Moderate abnormality of gas exchange frequently

Transcription

1 Prediction of Hypoxemia and Mechanical Ventilation After Lung Resection for Cancer Marc Filaire, MD, Mario Bedu, MD, Adel Naamee, MD, Sylvie Aubreton, PhT, Laurent Vallet, MD, Bernadette Normand, PhD, and Georges Escande, MD Departments of Thoracic Surgery, Respiratory Physiology and Biostatistics, Gabriel Montpied Hospital, Clermond-Ferrand, France Background. Hypoxemia usually occurs after thoracotomy, and respiratory failure represents a major complication. Methods. To define predictive factors of postoperative hypoxemia and mechanical ventilation (MV), we prospectively studied 48 patients who had undergone lung resection. Preoperative data included, age, lung volume, force expiratory volume in one second (FEV1), predictive postoperative FEV1 (FEV1ppo), blood gases, diffusing capacity, and number of resected subsegments. Results. On postoperative day 1 or 2, hypoxemia was assessed by measurement of PaO 2 and alveolar-arterial oxygen tension difference (A-aDO2) in 35 nonventilated patients breathing room air. The other patients (5 lobectomies, 9 pneumonectomies) required MV for pulmonary or nonpulmonary complications. Using simple and multiple regression analysis, the best predictors of postoperative hypoxemia were FEV1ppo (r 0.74, p < 0.001) in lobectomy and tidal volume (r 0.67, p < 0.01) in pneumonectomy. Using discriminant analysis, FEV1ppo in lobectomy and tidal volume in pneumonectomy were also considered as the best predictive factors of MV for pulmonary complications. Conclusions. These results suggest that the degree of chronic obstructive pulmonary disease in lobectomy and impairment of preoperative breathing pattern in pneumonectomy are the main factors of respiratory failure after lung resection. (Ann Thorac Surg 1999;67:1460 5) 1999 by The Society of Thoracic Surgeons Moderate abnormality of gas exchange frequently occurs within the first 48 h after thoracic surgery and usually returns to normal after 1 week [1 3]. Severe hypoxemia requires mechanical ventilation (MV) if not adjusted by additional oxygen or if associated with severe alveolar hypoventilation. In case of MV, the mortality rate is high, varying from 25% to 80% [4 7]. After pulmonary resection, no predictive factors of hypoxemia have been established, and conflicting results have been reported on determinants of MV [5, 6, 8]. Our purpose was to investigate the predictive factors of early postoperative hypoxemia and MV further to pulmonary resection. Material and Methods Accepted for publication Nov 30, Address reprint requests to Dr Filaire, Service de Chirurgie Thoracique, Hôpital Gabriel Montpied, 30 Place Henri Dunant, Clermont- Ferrand, France; mfilaire@chu-clermontferrand.fr. From March 1995 to October 1995, 47 patients, among those admitted for diagnosed or suspected lung malignancy in our institution, were considered operable according to the existing guidelines for pulmonary resections [9]. Pulmonary assessment included complete history and physical examination, and was objectively quantified by pulmonary function tests. We especially regarded a PaCO 2 45 mm Hg, a diffusing capacity 50% of predicted, and a calculated predictive postoperative force expiratory volume in 1 sec of 33% of predicted as requirements before resection. An additional patient presenting a poor general status, as well as a predictive postoperative force expiratory volume in 1 sec (FEV1ppo) below 33% of the predicted value, underwent pneumonectomy due to an ineffective and infected lung. Our study group comprised these 48 patients. There were 42 men and 6 women with a mean age of years (range 37 to 74 years) and a mean weight of kg (range 44 to 108 kg). Right pneumonectomy was performed in 13 patients, left pneumonectomy in 9, lobectomy in 24, and bilobectomy in 2. Preoperative Pulmonary Assessment Spirometry, diffusing capacity (DLCO), and breathing pattern were measured within the preoperative week with the Transfer Test Morgan Type C (P.K