Noninvasive Ventilation for Treatment of Acute Respiratory Failure in Patients Undergoing Solid Organ Transplantation a Randomized Trial

CARING FOR THE CRITICALLY ILL PATIENT

Noninvasive Ventilation for Treatment of Acute Respiratory Failure in Patients Undergoing Solid Organ Transplantation A Randomized Trial

Massimo Antonelli, MD Context Noninvasive ventilation (NIV) has been associated with lower rates of en- Giorgio Conti, MD dotracheal intubation in populations of patients with acute respiratory failure. Maurizio Bufi, MD Objective To compare NIV with standard treatment using supplemental oxygen administration to avoid endotracheal intubation in recipients of solid organ transplanta- Maria Gabriella Costa, MD tion with acute hypoxemic respiratory failure. Angela Lappa, MD Design and Setting Prospective randomized study conducted at a 14-bed, gen- Monica Rocco, MD eral intensive care unit of a university hospital. Alessandro Gasparetto, MD Patients Of 238 patients who underwent solid organ transplantation from Decem- ber 1995 to October 1997, 51 were treated for acute respiratory failure. Of these, 40 Gianfranco Umberto Meduri, MD were eligible and 20 were randomized to each group. Intervention Noninvasive ventilation vs standard treatment with supplemental oxy- N THE PAST 2 DECADES, ADVANCE- gen administration. ments in immunosuppressive strat- egies and major breakthroughs in Main Outcome Measures The need for endotracheal intubation and mechanical surgical and organ preservation ventilation at any time during the study, complications not present on admission, duration of ventilatory assistance, length of hospital stay, and intensive care unit mor- Itechniques have transformed organ tality. transplantation into a therapy for an in- creasing population of patients with end- Results The 2 groups were similar at study entry. Within the first hour of treatment, 14 patients (70%) in the NIV group, and 5 patients (25%) in the standard treatment stage organ failure. Although prevent- group improved their ratio of the PaO to the fraction of inspired oxygen (FIO ). Over ing rejection remains the principle focus 2 2 time, a sustained improvement in PaO2 to FIO2 was noted in 12 patients (60%) in the in improving overall survival statistics, NIV group, and in 5 patients (25%) randomized to standard treatment (P = .03). The pulmonary complications following use of NIV was associated with a significant reduction in the rate of endotracheal in- transplantation are responsible for most tubation (20% vs 70%; P = .002), rate of fatal complications (20% vs 50%; P = .05), morbidity and contribute substantially length of stay in the intensive care unit by survivors (mean [SD] days, 5.5 [3] vs 9 [4]; to the mortality associated with vari- P = .03), and intensive care unit mortality (20% vs 50%; P = .05). Hospital mortality ous organ transplantation procedures.1 did not differ. Approximately 5% of patients undergo- Conclusions These results indicate that transplantation programs should consider ing renal, hepatic, cardiac, or pulmonary NIV in the treatment of selected recipients of transplantation with acute respiratory transplantationdeveloppneumoniainthe failure. period after transplantation, which has JAMA. 2000;283:235-241 www.jama.com an associated crude mortality of 37%.1 In patients with acute respiratory failure the need for an invasive artificial air- Author Affiliations are listed at the end of this ar- (ARF), endotracheal intubation is the way.2 In ARF, when noninvasive ven- ticle. Corresponding Author and Reprints: Massimo Antonelli, single most important predisposing fac- tilation (NIV) is effective in avoiding MD, Istituto di Anestesiologia e Rianimazione, Univer- tor for developing nosocomial bacterial endotracheal intubation, the inci- sita` Cattolica del Sacro Cuore, 1 Largo F. Vito, 00168 2 Rome, Italy (e-mail: [email protected]). pneumonia. dence of bacterial pneumonia is ex- Caring for the Critically Ill Patient Section Editor: Noninvasive positive-pressure ven- tremely low.3 Small, uncontrolled stud- Deborah J. Cook, MD, Consulting Editor, JAMA. Advisory Board: David Bihari, MD; Christian Brun- tilation refers to the delivery of as- ies in recipients of lung transplants have Buisson, MD; Timothy Evans, MD; John Heffner, MD; sisted mechanical ventilation without reported that NIV permits ealier extu- Norman Paradis, MD.

Downloaded From: https://jamanetwork.com/ on 10/01/2021 NONINVASIVE VENTILATION IN PATIENTS WITH SOLID ORGAN TRANSPLANTS

