eAppendix 1. List of NIVAS Trial Investigators
eAppendix 2. Supplemental Methods
eFigure. Absolute Difference of Intubation at 30 Days
eTable 1. Gas Exchange According to Study Group
eTable 2. Settings, Monitored Parameters, Tolerance and Side Effects of Noninvasive Ventilation
eTable 3. Bivariate and Multivariable Analysis of Factors Associated With the Primary Outcome
eTable 4. Reasons for Re-intubation, as Defined in the Protocol Guidelines, According to Study Group
eTable 5. Primary and Secondary Outcomes According to Study Group
eTable 6. Clinical Pulmonary Infection Score (CPIS) and Microorganisms Causing Pneumonia According to Study Group
Statistical Analysis Plan
Jaber S, Lescot T, Futier E, Paugam-Burtz C, Seguin P, Ferrandiere M, Lasocki S, Mimoz O, Hengy B, Sannini A, Pottecher J, Abback P, Riu B, Belafia F, Constantin J, Masseret E, Beaussier M, Verzilli D, De Jong A, Chanques G, Brochard L, Molinari N, for the NIVAS Study Group. Effect of Noninvasive Ventilation on Tracheal Reintubation Among Patients With Hypoxemic Respiratory Failure Following Abdominal SurgeryA Randomized Clinical Trial. JAMA. 2016;315(13):1345-1353. doi:10.1001/jama.2016.2706
It has not been established whether noninvasive ventilation (NIV) reduces the need for invasive mechanical ventilation in patients who develop hypoxemic acute respiratory failure after abdominal surgery.
To evaluate whether noninvasive ventilation improves outcomes among patients developing hypoxemic acute respiratory failure after abdominal surgery.
Design, Setting, and Participants
Multicenter, randomized, parallel-group clinical trial conducted between May 2013 and September 2014 in 20 French intensive care units among 293 patients who had undergone abdominal surgery and developed hypoxemic respiratory failure (partial oxygen pressure <60 mm Hg or oxygen saturation [Spo2] ≤90% when breathing room air or <80 mm Hg when breathing 15 L/min of oxygen, plus either  a respiratory rate above 30/min or  clinical signs suggestive of intense respiratory muscle work and/or labored breathing) if it occurred within 7 days after surgical procedure.
Patients were randomly assigned to receive standard oxygen therapy (up to 15 L/min to maintain Spo2 of 94% or higher) (n = 145) or NIV delivered via facial mask (inspiratory pressure support level, 5-15 cm H2O; positive end-expiratory pressure, 5-10 cm H2O; fraction of inspired oxygen titrated to maintain Spo2 ≥94%) (n = 148).
Main Outcomes and Measures
The primary outcome was tracheal reintubation for any cause within 7 days of randomization. Secondary outcomes were gas exchange, invasive ventilation–free days at day 30, health care–associated infections, and 90-day mortality.
Among the 293 patients (mean age, 63.4 [SD, 13.8] years; n=224 men) included in the intention-to-treat analysis, reintubation occurred in 49 of 148 (33.1%) in the NIV group and in 66 of 145 (45.5%) in the standard oxygen therapy group within+ 7 days after randomization (absolute difference, −12.4%; 95% CI, −23.5% to −1.3%; P = .03). Noninvasive ventilation was associated with significantly more invasive ventilation–free days compared with standard oxygen therapy (25.4 vs 23.2 days; absolute difference, −2.2 days; 95% CI, −0.1 to 4.6 days; P = .04), while fewer patients developed health care–associated infections (43/137 [31.4%] vs 63/128 [49.2%]; absolute difference, −17.8%; 95% CI, −30.2% to −5.4%; P = .003). At 90 days, 22 of 148 patients (14.9%) in the NIV group and 31 of 144 (21.5%) in the standard oxygen therapy group had died (absolute difference, −6.5%; 95% CI, −16.0% to 3.0%; P = .15). There were no significant differences in gas exchange.
