Study of multiparameter respiratory pattern complexity in surgical critically ill patients during weaning trials
 Vasilios E Papaioannou†^{1}Email author,
 Ioanna G Chouvarda†^{2},
 Nikos K Maglaveras^{2} and
 Ioannis A Pneumatikos^{1}
DOI: 10.1186/14726793112
© Papaioannou et al; licensee BioMed Central Ltd. 2011
Received: 26 August 2010
Accepted: 21 January 2011
Published: 21 January 2011
Abstract
Background
Separation from mechanical ventilation is a difficult task, whereas conventional predictive indices have not been proven accurate enough, so far. A few studies have explored changes of breathing pattern variability for weaning outcome prediction, with conflicting results. In this study, we tried to assess respiratory complexity during weaning trials, using different nonlinear methods derived from theory of complex systems, in a cohort of surgical critically ill patients.
Results
Thirty two patients were enrolled in the study. There were 22 who passed and 10 who failed a weaning trial. Tidal volume and mean inspiratory flow were analyzed for 10 minutes during two phases: 1. pressure support (PS) ventilation (1520 cm H_{2}O) and 2. weaning trials with PS: 5 cm H_{2}O. Sample entropy (SampEn), detrended fluctuation analysis (DFA) exponent, fractal dimension (FD) and largest lyapunov exponents (LLE) of the two respiratory parameters were computed in all patients and during the two phases of PS. Weaning failure patients exhibited significantly decreased respiratory pattern complexity, reflected in reduced sample entropy and lyapunov exponents and increased DFA exponents of respiratory flow time series, compared to weaning success subjects (p < 0.001). In addition, their changes were opposite between the two phases of the weaning trials. A new model including rapid shallow breathing index (RSBI), its product with airway occlusion pressure at 0.1 sec (P_{0.1}), SampEn and LLE predicted better weaning outcome compared with RSBI, P_{0.1} and RSBI* P_{0.1} (conventional model, R^{2} = 0.874 vs 0.643, p < 0.001). Areas under the curve were 0.916 vs 0.831, respectively (p < 0.05).
Conclusions
We suggest that complexity analysis of respiratory signals can assess inherent breathing pattern dynamics and has increased prognostic impact upon weaning outcome in surgical patients.
Background
Several indices have been studied for estimation of weaning readiness [1–4]. However, their prognostic value has demonstrated modest accuracy so far, whereas two consensus conferences on weaning did not recommend their routine application in clinical practice and proposed decisionmaking based on clinical criteria of improvement [3, 5].
Recognition that physiologic time series contain hidden information related to an extraordinary complexity that characterizes physiologic systems, has led to the investigation of new techniques from statistical physics for the study of living organisms [6]. Through those techniques different 'physiomarkers' can be estimated, which include variability and complexity indices of different biological signals. Only a few studies have explored indices derived from breathing pattern variability analysis for the estimation of weaning readiness [7–10]. However, different weaning protocols were implemented in heterogeneous groups of patients, using only one and different from each other method for the assessment of breathing dynamics, with conflicting results.
In particular, one study that included medical patients found increased variability and complexity of various ventilatory parameters in those with weaning failure. Two other studies recruited subjects who underwent cardiac and abdominal surgery and found contradictory results in terms of respiratory complexity during weaning trials. Finally, another research group studied a mixed group of patients and showed increased respiratory variability in those who managed to separate from the ventilator. In conclusion, none of the above studies used a combination of different methods for the assessment of complex dynamics of respiratory signals; something that could have increased diagnostic accuracy of such approach.
Variability analysis is not only observing over a longer period of time but much more watching from a different perspective (i.e., how much and why the values are deriving from the mean) [11]. Moreover, it can provide continuous and real time information at any point of the different weaning phases. Coefficients of variation (CVs), spectral and autocorrelation analyses of different respiratory signals are called linear methods and have been implemented for assessing breathing pattern variability and predicting weaning readiness in different groups of mechanically ventilated patients. However, their application supposes stationary time series behaviour, meaning stability of statistical properties of signals along time [11]. Furthermore, they present insensitivity to the orderliness of data and lack the ability of describing system inherent dynamics. For instance, a time series can be very variable but not very complex (oscillation). Conversely, a time series can be less variable but highly complex. For the above reasons, nonlinear methods may better describe nonstationary and nonlinear (continuous and often unpredictable crosstalk between systems' components) properties of a signal [6, 11, 12].
In the present study and contrary to those that were mentioned previously, we tried to investigate respiratory pattern dynamics using a 'toolkit' of nonlinear methods, in a homogeneous group of surgical critically ill patients during weaning from mechanical ventilation. We wanted to test the hypothesis that reduced respiratory complexity might discriminate weaning failure or success groups. In addition, we examined whether these domains of measurements and their change during weaning trials can predict weaning outcome and therefore identify a unique value of such analysis.
Methods
Setting and studying population
This study was performed in a mixed 12bed Intensive Care Unit (ICU) in the University hospital of Alexandroupolis, Greece, after approval by local Scientific and Ethics Committee. A total of 32 consecutive patients admitted to the ICU from September 2009 to February 2010 who underwent major abdominal surgery [scheduled 18 (56%), urgent 14 (44%)], with a mean Acute Physiology and Chronic Health Evaluation (APACHE) II score upon admission 18.3 (standard deviation: 6.7), were enrolled. There were 25 men and 7 women, with a mean age of 66.4 (SD: 7.9) years.
The whole studying population was divided into successful (S, n = 22) and unsuccessful (U, n = 10) groups according to the weaning outcome. Briefly, the two groups included those who were successfully or unsuccessfully extubated and remained free from invasive or noninvasive ventilation for over 48 hours, respectively. All patients enrolled in the study received mechanical ventilation (model Evita 2 Dura, Dräger, Germany) for at least 48 hours and when they met the recommended weaning criteria [3], they underwent their first spontaneous breathing trial (SBT) using low pressure support ventilation (PSV) [4]. Those with cardiac arrhythmias, neurological diseases or premedication with cardiovascular drugs were excluded from the study. Moreover, none from our patients suffered from chronic obstructive pulmonary disease (COPD), minimizing possible effects of chronic hypercapnia on the respiratory centers control.
Weaning protocol
All patients were under synchronized intermittent mechanical ventilation (SIMV) before the weaning trials, whereas none of them required administration of neuromuscular blocking agents. Patients were ventilated with pressure support (PS) mode for 30 minutes, whereas the pressure level setting was between 15 and 20 cm H_{2}O to maintain a tidal volume (V_{T}) of approximately 810 ml/Kg (stage 1, high supportH). Positive endexpiratory pressure (PEEP) was 5 cm H_{2}O, fraction of inspired oxygen concentration was 40% and pressure triggering sensitivity was set on 2 cm H_{2}O. Sedatives and opioid analgesics were discontinued in all patients, 24 hours prior to the study, whereas nonsteroid antiinflammatory agents were used occasionally as pain relievers. At the end of this stage, minute ventilation (MV), respiratory rate (RR), V_{T}, heart rate (HR) and blood gases were measured in all patients and since they met the weaning criteria, the ventilator mode was switched to 5 cm H_{2}O PS plus 5 cm H_{2}O PEEP and the other settings remained the same (stage 2, low supportL) for other 30 minutes. When patients completed the 30min SBT with low PS they were either extubated and considered as weaning success group or were reconnected to the ventilator and considered as weaning failure group (2 of them required reinstitution of mechanical ventilation within 24 hours after extubation and 8 after the performance of weaning trials) [3]. All subjects were kept in semisitting position and left undisturbed throughout the study.
Respiratory signals acquisition
Respiratory data of the whole study population
Variables  Successful group (S)  Unsuccessful group (U)  p value 