bation after transplantation surgery,4 and neous presence of renal and cardiovas- ume of 8 to 10 mL/kg, a respiratory rate can prevent need for intubation in those cular failures)7; and tracheostomy, fa- of fewer than 25/min, the disappear- with ARF.1,5 Randomized studies of NIV cial deformities, or recent oral, ance of accessory muscle activity (as in solid organ transplant recipients with esophageal, or gastric surgery. The sim- evaluated by palpating the sternoclei- hypoxemic ARF are lacking. plified Acute Physiologic Score was cal- domastoid muscle),9 and patient com- In a previous randomized study we culated on admission to the study.8 fort. Positive end-expiratory pressure demonstrated the efficacy of NIV to To minimize the risk of bias due to was increased in increments of 2 to 3 treat immunocompetent patients with the obvious difficulty of blinding in this cm H2O repeatedly up to 10 cm H2O un- ARF of various origins, comparing NIV study, medical management of the ARF til the FIO2 requirement was 0.6 or less. delivered through a face mask with con- (eg, antibiotic, antiviral, or antifungal Ventilator settings were adjusted based ventional mechanical ventilation de- agents; bronchodilators; diuretics; fre- on continuous oximetry and measure- livered through an endotracheal tube.6 quent respiratory treatments and chest ments of arterial blood gases. Patients We studied solid organ transplant re- physiotherapy), immunosuppressive were not sedated. cipients with hypoxemic ARF and com- therapy (corticosteroids, azathio- Ventilation was standardized accord- pared NIV delivered through a face mask prine, cyclosporine), time of medical in- ing to the protocol of Wysocki et al.10 with standard treatment using oxygen terventions, central venous pressure and During the first 24 hours, ventilation supplementation to avoid endotra- cardiac output monitoring, frequency was continuously maintained until oxy- cheal intubation and decrease duration of blood gases, and other aspects of ICU genation and clinical status improved. of intensive care unit (ICU) stay. support (head of the bed kept el- Subsequently, each patient was evalu- evated at a 45° angle, nutrition, fluid ated daily while breathing supplemen- METHODS administration, and correction of elec- tal oxygen without ventilatory sup- Study Design and Patient Selection trolyte abnormalities) were similar in port for 15 minutes. Noninvasive We enrolled all consecutive adult re- the 2 groups. Both groups were treated ventilation was reduced progressively cipients of solid organ transplants ad- by the same medical and nursing staffs. in accordance with the degree of clini- mitted to the 14-bed general ICU of La Patients assigned to the standard treat- cal improvement and was discontin- Sapienza University Hospital (Rome, ment group received oxygen supple- ued if the patient maintained a respi- Italy) with acute hypoxemic respir- mentation via a Venturi mask starting ratory rate lower than 30/min and a atory failure. Patients enrolled were with an FIO2 equal to or greater than 0.4, PaO2 greater than 75 mm Hg with a FIO2 randomly assigned to receive either and adjusted to achieve a level of arte- of 0.5 without ventilatory support. standard treatment with oxygen supple- rial oxygen saturation (by oximetry) mentation delivered by Venturi mask above 90%. All patients had continu- Endotracheal Intubation or NIV through a face mask. Computer- ous electrocardiographic and arterial Patients who failed standard treat- generated random assignments were oxygen saturation monitoring (Biox ment or NIV underwent endotracheal concealed in sealed envelopes. A hos- 3700, Ohmeda, Boulder, Colo). We intubation with cuffed endotracheal pital ad hoc ethics committee ap- used 2 types of mechanical ventila- tubes (internal diameter of 7.5-8.5 mm) proved the protocol, and all patients or tors: the Puritan Bennett 7200 (Puri- and were mechanically ventilated. Pre- the next of kin gave written informed tan Bennett Co, Overland Park, Kan) determined criteria included failure to consent. and the Servo 900 C Siemens (Sie- maintain a PaO2 above 65 mm Hg with The criteria for eligibility were acute mens Elema, Uppsala, Sweden). an FIO2 equal to or greater than 0.6; de- respiratory distress; a respiratory rate velopment of conditions necessitating greater than 35/min, a ratio of the PaO2 Noninvasive Ventilation endotracheal intubation to protect the to the fraction of inspired oxygen (FIO2) For patients assigned to NIV, the ven- airways (coma or seizure disorders) or (PaO2:FIO2) of less than 200 while the pa- tilator was connected with conven- to manage copious tracheal secre- tient was breathing oxygen through a tional tubing to a clear, full face mask tions, hemodynamic or electrocardio- Venturi mask; and active contraction of with an inflatable soft cushion seal and graphic instability; inability to correct the accessory muscles of respiration or a disposable foam spacer to reduce dead dyspnea; or inability on the part of the paradoxical abdominal motion. space (Gibeck, Upplands, Sweden). The patient randomized to NIV to tolerate Exclusion criteria were a requirement mask was secured with head straps. In the face mask.6 for emergent intubation for cardiopul- most patients, a hydrocolloid sheet was monary resuscitation, respiratory arrest, applied over the nasal bridge. For pa- Conventional Ventilation severe hemodynamic instability, de- tients with a nasogastric tube, a seal Intravenous benzodiazepines (diaz- creased level of consciousness; respi- connector in the dome of the mask was epam, 0.2 mg/kg [bolus]) or propofol (2 ratory failure caused by neurological used to minimize air leakage. After the mg/kg) were used for sedation at the mo- disease or status asthmaticus; more than mask was secured, pressure support was ment of intubation, and none of the pa- 2 new organ failures (eg, the simulta- increased to obtain an exhaled tidal vol- tients received paralyzing agents. The

