedale di Cattinara, Dipartimento di Medicina Perioperatoria, Terapia Intensiva ed Emergenza, Strada di Fiume 447, I-34149 Trieste, Italy; e-mail: u.lucangelo@fmc.units.it DOI: 10.1378/chest.07-0935
www.chestjournal.org

outcome at ICU admission and after 24 to 48 h of treatment in mechanically ventilated patients with ALI or ARDS.

Materials and Methods We prospectively studied 36 mechanically ventilated patients with ALI who had been admitted to the ICU of the Hospital of Sabadell (Sabadell, Spain) [22 patients] and the Hospital of Cattinara, Trieste (Trieste, Italy) [14 patients]. The protocol was approved by the ethics committees at both institutions, and informed consent was waived because of the observational nature of the study. ALI was defined as a Murray lung injury score of 113 and a Pao2/ fraction of inspired oxygen (Fio2) ratio of 300.14 The exclusion criteria were age 18 years, moribund state, chronic hypercapnic disease, pregnancy, massive pulmonary embolism, chest wall deformity, bronchopleural fistula, and extraalveolar air on chest radiograph. Patients who died within 48 h were also excluded from the study. The simplified acute physiologic score (SAPS) II was calculated at ICU admission. Data were recorded after 30 min of stable clinical condition in the following three periods: within 12 h of ICU admission; after 24 h; and after 48 h. The sequential organ failure assessment (SOFA) score was also calculated at ICU admission, after 24 h, and after 48 h. Gasometric, hemodynamic, and respiratory mechanics parameters were recorded, and capnographic indexes were calculated. Patients were orally intubated and mechanically ventilated in assist/control mode with a lung-protective strategy (Vt of 6 to 8 mL/kg and moderate/high PEEP) based on the attending physicians criteria. Sedation was continuously maintained with midazolam (0.03 to 0.2 mg/kg/h). The patients clinical management and the decision to initiate invasive cardiac output monitoring (in 10 patients) was determined by intensivists not involved in the study. Breath-by-breath volumetric capnography (ie, CO2Vt curve) was obtained by a mainstream capnograph with a differential pressure pneumotachometer (Novametrix CO2SMO ; Novametrix Medical Systems; Wallingford, CT) connected between the endotracheal tube and the ventilator Y-piece. Airflow, pressure at the airway opening, and CO2 concentration were simultaneously recorded for a 3-min period by the capnograph at a sampling frequency of 100 Hz. CO2 concentration was measured by infrared absorption with a response time of 75 ms. The device was calibrated before each experiment according to the manufacturers instructions. Ten nonconsecutive cycles were randomly selected for analysis from the recordings in each patient and
CHEST / 133 / 1 / JANUARY, 2008

63

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

condition. Computerized analyses of these cycles were performed (using software specially developed in our laboratory for this purpose based on Labview, version 3.1; National Instruments; Austin, TX). Vco2 was calculated based on the mathematical integration of the measured flow and CO2 signal, and was normalized for minute ventilation. The slope of phase III of the CO2-Vt curve was determined by least-squares linear regression between 75% of the expired volume and the end-tidal point. To avoid differences due to different ventilatory frequencies, the slope of phase III was normalized by dividing it by Petco2 (ie, the normalized expired CO2 slope [SlopeN-75%]).15 The VdBohr/Vt was calculated from common equations.3 Physiologic dead space (Vdphys/Vt) was calculated from the Enghoff modification of the Bohr equation.3,4 The newest index, the Vae/Vt ratio, was calculated as previously de scribed.7 Briefly, Vco2 plotted as a function of expired volume originates a Vco2-Vt curve; from this curve, the last 50 points of every cycle were back-extrapolated by least-squares linear regression analysis representing the ideal lung behavior (straight dotted line). Assuming a fixed amount of dead space contamination of 5% (ie, dead space allowance [DSA]), a straight line having its maximal value at endexpiration and a slope of 0.95 (1 DSA) times that of the ideal line is plotted. Alveolar ejection begins at the intersection between the Vco2-Vt curve and the straight line. The volume between this point and end-expiration is the Vae, and this is expressed as the Vae/Vt ratio.7,16 Airflow and pressure curves obtained by the capnograph (CO2SMO ; Novametrix Medical Systems) were subsequently analyzed to obtain Vt by integrating flow over time. Then, respiratory system compliance (Crs) and total respiratory system resistance (Rrs), together with its partitioning on airway resistance and tissue resistance ( Rrs) components, were calculated as previously described.12 Finally, the oxygenation index (OI; ie, mean airway pressure Fio2 100/Pao2) was also calculated on ICU admission and after 24 and 48 h. Statistical Analysis Parametric and nonparametric values were expressed as the mean SD, and median and quartile, respectively. Student/Welch t tests were used to compare mean values, and Wilcoxon tests were used to compare ranks between survivors and nonsurvivors. A maximal linear model (explanatory variables and nonlinear interaction were included) was used to analyze the binary response outcome related to the explanatory variables studied. Receiver operating characteristic (ROC) curves were obtained for prog64