(n = 22)  (n = 10)  
Minute ventilation (MV), L  12.33 (11.2116.74)  12.95 (11.7813.2)  NS 
RR (respiratory rate) breaths/min  16 (1518)  17 (1519)  NS 
HR (heart rate) beats/min  82 (7786)  83 (8290)  NS 
Tidal volume (V_{T} ) ml  676.2 (543.4742.2)  665 (583.4834.8)  NS 
pH  7.41 (7.377.46)  7.39 (7.377.44)  NS 
PaCO_{2} mmHg  41.23 (36.245.6)  42.65 (38.346.8)  NS_{2aCOl} 
PaO_{2} mmHg  132.3 (98.4156.7)  131.84 (94.84148.2)  NS 
Male/female (Number)  17/5  8/2  <0.05 
Age (years)  65.7 (5872.3)  67.56 (56.274.45)  NS 
APACHE II score  16.4 (13.2319.34)  21.26 (18.7624.7)  <0.05 
Weight (Kg)  76.75 (64.382.54)  75.84 (63.287.2)  NS 
Time of ventilatory support before  
weaning trial (hours)  133.6 (112.5157.3)  175.6 (142.6214.2)  <0.05 
Time series analysis
Detrended fluctuation analysis (DFA)
DFA quantifies intrinsic fractallike (selfsimilar) correlation properties of dynamic systems, whose basic features is scale invariance, meaning that the same features repeat themselves on different measurement scales [6, 12]. The mean inspiratory flow and V_{T} interval data after integration were divided into windows of the same size n and subsequently, analysed in relation to a local trend in each window. This procedure was repeated for all different windows. The variability is depicted on a loglog scale as a function of different sizes of windows in a form of linear slope (or selfsimilar parameter) and characterises the fractallike correlation properties of the signal. DFA permits the detection of longrange correlations within a time series and has already been applied for assessing fractal properties in highly complex cardiovascular signals [12, 15]. Values higher than 1 and towards 1.5 tend to reflect a more periodic and predictable in its evolution time series whereas values lower than 1 and approaching 0.5 characterize a randomlike process. For DFA estimation, we used available software from physionet (http://www.physionet.org).
Sample Entropy (SampEn)
Approximate entropy (ApEn) was introduced by Pincus as a quantification of regularity in data and compares each group of consecutive measurements over a predefined time window to every other group of measurements of the same time length. ApEn is a measure of the likelihood that patterns recur over specified time intervals. Regular signals are expected to have low ApEn, while complex ones take on higher ApEn values [16, 17]. Due to ApEn's dependence on the record length an alternative statistic named sample entropy (SampEn) was introduced by Richmann and Moorman [18] with the benefit of reduced computational load.
Sample entropy that represents the negative natural logarithm of the conditional probability that two sequences similar for m points remain similar at the next point with a tolerance r, where selfmatches are not included [18], was calculated for flow and V_{T} time series. For entropy analysis, different values of parameters (m, r, N) are used for calculations. The N is the length of the time series. The parameter r that is the tolerance for accepting matches, is set between 1525% of standard deviation (SD) of the time series, after normalization (SD = 1). The parameter m (embedding dimension) is the length of sequences to be compared and its values vary between 1 and 2 for data length ranging from 100 to 5000 data points [16]. In our analysis, we computed SampEn assigning the values of 2 for m and 0.15 for r, according to criteria published elsewhere, in order to minimize the maximum of the relative errors of SampEn and of the conditional probability estimate [19], using software available from physionet (http://www.physionet.org).
Fractal dimension (FD)
The fractal dimension is another method of quantifying fractal properties of a time series. In this study, FD was estimated in Matlab by use of Higuchi method, which seems to provide more accurate results and incorporates a fast algorithm that requires only short time intervals [20]. FD is based on a measure of length L(k) of a time series, computed at different scales, by using a segment of k samples as a unit in each scale. The value of FD is calculated by a leastsquares linear bestfitting procedure as the angular coefficient of the linear regression of the loglog graph of the mean of k values Lm(k) for m = 1,2,3...k, with k being an interval time. The length Lm(k) originating from time m is calculated as the normalized sum of absolute differences between the values of point pairs that are 'k samples distant' and the length of curve of the time interval k, L(k) is calculated as the mean of the k values Lm(k). If the L(k) relates to the scale used (k) linearly in a loglog plot with slope FD, then the curve is said to show fractal dimension. High FD values reflect a high irregularity of the time series and an estimate of the scaleindependent complexity of the underlying system (over space or time).
Largest Lyapunov exponents (LLE )
whereas LLE is the largest Lyapunov exponent. In this study, we computed LLE of mean inspiratory flow and tidal volume signals, using the algorithm proposed by Rosenstein in Matlab, which seems to be useful, particularly in small data sets [21]. Values higher than 0 reflect an unstable and unpredictable system, where nearby points will diverge to any arbitrary separation. Increased LLEs reflect increased sensitivity to initial conditions and characterize unpredictable variations, whereas low values indicate regularity [21].
Finally, threedimensional distributions of different respiratory signals and phase spaces of mean inspiratory flow and MV, which describe all the possible states (trajectories) of a system, were determined. Briefly, different values of x(i) were plotted against the following ones in treedimensional axes: x, x+t and x+2t in Matlab, giving rise to the phase portrait of the signal. This graph is a complicated set of nonrepeating patterns in case of complex systems, whereas in periodic ones resemble a simple closed loop.
Statistical analysis
Data are presented as median values with 10^{th} and 90^{th} percentiles. Weaning success (S) and failure groups (U) were compared with the nonparametric MannWhitney test for continuous variables and the chisquare test for dichotomous variables, whereas different respiratory complexity indices over the 2 phases of PSV were compared with a Wilcoxon paired test. Spearman's ρ was computed for estimating relationships between all variables before and after the SBT and duration of ventilation. A stepwise multiple regression analysis was performed in order to test whether new indices add prognostic value to existing ones and finally, for building a new prognostic model. Moreover, for assessing prognostic accuracy upon weaning outcome of conventional (or model 1 that included RSBI, P_{0.1} and RSBI* P_{0.1}) versus new studied indices, a 40fold cross validation procedure was followed to assess each model's efficiency, using available software from Matlab. In each try, 75% of the dataset was chosen as training set and the remaining 25% as testing set. The regression model was trained with the training set to separate between the two classes. Bootstrapping (500 times) was applied in each try, to ensure more robust estimation of regression parameters, due to small sample size. The bootstrapping procedure involves the repetition of the experiment, each time with a slightly variant dataset produced by the replacement of a dataset sample with another existing one. A distribution is produced for each estimated variable, and the mean value is then used as a robust estimator of the variable in focuses, in this case the regression parameters. In each try, the area under the curve (AUC) and standard error were calculated, along with the best threshold for class separation, and the best sensitivityspecificity pair (in terms of receiver operating characteristic curveROC) [22]. All other tests were performed with SPSS Software Version 13.0 (SPSS Inc, Chicago III), whereas values of p < 0.05 were considered to be significant.
Results
The respiratory parameters, blood gases and demographic data did not differ between the 2 groups before the performance of the weaning trials (Table 1). In addition, mean APACHE II score upon admission and duration of ventilation before the start of SBTs were significantly higher in group U than in group S. Heart and respiratory rate at the end of SBT (phase L) in group U was significantly increased compared with group S [118 (83132) vs 98 (78113) and 35 (2843) vs 28 (2235), p < 0.001, respectively]. In both groups, HR and RR between the two phases showed significant increase. Correlations between all complexity indices with duration of ventilation were found to be insignificant.
Differences of complexity indices between patient subgroups
Parameter  Median  95% CI  Range  p value 

