Epidural analgesia improves outcome after multiple rib fractures Eileen M. Bulger, MD, Thomas Edwards, PhD, MD, Patricia Klotz, RN, and Gregory J. Jurkovich, MD, Seattle, Wash Background. Rib fractures are common and associated with significant pulmonary morbidity. We hypothesized that epidural analgesia would provide superior pain relief, and reduce the risk of subsequent pneumonia. Methods. A prospective, randomized trial of epidural analgesia versus IV opioids for the management of chest wall pain after rib fractures was carried out. Entry criteria included patients older than 18 years with more than 3 rib fractures and no contraindications to epidural catheter placement. Results. From March 2000 to December 2003, 408 patients were admitted with more than 3 rib fractures; 282 met exclusion criteria, 80 could not be consented, and 46 were enrolled (epidural n = 22, opioids n = 24). The groups were comparable for mean age, injury severity score, gender, chest Abbreviated Injury Scale, and mean number of rib fractures. The epidural group tended to have more flail segments (38% vs 21%, P =.20) and pulmonary contusions (59% vs 38%, P =.14), and required more chest tubes (95% vs 71%, P =.03) Despite the greater direct pulmonary injury in the epidural group, their rate of pneumonia was 18% versus 38% for the intravenous opioid group. When adjusted for direct pulmonary injury, there was a greater risk of pneumonia in the opioid group: OR, 6.0; 95% CI, 1.0-35; P =.05. When stratified for the presence of pulmonary contusion there was a 2.0-fold increase in the number of ventilator days for the opioid group: incident rate ratio, 2.0; 95% CI, 1.6-2.6; P <.001. Conclusions. The use of epidural analgesia is limited in the trauma population due to numerous exclusion criteria. However, when feasible, epidural analgesia is associated with a decrease in the rate of nosocomial pneumonia and a shorter duration of mechanical ventilation after rib fractures. (Surgery 2004;136:426-30.) From the Departments of Surgery and Anesthesiology, Harborview Medical Center, University of Washington, Seattle, Wash RIB FRACTURES ARE A COMMON INJURY in the blunt trauma population with a reported incidence of 10% among patients admitted to a regional trauma center. 1 The presence of 3 or more rib fractures has been associated with increased mortality and duration of care in intensive care units and hospitals. 2 Among the elderly, rib fractures have been associated with a 31% rate of nosocomial pneumonia. 3 The pain associated with rib fractures impairs ventilatory function and increases pulmonary morbidity. Management of these patients is therefore focused on achieving adequate analgesia and clearance of pulmonary secretions. Previous studies have demonstrated that epidural analgesia Presented at the 65th Annual Meeting of the Society of University Surgeons, St. Louis, Missouri, February 11-14, 2004. Reprint requests: Eileen M. Bulger, MD, Box 359796 Harborview Medical Ctr, 325 9th Ave, Seattle, WA 98005. 0039-6060/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.surg.2004.05.019 provides superior pain relief and improves pulmonary function tests when compared to intravenous (IV) opioids for patients with rib fractures. 4-6 However, no previous study has demonstrated that the use of epidural analgesia actually improves outcome for these patients. We sought to determine if epidural analgesia would reduce the risk of nosocomial pneumonia and duration of mechanical ventilation when compared to IV opioids. METHODS All patients admitted to our Level 1 trauma center who were older than 18 years and had 3 or more rib fractures were screened for enrollment from March 1, 2000, through December 15, 2003. Patients who had chest wall pain requiring IV opioids were evaluated for eligibility to receive thoracic epidural analgesia. Patients were excluded if they had any acute spine fracture or pre-existing spine deformity, severe traumatic brain or spinal cord injury, or severe altered mental status such that pain could not be assessed, unstable pelvic 426 SURGERY
Surgery Volume 136, Number 2 Bulger et al 427 fracture or open abdomen that would preclude positioning for epidural placement, ongoing cardiac instability or coagulopathy, active chest wall infection, and acute thoracic aortic transection. Patients whose chest wall pain was manageable with oral opioids or anti-inflammatory medications were also excluded. Eligible patients were approached for informed consent under terms approved by the Human Subjects Review Committee of the University of Washington. Informed consent was obtained from the patient or legal next of kin for patients who required mechanical ventilation. Prisoners were excluded from enrollment. Once consent was obtained, patients were randomized to receive either epidural analgesia or IV opioid analgesia. Randomization was performed