Clinical Investigation

Rescue Therapy by Switching to Total Face Mask After Failure of Face Mask-Delivered Noninvasive Ventilation in Do-Not-Intubate Patients in Acute Respiratory Failure* Malcolm Lemyze, MD1; Jihad Mallat, MD1; Olivier Nigeon, MD2; Stéphanie Barrailler, MD1; Florent Pepy, MD1; Gaëlle Gasan, MD1; Nicolas Vangrunderbeeck, MD1; Philippe Grosset, MD2; Laurent Tronchon, MD1; Didier Thevenin, MD1

Objective:  To evaluate the impact of switching to total face mask in cases where face mask-delivered noninvasive mechanical ventilation has already failed in do-not-intubate patients in acute respiratory failure. Design and Setting:  Prospective observational study in an ICU and a respiratory stepdown unit over a 12-month study period. Intervention:  Switching to total face mask, which covers the entire face, when noninvasive mechanical ventilation using facial mask (oronasal mask) failed to reverse acute respiratory failure. Patients:  Seventy-four patients with a do-not-intubate order and treated by noninvasive mechanical ventilation for acute respiratory failure. Main Results: Failure of face mask-delivered noninvasive mechanical ventilation was associated with a three-fold increase in in-hospital mortality (36% vs. 10.5%; p = 0.009). Nevertheless, 23 out of 36 patients (64%) in whom face mask-delivered noninvasive mechanical ventilation failed to reverse acute respiratory failure and, therefore, switched to total face mask survived hospital discharge. Reasons for switching from facial mask to total face mask included refractory hypercapnic acute respiratory failure (n = 24, 66.7%), painful skin breakdown or facial mask intolerance (n = 11, 30%), and refractory hypoxemia (n = 1, 2.7%). In the 24 patients switched from facial mask to total face mask because of refractory hypercapnia, encephalopathy score (3 [3–4]

*See also p. 675.  epartment of Respiratory and Critical Care Medicine, Schaffner HosD pital, Lens, France. 2 Department of Emergency Medicine, Schaffner Hospital, Lens, France. The authors have not disclosed any potential conflicts of interest. For information regarding this article, E-mail: [email protected] 1

Copyright © 2013 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0b013e31826ab4af

Critical Care Medicine

vs. 2 [2–3]; p < 0.0001), Paco2 (87 ± 25 mm Hg vs. 70 ± 17 mm Hg; p < 0.0001), and pH (7.24 ± 0.1 vs. 7.32 ± 0.09; p < 0.0001) significantly improved after 2 hrs of total face mask-delivered noninvasive ventilation. Patients switched early to total face mask (in the first 12 hrs) developed less pressure sores (n = 5, 24% vs. n = 13, 87%; p = 0.0002), despite greater length of noninvasive mechanical ventilation within the first 48 hrs (44 hrs vs. 34 hrs; p = 0.05) and less protective dressings (n = 2, 9.5% vs. n = 8, 53.3%; p = 0.007). The optimal cutoff value for face maskdelivered noninvasive mechanical ventilation duration in predicting facial pressure sores was 11 hrs (area under the receiver operating characteristic curve, 0.86 ± 0.04; 95% confidence interval 0.76–0.93; p < 0.0001; sensitivity, 84%; specificity, 71%). Conclusion: In patients in hypercapnic acute respiratory failure, for whom escalation to intubation is deemed inappropriate, switching to total face mask can be proposed as a last resort therapy when face mask-delivered noninvasive mechanical ventilation has already failed to reverse acute respiratory failure. This strategy is particularly adapted to provide prolonged periods of continuous noninvasive mechanical ventilation while preventing facial pressure sores. (Crit Care Med 2013; 41:481–488) Key Words: acute respiratory failure; chronic obstructive pulmonary disease; do-not-intubate order; noninvasive ventilation; total face mask

N

oninvasive mechanical ventilation (NIV) is now considered as the cornerstone of the initial therapeutic management of patients admitted for acute on chronic respiratory failure (1). When appropriately indicated, NIV significantly reduces mortality, mean hospital stay, and intubation rate (2, 3). However, NIV may fail to reverse acute respiratory failure (ARF) in 30% to 40% of cases, leading to a two-fold increase in associated mortality (4–6). This is especially dramatic for patients who refused intubation, or for whom intubation is www.ccmjournal.org

