Case Presentation

Hospital Course

Introduction

Physiology

Patient Selection

Application of NPPV

Gathering the Evidence

Lumpers

Splitters

Conclusion

Noninvasive Ventilation in Acute Respiratory Failure

Resident Grand Rounds

Case Presentation

A 28 year old white female at 22 weeks gestation presents to the MICU in transfer from an outside hospital with progressive shortness of breath. Seven days prior to admission, she had presented to her PCP with fever to 102, cough and dyspnea. She was treated with erythromycin for community-acquired pneumonia. The patient returned three days later and was admitted to the outside hospital with progressive shortness of breath. Despite IV antibiotics, she developed worsening infiltrates and hypoxemia resulting in her transfer to NCBH. Upon arrival she was noted to be anxious and dyspneic. She complained of fatigue, wheezing, and cough productive of purulent sputum. ROS was otherwise negative and the pregnancy had thus far been uncomplicated.

Asthma/Bronchitis IV Cefuroxime 20 pack years

G1P0 - 22wks Albuterol Nebs No EtOH, drugs

NKDA IV Solumedrol No pets, travel

Temp: 101.8 Pulse: 128 BP: 155/90 RR: 40 Sats: 92% on 100% FM

General: Young pregnant WF in respiratory distress

HEENT: PERRL, membranes moist

Resp: Bilateral diffuse rales, scattered wheezes, accessory muscle use

CV: tachycardic with II/VI crescendo-decrescendo murmur at LUSB

Abd: paradoxical movement, active BS, uterus at umbilicus

Ext: pale, cool, moist, 2+ pulses

Neuro: A&O 4/4, nonfocal

WBC: 22.1, 89% segs, 5% bands. Hgb: 12.4. Plts: 485. SMAC WNL. UA WNL

ABG: 7.48/24/58/22/90% on 100% NRBFM. ECG: ST, otherwise normal

CXR: Bilateral dense infiltrates consistent with pneumonia

Shortly after her arrival to the MICU, preparations were made for elective endotracheal intubation for hypoxemic respiratory failure before she developed respiratory muscle fatigue. In the presence of an Ob/Gyn, Anesthesiologist and Pulmunologist, she was anesthetized, intubated and mechanically ventilated. She was continued on IV steroids and inhaled bronchodilators and placed on IV Vancomycin, Rocephin and Azithromycin for treatment of community-acquired pneumonia. She showed gradual improvement and was weaned and extubated five days later. She was discharged from the MICU on hospital day seven and did well thereafter.

Acute respiratory failure refers to a severe deterioration of alveolar gas exchange that may require mechanical ventilation for life support. Patients with acute respiratory failure may present with primary hypoxemic failure (Type 1), primary ventilatory failure (Type 2) or a combination of the two. Since the 1950's, patients failing medical management of acute respiratory failure (ARF) have required insertion of an endotracheal tube for mechanical intermittent positive pressure ventilation. Noninvasive positive-pressure ventilation (NPPV) refers to the delivery of intermittent positive pressure ventilation via a nasal or full face mask. Continuous positive airway pressure alone via a mask (mask CPAP) differs from NPPV in that continuous positive airway pressure is delivered without inspiratory assistance. This paper seeks to examine the evidence supporting the use of NPPV in patients with ARF. Although mask CPAP for the treatment of obstructive sleep apnea and chronic ventilatory failure lead to the development and study of NPPV in ARF, it is beyond the scope of this paper.

In acute respiratory failure, the precise mechanisms for how NPPV works is not known. It is assumed that NPPV improves alveolar gas exchange and symptoms by relieving respiratory muscle fatigue.¹ Brochard et al in 1990 found significant reduction in trans-diaphragmatic pressure, diaphragmatic pressure-time product and EMG activity (measures of inspiratory muscle activity) in patients treated with NPPV.² Another possible mechanism includes decreased work of breathing by overcoming the effect of intrinsic positive end-expiratory pressure (PEEP) on the inspiratory muscles. Intrinsic PEEP places a load on inspiratory muscles that must be overcome to cause inspiratory airflow. It has been shown that the application of extrinsic PEEP and pressure support ventilation can significantly reduce the inspiratory threshold load and pressure-time product in COPD patients.² Respiratory muscle function may also be improved by increased tissue oxygenation, especially in patients whose initial PaO₂ is less than 50 mm Hg prior to treatment.