Morgan Ltd, Chatham, Kent, England). Spirometric variables included vital capacity (VC), residual volume (RV), total lung capacity (TLC), RV/TLC, forced expiratory volume in 1 sec (FEV1), FEV1/VC, and functional residual capacity (FRC). Residual volume, FRC, and TLC were assessed with the helium dilution method. DLCO and corrected DLCO for ventilation by minute (DLCO/VE) were measured by the steady-state method. Tidal volume (Vt), respiratory rate (RR), and VE were calculated during a DLCO test over a 3-min period. Arterial blood gases (PaO 2, PaCO 2 ) were measured in patients breathing room air within 2 min after sampling (BGE Electrolytes Instrumentation Laboratory, Paris, France). RR and VE 1999 by The Society of Thoracic Surgeons /99/$20.00 Published by Elsevier Science Inc PII S (99)

2 Ann Thorac Surg FILAIRE ET AL 1999;67: HYPOXEMIA AFTER LUNG RESECTION 1461 were corrected for 1 min. Predictive postoperative FEV1 was calculated using the Nakahara s formula [10]: FEV1ppo FEV1 1 b n / 42 n, where n is the number of obstructed subsegments, and b is the total number of removed subsegments. The total number of pulmonary subsegments is 42, with the left upper lobe and lower one containing 10 each, the right upper lobe 6, the right middle lobe 4, and the right lower lobe 12. The n value was calculated from the findings of preoperative bronchofibroscopy. In the presence of a subsegment with an above 75% occlusion, the obstruction was regarded as total. With a subsegment stenosis between 50% and 75%, the occlusion was regarded as 0.5, and below 50% the mild stenosis was ignored. The alveolar arterial oxygen tension difference (A-aDO 2 ) was calculated using a simplified alveolar gas equation: A-aDO PaO 2 PaCO 2 /0.8. Postoperative Course Patients requiring MV for less than 48 h without reintubation were regarded as having uncomplicated courses. Patients requiring either MV for 48 h and longer or reintubation were considered as having complicated courses. In this case, postoperative complications liable to prolong MV were observed. Pulmonary complications were defined as pneumonia (temperature 38 C for 48 h, purulent sputum production, and infiltrate on chest roentgenogram), lobar atelectasis, pulmonary embolus, noncardiac pulmonary edema, ventilatory inefficiency, and tracheostomy. Nonpulmonary complications were defined as other complications whatever their origin. Assessment of Hypoxemia In patients with uncomplicated course, arterial blood gases were studied within the first 48 postoperative hours after supplementary oxygen had been stopped 10 min earlier. Additional oxygen was provided under prescription with a face mask during the first postoperative night. Considering that a decrease in PaO 2 is generally observed in a large majority of patients in the first postoperative days, we define hypoxemia as the PaO 2 or the A-aDO 2 status of the patients within the first 48 postoperative hours. Statistical Analysis Simple regression analysis and stepwise regression analysis were used to determine whether age or any preoperative parameters of lung function were predictive of postoperative PaO 2 and A-aDO 2. Comparison between lung function preoperative mean values of patients with uncomplicated course and patients with prolonged MV was carried out by an unpaired t test. Discriminant analysis using the PCSM statistical package (Delta Consultants, Meylan, France) was applied to determine whether lung function preoperative parameters were predictive of prolonged MV for pulmonary complications. Preoperative parameters with a p 0.1 at the unpaired t test were selected to be entered into the Table 1. Surgical Procedures and Postoperative Course Group No. Procedures Postoperative course L BL, 5 LUL, 10 Uneventful LLL, 6 RUL L2 4 1 BL, 2 RLL, 4 pneumonia 1 LUL P RP, 6 LP Uneventful P2 5 3 RP, 2 LP 2 pneumonia, 3 ventilatory inefficiency P3 4 3 RP, 1 LP 1 operative bleeding, 2 cardiac failure, 1 broncho-pleural fistula BL bilobectomy; L1 lobectomy patients with uncomplicated