nostic variables. Statistical analysis was performed using an open-source statistical package (R [www. R-project.org]).17 Results We included 36 patients, 12 female and 24 male, with a mean age of 66 15.2 years. No patients died within the first 48 h. Patients coming from the two ICUs differed in mean age (Sabadell Hospital, 61.3 16.5 years; Trieste Hospital, 73.4 9.4 years; p 0.05). The clinical variables studied (Tables 1, 2) and the mortality rate were similar in the Sabadell and Trieste Hospitals (41% and 36%, respectively; difference was not significant). Consequently, the data were analyzed as if from a single population. The median length of mechanical ventilation was 10 days (minimum, 3 days; lower quartile, 7 days; upper quartile, 15.5 days; maximum, 73 days), and the median length of ICU stay was 12 days (minimum, 3 days; lower quartile, 8.5 days; upper quartile, 20.5 days; maximum, 73 days). The ICU mortality

Table 1Comparison of Physiologic Scores, Gas Exchange, Mechanical Ventilatory Parameters, and Volumetric Capnographic Indices at ICU Admission in Survivors and Nonsurvivors*
Variables SAPS II SOFA score Lung injury score OI Crs, mL/cm H2O Pao2/Fio2, mm Hg Paco2, mm Hg Mean arterial pressure, mm Hg Cardiac output, L/min (n 10) Minute volume, L Respiratory rate, breaths/min Vt, mL/kg Plateau pressure, cm H2O PEEP, cm H2O Rrs, cm H2O/L/s Rrs, cm H2O/L/s SlopeN-75% VDBohr/Vt VDphys/Vt Vae/Vt ratio Vco2, mL/min Survivors (n 22) 37.5 (3245) 7 (69) 1.63 (1.252) 7.6 3.6 38 11.7 201 60 38 9.2 82 12.8 6.3 8.3 15 7.0 25 9 3.9 13.5 0.67 0.41 0.47 0.42 214 2.7 2.3 3.7 1.2 3.5 4 2.5 5.8 0.38 0.19 0.19 0.17 82 Nonsurvivors (n 14) 49 (3654) 8.5 (510) 2.38 (1.253) 10.5 4.8 32 11.7 194 96 40 10.2 72 14.3 4.7 9.2 20 6.4 24 8.3 3.8 14.8 0.74 0.44 0.55 0.41 225 2.1 4.2 7.8 1.1 7 2.6 3.4 6.3 0.43 0.17 0.13 0.13 115 p Value 0.144 0.463 0.112 0.147 0.175 0.799 0.481 0.052 0.325 0.432 0.028 0.178 0.795 0.645 0.944 0.279 0.649 0.610 0.152 0.970 0.820

*Values are given as the mean SD or as median (interquartile range), unless otherwise indicated. Reached statistical significance (p 0.05) due to the application of a protective ventilatory strategy in ARDS patients.
Original Research

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

Table 2Time Series Data From ICU Admission to 24 and 48 h After in the Variables Studied Among Survivors and Nonsurvivors*
Variables SOFA D S OI D S Pao2/Fio2 ratio, mm Hg D S Plateau pressure, cm H2O D S Rrs, cm H2O/L/s D S Crs, mL/cm H2O D S Rrs, cm H2O/L/s D S MAP, mm Hg D S VDBohr/Vt D S Vdphys/Vt D S SlopeN-75% D S Vco2, mL/min D S Vae/Vt ratio D S Vt, mL/kg D S PEEP, cm H2O D S Respiratory rate, breaths/min D S At ICU Admission 8.5 (510) 7.0 (69) 10.5 7.6 4.8 3.6 At 24 h 9.5 (611) 8.0 (710) 11.2 7.7 6.4 3.6 At 48 h 7.5 (512) 7.0 (59) 10.7 6.5 7.0 2.8

194 201

96 60

191 199

104 57

192 231

100 83

24.4 24.9 16.2 15.0 32 37 3.9 4.1 72 82 0.45 0.43 0.55 0.48 0.74 0.69 225 214 0.42 0.42 6.4 7 8.3 9

7.2 3.9 5.2 9.9 12.2 12.6 3.48 2.78 14.3 12.8 0.18 0.18 0.13 0.17 0.43 0.39 115 82 0.13 0.17 1.1 1.2 2.6 4

24.9 23.4 17.6 15.2 33 38 4.4 3.9 78 81 0.43 0.43 0.54 0.48 0.77 0.62 220 215 0.36 0.44 7 6.8 7.4 8.9

6.8 4.6 3.2 6.7 16.1 10.3 2.73 2.32 11.7 11.3 0.18 0.16 0.14 0.14 0.37 0.31 81 80 0.13 0.17 2.2 1 3.3 3.7