(SE)  (Lowerupper bounds)  (1090% percentiles)  
Sample entropy of flow  
S  1.26 (0.035)  1.161.32  0.871.46  < 0.001 
U  0.88 (0.027)  0.770.94  0.651.21  
Sample entropy of V_{T}  
S  1.27 (0.075)  0.891.47  0.861.44  < 0.001 
U  0.79 (0.029)  0.541.13  0.641.92  
DFA exponent of flow  
S  0.81 (0.034)  0.431.35  0.720.95  < 0.001 
U  1.08 (0.05)  0.761.13  0.941.25  
Largest Lyapunov exponent of flow  
S  0.76 (0.045)  0.561.22  0.481.11  < 0.001 
U  0.27 (0.014)  0.140.55  0.090.42 
Inverse changes were found in SampEn values of both studied respiratory signals, compared to DFA findings, between and within groups. Thus, decreased entropy values of the inspiratory flow time series were exhibited in weaning failure patients after transition from high to low pressure support ventilation [0.88 (0.651.21) vs 1.18 (0.851.37), p < 0.001] and increased values in those who separated successfully from the ventilator [1.26 (0.871.46) vs 0.82 (0.71.1), p < 0.001]. Accordingly, patients from group U manifested reduced entropy values at the end of the SBT compared with those from group S (0.88 vs 1.26, p < 0.001). Entropy values of tidal volume were also found to change similarly with those of respiratory flow [0.79 (0.640.92) vs 1.34 (0.881.65) for group U and 1.27 (0.861.44) vs 0.78 (0.650.92) for group S, p < 0.001, respectively, Table 2, Figure 7].
Fractal dimension of both ventilatory signals did not differ significantly between and within groups; however, its values were found to increase at the end of SBT in group S, compared with group U [1.33 (1.251.47) vs 1.31 (1.201.45) for flow and 1.40 (1.31.52) vs 1.38 (1.221.49) for V_{T}].
Finally, the same trends of change were observed in largest lyapunov exponents of both studied respiratory signals in weaning success and failure patients. LLE of inspiratory flow increased in group S from phase H to phase L [0.76 (0.481.11) vs 0.43 (0.230.68), p < 0.001] and decreased respectively in group U [0.27 (0.090.42) vs 0.65 (0.341.04), p < 0.001]. Similar changes but without statistical significance were found in tidal volume signals (Table 2, Figure 7).
Conventional weaning predictors exhibited also significant differences between groups S and U. RSBI, P_{0.1} and their product RSBI* P_{0.1} were significantly increased in subjects who failed a SBT [112.2 (85143) vs 97.8 (73116), 1.73 (1.451.98) vs 1.48 (1.321.86) and 175.25 (134.43210.3) vs 102.32 (97.84145.2) respectively, p < 0.005 for all comparisons].
Crossvalidation and ROC curve analysis of the two predictive models
Models  AUC (SE)  Threshold  AUC  Specificity  Sensitivity 

