Continuous Positive Airway Pressure for Treatment of Postoperative Hypoxemia a Randomized Controlled Trial

CARING FOR THE CRITICALLY ILL PATIENT

Continuous Positive Airway Pressure for Treatment of Postoperative Hypoxemia A Randomized Controlled Trial

Vincenzo Squadrone, MD Context Hypoxemia complicates the recovery of 30% to 50% of patients after ab- Massimiliano Coha, MD dominal surgery; endotracheal intubation and mechanical ventilation may be re- Elisabetta Cerutti, MD quired in 8% to 10% of cases, increasing morbidity and mortality and prolonging intensive care unit and hospital stay. Maria Maddalena Schellino, MD Objective To determine the effectiveness of continuous positive airway pressure com- Piera Biolino, MD pared with standard treatment in preventing the need for intubation and mechanical Paolo Occella, MD ventilation in patients who develop acute hypoxemia after elective major abdominal surgery. Giuseppe Belloni, MD Design and Setting Randomized, controlled, unblinded study with concealed al- Giuseppe Vilianis, MD location conducted between June 2002 and November 2003 at 15 intensive care units Gilberto Fiore, MD of the Piedmont Intensive Care Units Network in Italy. Franco Cavallo, MD Patients Consecutive patients who developed severe hypoxemia after major elec- V. Marco Ranieri, MD tive abdominal surgery. The trial was stopped for efficacy after 209 patients had been enrolled. for the Piedmont Intensive Care Units Interventions Patients were randomly assigned to receive oxygen (n=104) or oxy- Network (PICUN) gen plus continuous positive airway pressure (n=105). Main Outcome Measures The primary end point was incidence of endotracheal ECOVERY FROM ABDOMINAL intubation; secondary end points were intensive care unit and hospital lengths of stay, surgery is usually fast and un- incidence of pneumonia, infection and sepsis, and hospital mortality. complicated, but postoperative hypoxemia complicates Results Patients who received oxygen plus continuous positive airway pressure had a lower intubation rate (1% vs 10%; P=.005; relative risk [RR], 0.099; 95% confidence betweenR 30% and 50% of cases, even interval [CI], 0.01-0.76) and had a lower occurrence rate of pneumonia (2% vs 10%, among those undergoing uneventful RR, 0.19; 95% CI, 0.04-0.88; P=.02), infection (3% vs 10%, RR, 0.27; 95% CI, 0.07- procedures.1 Although oxygen admin- 0.94; P=.03), and sepsis (2% vs 9%; RR, 0.22; 95% CI, 0.04-0.99; P=.03) than did pa- istration and incentive spirometry are tients treated with oxygen alone. Patients who received oxygen plus continuous positive effective in treating the vast majority of airway pressure also spent fewer mean (SD) days in the intensive care unit (1.4 [1.6] vs cases of postoperative hypoxemia,2 res- 2.6 [4.2], P=.09) than patients treated with oxygen alone. The treatments did not affect piratory failure may occur early in the the mean (SD) days that patients spent in the hospital (15 [13] vs 17 [15], respectively; postoperative course,3 requiring endo- P=.10). None of those treated with oxygen plus continuous positive airway pressure died in the hospital while 3 deaths occurred among those treated with oxygen alone (P=.12). tracheal intubation and mechanical ventilation in 8% to 10% of patients, thus Conclusion Continuous positive airway pressure may decrease the incidence of en- increasing morbidity and mortality and dotracheal intubation and other severe complications in patients who develop hypox- prolonging intensive care unit (ICU) emia after elective major abdominal surgery. and hospital stay.1-4 Loss of function- JAMA. 2005;293:589-595 www.jama.com ing alveolar units has been recognized Author Affiliations and PICUN Members are listed at Caring for the Critically Ill Patient Section Editor: as the underlying mechanism respon- the end of this article. Deborah J. Cook, MD, Consulting Editor, JAMA. sible for postoperative hypoxemia.5-7 Corresponding Author: V. Marco Ranieri, MD, Uni- Advisory Board: David Bihari, MD; Christian versità di Torino, Dipartimento di Anestesia, Azienda Brun-Buisson, MD; Timothy Evans, MD; John Pulmonary atelectasis after abdominal Ospedaliera S. Giovanni Battista-Molinette, Corso Heffner, MD; Norman Paradis, MD; Adrienne surgery is, in fact, common. It may ex- Dogliotti 14, 10126 Torino ([email protected]). Randolph, MD.

Downloaded From: http://jama.jamanetwork.com/ on 02/11/2013 CPAP FOR POSTOPERATIVE HYPOXEMIA