Downloaded From: https://jamanetwork.com/ on 10/01/2021 NONINVASIVE VENTILATION IN PATIENTS WITH SOLID ORGAN TRANSPLANTS

initial ventilator setting was an assisted- defined as ability to maintain the de- patients without organ transplantation controlled ventilation mode with a defined improvement in PaO2:FIO2 until and with acute hypoxemic respiratory livered tidal volume of 10 mL/kg and a mechanical ventilation was discontin- failure supported with NIV required en- respiratory rate of 14 to 18/min, a posi- ued, as confirmed by serial blood gas dotracheal intubation.6 A sample size of tive end-expiratory pressure of 5 cm measurements. 40 patients was chosen20 to detect, with H2O, and an FIO2 of 0.8. Positive end- Patients were monitored for the de- a 95% probability, a difference be- expiratory pressure was increased in in- velopment of infections or other com- tween the postulated 70% rate of intu- crements of 2 to 3 cm H2Oupto10cm plications. Sepsis, severe sepsis, and sep- bation in the standard treatment group H2O until the FIO2 requirements were tic shock were defined according to and a 30% rate in the NIV group, with less than or equal to 0.6. The head of the consensus guidelines.13 Infection was a power of 80%. The odds ratios (ORs), bed was kept elevated at 45° to mini- diagnosed using strict criteria.14 Pa- relative risks, and 95% confidence in- mize the risk of aspiration. When spon- tients in whom clinical manifestation tervals (CIs) are given with the ␹2 val- taneous breathing reappeared, the ven- of pneumonia developed15 underwent ues and P values.21 tilator settings were changed to bronchoscopy with bronchoalveolar la- intermittent mandatory ventilation (rate, vage. The methods and laboratory pro- RESULTS 4-7/min) with pressure support (14 to cedures followed consensus guide- Between December 1995 and October 16,17 20 cm H2O) titrated to achieve a spon- lines. Bacterial pneumonia was 1997, 238 adults received a solid or- taneous tidal volume of 8 to 10 mL/kg, diagnosed when more than 104 colony- gan transplant (liver, lung, or kidney). a respiratory rate less than 25/min, and forming units of bacteria per milliliter Fifty-one patients were treated in our disappearance of accessory muscle ac- were measured in bronchoalveolar la- ICU for hypoxemic ARF occurring at tivity.11 All patients were weaned from vage fluid.16 Diagnostic criteria for op- different time intervals after transplan- the ventilator by reducing the level of portunistic pneumonia were previ- tation (TABLE 1). Three patients had an 17,18 pressure support by 4 cm H2O twice and ously described. Because infections exclusion criterion (1 tracheostomy and then decreasing the ventilatory rate by in patients receiving mechanical ven- 2 cases of impaired consciousness) and 2/min at 2-hour intervals as tolerated. tilation are frequently associated with 8 refused to participate, thus 40 were If the patient tolerated an intermittent an invasive device,14 an index of inva- enrolled. Twenty patients were as- mandatory ventilation rate of 0.5/min, siveness was established by counting the signed to each group and all com- with a pressure support level of 8 cm number of devices (central venous, ar- pleted the study and follow-up. The H2O and an FIO2 of less than or equal terial, pulmonary artery and urinary baseline characteristics of the 2 groups to 0.5, a 2-hour T-piece trial was initi- catheters, drainage tubes, endotra- were similar (Table 1). Reasons for ated.11 Patients were extubated if they cheal, and nasogastric tubes) per pa- transplantation for the NIV group and maintained a respiratory rate less than tient at study entry. The duration of use standard treatment group were as fol- 30/min and a PaO2 greater than 75 of the invasive devices was calculated lows, respectively, liver transplanta- mm Hg.11 as the number of days during which all tion: posthepatic cirrhosis, 6 and 7; he- the invasive devices counted on admis- patic cancer, 2 and 2; cystic fibrosis, 1 End Points and Definitions sion were maintained per patient. Cri- and 0; amanita phalloides intoxica- The primary outcome variable was the teria for adult respiratory distress syn- tion, 1 and 0; and alcoholic cirrhosis, need for endotracheal intubation and drome (ARDS) followed consensus 0 and 3; and lung transplantation: cys- mechanical ventilation at any time dur- guidelines.19 Multiple organ failure was tic fibrosis, 2 and 2; ␣-1 antitrypsin defi- ing the study. Secondary end points in- defined as previously described.7 cit, 1 and 0; and severe bronchiectasis, cluded complications not present on ad- 1 and 0. One patient in the NIV group mission, duration of ventilatory Statistical Analysis received a single lung transplant. The assistance, length of the hospital stay, Results are given as mean (SD). Demo- other 3 in the NIV group and those in and ICU mortality. graphic and physiologic characteris- the standard treatment group had bilat- Arterial blood gas levels were deter- tics for the 2 groups were compared us- eral lung transplantation. Reasons for re- mined at baseline, at 1 hour, and at ing the t test for continuous data and nal transplantation were as follows: 4-hour intervals. Improvement in gas ex- with the Mantel-Haenszel extended ␹2 membranous glomerulonephritis, 2 and change was defined as ability to in- test for categorical data. The 2-tailed 3; Berger disease, 2 and 1; chronic py- crease PaO2:FIO2 above 200 or an in- Fisher exact test was used when the ex- elonephritis, 1 and 2; and polycystic kid- crease in this ratio of more than 100 from pected number of cases per cell was less ney disease, 1 and 0. All renal transplan- baseline.12 Improvement in gas ex- than 5. tation recipients had end-stage renal change was evaluated within 1 hour (ini- In the 2 years preceding this study, failure on hemodialysis. tial improvement) after study entry and 70% of recipients of solid organ trans- At study entry, 11 patients had a re- over time (sustained improvement). Sus- plants with ARF required endotracheal cent diagnosis of pneumonia made by tained improvement in gas exchange was intubation. In the same period, 30% of bronchoscopic bronchoalveolar la-