Endotracheal intubation is associated with direct laryngeal trauma, loss of the ability to speak, discomfort, need for sedation, and an increased incidence of sinusitis and pneumonia.^3-4 NPPV seeks to avoid these complications. Theoretical advantages include increased patient comfort, the preservation of speech, swallowing and natural airway defense mechanisms, and the ease of starting and stopping mechanical ventilation.⁵ Disadvantages include the need for an alert and cooperative patient, lack of direct airway access for suctioning secretions, gastric distention and difficulties with mask fit resulting in air leaks and facial necrosis.¹ There has also been concern that patients receiving NPPV require more time and attention by nursing staff and respiratory therapists.

Patient selection for NPPV is critical to success. Patients should be alert and cooperative in order to synchronize with the ventilator. Patients who are somnolent secondary to CO₂ narcosis can be given a trial of NPPV in assist control mode since their mental status will usually improve after 30 minutes of effective NPPV. ⁵ Many physicians hesitate to use NPPV in ARF because the patients appear too anxious to tolerate a face mask. The anxiety experienced by many patitents with ARF resolves after their ventilation and oxygenation improves with 15-30 minute of NPPV.⁵

Patients who require airway protection from excessive secretions, emesis and patients in comatose states should not receive NPPV. It should not be used in patients with facial trauma. A properly fitting mask must be available. Patients with morbid obesity are not good candidates for NPPV because of the high pressures needed for inspiration in these patients.

Patients with hemodynamic instability have been excluded from clinical trials as they frequently have mental status changes or acute decompensation requiring intubation and thus are not good candidates for NPPV. Some feel that NPPV should not be applied in patients with acute myocardial infarction; this recommendation comes after a trial of NPPV in patients with cardiogenic pulmonary edema was halted because of excess mortality in the NPPV group. ⁶

NPPV is applied via either a nasal or full face mask, although there are no studies directly comparing the two. Most studies of NPPV have been conducted using nasal masks. Most patients find nasal masks more comfortable; they allow easier communication and clearance of vomitus or respiratory secretions and there is less dead space ventilation.

In patients who are not able to keep their mouth closed, positive airway pressure rapidly escapes through the mouth of patients wearing a nasal mask. These patients can either be fitted with a chip strap that keeps the mouth closed or a full face mask, whichever is most comfortable for the individual patient. Full face mask NPPV is preferable in patients with nasal airway deformities or severe dyspnea because of the inability to move adequate air through the nasal passages.

A mask with a transparent dome is preferable to allow visual inspection of the airway. The mask should be lightweight and have a soft, pliable seal for comfort and reduction in trauma and air leakage. Duoderm can be used to protect the nasal bridge and seal leaks. Aerophagia is reduced when peak airway pressures are less than 25 cm H₂O. If a gastric tube is required, either a full face mask with a NG tube port or an orogastric tube should be used to prevent air leaks.

Any mode of ventilation can be performed in NPPV although pressure support ventilation (PSV) is most commonly used in clinical trials. Pressure limited modes of ventilation are better tolerated because they minimize peak airway pressure and reduce air leak. In two trials comparing assist-control and PSV modes of ventilation, there was no difference in physiologic parameters but PSV was better tolerated.^7-8 Pressure control ventilation can be used in patients with initially low respiratory rates. Flow trigger is preferred to pressure trigger for all modes.

Many ventilators can be attached to masks to provide NPPV. These include standard hospital ventilators such as the Puritan BennettÒ 7200 and SeimensÒ Servo 300 and Servo 900 used at WFUBMC and a dedicated NPPV ventilator such as the RespironicsÒ BiPAPÒ ventilator. Care must be given as to whether settings are for EPAP (airway pressure on expiration) and IPAP (airway pressure on inspiration) or PEEP (airway pressure on expiration) and pressure support (positive inspiratory pressure in addition to PEEP). Special precautions must be made with some versions of the Puritan-Bennett^® 7200 ventilator. Respiratory pauses can occur if mask leak or mouth opening prevents system flow from falling below 5 L/min and triggering the end of a respiratory cycle.⁵

Monitoring of patients receiving NPPV should be similar to that of intubated patients. This includes continues pulse-oximetry, frequent ABG’s and close supervision by nursing and respiratory therapy (RT) staff. The first thirty to sixty minutes of NPPV are labor intensive and nursing or RT staff trained in manipulation of the mask and ventilator settings are required at the bedside. Patients must be watched closely for signs of NPPV failure. (See Table below) As with other forms of mechanical ventilation, the monitoring requirements decrease once the patients and settings are stabilized.

