Article Text

Download PDFPDF

Variability of respiratory parameters and extubation readiness in ventilated neonates
  1. Jennifer Kaczmarek1,
  2. C Omar Farouk Kamlin2,
  3. Colin J Morley3,
  4. Peter G Davis2,
  5. Guilherme M Sant'Anna
  1. 1Department of Neonatology, McGill University, Montreal, Canada
  2. 2Division of Newborn Services, Royal Women's Hospital, Melbourne, Australia
  3. 3Neonatal Consultant, Cambridge, UK
  4. 4Department of Pediatrics, Mcgill University Health Center, Montreal, Canada
  1. Correspondence toGuilherme M Sant'anna, Mcgill University Health Center, Department of Pediatrics, 2300 Tupper Street, Room C-912, Montreal, Quebec H3H1P3, Canada; guilherme.santanna{at}


Background A spontaneous breathing trial (SBT) has been used to guide suitability of extubation in VLBW infants. Respiratory variability (RV) has been used to assess extubation readiness in adults but was never investigated in preterms. The combination of a SBT and RV may further improve prediction of successful extubation.

Study design Using data previously collected during the SBT, the following respiratory variables were analysed: inspiratory time (TI), expiratory time (TE), TI/total breath time, tidal volume (VT) and mean inspiratory flow (VT/TI). RV was quantified using time-domain analysis for each respiratory variable and expressed as a variability index (VI). The sensitivity, specificity, positive and negative predictive values (PPV and NPV) of the SBT, each VI and combined SBT+VI were calculated. Extubation failure was defined as need of re-intubation within 72 h.

Results A total of 44 infants were included. Successfully (n=36) and unsuccessfully (n=8) extubated infants had similar baseline characteristics and number of breaths analysed. VI for VT/TIwas significantly decreased in the failure group. The combination of the SBT and VI of either TIor VTwere the most accurate predictors of successful extubation with a sensitivity of 100% and specificity of 75% and a PPV and NPV for extubation success of 95% and 100%, respectively.

Conclusions A significant decrease in VT/TIvariability occurred in infants requiring re-intubation. The combination of a SBT failure and decreased variability in TIor VTwas highly predictive of failure. This combination is promising but requires prospective evaluation in a larger population.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

What is already known on this topic

  • Both prolonged endotracheal intubation and re-intubation of preterm infants are associated with increases in short- and long-term morbidity

  • A spontaneous breathing test can most accurately predict extubation readiness but lacks optimal specificity and negative predictive value

  • In adults, respiratory variability is decreased in patients that fail weaning from ventilation but this has never been assessed in the preterm infant population

What this study adds

  • Variability of respiratory parameters can be quantified in ventilated preterm infants undergoing a trial of ETT CPAP prior to extubation

  • The combination of a spontaneous breathing trial and measured variability in respiratory parameters is a promising tool for predicting successful extubation in preterm infants, which needs to be studied further


Many extremely preterm infants require endotracheal (ET) intubation and ventilation (MV) within the first days of life.1,2MV is associated with complications and should be discontinued as early as possible.3,,8However, 20%–40% of infants fail extubation and require re-intubation.9,10This is also associated with increased morbidity and mortality.11

Several strategies to assess extubation readiness in preterm infants have been investigated, none of which have demonstrated significant improvement over clinical decision making.12,,23Kamlin et aldeveloped a 3 min spontaneous breathing trial (SBT), where the ventilator was switched to ET CPAP for 3 min and changes in breathing, heart rate and oxygenation observed, to assess readiness for extubation in very low birth weight infants with a high predictive value.24After implementation in a small clinical practice, the extubation failure rate did not decrease.25

Respiratory variability (RV) represents the natural variability in the timing and magnitude of sequential breaths and results from rapidly occurring changes in respiratory mechanics, blood gases, chemosensitivity and inputs from airway receptors.26RV has been assessed as a potential predictor of extubation readiness in adults but never in preterm infants.27,,31Since RV can be quantified during the SBT, we designed this study to determine the predictive value of RV, alone and in combination with the SBT, with the primary outcome of successful extubation.