ceed 25% of the total lung volume and METHODS tional class of II, III, or IV; had valvu- is seen several days after surgery.5-7 Patients lar heart disease, history of dilated car- Continuous positive airway pres- From June 2002 to November 2003, pa- diomyopathy, implanted cardiac pace sure (CPAP) is a breathing mode by tients were recruited from the centers maker, unstable angina, or myocar- which the patient spontaneously of the Piedmont Intensive Care Units dial infarction and cardiac surgery breathes through a pressurized circuit Network (PICUN; members and insti- within the previous 3 months; had a his- against a threshold resistor that main- tutions are listed in the acknowledg- tory of chronic obstructive pulmo- tains a preset positive airway pressure ment). Ethics committees approved the nary disease, asthma, or sleep disor- during both inspiration and expira- protocol and written informed con- ders; had preoperative infection, sepsis, tion. Although several studies have sent was obtained from the patients. or both13; had a body mass index higher demonstrated the efficacy of CPAP to Patients scheduled for elective ab- than 40; had a presence of tracheos- reduce atelectasis and improve oxygen- dominal surgery and general anesthe- tomy, facial, neck, or chest wall abnor- ation in patients after abdominal sur- sia were eligible to participate in the malities; required an emergency pro- gery,8-11 no clinical trials have con- study if they met the following crite- cedure (operation that must be firmed that the improvement of gas ria: abdominal surgery requiring lapa- performed as soon as possible and no exchange with CPAP actually results in rotomy and time of viscera exposure longer than 12 hours after admis- a reduced need for intubation and me- longer than 90 minutes. At the end of sion); or had undergone abdominal aor- chanical ventilation in patients who de- the surgical procedure, patients were tic aneurysm surgery, chemotherapy, or velop hypoxemia after abdominal sur- extubated and underwent a 1-hour immunosuppressive therapy within the gery.12 We conducted a multicenter, screening test breathing oxygen previous 3 months. Patients were also prospective, randomized clinical trial through a Venturi mask at an inspira- excluded if before randomization they to compare the efficacy of CPAP with tory fraction of 0.3. Patients were in- had arterial pH lower than 7.30 with an standard oxygen therapy in the treat- cluded in the study if they developed arterial carbon dioxide tension higher ment of postoperative hypoxemia. We an arterial oxygen tension to inspira- than 50 mm Hg; arterial oxygen satu- also set out to examine the hypothesis tory oxygen fraction ratio (PaO2/FiO2) ration lower than 80% with the maxi- that early application of CPAP may pre- of 300 or less (FIGURE 1). Patients were mal fraction of inspiratory oxygen; clini- vent intubation and mechanical venti- excluded if before surgery they were cal signs of acute myocardial infarction; lation in patients who develop acute hy- older than 80 or younger than 18 years; systolic arterial pressure lower than 90 poxemia after major abdominal surgery. had a New York Heart Association func- mm Hg under optimal fluid therapy; presence of criteria for acute respiratory distress syndrome14; hemoglobin Figure 1. Patient Flow Chart concentration lower than 7 g/dL, se- rum albumin level lower than 3 g/dL; 1322 Patients Enrolled creatinine level higher than 3.5 mg/dL (309 µmol/L); or a Glasgow Coma Scale 12 Refused to Participate 1080 Excluded (Did Not Meet lower than 12. Postoperative Eligibility Criteria) Study Design 230 Met Postoperative Eligibility Criteria Concealed randomization was con-

21 Excluded ducted centrally through a dedicated 11 Lack of Intensive Care Web site using a computer-generated Unit Beds 6 Arterial Oxygen Saturation block randomization schedule. <80% With Maximal Fraction of Inspired Oxygen Patients were randomly assigned to 3 Arterial pH <7.30 With be treated for 6 hours with oxygen PaCO2 >50 mm Hg 1 Systolic Blood Pressure through a Venturi mask at an FiO2 of <90 mm Hg 0.5 (control patients) or with oxygen at an FiO2 of 0.5 plus a CPAP of 7.5 cm 209 Randomized 8,15 H2O. At the end of the 6-hour period, patients underwent a 1-hour 104 Assigned to Receive Oxygen Therapy by 105 Assigned to Receive Oxygen Therapy With Venturi Mask (Control) Continuous Positive Airway Pressure screening test breathing oxygen through a Venturi mask at an FiO2 of 2 Developed Treatment Intolerance and 4 Developed Treatment Intolerance and 8,15 Discontinued Study Treatment Discontinued Study Treatment 0.3. Patients returned to the assigned treatment if the PaO2/FiO2 ratio 104 Included in Analysis 105 Included in Analysis was 300 or less; treatment was interrupted if the PaO2/FiO2 ratio was higher

Downloaded From: http://jama.jamanetwork.com/ on 02/11/2013 CPAP FOR POSTOPERATIVE HYPOXEMIA