1. Inability of patient to tolerate the mask secondary to discomfort

2. Failure to improve gas exchange of symptoms of dyspnea

3. Need for endotracheal intubation to protect airway or clear secretions

4. Hemodynamic instability

1. Failure to improve mental status after 30 minutes of NPPV in patients with CO₂ narcosis

This review was sparked by an interest in the best evidence to support the use of NPPV in patients with acute respiratory failure. The main outcomes of interest were reduction in mortality, morbidity, hospital stay and costs. Avoidance of endotracheal intubation was a common primary end-point in most studies. However, it is a surrogate marker for the above more important outcomes. Avoidance of intubation as an endpoint assumes that intubation in all patients is associated with increased morbidity, mortality and costs. That assumption is based on data from case-control and cohort studies only and fraught with potential bias. With the exception of one meta-analysis, only randomized-controlled trials (RCT) of NPPV in patients various types of ARF that addressed the above outcomes were reviewed. The validity of these trials was assessed according to criteria from the Users’ Guides to the Medical Literature and details are discussed below. ^9-10

Most recently, Wood et al published a RCT of NPPV in 27 patients presenting to a university teaching hospital emergency department with acute respiratory distress.¹¹ Consecutive adult patients with clinical evidence of ARF, RR > 25 and one of the following: pH < 7.35, PaCO₂ > 45, PaO₂ < 55, O₂ Sat < 90% or A-a gradient > 100 on O₂ were enrolled. The usual exclusion criteria were used with the notable exclusion of patients who could not tolerate wearing a mask and patient who were DNR. All diagnoses of ARF were accepted. Patients were randomized to receive either conventional medical therapy (CMT) appropriate to their underlying disease or CMT plus continuous NPPV via nasal mask and BiPAP ventilator until their respiratory distress resolved. Main outcome measures were: the need for intubation (as established by predetermined objective criteria), hospital mortality, ICU length of stay, duration of mechanical ventilation, number of organ system failure, and staff utilization.

Twenty-seven out of 87 patients screened were enrolled in the study. There were no unusual reasons for patient exclusion. Sixteen were randomized to CMT and eleven to CMT plus NPPV. Patients in the NPPV group had more pneumonia (44% vs 18%) and higher predicted hospital mortality based on APACHE II scores (26% vs 19%). More patients in the control group had COPD (36% vs 12%). Although these numbers were not statistically significant because of the small study size, they are concerning.

There were no statistically significant differences between the two groups among any of the outcome measures except for an increased number of organ system failures in the NPPV group (average 1.8 vs 0.9 organ systems, p = 0.034). There was a trend towards increased mortality in the NPPV group (25% vs 0%) but this was not statistically significant (p = 0.123). Overall, the study was well designed except that the small study size lacked the power to detect clinically significant differences in the baseline characteristics or outcomes. In the authors defense, it was the first to look at patients presenting to the ED as opposed to the ICU and was meant to be a pilot study. There is no evidence in this study to support or deter the use of NPPV in patients with ARF presenting to the ED. Larger studies are needed in this patient population.

In 1995, Kramer et al conducted a randomized controlled trial of NPPV that showed a decreased need for endotracheal intubation.¹² This study screened consecutive non-intubated patients with respiratory distress (as defined by clinical symptoms and pH < 7.35, PaCO2 > 45 and RR > 24) who presented to two university medical center intensive care units in a one-year period. Patients were randomized to receive either standard medical therapy (SMT) for their diagnosis or SMT plus continuous positive pressure ventilation via a nasal mask and BiPAP® respirator. Patients were intubated and weaned according to pre-established guideline. The primary outcome measure was the need for endotracheal intubation. Secondary outcomes included physiologic parameters, Dyspnea Score, nursing and RT time, hospital length of stay, hospital charges, and mortality.