481

Lemyze et al

deemed inappropriate (7–10). Given their disastrous outcome, do-not-intubate patients may be considered as poor candidates for NIV (9, 10). On the other hand, NIV may relieve air hunger, feeling of impending death, and may be proposed as the treatment of last resort to these patients in spite of their poor prognosis (7, 8, 11, 12). In this setting, all efforts should be provided to improve patient’s comfort. Despite careful precautions, painful facial pressure sores may rapidly occur when a facial mask (FM) is applied for long periods of time (13, 14). The type of interface is widely recognized as a key point that can greatly influence the way the patient tolerates NIV. Thus, the interface is a critical determinant of NIV success or failure (1, 15, 16). We hypothesized that changing the interface to a total FM— after failure of FM-delivered NIV (FM-NIV)—would result in a clinical and gasometrical improvement in do-not-intubate patients suffering from ARF. MATERIALS AND METHODS Patients During a 12-month period, a prospective observational study was conducted in the Department of Emergency and Critical Care Medicine of Schaffner Hospital to evaluate all patients with a do-not-intubate order and treated by NIV for ARF. The local ethics committee approved the study and signed informed consent was obtained from all the patients or next of kin. Clinical criteria defining ARF included tachypnoea (> 24 breaths per min in obstructive, > 30 per min in restrictive), signs of increased work of breathing, accessory muscle use, and abdominal paradoxical motion. Blood gas criteria for entry included either pH < 7.35, Paco2 > 45 mm Hg, or Pao2/Fio2 ratio < 200. As recommended, patients could meet either clinical or blood gas criteria to be eligible to avoid delay in the application of assisted ventilation (1). Patients were excluded when having respiratory or cardiac arrest, multiple organ failure, need for vasopressors, fixed upper airway obstruction, facial burns or trauma, recent facial or upper airway surgery, swallowing impairment, and vomiting. Unresponsiveness, agitation, and uncooperativeness were considered exclusion criteria unless the treating physician made the judgment that they were exclusively related to hypercarbic encephalopathy. According to ethical practice and standard of care of our hospital, all patients admitted for acute or chronic respiratory or cardiac failure were classified with a do-not-intubate order or not. This decision was made by the patient himself whenever possible, or by a multidisciplinary team including physicians and nurses caring for the patient when the latter does not have the capacity to make such decision. Clinicians involved in the decision process included at least an intensivist and either a pulmonologist or a cardiologist, who did not participate in the present study. Patients were classified as do-not-intubate when their physical disability and their underlying debilitating conditions made them poor candidates for intubation. The patient’s family was informed in a clear and loyal manner, and all efforts were provided to make them understand and adhere to the medical decision. In the prehospital setting, some patients with 482

www.ccmjournal.org

terminal illness may nevertheless have been intubated in case of ARF especially when advance directives were unknown. These patients may have been given a “do-not-reintubate” order during the weaning period following similar ethical assessment by a multidisciplinary team. When extubation was possible, NIV could be proposed to those patients or their families as a noninvasive form of life support to help them recover from ARF. Except for invasive interventions, therapeutic management of do-not-intubate patients was similar as the one of every patient admitted for ARF. All the patients were admitted either to the ICU or to the step down unit at least for the first 2 days. Once stabilized, they were transferred to a medical ward, where NIV could be pursued if necessary. The need for antibiotic treatment, corticosteroids, or diuretics was evaluated by the attending physician. All patients with obstructive respiratory disease were under bronchodilator and corticosteroid therapy. Patients did not receive sedation and/or analgesia before and during NIV. NIV Technique NIV was started as soon as the patient was admitted to the emergency room and was carried on in the ICU. According to our local protocol, the same turbine-driven portable ventilator (BiPAP Vision, Philips Respironics, Murrysville, PA) and FM (oronasal mask) were used in all patients. The patients were gently placed in the sitting position and received reassuring explanations about the technique. The BiPAP Vision was equipped with its specific single branch circuit including an intentional leak, which was calibrated before each use. The BiPAP vision ventilator delivers a bilevel positive pressure targeted mode with a special auto-adaptive triggering and cycling called Auto-Trak, which ensures optimum sensitivity despite changing breathing patterns or circuit leaks. Active humidification (MR 850, Fischer & Paykel Healthcare, Auckland, New Zealand) was incorporated among the circuit of the ventilator when using the FM. Ventilator settings were adjusted at the discretion of the attending clinician. NIV was continuously applied until a significant improvement of the patient occurred. If effective, gradual reduction of the duration of NIV could then be considered by the attending physician. Interface Characteristics and Criteria Defining Failure of FM-NIV All the patients began NIV with the same FM (PerformaTrak, Philips Respironics, Murrysville, PA). The FM covers the nose and mouth and includes a double silicon layer and a safety valve that opens in case of unexpected ventilator failure (Fig. 1). Mask size was chosen according to patient’s face dimensions, and straps were tightened in order to be the most comfortable as possible without excessive leaks around the soft silicone seal. Changing the interface during the first 12 hrs of NIV was considered as an “early switch”, whereas “late switch” referred to switching from FM to total FM after the 12th hr. The decision to change the FM to a total FM was left up to the attending physician providing that it was motivated by any of the following: 1) failure of FM-NIV to reverse clinical signs of ARF; 2) failure of FM-NIV to improve pulmonary gas exchange; and 3) major discomfort or severe skin damage with the FM that compromised February 2013 • Volume 41 • Number 2