25.2 23.9 15.0 16.4 35 37 3.6 3.4 78 84 0.44 0.40 0.56 0.44 0.86 0.66 223 236 0.34 0.46 7.2 6.6 8.8 8.8

7.1 4.5 5.1 7.2 15.7 12.2 2.50 2.87 12.0 13.7 0.19 0.17 0.13 0.15 0.34 0.37 82 86 0.11 0.17 2 1 2 2.7

rate was 39% (14 patients died and 22 patients survived), while the SAPS II-predicted mortality rate was 27%. The median number of ventilation-free days (calculated on the 28th day) in the 22 patients who survived was 19 days (minimum, 2 days; lower quartile, 16 days; upper quartile, 22 days; maximum, 25 days) and 0 days (minimum, 0 days; lower quartile, 0 days; upper quartile, 0 days; maximum, 12 days) in the 14 patients with poor outcome (p 0.01). Baseline data are shown in Table 1, and repeated measures of recorded variables after 24 and 48 h are shown in Table 2. To investigate the prognostic values of explanatory variables (Table 2), a linear model analysis was applied. The main outcome predictor was derived from Vae/Vt ratio at ICU admission (Vae/Vt-adm) and after 48 h (Vae/Vt-48 h), as expressed by the minimal adequate model to predict outcome (p 0.013) in the following linear equation: Outcome 0.38 1.49 (Vae/Vt-adm) 2.06 (Vae/Vt-48 h) (1)

20 15

7.8 3.7

19 16

6.8 3.9

19 16

6.8 4.4

*Values are given as the median (interquartile range) or mean SD. D death; S survival; MAP mean arterial pressure; SlopeN75% normalized expired CO2 slope. Reached statistical significance (p 0.05).

Values of 0.5 can be interpreted as a prediction of survival, and values of 0.5 indicate poor outcome (due to the choice of levels, 0 D, 1 S). In our study, the model is characterized by four falsepositive results, who survived even though the predictor value was 0.5, and by five false-negative results (sensitivity, 82%; specificity, 64%). The linearity of equation 1 enables outcome prediction to be simplified, considering the difference between Vae/Vt-48 h and Vae/Vt-adm ( Vae/Vt). To find a Vae/Vt cutoff value maximizing sensitivity and specificity, an ROC curve analysis was performed. Using 0.011 as the cutoff for Vae/Vt yielded 73% sensitivity and 93% specificity with a likelihood ratio (LR) of 10.2 and an area under the ROC curve of 0.83 (95% confidence interval [CI], 0.67 to 0.93). In this case, outcome prediction reliability increases (odds ratio, 44.2) with five false positive results (survived with Vae/Vt 0.011) and only one false-negative result (ie, patient died with Vae/Vt of 0.011). Lung injury score, Pao2/Fio2 ratio at ICU admission, and Vdphys/Vt after 48 h (Vdphys/VT-48 h) were identified by linear model analyses. Lung injury score as an outcome predictor showed an area under the ROC curve of 0.66, an LR of 2.3, a sensitivity of 82%, and a specificity of 64%, with a cutoff of 2. Outcome related to OI (p 0.01) is expressed by the following formula: Outcome 1.16 0.02 (Pao2/Fio2-adm) (Pao2/Fio2-adm)2 (2)
65

0.000045
www.chestjournal.org

CHEST / 133 / 1 / JANUARY, 2008

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

showed an ROC curve area of 0.79, an LR of 3.0, a sensitivity of 86%, and a specificity of 71%, with a cutoff of 0.53. Vdphys/Vt-48 h was expressed by the following formula (p 0.02): Outcome 1.24 1.36 (Vdphys/Vt-48) (3) which exhibited an area under the ROC curve of 0.75, an LR of 8.8, a sensitivity of 63%; and a specificity of 93%, with a cutoff of 0.46. Finally, the results of a more complex linear model including interactions between the significant explanatory variables considered suggests that outcome could be predicted (p 0.003) by the following equation: Outcome 1.84 5.7 0.013 (Pao2/Fio2-adm) 0.03 (4)