postoperative course; L2 lobectomy patients with prolonged MV for 48 h or more for pulmonary complications; LLL left lower lobectomy; LP left pneumonectomy; LUL left upper lobectomy; P1 pneumonectomy patients with uncomplicated postoperative course; P2 pneumonectomy patients with prolonged MV for 48 h or more for pulmonary complications; P3 pneumonectomy patients with prolonged MV for 48 h or more for nonpulmonary complications; RLL right lower lobectomy; RP right pneumonectomy; RUL right upper lobectomy. regression model. For the lobectomy group, the selected variables included age, VR/TLC, FEV1, FEV1ppo, %FEV1ppo, as well as the number of resected subsegments. Selective criteria for the pneumonectomy group included age, VC, %VC, RV/TLC, FEV1, %FEV1, Vt, RR, DLCO/VE, and FEV1ppo. A probability below 0.05 was accepted as statistically significant. Results Postoperative course was uneventful in 22 patients after lobectomy (Group L1) and in 13 patients after pneumonectomy (Group P1). Four lobectomy patients (Group L2) and 5 pneumonectomy patients (Group P2) required prolonged MV for pulmonary complications. Prolonged MV for nonpulmonary complications was necessary in 4 pneumonectomy patients (Group P3). Lobectomy patients did not develop any nonpulmonary complications. Details on postoperative course are shown in Table 1. Prediction of Postoperative Hypoxemia in Patients with Uncomplicated Courses Regression analysis revealed no significant correlation between preoperative lung function and postoperative PaO 2 (PaO 2 po) (63 9 mm Hg, range 45 to 83) or postoperative A-aDO 2 (A-aDO 2 po) (38 9 mm Hg, range 23 to 56) in all patients (lobectomy and pneumonectomy). Using simple regression analysis allowed us to assess that FEV1, %FEV1, FEV1/VC, FEV1ppo, and %FEV1ppo were predictive of PaO 2 po and A-aDO 2 po for lobectomy patients (Table 2). The first step of the regression analysis showed significant correlation between FEV1ppo and PaO 2 po (r 0.74, p 0.001) (Fig 1), and between %FEV1ppo and PaO2po (r 0.67, p 0.001). At the second step, the prediction of PaO 2 po was improved when %DLCO was associated with FEV1ppo (r 0.8; regression equation: PaO 2 po FEV1ppo 0.07 %DLCO 22.06) or with %FEV1ppo (r 0.77; regres-

3 1462 FILAIRE ET AL Ann Thorac Surg HYPOXEMIA AFTER LUNG RESECTION 1999;67: Table 2. Correlations Between Preoperative Variables and Postoperative Values of PaO 2 and A-aDO 2 in Lobectomy and Pneumonectomy Patients With Uncomplicated Course L1 (n 22) P1 (n 13) PO 2 po A-aDO 2 po PO 2 po A-aDO 2 po Age VC %VC RV %RV TLC %TLC RV/TLC FEV c 0.64 c %FEV b 0.56 b FEV1/VC 0.52 b 0.51 a FRC %FRC Vt a 0.71 b RR a 0.68 b VE a DLCO 0.50 a %DLCO DLCO/VE a PaO A-aDO FEV1ppo 0.74 c 0.70 c %FEV1ppo 0.67 c 0.62 b Resected Subsegments Fig 1. Postoperative PaO 2 (PaO 2 po) is plotted against FEV1ppo after lobectomy. (Triangles) Patients with uncomplicated course (group L1); (bars) patients with prolonged mechanical ventilation (MV) for pulmonary complications (group L2). Of the 4 patients with FEV1ppo 1,500 ml, 3 belong to group L2. The position of the complicated patients in the x-axis is not reflecting their PaO 2 po. Prediction of Mechanical Ventilation Using unpaired t test, the best predictors of prolonged MV for pulmonary complications were FEV1ppo, %FEV1ppo, and number of resected subsegments for the lobectomy patients ( p 0.01; Table 3) and VC, FEV1, Vt, DLCO/VE, FEV1ppo, and RR for pneumonectomy patients ( p 0.05; Table 4). Preoperative selected variables of lung function were also assessed using discriminant analysis. For lobectomy patients, FEV1ppo was the best a p 0.05; b p 0.01; c p A-aDO2 alveolar arterial oxygen tension difference; A-aDO 2 po postoperative A-aDO 2 ; DLCO diffusing capacity; FEV1 forced expiratory volume in 1 second; FEV1ppo predictive postoperative force expiratory volume in one second; FRC functional residual capacity; L1 lobectomy patients with uncomplicated postoperative course; P1 pneumonectomy patients with uncomplicated postoperative course; PaO2 