SURGICAL INFECTIONSVolume 9, Number 5, 2008© Mary Ann Liebert, Inc.DOI: 10.1089/sur.2007.071

Clostridium difficile Infection in Critically Injured Trauma Patients

Kimberly Lumpkins, Grant V. Bochicchio, Manjari Joshi, Ryan Gens, Kelly Bochicchio, Anne Conway, Stacey Schaub, and Thomas Scalea

Abstract

Background: Clostridium difficile infection (CDI) is an increasing problem worldwide. To our knowledge, thereare no studies evaluating specifically the incidence of CDI in the critically injured trauma population.Methods: Five hundred eighty-one consecutive critically injured trauma patients were followed prospectivelyfor development of CDI, diagnosed by toxin assay. Testing was ordered on the basis of symptoms. Antibioticusage was classified as surgical prophylaxis or therapeutic.Results: Nineteen cases of CDI were diagnosed in 581 patients (3.3%). Age, sex, race, and injury severity score(ISS) were similar in patients with and without CDI (p � 0.2); the mean ISS in patients with CDI was 29 points.Intensive care unit length of stay (ICU LOS), ventilator days, and hospital length of stay (HLOS) were signifi-cantly higher in the CDI patients. The diagnosis was made a mean of 17 days after admission; however, in fourpatients (21%), the infections were diagnosed within six days of admission. Fourteen patients (74%) had re-ceived therapeutic antibiotics for confirmed or suspected infection prior to the appearance of colitis; four pa-tients (21%) received only intraoperative prophylaxis, and one patient had no antibiotic exposure. These fivepatients were significantly younger than those who developed CDI after therapeutic antibiotic usage and hadsignificantly shorter HLOS and ICU LOS (p � 0.05). This result persisted after controlling for age using multi-ple linear regression analysis.Conclusions: Clostridium difficile infection occurred in 3.3% of patients and was diagnosed as early as the fourthhospital day. We have identified a unique subgroup of younger patients who developed CDI after minimal orno antibiotic exposure. Further study is needed to characterize this population.

497

CLOSTRIDIUM DIFFICILE is the pathogen most commonlyidentified in cases of nosocomial antibiotic-associated di-

arrhea, and may have serious or even fatal consequences [1].Throughout North America and Europe, outbreaks of C. dif-ficile infection (CDI) are occurring more frequently and areassociated with increasing rates of toxic megacolon, septicshock, and death [2,3]. Investigators of recent outbreaks inthe United States and Canada have revealed the emergencea new and highly toxic strain of C. difficile [4–6].

Few studies have reported the incidence rates of CDIspecifically among intensive care unit (ICU) patients. In 2000,Chang and Nelson reported an incidence of 4.8% in a cohortof 554 ICU and cardiac care unit (CCU) patients, which washigher than the incidence rate found in their entire hospitalpopulation (2%) [7]. In a cohort study of 1,872 ICU patients,Lawrence et al. found that 4% had CDI; for one-half of these

patients, it was a new diagnosis during an ICU admissionfor other causes [8]. Marra et al. demonstrated clearly the ad-verse consequences of CDI in the ICU. In their series of 58ICU patients with CDI, the mortality rate was 27.6%; 62% ofpatients developed concurrent infections, especially bac-teremia [9].

Clostridium difficile also is an important pathogen in thepostoperative population. In a series of patients having elec-tive colorectal surgery, 4.2% developed CDI [10], and in an-other series, 5% of liver transplant recipients suffered fromCDI [11]. Clostridium difficile infection is well recognized asan infrequent but potentially severe complication of cardiacsurgery [12,13].

Severely injured trauma patients represent a complexgroup with characteristics of both ICU and postoperative pa-tients. Despite the increasing importance of C. difficile, to our

Division of Clinical & Outcomes Research, R. Adams Cowley Shock Trauma Center, Baltimore, Maryland.

knowledge, there are no studies of the epidemiology of CDIin the critically injured trauma population. The objective ofthis study was to evaluate the incidence of CDI in criticallyinjured trauma patients and to quantify the risk factors as-sociated with the acquisition of CDI.

Patients and Methods

Selection of study participants and sample collection

This study utilized a cohort design. The study populationconsisted of all critically injured trauma patients admitted tothe ICU of the R. Adams Cowley Shock Trauma Center fromJuly, 2004 to October, 2006. The only exclusion criterion wasa non-traumatic reason for admission. The demographic datacollected included age, sex, mechanism of injury, and InjurySeverity Score (ISS). Outcome data recorded prospectivelyincluded hospital length of stay (HLOS), intensive care unitlength of stay (ICU LOS), and ventilator days.

Diagnosis of CDI

Stool samples for evaluation for CDI were ordered by thetreating physicians on the basis of patient signs and symp-toms. The assays, performed by the hospital laboratory, werethe Tox A/B QuikChek (Inverness Medical, Princeton NJ),an immunoassay for C. difficile toxins A and B with reportedsensitivity of 90.2% and specificity of 99.7%. Initiation of anti-CDI pharmacotherapy was at the discretion of the treatingclinicians. No evaluation of C. difficile genotype was per-formed.

Antibiotic use

All prescribed courses of antibiotics were recordedprospectively. The indication for antibiotic use was deter-mined by discussion with clinicians and review of the med-

ical record. Operative, procedure, and medication adminis-tration records were reviewed to determine exposure to pro-phylactic antibiotics in CDI patients who did not receivetherapeutic antibiotics (standing antibiotic orders for sus-pected or confirmed infection). Patients with CDI were defined as having “minimal exposure” if they received onlyintraoperative prophylaxis or no antibiotics prior to devel-opment of CDI, and defined as having “therapeutic expo-sure” if they received standing-order antibiotics.