than 300. Nasal oxygen (8-10 L/min) tubate a patient was made by the at- analyses were conducted by an inde- was given if the treatment was not tending clinician, who recorded the rea- pendent data and safety monitoring tolerated. sons for intubation from the list of board after enrollment of each succes- In all centers, CPAP was generated possible reasons. sive group of approximately 200 pa- using a flow generator with an adjust- Secondary outcome variables were tients. The interim analysis was based able inspiratory oxygen fraction set to ICU and hospital length of stay; inci- on the comparison of intubation rate in deliver a flow of up to 140 L/min (Whis- dence of pneumonia, infection, and the 2 treatment groups with the use of perflow, Caradyne, Ireland) and a sepsis within the first month after sur- a normal approximation for a 2-sided spring-loaded expiratory pressure valve gery2,16; and hospital mortality. Pneu- ␣ level of .05. Stopping boundaries were (Vital Signs Inc, Totoma, NJ) and monia, infection, and sepsis were iden- designed to allow termination of the applied using a latex-free polyvinyl tified using standard definitions.13 study if the use of CPAP was found to chloride transparent helmet (CaStar, To attenuate effects on outcome vari- be either efficacious (PϽ.016)18 or in- Starmed, Italy).16 All centers mea- ables, anesthesia, postoperative pain effective (PϾ.022).19 sured the inspiratory oxygen fraction control, and respiratory physio- All analyses were conducted on an using an oxygen analyzer (Oxicheck, therapy were constrained by proto- intention-to-treat basis. Because val- Caradyne, Ireland) through the Ven- cols. Anesthesia was induced and main- ues were normally distributed, results turi mask or the helmet. tained with propofol, atracurium, are reported as mean (SD). Continu- If intolerance (defined as patient in- remifentanil, and sevoflurane. Lac- ous variables were compared with the ability to tolerate the helmet or the Ven- tated Ringer solution was given use of the unpaired t test. Categorical turi mask because of discomfort, claus- throughout surgery (8 mL/kg per hour) variables were compared with the use trophobia, or pain) was observed, both and to compensate for blood loss (3 mL of Fisher exact test or the ␹2 test, when treatments were interrupted and 8 to for every milliliter of blood loss); allo- appropriate. The Kaplan-Meier curve 10 L of oxygen was administered genic red blood cells were given for for intubation rate was plotted for the through nasal probe. blood losses exceeding 20% of circu- first 7 days after surgery2 and was com- lating blood volume. Tidal volume, pared by the log-rank test. A probabil- Outcome Variables minute ventilation, positive end- ity of .05 on 2-sided testing was re- The primary outcome variable was en- expiratory pressure, and inspiratory garded as being significant. Analyses dotracheal intubation within the first oxygen fraction were 8 to 10 mL/kg, were performed using SAS statistical 7 days after surgery since intubation, 100 mL/kg, 4 to 5 cm of H2O, and no software version 8.2 (SAS Institute, for respiratory failure is commonly seen more than 0.5, respectively; patients Cary, NC). during this period.2,17 Intubation was were extubated when awake and with performed when patients presented an oxygen saturation higher than 90% RESULTS with 1 of the following: (1) severe hy- at air temperature. Postoperative anal- The independent data and safety moni- poxemia, defined as arterial oxygen gesia was managed with intramuscu- toring board stopped the trial at the first saturation lower than 80% despite the lar opioids given to obtain values lower interim analysis because the observed use of the maximal FiO2; (2) respira- than 30 mm on a visual analog scale intubation rate in the group of pa- tory acidosis, defined as arterial pH level (from 0 mm corresponding to “no pain” tients treated with oxygen plus CPAP lower than 7.30 with a carbon dioxide to 100 mm corresponding to “the worst was lower than in the group of pa- tension higher than 50 mm Hg; (3) pain ever felt”). Chest physiotherapy tients treated with oxygen alone signs of patient distress with acces- (diaphragmatic breathing, pursed-lip (P=.005). sory muscle recruitment and paradoxi- breathing, and forced expiration) was cal abdominal or thoracic motion; (4) given once preoperatively, once on the Study Population hemodynamic instability defined as an day of operation, and twice daily for 3 Of the 1332 patients enrolled, 11 pa- 80- to 90-mm Hg increase or a 30- to days after the operation. tients refused to participate, 230 pa- 40-mm Hg decrease in systolic blood tients matched the study criteria (17%) pressure relative to the baseline value Power and Statistical Analysis and 209 underwent randomization. The or need for inotropic drugs for at least Based on previous data3 and on a ret- remaining 21 patients were not ran- 2 hours to maintain systolic blood pres- rospective evaluation of medical charts, domized for the following reasons: no sure higher than 85 mm Hg or electro- the predicted intubation rate of pa- bed available in the ICU (n=11); arte- cardiogram evidence of ischemia or sig- tients fulfilling study criteria was ap- rial oxygen saturation lower than 80% nificant ventricular arrhythmias; (5) proximately 10% to 15%. The trial was with the maximal fraction of inspira- need for sedation for major agitation; designed to enroll 600 patients to dem- tory oxygen (n=6); arterial pH lower (6) decreased alertness defined as a onstrate at least a 50% reduction in the than 7.30 with an arterial carbon diox- Glasgow Coma Score lower than 9; or intubation rate with a 5% risk of type I ide tension higher than 50 mm Hg (7) cardiac arrest.15 The decision to in- error and a power of 80%. Interim (n=3); systolic arterial pressure lower

Downloaded From: http://jama.jamanetwork.com/ on 02/11/2013 CPAP FOR POSTOPERATIVE HYPOXEMIA