arterial blood gas; PaO2po postoperative PaO 2 ; RR respiratory rate; RV residual volume; TLC total lung capacity; VC vital capacity; VE ventilation by minute; Vt tidal volume. sion equation: PaO 2 po 0.43 %FEV1ppo 1.06 %DLCO 18.17). For pneumonectomy patients, Vt and RR were predictive of PaO 2 po and A-aDO 2 po (Table 2). The first step of regression analysis showed that Vt was significantly correlated with either PaO 2 po (r 0.67, p 0.01; regression equation: PaO 2 po 0.05Vt 33.13) and A-aDO 2 po (r 0.71, p 0.01; regression equation: A-aDO 2 po 0.05RR 68.8) (Fig 2). Using stepwise regression analysis, prediction of PaO 2 po or A-aDO 2 po for either lobectomy or pneumonectomy patients could not be improved with others preoperative data. Fig 2. Postoperative A-aDO 2 (A-aDO 2 po) is plotted against tidal volume (Vt) after pneumonectomy. (Squares) Patients with uncomplicated course (group P1); (bars) patients with prolonged mechanical ventilation (MV) for pulmonary complications (group P2); (triangles) patients with prolonged mechanical ventilation (MV) for nonpulmonary complications (group P3). Of the 5 patients with Vt 550 ml, 4 belong to group P2. Patient no. 24 required reintubation on the fourth postoperative day. The position of the complicated patients on the x-axis does not reflect their A-aDO 2 po.

4 Ann Thorac Surg FILAIRE ET AL 1999;67: HYPOXEMIA AFTER LUNG RESECTION 1463 Table 3. Comparisons of Preoperative Data Between Groups L1 and L2 a L1 (n 22) L1 vs L2 L2 (n 4) Sex 19 M/3 F 3 M/1 F Age (years) NS Weight (Kg) NS 66 5 Height (cm) NS VC 3, (97) NS 3, (94) RV 2, (110) NS 2,940 1,200 (111) TLC 6,230 1,010 (99) NS 6,280 1,970 (99) RV/TLC NS FEV1 2, (86) NS 1, (75) FEV1/VC NS FRC 3, (103) NS 3,790 1,670 (111) Vt NS RR NS 18 6 VE NS DLCO (%) (108) NS (101) DLCO/VE NS ph NS PaO NS PaCO NS A-aDO NS FEV1ppo (%) 2, (70) p , (51) Resected p Subsegments a Data presented are mean standard deviation (% predictive value). AaDO2 alveolar arterial oxygen tension difference; DLCO diffusing capacity; FEV1 forced expiratory volume in 1 second; FEV1ppo predictive postoperative force expiratory volume in one second; FRC functional residual capacity; L1 lobectomy patients with uncomplicated course; L2 lobectomy patients with prolonged MV for pulmonary complications; NS nonsignificant; PaCO2 arterial blood gas; PaO2 arterial blood gas; RR respiratory rate; RV residual volume; TLC total lung capacity; VC vital capacity; VE ventilation by minute; Vt tidal volume. predictor (74% correct, p 0.05) at the first step of the analysis, and the number of resected subsegments further increased the prediction (85% correct, p 0.05) at the second step. For pneumonectomy patients, Vt was the best predictive factor (83%, p 0.05) at the first step of the analysis, and RR increased the prediction (89% correct, p 0.05) at the second step. Comment The results of this study indicate that for patients with uncomplicated postoperative course, postoperative fall in PaO 2 and A-aDO 2 can be preoperatively predicted by FEV1ppo for lobectomy and Vt and RR for pneumonectomy. These same preoperative data enabled us to identify patients who developed respiratory failure due to only pulmonary complications. Postoperative Hypoxemia Postoperative hypoxemia is common [2, 11, 12] and may be more severe between the first and third postoperative day [2]. Impaired gas exchange is mainly due to atelectasis, impairment of the chest wall mechanics, diaphragmatic dysfunction, and impaired ventilatory control [13, 14]. Patients who underwent thoracic surgery may be exposed to particular risk due to pulmonary resection, and because they generally have moderate emphysema. However, little is known about predictive factors of postoperative hypoxemia. In addition to major abdominal surgery, significant association was found between the mean preoperative overnight saturation and the nocturnal saturation until the fifth postoperative day, but this could not be predicted by routine preoperative spirometry [11, 12]. After