Conclusions and Relevance
Among patients with hypoxemic respiratory failure following abdominal surgery, use of NIV compared with standard oxygen therapy reduced the risk of tracheal reintubation within 7 days. These findings support use of NIV in this setting.
clinicaltrials.gov Identifier: NCT01971892
Quiz Ref IDPostoperative acute respiratory failure is a major contributor to the overall risk of surgery, leading to an increase in morbidity and mortality.1- 3 The early postoperative period following abdominal surgery is associated with diaphragmatic dysfunction and a decrease in lung vital capacity, which may lead to atelectasis formation and hypoxemia.4Quiz Ref ID Treating postoperative acute respiratory failure usually refers to tracheal reintubation and invasive mechanical ventilation.5 Tracheal reintubation for acute respiratory failure is associated with higher mortality and increased health care utilization, with a longer duration of both intensive care unit (ICU) and hospital stay.3 Reasons for the increase in mortality include complications during the reintubation period6,7 and health care–associated infections such as pneumonia.6- 9 This suggests that postoperative outcome may be improved by strategies aimed at avoiding reintubation and invasive mechanical ventilation.10
Noninvasive ventilation (NIV) has proven effective in nonsurgical cases of acute exacerbation of chronic obstructive pulmonary disease11 and cardiogenic pulmonary edema.12 However, to date, no evidence supports the use of NIV in surgical patients with hypoxemic acute respiratory failure after abdominal surgery. Indeed, NIV is sometimes considered a relative contraindication after recent upper gastrointestinal tract surgery.13
To our knowledge, no multicenter randomized clinical trials have evaluated whether NIV could reduce the need for invasive mechanical ventilation and its effect on the incidence of health care–associated infections in patients who develop hypoxemic acute respiratory failure after abdominal surgery.14,15
We hypothesized that application of NIV may prevent reintubation and invasive mechanical ventilation and may decrease the rate of health care–associated infections.
We thus conducted a multicenter randomized clinical trial of NIV in surgical patients who developed hypoxemic acute respiratory failure after abdominal surgery, comparing NIV against standard oxygen therapy.
The trial was an investigator-initiated, multicenter, stratified, parallel-group trial with a computer-generated allocation sequence and an electronic system–based randomization. The study protocol and statistical analysis plan (Supplement 2 and Supplement 3) were approved for all centers by a central ethics committee in accordance with French law. The trial was conducted in accordance with the Declaration of Helsinki. Written informed consent from the patient or consent from a relative was obtained on study inclusion. An independent data and safety monitoring committee oversaw the study conduct and reviewed blinded safety data, with interim analyses performed after the inclusion of 100 and 200 patients. Patients were screened and underwent randomization between May 2013 and September 2014 at 20 French ICUs. Randomization was performed centrally by the minimization method with the use of a computer-generated and blinded assignment sequence. Randomization was stratified according to study site, age (<60 vs ≥60 years), site of surgery (upper vs lower abdominal), and use of postoperative epidural analgesia, as this may influence outcomes.16
Patients were eligible for participation in the study if they were older than 18 years and had undergone laparoscopic or nonlaparoscopic elective or nonelective abdominal surgery under general anesthesia. Patients were included if they met the following criteria: a diagnosis of acute respiratory failure occurring within 7 days of the surgical procedure,17,18 defined as the presence and persistence for more than 30 minutes of hypoxemia (defined by a partial oxygen pressure <60 mm Hg when breathing room air or <80 mm Hg when breathing 15 L/min of oxygen or a peripheral oxygen saturation [Spo2] ≤90% when breathing room air plus either  a respiratory rate higher than 30/min or  clinical signs suggestive of intense respiratory muscle work and/or labored breathing, such as use of accessory respiratory muscles, paradoxical motion of the abdomen, or intercostal retraction). Exclusion criteria were withholding of life-sustaining treatment,19 contraindications to noninvasive ventilation, sleep apnea syndrome, immediate tracheal intubation, requirement for an emergent surgical procedure, and previous recruitment in another trial.
We assigned causes of acute respiratory failure following extubation using adapted published definitions11,20- 23 as follows: upper airway obstruction, aspiration or excess respiratory secretions, encephalopathy, congestive heart failure, pneumonia, and atelectasis.