Clinical Investigation

725, Radiometer, Copenhagen, Denmark) was calibrated several times a day. To ensure accurate comparison between the efficiency of the two interfaces, the same ventilator settings were used after changing the mask until arterial blood gases were performed 2 hrs later. The severity of hypercarbic encephalopathy was assessed according to Kelly and Matthay scale (23). NIV-associated adverse events recorded during the study included skin breakdown, conjunctivitis, nausea and gastric distention, claustrophobia, and patient’s discomfort. Whenever possible, the patient’s comfort was assessed using a 5-point score—1 = very uncomfortable; 2 = quite uncomfortable; 3 = acceptable; 4 = comfortable; 5 = very comfortable—as previously described (13). Pressure sores of the face were detected and staged by the Figure 1.  A, The facial mask (Performatrack, Philips) used in the present study; B, Example of a patient switched late to total face mask (Performax, Philips) in order to pursue noninvasive mechaninurse in charge according to the National cal ventilation (NIV), after prolonged facial mask-delivered NIV had caused painful pressure sore of Pressure Ulcer Advisory Panel guidelines the nasal bridge. (24). Briefly, a redness lasting longer than 30 mins after removal of the mask was considered as a skin breakdown stage 1, an intact or open/ the pursuit of NIV. Failure of FM-NIV to reverse clinical signs of ARF was defined as no improvement or increase in the score for ruptured serum-filled blister or an excoriation was classified encephalopathy (17) or in respiratory rate. Failure of FM-NIV to as stage 2, whereas ulcer and skin necrosis were respectively improve pulmonary gas exchange was defined as no improvement classified stages 3 and 4. They were specifically located on the nasal bridge, the nasolabial folds, the cheeks, the chin, the or fall in pH or Pao2/Fio2, no improvement or increase in Paco2. forehead, or the occipital protuberance, where the mask and its In that case, total FM (PerforMax, Philips Respironics) was comfortably placed on the patient’s face with the bottom of the straps might exert the maximal pressure. cushion sitting below the lower lip and above the chin, and the top of the cushion resting above the eyebrows in the middle of Statistical Analysis the front head (Fig. 1). The straps were adjusted to eliminate Data are presented as mean ± sd unless otherwise stated. Proporleaks without overtightening the headgear. The top headgear tions were used as descriptive statistics for categorical variables. strap was correctly fit to prevent the mask from falling along The normality of data distribution was assessed using the Kolthe patient’s face. We chose not to use active humidifier with mogorov–Smirnov test. Continuous data that were normally total FM to prevent the glass of the mask from being quickly distributed were compared using Student’s t test; otherwise, the covered with water (fog effect). Heat-and-moisture exchanger Mann–Whitney U test was applied. Analysis of the discrete data was not chosen because it can negatively impact on physiologi- was performed by chi-square test or Fisher’s exact test when the cal variables during NIV (18, 19). numbers were small. Multiple longitudinal comparisons were made by repeated measures analysis of variance with the linear Evaluation Criteria mixed model. The fixed effect was the time, and the random Anthropometric data, smoking history (pack years), risk of effect was the patient. The Bonferroni method was used to addeath assessed by the Simplified Acute Physiologic Score II (20), just for multiple comparisons. The comparison of the comfort and duration of NIV were recorded. Severity of respiratory score with total FM and FM was also performed with the linear disease was evaluated using Medical Research Council dyspnea mixed model in order to take into account patients having a score (21) 2 wk before admission by patient’s or family’s recall. missing data for a measure. Due to a non-normally distributed Preadmission health status was assessed by Knaus index (17) parameter, this analysis was done on the ranks according to 2 wk before admission and by Charlson’s comorbidity score Conover method. (25) The ability of the duration of ventilation (22). The underlying lung disease was classified as obstructive, under facial mask to predict the development of facial pressure restrictive, or mixed, based on medical history, chest radiograph, sores was determined by the area under the receiver operating and/or spirometric data if available. Arterial blood gases were characteristic curves (area under curve) with a 95% confidence measured before initiation of FM-NIV, and repeated after 1 to 2 interval. Area under curve > 0.8 indicates good diagnostic achrs, 6 to 12 hrs, 24 hrs, 48 hrs, and 2 hrs after every subsequent curacy. The cutoff that best discriminated the development of facial pressure was chosen to maximize both the sensitivity and change in the interface. Note that the blood gas analyzer (ABL Critical Care Medicine

www.ccmjournal.org

483

Lemyze et al

the specificity. Statistical analyses were performed using SPSS (SPSS for windows release 17.0, Chicago, IL) and SAS software (SAS Institute, Cary, NC 25513; version 9.2). A p value < 0.05 was considered statistically significant for single comparisons. All reported p values are two-sided.