95% CI  
Model 1  0.831 (0.14)  0.402  0.730.92  0.895  0.858 
(RSBI, P_{0.1} and RSBI* P_{0.1})  
Model 2  0.916 (0.006)  0.296  0.760.98  0.967  0.886 
(RSBI, RSBI* P_{0.1}, SampEn, LLE) 
Discussion
A considerable body of data suggests that healthy individuals exhibit breathtobreath variability of breath components in a breath series [23, 24]. Breathtobreath variations have been traditionally treated as random uncorrelated white noise superimposed on the output of the respiratory controller [24, 25]. According to Tobin, the random fraction aids respiratory system to perform tasks other than gas exchange, such as speaking [26]. Only simple statistics such as mean, variance and coefficient of variation can estimate random variational fraction after averaging over many breathing cycles. Since variability in complex living systems is not only an artefact of biological noise but also an intrinsic property of various control mechanisms, different types of deterministic (nonrandom) variability have been described in the pattern of breathing [27–29].
Neurons in the brain stem govern respiratory rhythm through a network of coupled oscillators. Critical components of this network are located in a specialised region of the brain stem called the preBotzinger complex (preBotC) [30]. Del Negro and colleagues showed that progressively elevating neuronal excitability of the preBotC of neonatal rats in vitro causes periodic modulation of the inspiratory rhythm, characterised by periodic oscillations, quasiperiodicity and ultimately disorganised aperiodic activity [31]. In another experimental study with anesthetised adult cat models, Chen et al found that both focal hypoxia and chemical stimulation of preBotC can produce a marked excitation of phasic phrenic nerve discharge, characterized by reduced complexity, estimated with approximate entropy (low ApEn values) [32]. The above studies support the hypothesis that central respiratory centers are responsible for different breathing patterns with various degrees of variability and complexity in different settings and levels of stimulation. In addition, they can also adapt ventilation to metabolic needs through integration of afferent information.
Apart from chemoreceptor signalling, chest wall and pulmonary receptors may continuously affect central neural output, especially during resistive breathing [33, 34]. Brack and Tobin measured breathing variability using CVs and autocorrelation analysis, over one hour in ten patients with restrictive lung disease and in seven healthy subjects. They found that variability of inspiratory time (Ti), expiratory time (Te) and V_{T}, were significantly reduced in the patients group compared with the healthy group [26].
Several approaches have been used for the study of respiratory complexity. Donaldson in 8 adults during resting breathing found that different respiratory parameters were characterized by positive lyapunov exponents [35]. Increased respiratory approximate entropy and lyapunov exponents have been reported in patients with panic disorders [36], whereas Akey and colleagues described a reduction in respiratory ApEn upon a hypoxic insult to the brain [37]. Peng who introduced the DFA algorithm for the study of noisy and nonstationary biological signals found that breathing intervals' DFA exponents were significantly decreased in elderly compared to young adults [38]. Finally, hypercapnia has been found to decrease complexity but increase largest lyapunov exponents of different respiratory time series [39].
Implementation of different mathematical tools derived from signal processing techniques for analyzing heart rate and respiratory pattern variability has been shown to provide prognostic information in the assessment of weaning readiness [40, 7–10]. Engoren studied 10 control patients who had undergone cardiac surgery and 21 patients who required prolonged (> 7 days) ventilatory support and found increased ApEn values of tidal volume in weaning failure subjects [7]. However and according to a study from Caminal, an inverse relation between the level of pressure support and the CVs of different ventilatory parameters has been established, supporting the view that unloading of respiratory muscles is associated with increased breathing pattern variability, at least in weaning success patients [41]. ElKhatib assessed 52 patients with various disorders during a continuous positive airway pressure (CPAP) trial of 5 cm H_{2}O for 60 minutes and found that the CVs, the KolmogorovSinai entropy (sum of largest lyapunov exponents) and the correlation dimension (measure of fractallike properties) of tidal volume and airway flow were all significantly smaller in the successfully weaning group compared with the failure weaning group [8]. On the contrary, Bien and Wysocki reported decreased variability of different ventilatory parameters in weaning failure patients [9, 10]. However, they did not perform nonlinear analysis of respiratory time series for assessing breathing complexity. In another interesting study, Vallverdu and colleagues examined heart rate and respiratory pattern complexity in 78 patients during weaning trials using information flow analysis, which describes the regularity of signals by estimating the auto and mutual information functions. The authors were able to find reduced complexity and a more coupled nonlinear oscillator behavior in weaning failure subjects [42].
To our knowledge, this is the first study in medical literature that applied a 'toolkit' of nonlinear methods in respiratory signals for estimating weaning outcome, in a cohort of surgical patients. In a recently published study that included a similar group of patients and implemented the same weaning protocol, we found reduced complexity and coupling of heart and respiratory rate signals derived from bedside monitors and estimated with DFA and different entropy metrics, in subjects who failed to separate from the ventilator [43]. However, since description of ventilatory complexity requires the assessment of both predictability (entropy) and sensitivity to initial conditions (Lyapunov exponents) of continuous oscillatory signals (flow) and time series of discrete values [24], we decided to apply these methods to different respiratory signals derived from the ventilator. In addition, we tried to assess their scaleinvariant properties by computing the fractal dimension. Finally, we applied the DFA algorithm for quantifying fractal properties of respiratory signals through the estimation of longrange correlations, which contrary to fractal dimension, is more suitable for analyzing nonstationary short time series. In conclusion, we assumed that the implementation of different methods for the mathematical description of respiratory complex dynamics that was not performed in our previous report (except for sample entropy), could add significant value in such analysis, in case of significant differences between patients with different outcome of interest. We used the same weaning protocol as Bien, who examined a sample of 78 patients with systemic inflammatory response syndrome (SIRS) [9].
All indices of respiratory complexity were found to exhibit inverse changes between weaning failure and success groups. After the performance of a SBT, increased unloading of the respiratory system was associated with increased breathing complexity in subjects who managed to liberate from the ventilator (figures 16), since lower values of DFA exponent and higher values of LLE and SampEn of inspiratory flow were found (figure 7). These results indicate reduced longrange correlations, increased sensitivity to initial conditions and augmented irregularity of flow. Tidal volume signals exhibited similar changes in DFA and LLE without reaching statistical significance, whereas fractal dimension of both signals increased insignificantly in the weaning success group. In conclusion, our findings support the hypothesis that increased ventilatory randomness was associated with weaning success and proved to be more reliable in discriminating patients with different weaning outcome in relation with conventional indices (figure 8). Moreover, ventilatory complexity must be estimated with a combination of nonlinear techniques, since respiratory time series are often very noisy and highly nonstationary, compared with cardiovascular signals [38, 43].