Quiz Ref IDPatients were randomly assigned to receive either NIV or standard oxygen therapy alone from randomization until day 30 or ICU discharge, whichever came first. Patients assigned to standard oxygen therapy received supplemental oxygen at a rate of up to 15 L/min to maintain an Spo2 of at least 94%. In the intervention group, NIV was delivered through a face mask connected to an ICU- or NIV-dedicated ventilator, using either a heated humidifier or heat and moisture exchanger to warm and humidify inspired gases.24 Noninvasive ventilation was started at an inspiratory positive airway pressure of 5 cm H2O, increasing to a maximum inspiratory pressure of 15 cm H2O, aiming to achieve an expiratory tidal volume between 6 and 8 mL/kg of predicted body weight and a respiratory rate lower than 25/min. Positive end-expiratory airway pressure (PEEP) was started at 5 cm H2O and increased as needed to a maximum of 10 cm H2O.4,10 PEEP and inspired oxygen fraction were titrated to maintain an Spo2 of at least 94%. Ventilator settings were subsequently adjusted as needed for patient comfort.4 Patients in this group were encouraged to use NIV for at least 6 hours, continuously or intermittently, during the first 24 hours after randomization. Between NIV sessions, patients received standard oxygen therapy as described above. The use of high-flow oxygen nasal cannulas was not permitted in either group. Any decision to discontinue NIV was left to the attending physician. All other aspects of patient care in both groups were conducted according to each center’s routine clinical practice.
To reduce the risk of delayed reintubation and to ensure consistency of indications for reintubation among all trial sites, predefined criteria were applied. Immediate reintubation was performed if patients had any of the following predefined major clinical events: respiratory or cardiac arrest, respiratory pauses with loss of consciousness or gasping for air, massive aspiration, persistent inability to clear respiratory secretions, heart rate of less than 50/min with loss of alertness, and severe hemodynamic instability without response to fluid and vasoactive drugs. After reintubation, all patients underwent ventilation with the same ventilation protocol, using a low-tidal-volume protective ventilatory strategy.25
The primary outcome for comparing NIV and standard oxygen therapy was any cause of reintubation within 7 days following randomization. Causes and time to reintubation were recorded. Secondary outcomes included gas exchange, health care–associated infection rate within 30 days, number of ventilator-free days (ie, days alive and without invasive mechanical ventilation) between days 1 and 30, antibiotic use duration, ICU and in-hospital length of stay, and 30- and 90-day mortality. Five of 7 secondary outcomes are reported in this article. Definitions for each health care–associated infection (pneumonia, urinary tract infection, central venous catheter–related infection, bacteremia, and surgical site infection, occurring both at least 48 hours after ICU admission and after study entry) are detailed in eAppendix 2 in Supplement 1.
We estimated that with a sample of 150 patients per group evaluated for the primary efficacy outcome, the study had at least 90% power to determine superiority of noninvasive ventilation compared with standard oxygen therapy. For the intention-to-treat analysis, the following assumptions were made: a 65% event rate in the standard oxygen therapy group10,26,27 and a 40% event rate in the noninvasive ventilation group14,28,29 (absolute risk reduction with NIV of at least 25% based on expert opinion). Further assumptions (15% of included patients) were made relating to patients randomized despite not being eligible for randomization according to inclusion/exclusion criteria and loss to follow-up for the primary end point. Two interim analyses were conducted after the first 100 and 200 patient randomizations by an independent data and safety monitoring committee for early stopping of the study for safety (mortality within 90 days) using a prespecified Haybittle-Peto efficacy boundary30 (α = .001 for the 2 interim analyses). A secondary modified intention-to-treat analysis was performed for the primary outcome including only patients who did not return to the operating room for reintervention. Unadjusted χ2 testing was used for primary outcome analysis. Multiple imputation was additionally performed if the frequency of missing data was greater than 5%. A Markov chain Monte Carlo method was used for the multiple imputation procedure; we generated m = 5 complete data sets. Multiple logistic regression analysis was used to identify relevant baseline covariates associated with the primary outcome. Variables tested in the model were selected if P<.15 and then presented as absolute difference for binary variables and mean differences for continuous variables with 95% confidence intervals. Kaplan-Meier curves for reintubation and for mortality rates were plotted for the first 30 and 90 days, respectively, after inclusion in the study and were compared by the log-rank test. We compared the primary outcome in prespecified subgroups defined by stratification criteria according to age (<60 vs ≥60 years), site of surgery (upper vs lower abdominal), and use/nonuse of epidural analgesia. A 2-tailed P<.05 was considered to indicate statistical significance. SAS software, version 9.3 (SAS Institute Inc), was used for all analyses.