thoracotomy without pulmonary resection, postoperative hypoxemia assessed by oxygen saturation failed to be predicted in 50% of cases [2]. To our knowledge, this issue has not been prospectively investigated after pulmonary resection. We have undertaken this study to identify the usual predictive factors of hypoxemia from routine pulmonary function tests. Most of our patients did not develop postoperative hypercapnia and consequently did not require mechanical ventilation for ventilatory failure. Therefore, we observed a wide range of postoperative PaO 2 values, reflecting differences in V/Q impairment among these patients. Globally, we did not find any predictive factors of hypoxemia, but considering lobectomy and pneumonectomy separately, results were more revealing. Using simple regression analysis for lobectomy, FEV1, %FEV1, FEV1/VC, FEV1ppo, and %FEV1ppo significantly correlated with hypoxemia. The resection extent was not a predictive factor, but FEV1ppo, which takes account of the number of resected segments, has proved to be a better predictive factor than FEV1. The stepwise regression analysis showed that FEV1ppo was the best predictive factor and that DLCO slightly improved the prediction. These results indicate that the degree of chronic obstructive pulmonary disease (COPD) is the main factor for the early decrease of postoperative PaO 2 and that the amount of resection is of little influence. We also confer to FEV1ppo a new interest to evaluate the risk for postlobectomy complications. After pneumonectomy, FEV1, FEV1ppo, FEV1/VC, and RV/TLC were not predictive of hypoxemia, thus suggesting that COPD degree cannot be considered as a reliable rationale of early postpneumonectomy hypoxemia. On the other hand, preoperative Vt and RR provided a valuable prediction of postpneumonectomy PaO 2 and A-aDO 2. No other factors were predictive of hypoxemia. We know that the amount of resection is correlated with the decrease in thoracopulmonary compliance [15], which is more significant after pneumonectomy than after lobectomy. Moreover, there is a relationship between decreased compliance and breathing pattern impairment resulting in rapid RR, small Vt, and high respiratory work [16]. Based on these results, it is likely that the breathing pattern impairment resulting from ventilatory mechanic changes is more damaging after pneumonectomy than after lobectomy. Such breathing pattern modifications are deleterious to gas exchange, increasing ventilation of the low perfusion lung area [17], dead space ventilation [18], and O 2 consumption. We

5 1464 FILAIRE ET AL Ann Thorac Surg HYPOXEMIA AFTER LUNG RESECTION 1999;67: Table 4. Preoperative Data of Pneumonectomy Patients for Groups P1, P2, and P3 a P1 (n 15) P2 vs P2 P2 (n 5) P2 vs P3 P3 (n 4) Sex 13 M/2 F 5 M/0 F 4 M/0 M Age (years) NS NS Weight (Kg) NS NS Height (cm) NS NS VC 4, (100) p , (78) NS 3, (91) RV 2,490 1,120 (109) NS 2,660 1,100 (108) NS 2, (93) TLC 6,660 1,660 (100) NS 5,770 1,270 (88) NS 5,780 1,060 (89) RV/TLC NS NS FEV1 2, (76) p , (58) NS 2, (77) FEV1/VC NS NS FRC 3,840 1,240 (105) NS 3,370 1,010 (96) NS 3, (100) Vt p NS RR p NS VE NS NS DLCO (104) NS (97) NS (121) DLCO/VE p NS ph p NS PaO NS NS PaCO NS NS A-aDO NS NS FEV1ppo 1, (42) p (33) p , (45) Resected NS NS Subsegments a Data presented are mean standard deviation (% predictive value). AaDO2 alveolar arterial oxygen tension difference; DLCO diffusing capacity; FEV1 forced expiratory volume in 1 second; FEV1ppo predictive postoperative force expiratory volume in one second; FRC functional residual capacity; NS nonsignificant; P1 pneumonectomy patients with uncomplicated course; P2 pneumonectomy patients with prolonged mechanical ventilation for pulmonary complications; P3 pneumonectomy patients with prolonged mechanical ventilation for nonpulmonary complications; PaO2 arterial blood O2; TLC total lung capacity; VC vital capacity; VE ventilation by minute; Vt tidal volume. think that these results, together with ours, suggest that the mechanism of hypoxemia