Downloaded From: https://jamanetwork.com/ on 10/01/2021 NONINVASIVE VENTILATION IN PATIENTS WITH SOLID ORGAN TRANSPLANTS

vage prior to admission to the ICU, and pneumonia were caused by Staphylo- plantation), Acinetobacter (1 case; 16 5 had an extrapulmonary infection. coccus aureus (1 case; 10 days after liver days after liver transplantation), P aeru- Seven of these 11 diagnostic bronchos- transplantation), Aspergillus (1 case; 15 ginosa (1 case; 8 days after renal trans- copies were performed on patients more days after renal transplantation), Pneu- plantation); 2 cases of pneumonia (1 than 24 hours prior to ICU admission mocystis carinii (1 case; patient non- case of Cytomegalovirus and 1 S aureus) with the aid of NIV.15 Noninvasive ven- compliant with antipneumocystis that precipitated ARDS. tilation was applied for less than 45 prophylaxis, 45 days after liver trans- Changes in PaO2:FIO2 and PaCO2 are minutes, and arterial blood gas obtained plantation), Nocardia (1 case; cystic shown in FIGURE 1. Within the first within 30 minutes of NIV withdrawal fibrosis patient, 53 days after lung trans- hour of treatment, 14 patients (70%) in were similar to prebronchoscopy val- plantation), and Pseudomonas aerugi- the NIV group and 5 patients (25%) in ues. Nine patients had definitive diag- nosa (1 case; 45 days after liver the standard treatment group had an nosis of pneumonia established: 6 were transplantation). The latter 3 pneumo- improvement in PaO2:FIO2 by study cri- bacterial and 3 opportunistic. In the NIV nia cases precipitated ARDS. In the stan- teria (OR, 7; 95% CI, 1.4-37; P = .005). group, 2 patients had radiographic dard treatment group, pneumonia was A sustained improvement in PaO2: manifestations suggestive of pneumo- caused by Cytomegalovirus (1 case; FIO2 over time (TABLE 2) was ob- nia before developing ARDS, but had positive viremia, 48 days after renal served in 12 patients randomized to negative bronchoalveolar lavage find- transplantation), S aureus (1 case; NIV (mean [SD], 142 [29] at baseline ings, blood culture samples, and sero- patient with renal failure requiring vs 271 [98] at the end of treatment; logical test results; the other cases of hemodialysis 60 days after renal trans- PϽ.001) and in 5 patients randomized to standard treatment (149 [22] at baseline vs 270 [18] at the end of treat- Table 1. Baseline Characteristics of the Patients and Causes of Acute Respiratory Failure* ment; PϽ.001). All patients with sus- Noninvasive Standard O IO Ventilation Treatment P tained improvement in Pa 2:F 2 over Group (n = 20) Group (n = 20) Value time avoided endotracheal intubation Age, y 45 (19) 44 (10) .89 (Table 2). Four patients in the NIV No. (%) of men 13 (65) 12 (60) .50 group did not have sustained improve- Simplified Acute Physiologic Score 13 (4) 13 (3) .93 ment in PaO2:FIO2, did not meet the pre- No. of invasive devices per patient 5 (1) 5 (1) .90 selected criteria for intubation, avoided Heart rate, beats/min 96 (20) 101 (14) .38 intubation, and survived. Respiratory rate, breaths/min 38 (3) 37 (1) .32 Body temperature, °C 37.2 (0.9) 37 (0.7) .35 Figure 1. Changes in the Ratio of PaO2 to 9 White blood cells, ϫ10 /L 0.005 (0.002) 0.007 (0.005) .12 Fraction of Inspired Oxygen (FIO2) and PaCO2 No. (%) of infections prior to entry 8 (40) 9 (45) .19 Over Time Systolic blood pressure, mm Hg 135 (23) 140 (24) .53

Arterial pH 7.46 (0.05) 7.43 (0.04) .13 PaO2:FIO2 PaCO2 Noninvasive Ventilation Noninvasive Ventilation PaCO2, mm Hg 42 (10) 38 (9) .14 Standard Treatment Standard Treatment

No. (%) of patients with PaCO2 Ͼ45 mm Hg 7 (35) 3 (15) .13

Ratio of PaO2 to fraction of inspired oxygen 129 (30) 129 (30) .96 300 100 No. (%) of patients who received 90 an organ transplant 250 Liver 10 (50) 12 (60) .37 ∗ ∗ 80 Mean Pa Lung 4 (20) 2 (10) .33 200 Ratio

Kidney 6 (30) 6 (30) .63 2 † 70 IO CO F

Time from transplantation, d† 23 (14) 22 (15) .88 2: 150 60 2 O , mm Hg Causes of acute respiratory failure‡ 50 Pneumonia 2 (10) 2 (10) .69 100

Cardiogenic pulmonary edema 4 (20) 5 (25) .50 Mean Pa 40 Acute respiratory distress syndrome§ 8 (40) 7 (35) .50 50 30 Mucous plugging or atelectasis 5 (25) 5 (25) .64 0 20 Pulmonary embolism 1 (5) 1 (5) .75 Baseline 1 h After Start Termination *Values are expressed as mean (SD) unless otherwise indicated. of Treatment of Treatment †Median time from transplantation to study entry was 18 days (range, 5-60 days). All transplant recipients were treated in our intensive care unit (ICU) both for their immediate postoperative care and for complications. All patients were Data are presented as mean (SD). Asterisk indicates already hospitalized at the time of ICU admission. Five patients (25%) in the noninvasive group and 4 (20%) in the PϽ.001; dagger, PϽ.005 vs baseline. A paired t test standard treatment group were discharged from the hospital after transplantation and readmitted. ‡Values are the number (percentage) of patients who experienced acute respiratory failure. was used for the statistical comparison. Termination §Conditions for acute respiratory distress syndrome are reported for the noninvasive ventilation group and standard of treatment refers to the last arterial blood gas value treatment group, respectively: complicated pneumonia, 5 and 2; extrapulmonary sepsis, 2 and 4; massive blood obtained prior to intubation or prior to removal of oxy- transfusion, 1 and 0; and acute pancreatitis, 0 and 1. gen supplementation.