From May 2013 through September 2014, 535 patients with acute respiratory failure within 7 days following abdominal surgery were eligible, of whom 300 underwent randomization, 150 to standard oxygen therapy and 150 to NIV (Figure 1). Seven patients were excluded after randomization because of withdrawn consent (n = 4) or ineligibility (n = 3). Data on the primary outcome were available for all 293 remaining patients (mean age, 63.4 [SD, 13.8] years; n = 224 men). Groups were similar with respect to inclusion, site, duration of surgery, causes of acute respiratory failure, time from surgery, time from extubation to acute respiratory failure (Table 1 and Table 2), and gas exchange (Table 2 and eTable 1 in Supplement 1). The initial settings were as follows: for the standard oxygen therapy group, mean oxygen flow was 10.4 L/min (SD, 5.1 L/min); for the NIV group, mean inspiratory pressure was 6.7 cm H2O (SD, 2.9 cm H2O), mean PEEP was 5.4 cm H2O (SD, 1.3 cm H2O), and mean fraction of inspired oxygen was 50% (SD, 16%), resulting in a mean tidal volume of 8.3 mL/kg (SD, 3.1 mL/kg) of predicted body weight (eTable 2 in Supplement 1).
Noninvasive ventilation improved the primary outcome of the 293 patients included in the intention-to-treat analysis; reintubation occurred in 49 of 148 patients (33.1%) in the NIV group and 66 of 145 (45.5%) in the standard oxygen therapy group at 7 days after randomization (absolute difference, −12.4%; 95% CI, −23.5% to −1.3%; P = .03) (Table 3, Figure 2, and the eFigure in Supplement 1). The multivariable analysis is shown in eTable 3 in Supplement 1, and NIV was significantly associated with reduced reintubation. Time from inclusion to reintubation (Table 3) and reasons for reintubation (eTable 4 in Supplement 1) did not significantly differ between the groups.
No significant difference was observed in the number of patients requiring reintubation for reoperation (16/148 [10.8%] in the NIV group and 16/145 [11%] in the standard oxygen therapy group; P = .95).
In the modified intention-to-treat analysis including all patients except those reintubated for reoperation (n = 261), NIV also improved the primary outcome. Reintubation occurred in 33 of 132 patients (25.0%) in the NIV group and 50 of 129 patients (38.8%) in the standard oxygen therapy group (P = .02).
Among patients subsequently reintubated, patients who received NIV spent less time under invasive mechanical ventilation than patients treated with standard oxygen therapy alone (Table 3). There were no significant differences in gas exchange between groups (eTable 1 in Supplement 1). At 30 days, compared with standard oxygen therapy, NIV was associated with significantly more ventilator-free days (25.4 vs 23.2 days; absolute difference, −2.2 days; 95% CI, −0.1 to 4.6 days; P = .04). Patients treated with NIV also experienced significantly fewer health care–associated infections (43/137 patients [31.4%] vs 63/128 [49.2%]; absolute difference, −17.8%; 95% CI, −30.2% to −5.4%; P = .003), especially less ICU-acquired pneumonia (20/137 patients [14.6%] vs 38/128 [29.7%]; P = .003) (Table 3 and eTable 5 in Supplement 1). Microorganisms causing pneumonia are detailed in eTable 6 in Supplement 1. At 90 days, 22 of 148 patients (14.9%) in the NIV group and 31 of 144 (21.5%) in the standard oxygen therapy group had died (absolute difference, −6.5%; 95% CI, −16.0% to 3.0%; P = .15) (Table 3 and Figure 3).