This study was a retrospective analysis, done in Montreal, Canada, of infant data previously collected in detail during a 3 min endotracheal continuous positive airway pressure (ETT CPAP) SBT test, done in Melbourne, Australia.24Briefly, when the clinical team decided an infant was ready for extubation, the ventilator was switched to ETT CPAP at the same pressure as the positive-end expiratory pressure setting. A SBT failure was recorded if the infant had either a bradycardia lasting longer than 15 s, defined as a drop in heart rate below 100 beats per minute, and/or a fall in oxygen saturation below 85% despite a 15% absolute increase in the fraction of inspired oxygen. Changes in FiO2 were made by the bedside nurse, research physician or research nurse in response to a fall in SpO2 below 85%. Physicians responsible for the care of the infant were not present at the time of the SBT. If this occurred, the 3 min trial was stopped and ventilation restarted. Heart rate was displayed on the HP M1205A Omnicare monitors (Hewlett-Packard, Andover, Massachusetts, USA), averaging every five beats. Pulse oximetry was monitored using Nellcor technology incorporated into the Omnicare monitor and set to average every 4 s. Methylxanthines were used before or after extubation at the clinician's discretion and infants were extubated to either nasal CPAP or nasal intermittent positive pressure ventilation using bi-nasal Hudson prongs. Extubation failure was defined as the need for re-intubation within 72 h of initial disconnection from MV.

During the SBT, the following respiratory variables were recorded on a breath-to-breath basis: inspiratory time (TI), expiratory time (TE) and tidal volume (VT) from which TI/Total breath time (TTOT) and mean inspiratory flow (VT/TI) were calculated. Data were measured using the Dräger Babylog 8000 plus (Lubeck, Germany) with its hot wire anemometer just proximal to the ET tube. Data were downloaded from the RS232 port of the ventilator for analysis.

Data and statistical analysis

The variability index (VI) of each respiratory variable was quantified using a time-domain analysis previously described by Cameron et al.26This analysis was done blinded to patient outcome of extubation failure or success and included the maximum number of consecutive breath-to-breath values recorded during the SBT of each infant, regardless of SBT pass or fail. Data was filtered for artefacts using TIand VT. Any measured TIgreater than 1 s or VTless than 1 ml/kg was excluded.

Five calculations of variability were performed for each respiratory variable: (1) Coefficient of Variation, (2) Triangular Index: construction of a histogram with bins of 0.05 s for TIand TE, 0.25 ml/kg for VT, 0.05 for TI/TTOTand 2 ml/kg/s for VT/TI, corresponding to the total number of breaths divided by the number occurring at the peak of the histogram, (3) DIFF >30%: percent of breaths differing from the mean value by more than 30%, (4) B-Babsolute: percent of breaths differing from the one immediately preceding it by more than the SD of all breaths and (5) B–B%mean: percent of breaths differing from the one immediately preceding it by more than 30% of the overall mean. For each of these calculations, higher values indicate greater variability.

Once this analysis was complete, values were stratified based on extubation failure or success. All five calculations of variability for each respiratory variable in each infant were then expressed as a percentage of the mean value of successfully extubated infants. Following this transformation, it was possible to average these newly calculated percentages into one overall VI for TI, TE, VT, TI/TTOTand VT/TI.

Continuous variables are expressed as mean±SD. The student t test was used to compare continuous variables between infants that were successfully extubated or failed. A p value <0.05 was considered significant. The ability of each VI to accurately discriminate between successful and failed extubation was assessed using receiver operating characteristic (ROC) curves. The first inflection point of each curve was identified as a cut-off to discriminate success and failure of extubation for the VI of each respiratory variable. Standard formulae were used to calculate sensitivity, specificity, positive predictive values (PPV) and negative predictive values (NPV) for the following: (1) SBT alone, (2) VI of each respiratory variable and (3) combination of the SBT and VI of each respiratory variable.


Of the 50 infants included the original study, a subset of respiratory data from 44 infants were included in the current analysis.24Data from one patient could not be recovered and data from five patients were excluded because less than 40 breaths were available for analysis. Of the 44 patients included, 36 infants were successfully extubated and eight (18%) failed.

The population characteristics and absolute values for all respiratory variables, averaged over the 3 min period of the SBT, were similar between success and failure patients (table 1). A comparable number of breaths were analysed in success and failure patients (84 (55–143) vs 70 (56–112), table 1). Differences in the VI of each respiratory parameter are presented in figure 1.

Figure 1

Variability of respiratory parameters during the spontaneous breathing trial (SBT).

