| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the *Department of Anesthesia and Intensive Care, Hopital Cardio-Vasculaire et Pneumologique Louis Pradel, Hospices Civils de Lyon, Lyon-Bron, France; Anesthesiology Department, Thoracic Surgery Unit, Department of Public Health, Biostatistic and Methodology Unit, and ||Department of Respiratory, Cardiocirculatory, and Exercise Physiology, University Hospital Strasbourg, Strasbourg, France.
Address correspondence and reprint requests to Christian Bauer, MD, Hospices Civils de Lyon, Hopital Cardio-Vasculaire et Pneumologique Louis Pradel, 28 Ave. du Doyen Lepine, 69677 Bron Cedex, France. Address e-mail to christian.bauer{at}chu-lyon.fr.
Abstract
BACKGROUND: Although thoracic epidural analgesia (TEA) is considered superior to IV opioids for postoperative analgesia after thoracic surgery, a few studies clearly demonstrate an improvement in pulmonary function attributable to TEA using a local anesthetic in combination with an opioid.
METHODS: In this prospective, randomized, double-blind study, we compared the effects of TEA with ropivacaine and sufentanil (TEA group) to IV morphine (IV group), as they affected pain and pulmonary function after lobectomy in 68 patients. Pain intensity, forced vital capacity (FVC), forced expiratory volume in 1 s (FEV1), FEV1/FVC ratio, forced expiratory flows, and sniff nasal inspiratory pressure as a marker of inspiratory muscle strength were measured from the first to the fourth postoperative day.
RESULTS: Pain relief was better in the TEA group at rest and on coughing (P < 0.001). The impairment of FVC and FEV1 was less in the TEA group when compared with that in the IV group (P < 0.001 and P = 0.003, respectively). Sniff nasal inspiratory pressure, FEV1/FVC ratio, and expiratory flow values decreased similarly in both groups. In-hospital mortality, as well as postoperative pulmonary complications, was not different between groups.
CONCLUSION: After lobectomy, TEA enables a significant increase in pulmonary function concomitant with better pain relief than systemic morphine, although a modest intercostal motor block may occur.
Pain due to thoracotomy, combined with the loss of parenchyma, drastically reduces respiratory function for the first postoperative days (1–3). Analgesia alone does not prevent reduction of pulmonary function after thoracotomy and lung resection (4,5). Although epidural analgesia is the preferred method of postoperative analgesia (6), the segmental epidural block induced by thoracic epidural analgesia (TEA) may negatively influence ventilatory mechanics. Few controlled trials have examined the effects of TEA on lung function in the immediate postoperative period as a primary outcome variable. The aim of the present study was to investigate whether the impairment of pulmonary function observed in patients after lung resection can be decreased by TEA.
METHODS
Patients
The study group consisted of patients undergoing elective pulmonary lobectomy or bilobectomy via a lateral or posterolateral thoracotomy, without chest-wall resection. After obtaining approval from the Institutional Ethical Committee, 94 consecutive patients were evaluated for eligibility. Exclusion criteria were age (younger than 18 yr or older than 75 yr), previous epidural analgesia or anesthesia, previous spinal anesthesia, forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) <70% of normal value, significant cardiovascular disease (severe valvular heart disease or congestive failure with New York Heart Association class III–IV) or severe coronary heart disease (Canadian Cardiovascular Society class III–IV), allergy to local anesthetics or opioids, current opioid use, antiplatelet therapy for the last 10 days, active infectious process, neurological disorders, previous surgery to or severe deformity of the thoracic spine, abnormal coagulation tests, renal or hepatic failure, lack of cooperation, and inability to understand or perform verbal or physical assessments. Ninety-three patients accepted and gave their written informed consent before entering the study. To obtain homogeneity in the sampling, it was decided a priori to drop patients from analysis if their procedure required pneumonectomy, only wedge resection, or if no lung parenchyma was resected. The process of inclusion into the study went on until the requested number of patients was reached, 34 in each group. The protocol specified that the study data would be analyzed per-protocol (i.e., only include patients who met the surgical inclusion criteria specified above), rather than by intention to treat.