Thirty-one patients were enrolled. There is no data is given on the number screened or breakdown by exclusion criteria. Patients in each group were similar with the majority of patients 23/31 (74%) having the diagnosis of COPD. Patients in the control group had a significantly higher need for intubation than the NPPV group (73% vs. 31%, p < 0.05, ARR = 52%, NNT = 2). Most of this difference was seen in patients with COPD. NPPV was tolerated for 24 hours by 69% of patients and that number is consistent with earlier case series. Respiratory therapists tended to spend more time with patients on NPPV in the first eight hours (100 min. vs. 44 min., NS). This decreased in the second eight hour period such that the total time spent by RT in the first 16 hours was similar (134 min for NPPV vs 103 min for controls, NS). Nursing time and complexity was also similar between the two groups. Length of stay, ventilator time and hospital charges were very similar between the two groups. Mortality was not significantly different between the two groups (6.5% in NPPV vs 13.3% in SMT). This study was well designed and carried out except that they failed to give us a percentage of patients screened who were not enrolled. This study lacked the power to detect clinically important differences in mortality between the two groups. Therefore this study does not support the use of NPPV in patients with respiratory failure.

In 1997, a meta-analysis was performed by Keenan et al to evaluate the effect of NPPV on mortality in patients with ARF.¹³ This analysis looked at randomized, controlled trials of NPPV in ARF published prior to September, 1995. Strict validity criteria for inclusion were laid out and seven studies were located (four articles and three in abstract form). ^{12, 14-19} The authors of all selected trials were contacted to clarify data and asked if they new of other un-selected or unpublished trials. Of the trials, four included only COPD patients ^14,16-18, two included all types of ARF ^12,19 and one excluded patients with COPD ¹⁵. Only five trials published mortality data and were included in the analysis. ^12,14-17 Though possibly limited by publication bias, all studies showed a trend toward reduced mortality with NPPV. The meta-analysis reported a significant reduction in mortality in all patients (Odds ratio 0.29, 95% CI: 0.15-0.59). The overall mortality in the control group was 50.9% compared to 15.3% in the NPPV groups (ARR = 35.4%, NNT = 2.8). This meta-analysis was affected by the large percentage of COPD patients (86%) in the included trials making generalization to all patients with ARF difficult. Analysis of the three including only COPD patients also showed a significant reduction in mortality (ARR = 22.3%, NNT = 4.5). There were not enough non-COPD patients to conduct a significant analysis. This trial strongly supports the use of NPPV in patients with COPD and ARF. No conclusions can be made regarding patients with other forms of ARF.

It may be more than coincidence that most of the mortality benefit in the above trials was seen primarily with COPD patients. Patients with COPD present with primarily Type 2 (hypoventilatory) respiratory failure. This differs from the other more common causes of ARF such as pneumonia, CHF and ARDS in which type 2 failure typically come only secondary to fatigue from primary type 1 (hypoxemic) respiratory failure. This lead many to believe that COPD and non-COPD patients should separated in clinical trials of NPPV.

In 1995, Wysocki et al conducted a small randomized, controlled trial of NPPV in patients with ARF excluding COPD.¹⁵ They screened 116 consecutive nonintubated patients admitted to their university medical center with ARF (as defined by two of the following (pH < 7.38, PaO₂ on room air < 60 mm Hg, PCO₂ > 50, RR > 25). Sixty-four per cent of patients were excluded for the following pre-determined criteria: ARF secondary to COPD (40%), neurologic disease (21%), multi-organ system failure (21%), intubation for cardiac arrest (10%), status asthmaticus (3%) and intubation for a surgical procedure (3%). Twenty patients were randomized to standard therapy 21 to NPPV via a full-face mask and a Puritan-Bennett 7200 ventilator. The primary endpoint was the rate of endotracheal intubation. Secondary endpoints included length of ICU stay and ICU mortality.