Downloaded From: https://jamanetwork.com/ on 10/01/2021 NONINVASIVE VENTILATION IN PATIENTS WITH SOLID ORGAN TRANSPLANTS

Overall, 18 patients underwent intu- Four patients (20%) in the NIV group or without septic shock) developed as bation (10 orotracheal and 8 nasotra- and 10 patients (50%) in the standard frequently in the NIV group than in the cheal) at a mean (SD) of 43 (45) hours treatment group (all of whom re- standard treatment group (4 vs 10; into the study (Table 2), 4 patients (20%) quired intubation) died in the ICU (OR, P = .05). Four patients randomized to in the NIV group and 14 patients (70%) 4; 95% CI, 0.8-20; P = .05). A sub- NIV developed criteria for severe sep- in the standard treatment group group death rate is reported in Table sis or septic shock after study entry. The (P = .002). None required emergent in- 2. Four patients (2 patients with myo- sources of sepsis included 2 cases of tubation. The reasons for intubation by cardial infarction and 1 with a new pul- pneumonia present at study entry that NIV group and standard treatment monary embolism in the NIV group and worsened, and 2 ventilator-associated group, respectively, included failure to 1 with septic shock in the standard pneumonia cases that developed after maintain PaO2 level above 65 mm Hg (3 treatment group) died in the hospital endotracheal intubation. Ten patients and 5), hemodynamic instability (1 and after ICU discharge. in the standard treatment group de- 3), management of secretions (0 and 3), The complications and events lead- veloped criteria for severe sepsis (6) and severe persistent dyspnea (0 and 3). ing to death are shown in TABLE 3. Fa- or septic shock (4) after study entry Thirteen patients required intubation tal complications were less frequent in and developed multiple organ failure within 24 hours of study entry, 10 in the the NIV group than in the standard including renal failure. The sources of standard treatment group and 3 in the treatment group (4 vs 10; P = .05). As sepsis included 2 cases of pneumonia NIV group (P = .02; FIGURE 2). In a sub- shown in Table 3, severe sepsis (with and 1 pancreatic abscess present at group analysis shown in Table 2, patients with ARDS due to either pulmonary or Table 2. Outcome Variables* nonpulmonary causes randomized to Noninvasive Standard NIV had an intubation rate of 38% vs Ventilation Treatment P 86% in the standard treatment group Variable Group (n = 20) Group (n = 20) Value (P = .08). Irrespective of randomization, Initial improvement in ratio of PaO2 to fraction of 14 (70) 5 (25) .005 inspired oxygen patients with pneumonia (opportunis- Sustained improvement in ratio of PaO2 to fraction of 12 (60) 5 (25) .03 tic or nosocomial) had a similar intuba- inspired oxygen, without intubation tion rate. Among patients with pulmo- Patients intubated within 24 h of study entry 3 (15) 10 (50) .02 nary edema or pulmonary embolism, all Patients requiring intubation 4 (20) 14 (70) .002 those randomized to NIV avoided intu- Failures per subgroup of patients bation, while 5 (83%) of the 6 patients Acute respiratory distress syndrome 2/5 (40) 2/2 (100) .28 (pulmonary etiology)† randomized to standard treatment re- Acute respiratory distress syndrome 1/3 (33) 4/5 (80) .28 quired intubation (P = .01). (extrapulmonary etiology)† Positive end-expiratory pressure ap- Pneumonia† 1/2 (50) 1/2 (50) .83 plied to the patients in the NIV group Cardiogenic pulmonary edema† 0/4 (0) 5/5 (100) .007 was lower than that used for the 14 pa- Pulmonary embolism 0/1 (0) 0/1 (0) .99 tients who failed standard treatment and Mucous plugging or atelectasis† 0/5 (0) 2/5 (40) .22 required intubation (mean [SD], 6 [1] Duration of mechanical ventilation, d‡§ 4 (5) 5 (6) .58 vs 8 [2] cm H2O; P = .02). The mean du- Duration of mechanical ventilation in survivors, d‡ 2 (0.7) 1.6 (2) .50 ration of NIV was 50 hours (range, 16- Duration of use for all invasive devices present at 5 (5) 9 (6) .05 94) for the 16 patients whose treat- study entry, d‡ ments were successful and 43 hours Length of intensive care unit stay, d‡ 7 (5) 10 (6) .18 (range, 10-120) for the 4 patients whose Length of intensive care unit stay in survivors, d‡ 5.5 (3) 9 (4) .03 treatments failed. As shown in Table 2, Intensive care unit deaths࿣ 4 (20) 10 (50) .05 Intensive care unit deaths per subgroup of patients† the invasive devices present at study en- Acute respiratory distress syndrome 3/8 (37) 4/7 (57) .40 try were used for a shorter period of time Pneumonia 1/2 (50) 1/2 (50) .80 in the group randomized to NIV than in Cardiogenic pulmonary edema 0/4 (0) 4/5 (80) .04 the group randomized to standard treat- Pulmonary embolism 0/1 (0) 0.1 (0) .99 ment (mean [SD] days, 5 [5] vs 9 [6]; Mucous plugging or atelectasis 0/5 (0) 1/5 (20) .50 P = .05). Hospital deaths¶ 7 (35) 11 (55) .17 Length of stay in the ICU was not dif- *Values are expressed as number (percentage) unless otherwise indicated. ferent in the 2 groups, but the 16 sur- †Values are expressed as No./total (percentage). ‡Values are expressed as mean (SD). vivors in the NIV group stayed in the §Duration of mechanical ventilation in patients randomized to standard treatment group refers to those patients who failed standard treatment and were intubated. ICU shorter than 10 survivors in the ࿣All deaths were due to complications that occurred after intubation. standard treatment group (mean [SD] ¶In the 2 years preceding this study, our overall institutional mortality for solid organ transplant recipients was 24%, and for those developing acute hypoxemic respiratory failure, 53%. days, 5.5 [3] vs 9 [4]; P = .03).