than 90 mm Hg under optimal fluid Table 1. Baseline Characteristics of the Patients at Study Inclusion Before Randomization therapy (n=1; Figure 1). Extubation Control CPAP (n = 104) (n = 105) and the screening test were performed Sex, No. (%) in the recovery room immediately af- Men 64 (62) 71 (68) ter surgery in 86% of patients while in Women 40 (38) 34 (32) the remaining 14% of patients they were Age, mean (SD), y 65 (10) 66 (9) performed in the ICU in a mean (SD) Body mass index, mean (SD)* 26.3 (4.5) 26.5 (4.7) of 9 (4) hours after surgery. Current smoker, No. (%)† 21 (20) 19 (18) Baseline characteristics, site, and du- SAPS II, mean (SD)‡ 28 (8) 27 (7) ration of surgery did not differ be- Type of surgery, No. (%) tween groups (TABLE 1). The visual Colectomy 38 (36) 39 (36) analog scale values for postoperative an- Gastrectomy 7 (6) 6 (6) algesia were a mean (SD) of 20 (5) mm Pancreatico-duodenectomy 18 (17) 19 (18) in the control group and 22 (4) mm in Retroperitoneal mass 4 (3) 3 (4) the CPAP group. Liver resection 24 (22) 22 (21) The mean (SD) PaO2/FiO2 ratio at the Liver transplant 13 (12) 16 (15) end of treatment was higher in pa- Pathology, No. (%) tients treated with CPAP plus oxygen Cancer 64 (62) 67 (64) than in patients treated with oxygen Noncancer 40 (38) 38 (36) alone (432 [45] and 341 [32], respec- Comorbidities, No. (%) Ͻ Diabetes 11 (11) 16 (15) tively, P .001). The mean (SD) time Hypertension 42 (40) 37 (35) of treatment required to obtain the oxy- Postoperative gases, mean (SD) genation goal was 19 (22) hours in the PaO2/FiO2 255 (31) 247 (33) CPAP group and 28 (27) hours in the Arterial, pH 7.39 (0.05) 7.38 (0.04) control group (P=.006). Two of the 104

PaCO2, mm Hg 39 (5) 39 (7) (2%) patients in the control group and Mean arterial blood pressure, mean (SD), mm Hg 86 (10) 85 (11) 4 of the 105 (4%) patients in the CPAP Time of surgical procedure, mean (SD), h§ 226 (95) 227 (91) group developed intolerance. All intol-

Abbreviations: CPAP, continuous positive airway pressure; PaO2/FiO2, arterial oxygen to inspiratory oxygen fraction ra- erances were observed during the first tio; PaCO2, arterial carbon dioxide. *Body mass index is the weight in kilograms divided by the square of the height in meters. 6 hours of treatment. None of the pa- †Patients who had smoked within 8 weeks of surgery were defined as current smokers. tients who developed intolerance to ‡Simplified Acute Physiology Score (SAPS II) (range, 0-63) is an index of the severity of illness; higher values indicate greater severity. treatment required intubation. §Time of surgical procedure is the time encompassed between the skin incision and the first suture placed to close skin incision. Primary End Points FIGURE 2 shows the Kaplan-Meier estimates of intubation rate by treat- Figure 2. Kaplan-Meier Estimates of Intubation Rate ment group. The cumulative probabil- 10 ity of remaining unintubated was higher in patients treated with CPAP (P=.005; 8 log-rank test): 10 patients (10%) in the control group, and 1 patient (1%) in the 6 CPAP group required intubation Control (P=.005). The reasons for intubation 4 were severe hypoxemia (8 patients), he- Intubation, % modynamic instability (1 patient), and 2 cardiac arrest (1 patient) in control pa- CPAP tients and severe hypoxemia in the CPAP patient. Relative risk (RR) for in- 0 20 40 60 80 100 120 140 160 tubation was 0.099 (95% confidence in- Time, h No. at Risk terval [CI], 0.01-0.76). Control 104 102 99 99 97 96 96 95 95 CPAP 105 104 104 104 104 104 104 104 104 Secondary End Points Estimates of intubation rates are according to whether or not patients received oxygen alone (control) or oxy- The mean (SD) ICU length of stay was gen plus continuous positive airway pressure (CPAP). The cumulative probability of remaining without intu- shorter in the oxygen plus CPAP group bation was higher in patients treated with CPAP (P=.005; log-rank test). than in the oxygen alone group (2.6

Downloaded From: http://jama.jamanetwork.com/ on 02/11/2013 CPAP FOR POSTOPERATIVE HYPOXEMIA

