|
Indian Journal of Critical Care Medicine, Vol. 9, No. 4, October-December, 2005, pp. 217-224 Original Article Noninvasive ventilation for hypercapnic respiratory failure in COPD and initial post-support deterioration of pH and PaCO2 may not predict failure Mani Raj Kumar Delhi Heart and Lung Institute, Indraprastha Apollo
Hospitals, Sarita Vihar, New Delhi 110 044 Abstract Objectives: To correlate the degree of encephalopathy, baseline values of PaCO2 and pH, and their early response to NIV with eventual in-hospital outcome in patients of severe acute-on-chronic hypercapnic respiratory failure in COPD.Design: Retrospective review. Setting: Intensive care unit. Material and methods: 24 episodes of acute exacerbation of COPD in 17 patients (10 females, 7 males) with a mean age of 59.5 years (range 48 - 82) where NIV was initiated. Data collected: encephalopathy score at baseline and at 24 hours, respiratory rate, breathing pattern, serial arterial blood gases, duration of NIV support per day and hospital days. Results: All patients had severe hypercapnia (mean peak PaCO2 89.0 mm Hg ± 21; range 66-143), respiratory acidosis (mean nadir pH 7.24 ± 0.058, range 7.14 - 7.33) and tachypnoea (mean respiratory rate 29.5 ± 4.69/mt; range 24 - 40). In 17 episodes, altered mental state was present (encephalopathy score 1.92 ± 1.32, median 2.5). Clinically stable condition occurred over several days (mean 13± 9.6 days; range 5 - 40). Intubation was avoided in 22 out of 24 episodes (91.6%) despite significant initial worsening of PaCO2 and pH. Two patients died. The mean time on NIV was16.5 hours/day (range 4 - 22). Conclusions: In selected patients of COPD with acute hypercapnic failure on NIV worsening PaCO2 and pH in the initial hours may not predict failure provided the level of consciousness and respiratory distress improve. Keywords: Positive pressure ventilation, Noninvasive ventilation, Chronic obstructive pulmonary disease, Respiratory failure, Hypercapnia, Mechanical ventilation Introduction It is now well established that most patients of COPD with respiratory failure due to acute exacerbation should receive a trial of NIV.[1] Recent metaanalyses of 8 and 15 randomized controlled studies respectively[2],[3] showed that, compared with usual care, adjunctive NIV therapy was associated with improvements in mortality, need for intubation, values of pH, PaCO2 and respiratory rate at 1 hr., complication rates and length of hospital stay. While early use of NIV is recommended, concerns exist that delay in detecting failure may increase mortality.[4],[5],[6] Current opinion favors switching to endotracheal mechanical ventilation (ETMV) in the event of deteriorating pH, PaCO2 and persistence of altered sensorium despite initial use of NIV.[7],[8],[9] Review of our early experience with NIV in India was prompted by current recommendations in literature at variance with our observations. In the last decade, when the full scope of this form of support was relatively unexplored, we found remarkable success with severe exacerbations of COPD where there was no prompt initial response to NIV. The threshold for intubation in our group of patients was higher than in current recommendations. Our later practice was conditioned by key publications in subsequent years that cautioned against delaying intubation and by our growing experience with more aggressive support. This account of our early experience is to stimulate further exploration of predictive factors relating to success with NIV in subsets of COPD patients presenting with severe respiratory acidosis. Current Indian experience on the use of NIV in COPD patients is limited to a single abstract.[10] Materials and Methods Patients of COPD managed by one team who received NIV between August 1994 and July 1996 at Batra Hospital and Medical Research Centre, New Delhi, India. Care was initiated in the intensive care unit with NIV at admission within a few hours of admission. Standard care included oxygen supplementation, corticosteroids, bronchodilators, antibiotics, rehydration and correction of electrolyte disturbances. Patients included Patients excluded Ventilators and interface Mode and settings Duration of support Intubation criteria Measurements The mental state was scored from 0 - 4 (modified Parsons-Smith criteria) (11) for encephalopathy: grade 0 - no abnormality; grade I - trivial lack of awareness, anxiety, shortened attention span; grade 2 - lethargy, tremor, asterixis, disorientation, personality change, inappropriate behaviour; grade 3 - somnolence to semistupor, responsive to stimuli, confused, gross disorientation and agitation; grade 4 - coma. Blood gases were measured at 1 hour