Preoperative Management
Patients were instructed in the use of a visual analog scale (VAS) graduated from 0 cm (no pain) to 10 cm (worst possible pain), a patient-controlled analgesia pump device (Graseby 9300®, Graseby Medical Limited, Watford, UK), and trained in spirometric testing (ZAN 100®-400® Spirometer, ZAN Messgeräte GmbH, Waldfenster, Germany) before surgery. Patients were randomized into one of two groups using a table of random numbers. The flow diagram of the study is shown in Figure 1. Thirty-four patients remained for final analysis in each group. The hospital pharmacy prepared a 650-mL bag of ropivacaine at 2 mg/mL with 0.5 µg/mL of sufentanil for the TEA group and a 200-mL bag of morphine at 1 mg/mL with 0.05 mg/mL of droperidol for the IV group.
|
Anesthetic Management
Anesthesia was always performed by the same staff anesthesiologist who did not participate in postoperative evaluation and who was aware of the patient’s treatment group. A subcutaneous injection of 5 mL of lidocaine at 1% was performed at the T3–4 interspace on all patients. In the TEA group, a 20-gauge epidural catheter (Perifix®, B. Braun Melsungen, Germany) was placed via a midline approach, using the hanging drop technique and threaded 4 cm cephalad in the epidural space. Three to 5 mL of ropivacaine at 7.5 mg/mL, depending on the height of the patient (3 mL for a height less to 160 cm, 4 mL for between 160 and 180 cm, and 5 mL for more than 180 cm) combined with 20 µg of sufentanil was injected to induce bilateral sensory block, 20 min after the puncture, which included two or more dermatomes around T4 assessed by loss of temperature discrimination to ice. In the IV group, a fake epidural catheter was taped on the patient’s back with the filter fixed on the shoulder in the same manner as the TEA group.
General anesthesia was induced with 2–4 mg/kg thiopental and 0.5 µg/kg sufentanil. A radial artery catheter was placed and tracheal intubation with a double-lumen tube was facilitated by 0.1 mg/kg vecuronium. The tidal volume under one-lung ventilation was 8 mL/kg, respiratory rate was between 12 and 14 breaths/min and positive end-expiratory pressure was not applied. Anesthesia was maintained with isoflurane and repeated boluses of IV sufentanil as necessary. In the TEA group and before thoracotomy, the epidural catheter was connected to the ropivacaine-sufentanil bag and the infusion rate set at 4 mL/h. Thirty minutes before chest closure, 1 g of paracetamol was given IV to all patients. Anesthesia was targeted to allow tracheal extubation in the operating room immediately after the surgical procedure and after fulfilling standard extubation criteria.
Masked Postoperative Patient-Controlled Pain Management
All patients used the same Graseby 9300 analgesia pump device and had a peripheral IV line throughout the study period. On the infusion line from the pump to the patient, was a three-way stopcock which permitted infusion towards either the epidural catheter (TEA group) or the IV access (IV group). The medication bag, the pump, and the three-way stopcock were masked inside a closed, opaque bag, so that neither the observer nor the patient could identify in which catheter, IV or epidural, the pain medication was infused.
The analgesic objective was to maintain VAS scores below three at rest and six at movement. In the TEA group, the rate of the continuous infusion was set according to the patient’s height. Patients shorter than 160 cm started with an infusion rate of 4 mL/h and boluses of 4 mL, patients between 160 and 180 cm had 5 mL/h and 5 mL, respectively, and those taller than 180 cm had 6 mL/h and 6 mL. In the IV group, the pump delivered a bolus of 1 mg with a lockout time of 7 min without continuous infusion. In the first hour after tracheal extubation, the studied analgesic medications were titrated, if necessary, to reach the analgesic objective. Titration consisted, in the TEA group, of an epidural bolus of 5 mL of ropivacaine at 3.75 mg/mL every 20 min and, in the IV group, of an IV bolus of 2 mg of morphine every 5 min. No supplemental analgesics were given during that period. Later and each time the analgesia goal could not be fulfilled, rescue medication consisted of IV 100 mg of ketoprofen every 12 h. Simultaneously, the epidural rate was increased by 2 mL/h and the bolus by 2 mL, and the morphine bolus was set at 1.5 mg. Systematically, all patients received 1 g of paracetamol every 6 h. Following the standard practice of the unit, pain was evaluated three times a day from the first postoperative day until discharge was performed by the nursing staff. The nursing staff was blinded to the pain treatment. The VAS pain scores used for analysis were recorded before spirometric testing by the same physician, who was blinded to the pain treatment.
Spirometric Measurements
For the entire study, all tests were performed at the bedside with the patient in a sitting or semirecumbent position, by the same physician who was unaware of the patient’s pain treatment protocol. FVC, FEV1, peak expiratory flow (PEF), mean forced expiratory flow between 25% and 75% of FVC, and sniff nasal inspiratory pressure test (SNIP) were measured. The spirometric measurements were performed the day before the operation and once a day from postoperative day 1 to postoperative day 4. Maximum effort was verbally encouraged. Maneuvers were separated by at least 30 s rest and continued until no further increase in volume, flow, or pressure could be obtained.