Table 3Comparison of Physiologic Scores, Gas Exchange, Mechanical Ventilatory Parameters, and Volumetric Capnographic Indices at ICU Admission in ALI and ARDS Patients*
Variables Age, yr SAPS II SOFA score Lung injury score OI Crs, mL/cm H2O Pao2/Fio2 ratio, mm Hg Paco2, mm Hg Mean arterial pressure, mm Hg Cardiac output, L/min (n 10) Minute volume, L Respiratory rate, breaths/min Vt, mL/kg Plateau pressure, cm H2O PEEP, cm H2O Rrs, cm H2O/L/s Rrs, cm H2O/L/s SlopeN-75% VDBohr/Vt VDphys/Vt Vae/ Vt ratio Vco2, mL/min ALI Patients (n 26) 64.3 15.5 37.5 (3252) 6.5 (59) 1.3 (12) 7.0 2.7 37 10.8 225 68 37 80 5.7 8.5 15.9 7.0 23.9 8.6 3.7 13.8 0.68 0.42 0.49 0.42 226 7.9 12.7 1.8 2.9 4.6 1.6 4.2 3.6 2.4 5.9 0.36 0.18 0.18 0.16 111 ARDS Patients (n 10) 70.3 14.5 44 (4153) 10 (811) 2.63 (2.253) 13.2 4.5 31.4 14.1 128 37 42.6 75 5.3 8.9 20 6.4 26 9.8 4.1 15.1 0.75 0.42 0.54 0.40 232 12.7 17.3 3.2 3.9 8.3 1.2 6.9 3.3 3.9 6.1 0.49 0.17 0.14 0.14 106 p Value 0.300 0.322 0.015 0.01 0.01 0.203 0.01 0.139 0.400 0.812 0.783 0.068 0.243 0.271 0.362 0.694 0.353 0.644 0.950 0.444 0.632 0.877

(Vae/Vt-adm)

(Pao2/Fio2-adm

Vae/Vt-adm)

with an area under the ROC curve of 0.84, an LR of 2.3, a sensitivity of 100%, and a specificity of 57%, with a cutoff of 0.37. Indexes recorded after 24 h from ICU admission were not useful in explaining the outcome. With the aim of further assessing its validity in predicting outcome, we checked our simplest model ( Vae/Vt) in the subpopulations of ALI and ARDS patients defined according to the European-North American Consensus14 (Table 3). Applying the previously obtained 0.011 Vae/Vt cutoff level in 10 ARDS patients (pulmonary ARDS, 8 patients; extrapulmonary ARDS, 2 patients) [Fig 1], the ROC curve analysis showed 67% sensitivity, 100% specificity, and a 0.72 area under the ROC curve (95% CI, 0.36 to 0.94). The 26 ALI patients (pneumonia, 5 patients; severe sepsis, 7 patients; lung contusion, 4 patients; postoperative respiratory failure, 10 patients) showed 74% sensitivity and 86% specificity (ie, there was only one false-negative result over 7 patients) with an LR of 5.16 and an 0.83 area under the ROC curve (95% CI, 0.63 to 0.95).

*Values are given as the mean unless otherwise indicated. p 0.05. p 0.01.

SD or median (interquartile range),

for ventilator adjustments to obtain dead space measurements. In the past few years, slight improvements in mortality prediction, obtained by serial measurements of Vdphys/Vt during the first week of ARDS, have been reported.5 Our study confirms previous data on the prognostic value of dead space, and, more interestingly, shows that variations in dead space and in Vae/Vt index measured without ventilator adjustments are also associated with outcome in patients ventilated with a nonaggressive ventilatory strategy.5,7,19 22 Limitations of the Study The limitations of our study are mainly related to sample size, the length of serial measurements, and the high mortality rate. Our sample size was based only on the availability of patients meeting the inclusion criteria and the affordability of the study. Whereas the small sample may explain the failure of some variables to reach significance, it does not affect our message based on the variables that easily
Original Research

Discussion In ARDS patients, pioneering studies18 have shown that physiologic dead space and its evolution in the first days of the disease were associated with death. Only in the last decade have Vdphys/Vt measurements regained the attention of researchers. Nuckton et al2 published the most successful study suggesting that high Vdphys/Vt is independently associated with increased risk of death in ARDS patients. The disadvantage of this method is the need
66

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

reached statistical significance. The length of the serial assessment is always open to discussion, but the fact that our data on Vae/Vt ratio already yielded impressive results at 48 h allows us to accept this interval as a convenient predictive tool. The usefulness of an index for detecting patients with early mortality will decrease as the interval is increased. Our ALI/ARDS patients had a 39% mortality rate for a mean SAPS II of 41, whereas patients in the study by Nuckton et al2 had a 42% mortality rate for a mean SAPS II of 47. Whether this apparent excess in mortality is related to our small sample size or to a different case mix is not easily recognizable. Bearing in mind all of the above-mentioned limitations, this clinical research should be considered to be a pilot study. Outcome Prediction in ARDS Outcome prediction in patients with ALI is an unresolved issue. The common severity scores correlate with survival but have low predictive power not only for a given patient, but also for different groups. In a 2003 study23 in ARDS patients, even at the highest SAPS II (ie, 49), the risk was increased fourfold, but with a 95% CI from 1.7 to 11.1. In our study, the SAPS II and SOFA score of survivors were lower than those of nonsurvivors, but, again, their predictive power was low. Oxygenation derangement, commonly expressed as Pao2/Fio2 ratio and OI have also been correlated with survival, but, again, with low predictive value. On the other hand, a statistically significant difference in SOFA, Pao2/ Fio2 ratio, and OI at ICU admission is seen between ALI and ARDS patients, as expected24,25 (Table 3). Patients with poor outcome had higher Vdphys/Vt values than survivors at ICU admission (15%) and Vdphys/Vt-48 h (27%). This trend, supported by a concomitant increment of the SlopeN-75% and by a constant behavior of VdBohr/Vt, suggests a possible link between Vdphys/Vt variations and mortality. In our study, Vdphys/Vt-48h reached a specificity of 93%. From a clinical viewpoint, it is important to reach the highest possible specificity early to minimize false-positive results. On the other hand, using a more sophisticated statistical model could increase specificity, but complex formulas often lose clinical relevance. Therefore, choosing a higher grade mathematical model conflicts with the principles of daily clinical application, and such a model would have to be tailored for fast applicability and easy comprehension. Although the complex model in equation 4 that considers Pao2/Fio2 ratio at ICU admission and Vae/Vt-adm reached 100% sensitivity, the specificity remains very poor, in disagreement with our goals.
CHEST / 133 / 1 / JANUARY, 2008