RESULTS During the 12-month study period, 573 patients received NIV for ARF either in the ICU or in the stepdown respiratory unit of Schaffner Hospital. Among them, 74 (13%) patients were classified with a do-not-intubate order. Their main characteristics at admission are displayed in Table 1. As shown in Figure 2, the interface was changed because of failure of FM-NIV (“FM failure” group) in 36 (48.6%) patients. These patients were most likely to have hypercapnic ARF, restrictive respiratory disorders, greater Paco2, and higher encephalopathy score at admission. A significantly larger number of patients died during their hospital stay in the “FM failure” group compared those of “FM success” group (n = 13/36, 36% vs. n = 4/38, 10.5%; p = 0.009). Nevertheless, 23 (64%) out of 36 patients in whom FMNIV failed to reverse ARF and, therefore, switched to total FM survived hospital discharge. Patients of the “FM failure” group, compared to those of the “FM success” group, required longer time of NIV during the first 48 hrs (40 [30–45] hrs vs. 22 [15– 28] hrs; p < 0.001, respectively) and a longer stay in critical care setting (8 [5–0] days vs. 5 [3–8] days; p = 0.007, respectively), and needed higher inspiratory positive airway pressure levels (18 [16–20] cm H2O vs. 16 [16–18] cm H2O; p = 0.003, respectively) and higher expiratory positive airway pressure levels (8 [6–8] cm H2O vs. 6 [6–8] cm H2O; p = 0.01, respectively). Reasons for Changing the Mask Among the patients who have been switched from FM to total FM, the change of the mask occurred within the first 12 hrs in 21 (58.3%) of them (switched early), whereas 15 (41.7%) patients were switched later (switched late; Fig. 2). The reasons for changing the mask are shown in Table 2. Refractory Hypercapnia Two thirds of the patients were switched to total FM because of persistent hypercapnia, most of them being switched early. Regarding these patients, changing the interface resulted in a significant decrease in Paco2 (87 ± 25 mm Hg vs. 70 ± 17 mm Hg; p < 0.0001) with a concomitant increase in pH (7.24 ± 0.1 vs. 7.32 ± 0.09; p < 0.0001) and a reduction of encephalopathy score (3 [3–4] vs. 2 [2–3]; p < 0.0001) after 2 hrs of total FM-NIV (Fig. 3). Nine out of these 24 patients eventually died, whereas 15 survived hospital discharge. Pressure Sores The second cause for changing the interface was facial skin breakdown and FM intolerance. In that case, a large majority of the patients were switched later to total FM (Table 2). Three out of these 11 patients died, whereas eight survived hospital discharge. Patients switched early to total FM were less likely to develop facial pressure sores, in spite of greater duration of NIV 484

www.ccmjournal.org

and less protective dressings use (Table 3). The risk of developing facial pressure sores was closely related to the duration of FM-NIV. Receiver operating characteristic curve analysis was used to determine if the duration of ventilation with FM could predict the development of facial pressure sores. The area under the curve for the length of stay under FM-NIV was statistically significant (n = 74, 0.86 ± 0.043; 95% confidence interval 0.76–0.93). The optimal cutoff value for the duration of FMNIV to predict facial pressure sores was 11 hrs with a sensitivity of 84% (95% confidence interval 69–93) and specificity of 71% (95% confidence interval 52–86). The positive likelihood ratio and negative likelihood ratio for this cutoff were 2.9 and 0.23, respectively. Switching the interface from FM to total FM was required in five out of six patients who developed stage 3 pressure ulcer of the nasal bridge in order to continue NIV (Fig. 1). Comfort Among the 11 patients who have been switched to total FM because of pressure sores or FM intolerance, comfort could be evaluated in only seven of them because two patients died and the two others were unfit for interrogation. In this small number of patients, the comfort score was higher with total FM than with FM although it did not reach statistical significance (4 [3–5] vs. 2 [1–3]; p = 0.08, respectively). Only one patient found the total FM less comfortable, but it could not be adjusted to the patient’s face because, even in its larger size, it was too small. Regarding all patients switched to total FM and fit for interrogation, total FM was found significantly more comfortable (4 [3–5] vs. 3 [2–3]; p = 0.004). Other Complications Except for pressure sores, other complications were either equally distributed between patients switched to total FM or not, that is, gastric distention (n = 7, 18.4% vs. n = 8, 22.2%; p = 0.7, respectively), or insignificant, as no conjunctivitis or claustrophobia occurred in both groups.