Arterial blood gases were measured 2 h after tracheal extubation and on the morning of postoperative day 1, with the patients breathing 3 L/min of oxygen via a nasal catheter. Respiratory rate, heart rate, and arterial blood pressure were recorded hourly in the intensive care unit and every 8 h by the ward nursing staff. Respiratory depression and hypotension were defined by a respiratory rate <8 bpm and a systolic arterial blood pressure <80 mm Hg. Patients were observed for sensory or motor blocks, Horner syndrome, urinary retention, nausea, vomiting, or itching. Chest radiographs were obtained systematically on the operative day and on postoperative days 1 and 3. Postoperative pulmonary complications were monitored until discharge and defined as the appearance of atelectasis or new lung densities associated or not with fever. Electrocardiograms were recorded on the day of the operation, on postoperative day 1 and as soon as an irregular heart rate, a rate higher than 100 bpm, or symptoms of myocardial ischemia were documented. The resort to ketoprofen, the time of chest tube removal, the length of hospital stay, and the in-hospital mortality were determined.
Statistical Analysis
We considered a reduction of 2.5 cm on the VAS to be clinically relevant. A standard deviation of three was estimated for this difference. On the basis of the formula for normal theory and assuming a two-sided error type I error of 0.05 and a power of 0.90, 31 patients in each group were required to show a reduction in the mean VAS of 2.5 in any group. The final number of patients needed was increased to 34 in each group to account for a possible rate of missing data of about 10%. Thus, the final number of patients needed was 68. A sample size calculation based on spirometric data, our primary end-point, would have been more relevant. Unfortunately, when performing the study, lung function data after lobectomy with homogeneous surgical samples were nonexistent. The results presented here, together with other data published since then (4,7), may be helpful for future sample size calculations.
For preoperative and postoperative patient characteristics, the comparison of qualitative data was done using either 
2 test or Fisher’s exact test, depending on the expected values for each group. The comparison of continuous data was done using either Student’s t-test or the nonparametric Mann–Whitney U-test depending on whether the data distribution was compatible with a Gaussian distribution or not. Continuous data are described with their mean, median, and standard deviation. Categorical and ordered data are described with percentages. Spirometric data were analyzed using linear mixed models. Mixed models are an extension of the analysis of variance where one can specify random effects apart from the usual fixed effects. Moreover, a given variance–covariance matrix can be specified in order to apply a model to the data. This makes it possible to consider the correlation between data that occurs when using repeated data on the same subjects as in our spirometric data. For each spirometric parameter, four different variance– covariance matrices have been tested (unstructured, Toeplitz, autoregressive, and Compound Symmetry). The choice of the best model depended on Bayesian information criterion, with which it is possible to compare nontested variance–covariance matrix. In these analyses, a different variance–covariance matrix was specified for each treatment group. The interaction between treatment and time was estimated in every model. The dependent variable was the variable itself and not a change from baseline. For each model, preoperative (baseline) spirometric data have been introduced as a covariate. Analyses were run with PROC MIXED from SAS 8.0 software were the variance– covariance structures were given in a REPEATED statement. Time and group were both considered as a categorical fixed effect factor.
VAS data had a non-Gaussian distribution in this study. Thus, it was not possible to use linear models such as linear regression or analysis of variance to fit the data. The data were modeled using an ordinal logistic regression with proportional odds. Like the standard logistic regression, this model provides odds ratio. The given odds ratio is the mean odds ratio over all possible partitions for the outcome (VAS scores). In other words, an equal odds ratio is assumed when comparing first and second categories versus a third one, and in comparing first versus second and third categories. In our results, the odds ratio must be given the following interpretation: an odds ratio significantly higher (respectively lower) than one shows higher (respectively lower) pain levels for the treatment group under study than for the control group. Odds ratios are given with their confidence intervals.
Analyses were run using GEE2 SAS 8.0 MACRO based on the General Estimation Equation (8). The choice of the best model is based solely on the comparison of the different structured variance– covariance matrix versus the unstructured one, because in the GEE2 MACRO likelihood values are not produced for these models (9). VAS was recorded as a continuous variable and was rounded to the next integer. A recoding was then considered and the data were put into one of three categories: "at rest" 0 for 0 values, 1 for values higher than 0 and equal or lower than 3, 2 for values higher than 3; "at cough" 0 for values higher than 0 and lower or equal than 3, 1 for values higher than 3 and lower or equal than 6 and 2 for values higher than 6. The choice of categories thresholds was made on unblinded data by considering the best balance between category sizes over the four postoperative days. Using the rounded values in the GEE ordinal logistic regression models gave no satisfactory results with either a bad fit of the model leading to the rejection of the proportional odds assumption or absence of convergence. Thus, the fitted models are based on the VAS categories only. The proportional odds assumption was not tested for VAS categories. The selected variance–covariance structure was the banded structure for both "at rest" and "at cough" VAS. The VAS and the spirometric analyses were done per protocol. A P value equal to or <0.05 was considered significant.