is different for lobectomy and pneumonectomy with uncomplicated postoperative course. In the case of lobectomy, hypoxemia seems to be more influenced by the degree of COPD and by its deleterious effects on the ventilation-perfusion ratio than the breathing pattern impairment resulting from alteration in thoracopulmonary compliance. In contrast, after pneumonectomy, hypoxemia seems to be more influenced by the breathing pattern impairment than the ventilation-perfusion ratio impairment resulting from COPD of the remaining lung. Evaluation of PaO 2 for lobectomy and of A-aDO 2 for pneumonectomy provided a better prediction of hypoxemia. We do not have a scientific explanation for this result. We hypothesize that a postoperative increase in O 2 consumption or cardiac output decrease resulting in venous oxygen saturation reduction is probably more accentuated after pneumonectomy than after lobectomy. Age is considered as a risk factor of postoperative hypoxemia after abdominal surgery [3, 19]. We did not find any significant correlation between age and postoperative PaO 2 or A-aDO 2, suggesting that age is not a major risk factor of hypoxaemia after pulmonary resection. Prediction of Mechanical Ventilation Data on this issue, after lung resection, are controversial. Hirschler-Schulte and associates found that complications leading to acute respiratory failure were unpredictable [6]. For other authors, FEV1ppo is considered as predictive of MV [5, 8, 10]. Wahi and associates [8] have found that FEV1ppo in patients ventilated for over 48 h was slightly but significantly decreased compared with nonventilated patients after pneumonectomy (43% vs 48%). According to Nakahara and coworkers [5], FEV1ppo in prolonged ventilated patients or in patients needing tracheostomy was significantly decreased compared with either nonventilated patients needing postoperative bronchoscopy for atelectasis or major sputum production (37.6% vs 55.3%) or to patients with uncomplicated outcomes (37.6% vs 65.1%). Reasons for MV were not specified in these studies. Considering it doubtful that MV could be predicted by a routine preoperative pulmonary assessment whatever the occurrence of postoperative complications, we tried to evaluate the probability of MV according to postoperative complication patterns. After lobectomy, only pulmonary complications occurred, and FEV1ppo was the single predictive factor of MV for patients with pneumonia. After pneumonectomy, FEV1ppo was significantly decreased in patients requir-

6 Ann Thorac Surg FILAIRE ET AL 1999;67: HYPOXEMIA AFTER LUNG RESECTION 1465 ing MV for pulmonary complications compared with patients with uncomplicated outcome. Our data support FEV1ppo as a valuable test of identifying high-risk patients for MV. Nevertheless, routine pulmonary function tests failed to predict MV in patients undergoing pneumonectomy with developed nonpulmonary complications (perioperative bleeding: 1 patient; arrhythmia and heart failure: 2 patients; bronchopleural fistula: 1 patient). This lack of prediction can be explained by the fact that these complications have specific risk factors [20, 21] that are not considered in routine function tests. Small tidal volume and high respiratory rate were predictive of hypoxemia and MV after pneumonectomy. We have previously mentioned that such breathing pattern modifications can result from a thoracopulmonary compliance decrease and an increase in respiratory work. It can also reflect a fatiguing domain load for respiratory muscles, which provokes breathing control to decrease Vt and to accelerate RR so as to minimize energy expenditure [22]. Consequently, it is not surprising for patients with preoperative altered breathing pattern to present a high risk of respiratory failure after pneumonectomy. Previous study indicated that preoperative respiratory work was higher in patients requiring mechanical ventilation for over 48 h than for patients needing mechanical ventilation for less than 48 h [23]. In our study, patients