DISCUSSION The main findings of the present study can be summarized as follows: 1. When FM-NIV failed to reverse hypercapnic ARF in donot-intubate patients, switching from FM to total FM can be proposed to prolong continuous application of NIV. In spite of their poor prognosis, a substantial number of these patients survived hospital discharge. 2. This strategy resulted in better correction of arterial blood gases. 3. The risk of developing facial pressure sores was closely related to the duration of FM-NIV. Patients switched early to total FM developed less facial pressure sores compared to patients ­managed with FM for long periods of time. 4. In case of painful FM-induced facial skin damage switching to total FM enabled the continuation of NIV with better patient comfort. February 2013 • Volume 41 • Number 2

Clinical Investigation

Table 1.  Patients

Characteristics at Admission Overall Population (n = 74)

Facial Mask Success (n = 38)

Age (yr)

75 [64–80]

78 [67–80]

74 [62–79]

NS

Male, n (%)

52 (70.3%)

30 (79%)

22 (61%)

NS

Body mass index

27 [22–35]

27.8 [22–35.6]

27 [22–35]

NS

26.5 [23–29.7]

26.2 [23–29.4]

27 [23.2–29.8]

NS

49 (66.2%)

21 (55.3%)

28 (78%)

0.04

8 (10.8%)

4 (10.5%)

4 (11%)

NS

Parameters

Albumin (g/L)

Facial Mask Failure (n = 36)

p

Noninvasive ventilation indication, n (%)   Initial hypercapnic ARF   Initial hypoxemic ARF   Postextubation ARF

17 (23%)

13 (34%)

4 (11%)

0.02

 Obstructive

29 (39.2%)

16 (42%)

13 (36%)

NS

 Restrictive

19 (25.7%)

6 (16%)

13 (36%)

0.04

 Mixed

26 (35%)

16 (42%)

10 (28%)

NS

Chronic respiratory failure, n (%)

58 (78.4%)

30 (79%)

28 (78%)

NS

  Exacerbation of chronic respiratory failure

30 (40.5%)

17 (44.7%)

13 (36%)

NS

 Pneumonia

29 (39.2%)

11 (29%)

18 (50%)

NS

  Cardiogenic pulmonary edema

15 (20.3%)

10 (26.3%)

5 (14%)

NS

5 [4.2–5]

5 [4–5]

5 [4.2–5]

NS

Knaus index C, n (%)

24 (32.4%)

15 (39.5%)

9 (25%)

NS

Knaus index D, n (%)

48 (64.9%)

23 (60.5%)

25 (69.4%)

NS

4 [3–5]

4 [3–5]

3 [2–5]

NS

43 [34–53]

NS

Type of chronic lung disease, n (%)

Cause of ARF, n (%)

Medical Research Council Dyspnea Score

Charlson’s Comorbidity Score Simplified Acute Physiologic Score 2

40 [35–49]

39 [35–47.5]

Respiratory rate (breaths/min)

29 ± 8

28 ± 7

Encephalopathy score

3 [2–4]

2 [2–3]

Systolic arterial blood pressure (mm Hg)

132 ± 19

130 ± 19

30 ± 9 4 [3–4]

NS 12 hrs of acidosis after the change of the interface, but we cannot rule FM-NIV within the first 2 days developed facial pressure sores. out this hypothesis. Secondly, we have not continuously and Protective dressings on the nasal bridge and nasolabial folds reliably monitored leaks around the mask. The latter may promay be proposed to significantly reduce facial pressure sores in patients who required prolonged FM-NIV. Nevertheless, about mote patient–ventilator asynchrony and may be a source of half of patients still develop nasal skin breakdown though (14). patient’s discomfort (15). However, our nursing team is experienced to regularly check this point, by placing a hand around The present study confirms this finding by reporting a high Table 3.  Comparison

of Patients Switched Early to Total Face Mask With Those Switched Late Regarding Facial Pressure Sores and Length of NIV Switched Early to Total Switched Late to Total Face Mask Face Mask (n = 21) (n = 15)

Parameters

p

Pressure sore, n (%)

5 (24%)

13 (87%)

0.0002

Pressure sore score

0 [0–0]

1 [1–3]