RESULTS
The flow diagram of the study is shown in Figure 1. The two groups had similar baseline characteristics (Table 1). The duration of surgery was similar in both groups (205 ± 43 min for the IV group versus 220 ± 34 min for the TEA group). Thirty patients (88%) in the TEA group and 18 patients (53%) in the IV group were tracheally extubated in the operating room. Sufentanil consumption (125 ± 40 µg vs 53 ± 25 µg, P < 0.001) and time to extubation (50 ± 71 min vs 11 ± 32 min, P = 0.006) were increased in the IV group when compared with those in the TEA group.
|
VAS scores at rest and on coughing expressed as odds ratios revealed significantly less pain for patients in the TEA group than in the IV group, throughout the four postoperative days (Table 2).
|
Patients in the TEA group had significantly greater postoperative FVC and FEV1 values (Fig. 2) when compared with those in the IV group (P < 0.001 and P = 0.003, respectively). The pattern of recovery was steady throughout the study, beginning on the first postoperative day and followed the pattern of the VAS scores. The day-to-day increase in FVC and FEV1 values beyond postoperative day 1 is statistically significant at all times (P < 0.0001). FEV1/FVC, PEF, and mean forced expiratory flow between 25% and 75% of FVC postoperative values were similar between groups. SNIP postoperative values were not significantly different between the two groups (P = 0.089). Postoperative Paco2 values 2 h after extubation (42.1 [40.5–43.6; 95% confidence interval] versus 44.4 mm Hg [42.9–46]) and on postoperative day 1 (41.7 [40.3–43.1] vs 44.8 mm Hg [43.4–46.2]) were significantly lower with TEA compared with those with IV (P = 0.03 and 0.001, respectively). Arterial blood pressure of oxygen was similar between groups. A treatment by time interaction was introduced in each model but was never significant in the finally selected model. The selected variance–covariance structures were autoregressive of first order for FVC and mean expiratory flow25%–75%, unstructured for FEV1 and PEF and compound symmetry for FEV1/FVC, SNIP, and Paco2.
|
In-hospital mortality (Table 3) was not significantly different between groups (P = 0.493). Two deaths occurred in the IV group. One patient died of a bronchial fistula that led to a fatal pneumopathy on postoperative day 17. The other patient died on postoperative day 19 from the consequences of a chylothorax. IV ketoprofen as a rescue medication was used significantly more frequently in the IV group (P = 0.026). Transient neurological side effects, such as paresis, paresthesias in the upper limbs, and Horner syndrome, were significantly more frequent in the TEA group (P = 0.006). The incidence of cardiac arrhythmias, pulmonary complications, nausea or vomiting, itching, urinary retention, and hypotension was not statistically different between the groups. We found no respiratory depression in our patients. The length of time of chest drainage and the length of stay in the hospital were not different between the groups. Two technical problems, due to the analgesia pump device, occurred in each group and were managed without violation of the study protocol.
|
DISCUSSION
The major findings of this study are that the patients treated with TEA had better preservation of lung volume, no increase in airway resistance, and lower pains scores at rest and on coughing than the patients treated with IV morphine. The pain score results are in accordance with other studies comparing a mixture of thoracic epidural local anesthetics and opioids to IV morphine after thoracotomy (6,10,11).
On postoperative day 1, all patients experienced a drastic decrease of 70% in FVC concomitant with a 60% decrease in FEV1 that represents more than the anatomical loss of lung tissue. These results were similar to those found by previous investigators (3,10,12,13).
A cumulative meta-analysis performed by Ballantyne et al. (14) showed that epidurally injected opioids and local anesthetics significantly improved pulmonary outcome, but failed to demonstrate a any benefit of the various pain therapies on pulmonary function. Only two of the trials reviewed by the authors studied thoracotomies with the use of thoracic epidural opioid plus local anesthetic or local anesthetic alone, versus systemic opioid (13,15). Of these two trials, only the study by Brichon et al. (13) reported data on pulmonary function. These investigators reported significantly higher values of FEV1 in the epidurally treated patients compared with those in a cryoanalgesia group and a control group.