These results parallel those from Schmidt and colleagues who reported increased LLE and KolmogorovSinai entropy values of mean inspiratory flow signals in mechanically ventilated patients, after switching the ventilator from the pressure support mode to neurally adjusted ventilatory assist mode (NAVA) [44]. According to these authors, successful spontaneous breathing trials unmask underlying variability and complexity of central neural output, since inspiratory pressure inhibits the respiratory drive. This effect is nicely reflected through the increased complexity indices of flow and is responsible for better neuromechanical coupling.
In another study, Mangin and colleagues investigated ventilatory chaotic dynamics in 17 mechanically ventilated patients during switching the ventilator from the assistcontrol mode to pressure support mode [45]. They were able to show that both fractal dimension and LLE were increased, particularly in 5 patients who were successfully extubated. Furthermore, the authors supposed that increased breathing complexity may also be attributed to higher vagal afferent feedback during unassisted breathing, as has already been shown by Sammon and Bruce [46].
These studies support our findings that transition between mechanical and spontaneous ventilation is associated with increased complexity of respiratory signals in weaning success patients, since duration of ventilation before the SBTs was similar between groups with different weaning outcome. Moreover, in a study of Burykin and Buchman investigating cardiorespiratory dynamics and synchronization during controlled and unassisted breathing in 13 surgical patients, it was demonstrated that mechanical ventilation reduces significantly both heart and respiratory rate complexity whereas spontaneous respiration is more irregular with increased uncoupling of cardiorespiratory rhythms in weaning success patients [47].
Higher variability and complexity of breathing pattern during controlled ventilatory support has been found to ameliorate oxygenation. In an oleic acid injury animal model, Mutch introduced fluctuations according to an algorithm, to mechanical respiration and found increased respiratory arrhythmia and oxygenation and decreased dead space compared with conventional ventilation (with similar MV) [48]. According to Suki, when fluctuations in the form of symmetrically distributed random noise is added to peak airway pressures, the mean does not change but isolated values can be augmented, leading to significant alveolar recruitment [49]. Moreover, low respiratory variability during both controlled and unassisted breathing could deteriorate respiratory mechanics by promoting microatelectasis [10].
Reduced respiratory complexity of flow signals in group U related to group S and between the two phases of PS within weaning failure patients might also reflect loss of effective control mechanisms that govern respiratory rhythms through a network of coupled oscillators [50]. It seems that increased respiratory load reduces complexity of central oscillator output, as it has been suggested from the different studies discussed so far. Moreover, Preas in a clinical study estimating endotoxin effect upon respiratory variability and complexity found a decrease in RR random variability in patients with restrictive lung diseases, a similar pattern of change with Tobin's study [26] and attributed dyspnoea to the endotoxin effect upon brain stem neurons [51]. In our study, the majority of weaning failure patients exhibited dyspnoea and rapid shallow breathing, whereas most of them had high APACHE II scores upon admission. Their characteristics parallel those of the Bien's study, implying the possible presence of SIRS during the weaning trials.
Furthermore, we assume that the observed significant results in different complexity properties for inspiratory flow and not for tidal volume in weaning failure patients could also be attributed to the decreased random variability of RR, which was found in Preas and Tobin's studies. According to Bruce, any respiratory activity includes variability of different types, such as random correlated and uncorrelated, periodic and nonlinear deterministic [24]. Alterations in the random fraction, although not measured in our study, could affect the reliable mathematical description of the nonrandom one, something that might limit the relevance of these results [24].
For that reason, different techniques, such as the noise titration method, have been implemented for detecting chaotic dynamics [45, 52]. However, in this study we did not adopt this technique, since our aim was not to detect the existence of ventilatory chaos but to investigate possible alterations in respiratory dynamics after a specific intervention (SBT). Moreover, we recruited a homogeneous group of patients and it is reasonable to assume that there were no intersubject variations at the level of noise.
Our study suffers several other limitations due to small sample size, which could increase false negative results and be responsible for lack of statistical significance in different nonlinear properties of ventilatory signals. In addition, implementation of sophisticated mathematical techniques remains a challenge for average physicians, whereas their standardization is urgently needed, since there is a lack of guidelines for parameter choice and bias to low values, in some cases. Concerning methodological issues, nonlinear characterization methods are extremely sensitive to noise and biased when applied to short data sets [24, 28]. Another important issue concerns nonfiltering of signals, since it has been proven that using filters can distort the characterization of nonlinearities [53]. Finally, using low sampling frequency avoids introduction of linearities within the time series, which has been found to occur during oversampling [53].
In conclusion and despite the fact that inspiratory flow and V_{T} time series preserved their fractallike properties, weaning failure patients exhibited reduced breathing pattern complexity during weaning trials, compared with subjects who were successfully separated from the ventilator. Increased respiratory load due to unresolved inflammatory response could be responsible for reducing effective neuromechanical coupling. The fact that multiplying the methods and studied parameters did not increase the chance of getting significant results for both flow and tidal volume could be associated with inherent limitations of such methods applied to short and highly noisy time series, patient characteristics or possible presence of unknown covariates. For that reason and based on findings form this and our previous study, we believe that a multimodal monitoring using in addition, both cardiovascular and electroencephalographic signals might increase diagnostic accuracy of such approach. Furthermore, the implementation of other methods such as information flow could be more suitable for studying the highly noisy and nonstationary ventilatory signals. Finally, a comparison between nonlinear properties of heart and different respiratory time series for predicting weaning outcome could shed more light into complex cardiorespiratory interactions during weaning trials.
Conclusions
Complexity analysis must incorporate many methods that capture different properties of respiratory dynamics. However, we suggest that nonlinear analysis of respiratory time during weaning trials might suffer some limitations despite increased diagnostic accuracy compared with conventional weaning indices. A multimodal monitoring of different biosignals derived from both the cardiovascular and respiratory system could increase the value of such methods. Sequential characterization of complex system's complexity could also provide a monitoring tool during weaning trials, at least in surgical patients. The perspective of adopting such techniques as descriptors of the effects of an intervention (SBT) may enhance effectiveness of early extubation. Nevertheless, these findings cannot yield information about weaning prediction in different groups of patients. More studies are needed for the estimation of their value in other subgroup categories, and for the quantitative assessment of changes during different weaning protocols.
Notes
List of abbreviations
 APACHE :