[4.2] days vs 1.4 [1.6] days, P=.09). Table 2. Secondary Outcomes Four patients in the oxygen plus CPAP Control CPAP Difference P group remained in the ICU for at least (n = 104) (n = 105) of Means (95% CI) Value* 4 days after surgery vs 13 patients in ICU length of stay, mean, d 2.6 1.4 –1.2 (–2.0 to –0.3) .09 the oxygen alone group (RR, 0.30; 95% Median (95% CI), d 1 (1-11) 1 (1-4) CI, 0.10 - 0.90; P=.02). All cases of Hospital length of stay, 17 (15) 15 (13) –2 (–6 to 2) .10 pneumonia were observed within the mean (SD), d first week after surgery.Two percent of Median (95% CI) 12 (7-47) 11 (6-35) patients treated with oxygen plus CPAP Relative Risk (95% CI) had pneumonia with in the first week Pneumonia, No. (%)† 10 (10) 2 (2) 0.19 (0.04 to 0.88) .02 of surgery vs 10 in the oxygen alone Infection, No. (%)‡ 11 (10) 3 (3) 0.27 (0.07 to 0.94) .03 group (RR, 0.19; 95% CI, 0.04-0.88; Sepsis, No. (%)§ 9 (9) 2 (2) 0.22 (0.04 to 0.99) .03 P=.02). Cases of infection (3% vs 10%; Anastomotic leakage, No. 6 1 RR, 0.27; 95% CI, 0.07-0.94; P=.03) Pneumonia, No. 3 1 and sepsis (2% vs 9%; RR, 0.22; 95% Deaths, No. (%) 3 (3) 0 (0) .12 Abbreviations: CI, confidence interval; CPAP, continuous positive end-expiratory pressure; ICU, intensive care unit. CI, 0.04-0.99; P=.03) were less fre- *All P values are 2-tailed. Comparisons between control and CPAP by Fisher exact test for categorical variables and quent in patients treated with oxygen 2-tailed t test for continuous variables. †Pneumonia was defined as the presence of an infiltrate on chest radiograph, plus at least 1 of the following: purulent plus CPAP than in patients treated with endotracheal aspirate; known pathogens on a Gram stain, or cultured from sputum or blood; temperature higher than 38.5°C or lower than 36°C, white blood cell count higher than 12 ϫ 103/µL or less than 3.5 ϫ 103/µL, or 20% oxygen alone, respectively; all oc- immature forms. curred after the first week after sur- ‡Infection was defined as the presence of a pathogenic microorganism, clinical infection, or both, necessitating anti- biotic administration. gery. All cases of infection were re- §Sepsis was defined as infection plus at least 3 of the following: core temperature higher than 38°C or lower than 36°C, heart rate higher than 90/min, a respiratory rate higher than 20/min, or an arterial carbon dioxide lower than 32 mm Hg lated to infection of the surgical wound. or the use of mechanical ventilation; white cell count higher than 12 ϫ 103/µL or lower than 4 ϫ 103/µL, or a differ- Sixty-seven percent of the sepsis were ential count showing more than 10% immature neutrophils. related to anastomotic leakage while the remaining 33% were related to pneumonia. The mean (SD) hospital length intubation rate in surgical patients can velop hypoxemia after abdominal sur- of stay did not differ between groups vary between 5% and 15% based on pre- gery.12,25,26 A recent study of 123 patients (15 [13] days in patients treated with vious physical status and the complex- showed that, despite early physiologi- CPAP vs 17 [15] days in patients treated ity of the surgical procedure.3,21,22 cal benefits, treatment with CPAP did with oxygen, P =.10). All patients The most evoked mechanism of hy- not reduce the need for intubation in treated with CPAP left the hospital alive poxemia after abdominal surgery is the patients with acute hypoxemic respi- while 3 patients died in the group impairment of the pulmonary ventila- ratory failure.15 Several factors could ex- treated with oxygen alone (P =.12; tion-perfusion ratio due to atelectasis plain the results of our trial. TABLE 2). caused by recumbent position, high First, we excluded patients who, be- oxygen concentration, temporary dia- fore randomization, presented with hy- COMMENT phragmatic dysfunction, impairment of percapnia and respiratory acidosis, se- Our study demonstrates that early treat- pulmonary secretion clearance, and vere hypoxemia, presence of criteria for ment with CPAP reduces the need for pain.5-7 Several studies have shown that, acute respiratory distress syndrome, intubation, the ICU length of stay and in patients with postoperative hypox- clinical signs of acute myocardial in- the incidence of pneumonia, infec- emic respiratory failure, CPAP im- farction, hypotension, and conscious- tion, and sepsis in patients who de- proves gas exchange, minimizes atel- ness impairment since controversies ex- velop acute hypoxemia after elective ectasis formation, and increases ist on the use of CPAP under these major abdominal surgery. functional residual capacity.8-11 How- conditions12,26 and included patients Of the 1332 patients enrolled, 209 ever, although randomized clinical trials who likely developed hypoxemia only developed hypoxemia an hour after ex- involving patients with cardiogenic pul- because of postoperative atelectasis.5-7 tubation and underwent randomiza- monary edema23 and with congestive Second, among factors that may limit tion (16%); the incidence of severe hy- heart failure and sleep-related breath- the application of noninvasive CPAP, poxemia in our study is consistent with ing problems24 showed that improve- the use of a face or nasal mask appears previous reports: Moller and cowork- ment of physiological parameters with to be particularly relevant. Inability to ers20 found that hypoxemia occurred in CPAP corresponded to a better clini- fit the mask, leaks, and patient discom- 13% of 200 patients who had under- cal outcome, no clinical trials have con- fort may limit continuous and long- gone elective surgery. The intubation firmed that the improvement of gas ex- term application of noninvasive CPAP rate within the first week after surgery change with CPAP actually results in and account for a large proportion of in the control group in our study was a reduced need for intubation and me- the failures.15,27 A transparent latex- 10%. Previous reports indicate that the chanical ventilation in patients who de- free polyvinyl chloride helmet origi-

Downloaded From: http://jama.jamanetwork.com/ on 02/11/2013 CPAP FOR POSTOPERATIVE HYPOXEMIA