and 3 - 8 hours after application of NIV and every morning thereafter, or if there was clinical deterioration. Both baseline PaCO2 and pH values and their worst values in the clinical course were recorded. Oxygen supplementation was given to all patients starting at 1-2L/mt adjusting flow to maintain SPO2 of 88-92%. Biochemical tests conducted at admission and thereafter periodically included: blood sugar, renal profile, electrolytes and liver function tests. Chest X-rays were performed daily in the initial unstable period. Success or Failure of NIV Statistical analysis Results During the study period, a total of 44 patients admitted under one team were given NIV support either in the medical intensive care unit or wards depending on illness severity [Table - 2]. The study group included 10 women and 7 men with a mean age of 59.5 years (range 48 - 82). All patients had severe hypercapnia: mean peak (i.e., the worst) PaCO2 89 ± 21 mm Hg (range 66 - 143) with severe respiratory acidosis: mean nadir (i.e., the worst) pH 7.24 ± 0.058 (range 7.14 - 7.33). All had tachypnoea - mean respiratory rate of 29.5 ± 4.69/mt (range 24 - 40). In 17 episodes, there was altered mental status with an Encephalopathy score of grade 3 in 13 episodes and grade 2 in 4 episodes [Table - 1]. The mean encephalopathy score at admission was 1.92 ± 1.32 (median 2.5). The mean time on NIV was 16.5 hrs/day (range 4-22). Clinical stability occurred over several days: mean 13 ± 9.6 days (range 5-40). After NIV, there was initially a rise in the PaCO2 levels in all patients. The mean baseline (immediately before NIV) PaCO2 levels, was 64.7± 14.2 mm Hg (range 46.5 - 98.6). Post NIV PaCO2 rose to peak levels of 89.0 ± 21 mm Hg (range 66 - 143). There was significant difference between the baseline and peak PCO2 levels (20.7 ± 18.4 mm Hg P < 0.001) [Figure - 1]. Correspondingly, the pH fell from a mean baseline value of 7.330 ± 0.96 to a mean nadir level of 7.24 ± 0.58 (P< 0.001) [Figure - 2]. The mean IPAP was 15.43 ± 0.62 and mean EPAP was 5.81± 0.62. NIV was successful in 22 out of 24 episodes (91.67%). The final PaCO2 levels at discharge in these 22 episodes decreased significantly to a mean level of 56.16 ± 9 mm Hg as compared to the peak PaCO2 levels (P< 0.001). The final pH rose significantly to 7.375 ± 0.06, as compared to nadir values (P< 0.001). The mean encephalopathy score after 24 hours was 0.70 ± 0.88 with a median of 0. The difference between mean baseline and 24-hour encephalopathy scores was significant (P = 0.001). The respiratory rate was significantly reduced from 29.5 ± 4.69/mt at baseline to 22.3 ± 2.1/mt after NIV (P< 0.001). In 6 of the episodes where the peak PaCO2 levels were 100 mm Hg or more NIV was successful and the patients could be discharged to nocturnal home NIV support [Figure - 2] and [Table - 3]. Two patients out of the 17 died, one after intubation and mechanical ventilation for 15 days following a trial of NIV for 2 days, and the other while receiving NIV, as she refused intubation. No correlation was found between initial pH and PaCO2 with the degree of encephalopathy or with response to NIV. Few complications occurred in this series. One patient who was indeed the first case to be supported with NIV developed a deep ulcer on the bridge of the nose. Notably this patient was ventilated with a volume ventilator and a less sophisticated mask than presently available. One patient had gastric distension that settled with reduction in pressures and time on NIV support. Discussion NIV has been studied extensively in the context of acute exacerbation of COPD and several RCTs have proved its efficacy.[2],[3] NIV has not been universally successful, with reported failure rates of 7-50% due mainly to the heterogeneity of the study population.[3] There are concerns about incorrect selection of patients leading to delay in intubation.[4],[5],[6] Understanding the determinants of success would facilitate accurate patient selection, determine the setting for its application and the timing of intubation when the risk of failure is high.[3],[7],[9],[13] This review of our earliest experience with NIV focuses on the criteria that are said to be predictive of outcome. The severity of respiratory acidosis at the outset is reported to be predictive of NIV failure. Ambrosino et al[4] in a retrospective review of COPD patients found that lower baseline PaCO2 values (79 vs. 98) and higher pH values (7.28 vs. 7.22) correlated with success. This was confirmed in other prospective randomized studies,[12],[13] but not in all.[14],[15],[16] Hypercapnia in this study (mean peak PaCO2 89 ± 21 mm Hg, range 66 - 143) is among the highest levels reported hitherto.