Table 1

Population characteristics and mean respiratory variables during the 3 min SBT

The SBT alone correctly classified the large majority of infants. All infants that failed the SBT failed extubation (sensitivity = 100% and NPV = 100%). However, three infants that passed the SBT failed extubation representing false positives (specificity = 63% and PPV = 92%). In combination, the SBT and VI of TIor VTenhanced the predictive accuracy of the SBT alone, increasing the specificity to 75% and PPV to 95% (table 2). Overall, a combination of the SBT and VI of either TIor VTwere the most accurate predictors of successful extubation.

Table 2

Predictive value of the SBT, Variability Index (VI) of each respiratory variable and the combination of the SBT and VI of each respiratory variable for successful extubation


This study is the first to investigate variability of respiratory parameters of extremely preterm infants undergoing a trial of ETT CPAP prior to extubation from ventilation. We demonstrated that calculations of variation of respiratory variables can be performed during a spontaneous breathing trial and add useful information. A decreased VI of most respiratory parameters was observed in infants that failed extubation. Measurements of RV are simple and non invasive and could become available at the bedside with the development of automated analysis.

The application of physiological variability measurements as markers of well-being has a long tradition. Illness usually causes a loss of variability and low variability can predict a poor outcome.32,,34In neonates, early and accurate prediction of any clinical event that may increase the morbidity is important because it enables customised management of those at highest risk. Reduced variability and transient decelerations in heart rate occur at increased frequency in the preclinical phases of sepsis and systemic inflammatory response syndrome.35,,37Another risk-assessment model called the ‘Physiscore’, which incorporates physiological signals including heart rate and RV recorded during the first 3 h of life was able to accurately predict an increased risk of morbidity due to infections and cardiopulmonary complications, even when these events were not diagnosed until days or weeks later.38

RV measurements have been evaluated prospectively in mechanically ventilated adults and reduced RV of TIor VTwas associated with an increased incidence of weaning failure.28,29In another study, during a 60 min SBT, RV was decreased in patients who failed extubation from MV. From a physiological standpoint, a SBT clearly means that the individual is exposed to a respiratory load, which is not physiological and actually increases work of breathing to a level higher than being extubated. Therefore, a decreased RV could have been due to an acute deterioration of the load-capacity balance of the respiratory system during the SBT or an unfavourable load-capacity balance before the SBT.31In preterm infants, very little data on RV is available. Most of the studies have investigated the effects of sleep stages and positioning on convalescent preterm infants, at postmenstrual ages (PMA) between 32–37 weeks. A decreased RV during quiet sleep was observed but no effect of positioning was noted.39Recently, the effect of weight and age (maturation) on respiratory complexity has been investigated in preterm neonates at an earlier median PMA of 29 weeks and 11 days of life. The complexity of the VTand respiratory rate time series increased with increasing age and weight.40Data evaluating the influence of sleep stage or position upon RV measurements in extreme preterm infants, under mechanical ventilation, and during the first days of life is not available. In our study, there were no differences in maturation (GA and PMA) and weight between success and failure infants.

The challenge in the decision-making process towards extubation of an extreme preterm infant is quite complex and involves the capability to capture several factors that are not necessarily directly related to the respiratory system but can affect the outcome of success or failure. It is known, for example, that not all extubations are going to be successful because of unknown sepsis that will manifest within the next two days or a not yet diagnosed subglottic stenosis.

Our study has several limitations since the analysis was performed, retrospectively, and on a small sample size. Also, a reduced number of breaths were available for analysis in certain patients because during the SBT, when significant episodes of bradycardia or desaturation occurred, infants were switched back to ventilation before the end of the 3 min, thus limiting the number of breaths available. Furthermore, the SBT was done only in infants deemed ready for extubation by the clinical team. The utility of the SBT/RV test as a prospective tool to allow possible earlier attempt at extubation needs to be evaluated. Therefore, this analysis should be interpreted as exploratory findings which should not be applied to clinical practise outside a larger clinical trial to further assess the usefulness of this potential prediction tool.

In conclusion, infants weighing <1250 g that required re-intubation had a significantly lower variability index of mean inspiratory flow prior to extubation. A combination of SBT failure and decreased variability of TIor VTmaintained a high NPV and sensitivity and increased the PPV and specificity in predicting successful extubation. These combinations appear promising as predictive tools but require further evaluation in a larger, prospectively studied population.



  • Funding JK was supported by scholarship and GSA received start-up funds from the Research Institute of the McGill University Health Center for this research.

  • Competing interests None.

  • Ethics approval Human Research Ethics Committee at the Royal Women's Hospital, Melbourne, Australia.

  • Provenance and peer review Not commissioned; externally peer reviewed.