Acute Physiology and Chronic Health Evaluation
 ApEn :

approximate entropy
 AUC :

area under the curve
 DFA :

detrended fluctuation analysis
 COPD :

chronic obstructive pulmonary disease
 CPAP :

continuous positive airway pressure
 CV :

coefficient of variation
 FD :

fractal dimension
 FW :

failure weaning
 HR :

heart rate
 ICU :

intensive care unit
 LLE :

lyapunov exponents
 MV :

minute ventilation
 PEEP :

positive endexpiratory pressure
 P _{ 0.1 } :

airway occlusion pressure at 0.1 sec
 PSV :

pressure support ventilation
 ROC :

receiver operating characteristic
 RR :

respiratory rate
 RSBI :

rapid shallow breathing index
 SampEn :

sample entropy
 SBT :

spontaneous breathing trial
 SIMV :

synchronized intermittent mandatory ventilation
 SIRS :

systemic inflammatory response syndrome
 SW :

successful weaning
 V _{ T } :

tidal volume.
Declarations
Acknowledgements
This study was awarded from the European Society of Intensive Care Medicine with the advanced treatment of acute respiratory failure award for 2009, funded by Dräger Medical AG & Co, Lübeck, Germany.
Authors’ Affiliations
References
 Mead M, Guyatt G, Cook D, Griffith L, Sinuff T, Kergl C, Mancebo J, Esteban A, Epstein SK: Predicting success in weaning from mechanical ventilation. Chest. 2001, 120: 400424. 10.1378/chest.120.6_suppl.400S.View ArticleGoogle Scholar
 Epstein SK, Ciubotaru RL, Wong JB: Effect of failed extubation on the outcome of mechanical ventilation. Chest. 1997, 112: 186192. 10.1378/chest.112.1.186.PubMedView ArticleGoogle Scholar
 Boles JM, Bion J, Connors A, Herridge M, Marsh B, Melot C, Pearl R, Silverman H, Stanchina M, VieirrardBaron A, Welte T: Weaning from mechanical ventilation. Eur Respir J. 2007, 29: 10331056. 10.1183/09031936.00010206.PubMedView ArticleGoogle Scholar
 Esteban A, Alia I, Gordo F, Fernandez R, Solsona JF, Vallverdu I, Macias S, Allegue JM, Blanco J, Carriedo D, Leon M, de la Cal MA, Taboada F, Gonzales de Velasco J, Palazon E, Carrizosa F, Tomas R, Suarez J, Goldwasser RS: Extubation outcome after spontaneous breathing trials with Ttube or pressure support ventilation. The Spanish Lung Failure Collaborative Group. Am J Respir Crit Care Med. 1997, 156: 459465.PubMedView ArticleGoogle Scholar
 MacIntyre NR, Cook DJ, Ely EW, Epstein SK, Fink JB, Heffner JE, Hess D, Hubmayer RD, Scheinhorn DJ: Evidencebased guidelines for weaning and discontinuing ventilatory support: a collective task force facilitated by the American College of Chest Physicians; the American Association for Respiratory Care and the American College of Critical Care Medicine. Chest. 2001, 120: 375395. 10.1378/chest.120.6_suppl.375S.View ArticleGoogle Scholar
 Buchman TG: The community of the self. Nature. 2002, 420: 246251. 10.1038/nature01260.PubMedView ArticleGoogle Scholar
 Engoren M: Approximate entropy of respiratory rate and tidal volume during weaning from mechanical ventilation. Crit Care Med. 1998, 26: 18171823.PubMedView ArticleGoogle Scholar
 El Khatib M, Jamaleddine G, Soubra R, Muallem M: Pattern of spontaneous breathing: potential marker for weaning outcome. Intensive Care Med. 2001, 27: 5258. 10.1007/s001340000758.PubMedView ArticleGoogle Scholar
 Bien MY, Hseu SS, Yien HW, IngTiau Kuo B, Lin YT, Wang JH, Kou YR: Breathing pattern variability: a weaning predictor in postoperative patients recovering from systemic inflammatory response syndrome. Intensive Care Med. 2004, 30: 241247. 10.1007/s0013400320738.PubMedView ArticleGoogle Scholar
 Wysocki M, Cracco C, Teixeira A, Mercat A, Diehl JL, Lefort Y, Derenne JP, Similowski T: Reduced breathing variability as a predictor of unsuccessful patient separation from mechanical ventilation. Crit Care Med. 2006, 34: 20762083. 10.1097/01.CCM.0000227175.83575.E9.PubMedView ArticleGoogle Scholar
 CasasecadelaHiguera P, MartinFernandez M, AlberolaLopez C: Weaning from mechanical ventilation: a retrospective analysis leading to a multimodal perspective. IEEE Trans Biomed Eng. 2006, 53: 13301345. 10.1109/TBME.2006.873695.PubMedView ArticleGoogle Scholar
 Goldberger AL, Amaral LA, Glass L, Hausdorff JM, Ivanov PC, Mark RG, Mietus JE, Moody GB, Peng CK, Stanley HE: PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals. Circulation. 2000, 101: 215220.View ArticleGoogle Scholar