[3],[16],[17],[18] The acuity of the exacerbations is also reflected in the presence of severe respiratory acidosis (mean nadir pH 7.24 ± 0.058, range 7.14 - 7.33). NIV was successful in 22 out of the 24 episodes (91.67%) testifying to its remarkable efficacy. Notably, in all of the 6 episodes with PaCO2 more than 100 mm Hg, NIV was successful. Previous reports have shown a success rate of 71-96%[9],[12],[15],[13],[19] in patients with lesser degrees of respiratory acidosis. The level of consciousness (LC) at admission has been used to predict outcome. Most studies[3] and guidelines[8],[20] excluded patients with altered sensorium due to theoretical concerns about the risk of aspiration. Baseline LC and pH values have been reported to correlate with success.[4],[9],[13],[16] In the study by Brochard et al ,[12] the encephalopathy score among patients in the NIV group who needed intubation was 1.9 ± 1.2. Based on data from 1,033 patients Confalonieri et al[9] have developed a risk prediction model in which GCS< 11 and pH < 7.25 correlate with the highest risk of failure. Contrary to this view in our series with a high rate of success, disturbed consciousness was present in 17 of the 24 episodes (mean ES 1.92, median 2.5). Several others have shown successful outcome in this group of patients. Benhamou et al[21] showed a relatively high success rate of 65% in elderly pati ents with drowsiness, agitation or coma. Intermittent negative pressure ventilation had a high rate of success in the presence of coma (GCS range 3-8).[22] Other studies[23],[24] have also found no impact of baseline sensorium on patients'response to NIV. More recently, Gonzalez-Diaz et al[18] showed that hypercapnic coma with GCS < 8 can be treated as successfully as awake patients with NIV. The largest subgroup in this series was also COPD in which the success rate of those with severe encephalopathy was 86%, similar to our results. Once initiated, the response to NIV may also indicate the chances of success. Studies suggest that this can be gauged early within the first 4 hours. In the study by Ambrosino et al[4] physicians intubated patients if the pH value remained < 7.35 after 1-2 hrs of NIV. However, after using logistic regression analysis only the baseline pH had a reliable predictive effect among several variables. Moreover, there was no control group of patients with different threshold values for intubation to substantiate this conclusion. Further, pneumonia was far more frequent among those who failed NIV trial as compared to those who succeeded. In our study, patients with worsening levels of PaCO2 and pH initially after NIV had successful outcomes when support was continued for more sustained periods. Patients of pneumonia were not selected for our study. In our series it was frequently observed that while the patients were on NIV, paradoxically, the PaCO2 levels rose initially, and this difference between the baseline and peak levels was highly significant (P< 0.001). Many prospective studies have demonstrated remarkable reduction of PaCO2 and pH levels within the first few hours[13],[15],[25],[26],[27] and the failure of this to occur has been regarded as a predictor of unsuccessful outcome.[4],[9],[13] Benhamou et al[21] noticed that the PaCO2 and pH levels remained unchanged during the first hour, but improved later. Meecham Jones et al[28] found that the most severely hypercapnic patients showed a rise in PaCO2 with the addition of nasal ventilation. In our series, all were severely hypercapnic, with possible preexisting chronic stable respiratory failure, showing similar initial worsening. We also found that the PaCO2 levels fluctuated considerably for a variable period (hours-days) before settling down to stable levels [Figure - 3]. The average period for clinical and biochemical stability was 13 ± 9.6 days (range 5 - 40). Despite the initial rise in PaCO2 often to alarming levels, we persevered with NIV because the patients were clinically better or without deteriorating sensorium. This is in contrast to studies discussed above where support was escalated to invasive ventilation when the expected biochemical response did not occur in the initial hours. Sensorium and dyspnoea in our study improved over several hours even while the blood gas levels were grossly abnormal. Attention to compliance with NIV or mask fit often produced a favorable response in those who had clinical deterioration after initial improvement. In all cases, the inspiratory pressure and the duration of NIV were increased to the maximum tolerated to improve ventilation. A minimum EPAP of 4cms H2O was applied in most episodes which would perhaps have reduced the chances of carbon-dioxide rebreathing, as a standard whisper swivel was used.