Downloaded From: https://jamanetwork.com/ on 10/01/2021 NONINVASIVE VENTILATION IN PATIENTS WITH SOLID ORGAN TRANSPLANTS

study entry that worsened. Seven in- Two patients in the NIV group and physiology of hypoxemic respiratory fections that developed after endotra- 4 in the standard treatment group de- failure caused by pneumonia, cardio- cheal intubation included 4 cases of veloped ventilator-associated pneumo- genic pulmonary edema, atelectasis, and ventilator-associated pneumonia, 1 nia (2 cases of P aeruginosa, 1 case of postoperative changes in pulmonary urinary tract infection (septic shock), Acinetobacter, 2 cases of methicillin- function.2 Similar to our prior report in and 2 intra-abdominal infections (1 resistant S aureus, and 1 case of Serra- immunocompetent patients with hy- case of septic shock). Septic complica- tia marcescens) diagnosed 5.6 (1) days poxemic respiratory failure, the venti- tions were associated with a duration after intubation, and all died due to lator protocol for NIV achieved a rapid of use of 6 or more days of the inva- multiple organ failure. None devel- and sustained improvement in gas- sive devices (OR, 5; 95% CI, 0.9-27; oped clinical or radiographic manifes- exchange abnormalities.6,12 We ex- P = .02). tations of pneumonia during NIV. One panded on prior uncontrolled reports patient in the NIV group had facial skin of 4 patients with lung and heart-lung necrosis that resolved within 8 days. transplants supported with NIV after Figure 2. Timing of Endotracheal Intubation developing severe respiratory infec- in the 2 Groups COMMENT tions leading to ARF.3,5 Our findings are

Noninvasive Ventilation In this randomized trial that was pow- similar to prior randomized studies of Standard Treatment ered to address intubation differences immunocompetent patients with acute and not mortality, early application of hypoxemic6,12,22 and hypercapnic res- 5 NIV in a group of solid organ recipi- piratory failure,12,22,23 and provide ad- 4 ents with hypoxemic ARF was well tol- ditional evidence of inherent advan- erated and associated with a rapid and tages of appropriately applied NIV.2 3 sustained improvement in gas ex- Few studies have reported on the ap- change. Compared with standard treat- plication of NIV in ARDS.6,12,24-26 In a re- 2 ment with supplemental oxygen, cent randomized study of patients with 1 patients randomized to NIV had signifi- hypoxemic ARF requiring mechanical 6 No. of Transplant Recipients No. of Transplant cantly lower rates of endotracheal intu- ventilation, we reported that 7 (22%) 0 3 h 12 h 12-24 h 3-5 d bation, septic complications, fatal com- of 32 patients randomized to NIV had Time From Study Entry plications, and ICU mortality. ARDS of varied etiology. Four (58%) of Half of the patients in the standard the 7 patients with ARDS avoided intu- Eighteen of the 40 patients required intubation, 14 (70%) in the standard treatment group and 4 (20%) treatment group necessitated endotra- bation and survived, while 3 patients in the noninvasive ventilation group (P = .002). Thir- cheal intubation within 24 hours of (42%) required intubation and died. teen patients required intubation within 24 hours from 26 the study entry (10 in the standard and 3 in the non- study entry. Physiological studies have Rocker et al recently reported the use invasive ventilation group; P = .02). shown that NIV can improve the patho- of NIV during 12 episodes of hypoxemic ARF occurring in hemodynami- cally stable patients with acute lung in- Table 3. Serious Complications and Fatal Events in the 2 Groups jury or ARDS. Intubation was required Noninvasive Standard Ventilation Treatment P in 34% of the episodes, and ICU mor- Group (n = 20) Group (n = 20) Value tality was 30%. In our study, the 8 ARDS No. (%) of patients with complications 8 (40) 12 (60) .17 patients randomized to NIV had an in- No. of complications occurring after intubation and 4 10 .05 tubation rate of 37.5% and a mortality causing death in intensive care unit rate of 37%. These findings are limited No. of complications per patient 1.12 1.4 .60 to a small selected patient population Total No. of complications/No. causing death in intensive care unit (%) and are insufficient for evaluating the Cardiogenic shock 1/0 (5) 2/2 (10) .50 role of NIV in ARDS. The studies pub- Organ rejection 4/0 (20) 3/0 (15) .25 lished to date should not be inter- Primary liver malfunction 1/0 (5) 1/0 (5) .75 preted to support the use of NIV in Worsening of infections present at study entry* 2/2 (10) 3/3 (15) .50 ARDS, but should provide the ratio- Ventilator-associated pneumonia 2/2 (10) 4/4 (20) .33 nale for a prospective randomized study. Severe sepsis and septic shock with multiple 4/4 (20) 10/8 (50) .05 The necessity of using immunosup- organ failure after study entry pressive therapy to prevent rejection in Gastrointestinal bleeding 1/0 (5) 0/0 (0) .99 recipients of solid organ transplants in- *At study entry, 5 patients had an identified infection without meeting clinical criteria for severe sepsis; 4 had pneumonia; and 1 had pancreatic abscess. These infections worsened; at 3 (2) days, expressed as mean (SD), after study creases morbidity and mortality asso- entry the patients met criteria for severe sepsis (2 pneumonia, 1 in each group), or septic shock (2 pneumonia and 1 1 pancreatic abscess). The other 9 septic complications occurred 12 (5) days after study entry. Note that the num- ciated with pulmonary infections. ber of septic complications occurring after the study entry included those infections present at study entry that wors- Nosocomial pneumonia is a frequent ened and pneumonia leading to severe sepsis and septic shock. complication of mechanical ventila-