nally designed to deliver oxygen dur- surgical reintervention predominantly outcome variables such as anesthesia, ing hyperbaric treatments has been occur a week after surgery.2,17 In our postoperative pain control, and respi- proposed recently as an alternative in- study, all cases of pneumonia occurred ratory physiotherapy. Moreover, lack of terface to deliver CPAP.16,28-30 The hel- within the first week from surgery while blindness unlikely influenced the re- met contains the head and the neck of all cases of infection and sepsis oc- duced ICU length of stay observed in pa- the patient and is secured by 2 armpit curred after the first week; all were more tients treated with CPAP since short- braces at an anterior and a posterior frequent in patients treated with oxy- age of ICU beds is such that ICU hook on a plastic ring joining the hel- gen than in patients treated with CPAP. attending physicians (not involved in the met to a soft collar that adheres to the These findings may be explained by re- study) always tried to discharge the pa- neck and ensures a sealed connection cent data showing that atelectasis pro- tients as soon as was safely possible. once the helmet is inflated. The flow motes bacterial growth in the lung and We originally intended to enroll 600 generator and the spring-loaded expi- increases lung permeability; reducing at- patients and conduct interim analyses af- ratory pressure valve are connected to electasis by positive end-expiratory pres- ter enrollment of each successive group the inlet and outlet of the helmet; a sure decreases bacterial growth in the of approximately 200 patients. Al- 2-way security valve allows the pa- lung, mitigates bacterial translocation though commonly used in randomized tient to breathe from outside should the from the lung into the bloodstream, and clinical trials,37,38 repeated interim analy- helmet deflate due to loss in circuit pres- normalizes the permeability of the epi- ses may lead to the experimental treat- sure or disconnection.16,28-30 Recent thelial-endothelial barrier.33,34 ment being incorrectly declared as dif- studies demonstrated that although im- In a recent study, Plant et al35 showed ferent from control.37,38 We used classic provement in oxygenation and func- that use of noninvasive ventilation on methods developed by Gordon and tional residual capacity was similar,29 general wards was feasible and clini- DeMets18 and Pocock19 for performing intolerance to ventilatory treatment, in- cally effective at reducing the need for interim analyses, using “stopping rules” cidence of skin necrosis, gastric disten- intubation and the mortality associated that were less stringent with respect to sion, and eye irritation were less com- with acute exacerbations of chronic ob- ineffectiveness than with respect to ef- mon with the helmet than with the face structive pulmonary disease. A recent ficacy with a P value required to stop the mask.16 The use of the helmet to de- consensus conference recommends that study for efficacy at the first interim liver CPAP could therefore explain the treatment of severe hypoxemia with non- analysis (PϽ.016) more stringent than reduced incidence of intolerance seen invasive ventilation should be per- the conventional PϽ.05. An indepen- in our study (4%) compared with a pre- formed in an ICU or within a system of dent data and safety monitoring board, vious trial (14%) that administered care capable of providing high levels of composed of 3 biostatisticians, a bio- CPAP through a full face mask.15 How- monitoring, with immediate access to ethicist, and a clinician who was knowl- ever, the helmet used to deliver CPAP staff skilled in invasive airway manage- edgeable about the study question, in our study is presently available in ment.12 Antonelli et al36 recently ob- stopped the trial after the first interim only a few European countries; fur- served that the frequency of failure of analysis because CPAP was found to be ther studies are therefore required to noninvasive ventilation requiring imme- efficacious for the primary outcome vari- evaluate whether use of more widely diate intubation was higher in patients able (P=.005). Despite this being, to the available interfaces for CPAP adminis- with severe hypoxemia associated with best of our knowledge, the largest mul- tration, such as full face or nasal masks, atelectasis or postsurgical sepsis than in ticenter trial demonstrating a clinically would provide results similar to those patients with severe hypoxemia associ- important outcome benefit of CPAP in that we describe herein. ated with cardiogenic pulmonary edema. this cohort of patients, its early termi- Third, previous studies showed that Application of noninvasive CPAP in pa- nation resulted in a relatively small num- application of CPAP several hours af- tients who develop severe hypoxemia af- ber of end points being reached and ana- ter the end of surgery31 or for a short ter major abdominal surgery, therefore, lyzed in each group. Nevertheless, the period32 did not result in any clinical should be limited to ICUs or high- results show that compared with stan- benefit. In our study, hypoxemia was dependency units where immediate in- dard treatment, early use of CPAP de- identified and treated immediately af- tubation for worsening of hypoxemia can creased the incidence of endotracheal inter surgery and CPAP was applied for be rapidly accomplished. tubation and other complications in a mean (SD) of 19 (22) hours and in- Allocation to treatment with oxygen patients with postoperative hypoxemia terrupted only when the oxygenation or oxygen plus CPAP was not blinded. after major abdominal surgery. Be- target for stopping treatment was To minimize potential bias in the as- cause of the low costs and the reduced reached. sessment of some of the study end risks associated with this approach, early Postoperative complications occur points, we used measures such as ob- use of CPAP appears to be a practical with a distinct temporal pattern: respi- jective criteria for endotracheal intuba- method for treating postoperative hy- ratory failure and pneumonia are seen tion and standardization of all cointer- poxemia in patients recovering from within the first week while sepsis and ventions that could have influenced elective major abdominal surgery.

Downloaded From: http://jama.jamanetwork.com/ on 02/11/2013 CPAP FOR POSTOPERATIVE HYPOXEMIA