No significant difference was seen between the 2 groups in the overall incidence of serious adverse events (eTable 2 in Supplement 1). Seven patients received NIV as rescue therapy in the standard oxygen therapy group, of whom 3 (42.9%) were subsequently intubated. There were 3 episodes of cardiac arrest, 2 occurring before intubation (1 in the standard oxygen therapy group and 1 in the NIV group). Tolerance and adverse effects of NIV after the first trial in the NIV group are reported in eTable 2 in Supplement 1.
Quiz Ref IDIn this multicenter randomized clinical trial conducted among patients with hypoxemic acute respiratory failure after abdominal surgery, noninvasive ventilation delivered via face mask reduced the need for reintubation and for invasive mechanical ventilation and was associated with fewer episodes of health care–associated infection compared with standard oxygen therapy.
Hypoxemia develops in 30% to 50% of patients after abdominal surgery and can in some patients be well tolerated without symptoms.4,22,28 In others, however, hypoxemia can progress to severe acute respiratory failure. The genesis of hypoxemic acute respiratory failure postoperatively is multifactorial and partly related to atelectasis due to hypoventilation and collapsed alveoli, retained secretions, and diaphragmatic dysfunction.29,31 Atelectasis promotes bacterial growth and increases lung permeability, leading to pneumonia.32 In our study, NIV significantly decreased overall health care–associated infections and halved the rate of pneumonia. Noninvasive ventilation can reverse loss of pulmonary volume through the combined positive effects of PEEP and inspiratory pressure support, which increase lung ventilation, reopen atelectatic alveoli, and improve gas exchange.4 Reducing atelectasis by NIV could also decrease bacterial growth, thus mitigating bacterial translocation from the lung into the bloodstream.32 Avoidance of endotracheal intubation, bypassing the upper airways, is probably the major reason for the pneumonia reduction observed in patients treated by NIV.33 Moreover, NIV has been shown to reduce overall nosocomial infection rates through reduction in both the number and duration of invasive devices such as intravenous and bladder catheters.33 Finally, reducing health care–associated infections, especially pneumonia, could contribute to the trend toward lower mortality observed in the NIV group (Table 3 and Figure 3).
Quiz Ref IDComplications have been reported with NIV, such as gastric distention and pulmonary aspiration. Noninvasive ventilation may also potentially impede patients’ ability to cough and expectorate postoperatively. In the present study, no adverse events were reported in either group. We did not observe higher morbidity and mortality in the NIV group, contrary to that reported in another postextubation study because of delayed reintubation in the NIV group.20 The selection of appropriate postoperative patients who may benefit from postextubation NIV is a key factor.4,34 One randomized clinical study by Squadrone et al22 evaluated the use of noninvasive continuous positive airway pressure delivered via helmet after abdominal surgery. They studied 209 patients who developed hypoxemia immediately after extubation without necessarily having signs of respiratory distress. Nonetheless, their early use of noninvasive continuous positive airway pressure significantly decreased the incidence of reintubation from 10% to 1%. Our study presents several differences: (1) we evaluated the efficacy of NIV delivered with face mask using 2 levels of positive airway pressure and (2) NIV was used as a therapeutic application in a more severe patient cohort with hypoxemic acute respiratory failure and not prophylactically in patients with hypoxemia alone. As a result, the respective incidence of reintubation (standard oxygen therapy control group rate of 10% for the study by Squadrone et al vs 45.5% in the current study) and mortality (3% in the control group of the study by Squadrone et al vs 22% in the current study) differed between the 2 studies (Table 3).
The strengths of the present study are its large sample size, the selected population base, multicenter design, the explicit criteria for reintubation, and a complete postoperative follow-up. Baseline characteristics in the 2 groups were well matched, and the criteria for health care–associated infection diagnosis are both validated and robust. The trial excluded patients who underwent another immediate surgical procedure, and stratification was performed according to study site, age, site of surgery, and use/nonuse of postoperative epidural analgesia.