 Yang KL, Tobin MJ: A prospective study of indexes predicting the outcome of trials of weaning from mechanical ventilation. N Engl J Med. 1991, 324: 14451450. 10.1056/NEJM199105233242101.PubMedView ArticleGoogle Scholar
 Sassoon CS, Te TT, Mahutte CK, Light RW: Airway occlusion pressure: an important indicator for successful weaning in patients with chronic obstructive pulmonary disease. Am Rev Respir Dis. 1987, 135: 107113.PubMedView ArticleGoogle Scholar
 Peng CK, Havlin S, Stanley HE, Goldberger AL: Quantification of scaling exponents and crossover phenomena in nonstationary heartbeat time series. Chaos. 1995, 5: 8287. 10.1063/1.166141.PubMedView ArticleGoogle Scholar
 Pincus SM, Goldberger AL: Physiological timeseries: what does regularity quantify. Am J Physiol Heart Circ Physiol. 1994, 266: 16431645.Google Scholar
 Pincus SM, Mulligan T, Iranmanesh A, Gheorghiu S, Godschalk M, Veldhuis JD: Older males secrete luteinizing hormone (LH) and testosterone more irregularly and jointly more asynchronously, than younger males. Proc Natl Acad Sci USA. 1996, 93 (24): 1410014105. 10.1073/pnas.93.24.14100.PubMedPubMed CentralView ArticleGoogle Scholar
 Richmann JS, Moorman JR: Physiological timeseries analysis using approximate entropy and sample entropy. Am J Physiol Heart Circ Physiol. 2000, 278: 20392049.Google Scholar
 Lake DE, Richman JS, Griffin MP, Moorman JR: Sample entropy analysis of neonatal heart rate variability. Am J Physiol Regul Comp Physiol. 2002, 283: 789797.View ArticleGoogle Scholar
 Higuchi T: Approach to an irregular time series on the basis of the fractal theory. Physica D. 1988, 31: 277283. 10.1016/01672789(88)900814.View ArticleGoogle Scholar
 Rosenstein M, Collins J, De Luca C: A practical method for calculating largest Lyapunov exponents for small data sets. Physica D. 1993, 65: 117134. 10.1016/01672789(93)90009P.View ArticleGoogle Scholar
 Efron B, Tibshirani RJ: An introduction to the bootstrap. 1993, London: Chapman and HallView ArticleGoogle Scholar
 Tobin MJ, Mador MJ, Guenther SM, Lodato RF, Sackner MA: Variability of resting respiratory drive and timing in healthy subjects. J Appl Physiol. 1988, 65: 309317.PubMedGoogle Scholar
 Bruce EN: Temporal variations in the pattern of breathing. J Appl Physiol. 1996, 84: 10791087.Google Scholar
 Khoo MC, Kronauer RE, Strohl KP, Slutsky AS: Factors inducing periodic breathing in humans. J Appl Physiol. 1982, 53: 644659.PubMedGoogle Scholar
 Brack T, Jubran A, Tobin MJ: Dyspnea and decreased variability of breathing in patients with restrictive lung disease. Am J Respir Crit Care Med. 2002, 165: 12601264. 10.1164/rccm.2201018.PubMedView ArticleGoogle Scholar
 Tobin MJ, Yang KL, Jubran A, Lodato RF: Interrelationship of breath components in neighbouring breaths of normal eupneic subjects. Am J Respir Crit Care Med. 1995, 152: 19671976.PubMedView ArticleGoogle Scholar
 Marchant BP: Timefrequency analysis for biosystems engineering. Biosyst Eng. 2003, 85: 261281. 10.1016/S15375110(03)000631.View ArticleGoogle Scholar
 Perez W, Tobin MJ: Separation of factors responsible for change in breathing pattern induced by instrumentation. J Appl Physiol. 1985, 59: 15151520.PubMedGoogle Scholar
 Koshiya N, Smith JC: Neuronal pacemaker for breathing visualized in vitro. Nature. 1999, 400: 360363. 10.1038/22540.PubMedView ArticleGoogle Scholar
 Del Negro CA, Wilson CG, Butera RJ, Rigatto H, Smith JC: Periodicity, mixedmode oscillations and quasiperiodicity in a rhythmgenerating neural network. Biophysical Journal. 2002, 82: 206214. 10.1016/S00063495(02)753873.PubMedPubMed CentralView ArticleGoogle Scholar
 Chen X, Chon KH, Solomon IC: Chemical activation of preBotzinger complex in vivo reduces respiratory network complexity. Am J Physiol Regul Integr Comp Physiol. 2005, 288: 12371247.View ArticleGoogle Scholar
 Daubenspeck JA: Influence of small mechanical loads on variability of breathing pattern. J Appl Physiol. 1981, 50: 299306.PubMedGoogle Scholar
 Weinberger SE, Schwartzstein RM, Weiss JW: Hypercapnia. N Engl J Med. 1989, 321: 12231231. 10.1056/NEJM198910193211620.PubMedView ArticleGoogle Scholar