Figure 1. The capability of Vae/Vt for predicting outcome evaluated in subpopulations of ARDS and ALI patients. Top, A: puts in evidence 100% specificity in predicting outcome in ARDS patients (no false-negative results). Bottom, B: shows 86% specificity in predicting outcome in ALI patients (one false-negative result in seven cases). The straight line represents a cutoff of 0.011, as predicted by our simplified generic model.

www.chestjournal.org

67

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

In our study, a simple model based only on the difference between volumetric capnographic index values recorded at admission and two days after ICU admission showed the highest specificity (93%) combined with a high LR (10.2) and very favorable odds ratio (44.2). Therefore, our results show that it could be useful to evaluate Pao2/Fio2 and Vae/Vt ratio in ALI and ARDS patients at ICU admission by applying our formula 4 (see the Materials and Methods section) [sensitivity, 100%]. Moreover, to ensure the highest possible specificity (93%), clinicians need to compute the difference between the noninvasive volumetric capnographic index proposed (ie, Vae/Vt) after 48 h. The role of dead space as a risk factor for mortality is widely accepted,2 but few data are available about its power as a predictive tool. Our results reinforce the value of most of the dead space evaluations because their area under the ROC curve was clinically significant (ie, 0.7). Using dead space parameters that require no change in ventilator settings to standardize measurement, unlike the method used in the study by Nuckton et al,2 can facilitate the use of these predictors in nonspecialized environments. The advantage of the Vae/Vt ratio is magnified by its being operator independent, so the need for expertise is obviated; this variable might be useful for automated and continuous recordings when a built-in ventilator device appears on the horizon. The improved accuracy of the trend along the first 48 h of ventilation for the Vae/Vt ratio also favors its introduction in the daily routine as a monitoring tool. In our limited ARDS subpopulation, Vae/Vt reaches a very high specificity in predicting outcome. If these findings are confirmed in a larger patient population, Vae/Vt may assume clinical relevance as a useful prognostic index among the existing measures of pulmonary function. Clinical Meaning of the Capnographic Index VAE/VT Carbon dioxide kinetics depend on the following three factors: peripheral production; blood transport (cardiac output); and lung elimination (alveolar ventilation).26 If catabolism is assumed to be constant, hemodynamic or alveolar ventilation perturbations produce peculiar volumetric capnographic curves. It is generally known that all situations producing a decrease in lung flow (eg, pulmonary embolism or severe hemorrhage) affect the capnographic wave, which decreases in width. This phenomenon is due to the so-called CO2 dilution effect, which is caused by a decrease in pulmonary blood flow, with alveolar ventilation being equal. In this situation, the shape of phase III on volumetric capnograms (ie, the CO2-Vt curve) does not vary except in width, and Vco2
68

decreases.27,28 Lung pathologies globally interact with CO2 washout, thus altering both convective and diffusive processes, as well as the time, available cross-section, and intraairway concentration gradient. Bronchial obstruction makes regional alveolar ventilation inhomogeneous, thus altering the normal ventilation/perfusion ratio. This determines different readings of CO2 alveolar pressure that are asynchronously exhaled, changing the shape of the CO2-Vt curve, with a prevalent increase in the slope of phase

Figure 2. Top, A: a CO2-Vt curve obtained in baseline condition (curve 2) in one representative patient with ARDS. CO2-Vt curves 3 and 1 are mathematical simulations in which pulmonary blood flow was decreased by half and increased twofold, respectively, while expired volume remained constant. Bottom, B: the resultant Vae/Vt ratio from the CO2-Vt curves depicted in top, A. Variations in pulmonary blood flow are associated with changes in Vco2 production (ie, area under the CO2-Vt curve), but the resultant capnographic index Vae/Vt ratio is not affected, suggesting less dependence on alterations in hemodynamics (bottom, B).
Original Research

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

III.28,29 The hypothetical physiologic meaning of the Vae/Vt ratio and its potentially lower dependence on pulmonary blood flow is shown in Figures 2 and 3. The variation of pulmonary blood flow was obtained by mathematical simulation and the Vae/Vt ratio for each simulated curve is equal to the original curve (0.33) even when Vco2 changed accordingly with pulmonary blood flow. These hypothetical results could suggest less dependence of the Vae/Vt ratio on cardiac output and hemodynamics (Fig 2).