Author Affiliations: Dipartimento di Anestesia, Azienda Drafting the manuscript: Squadrone, Cavallo, Ranieri. Molinette, Anestesia e Rianimazione 1-3, Torino: A.G. Ospedaliera S.Giovanni Battista-Molinette, Università di Critical revision of the manuscript for important in- Viale; Ospedale Civile, Chivasso: E. Castenetto; Osped- Torino (Drs Squadrone, Coha, Cerutti, Schellino, Oc- tellectual content: Coha, Cerutti, Schellino, Biolino, ale Civile Ivrea: M. R. Salcuni; Ospedale E. Agnelli, cella, and Ranieri), Servizio di Anestesia e Rianimazi- Occella, Belloni, Vilianis, Fiore. Pinerolo: L. Del Piano; Ospedale Civile S. Croce, Mon- one, Ospedale Civile di Chivasso (Dr Biolino), Servizio Statistical analysis: Squadrone, Cavallo, Ranieri. calieri: P. Buffa; Ospedale Civile, Asti: S. Cardellino, di Anestesia e Rianimazione, Ospedale Civile di Ivrea (Dr Obtaining funding: Ranieri. S. Perno; Ospedale Civile, Ciriè: P. Giugiaro; Osp. S. Belloni), Servizio di Anestesia e Rianimazione, Osped- Administrative, technical, or material support: Annunziata, Savigliano: G. Vai, M. Testa; Ospedale ale Civile di Pinerolo (Dr Vilianis), Servizio di Anestesia Squadrone, Coha, Cerutti, Schellino, Biolino, Occella, degli Infermi, Rivoli: B. Barberis, B. Babuin; Ospedale e Rianimazione, Ospedale S. Croce di Monacalieri (Dr Belloni, Vilianis, Fiore. S.Giovanni Antica Sede, Torino: L. Musso; Ospedale Fiore), and Dipartimento di Sanità Pubblica e Microbio- Supervision: Ranieri. Martini, Torino: A. Parigi, M.Torta; Ospedale Mag- logia, Università di Torino (Dr Cavallo), Italy. Financial Disclosures: None reported. giore della Carità, Novara: F. Della Corte, A. Gratarola; Author Contributions: Dr Ranieri had full access to all Funding/Support: This study was supported by grants Ospedale S.Giovanni Battista-Molinette, Anestesia e Ri- of the data and takes complete responsibility for the in- 3ACS-03 of Regione Piemonte: Ricerca Sanitaria animazione 6, Torino: P. Donadio, M. Vaj; IRCC tegrity of the data, and the accuracy of data analysis. Finalizzata and PR60ANRA03 of Università di Torino: Candiolo, Torino: F. De Bernardi, E. Moselli; Ospedale Study conception and design: Squadrone, Cavallo, Progetti di Ricerca Locali. Giovanni Bosco, Torino: E. Guglielmotti. Clinical coor- Ranieri. Role of the Sponsor: Neither grant agency influenced dinating center: Ospedale S. Giovanni Battista Moli- Acquisition of data: Squadrone, Coha, Cerutti, the design and conduct of the study, collection, man- nette, Università di Torino: G. Rosboch, A. Rossi, S. Schellino, Biolino, Occella, Belloni, Vilianis, Fiore, agement, analysis, and interpretation of the data, or Vendramin. Data safety and monitoring board: C. Fil- Ranieri. preparation, review, or approval of the manuscript. ippini, M. Pizzio, M. Eandi, G. Costa, and A. Pesenti. Analysis and interpretation of data: Squadrone, Participating Members of the Piedmont Intensive Care Protocol Review Committee: C. Gregoretti, A. Tempia, Cavallo, Ranieri. Unit Network (PICUN): Ospedale S. Giovanni Battista- G. Carbone.