A number of studies have used FVC and FEV1 to analyze overall respiratory function and particularly the effects of epidurally injected local anesthetics in nonsurgical subjects. In healthy volunteers, Sundberg et al. (16) reported that high TEA performed with bupivacaine at 5 mg/mL caused a decrease in vital capacity and inspiratory capacity, but not FEV1. In subjects with severe obstructive pulmonary disease, segmental high TEA performed with 6 mL of 0.75% ropivacaine showed evidence of a slight respiratory motor block with a decrease of 10% and 13%, respectively, in FEV1 and FVC (17). The higher values of FVC and FEV1 found in our patients treated with TEA using 0.2% ropivacaine shows that the analgesic effect outweighs the potential negative effects on ventilatory function. Even if the difference in VAS scores were not clinically meaningful, the patients treated with IV morphine had a delayed recovery in lung volumes that could have been disadvantageous. The first postoperative days are crucial to successful recovery. Failure to mobilize lung volume can lead to atelectasis, impaired gas exchange, pneumonia, and respiratory failure. Postoperative FEV1 is a useful predictor of postoperative outcome. A low predicted postoperative value of FEV1 has been shown to correlate significantly with a complicated postoperative course and poor surgical outcome (18,19). By increasing lung volume, TEA may help reduce the incidence of postthoracothomy pulmonary complications (20). The two evaluated analgesia techniques did not influence either the length of time of chest drainage or the length of hospital stay in our study.
A trend to more in-hospital mortality and more postoperative pulmonary complications was found in the IV group, but without statistical significance. Our study lacked statistical power to find a significant difference.
Groeben et al. (21) showed that sympathetic blockade with bupivacaine concentrations as high as 0.75% did not affect the tone of bronchial muscles. Conversely in our patients, we did not observe any bronchial obstruction, assessed by the FEV1 to FVC ratio, with ropivacaine at 2 mg/mL.
For patients with postthoracotomy pain, the FVC maneuver is demanding. The SNIP test has been shown to be a reliable and useful noninvasive test for inspiratory muscle strength both in healthy subjects and in patients with neuromuscular or skeletal disorders (22–24). We decided to use the SNIP test to assess inspiratory muscle strength. We believe this is the first study that analyzes the influence of lobectomy and postthoracotomy analgesia on SNIP. The SNIP values decreased after surgery by 50% and remained higher in the TEA group, but the difference did not reach statistical significance (P = 0.0886). Thus, we cannot exclude that TEA with 0.2% ropivacaine (mL) could have produced a certain amount of motor block of the intercostal muscles or hindered rib cage expansion by decreasing the activity of the diaphragm.
A limitation of this study was the absence of validation at the time of the SNIP test in patients who had undergone a resection of lung parenchyma. Further studies are needed in this setting to test agreement between SNIP and a standard method like transdiaphragmatic pressure. Our study was statistically under-powered to show a difference between the groups in SNIP values. Another important limitation is the lack of sedation monitoring in our patients. The amount of morphine used, and thus morphine analgesia, may have been suboptimal in the IV group. This is suggested by the more frequent use of rescue ketoprofen in the IV morphine group. However, mean VAS scores stayed within acceptable ranges with 68% of patients with <3 at rest on postoperative day 1 and increasing to 88% on postoperative day 4.
In conclusion, this study showed superior analgesia provided by TEA, especially at mobilization, and a better preservation of FVC and FEV1, without increasing airway resistance. Undoubtedly, a larger series is needed before conclusive results can be drawn on the effects of TEA on SNIP. After pulmonary lobectomy, TEA is a useful postoperative analgesic technique to provide high-quality analgesia while minimizing ventilation impairment.
Footnotes
Accepted for publication March 12, 2007.
Supported by a grant from the French Ministry of Health n° 1990 PHRC 97, a grant from AstraZeneca and Janssen companies.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
, IV (IV morphine analgesia); light gray bars 
, TEA (thoracic epidural analgesia); FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; PEF, peak expiratory flow; MEF25%–75%, mean expiratory flow between 25% and 75% of FVC; SNIP, Sniff nasal inspiratory pressure; preop, preoperative values; POD, postoperative day. P values for preoperative values comparisons were obtained independently of mixed models, using Student’s t-test, the preoperative values in the model essentially were a covariate and not particularly used for its P value. Means and sem were observed values while P values were obtained using models in which the sem was slightly different due to the process of model estimation. *P < 0.05 versus IV.