On the other hand, Vae/Vt ratio seems to change according to lung dishomogeneity, where the slope of phase III increases with respect to baseline (Fig 3). Another advantage of the mathematical analytic integration of the Vco2-Vt curve is the elimination of the ripples during phase III of the volumetric capnographic curve. As a matter of fact, the resulting curve is smooth, representing an integral-mean value of the previous CO2-Vt curve affected by noise (ie, cardiac oscillations and secretions). In summary, we found that a group of nonsurvivors from ALI and ARDS were clearly detected by Pao2/Fio2 ratio and dead space measurements, and worsening Vae/Vt was the most accurate. These results suggest that the combined analysis of indexes related to oxygenation and carbon dioxide kinetics in a patient with acute respiratory failure could provide the basis for a correct prognosis of outcome, because these two items characterize lung physiologic function. If this rising hypothesis is confirmed by means of larger sample studies, Vae/Vt may become a useful prognostic index in ALI and ARDS patients. Appendix
In this appendix we explain how to calculate Vae/Vt using a spreadsheet (eg, OpenOffice Calc [www.openoffice.org]; or Microsoft Excel; Microsoft; Redmond, WA). Suppose we have collected the expiratory volume (volume_exp) and CO2, in milliliters and in millimeters of mercury, respectively, in the first two columns (Table 4). The first step is to apply a numerical method (eg, the trapezoidal rule [ref_1]) to calculate the integral of the CO2 vs volume_exp. It is sufficient to calculate for every k 1, . . ., n 1: aux trapezoid k 1 (CO2k 1 CO2k) (volume k and, consequently, for every k quantities: Vco2i Vco2k 1 Vco2k 1, . . ., n 1 2 volume k)

1, to cumulate these

aux trapez1/760 aux trapezk 1/760

Table 4
Variables 1 2 3 4 5 6 7 8 9 10 11 12 13 A Volume_exp 54.8 60.1 65.8 71.5 77.1 83 88.9 94.8 100.8 106.9 113 B C D

Figure 3. Top, A: a CO2-Vt curve obtained in baseline condition (curve 2) in the same representative patient with ARDS. CO2-Vt curve 3 is a simulation of the increased slope of phase III with respect to CO2-Vt curve 2 (baseline condition). Pulmonary blood flow and expired volume are constant. Bottom, B: the resultant Vae/Vt ratio from CO2-Vt curves depicted in top, A. Variation in the slope of phase III on CO2-Vt curves produced concomitant changes in the Vae/Vt ratio. In this example, the Vae/Vt ratio decreased from 0.33 to 0.18, showing a strong influence on pure lung pathology.
www.chestjournal.org

CO2 aux_trapezoid Vco2 2.897 4.498 19.597 0.026 6.499 31.341 0.067 8.796 43.591 0.124 10.996 55.418 0.197 13.093 71.063 0.291 14.694 81.972 0.399 15.894 (B9 B8) (A9 A8)/2 0.517 16.495 97.167 0.645 16.895 101.840 0.779 17.199 103.789 0.916

CHEST / 133 / 1 / JANUARY, 2008

69

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

Table 5
Variables 1 2 3 4 5 6 7 8 9 10 11 12 13 A volume_exp 54.8 60.1 65.8 71.5 77.1 83 88.9 94.8 100.8 106.9 113 B CO2 2.897 4.498 6.499 8.796 10.996 13.093 14.694 15.894 16.495 16.895 17.199 C aux_trapezoid 19.597 31.341 43.591 55.418 71.063 81.972 90.235 97.167 101.840 103.789 D Vco2 0.026 0.067 0.124 0.197 0.291 0.399 D4 C5/760 0.645 0.779 0.916