REFERENCES 1. Arozullah AM, Daley J, Henderson WG, Khuri SF; definitions, mechanisms, relevant outcomes, and clini- 27. Mehta S, Hill NS. Noninvasive ventilation. Am J National Veterans Administration Surgical Quality Im- cal trial co-ordination. Am J Respir Crit Care Med. 1994; Respir Crit Care Med. 2001;163:540-577. provement Program. Multifactorial risk index for pre- 149:818-824. 28. Chiumello D, Pelosi P, Carlesso E, et al. Nonin- dicting postoperative respiratory failure in men after ma- 15. Delclaux C, L’Her E, Alberti C, et al. Treatment vasive positive pressure ventilation delivered by hel- jor noncardiac surgery. Ann Surg. 2000;232:242-253. of acute hypoxemic nonhypercapnic respiratory in- met vs standard face mask. Intensive Care Med. 2003; 2. O’Donohue WJ Jr. National survey of the usage sufficiency with continuous positive airway pressure 29:1671-1679. of lung expansion modalities for the prevention and delivered by a face mask: a randomized controlled trial. 29. Patroniti N, Foti G, Manfio A, Coppo A, Bellani G, treatment of postoperative atelectasis following ab- JAMA. 2000;284:2352-2360. Pesenti A. Head helmet vs face mask for non-invasive dominal and thoracic surgery. Chest. 1985;87:76-80. 16. Antonelli M, Pennisi MA, Pelosi P, et al. Non- continuous positive airway pressure: a physiological 3. Thompson JS, Baxter T, Allison JG, Johnson FE, Lee invasive positive pressure ventilation using a helmet study. Intensive Care Med. 2003;29:1680-1687. KK, Park WY. Temporal patterns of postoperative in patients with acute exacerbation of chronic ob- 30. Tonnelier JM, Prat G, Nowak E, et al. Noninva- complications. Arch Surg. 2003;138:596-603. structive pulmonary disease: a feasibility study. sive continuous positive airway pressure ventilation us- 4. Lang M, Niskanen M, Miettinen P, Alhava E, Takala Anesthesiology. 2004;100:16-24. ing a new helmet interface: a case-control prospec- J. Outcome and resource utilization in gastroenter- 17. Collins TC, Daley J, Henderson WH, Khuri SF. Risk tive pilot study. Intensive Care Med. 2003;29:2077- ological surgery. Br J Surg. 2001;88:1006-1014. factor for prolonged length of stay after major elec- 2080. 5. Bendixen HH, Hedley-White J, Laver MB. Im- tive surgery. Ann Surg. 1999;230:251-259. 31. Drummond GB, Stedul K, Kingshott R, et al. Au- paired oxygenation in surgical patients during gen- 18. Gordon Lan KK, DeMets DL. Discrete sequential tomatic CPAP compared with conventional treat- eral anesthesia with controlled ventilation: a concept boundaries for clinical trials. Biometrika. 1983;70:659- ment for episodic hypoxemia and sleep disturbance of atelectasis. N Engl J Med. 1963;269:991-996. 663. after major abdominal surgery. Anesthesiology. 2002; 6. Lindberg P, Gunnarsson L, Tokics L, et al. Atelec- 19. Pocock SJ. Group sequential methods in the de- 96:817-826. tasis and lung function in the postoperative period. sign and analysis of clinical trials. Biometrika. 1977;64: 32. Carlsson C, Sonden B, Thylen U. Can postopera- Acta Anaesthesiol Scand. 1992;36:546-553. 191-199. tive continuous positive airway pressure (CPAP) pre- 7. Hedenstierna G. Gas exchange during anaesthesia. 20. Moller JT, Wittrup M, Johansen SH. Hypoxemia vent pulmonary complications after abdominal surgery? Acta Anaesthesiol Scand Suppl. 1990;94:27-31. in the postanesthesia care unit: an observer study. Intensive Care Med. 1981;7:225-229. 8. Stock MC, Downs JB, Gauer PK, Alster JM, Imrey Anesthesiology. 1990;73:890-895. 33. Van Kaam AH, Lachmann RA, Herting E, et al. PB. Prevention of postoperative pulmonary compli- 21. Rothaar RC, Epstein SK. Extubation failure: mag- Reducing atelectasis attenuates bacterial growth and cations with CPAP, incentive spirometry, and conser- nitude of the problem, impact on outcomes, and translocation in experimental pneumonia. Am J Respir vative therapy. Chest. 1985;87:151-157. prevention. Curr Opin Crit Care Med. 2003;9:59-66. Crit Care Med. 2004;169:1046-1053. 9. Dehaven CB, Hurst JM, Branson RD. Postextuba- 22. Glanemann M, Langrehr J, Kaisers U, et al. Post- 34. Duggan M, McCaul CL, McNamara PJ, Engel- tion failure treated with a continuous positive pres- operative tracheal extubation after orthopic liver berts D, Ackerley C, Kavanagh BP. Atelectasis causes sure airway mask. Crit Care Med. 1985;13:46-48. transplantation. Acta Anaesthesiol Scand. 2001;45: vascular leak and lethal right ventricular failure in un- 10. Lindner KH, Lotz P, Ahnefeld FW. Continuous posi- 333-339. injured rat lungs. Am J Respir Crit Care Med. 2003;167: tive airway pressure effect on functional residual ca- 23. Bersten AD, Holt AW, Vedig AE, Skowronski GA, 1633-1640. pacity, vital capacity and its subdivisions. Chest. 1987; Baggoley CJ. Treatment of severe cardiogenic pulmo- 35. Plant PK, Owen JL, Elliott MW. Early use of non- 92:66-70. nary edema with continuous positive airway pres- invasive ventilation for acute exacerbations of chronic 11. Rusca M, Proietti S, Schnyder P, et al. Preven- sure delivered by face mask. N Engl J Med. 1991;325: obstructive pulmonary disease on general respiratory tion of atelectasis formation during induction of gen- 1825-1830. wards: a multicentre randomised controlled trial. Lancet. eral anesthesia. Anesth Analg. 2003;97:1835-1839. 24. Kaneko Y, Floras JS, Usui K, et al. Cardiovascular 2000;355:1931-1935. 12. Evans TW. International Consensus Confer- effects of continuous positive airway pressure in pa- 36. Antonelli M, Conti G, Moro ML, et al. Predictors ences in Intensive Care Medicine: non-invasive posi- tients with heart failure and obstructive sleep apnea. of failure of noninvasive positive pressure ventilation tive pressure ventilation in acute respiratory failure. N Engl J Med. 2003;348:1233-1241. in patients with acute hypoxemic respiratory failure: Intensive Care Med. 2001;27:166-178. 25. Keenan SP. Noninvasive positive pressure venti- a multi-center study. Intensive Care Med. 2001;27: 13. ACCP-SCCM Consensus Conference. Defini- lation in acute respiratory failure. JAMA. 2000;284: 1718-1728. tions of sepsis and multiple organ failure and guide- 2376-2378. 37. Morse MA, Califf RM, Sugarman J. Monitoring lines for the use of innovative therapies in sepsis. Crit 26. Pang D, Keenan SP, Cook DJ, Sibbald WJ. The and ensuring safety during clinical research. JAMA. Care Med. 1992;20:864-874. effect of positive pressure airway support on mortal- 2001;285:1201-1205. 14. Bernard GR, Artigas A, Brigham KL, et al. The ity and the need for intubation in cardiogenic pulmo- 38. Slutsky AS, Lavery JV. Data safety and monitor- American-European Consensus Conference on ARDS: nary edema. Chest. 1998;114:1185-1192. ing boards. N Engl J Med. 2004;350:1143-1147.

Downloaded From: http://jama.jamanetwork.com/ on 02/11/2013