There were no statistically significant differences in any endpoints between the two groups though the trend was for better outcomes in the NPPV group (See Table 1). Though this may represent a type 2 error secondary to the small study population, this study does not support the use of NPPV in patients with ARF not secondary to COPD. The post hoc analysis showed a significant improvement in all outcomes in patients with PaCO₂ > 45 mm Hg (See Table 2). Particular striking was the 55% absolute risk reduction in mortality in the NPPV group (66% vs. 9%, NNT = 1.8). There is a risk that the results of a post hoc analysis are due to chance, especially if multiple analyses were done. The authors speculate that NPPV may be of more benefit to patient with ventilatory failure than hypoxemic failure because of its ability to improve respiratory muscle function. It may be more than coincidence that NPPV again appears to work in patients with a type 2 component to their respiratory failure. This is consistent with the success of NPPV in COPD patients. Patients with hypercapnic respiratory failure not secondary to COPD are a subgroup that should be prospectively studied in the future.

Table 1

Table 2

Recently, Antonelli et al published a randomized controlled trial comparing NPPV to endotracheal intubation and mechanical ventilation in patients with hypoxemic respiratory failure.²⁰ This represents a departure from most studies which compared NPPV to medical therapy and then examined the rate of intubation in the two groups. By using this method, the authors hoped to increase the outcomes of importance (morbidity and mortality) and reach statistical significance with a smaller study size. For one year this study enrolled consecutive patients admitted to a university ICU with hypoxemic respiratory failure felt to require mechanical ventilation (RR > 35, PaO₂ : FiO₂ < 200 and accessory respiratory muscle use). Patients with ARF from COPD, neuromuscular disease, status asthmaticus or respiratory arrest; cardiopulmonary arrest, multi-organ system failure and facial deformities of recent facial surgery were excluded. Patients were randomized to either intubation and mechanical ventilation or noninvasive positive pressure ventilation via a full face mask and standard mechanical ventilators. Changes in ventilator settings and weaning were performed according to predefined protocols which are included in the published study. Primary endpoints were physiologic parameters, infections (especially sepsis, pneumonia and sinusitis) and other complications. Secondary endpoints were mortality, length of ICU stay and duration of mechanical ventilation.

A total of 486 patients were screened for the study, of whom 295 were intubated on arrival, 19 had tracheostomy and 95 had COPD. Of the 77 eligible patients, 13 refused to participate and thus 64 (83% of eligible patients) were randomized (32 in each group). Physiologic parameters were similar in the two groups at baseline and sixty minutes except that the conventional ventilation group had a statistically significant difference in pH (7.37 in the conventional group vs 7.45 in the NPPV group) and PaCO₂ (42 vs 38 mm Hg respectively). Ten patients (31%) in the NPPV group required intubation. There was a significant reduction in the complication rate in patients receiving NPPV (38% vs 66%, ARR = 28%, NNT = 3.6). Patients in the NPPV had a shorter average duration of stay in the ICU (9 +/- 7 day vs 16 +/- 17 days, p = 0.04). There was a trend towards decreased ICU mortality in the NPPV treated group but it was not statistically significant (47% vs 28%, ARR = 19%, NNT = 5.3, p = 0.19). The authors stated that post hoc analysis showed a significantly decreased mortality among patients in the NPPV group with Simplified Acute Physiology Scores (SAPS) than in the control group but that data was not published.

This was a much more promising study of NPPV in patients with respiratory failure. Although the mortality difference was not statistically significant, it was promising. The presence of critically ill patients may have prevented seeing a significant difference as these patients tend to do poorly regardless of intervention. The authors note about improved survival in patents with lower SAPS scores cannot be judged without the ability to view the data although it elucidates an area for further study. The reduction in ICU stay and complication rate and the trend towards improved survival in this study support the use of NPPV in patients with hypoxemic respiratory failure.

Based on the assumption that respiratory failure in COPD is clinically and physiologically unique from other types of respiratory failure, there is no convincing evidence that NPPV improves survival in patients with ARF. However, there is good evidence that it reduces ICU stay and complications in patients. No studies have shown evidence of harm in this highly selected patient population. Thus the use of NPPV can be recommended in selected patients with ARF not secondary to COPD as an alternative to endotracheal intubation and mechanical ventilation. Further large randomized controlled trials are needed to help define which patients are most likely to benefit from NPPV.