References
1 Tyburski JG, Carlin AM, Harvey EH, et al. End-tidal CO2arterial CO2 differences: a useful intraoperative mortality marker in trauma surgery. J Trauma 2003; 55:892 896 2 Nuckton TJ, Alonso JA, Kallet RH, et al. Pulmonary deadspace fraction as a risk factor for death in the acute respiratory distress syndrome. N Engl J Med 2002; 346:12811286 3 Fletcher R, Jonson B, Cumming G, et al. The concept of dead space with special reference to the single breath test for carbon dioxide. Br J Anesth 1981; 53:77 88 4 Lucangelo U, Blanch L. Dead space. Intensive Care Med 2004; 30:576 579 5 Kallet RH, Alonso JA, Pittet JF, et al. Prognostic value of the pulmonary dead-space fraction during the first 6 days of acute respiratory distress syndrome. Respir Care 2004; 49:1008 1014 6 Fowler WS. Lung function studies II: the respiratory deadspace. Am J Physiol 1948; 154:405 410 7 Romero PV, Lucangelo U, Lopez Aguilar J, et al. Physiologically based indices of volumetric capnography in patients receiving mechanical ventilation. Eur Respir J 1997; 10: 1309 1315 8 Murias GE, Villagra A, Vatua S, et al. Evaluation of a noninvasive method for cardiac output measurement in critical care patients. Intensive Care Med 2002; 28:1470 1474 9 Hoffbrand B. The expiratory capnogram: a measure of ventilation-perfusion inequalities. Thorax 1966; 21:518 523 10 Verschuren F, Liistro G, Coffeng R, et al. Volumetric capnography as a screening test for pulmonary embolism in the emergency department. Chest 2004; 125:841 850 11 Verschuren F, Heinonen E, Clause D, et al. Volumetric capnography as a bedside monitoring of thrombolysis in major pulmonary embolism. Intensive Care Med 2004; 30: 2129 2132 12 Blanch L, Lucangelo U, Lopez-Aguilar J, et al. Volumetric capnography in patients with acute lung injury: effects of positive end-expiratory pressure. Eur Respir J 1999; 13:1048 1054 13 Murray JF, Matthay MA, Luce JM, et al. An expanded definition of the adult respiratory distress syndrome. Am Rev Respir Dis 1989; 138:720 723 14 Bernard GR, Artigas A, Brigham KL, et al. The AmericanEuropean Consensus Conference on ARDS: definitions, mechanism, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 1994; 149:818 824 15 Fletcher R, Jonson B. Deadspace and the single breath test for carbon dioxide during anaesthesia and artificial ventilation: effects of tidal volume and frequency of respiration. Br J Anaesth 1984; 56:109 119 16 Blanch L, Romero PV, Lucangelo U. Volumetric capnography in the mechanically ventilated patient. Minerva Anestesiol 2006; 72:577585 17 R Foundation for Statistical Computing Development Core Team. R: a language and environment for statistical computing. Available at: http://www.R-project.org. Accessed November 14, 2007 18 Shimada Y, Yoshiya I, Tanaka K, et al. Evaluation of the progress and prognosis of adult respiratory distress syndrome: simple respiratory physiologic measurement. Chest 1979; 76:180 186 19 Kiiski R, Takala J, Kari A, et al. Effect of tidal volume on gas exchange and oxygen transport in the adult respiratory distress syndrome. Am Rev Respir Dis 1992; 146:11311135 20 Beydon L, Uttman L, Rawal R, et al. Effects of positive end-expiratory pressure on dead space and its partitions in acute lung injury. Intensive Care Med 2002; 28:1239 1245
Original Research

Table 6
D Ral 19.6 50.94 94.53 6,930.07 7,025.99 7,121.72 7,217.44 7,311.10 E Vco2 0.03 0.07 0.12 9.12 9.24 9.37 9.5 9.62 F G H

Slope Line Whereline 0.95*(E193 E143)/(A193 A143) Quote 2.3 Below 4.34 2.1 Below 1.91 Below 9.1 Below 9.24 Below 9.37 Below 9.5 Above 9.63 Above

to obtain Vco2, which is, in other words, the sum of all adjacent cells (eg, D5 D4 C5/760). Now, one can consider Vco2 vs volume_exp to obtain Table 5. The last 50 points sampled are back-extrapolated by least-squares linear regression analysis. In order to simplify the procedure, without a loss of generality, it is possible to consider the slope of the straight line joining the end-tidal point (volumen, Vco2n) and the point (volumen 50, Vco2n 50) as a good approximation: Vco2n volumen Vco2n 50 volumen 50

Slope

and to reduce the slope by multiplying it by 0.95 in order to consider the DSA phenomenon. Nevertheless, for more precise analysis, one can consider the linear regression of the last 50 points (obtained directly from the algorithms available in the spreadsheet) and to multiply the slope by 0.95. Now it is sufficient to check the volumecross where the straight line crosses the Vco2 curve (Table 6) by an if . . ., then . . . statement: IF(G3 Therefore, Vae Vae/Vt volumen volumecross E3; above; below)

volumen volumecross volumen volumel

[ref_1] (see http://en.wikipedia.org/wiki/Trapezoidal_ rule).