 Donaldson GC: The chaotic behavior of resting human respiration. Respir Physiol. 1992, 88: 313321. 10.1016/00345687(92)90005H.PubMedView ArticleGoogle Scholar
 Yeragani VK, Radhakrisha RK, Tancer M, Uhde T: Nonlinear measures of respiration: respiratory irregularity and increased chaos of respiration in patients with panic disorder. Neuropsychobiology. 2002, 46: 111120. 10.1159/000066388.PubMedView ArticleGoogle Scholar
 Akay M, Sekine N: Investigating the complexity of respiratory patterns during recovery from severe hypoxia. J Neural Eng. 2004, 1: 1620. 10.1088/17412560/1/1/003.PubMedView ArticleGoogle Scholar
 Peng CK, Mietus JE, Liu Y, Lee C, Hausdorff JM, Stanley HE, Goldberger AL, Lipsitx LA: Quantifying fractal dynamics of human respiration: age and gender effects. Ann Biomed Eng. 2002, 30: 683692. 10.1114/1.1481053.PubMedView ArticleGoogle Scholar
 Fiamma MN, Straus C, Thibault S, Wysocki M, Baconnier P, Similowski T: Effects of hypercapnia and hypocapnia on ventilatory variability and the chaotic dynamics of ventilatory flow in humans. Am J Physiol Regul integr Comp Physiol. 2007, 292: 19851993.View ArticleGoogle Scholar
 Shen HN, Lin LY, Chen KY, Kuo PH, Yu CJ, Wu HD, Yang PC: Changes of heart rate variability during ventilator weaning. Chest. 2003, 123: 12221228. 10.1378/chest.123.4.1222.PubMedView ArticleGoogle Scholar
 Caminal P, Domingo L, Giraldo BF, Vallverdu M, Benito S, Vazquez G, Kaplan D: Variability analysis of the respiratory volume based on nonlinear prediction methods. Med Biol Eng Comput. 2004, 42: 8691. 10.1007/BF02351015.PubMedView ArticleGoogle Scholar
 Vallverdu M, Tibaduisa O, Claria F, Hoyer D, Giraldo B, Benito S, Caminal P: Information flow to assess cardiorespiratory interactions in patients on weaning trials. Conf Proc IEEE Eng Med Biol Soc. 2006, 1: 14621465.PubMedGoogle Scholar
 Papaioannou V, Chouvarda I, Maglaveras N, Dragoumanis C, Pneumatikos I: Changes of heart and respiratory rate dynamics during weaning from mechanical ventilation: a study of physiologic complexity in surgical critically ill patients. J Crit Care.
 Schmidt M, Demoule A, Cracco C, Gharbi A, Fiamma MN, Straus C, Duguet A, Gottfried SB, Similowski T: Neurally adjusted ventilatory assist increases respiratory variability and complexity in acute respiratory failure. Anesthesiology. 2010, 112: 670681. 10.1097/ALN.0b013e3181cea375.PubMedView ArticleGoogle Scholar
 Mangin L, Fiamma MN, Straus C, Derenne JP, Zelter M, Clerici C, Similowski T: Source of human ventilatory chaos: lessons from switching controller mechanical ventilation to inspiratory pressure support in critically ill patients. Respiratory Physiology & Neurobiology. 2008, 161: 189196.View ArticleGoogle Scholar
 Sammon MP, Bruce EN: Vagal afferent activity increases dynamical dimension of respiration in rats. J Appl Physiol. 1991, 70: 17481762.PubMedGoogle Scholar
 Burykin A, Buchman TG: Cardiorespiratory dynamics during transition between mechanical and spontaneous ventilation in intensive care. Complexity. 2008, 13 (6): 4059. 10.1002/cplx.20229.View ArticleGoogle Scholar
 Mutch WA, Harms S, Ruth Graham M, Kowalski SE, Girling LG, Lefevre GR: Biologically variable or naturally noisy mechanical ventilation recruits atelectatic lung. Am J Respir Crit Care Med. 2000, 162: 319323.PubMedView ArticleGoogle Scholar
 Suki B, Alencar AM, Sujeer MK, Lutchen KR, Collins JJ, Andrade JS: Lifesupport system benefits from noise. Nature. 1998, 393: 127128. 10.1038/30130.PubMedView ArticleGoogle Scholar
 Pincus SM, Gladstone IM, Ehrenkranz RA: A regularity statistic for medical data analysis. J Clin Monit. 1991, 7: 335345. 10.1007/BF01619355.PubMedView ArticleGoogle Scholar
 Preas HL, Jubran A, Vandivier RW, Reda D, Godin PJ, Banks SM, Tobin MJ, Suffredini AF: Effect of endotoxin on ventilation and breath variability. Am J Respir Crit Care Med. 2001, 164: 620626.PubMedView ArticleGoogle Scholar
 Poon CS, Barahona M: Titration of chaos with added noise. Proc Natl Acad Sci USA. 2001, 48: 71077112. 10.1073/pnas.131173198.View ArticleGoogle Scholar
 Barahona M, Poon CS: Detection of nonlinear dynamics in short, noisy time series. Nature. 1996, 381: 215217. 10.1038/381215a0.View ArticleGoogle Scholar
Copyright
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.