Downloaded From: https://jamanetwork.com/ on 10/01/2021 NONINVASIVE VENTILATION IN PATIENTS WITH SOLID ORGAN TRANSPLANTS

tion and is an important factor in de- operatively with NIV.30 In the present fied Acute Physiology Score for ICU patients. Crit Care Med. 1984;12:975-977. termining outcome of respiratory fail- study, 1 of the 4 lung recipients who 9. Brochard L, Harf A, Lorino H, Lemaire F. Inspira- ure. In the present study, ventilator- had cystic fibrosis received NIV as a tory pressure support prevents diaphragmatic fa- tigue during weaning from mechanical ventilation. Am associated pneumonia developed after bridge for transplantation. He was ran- Rev Respir Dis. 1989;139:513-521. intubation in one third of patients and domized to the NIV group after ARF in 10. Wysocki M, Tric L, Wolff MA, et al. Noninvasive was associated with 100% mortality. As the posttransplantation period, avoided pressure support ventilation in patients with acute respiratory failure. Chest. 1995;107:761-768. previously observed in immunocom- intubation, and was successfully dis- 11. Brochard L, Rauss A, Benito S, et al. Comparison petent patients with hypoxemic or hy- charged from the hospital. In this study, of three methods of gradual withdrawal from ventila- 6,12 tory support during weaning from mechanical venti- percapnic ARF, patients random- hospital mortality was similar in the 2 lation. Am J Respir Crit Care Med. 1994;150:896-903. ized to NIV had fewer fatal septic groups. Individual factors and evolu- 12. Meduri GU, Turner RE, Abou-Shala N, et al. Non- tion of the surgical complications might invasive positive pressure ventilation via face mask. complications than patients random- Chest. 1996;109:179-193. ized to standard treatment, with a lower be important determinants of the final 13. Bone RC, Balk RA, Cerra FB, et al. Definitions for mortality in the ICU. These findings are outcome. Studies specifically powered sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest. 1992;101:1644- in agreement with the observations of to address mortality are needed before 1655. a recent prospective epidemiological drawing conclusions on this issue. 14. Meduri GU, Mauldin GL, Wunderink RG, Lee KV, Jones C, Trolley E. Causes of fever and pulmonary den- study of patients with ARF requiring In conclusion, in a group of organ sities in patients with clinical manifestations of ventilator- mechanical ventilation. Nourdine et al27 transplant recipients with ARF of vari- associated pneumonia. Chest. 1994;106:221-235. ous origins, early administration of NIV 15. Antonelli M, Moro ML, Capelli O, et al. Risk fac- reported that patients supported non- tors for early onset pneumonia in trauma patients. invasively vs those that received intu- was well tolerated and associated with Chest. 1994;105:224-228. bation had a lower incidence of noso- a significant reduction in the rate of en- 16. Meduri GU, Chastre J. The standardization of bronchoscopic techniques for ventilator-associated pneu- comial infections (pulmonary and dotracheal intubation, fatal complica- monia. Chest. 1992;102(suppl):557-564. extrapulmonary; PϽ.01), a shorter du- tions, and ICU mortality. Active trans- 17. Antonelli M, Conti G, Riccioni L, Meduri GU. Non- invasive positive-pressure ventilation via face mask dur- ration of ICU stay (PϽ.01), and a lower plantation programs should consider ing bronchoscopy with BAL in high-risk hypoxemic pa- mortality (PϽ.01). In our study, we NIV in the treatment of eligible patients tients. Chest. 1996;110:724-728. found that a longer use of invasive de- with ARF who have no contraindica- 18. Gentile G, Micozzi A, Girmenia C, et al. Pneumo- nia in allogenic and autologous bone marrow recipi- vices was associated with a higher in- tions and who can be monitored safely ents: a retrospective study. Chest. 1993;104:371-375. cidence of septic complications. Trans- in the appropriate environment. 