Our study has several limitations. First, the observed rate of reintubation in our study was lower than predicted in the standard oxygen therapy group.2,3,17,35 This could be due, in part, to exclusion of patients in whom acute respiratory failure was an early symptom of surgical complications needing immediate reintervention. Although invasive endotracheal mechanical ventilation has remained the cornerstone of ventilatory strategy for severe acute respiratory failure for many years,2 several studies have shown that mortality associated with pulmonary complications is largely related to the risks of postoperative reintubation and mechanical ventilation.2,3,36- 38 Second, the present study was not designed to show a significant decrease in mortality in the NIV group. The lack of significance for lower mortality observed in the NIV group (from 22% to 15% at day 90) could be due to an underpowered design. The low mortality rate in the NIV group may result from the cumulative effects of a decreased reintubation rate, a shorter duration of invasive mechanical ventilation, and a reduced rate of health care–associated infections, especially pneumonia (Table 3). Third, although we applied predefined criteria for reintubation, bias cannot be completely ruled out because blinding with NIV was not feasible. Fourth, the clinically relevant effect size used in the power analysis was 25%. Because we were not able to identify randomized clinical trials that included similar patients, this was based on expert opinion and was chosen to limit the likelihood of a type 1 error.
Recent high-impact trials have demonstrated the benefits in nonsurgical hypoxemic respiratory failure21 or equivalence of high-flow nasal cannula compared with NIV in patients after cardiothoracic surgery with moderate to severe hypoxemia.23 Future studies comparing use of high-flow oxygen cannula vs standard oxygen therapy and NIV for patients after abdominal surgery as preventive (prophylactic)39 or curative application are needed.40
Among patients with hypoxemic respiratory failure following abdominal surgery, use of NIV compared with standard oxygen therapy reduced the risk of tracheal reintubation within 7 days. These findings support use of NIV in this setting.
Group Information: The Noninvasive Ventilation for Postextubation Respiratory Failure After Abdominal Surgery (NIVAS) study investigators are listed in eAppendix 1 in Supplement 1.
Corresponding Author: Samir Jaber, MD, PhD, Département d’Anesthésie Réanimation B (DAR B), 80 Avenue Augustin Fliche, 34295 Montpellier, France (email@example.com).
Published Online: March 15, 2016. doi:10.1001/jama.2016.2706.
Author Contributions: Dr Molinari had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Jaber, Paugam-Burtz, Riu-Poulenc, Belafia, Molinari.
Acquisition, analysis, or interpretation of data: Jaber, Lescot, Futier, Paugam-Burtz, Seguin, Ferrandiere, Lasocki, Mimoz, Hengy, Sannini, Pottecher, Abback, Riu-Poulenc, Constantin, Masseret, Beaussier, Verzilli, De Jong, Chanques, Brochard, Molinari.
Drafting of the manuscript: Jaber, Futier, Riu-Poulenc, Verzilli, Molinari.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: De Jong, Molinari.
Obtained funding: Jaber.
Administrative, technical, or material support: Jaber, Seguin, Belafia, Masseret, Verzilli, Chanques, Molinari.
Study supervision: Jaber, Futier, Mimoz, Belafia, Constantin, Beaussier, Verzilli, Molinari.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Jaber reports consulting fees from Drager, Maquet, Hamilton, and Fisher & Paykel Healthcare. Dr Lescot reports grants, personal fees, and/or nonfinancial support from Baxter, Fresenius, and Merck Sharpe & Dohme. Dr Futier reports advisory board membership for Baxter, consulting and/or lecture fees from Fresenius, General Electric, and Drager, and travel reimbursement from Fisher & Paykel Healthcare. Dr Brochard reports grants from Covidien, General Electric, Fisher & Paykel Healthcare, and Maquet, and nonfinancial/other support from Philips and Maquet. No other disclosures were reported.
Funding/Support: This investigator-initiated trial was funded by Montpellier University Hospital and the APARD Foundation.
Role of the Funders/Sponsors: The study funders/sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.
Previous Presentation: Selected data in this article were presented at the International Symposium on Intensive Care and Emergency Medicine meeting in Brussels, Belgium, March 15-18, 2016.
Additional Contributions: We thank Mervyn Singer, MD, PhD, Bloomsbury Institute of Intensive Care Medicine, University College London, for critical revision of the manuscript and for English-language editing. He did not receive financial compensation for his contributions.