- Email: [email protected]
Pneumonia: Cause or symptom of postinjury multiple organ failure?
Pneumonia: Cause or Symptom of Postinjury Multiple Organ Failure? Angela Sauaia, MD, Frederick A. Moore, MD, Ernest E. Moore, MD, James B. Haenel, RRT, Robert A. Read, MD, Denver,Colorado
Recent studies have shown that selective gut decontamination can reduce the incidence of pneumonia, but this does not decrease multiple organ failure (MOF) or mortality. These findings have prompted the hypothesis that pneumonia is an inconsequential symptom of MOF. To test this, we prospectively evaluated 123 high-risk trauma patients (mean Injury Severity Score = 36.2 q- 1.5). Organ dysfunction, scored daily according to a 12-point scale, ultimately developed in 28 (23%) patients. Major infectious were diagnosed, based on strict criteria, in 59 patients (48%), and pneumonia developed in 52 patients (43%). Pneumonia was significantly associated with MOF (82% of patients with MOF versus 30% of patients without MOF, p < 0 . 0 0 0 1 ) . In 14 (50%) of the patients with MOF, pneumonia preceded a significant rise (greater than or equal to 3) in serial MOF scoring. Of note, 10 (71%) of these patients died. Among the remaining 14 patients with MOF, 10 developed pneumonia, but this was associated with a minimal increase (less than or equal to 2) in MOF scoring (3 patients died). These data, by temporal association with MOF scoring, implicate pneumonia in precipitating or significantly worsening organ failure in 50% of the patients who developed MOF.
' ultiple organ failure (MOF) remains the leading M . cause of late postinjury death, but, despite intensive investigation, its pathogenesis remains elusive [1-4]. Following resuscitation, heterogenous trauma patients enter a similar phase of hyperinflammation. This is presumably beneficial (i.e., the injury stress response) and, in most cases, resolves as the patient recovers. What triggers the transition into the autodestructive systemic inflammation that characterizes postinjury MOF is not clear [5,6]. Traditionally, infection has been viewed as the pivotal delayed event that precipitates MOF [7-9]. However, it has been recognized that MOF can occur without an identifiable focus of infection and that MOF will frequently not resolve despite successful drainage of abscesses [10-13]. More recently, prospective, randomized trials have shown that selective gut decontamination (SGD) can reduce the incidence of infections (principally pneumonia) in patients requiring prolonged mechanical ventilation but offer conflicting results regarding the impact on organ failure or mortality [14-18]. These findings have prompted the hypothesis that pneumonia is merely an inconsequential symptom of MOF. This has important implications in directing future strategies in preventing MOF. Consequently, we designed this study to investigate the potential causal relationship between infectious complications (principally pneumonia) and MOF in patients sustaining major trauma to the torso who required prolonged mechanical ventilation. Our specific aims were as follows: (1) to prospectively document the incidence of pneumonia in this trauma population; (2) to determine if pneumonia is associated with MOF; and (3) to characterize the temporal relationship between pneumonia and organ failure scoring to determine whether pneumofiia could precipitate or worsen MOF.
PATIENTS AND METHODS During the 2-year period ending December 1992, all patients with injuries to the torso admitted to the surgical intensive care unit (SICU) of the Denver General Hospital who had an Injury Severity Score (tSS) greater than From the Denver General Hospital, Departmentsof Surgical Critical 15 and who required more than 24 hours of mechanical Care (FAM), Surgery (AS, FAM, EEM, RAR), and Respiratory ventilation were prospectively evaluated for the developTherapy (JBH), Universityof Colorado Health SciencesCenter, Den- ment of infections and MOF. Exclusion criteria included ver, Colorado. Glasgow Coma Score lower than 8 due to head trauma This study was supported in part by grant No. P50-GM49222from the National Institutes of Health. Dr. Sauaia was supported by CNPq and age less than 16 years. The function of four organs (lung, liver, kidney, and heart) was evaluated on a daily grant No. 201113/910 (RE) from the BrazilianInstitute. Requests for reprints should be addressed to Frederick A. Moore, basis, and organ dysfunction was graded from 0 to 3 [19]. MD, Department of Surgery, DenverGeneral Hospital, 777 Bannock Lung dysfunction was quantitated by the adult respiraStreet, Denver,Colorado 80204-4507. Presented at the 45th Annual Meeting of the SouthwesternSurgi- tory distress syndrome (ARDS) score shown in Table L MOF was defined as the sum of the four individual organ cal Congress,Monterey,California,April 18-21, 1993. dysfunction grades, obtained simultaneously after 48 606
THE AMERICAN JOURNAL OF SURGERY VOLUME 166 DECEMBER 1993
PNEUMONIA AND POSTINJURY MULTIPLE ORGAN FAILUBF.
TABLE I
Postinjury Adult Respiratory Distress Syndrome Score
Pulmonary/radiographic
PaOa/FiO2 (mm Hg) Minute ventilation (L/min) PEEP (cm H20) Static compliance
Grade 1
Grade 2
Grade 3
Grade 4
Diffuse, mild interstitial marking/opacities 175-250 11-13 6-9 40-50
Diffuse, marked interstitial/ mild air-space opacities 125-174 14-16 10-13 30-39
Diffuse, moderate airspace consolidation 80-124 17-20 14-17 17-20
Diffuse, severe air-space consolidation < 80 > 20 > 17 < 20
PaO2/FiO2 = arterial partial pressure of oxygen/inspired fraction of oxygen; PEEP = positive end-expiratory pressure.
T A B L E II
Postinjury Multiple Organ Failure Score
Pulmonary dysfunction Renal dysfunction Hepatic dysfunction* Cardiac dysfunction1
Grade 1
Grade 2
Grade 3
ARDS score > 5 Creatinine > 1.8 mg/dL Bilirubin >2.0 mg/dL Minimal inotropes
ARDS score > 9 Creatinine > 2.5 mg/dL Bilirubin >4,0 mg/dL Moderate inotropes
ARDS score > 13 Creatinine > 5 mg/dL Bilirubin >8.0 mg/dL High inotropas
ARDS = adult respiratory distress syndrome, *Biliary obstruction and resolving hematoma not involved. TCardiac index < 3.0 L/min.m 2 requiring inotrop=csupport, Minimal dose = dopamine or dobutamine < 5 ~g/kg/min: moderate dose = dopamine or dobutamine 5 to 15 i~g/kg/min; high dose = greater than moderate doses of above agents.
hours of admission, greater than or equal to 4 (Table II). MOF was defined as early if present on postinjury day 3, or late if it occurred after day 3. Patients were monitored for the development of infectious and noninfectious complications. Septic complications were categorized as major or minor. Major infections included pneumonia, abscess, and fasciitis. Pneumonia was diagnosed based on the following criteria: (A) positive blood or pleural cultures for the same microorganism recovered in the tracheal aspirate; (B) new or progressive pulmonary infiltrate; (C) fever (higher than 38~ (D) leukocytosis (more than 10,000/mm3); (E) sputum Gram's Stain with more than 10 polymorphonuclear cells per high-power field; and (F) no other source of infection but the lungs. Pneumonia was defined as one of the following combinations:A + Bor B + C + D + Eor B + twoofC, D , E , + F [20]. Pneumonia was excluded when there was clinical resolution without antimicrobial therapy or when an alternative diagnosis was definitely made clinically or by autopsy. Abscess was defined as a purulent collection that required drainage. Fasciitis was defined as a wound infection requiring d6bridement. Minor infections were urinary tract infection (UTI), catheter and wound infections not requiring d6bridement, sinusitis, and conjunctivitis, diagnosed using the Centers for Disease Control definitions [21]. These complications were recorded on the day the definition criteria were met. To determine the potential causal relationship between the complication and MOF, we examined its temporal relationship to serial MOF scoring. We specifically compared the MOF score obtained on the day the complication was diagnosed and the MOF score obtained 48 to 72 hours later. Therefore,
complications could be classified in four categories: (1) not related because it occurred 4 or more days before the onset of MOF; (2) potential "trigger" if the MOF score on the day of diagnosis was less than 4 (i.e., no MOF) and rose 3 or more points within 48 to 72 hours; (3) worsening MOF if MOF was present on the day of diagnosis (i.e., MOF score greater than or equal to 4) and rose 3 or more points within 72 hours later; or (4) potential "symptom" if the complication occurred after MOF was present and was associated with a rise in the MOF score of less than 3. The results are expressed as mean 4- SEM. Comparisons of categorical variables were made using the x 2 test with Yates' correction or Fisher's exact test if any expected cell value was less than 5. The Mann-Whitney nonparametric test was used in comparing continuous variables. Differences were considered significant when the p value was less than 0.05. RESULTS During the study period, 123 trauma patients met the entry criteria. The mean patient age was 36.2 4- 1.5 years (range: 16 to 81 years), and there were 96 male patients (78%). The mechanism of injury was blunt in 82 patients (67%). The Injury Severity Score (ISS) ranged from 16 to 75 (mean 4- SEM: 26.4 4- 1.0). Ninety-three patients (76%) required at least 1 emergent/urgent operation (65 laparotomies, 19 thoracotomies, 17 orthopedic procedures, 6 neck explorations, and 5 major vascular repairs). Thirty-one patients (25%) had a systolic blood pressure lower than 90 mm Hg in the emergency department, and 29 (24%) required more than 6 units of packed red blood cells during the first 12 hours of admission. Twenty-eight
THE AMERICAN JOURNAL OF SURGERY
VOLUME 166 DECEMBER 1993
607
SAUAIA ET AL
MOF = multiple organ failure; NA = not applicable because the patient did not have a major infection (in these patients, MOF score "48 to 72 h later" denotes the worst score during hospital course).
cases of intra-abdominal abscess (IAA), 4 cases of fasciitis, and 1 case of empyema) occurred in 59 patients (48%). Thirty-three minor infections (13 UTIs, 9 wound infections, 6 catheter infections, 4 cases of sinusitis, and 1 case of conjunctivitis) occurred in 26 (21%) patients. An analysis of the breakdown of septic and nonseptic complications in patients with M O F versus patients without MOF is depicted in Table HI. MOF was associated with major infections (p <0.0001), minor infections (p = 0.003), and nonseptic complications (p = 0.004). Specific sources of infectious complications were then tested for association with MOF. Pneumonia was the only major infection associated with MOF: 27 episodes occurred in 23 patients with MOF (82%) compared with 32 episodes in 29 patients without M O F (82%) (p <0.0001). Among minor infections, only UTI was associated with MOF; eight patients with M O F (29%) developed UTI compared with five patients without MOF (5%) (p = 0.0015). No specific type of nonseptic complication was associated with MOF. Finally, we analyzed the temporal association of major infections, minor infections, and nonseptic complications with the onset or progression of MOF. Data for the 14 patients who developed early MOF are shown in Table IV. In four cases, pneumonia was temporally associated with worsening MOF; all patients died. In the other seven patients who developed pneumonia, pneumonia appeared to be a "symptom" (i.e., it was not related to a significant rise in the MOF score). Table V depicts the data for the 14 patients with late MOF. Four of these patients developed an IAA with or without pneumonia, and, in all cases, this appeared to "trigger" MOF. Nine other patients developed pneumonia, and, in eight patients, this was temporally related to the onset of MOF (five patients died). In regard to the 17 minor infections that occurred in 10 (36%) patients with MOF, 1 patient had a wound infection that preceded the onset of MOF by 4 days (i.e., "not related"), 2 patients had a UTI diagnosed at the same time as a major infection, and the remaining 7 patients had minor infections detected more than 72 hours after MOF occtirred that were associated with a minimal (less than 2) or no rise in MOF scoring. Similarly, nonseptic complications occurred frequently in patients with MOF. Two cases (one small bowel ischemia and one small bowel obstruction) were temporally related to a significant rise in the MOF score (Table IV) and, thus, may have "triggered" MOF. However, in the remaining MOF patients, nonseptic complications developed after MOF was established and were not associated with a significant rise in MOF scoring.
patients (23%) developed MOF. In 14 patients (50%), this occurred early (6 died), whereas in the remaining 14 patients, MOF occurred late (7 died). There were four additional deaths: three patients died of neurologic complications (nontrauma-related cerebral infarction, cerebral edema, and anoxic encephalopathy due to aspiration), and one died of isolated fulminant ARDS. Overall, 73 major infections (59 cases of pneumonia, 9
COMMENTS Despite extensive investigation, the pathogenesis of postinjury MOF is not clear [1-4]. A variety of anatomic insults (major abdominal trauma, multiple long bone or pelvic fractures, flail chest/pulmonary contusion), coupled with an early oxygen supply/demand imbalance, are known to place patients at risk for this morbid syndrome [22]. Although a single massive insult (one-hit model) can cause MOF, the more classic presentation is multiple
TABLE lIl
Septic and Nonseptic Complications in Torso Trauma Patients With and Without MOF MOF (n = 28) Major infections Pneumonia IAA Fasciitis Empyema
No MOF (n = 95)
27* } 4 25 patients 2 (89%)* 0
Minor infections UTI Wounr Catheter Sinusitis Conjunctivitis
8* 4 3 1 1
Nonseptic complications Cardiac Gastrointestinal Pulmonary Vascular Neurologic
3 3 3 2
32 5 2 1
34 patients (36%)
5 12 patients (43%)*
5 3 3 0
14 patients (15%)
10 patients
3 1
10 patients (10%)
(36%)*
1
1} 1
4
MOF = multiple organ failure: AA = intra-abdominal abscess; UTI = urinary tract infection. *o <0.05. MOF compared with no MOF group.
TABLE IV
Temporal Relationship Between MOF Scores and Major Infections in the 14 Patients With Torso Trauma Who Developed Early (_<3 Days) MOF
Major Infection None Pneumonia None Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Fasciitis
608
MOF Score 48 tc 72 h Onset Later NA 8 NA 6 5 4 4 4 6 4 4 6 6 6
4 11 4 2 2 11 11 5 9 6 5 6 3 8
Outcome Survived Died Survived Survived Survived Died Died Died Died Survived Survived Survived Survived Died
T H E A M E R I C A N J O U R N A L OF S U R G E R Y
V O L U M E 166
DECEMBER 1993
PNEUMONIAAND POSTIN~URYMULTIPLE ORGAN FAILURE
sequential insults (two-hit model) [23]. Following resuscitation, heterogenous trauma patients enter a similar phase of inflammation. This "injury stress response" is presumably beneficial and resolves in most patients as they recover clinically. However, in certain patients, this state becomes exaggerated, resulting in malignant systemic inflammation that culminates in MOF [5,6]. Traditionally, MOF has been viewed as the sequential cascade of organ failure that occurs as a result of uncontrolled sepsis [7,8]. This thought led to an aggressive policy of mandatory laparotomy in MOF patients to exclude an occult IAA [9]. Clearly, an untreated IAA can cause MOF; however, successful drainage of purulence once MOF is established does not ensure survival [13]. Moreover, it has become evident that patients can die of clinical sepsis (i.e., the "sepsis syndrome") without a culprit focus of infection [11,12]. More recently, SGD has been shown to reduce the incidence of infections (principally pneumonia) but has not consistently attenuated organ failure or decreased mortality [14-18]. Thus, these data have challenged the common belief that major infections are the prime delayed events that precipitate as well as perpetuate MOF. In our study, we observed a 48% incidence of major septic complications (83% were pneumonia) in patients with trauma to the torso with an ISS greater than 15 who required more than 24 hours of mechanical ventilation. The 43% incidence of pneumonia is nearly identical to those observed in two recent studies in similar groups of trauma patients [24,25]. Second, we observed that major infections (principally pneumonia), minor infections (principally UTI), and nonseptic complications were associated with MOF. Although the 89% incidence of major infections in MOF patients is high, this rate is similar to that reported by Fry et al [8] who were the first to emphasize the association between sepsis and MOF. A notable difference in our study, however, is that the incidence of IAA in MOF patients was only 14%, whereas the Louisville group observed a 44% incidence. This can be explained in part by the difference in patient populations, i.e., approximately one third of their patients underwent laparotomy for nontraumatic gastrointestinal emergencies. Moreover, in the past decade, we have witnessed tremendous advances in critical care, antibiotic therapy, and nutritional support as well as therapeutic and diagnostic radiology. Consequently, an IAA is less likely to be implicated in the pathogenesis of MOF. Indeed, Meakins et al [3] reported that, in 1979, two thirds of their patients dying of MOFhad IAA, whereas, in 1987, only 1 of their 26 patients with MOF had an IAA [3,12]. In contrast, pneumonia remains a frequent complication in these patients. The recent studies in SGD, however, suggest that pneumonia may not be the cause but rather a symptom of MOF [18]. Some of these studies can be criticized because they include patient populations in which pneumonia has a low incidence or little impact on patient outcome, e.g., patients with brain injuries [14-17]. In fact, we recently completed a prospective analysis of pneumonia in ventilated trauma patients [19]. As expected, pneumonia occurred frequently, but, although pneumonia was
TABLE V Temporal Relationship Between MOF Scores and Major Infections in the 1 4 Patients With Torso Trauma Who Developed L a t e ( > 3 Days) MOF MOF Score Major InfeCtion
Onset
48 to 72 h Later
Outcome
Pneumonia None IAA + pneumonia Pneumonia Pneumonia Pneumonia Pneumonia IAA Pneumonia Pneumonia Pneumonia Pneumonia IAA + pneumonia IAA
5 NA 3 3 0 0 2 3 2 1 3 1 0 1
6 4 8 11 12 4 7 11 7 4 7 5 6 4
Died Survived Died Survived Died Died Survived Survived Died Survived Died Died Survived Survived
MOF = multiple organ failure; IAA = intra-abdorninal abscess; NA = not applicable because the t~atient did not have a major infection (in these patients, MOF Score "48 to 72 h later" denotes the worst score during hospital course).
associated with MOF and death in patients with trauma to the torso, this association was not observed in patients with brain injuries. Consequently, in the present study, we excluded patients with isolated or severe head trauma. Again, we observed that pneumonia was associated with MOF in patients with major trauma to the torso. Moreover, by characterizing the temporal relationship between pneumonia and MOF scoring in this study, half of the cases of pneumonia can be implicated as triggering or worsening MOF. To better appreciate the impact of pneumonia in the pathogenesis of MOF, it is important to differentiate early and late MOF. Presumably, in early MOF, the initial tissue injury and shock are sufficient to initiate a malignant systemic inflammation response that culminates in MOF. These patients frequently develop pneumonia. Perhaps these represent symptoms of MOF (i.e., failure of barrier function and the immune system); however, one third are associated with significant worsening in MOF scoring and death. On the other hand, in patients who developed late MOF, pneumonia frequently precedes the onset of MOF by 48 to 72 hours and, therefore, can be implicated as a potential pathogenic factor. These results are similar to those reported by Waydhas et al [25] who also examined the temporal relationship between infection and postinjury organ failure. These authors found that late organ failure was preceded by an infection in 50% of their patients, and, in another 25%, it was detected at the s~ime time. In the group of patients with early onset MOF; infection followed the diagnosis of MOF in 35% of the patients, and it was associated with further deterioration of organ function in 44% of these patients. Similar findings were described by Faist et al [10] in a series of polytrauma patients. Collectively, these findings suggest that infection is related to the pattern of
THE AMERICAN JOURNAL OF SURGERY VOLUME 166 DECEMBER 1993 609
SAUAIA ET AL
onset of postinjury organ failure. Further investigation to characterize these two populations is critical, since different approaches may be necessary to prevent MOF. REFERENCES 1. Border JR. Multiple systems organ failure. Ann Surg 1992: 16: 111-6. 2. Deitch EA. Multiple organ failure: pathophysiology and potential future therapy. Ann Surg 1992; 216:117-34. 3. Meakins JL. Etiology of multiple organ failure. J Trauma 1990; 30: S165-8: 4. Carico JC, Meakins JL, Marshall JC, et al. Multiple organ failure. Arch Surg 1986; 121: 196-208. 5. Cerra FA. The systemic septic response: concepts of pathogenesis. J Trauma 1990; 30: S129-74. 6. Bone RC, Balk RA, Cerra FB, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest 1992; 101: 1644-55. 7. Eiseman B, Beart R, Norton L. Multiple organ failure. Surg Gynecol Obstet 1977; 144: 323-6. 8. Fry DE, Pearlstein L, Fulton RL, et al. Multiple system organ failure: the role of uncontrolled infection. Arch Surg 1980; 115: 136-40.
9. Ferraris VA. Exploratory laparotomy for potential abdominal sepsis in patients with multiple organ failure. Arch Surg 1983; 118: 1130-3.
10~ Faist E, Baue AE, Dittmer H, et aL Multiple organ failure in polytrauma patients. J Trauma 1983; 23: 775-86. 11. Gods ILIA, Boekholtz WKF, Bebber IPT, et al. Multiple organ failure and sepsis without bacteria. Arch Surg 1986; 121: 897-901. 12. Marshall JC, Chrislow NV, Horn R, et al. The microbiologyof multiple organ failure. The proximal gastrointestinal tract as an occult reservoir of pathogens. Arch Surg 1988; 123: 309-15. 13, Norton LW. Does drainage of intraabdominal pus reverse multiple organ failure? Am J Surg 1985; 149: 347-50. 14, Reidy J J, Ramsay G. Clinical trials of selective decontamination of the digestive tract: review. Crit Care Med 1990; 18: 1449-56. 15. Poole GV, Muakkassa FF, Griswold JA. Pneumonia, selective decontamination, and multiple organ failure. Surgery 1992; 111: 1-3. 16. van Saene HKF, Stoutenbeek CC, Stoiler JK. Selectivedecontamination of the digestive tract in the intensive care unit: current status and future prospects. Crit Care Med 1992; 20: 691-703. 17. Gastirme H, Wolf M, Delatour F, et aL A controlled trial of intensivecare units of selectivegut decontamination of the digestive tract with nonabsorbable antibiotics. N Engl J Met] 1992; 326: 594-9. 18. Cerra FB, Maddaus MA, Dunn DL, et al. Selectivegut decontaminationreduces nosocomialinfections and length of stay but not mortality or organ failure in surgical intensive care unit patients. Arch Surg 1992; 127: 163-9. 19. Sauaia A, Moore FA, Moore EE, et al. Pneumonia related multiple organ failure is not a primary cause of death in head trauma. Panamerican Journal of Trauma 1992; 3: 90-6. 20. Sauaia A, Moore FA, Moore EE, et al. Diagnosing pneumonia in mechanically ventilated trauma patients: quantitative cultures of endotracheal aspirate versus bronchoalveolar lavage. J Trauma 1993; 35: 512-7. 21. Gainec JS, Jarvis WR, Horan TC, et al. CDC definitions for nosecomial infections. Am J Infect Control 1988; 16: 128-40. 22. Moore FA, Haenel JB, Moore EE, et aL Incommensurate oxygen consumption in response to maximal oxygen availability predicts postinjury multiple organ failure. J Trauma 1992; 33: 1-9. 23. Anderson BO, Harken AH. Multiple organ failure: inflamma610
tory priming and activation sequences promote autogenous tissue injury. J Trauma 1990; 30: $44-9. 24, Rodriguez JL, Gibbons KJ, Bitzer LG, et aL Pneumonia: incidence, risk factors and outcome in injured patients. J Trauma 1991; 31: 907-14. 25. Waydhas C, Nast-Kolb D, Jochum M, et al. Inflammatory mediators, infection, sepsis, and multiple organ failure after severe trauma. Arch Surg 1991; 127: 460-7.
DISCUSSION M. Michael Shabot (Los Angeles, CA): In an era that long preceded the descriptions of intensive care units, adult respiratory distress syndrome, and multiple organ failure (MOF), Sir William Osier described pneumonia as "the captain of the men of death." You and your colleagues have presented an interesting study that tests this statement directly. Surely, Osier would recognize that MOF represents the "four horsemen" of current intensive care units (ICUs). You tested the null hypothesis that the development of pneumonia in patients with trauma to the torso is unrelated to the development of MOF. Your question is, does pneumonia cause MOF in trauma patients, or is it merely an epiphenomenon, a fellow traveler, in otherwise sick patients? I think this is an important issue. Despite, or because of, the development of potent antibiotics and sophisticated methods of ICU respiratory support, nosocomial pneumonia is now the third most common nosocomial infection in the United States and the most common cause of nosocomial death (Pennington JE. Hospital-acquired pneumonia. In: Wenzel RP, editor. Prevention and control of nosocomial infections. Baltimore: Williams & Wilkins, 1987:321-4). Fry and colleagues (Fry DE, et al: Multiple system organ failure. The role of uncontrolled infection. Arch Surg 1980; 115: 136-40) reported that invasive pulmonary infections occurred in more than half of the patients with MOF. In 1991, Bjerke and Shabot (Bjerke HS, et al: Impact of ICU nosoeomial infections on outcome from surgical care. Am Surg 1991; 57: 798802) reported that the development of ICU nosocomial infection was associated with a 12-fold increase in ICU stay, a 4-fold increase in hospital stay, and a 4-fold increase in mortality, significant increases that persisted even when the patients were stratified by severity of illness. Thus, on an a priori basis, it seems unlikely that you would find pneumonia to be an "innocent bystander" with regard to MOF in the ICU. In the present study, infectious complications were tabulated in 123 adult patients with torso trauma who had admission Glasgow Coma Scale scores greater than 8 and no isolated head injury. MOF was scored according to a four-organ dysfunction scale devised by the authors. The temporal relationship between the diagnosis of an infectious complication and MOF was determined by comparing the MOF score and the day the infection was diagnosed with the highest MOF score recorded 3 or more days later. Twenty-eight of the 123 study patients developed MOF, which was associated with pneumonia 82% of the time. Of the 14 patients who developed MOF early, within 3 days of ICU admission, only 4 had higher MOF
THE AMERICAN JOURNAL OF SURGERY VOLUME 166 DECEMBER 1993
PNEUMONIA AND POSTINJURY MULTIPLE ORGAN FAILURE
scores 3 or more days after developing pneumonia. In the more typical scenario is that, following resuscitation, less other 10 patients who developed MOF early, the MOF severely injured patients enter a beneficial state of hyperscore either declined or did not rise significantly after the inflammation, i.e., the injury stress response. However, development of pneumonia. Of the 14 patients who devel- certain patients appear to be vulnerable (i.e., primed) oped MOF late, 3 or more days after ICU admission, all such that a second inflammatory insult triggers delayed but 2 had a significant rise in the MOF score after the exaggerated systemic inflammation that culminates in onset of an infectious complication, principally pneumo- late MOF. Our data, by temporal association, implicate pneumonia as being one of the primary delayed activation nia. With that said, my questions to you are as follows: (1) stimuli for late MOF. Of course, a prospective, randomWhy does the development of a nosocomial infection or ized trial aimed at reducing pneumonia in patients at high pneumonia late in the ICU course have a greater impact risk for late MOF would be required to prove this. A on the MOF score, compared with that same infection major limitation of the SGD studies is that they include occurring early in the ICU course? (2) Has a causal patients in whom pneumonia does not impact on outrelationship between nosocomial pneumonia and organ come. For example, in a recent study, we observed that failure been proven by this study? Might both entities pneumonia occurs frequently in patients with head injusimply be the expression of a common, underlying factor, ries; however, the prime determinant of death was the perhaps acute depletion of immunoproteins, septic over- severity of the initial brain injury rather than late sepsisload, circulatory shock, or other insult? Finally, do you related MOF. Consequently, it is not surprising that trauhave any evidence that preventing pneumonia in torso ma-related SGD studies, in which half of the enrolled trauma patients with selective gut decontamination patients have head injuries, have failed to show a favor(SGD) or any other therapy will also prevent the develop- able impact on outcome. However, these data should not be used to conclude that SGD is not valuable. In fact, ment of MOF? David Felieiano (Atlanta, GA): Did you use any type SGD has consistently been shown to reduce nosocomial of prophylaxis, and did you look at pneumonia as an infection, which is associated with reduced cost and independent variable in the patients who died of MOF? length of ICU stay. We do not use SGD because we Frederick A. Moore (closing): Infection has long believe that early enteral nutrition achieves the same been recognized to be a prime inciting event for MOF, benefits. In regard to the question of whether pneumonia but MOF can occur in the absence of infection. There- is an independent predictor of MOF, we are currently fore, it is not surprising that pneumonia plays a less signif- involved in a prospective epidemiologic study of postinjuicant role in early MOF where a massive traumatic insult ry MOF. One of the goals is to develop a predictive model precipitates early malignant systemic inflammation. In for MOF, and the addition of pneumonia in our logistic this setting, pneumonia is frequently a delayed event, but, regression analysis does increase our ability to predict MOF. This logistic regression analysis, however, does not in two thirds of the cases, it appears to be inconsequential, presumably because it does not amplify the existing auto- answer the question of whether pneumonia is the cause or destructive inflammatory state. On the other hand, the just a sign of MOF. Consequently, we did this study.
THE AMERICAN JOURNAL OF SURGERY
VOLUME 166 DECEMBER 1993
611
Recent studies have shown that selective gut decontamination can reduce the incidence of pneumonia, but this does not decrease multiple organ failure (MOF) or mortality. These findings have prompted the hypothesis that pneumonia is an inconsequential symptom of MOF. To test this, we prospectively evaluated 123 high-risk trauma patients (mean Injury Severity Score = 36.2 q- 1.5). Organ dysfunction, scored daily according to a 12-point scale, ultimately developed in 28 (23%) patients. Major infectious were diagnosed, based on strict criteria, in 59 patients (48%), and pneumonia developed in 52 patients (43%). Pneumonia was significantly associated with MOF (82% of patients with MOF versus 30% of patients without MOF, p < 0 . 0 0 0 1 ) . In 14 (50%) of the patients with MOF, pneumonia preceded a significant rise (greater than or equal to 3) in serial MOF scoring. Of note, 10 (71%) of these patients died. Among the remaining 14 patients with MOF, 10 developed pneumonia, but this was associated with a minimal increase (less than or equal to 2) in MOF scoring (3 patients died). These data, by temporal association with MOF scoring, implicate pneumonia in precipitating or significantly worsening organ failure in 50% of the patients who developed MOF.
' ultiple organ failure (MOF) remains the leading M . cause of late postinjury death, but, despite intensive investigation, its pathogenesis remains elusive [1-4]. Following resuscitation, heterogenous trauma patients enter a similar phase of hyperinflammation. This is presumably beneficial (i.e., the injury stress response) and, in most cases, resolves as the patient recovers. What triggers the transition into the autodestructive systemic inflammation that characterizes postinjury MOF is not clear [5,6]. Traditionally, infection has been viewed as the pivotal delayed event that precipitates MOF [7-9]. However, it has been recognized that MOF can occur without an identifiable focus of infection and that MOF will frequently not resolve despite successful drainage of abscesses [10-13]. More recently, prospective, randomized trials have shown that selective gut decontamination (SGD) can reduce the incidence of infections (principally pneumonia) in patients requiring prolonged mechanical ventilation but offer conflicting results regarding the impact on organ failure or mortality [14-18]. These findings have prompted the hypothesis that pneumonia is merely an inconsequential symptom of MOF. This has important implications in directing future strategies in preventing MOF. Consequently, we designed this study to investigate the potential causal relationship between infectious complications (principally pneumonia) and MOF in patients sustaining major trauma to the torso who required prolonged mechanical ventilation. Our specific aims were as follows: (1) to prospectively document the incidence of pneumonia in this trauma population; (2) to determine if pneumonia is associated with MOF; and (3) to characterize the temporal relationship between pneumonia and organ failure scoring to determine whether pneumofiia could precipitate or worsen MOF.
PATIENTS AND METHODS During the 2-year period ending December 1992, all patients with injuries to the torso admitted to the surgical intensive care unit (SICU) of the Denver General Hospital who had an Injury Severity Score (tSS) greater than From the Denver General Hospital, Departmentsof Surgical Critical 15 and who required more than 24 hours of mechanical Care (FAM), Surgery (AS, FAM, EEM, RAR), and Respiratory ventilation were prospectively evaluated for the developTherapy (JBH), Universityof Colorado Health SciencesCenter, Den- ment of infections and MOF. Exclusion criteria included ver, Colorado. Glasgow Coma Score lower than 8 due to head trauma This study was supported in part by grant No. P50-GM49222from the National Institutes of Health. Dr. Sauaia was supported by CNPq and age less than 16 years. The function of four organs (lung, liver, kidney, and heart) was evaluated on a daily grant No. 201113/910 (RE) from the BrazilianInstitute. Requests for reprints should be addressed to Frederick A. Moore, basis, and organ dysfunction was graded from 0 to 3 [19]. MD, Department of Surgery, DenverGeneral Hospital, 777 Bannock Lung dysfunction was quantitated by the adult respiraStreet, Denver,Colorado 80204-4507. Presented at the 45th Annual Meeting of the SouthwesternSurgi- tory distress syndrome (ARDS) score shown in Table L MOF was defined as the sum of the four individual organ cal Congress,Monterey,California,April 18-21, 1993. dysfunction grades, obtained simultaneously after 48 606
THE AMERICAN JOURNAL OF SURGERY VOLUME 166 DECEMBER 1993
PNEUMONIA AND POSTINJURY MULTIPLE ORGAN FAILUBF.
TABLE I
Postinjury Adult Respiratory Distress Syndrome Score
Pulmonary/radiographic
PaOa/FiO2 (mm Hg) Minute ventilation (L/min) PEEP (cm H20) Static compliance
Grade 1
Grade 2
Grade 3
Grade 4
Diffuse, mild interstitial marking/opacities 175-250 11-13 6-9 40-50
Diffuse, marked interstitial/ mild air-space opacities 125-174 14-16 10-13 30-39
Diffuse, moderate airspace consolidation 80-124 17-20 14-17 17-20
Diffuse, severe air-space consolidation < 80 > 20 > 17 < 20
PaO2/FiO2 = arterial partial pressure of oxygen/inspired fraction of oxygen; PEEP = positive end-expiratory pressure.
T A B L E II
Postinjury Multiple Organ Failure Score
Pulmonary dysfunction Renal dysfunction Hepatic dysfunction* Cardiac dysfunction1
Grade 1
Grade 2
Grade 3
ARDS score > 5 Creatinine > 1.8 mg/dL Bilirubin >2.0 mg/dL Minimal inotropes
ARDS score > 9 Creatinine > 2.5 mg/dL Bilirubin >4,0 mg/dL Moderate inotropes
ARDS score > 13 Creatinine > 5 mg/dL Bilirubin >8.0 mg/dL High inotropas
ARDS = adult respiratory distress syndrome, *Biliary obstruction and resolving hematoma not involved. TCardiac index < 3.0 L/min.m 2 requiring inotrop=csupport, Minimal dose = dopamine or dobutamine < 5 ~g/kg/min: moderate dose = dopamine or dobutamine 5 to 15 i~g/kg/min; high dose = greater than moderate doses of above agents.
hours of admission, greater than or equal to 4 (Table II). MOF was defined as early if present on postinjury day 3, or late if it occurred after day 3. Patients were monitored for the development of infectious and noninfectious complications. Septic complications were categorized as major or minor. Major infections included pneumonia, abscess, and fasciitis. Pneumonia was diagnosed based on the following criteria: (A) positive blood or pleural cultures for the same microorganism recovered in the tracheal aspirate; (B) new or progressive pulmonary infiltrate; (C) fever (higher than 38~ (D) leukocytosis (more than 10,000/mm3); (E) sputum Gram's Stain with more than 10 polymorphonuclear cells per high-power field; and (F) no other source of infection but the lungs. Pneumonia was defined as one of the following combinations:A + Bor B + C + D + Eor B + twoofC, D , E , + F [20]. Pneumonia was excluded when there was clinical resolution without antimicrobial therapy or when an alternative diagnosis was definitely made clinically or by autopsy. Abscess was defined as a purulent collection that required drainage. Fasciitis was defined as a wound infection requiring d6bridement. Minor infections were urinary tract infection (UTI), catheter and wound infections not requiring d6bridement, sinusitis, and conjunctivitis, diagnosed using the Centers for Disease Control definitions [21]. These complications were recorded on the day the definition criteria were met. To determine the potential causal relationship between the complication and MOF, we examined its temporal relationship to serial MOF scoring. We specifically compared the MOF score obtained on the day the complication was diagnosed and the MOF score obtained 48 to 72 hours later. Therefore,
complications could be classified in four categories: (1) not related because it occurred 4 or more days before the onset of MOF; (2) potential "trigger" if the MOF score on the day of diagnosis was less than 4 (i.e., no MOF) and rose 3 or more points within 48 to 72 hours; (3) worsening MOF if MOF was present on the day of diagnosis (i.e., MOF score greater than or equal to 4) and rose 3 or more points within 72 hours later; or (4) potential "symptom" if the complication occurred after MOF was present and was associated with a rise in the MOF score of less than 3. The results are expressed as mean 4- SEM. Comparisons of categorical variables were made using the x 2 test with Yates' correction or Fisher's exact test if any expected cell value was less than 5. The Mann-Whitney nonparametric test was used in comparing continuous variables. Differences were considered significant when the p value was less than 0.05. RESULTS During the study period, 123 trauma patients met the entry criteria. The mean patient age was 36.2 4- 1.5 years (range: 16 to 81 years), and there were 96 male patients (78%). The mechanism of injury was blunt in 82 patients (67%). The Injury Severity Score (ISS) ranged from 16 to 75 (mean 4- SEM: 26.4 4- 1.0). Ninety-three patients (76%) required at least 1 emergent/urgent operation (65 laparotomies, 19 thoracotomies, 17 orthopedic procedures, 6 neck explorations, and 5 major vascular repairs). Thirty-one patients (25%) had a systolic blood pressure lower than 90 mm Hg in the emergency department, and 29 (24%) required more than 6 units of packed red blood cells during the first 12 hours of admission. Twenty-eight
THE AMERICAN JOURNAL OF SURGERY
VOLUME 166 DECEMBER 1993
607
SAUAIA ET AL
MOF = multiple organ failure; NA = not applicable because the patient did not have a major infection (in these patients, MOF score "48 to 72 h later" denotes the worst score during hospital course).
cases of intra-abdominal abscess (IAA), 4 cases of fasciitis, and 1 case of empyema) occurred in 59 patients (48%). Thirty-three minor infections (13 UTIs, 9 wound infections, 6 catheter infections, 4 cases of sinusitis, and 1 case of conjunctivitis) occurred in 26 (21%) patients. An analysis of the breakdown of septic and nonseptic complications in patients with M O F versus patients without MOF is depicted in Table HI. MOF was associated with major infections (p <0.0001), minor infections (p = 0.003), and nonseptic complications (p = 0.004). Specific sources of infectious complications were then tested for association with MOF. Pneumonia was the only major infection associated with MOF: 27 episodes occurred in 23 patients with MOF (82%) compared with 32 episodes in 29 patients without M O F (82%) (p <0.0001). Among minor infections, only UTI was associated with MOF; eight patients with M O F (29%) developed UTI compared with five patients without MOF (5%) (p = 0.0015). No specific type of nonseptic complication was associated with MOF. Finally, we analyzed the temporal association of major infections, minor infections, and nonseptic complications with the onset or progression of MOF. Data for the 14 patients who developed early MOF are shown in Table IV. In four cases, pneumonia was temporally associated with worsening MOF; all patients died. In the other seven patients who developed pneumonia, pneumonia appeared to be a "symptom" (i.e., it was not related to a significant rise in the MOF score). Table V depicts the data for the 14 patients with late MOF. Four of these patients developed an IAA with or without pneumonia, and, in all cases, this appeared to "trigger" MOF. Nine other patients developed pneumonia, and, in eight patients, this was temporally related to the onset of MOF (five patients died). In regard to the 17 minor infections that occurred in 10 (36%) patients with MOF, 1 patient had a wound infection that preceded the onset of MOF by 4 days (i.e., "not related"), 2 patients had a UTI diagnosed at the same time as a major infection, and the remaining 7 patients had minor infections detected more than 72 hours after MOF occtirred that were associated with a minimal (less than 2) or no rise in MOF scoring. Similarly, nonseptic complications occurred frequently in patients with MOF. Two cases (one small bowel ischemia and one small bowel obstruction) were temporally related to a significant rise in the MOF score (Table IV) and, thus, may have "triggered" MOF. However, in the remaining MOF patients, nonseptic complications developed after MOF was established and were not associated with a significant rise in MOF scoring.
patients (23%) developed MOF. In 14 patients (50%), this occurred early (6 died), whereas in the remaining 14 patients, MOF occurred late (7 died). There were four additional deaths: three patients died of neurologic complications (nontrauma-related cerebral infarction, cerebral edema, and anoxic encephalopathy due to aspiration), and one died of isolated fulminant ARDS. Overall, 73 major infections (59 cases of pneumonia, 9
COMMENTS Despite extensive investigation, the pathogenesis of postinjury MOF is not clear [1-4]. A variety of anatomic insults (major abdominal trauma, multiple long bone or pelvic fractures, flail chest/pulmonary contusion), coupled with an early oxygen supply/demand imbalance, are known to place patients at risk for this morbid syndrome [22]. Although a single massive insult (one-hit model) can cause MOF, the more classic presentation is multiple
TABLE lIl
Septic and Nonseptic Complications in Torso Trauma Patients With and Without MOF MOF (n = 28) Major infections Pneumonia IAA Fasciitis Empyema
No MOF (n = 95)
27* } 4 25 patients 2 (89%)* 0
Minor infections UTI Wounr Catheter Sinusitis Conjunctivitis
8* 4 3 1 1
Nonseptic complications Cardiac Gastrointestinal Pulmonary Vascular Neurologic
3 3 3 2
32 5 2 1
34 patients (36%)
5 12 patients (43%)*
5 3 3 0
14 patients (15%)
10 patients
3 1
10 patients (10%)
(36%)*
1
1} 1
4
MOF = multiple organ failure: AA = intra-abdominal abscess; UTI = urinary tract infection. *o <0.05. MOF compared with no MOF group.
TABLE IV
Temporal Relationship Between MOF Scores and Major Infections in the 14 Patients With Torso Trauma Who Developed Early (_<3 Days) MOF
Major Infection None Pneumonia None Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Pneumonia Fasciitis
608
MOF Score 48 tc 72 h Onset Later NA 8 NA 6 5 4 4 4 6 4 4 6 6 6
4 11 4 2 2 11 11 5 9 6 5 6 3 8
Outcome Survived Died Survived Survived Survived Died Died Died Died Survived Survived Survived Survived Died
T H E A M E R I C A N J O U R N A L OF S U R G E R Y
V O L U M E 166
DECEMBER 1993
PNEUMONIAAND POSTIN~URYMULTIPLE ORGAN FAILURE
sequential insults (two-hit model) [23]. Following resuscitation, heterogenous trauma patients enter a similar phase of inflammation. This "injury stress response" is presumably beneficial and resolves in most patients as they recover clinically. However, in certain patients, this state becomes exaggerated, resulting in malignant systemic inflammation that culminates in MOF [5,6]. Traditionally, MOF has been viewed as the sequential cascade of organ failure that occurs as a result of uncontrolled sepsis [7,8]. This thought led to an aggressive policy of mandatory laparotomy in MOF patients to exclude an occult IAA [9]. Clearly, an untreated IAA can cause MOF; however, successful drainage of purulence once MOF is established does not ensure survival [13]. Moreover, it has become evident that patients can die of clinical sepsis (i.e., the "sepsis syndrome") without a culprit focus of infection [11,12]. More recently, SGD has been shown to reduce the incidence of infections (principally pneumonia) but has not consistently attenuated organ failure or decreased mortality [14-18]. Thus, these data have challenged the common belief that major infections are the prime delayed events that precipitate as well as perpetuate MOF. In our study, we observed a 48% incidence of major septic complications (83% were pneumonia) in patients with trauma to the torso with an ISS greater than 15 who required more than 24 hours of mechanical ventilation. The 43% incidence of pneumonia is nearly identical to those observed in two recent studies in similar groups of trauma patients [24,25]. Second, we observed that major infections (principally pneumonia), minor infections (principally UTI), and nonseptic complications were associated with MOF. Although the 89% incidence of major infections in MOF patients is high, this rate is similar to that reported by Fry et al [8] who were the first to emphasize the association between sepsis and MOF. A notable difference in our study, however, is that the incidence of IAA in MOF patients was only 14%, whereas the Louisville group observed a 44% incidence. This can be explained in part by the difference in patient populations, i.e., approximately one third of their patients underwent laparotomy for nontraumatic gastrointestinal emergencies. Moreover, in the past decade, we have witnessed tremendous advances in critical care, antibiotic therapy, and nutritional support as well as therapeutic and diagnostic radiology. Consequently, an IAA is less likely to be implicated in the pathogenesis of MOF. Indeed, Meakins et al [3] reported that, in 1979, two thirds of their patients dying of MOFhad IAA, whereas, in 1987, only 1 of their 26 patients with MOF had an IAA [3,12]. In contrast, pneumonia remains a frequent complication in these patients. The recent studies in SGD, however, suggest that pneumonia may not be the cause but rather a symptom of MOF [18]. Some of these studies can be criticized because they include patient populations in which pneumonia has a low incidence or little impact on patient outcome, e.g., patients with brain injuries [14-17]. In fact, we recently completed a prospective analysis of pneumonia in ventilated trauma patients [19]. As expected, pneumonia occurred frequently, but, although pneumonia was
TABLE V Temporal Relationship Between MOF Scores and Major Infections in the 1 4 Patients With Torso Trauma Who Developed L a t e ( > 3 Days) MOF MOF Score Major InfeCtion
Onset
48 to 72 h Later
Outcome
Pneumonia None IAA + pneumonia Pneumonia Pneumonia Pneumonia Pneumonia IAA Pneumonia Pneumonia Pneumonia Pneumonia IAA + pneumonia IAA
5 NA 3 3 0 0 2 3 2 1 3 1 0 1
6 4 8 11 12 4 7 11 7 4 7 5 6 4
Died Survived Died Survived Died Died Survived Survived Died Survived Died Died Survived Survived
MOF = multiple organ failure; IAA = intra-abdorninal abscess; NA = not applicable because the t~atient did not have a major infection (in these patients, MOF Score "48 to 72 h later" denotes the worst score during hospital course).
associated with MOF and death in patients with trauma to the torso, this association was not observed in patients with brain injuries. Consequently, in the present study, we excluded patients with isolated or severe head trauma. Again, we observed that pneumonia was associated with MOF in patients with major trauma to the torso. Moreover, by characterizing the temporal relationship between pneumonia and MOF scoring in this study, half of the cases of pneumonia can be implicated as triggering or worsening MOF. To better appreciate the impact of pneumonia in the pathogenesis of MOF, it is important to differentiate early and late MOF. Presumably, in early MOF, the initial tissue injury and shock are sufficient to initiate a malignant systemic inflammation response that culminates in MOF. These patients frequently develop pneumonia. Perhaps these represent symptoms of MOF (i.e., failure of barrier function and the immune system); however, one third are associated with significant worsening in MOF scoring and death. On the other hand, in patients who developed late MOF, pneumonia frequently precedes the onset of MOF by 48 to 72 hours and, therefore, can be implicated as a potential pathogenic factor. These results are similar to those reported by Waydhas et al [25] who also examined the temporal relationship between infection and postinjury organ failure. These authors found that late organ failure was preceded by an infection in 50% of their patients, and, in another 25%, it was detected at the s~ime time. In the group of patients with early onset MOF; infection followed the diagnosis of MOF in 35% of the patients, and it was associated with further deterioration of organ function in 44% of these patients. Similar findings were described by Faist et al [10] in a series of polytrauma patients. Collectively, these findings suggest that infection is related to the pattern of
THE AMERICAN JOURNAL OF SURGERY VOLUME 166 DECEMBER 1993 609
SAUAIA ET AL
onset of postinjury organ failure. Further investigation to characterize these two populations is critical, since different approaches may be necessary to prevent MOF. REFERENCES 1. Border JR. Multiple systems organ failure. Ann Surg 1992: 16: 111-6. 2. Deitch EA. Multiple organ failure: pathophysiology and potential future therapy. Ann Surg 1992; 216:117-34. 3. Meakins JL. Etiology of multiple organ failure. J Trauma 1990; 30: S165-8: 4. Carico JC, Meakins JL, Marshall JC, et al. Multiple organ failure. Arch Surg 1986; 121: 196-208. 5. Cerra FA. The systemic septic response: concepts of pathogenesis. J Trauma 1990; 30: S129-74. 6. Bone RC, Balk RA, Cerra FB, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest 1992; 101: 1644-55. 7. Eiseman B, Beart R, Norton L. Multiple organ failure. Surg Gynecol Obstet 1977; 144: 323-6. 8. Fry DE, Pearlstein L, Fulton RL, et al. Multiple system organ failure: the role of uncontrolled infection. Arch Surg 1980; 115: 136-40.
9. Ferraris VA. Exploratory laparotomy for potential abdominal sepsis in patients with multiple organ failure. Arch Surg 1983; 118: 1130-3.
10~ Faist E, Baue AE, Dittmer H, et aL Multiple organ failure in polytrauma patients. J Trauma 1983; 23: 775-86. 11. Gods ILIA, Boekholtz WKF, Bebber IPT, et al. Multiple organ failure and sepsis without bacteria. Arch Surg 1986; 121: 897-901. 12. Marshall JC, Chrislow NV, Horn R, et al. The microbiologyof multiple organ failure. The proximal gastrointestinal tract as an occult reservoir of pathogens. Arch Surg 1988; 123: 309-15. 13, Norton LW. Does drainage of intraabdominal pus reverse multiple organ failure? Am J Surg 1985; 149: 347-50. 14, Reidy J J, Ramsay G. Clinical trials of selective decontamination of the digestive tract: review. Crit Care Med 1990; 18: 1449-56. 15. Poole GV, Muakkassa FF, Griswold JA. Pneumonia, selective decontamination, and multiple organ failure. Surgery 1992; 111: 1-3. 16. van Saene HKF, Stoutenbeek CC, Stoiler JK. Selectivedecontamination of the digestive tract in the intensive care unit: current status and future prospects. Crit Care Med 1992; 20: 691-703. 17. Gastirme H, Wolf M, Delatour F, et aL A controlled trial of intensivecare units of selectivegut decontamination of the digestive tract with nonabsorbable antibiotics. N Engl J Met] 1992; 326: 594-9. 18. Cerra FB, Maddaus MA, Dunn DL, et al. Selectivegut decontaminationreduces nosocomialinfections and length of stay but not mortality or organ failure in surgical intensive care unit patients. Arch Surg 1992; 127: 163-9. 19. Sauaia A, Moore FA, Moore EE, et al. Pneumonia related multiple organ failure is not a primary cause of death in head trauma. Panamerican Journal of Trauma 1992; 3: 90-6. 20. Sauaia A, Moore FA, Moore EE, et al. Diagnosing pneumonia in mechanically ventilated trauma patients: quantitative cultures of endotracheal aspirate versus bronchoalveolar lavage. J Trauma 1993; 35: 512-7. 21. Gainec JS, Jarvis WR, Horan TC, et al. CDC definitions for nosecomial infections. Am J Infect Control 1988; 16: 128-40. 22. Moore FA, Haenel JB, Moore EE, et aL Incommensurate oxygen consumption in response to maximal oxygen availability predicts postinjury multiple organ failure. J Trauma 1992; 33: 1-9. 23. Anderson BO, Harken AH. Multiple organ failure: inflamma610
tory priming and activation sequences promote autogenous tissue injury. J Trauma 1990; 30: $44-9. 24, Rodriguez JL, Gibbons KJ, Bitzer LG, et aL Pneumonia: incidence, risk factors and outcome in injured patients. J Trauma 1991; 31: 907-14. 25. Waydhas C, Nast-Kolb D, Jochum M, et al. Inflammatory mediators, infection, sepsis, and multiple organ failure after severe trauma. Arch Surg 1991; 127: 460-7.
DISCUSSION M. Michael Shabot (Los Angeles, CA): In an era that long preceded the descriptions of intensive care units, adult respiratory distress syndrome, and multiple organ failure (MOF), Sir William Osier described pneumonia as "the captain of the men of death." You and your colleagues have presented an interesting study that tests this statement directly. Surely, Osier would recognize that MOF represents the "four horsemen" of current intensive care units (ICUs). You tested the null hypothesis that the development of pneumonia in patients with trauma to the torso is unrelated to the development of MOF. Your question is, does pneumonia cause MOF in trauma patients, or is it merely an epiphenomenon, a fellow traveler, in otherwise sick patients? I think this is an important issue. Despite, or because of, the development of potent antibiotics and sophisticated methods of ICU respiratory support, nosocomial pneumonia is now the third most common nosocomial infection in the United States and the most common cause of nosocomial death (Pennington JE. Hospital-acquired pneumonia. In: Wenzel RP, editor. Prevention and control of nosocomial infections. Baltimore: Williams & Wilkins, 1987:321-4). Fry and colleagues (Fry DE, et al: Multiple system organ failure. The role of uncontrolled infection. Arch Surg 1980; 115: 136-40) reported that invasive pulmonary infections occurred in more than half of the patients with MOF. In 1991, Bjerke and Shabot (Bjerke HS, et al: Impact of ICU nosoeomial infections on outcome from surgical care. Am Surg 1991; 57: 798802) reported that the development of ICU nosocomial infection was associated with a 12-fold increase in ICU stay, a 4-fold increase in hospital stay, and a 4-fold increase in mortality, significant increases that persisted even when the patients were stratified by severity of illness. Thus, on an a priori basis, it seems unlikely that you would find pneumonia to be an "innocent bystander" with regard to MOF in the ICU. In the present study, infectious complications were tabulated in 123 adult patients with torso trauma who had admission Glasgow Coma Scale scores greater than 8 and no isolated head injury. MOF was scored according to a four-organ dysfunction scale devised by the authors. The temporal relationship between the diagnosis of an infectious complication and MOF was determined by comparing the MOF score and the day the infection was diagnosed with the highest MOF score recorded 3 or more days later. Twenty-eight of the 123 study patients developed MOF, which was associated with pneumonia 82% of the time. Of the 14 patients who developed MOF early, within 3 days of ICU admission, only 4 had higher MOF
THE AMERICAN JOURNAL OF SURGERY VOLUME 166 DECEMBER 1993
PNEUMONIA AND POSTINJURY MULTIPLE ORGAN FAILURE
scores 3 or more days after developing pneumonia. In the more typical scenario is that, following resuscitation, less other 10 patients who developed MOF early, the MOF severely injured patients enter a beneficial state of hyperscore either declined or did not rise significantly after the inflammation, i.e., the injury stress response. However, development of pneumonia. Of the 14 patients who devel- certain patients appear to be vulnerable (i.e., primed) oped MOF late, 3 or more days after ICU admission, all such that a second inflammatory insult triggers delayed but 2 had a significant rise in the MOF score after the exaggerated systemic inflammation that culminates in onset of an infectious complication, principally pneumo- late MOF. Our data, by temporal association, implicate pneumonia as being one of the primary delayed activation nia. With that said, my questions to you are as follows: (1) stimuli for late MOF. Of course, a prospective, randomWhy does the development of a nosocomial infection or ized trial aimed at reducing pneumonia in patients at high pneumonia late in the ICU course have a greater impact risk for late MOF would be required to prove this. A on the MOF score, compared with that same infection major limitation of the SGD studies is that they include occurring early in the ICU course? (2) Has a causal patients in whom pneumonia does not impact on outrelationship between nosocomial pneumonia and organ come. For example, in a recent study, we observed that failure been proven by this study? Might both entities pneumonia occurs frequently in patients with head injusimply be the expression of a common, underlying factor, ries; however, the prime determinant of death was the perhaps acute depletion of immunoproteins, septic over- severity of the initial brain injury rather than late sepsisload, circulatory shock, or other insult? Finally, do you related MOF. Consequently, it is not surprising that trauhave any evidence that preventing pneumonia in torso ma-related SGD studies, in which half of the enrolled trauma patients with selective gut decontamination patients have head injuries, have failed to show a favor(SGD) or any other therapy will also prevent the develop- able impact on outcome. However, these data should not be used to conclude that SGD is not valuable. In fact, ment of MOF? David Felieiano (Atlanta, GA): Did you use any type SGD has consistently been shown to reduce nosocomial of prophylaxis, and did you look at pneumonia as an infection, which is associated with reduced cost and independent variable in the patients who died of MOF? length of ICU stay. We do not use SGD because we Frederick A. Moore (closing): Infection has long believe that early enteral nutrition achieves the same been recognized to be a prime inciting event for MOF, benefits. In regard to the question of whether pneumonia but MOF can occur in the absence of infection. There- is an independent predictor of MOF, we are currently fore, it is not surprising that pneumonia plays a less signif- involved in a prospective epidemiologic study of postinjuicant role in early MOF where a massive traumatic insult ry MOF. One of the goals is to develop a predictive model precipitates early malignant systemic inflammation. In for MOF, and the addition of pneumonia in our logistic this setting, pneumonia is frequently a delayed event, but, regression analysis does increase our ability to predict MOF. This logistic regression analysis, however, does not in two thirds of the cases, it appears to be inconsequential, presumably because it does not amplify the existing auto- answer the question of whether pneumonia is the cause or destructive inflammatory state. On the other hand, the just a sign of MOF. Consequently, we did this study.
THE AMERICAN JOURNAL OF SURGERY
VOLUME 166 DECEMBER 1993
611