70

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

21 Smith RPR, Fletcher R. Positive end-expiratory pressure has little effect on carbon dioxide elimination after cardiac surgery. Anesth Analg 2000; 90:85 88 22 Langley F, Even P, Duroux P, et al. Ventilatory consequences of unilateral pulmonary artery occlusion. In: Distribution des exchanges gaseaux pulmonaires. INSERM, WF D613: Paris, France, 1976; 209 212 23 Venet Ch, Guyomarch S, Pingat J, et al. Prognostic factors in acute respiratory distress syndrome: a retrospective multivariate analysis including prone positioning in management strategy. Intensive Care Med 2003; 29:14351441 24 Ferreira FL, Bota DP, Bross A, et al. Serial evaluation of the SOFA score to predict outcome in critically ill patients. JAMA 2001; 286:1754 1758 25 Monchi M, Bellenfant F, Cariou A, et al. Early predictive

26 27

28

29

factors of survival in the acute respiratory distress syndrome: a multivariate analysis. Am J Respir Crit Care Med 1998; 158:1076 1081 Bhavani-Shankar K, Moseley H, Kumar AY. Capnometry and anaesthesia. Can J Anaesth 1992; 39:617 632 Arnold JH, Stenz RI, Grenier B, et al. Single-breath CO2 analysis as a predictor of lung volume change in a model of acute lung injury. Crit Care Med 2000; 28:760 764 Stenz RI, Grenier B, Thompson JE, et al. Single breath CO2 analysis as a predictor of lung volume in a healthy animal model during controlled ventilation. Crit Care Med 1998; 26:1409 1413 Neufeld GR, Gobran S, Baumgardner JE, et al. Diffusivity, respiratory rate and tidal volume influence inert gas expirograms. Respir Physiol 1991; 84:31 47

www.chestjournal.org

CHEST / 133 / 1 / JANUARY, 2008

71

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

CHEST
Erratum: Chest 2008: 133:6271
In the January 2008 issue, in the article by Lucangelo et al, titled Prognostic Value of Different Dead Space Indices in Mechanically Ventilated Patients With Acute Lung Injury and ARDS (Chest 2008: 133:6271), the correct affiliation for Lluis Blanch, should be as follows: Lluis Blanch, MD, PhD; Critical Care Centre; CIBER Enfermedades Respiratorias: Hospital de Sabadell; Corporacio Parc Taul: Sabadell.

Errata

reports of meetings are also not precluded, but additional data or copies of tables should not amplify such reports.

Erratum: Chest 2009; 135:276 286

In the February 2009 issue, in the article by Lellouche et al, titled Humidification Performance of 48 Passive Airway Humidifiers (Chest 2009: 135:276 286), in Table 1, the manufacturer of Device Nos. 1 and 3 should be listed as Medisize.

Erratum: Chest 2009; 135:233237

In the January 2009 issue, in the article by Langdon-Neuner titled When Does Previous Disclosure Become a Prior Publication? (Chest 2009; 135:233237), the last sentence of the first full paragraph on page 234, the sentence should read, Press

Erratum: Chest 2009: 135:330 336

In the February 2009 issue, in the article by Tonelli de Oliveria et al titled Diagnosis of Obstructive Sleep Apnea and Its Outcomes with Home Portable Monitoring (Chest 2009; 135: 330 336), the name of the third author is misspelled. It should be Luiz Felipe Teer-Vasconcellos.

www.chestjournal.org

CHEST / 135 / 6 / JUNE, 2009

1699

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians

Prognostic Value of Different Dead Space Indices in Mechanically Ventilated Patients With Acute Lung Injury and ARDS * Umberto Lucangelo, Francesca Bernab, Sara Vatua, Giada Degrassi, Ana Villagr, Rafael Fernandez, Pablo V. Romero, Pilar Saura, Massimo Borelli and Lluis Blanch Chest 2008;133; 62-71; Prepublished online November 7, 2007; DOI 10.1378/chest.07-0935 This information is current as of June 28, 2011
Updated Information & Services Updated Information and services can be found at: http://chestjournal.chestpubs.org/content/133/1/62.full.html References This article cites 27 articles, 12 of which can be accessed free at: http://chestjournal.chestpubs.org/content/133/1/62.full.html#ref-list-1 Cited Bys This article has been cited by 1 HighWire-hosted articles: http://chestjournal.chestpubs.org/content/133/1/62.full.html#related-urls Permissions & Licensing Information about reproducing this article in parts (figures, tables) or in its entirety can be found online at: http://www.chestpubs.org/site/misc/reprints.xhtml Reprints Information about ordering reprints can be found online: http://www.chestpubs.org/site/misc/reprints.xhtml Citation Alerts Receive free e-mail alerts when new articles cite this article. To sign up, select the "Services" link to the right of the online article. Images in PowerPoint format Figures that appear in CHEST articles can be downloaded for teaching purposes in PowerPoint slide format. See any online figure for directions.

Downloaded from chestjournal.chestpubs.org by guest on June 28, 2011 2008 American College of Chest Physicians