[29] The higher success rate in our series is attributable to a more sustained and aggressive application of NIV together with a disregard for worsening PaCO2 and pH, if the clinical state in terms of sensorium and dyspnoea was stable or improving. This raises a possibility (hitherto not studied or defined accurately) that the trial period for NIV was far too short in the preceding studies and that longer trials can be safely attempted in patients of COPD. Possible mechanisms for the paradoxic initial rise in PaCO2 after NIV may be a combination of several factors including resetting of the respiratory drive to high PaCO2 levels in chronic stable hypercapnic failure,[30] loss of hypoxic drive after oxygenation, reduction of minute ventilation from NIV-induced resistive loading of the upper airway[31] and a rebound increase in REM sleep with reduction of alveolar ventilation.[32] The net effect would be a rise in the PaCO2, although the patient is better rested. Improvement in sensorium despite increasing hypercapnia suggests that other factors such as fatigue may also influence mental alertness. The decision to intubate patients of COPD is always a difficult one because of the high risk of complications, and the possibility of prolonged mechanical ventilation with attendant morbidity and mortality. In the Third World, given the high cost of intensive care and the scarcity of ventilators this decision is even more difficult. It is disappointing that out of those who can be weaned from invasive ventilation only 38% - 56% survive at the end of one year[33],[34] questioning the overall cost-effectiveness of prolonged mechanical ventilation. Often the family, when informed of the uncertain course and the expenses involved, is reluctant to consent to invasive ventilation. For these reasons, the decision for ventilatory support often poses a moral dilemma in those with advanced chronic disability who present with acute respiratory decompensation. The trial with NIV should therefore be aggressive and should not be prematurely abandoned, taking care that intubation, if considered an option, is not delayed. Experience with this series of patients suggests that longer trials of NIV can be safely attempted in a subset who may improve clinically while they may temporarily worsen in terms of biochemical parameters. Although the patients in this series were spared the problems of prolonged invasive mechanical ventilation, they, nevertheless required prolonged support with NIV. This, in contrast to invasive support was associated with few serious complications. With close monitoring, good mask fit, trained personnel and skilful handling of patients high success rates can be achieved. The main limitation of the study is that it is retrospective. However, given the well-proven efficacy of NIV in the setting of COPD patients a prospective, randomized study with controls supported invasively without prior trial of NIV cannot be ethically supported. It would appear that the current recommendations for predicting failure of NIV might not be accurate for an individual patient. The question of the optimal duration of trial can perhaps be addressed by a randomized controlled trial comparing different cut-off points for switching from noninvasive to invasive ventilation, rather than using pre-defined rigid and arbitrary criteria for intubation. Our observations certainly challenge the currently accepted boundaries set for NIV since a high proportion of this series of patients with altered sensorium and initially deteriorating biochemical parameters went on to have a favorable outcome. Conclusions In selected patients of COPD with severe hypercapnia and respiratory acidosis, NIV may be continued even if PaCO2 and pH deteriorate initially provided there is improvement in clinical status as reflected by sensorium and degree of respiratory distress. As there may be a an initial worsening of PaCO2 and pH without signifying failure of NIV, these values alone should not influence the decision to escalate treatment to invasive support. This decision may be more reliably based on clinical status parameters. The tolerable thresholds for NIV trial may be set too low and need further study.Acknowledgment I wish to express my gratitude to Dr. S. Bal, MS, DNB, FRCS, for his valuable comments and help with the manuscript and to Mrs. Sobhana. S for secretarial help.References
Copyright 2005 - Indian Journal of Critical Care Medicine The following images related to this document are available:Photo images[cm05034t2.jpg] [cm05034f2.jpg] [cm05034f1.jpg] [cm05034t1.jpg] [cm05034t3.jpg] [cm05034f3.jpg] |
|