with an RR 24/min and a Vt 550 ml developed respiratory failure. Despite the fact that our results have to be confirmed with a larger population, we postulate that the increase in respiratory work resulting from pneumonectomy was too important for these patients with a preoperative respiratory muscle fatigue. Thus, we believe that Vt and RR measurements are simple and useful preoperative tests to assess respiratory failure risk in patients eligible for pneumonectomy. We thank Mrs. Martine Collomb, Laurent Chiche, MD, and Philippe Hervé, MD, for their help in the preparation of the manuscript. References 1. Tisi GN. Preoperative evaluation of pulmonary function. Am Rev Respir Dis 1979;119: Entwistle MD, Roe PG, Sapsford DJ, Berrisford RG, Jones JG. Patterns of oxygenation after thoracotomy. Br J Anaesth 1991;67: Aldren CP, Barr LC, Leach RD. Hypoxaemia and postoperative pulmonary complications. Br J Surg 1991;78: Markos J, Mullan BP, Hillman DR, et al. Preoperative assessment as a predictor of mortality and morbidity after lung resection. Am Rev Respir Dis 1989;139: Nakahara K, Ohno K, Hashimoto J, et al. Prediction of postoperative respiratory failure in patients undergoing lung resection for lung cancer. Ann Thorac Surg 1988;46: Hirschler-Schulte CJW, Hylkema BS, Meyer RW. Mechanical ventilation for acute postoperative respiratory failure after surgery for bronchial carcinoma. Thorax 1985;40: Nakagawa K, Nakahara K, Miyoshi S, Kawashima Y. Oxygen transport during incremental load as a predictor of operative risk lung cancer patients. Chest 1992;101: Wahi R, McMurtrey MJ, DeCaro LF, et al. Determinants of perioperative morbidity and mortality after pneumonectomy. Ann Thorac Surg 1989;48: Celli BR. What is the value of preoperative pulmonary function testing? Med Clin North Am 1993;77: Nakahara K, Monden Y, Ohno K, Kawashima Y. A method for predicting postoperative lung function and its relation to postoperative complications in patients with lung cancer. Ann Thorac Surg 1985;39: Rosenberg J, Ullstad T, Rasmussen J, Hjorne FP, Poulsen NJ, Goldman MD. Time course of postoperative hypoxaemia. Eur J Surg 1994;160: Reeder MK, Goldman MD, Loh L, et al. Postoperative hypoxaemia after major abdominal vascular surgery. Br J Anaesth 1992;68: Jones JG, Sapsford DJ, Wheatley RG. Postoperative hypoxaemia: mechanisms and time course. Anaesthesia 1990; 45: Catley DM, Thornton C, Jordan C, Lehane JR, Royston D, Jones JG. Pronounced, episodic oxygen desaturation in the postoperative period: its association with ventilatory pattern and analgesic regimen. Anesthesiology 1985;63: Franck NR, Siebens AA, Newman MM, St Albans. The effect of pulmonary resection on the compliance of the human lungs. J Thorac Cardiovasc Surg 1959;38: Easton PA, Arnup NE, de la Rocha A, Fleetham JA, Anthonisen NR. Ventilatory control after pulmonary resection. Am Rev Respir Dis 1983;128: Lutchen KR, Saidel GM, Pririano FP, Horowitz JG, Deal EC. Mechanics and gas distribution in normal and obstructed lungs during tidal breathing. Am Rev Respir Dis 1984;130: Hsia CCW, Peshock RM, Estrera AS, McIntire DD, Ramanathan M. Respiratory muscle limitation in patients after pneumonectomy. Am Rev Respir Dis 1993;147: Kitamura H, Sawa T, Ikesono E. Postoperative hypoxemia: the contribution of age to the maldistribution of the ventilation. Anesthesiology 1972;36: Asamura H, Naruke T, Tsuchiya R, Goya T, Kondo H, Suemasu K. Bronchopleural fistulas associated with lung cancer operations: univariate and multivariate analysis of risk factors, managements and outcome. J Thorac Cardiovasc Surg 1992;104: Goldman L, Caldera DL, Nussbaum SR, et al. Multifactorial index of cardiac risk in noncardiac surgical procedures. N Engl J Med 1977;297: Roussos C. Ventilatory muscle fatigue governs breathing frequency. Bull Eur Physiopathol Respir. 1984;20: Maeda H, Nakahara K, Ohno K, Tetsuo K, Ikeda M, Kawashima Y. Diaphragm function after pulmonary resection: relationship with postoperative respiratory failure. Am Rev Respir Dis 1988;137: