Risk Factors for Early Onset Pneumonia in Trauma Patients

Risk Factors for Early Onset Pneumonia in Trauma Patients* Massimo Antonelli, M.D.; Maria Luisa Moro, M.D.; Ornella Capelli, M.D.; Roberto A. De Blasi, M.D.; Rosa Rita D'Errico, M.D.; Giorgio Conti, M.D.; Maurizio Bufi., M.D.; and Alessandro Gasparetto, M.D. Study objectives: The aim of the study was to identify ruk factors for early onset pneumonia (EOP) in trauma patients.

in order to seek possible intervention strategies. Study population: Participants included 124 consecutive trauma patients admitted to a general intensive care unit (ICU) of a university hospital from December 1990 to February 1992 inclusive. Data collection: The foDowing data were prospectively collected for each patient: demographics, severity oftrauma according to the abbreviated injury scale (AIS), severity of coma according to the Glasgow coma scale (GCS), presence of pnemnotborax, pulmonary contusion. rib fractures, hemothorax, and mechanical ventilation. AD patients were monitored daily during the ICU stay for the onset of pneumonia, sepsis syndrome, septic shock, and adult respiratory distress syndrome (ARDS). Criteria for the
infections represent a common compliN osocomial cation among critically ill patients, especially those requiring mechanical ventilation ( MV) .1 Green et al 2 and Daniele3 reported a high incidence of pulmonary infections among multiple trauma patients, attributable to a decrease in immunologic defenses. The risk of bacterial infections in trauma patients seems to depend upon several factors. (1) Upper airway colonization is a relatively frequent phenomenon in this group of patients. (2) Trauma lesions requiring surgical treatment are at high risk for *From the lstituto di Anestesiologia e Rianimazione, Universita "La Sapienza" (Drs. Antonelli, Capelli, De Blasi, D'Errico, Conti, Bufi, and Gasparetto), and Laboratorio di Epidemiologia e Biostatistica, Istituto Superiore di Sanita (Dr. Moro), Rome,ltaly. Manuscript received February 17, 1993; revision accepted May 27. Reprint requests: Dr. Antonelli, Istituto di Anestesiologia e Rianimazione, Policlinico Umberto I, Viole del Policlinico 155, 00161 Rome, Italy

224

Results: Overall mortality was 43.5 percent: mortality increased by age and AIS score. Forty one patients (33.1 percent) developed pneumonia: 26 (63.4 percent) were EOP and 15 (36.6 percent) were LOP. In the univariate analysis, an age greater than 40 years, the presence of pulmonary contusion, AIS of more than 4 for thorax and of more than 9 for abdomen, and the absence of mechanical ventilation (MV) during the first 4 days of hospitalization or MV lasting less than 24 h were significantly associated with an increased risk of acquiring EOP. Logistic regression analysis showed that the strongest risk factor for EOP was a combined severe abdominal and thoracic trauma, which increased the risk of EOP by 11 times; an age of more than 40 years and MV of less than 24 h during the first 4 days of hospitalization were also independent risk factors for EOP. Factors associated with LOP were an AIS score of more than 4 for abdomen and a length of MV of more than 5 days. Conclusion: In a trauma population, a combined severe abdominal and thoracic trauma represents a major risk factor for EOP. Mechanical ventilation administered during the first days after trauma seems to reduce the risk ofEOP. As reported in previous studies, mechanical ventilatory support lasting more than 5 days is associated with an increased risk of LOP. (Chest 1994; 105:224-28) AIS =abbreviated injury scale; ARDS = adult respiratory distress syndrome; EOP = early onset pneumonia; GCS = GlaSgow coma scale; LOP = late onset pneumonia; MV =mechanical ventilation

infection.4 (3) Trauma per se induces an alteration of the natural antimicrobial barrier. (4) Sedatives and barbiturates used in patients with cerebral edema exert an immunosuppressive effect. 5·6 (5) Massive blood loss results in a marked decrease of immunodefense proteins. 7·8 In other words, multiple trauma patients suffer from a secondary immunodeficiency. 2·7·9 The most accepted pathophysiology for pneumonia in intensive care settings is the aspiration of oropharyngeal secretions. 10 The oropharynx, colonized by microflora, 11 is believed to be a major source of pathogens for respiratory tract infections. Four major risk factors for pneumonia in critically ill and mechanically ventilated patients have been described 12. 15: (a) massive gastric aspiration; (b) intracranial pressure monitoring; (c) cimetidine administration; and (d) duration of MV. However, few data are Early Onset Pneumonia in Trauma Patients (Antonelli eta/)

available on the incidence of early onset pneumonia (EOP) and late onset pneumonia (LOP) and related risk factors in multiple trauma patients. The aim of this study was to estimate the frequency ofEOP and LOP as well as risk factors for EOP in a group of multiple trauma patients admitted to the general ICU of the "La Sapienza" University Hospital in Rome, Italy. METHODS

Study Population and Data Collection One hundred twenty four consecutive multiple trauma patients admitted to a our general ICU were prospectively enrolled from December 1990 to February 1992 inclusive.

Data Recorded The following information was collected for each patient: demographics, severity of trauma according to the abbreviated injury scale (AIS), 16 severity of coma according to the Glasgow coma scale (GCS),17 presence of pneumothorax, pulmonary contusion, rib fractures, hemothorax, and MV. During ICU stay, patients were monitored daily for onset of pneumonia, sepsis syndrome, septic shock, and adult respiratory distress syndrome (ARDS).

Criteria for the Diagnosis of Pneumonia Patients with pneumonia satisfied all the following criteria: (a) core temperature of greater than 38.3°C; (b) a WBC count of more than 10,000 cellslmm3 ; (c) purulent tracheobronchial secretions; (d) worsening of pulmonary gas exchange levels; (e) persistent radiographic pulmonary infiltrate ( >24 h); if) cultures ofbronchoalveolar lavage (BAL) samples obtained during a bronchoscopy that yielded 1()5 or more colony-forming unit (cfu)/ml of at least one microorganism; for those patients whose BAL had quantitative cultures less than 1()5, we considered the diagnosis of pneumonia reliable only if the same microorganism was isolated also in blood cultures. Criteria for defining the absence of pneumonia were as follows: Lack of significant growth on quantitative cultures of the BAL ( <1()5 cfu/ml) with resolution of one of the following: (1) fever; (2) radiographically demonstrated infiltrates; (3) fever and radiographically shown infiltrates; (4) radiographically shown infiltrates and a definitive alternative diagnosis; or (5) persistent fever or radiographically demonstrated infiltrates with a definitive Bltemative diagnosis. Pneumonia was considered EOP if onset was within the first 4 days after trauma and LOP when occurrence was after the first 96 h.14 Due to the presence of possibly contaminated wounds, empiric, broad-spectrum antibiotic treatment (piperacillin, amikacin, and metronidazole) was begun for all patients upon admission to the ICU. None of the patients received selective decontamination of the digestive system.

Bronchoalveolar Lavage Technique and Sample Processing A fiberoptic bronchoscope (FB) (Olympus 1T20) was inserted into the endotracheal tube (
laboratory for bacterial quantitative cultures.

Microbiological Processing The BAL fluil;l obtained was homogenized using repeated aspirations with a Pasteur pipette. The BAL effiuent was then cultured quantitatively by serial dilutions (ll10, lllOO, ll1,000) of the original specimen, prepared in saline solution. Diluted specimens were placed in culture media within 60 min after collection; a 0.1-ml aliquot of each sample dilution was inoculated into the following media: Columbia blood agar, chocolate agar with bacitracin, S.S.A., M.S.A., Sabouraud's agar, Enterococcal, MacConkey agar for aerobic cultures; Wilkins, regenerated after 20 min boiling, for anaerobic cultures. Gram and Ziehl-Neelsen stains and direct immunofluorescence against Legtonella pneumophila were performed using undiluted samples.

Definitions Sepsis syndrome was diagnosed using the following criteria: (a) presence of a clinical condition which may produce sepsis; (b ) fever with a temperature of more than 38.3°C or hypothermia with a temperature of less than 35.6°C (core temperature); (c) heart rate greater than 90 beats per minute; (d) respiratory rate more than 20 breaths per minute; (e) evidence of altered organ perfusion, characterized by one or more of the following: metabolic acidosis (pH < 7 .3or elevated blood lactate levels), aratio of Pa()2 to fraction of inspired oxygen of less than 250, oliguria (urine output< 0.5 mV kglh, for at least 1 h), disseminated intravascular coagulation (prothrombin time or partial thromboplastin time longer than normal, platelet count, <100,000 cellslmm3 ) , deterioration of mental state; if) hypotension (systemic arterial pressure < 90 mm Hg or a drop of> 40 mm Hg with respect to basal values), lasting at least 1 h and not responding to conventional administration of fluids and/or requiring doses of dopamine of more than 6 ~glkglmin . Septic shock was defined as hypotension (systemic arterial pressure < 90 mm Hg or a drop of > 40 mm Hg, with respect to basal values), lasting more than 1 h and not responding to the administration of fluids and/or requiring doses of dopamine of 6 ~glkglmin or vasopressors. The ARDS was diagnosed when the Murray score was> 2.5,18 taking into account the score assigned to the ratio of Pa02 to the fraction of inspired oxygen. positive end-expiratory pressure, pulmonary compliance,

and radiographic evidence of diffuse pulmonary inflltrates.

Data Analysis Data were analyzed using the BMDP statistical package (BMDP, Los Angeles, 1990). Differences between proportions were elevated by the 'X,2 test or Fisher's exact test; differences for continuous variables were evaluated by the t test, if the variables were normally distributed, or by the Mann-Whitney test, if not. Variables found to be associated with EOP and LOP with a probability of less than 0.25 unadjusted for multiple comparisons were included as independent variables in a conditional stepwise logistic regression analysis. The presence of interaction among the variables entered in the models was assessed by the likelihood ratio test statistic for the models with and without the interaction terms. RESULTS

One hundred twenty four patients were admitted for trauma to the ICU during the study period. These patients mainly consisted of young men who were transferred from other hospitals after receiving first aid and were admitted to the ICU for multiple trauma (Table I); 83.9 percent of the patients were admitted within the first 24 h after trauma. CHEST I 105 I 1 I JANUARY, 1994

225

Table 1-CharactnVtica of the Studg Population• 36.1± 19.7 100 (80.6) 24 (19.4)

Age, mean±SD Sex, No.{%) M

F

Trauma, No. {%) Multiple

Head AIS, mean±SD GCS, mean± SD Transferredt MV, No. {%) Length ofMV (days), mean± (SD) Antimicrobial treatment, No.(%) Length of stay, (days), mean± SD Deaths, No. {%)

80 (64.5) 44 (35.5) 25.4±10.7 10.7±4.3 lll (89.5) 83 (66.9) 11.9±13.0 116 (93.5) 16.8± 17.6 54 (43.5)

*Total number is 124.

Overall mortality was 43.5 percent; mortality increased by age (32.9 percent less than 40 years, 60 percent for the 41- to 65-year age group, and 75 percent for those more than 65 years) and by AIS (28.1 percent for those with an AIS score under 17, 46.3 percent for those with scores of 17 to 32, and 56 percent for those with scores above 32). Forty-one patients (33.1 percent) developed pneumonia during the ICU stay: 26 (63.4 percent) had Table 2-Bronchoalveolar Ltlvoge tmd Blood Culture Bault. of lbtient. With &rig an-t Pneumonia Bronchoalveolar Lavage Cultures Patient No.

cfulml> 1 x 10"

cfulml<1 x 10"

Blood Cultures

1

Staphylococcw aum.~a

Saum.~a

2 3

Pseudomonas ~no&O Pseudomonas aerugino&O Pseudomonas aerugi0086

Paeruginosa

4 5 6 7

-·

Ssimulans Aanltratw S OUI11UI

Aanitratus

Enterobocter; cloacae E cloacae SAum.~a

Saum.~a

8 9 10 11 12 13 14 15 17 18 19 20 21 22 23 24 25 26

Klebnella pneurnonioe Candida S OUI11UI

Aanitratus Aanitratus S OUI11UI

Aanitratus

CUrobacter .freundl

Aanltratw S OUI11UI

s epidermldis

S OUI11UI Pseudomonas species Citrobacter species

Escherichia coli

Pseudomonas species

•- =sterile.

226

p jltiOf'eScens

Pfluorescens

Ecoli

Ecoli

S OUI11UI

EOP and 15 (36.6 percent) h~d LOP. Among the 26 EOP patients, 13 developed ARDS and 10 developed sepsis syndrome, which in 8 cases was most likely due to pneumonia, given the temporal relationship and the absence of other foci of infection. Among the 15 LOP patients, 4 developed ARDS and 6 developed sepsis syndrome. In 23 of the 26 EOP patients, the BAL showed quantitative cultures of more than lOS cfulml (Table 2). Three patients had quantitative BAL cultures of less than lOS cfulml, but their blood cultures were positive for the same microorganism isolated in the BAL fluid and their clinical evolution was consistent with the di~osis of pneumonia. Of 26 isolated microorganisms in the BAL with cultures of more than lOS cfu/ml, 10 were Gram-positive (all staphylococci); 15 were Gramnegative, mostly Acinetobacter anitratus (6 of 15); one was Candida albicans. In the three BAL samples with cultures less than lOS all isolated bacteria were Gramnegative. In 12 of the 15 LOP patients, the BAL fluid had quantitative cultures of more than lOS; as in the EOP group, 3 patients were considered to have pneumonia on the basis of consistency between BAL and blood cultures. Overall, 21 microorganisms were isolated: 10 were staphylococci and 11 Gram-negative bacilli. The univariate analysis showed that age, presence of pulmonary contusion, overall AIS for thorax greater than 4, AIS for abdomen greater than 9, and absence of MV during the first 4 days of hospitalization or MV lasting less than 24 h were significant risk factors for acquiring EOP. It is interesting to note that p~tients admitted with both high abdomen and high thorax trauma scores had the highest risk of EOP (Table 3). A low GCS score was not associated with an increased risk of EOP; in fact, the incidence of EOP did not vary by GCS category. Variables which were significant in the univariate analysis were entered in a stepwise logistic regression model: age greater than 40 years, severe thoracic and abdominal trauma as reflected by an elevated AIS score, and MV lasting less than 24 h were independently associated with an increased risk of EOP. The strongest risk factor was a combined abdominal and thoracic trauma which increased the risk of EOP by 11 times (Table 4). Factors significantly associated with LOP in the univariate analysis were AIS score for abdomen greater or equal to 4 and exposure to MV for more than 5 days prior to the onset of pneumonia. The relevance of these factors was confirmed by the multivariate analysis: both factors increased the likelihood of LOP by at least 3 times (AIS for abdomen: OR 3.7; probability value= 0.04, confidence limit, 95 percent 1.04-13; MV more than 5 days: or 4.0 probability value = 0.035, confidence limit, 95 percent 1.09-14.3). Early Onset Pneumonia In Trauma Patients (AntofHIIH eta/)

Table 3-.Riak Factonfor Early Onlet Pneumonia: Unicariate AnGlpa EOP

Factors Age, yr <=40 41-65 >65 Sex M F Type of trauma

Head Multiple Pulmonary contusion No Yes Pneumothorax No Yes Hemothorax No Yes Rib fractures No Yes A/S overall <=16 17-32 >32 A/S, thorax and abdomen Thorax> = 4 , abdomen >9 Thorax> = 4, abdomen <9 Thorax <4, abdomen any score LengthofMV >24h :S24 h

No. of Patients No.

%

~ility Value*

OR

82 30 12

12 9 5

14.6 1.0 30.0 2.5 41.7 4.2

0.04

100 24

19 7

19.0 1.0 29.2 1.75

0.27

44

80

6 20

13.6 1.0 25.0 2.1

0.14

75 49

11 15

14.7 1.0 30.6 2.6

0.06

89 35

16 10

18.0 1.0 28.6 1.8

0.19

106 18

22 4

20.8 1.0 22.2 1.1

0.88

80 44

13 13

16.2 1.0 29.5 2.2

0.08

32 52 25

3 15 8

9.4 1.0 22.4 3.9 32.0 4.6

0.10

57

5

8.7 1.0

52

14

26.9 3 .8

15

7

46.7 9.1

0.002

61 63

5 21

8.2 1.0 33.3 5.6

0.0006

*All the probability values are calculated by Pearson x• test except the one for pulmonary contusion which was calculated by Fisher's exact test. DISCUSSION

Trauma patients have been shown to be at high risk for pneumonia,19 but no clear distinction has been

made between EOP and LOP. Our study confirms that pneumonia is a commonly occurring complication in multiple trauma patients and shows that most cases are EOP, occurring within 4 days after trauma. Similar findings also have been described among general intensive care populations. 7•16•17 Early and late onset pneumonia probably result from different underlying causal models; therefore, identification of specific risk factors is essential in developing effective preventive and therapeutic measures. Our study shows that a combined abdominal and thoracic trauma is the strongest risk for EOP, resulting in a more than tenfold increase in the risk of infectious complications of the lungs. Hence, postcontusive inflammatory mechanisms are important factors conducive to bacterial colonization and development of pneumonia. Another important pathophysiologic mechanism may be the altered diaphragmatic dynamics induced by abdominal and thoracic trauma, resulting in impaired ventilatory capability and a decrease in functiona! residual capacity. This hypothesis seems to be confirmed by the finding that the incidence of EOP, among patients ventilated within 4 days after trauma, was significantly higher for patients receiving either mechanical ventilation for less than 24 h or not receiving MV at all. Maintaining positive pressure in the respiratory system, and consequently improving aeration of the contused and partially atelectatic portions of the lung, may restore functional residual capacity to adequate levels and reduce early bacterial colonization. Moreover, endotracheal intubation guarantees easy suction of bronchial secretions, preventing another potential and important risk factor for EOP. Hence, MV during the first posttraumatic phase could contribute to a decrease in the deleterious effects of abdominal and thoracic injuries on deranged diaphragmatic dynamics, improving the ventilation of the lower lobes. This, in turn, could further reduce the likelihood of bacterial colonization. For LOP, mechanical ventilation has the opposite

Table 4-.Riak Factonfor Early Onlet Pneumonia: Multiooriate Anolgaia Factors Age, yr >=40 >40 A/S, thorax and abdomen 1* 2 3 Length of MV, h >24 :S24

No. of Patients

Regression Coefficient

SE

Probability Value

OR

Confidence Limit 95%

82 42

1.319

0.529

0.012

1 3.74

131-10.7

57 52 15

1.522 2.423

0.609 0.777

0.012 0.001

1 4.58 11.3

1.37-15.3 2.42-52.5

61 63

1.934

0.586

0.001

1.0 6.92

2.17-22.1

*1: thorax <4, abdomen any score, 2:thorax >4, abdomen< =9; 3:thorax >4, abdomen >9. CHEST I 105 I 1 I JANUARY, 1994

227

effect with respect to EOP. As shown in other studies,12•15 we obseiVed a fourfold increase in LOP risk when mechanical ventilation lasted more than 5 days. This can be attributed to increased likelihood of oropharyngeal colonization and/or direct inoculation into the trachea by manipulation of the endotracheal tube (eg, introducing a suction catheter). Thus, the need for MV may have contrasting effects, constituting a protective factor for EOP and a risk factor for LOP. In our study group the state of consciousness, as measured by the GCS, did not seem to be a risk factor for EOP and LOP. This is not in accordance with the data reported by Mosconi et al,14 who found an association between impaired airway reflexes and onset of pneumonia. The different methods used for evaluation of coma may explain the disagreement in the results obtained. Repeatability of the diagnosis of impaired airway reflexes may be low due to a subjective evaluation; for this reason, we preferred to adopt the GCS as a more accurate and repeatable index of the severity of coma. Given the high incidence of pneumonia in trauma patients, future research should target these populations in order to better estimate the relative impact of EOP and LOP and to identify risk factors and effective control measures. According to the results of Stoutenbeck et al,20 selective digestive decontamination could be effective in reducing the incidence of both EOP and LOP in this group of patients. Mechanical ventilation during the first days after trauma could be a valuable preventive measure, and it should be further investigated in an experimental setting. Scoring of the severity of abdominal and thoracic trauma at admission to the ICU, according to the AIS, appears to be a useful tool for identifying patients at increased risk of EOP who could benefit from MV during the first days after trauma. ACKNOWLEDGMENTS: The authors are grateful to Drs. Maria Segneri and Luca Severi and to Luigi Riccioni for their assistance in collecting and delineating the data of this article. REFERENCES

LaForce FM. Lower respiratory tract infections. In: Bennett JV. Brachman PS, eds. Hospital acquired infections, 3rd ed. Boston: Little Brown, 1992 2 Green GM, Jacob GL, Low RB, et al. Defense mechanism of the respiratory membrane. Am Rev Respir Dis 1977; 115:479

228

3 Daniele RP. Immunodefenses of lung. In: Fishman AP. Pulmonary diseases and disorders. New York: McGraw Hill, 1980; 624 4 Weigelt JA. Risk of wound infections in trauma patients. Am J Surg 1985; 150:782 5 Berghem L, Lanhborg C, Janstrad C. Role of phagocitic cells in host defense in relation to trauma: a brief review. Acta Chir Scand 1979; 489(suppl):231 6 Ledingham I, McAwatt I. Influence of sedation on mortality in critically ill multiple trauma patients. Lancet 1983; 1:1270 7 Howard RJ, Simmons RM. Acquired immunologic deficiencies after trauma and surgical procedure. Surg Gynec Obstet 1974; 139:771 8 Holch N, Grob PJ, Glinz W. Geroulanas S. Posttraumatisches/ post-operatives lmmunodefektsyndrom: Vortrag Lahres-versannlung. Schweiz Ges Intensiv Med Zurich 1987 9 Sleigh JD, Dich HM . Problems of defective immunity: the dimished immune response. In: Wilson G, Miles A, Parker MT. Principles of bacteriology and immunity. London: Edward Arnold, 1993; 402 10 Craven DE, Steger KA. Nosocomial pneumonia in the intubated patients: new concepts on pathogenesis and prevention. Surg Clin North Am 1989; 3:843-66 11 Johanson WG, Pierce AK, Sanford JP. Changing pharyngeal bacterial flora of hospitalized patients: emergence of Gram negative bacilli. N Eng! J Med 1969; 281 :1137-40 12 Craven DE, Kunches LM, Klinsky V, Lichtenberg DA, Make BJ, McCabe WR. Risk factors for pneumonia and fatality in patients receiving continuous mechanical ventilation. Am Rev Respir Dis 1986; 133:792-96 13 Mandelli M, Mosconi P, Langer M, Cigada M. Prevention of pneumonia in an intensive care unit: a randomized multicenter clinical trial. Crit Care Med 1989; 17:501-05 14 Mosconi P, Langer M, Cigada M, Mandelli M, I.C.U.G.I.C. Epidemiology and risk factors of pneumonia in critically ill patients. Eur J Epidemioll991; 7:320-27 15 Torres A, Aznar R, Gatell JM, Jimenez P, Gonzhles J, Ferrer A, et al. Incidence, risk and prognosis factors of nosocomial pneumonia in mechanically ventilated patients. Am Rev Respir Dis 1990; 142:523-26 16 Committee on medical aspects of automotive safety. Rating the severity of tissue damage: the abbreviated injury scale. JAMA 1971; 215:277-80 17 Teasdale G, Jennett B. Assessment of coma and impaired consciousness: a practical scale. Lancet 1974; 2:81-4 18 Murray JF, Matthay MA, Luce JM, Lock MR. An expanded definition of the adult respiratory distress syndrome. Am Rev Respir Dis 1988; 138:720-23 19 Walker WE, Kapelanski DP, Weiland AP, Stewart JD, Duke JH. Pattern of infections and mortality on thoracic trauma. Ann Surg 1985; 201:752-57 20 Stoutenbeck CP, Van Saene HKF, Miranda DR, Zandstra DF, Langrehr D. The effect of oropharyngeal decontamination using topical nonahsorbable antibiotics on the incidence of nosocomial respiratory tract infections in multiple trauma patients. JTrauma 1987; 27:357-64

Early Onset Pneumonia in Trauma Patients (Antonelli eta/)