in blocks of 20, with a plan for interim analysis after 40 patients. Initial power calculations were hampered by the lack of preliminary data defining the anticipated effect of epidural analgesia on pneumonia rates. Randomization assignments were placed in sealed envelopes and shuffled by an individual independent of the study. They were then numbered sequentially and required to be taken in order at the time of randomization. This prevented any potential for selection bias by the research nurse. Epidural catheters were placed and all analgesia was managed by consultants from the Pain Relief Service, a division of the Department of Anesthesiology. The primary medications administered via the epidural route included bupivicaine, morphine, and fentanyl. IV opioids, including morphine, hydromorphone, and fentanyl, were administered by patient-controlled analgesia for alert patients and with nurse assistance for patients who could not participate in self-administration. The medications administered were not regulated by the study, but all patients were managed by the Pain Relief Service, which minimized variability among providers. Baseline data included demographics, mechanism of injury, injury severity based on the injury severity score (ISS), number and location of rib fractures, presence of a flail segment, presence of a pulmonary contusion based on the review of chest radiographs over the first 48 hours, hemodynamic status on admission, and associated injuries and operations. The primary outcome variable was the development of nosocomial pneumonia within the first 28 days as defined by the Centers for Disease Control and Prevention s definition for pneumonia, which includes a new infiltrate on chest x-ray corresponding in size to 1 or more segments of the lung, fever (T > 38.38C) or hypothermia ( < 368C), leukocytosis (white blood cell count [WBC] > 10,000/mm 3 or a 25% increase over last available value, or bands >10% of total WBC) or leukopenia (WBC < 4,000/mm3) and bacteriologic confirmation (BAL with >10 4 CFU/mL or positive sputum culture with > 3+ growth of 1 type of pathogenic bacteria). 7 Secondary outcome variables included duration of mechanical ventilation, length of stay in hospital or intensive care unit, and mortality. The development of the acute respiratory distress syndrome (ARDS) was determined based on the AmericaneEuropean consensus conference definition. 8 Pulmonary contusion was defined as a dominant infiltrate on chest x-ray within the first 48 hours consistent with the pattern of injury. Data were also collected detailing any complications associated with the analgesia technique. Statistical analysis was conducted on an intention to treat basis. If a patient in either study group failed to achieve adequate pain relief with the assigned treatment, they were allowed to cross over to the other arm of the study. In this event, they were analyzed based on the initial randomization assignment. Data were collected on a standardized data abstraction form and entered into a computerized database. The statistical software program, Stata Version 8 (Stata Corp, College Station, Tex) was used for analysis. Differences between treatment groups were determined with the chi-square test for dichotomous variables and the Student t test for continuous variables. Significance was defined as P <.05. This allowed for identification of confounding variables that altered the outcome variable estimate by greater than 10%. These variables were included in a logistic regression model to assess the risk of developing nosocomial pneumonia as the primary outcome. Poisson regression was used to evaluate the effect of the treatment group on duration of mechanical ventilation. The effect of treatment was stratified based on the presence of pulmonary contusion. RESULTS During the study period, 408 patients who were older than 18 years and had more than 3 rib fractures were admitted. Of these, 282 were excluded, 80 refused participation or could not be consented, and 46 were enrolled (Figure). Twenty-two patients were randomized to the epidural group and 24 to the systemic opioid group. The groups were comparable for gender, mean age, mean ISS score, chest Abbreviated Injury Scale (AIS), head AIS, hypotension on admission, and mean number of ribs fractured (Table I). The
428 Bulger et al Surgery August 2004 Figure. Patient screening and enrollment. All adult patients with 3 or more rib fractures were screened for enrollment. Patients with insufficient chest wall pain to require IV opioids were excluded. Additional exclusion criteria included associated injuries that precluded epidural placement. Two patients were transferred to our facility with an epidural catheter already in place. majority of patients were injured as a result of a motor vehicle collision (68%). There tended to be more flail segments (38% vs 21%, P =.20) and pulmonary contusions (59% vs 38%, P =.14) in the epidural group. In addition, patients in the epidural group were more likely to have a chest tube (95% vs 71%, P =.03). The Acute Physiology and Chronic Health Evaluation (APACHE) II score on admission was no different in the epidural group (mean of 13 vs 11, P =.20). Three patients (13%) in the systemic opioid group failed to achieve adequate pain control and required cross over to the epidural group, and 3 patients (13%) in the epidural group required cross over to IV opioids. Two of the 3 patients who had their epidural catheters removed were removed due to deteriorating neurologic status related to traumatic brain injury; the remaining catheter was removed due to the need to start the patient on systemic anticoagulation. Patients in the epidural group had the catheter in place for an average of 4.4 days (range, 1 to 7). Complications of epidural analgesia were observed in 6 patients (27%). These included pruritis (5 patients, 23%), transient motor block (2 patients, 9%), catheter site inflammation or superficial infection (1 patient, 5%), hypotension (1 patient, 5%), and allergic reaction (1 patient, 5%). Two patients required replacement of the catheter to achieve better pain control. Eight patients (33%) suffered complications from IV opioids, including pruritis (5 patients, 21%), nausea/vomiting (6 patients, 25%), and depressed level of consciousness (1 patient, 4%). The unadjusted rate of pneumonia was 18% in the epidural analgesia group and 38% in the systemic analgesia group (P =.15) (Table II). When adjusted for the difference in direct pulmonary injury, patients in the systemic opioid group had a 6.0-fold increase in the risk of pneumonia (OR, 6.0; 95% CI, 1.0-35; P =.05) (Table III). Patients in the epidural group spent a mean of 7.6 days on the ventilator compared to 9.1 days for those in the systemic opioid group. This finding resulted in an unadjusted incident rate ratio (IRR) of 1.19 (95% CI, 0.97-1.45; P =.09), which indicates a 19% increase in ventilator days for the systemic opioid group. When stratified based on the presence of a pulmonary contusion, patients in the opioid group had a 2.0-fold increase in
Surgery Volume 136, Number 2 Bulger et al 429 ventilator days (IRR, 2.0; 95% CI, 1.6-2.6; P <.001). There was no difference in mortality or duration of hospital or ICU stay between the groups. Patients in the epidural group tended to have a higher rate of ARDS than those in the systemic opioid group (P =.15). DISCUSSION Previous studies have demonstrated that epidural analgesia after rib fractures is associated with improved pain relief scores and improved pulmonary function tests compared to IV opioids. 4,5 These studies have not, however, identified a definitive association between these factors and improved patient outcome. Retrospective reviews of epidural versus opioid administration are hampered by significant selection bias because patients who receive epidural catheters tend to be older and have more significant chest wall injury. 6 Our study is the first to demonstrate that epidural analgesia improves outcome for patients with chest wall pain associated with multiple rib fractures. This is evident by the decreased risk of nosocomial pneumonia and shorter duration of mechanical ventilation. Rib fractures among the blunt trauma population are associated with significant multisystem injury. As a result, the applicability of epidural analgesia is limited by numerous exclusion criteria. Each of the exclusion criteria used in this study was derived from a set of clinical practice guidelines considered to be conservative standards; the criteria are based on specific risks inherent in the technique of epidural analgesia. 9 Of the 408 patients identified, 352 had sufficient chest wall pain to require IV therapy. Among these, 224 (64%) had associated injuries that excluded epidural placement, the majority of which were spine fractures (61%). This fact points out the limitations associated with epidural use in this population. This may warrant the further development of other techniques to provide regional analgesia or re-evaluation of exclusion criteria for epidural placement in these patients. Application of theses standards was deliberately conservative in this study. It may be that, for example, in certain types of spinal column injuries associated with rib fractures (eg, transverse process fractures) liberalization of the criteria may allow for wider use of a beneficial technique with no significant increase in adverse events. Of note, complications from epidural use were no more frequent than those related to IV opioids, with the predominant complication of pruritis in both groups. No Table I. Demographics and injury severity Epidural (n = 22) Opioids (n = 24) P value % male 77% 67%.42 Age (y)* 49 ± 18 46 ± 16.55 ISS* 26 ± 8 25 ± 8.54 APACHE II* 13 ± 5 11 ± 5.20 Chest AIS* 3.7 ± 0.7 3.7 ± 0.8 1.0 Head AIS* 1.2 ± 1.2 1.1 ± 1.2.78 No. of rib Fx* 7.2 ± 3.2 6.8 ± 3.3.76 Flail segment 8 (38%) 5(21%).20 Chest tube 21 (95%) 17 (71%).03 8 (33%).65 Shock on admission 6 (27%) Mechanical ventilation 16 (72%) 13 (54%).19 Pulmonary contusion 13 (59%) 9 (38%).14 AIS, Abbreviated Injury Scale; APACHE, Acute Physiology and Chronic Health Evaluation; ISS, Injury Severity Score; Fx, fracture. *Mean ± SD. Table II. Unadjusted outcome parameters Epidural (n = 22) Opioids (n = 24) P value Pneumonia 4 (18%) 9 (38%).15 No. of vent days* 8±16 9±26.41 ARDS 10 (45%) 6 (25%).15 Mortality 2 (9%) 1 (4.2%).50 LOS (d)* 18 ± 16 16 ± 13.60 LICU (d)* 10 ± 15 12 ± 26.78 ARDS, Acute respiratory distress syndrome; LOS, length of hospital stay; LICU, length of intensive care unit stay. *Mean ± SD. Table III. Adjusted outcome parameters OR/IRR 95% CI P value Nosocomial pneumonia* OR, 6.0 1.0-35.05 Ventilator daysy IRR, 2.0 1.6-2.6 <.001 OR, Odds ratio; IRR, incident rate ratio; CI, confidence interval. *Logistic regression IV opioid vs epidural (confounding variables: pulmonary contusion, flail chest, chest tube, APACHE II). y Poisson regression IV opioid vs epidural (stratified for pulmonary contusion). epidural hematomas or epidural abscesses were identified. A limitation of our trial is the sample size, which represents only 37% of the patients who met eligibility for the study. This was due in large part to the inability to obtain consent due to either patient or family refusal. Ten patients were
430 Bulger et al Surgery August 2004 excluded based on physician refusal, which could have introduced some selection bias. Although the difference in pneumonia rates did not reach statistical significance on univariate analysis, adjusting for the difference in pulmonary injury between the groups resulted in a significant effect. An additional limitation to the trial was the inability to blind the care providers or patients to the therapy group. This could have led to subtle treatment modifications biased by the analgesia technique employed. For example, it is possible that the care providers were more aggressive about weaning mechanical ventilation based on the presence of an epidural catheter; thus, this action could account for the shorter duration of mechanical ventilation in this group. All ventilator weaning was managed by a standardized protocol conducted by the respiratory therapists; however, the final decision to extubate the patient was left to the physician. Because of the need to provide each patient with adequate pain relief, we designed the study to allow cross over to the alternative treatment arm should the assigned treatment fail. Allowing cross over is another potential limitation, but cross over appeared to occur equally between the 2 groups, and the data were analyzed on an intention to treat basis. CONCLUSION We demonstrated that epidural analgesia, when feasible, improves the outcome for patients with chest wall pain associated with multiple rib fractures. Use of this therapy is limited by associated injuries in this patient population. REFERENCES 1. Ziegler DW, Agarwal NN. The morbidity and mortality of rib fractures. J Trauma 1994;37:975-9. 2. Lee RB, Bass SM, Morris JA, Jr, MacKenzie EJ. Three or more rib fractures as an indicator for transfer to a Level I trauma center: a population-based study. J Trauma 1990;30: 689-94. 3. Bulger EM, Arneson MA, Mock CN, Jurkovich GJ. Rib fractures in the elderly. J Trauma 2000;48:1040-6; discussion 46-7. 4. Mackersie RC, Karagianes TG, Hoyt DB, Davis JW. Prospective evaluation of epidural and IV administration of fentanyl for pain control and restoration of ventilatory function following multiple rib fractures. J Trauma 1991;31: 443-9; discussion 49-51. 5. Moon MR, Luchette FA, Gibson SW, Crews J, Sudarshan G, Hurst JM, et al. Prospective, randomized comparison of epidural versus parenteral opioid analgesia in thoracic trauma. Ann Surg 1999;229:684-91; discussion 91-2. 6. Wu CL, Jani ND, Perkins FM, Barquist E. Thoracic epidural analgesia versus IV patient-controlled analgesia for the treatment of rib fracture pain after motor vehicle crash. J Trauma 1999;47:564-7. 7. Garner J, Jarvis W, Emori T, Horan T, Hughes J. CDC definitions for nosocomial infections. In: Olmsted R, ed. APIC Infection Control and Applied Epidemiology: Principles and Practice. St Louis: Mosby; 1996:A1-20. 8. Bernard GR, Artigas A, Brigham KL, Carlet J, Falke K, Hudson L, et al. Report of the American-European consensus conference on ARDS: definitions, mechanisms, relevant outcomes and clinical trial coordination. The Consensus Committee. Intensive Care Med 1994;20:225-32. 9. Edwards W. Posttrauma pain. In: Loeser J, Butler S, Chapman C, Turk D, eds. Bonica s Management of Pain. 3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2001: 788-93.