19. Bernard GR, Artigas A, Brigham KL, et al. The American-European Consensus Conference on ARDS. plant recipients randomized to NIV had Author Affiliations: Istituto di Anestesiologia e Rian- Am J Respir Crit Care Med. 1994;149:818-824. a shorter use of invasive devices and a imazione, Università Cattolica del Sacro Cuore Rome, 20. Kraemer H, Thiemann S. How Many Subjects? Sta- Italy (Drs Antonelli and Conti), Universita` La Sapi- tistical Power Analysis in Research. Newbury Park, En- lower rate of nosocomial infections. enza Policlinico Umberto I, Rome (Drs Bufi, Costa, gland: Sage Publications; 1987. Avoiding intubation with early imple- Lappa, Rocco, and Gasparetto); and the Department 21. Snedecor GW, Cochran WG. Statistical Meth- mentation of NIV should be an impor- of Medicine, Pulmonary and Critical Care Division, ods. 6th ed. Ames: Iowa State University Press; 1967. Memphis Lung Research Program, University of Ten- 22. Keenan SP, Kernerman PD, Cook DJ, et al. Effect tant objective in the management of nessee, Memphis (Dr Meduri). of noninvasive positive pressure ventilation on mor- respiratory failure after solid organ tality in patients admitted with acute respiratory failure. Crit Care Med. 1997;25:1685-1692. REFERENCES transplantation, and NIV may help 23. Brochard L, Mancebo J, Wysocki M, et al. Non- achieve that goal. 1. Mermel LA, Maki DG. Bacterial pneumonia in solid invasive ventilation for acute exacerbations of chronic organ transplantation. Semin Respir Infect. 1990;5: obstructive pulmonary disease. N Engl J Med. 1995; In this study, 3 of the 4 patients with 10-29. 333:817-822. cystic fibrosis and who received lung 2. Meduri GU. Noninvasive ventilation. In: Marini J, 24. Pennock BE, Kaplan PD, Carlin BW, et al. Pres- transplantation required endotracheal Slutsky A, eds. Physiological Basis of Ventilatory Sup- sure support ventilation with a simplified ventilatory port: A Series on Lung Biology in Health and Dis- support system administered with a nasal mask in pa- intubation, and 2 died of pneumonia (1 ease. New York, NY: Marcel Dekker; 1998:921-998. tients with respiratory failure. Chest. 1991;100:1371- patient from each group). In patients 3. Ambrosino N, Rubini F, Callegari G, Nava S, Frac- 1376. chia C, Rampulla C. Noninvasive mechanical ventila- 25. Meduri GU, Conoscenti CC, Menashe P, Nair S. with cystic fibrosis who are heavily colo- tion in the treatment of acute respiratory failure due Noninvasive face mask ventilation in patients with acute nized with P aeruginosa, endotracheal to infectious complications of lung transplantation. respiratory failure. Chest. 1989;95:865-870. Monaldi Arch Chest Dis. 1994;49:311-314. 26. Rocker GM, Mackenzie MG, Williams B, Logan intubation with conventional ventila- 4. Kilger E, Briegel J, Haller M, et al. Noninvasive ven- PM. Noninvasive positive pressure ventilation. Chest. tion is frequently associated with dis- tilation after lung transplantation. Med Klin. 1995; 1999;115:173-177. 90:26-28. 27. Nourdine K, Combes P, Carton MJ, et al. Does semination of the pulmonary infec- 5. Rubini F, Nava S, Callegari G, Fracchia C, Ambro- noninvasive ventilation reduce the ICU nosocomial in- 28 tion and development of septic shock. sino N. Nasal pressure ventilation (NPSV) in a case of fection risk? Intensive Care Med. 1999;25:567-573. Several reports have described the suc- Pneumocystis carinii pneumonia in single lung trans- 28. Swami A, Evans TW, Morgan CJ, et al. Conven- plantation. Minerva Anestsiol. 1994;60:139-142. tional ventilation as bridge to heart-lung transplanta- cessful implementation of NIV as a 6. Antonelli M, Conti G, Rocco M, et al. A compari- tion in cystic fibrosis. Eur Respir J. 1991;4(suppl):188. bridge to transplantation in patients son of noninvasive positive-pressure ventilation and 29. Quaranta AJ, Roy B, O’Brien G, et al. Noninva- 29,30 conventional mechanical ventilation in patients with sive positive pressure ventilation as a bridge therapy with cystic fibrosis. In one study, acute respiratory failure. N Engl J Med. 1998;339: to lung transplantation or volume reduction surgery. duration of intubation and ICU stay after 429-435. Am J Respir Crit Care Med. 1996;153:A608. 7. Goris RJ, te Boekhorst TP, Nuytinck JK, et al. Mul- 30. Hodson ME, Madden BP, Steven MH, et al. Non- transplantation were much shorter in tiple-organ failure. Arch Surg. 1985;120:1109-1115. invasive mechanical ventilation for cystic fibrosis pa- cystic fibrosis patients supported pre- 8. Le Gall JR, Loirat P, Alperovitch A, et al. A simpli- tients. Eur Respir J. 1991;4:524-527.

Downloaded From: https://jamanetwork.com/ on 10/01/2021