Statistical analysis

The Pearson chi-square and Fisher exact tests were usedas appropriate to compare categorical demographic variablesby CDI status. The Student t-test was used for univariateanalysis of continuous demographic variables by CDI status.Multiple logistic regression analysis was performed to de-termine risk factors for the development of CDI. Age, gen-der, ISS, mechanism of injury, HLOS, and laparotomy statuswere included as risk factors in this model.

After determination of antibiotic exposure status in pa-tients with CDI, the Fisher exact test and the Student t-testwere utilized to compare demographic and outcome vari-ables (HLOS, ICU LOS, ventilator days, and death) betweenpatients having minimal and therapeutic antibiotic exposure.Individual multiple linear regression models were employedto control for demographic differences between the thera-peutic and minimal antibiotic exposure groups. Analysis wasperformed using Stata version 9 (Stata Corporation, Lake-wood, TX).

Results

Patient demographics and CDI status

Five hundred eighty-one consecutive trauma patients ad-mitted to the ICU were enrolled in this study. The average

LUMPKINS ET AL.498

TABLE 1. DEMOGRAPHIC AND OUTCOME DATA FOR PATIENTS

WITH AND WITHOUT CLOSTRIDIUM DIFFICILE INFECTION (CDI)a

CDI (n � 19) Non-CDI (n � 562)

Mean age 48.7 � 20.7 43.7 � 20.2Percent male 84.2 75.4Percent with blunt trauma 89.5 85.7Injury Severity Score 33.2 � 14.5 28.8 � 13.1Percent having laparotomy 22.0 23.0

aAll p values are non-significant.

TABLE 2. PROPHYLACTIC ANTIBIOTIC EXPOSURE IN PATIENTS WITH CLOSTRIDIUM DIFFICILE INFECTION

WHO DID NOT RECEIVE THERAPEUTIC ANTIBIOTICS PRIOR TO DEVELOPMENT OF INFECTION

Hospital day Patient Antibiotic exposure (hospital day) of diagnosis

75 Intraoperative cefazolin (4, 6) 8105 Intraoperative ceftriaxone and vancomycin (0) 6126 No documented antibiotic exposure 4322 Cefotetan in the trauma resuscitation unit (0) 6491 Intraoperative cefazolin (0, 3, 6, 9) 11

age overall was 43.8 � 20 years (mean � standard devia-tion), and the mean ISS was 28.9 � 13.1 points. Male patientsaccounted for 75.7% of the total (n � 439), and 85.9% of thecohort (n � 498) had suffered blunt injuries. Nineteen pa-tients (3%) were found to have CDI by laboratory assay dur-ing their hospital stays (CDI group); 562 patients did not de-velop CDI (non-CDI group). There were no significantdifferences between patients with and without CDI in anydemographic variable (all p values � 0.05) (Table 1). Twenty-three percent of the non-CDI patients underwent a laparot-omy for abdominal injuries (129 of 562), compared with 22%of CDI patients (4 of 19; p � 0.05). No patient underwent acolectomy for the treatment of C. difficile infection. The over-all cumulative incidence of CDI was 1.7 cases per 1,000 pa-tient-days.

Logistic regression analysis for development of CDI

In a model controlling for age, sex, mechanism of injury,ISS, HLOS, and laparotomy status, only HLOS was found toconvey a higher risk for the development of CDI (odds ra-tio [OR] 1.04; 95% confidence interval 1.02, 1.07, p � 0.002;all other p values � 0.05).

Antibiotic usage in patients with CDI

Of 19 patients with laboratory-confirmed CDI, 14 (74%)had received therapeutic antibiotics prior to the diagnosis.Of the five patients who did not receive therapeutic anti-biotics, four received one or more doses of prophylacticantibiotics during surgery or other invasive proceduresprior to the diagnosis of CDI. In the remaining patient, noantibiotic use was documented prior to the onset of infec-tion. The nature of the antibiotic exposure for the five patients who did not receive therapeutic antibiotics is de-scribed in Table 2. Patients who had minimal or no anti-biotic exposure were significantly younger than patientsreceiving therapeutic antibiotics (average age 29.6 vs. 55.5years; p � 0.005), but had similar ISS scores and mecha-nisms of injury (p � 0.05).

Outcome data by infection status and antibiotic exposure

As expected, the mean HLOS was 15.9 days greater in pa-tients who developed CDI than in those who did not (34.9vs. 19.0 days; p � 0.003). The ICU LOS and ventilator dayswere significantly greater as well (mean difference 12.9 and16.9 days, respectively; both p � 0.05) (Table 3). The mortal-ity rate was not higher in the CDI group (14.7% without CDIvs. 21% with CDI; p � 0.5).

When CDI patients were compared by extent of antibioticexposure, those with minimal exposure were found to havea significantly shorter HLOS, ICU LOS, and ventilator timethan patients having therapeutic exposure (Table 4). Becauseyounger age might be a significant confounder for outcomevariables in the prophylactic exposure population, multiplelinear regressions controlling for age were performed indi-vidually for HLOS, ICU LOS, and ventilator days. Prophy-lactic antibiotic status continued to have a significant impacton HLOS, ICU LOS, and ventilator days when controllingfor age (coefficients for prophylactic exposure �33.4, �26,and �33.4, respectively; all p � 0.05). Although all fourdeaths among CDI patients occurred in the group receivingtherapeutic antibiotics, this was not statistically significant(p � 0.53).

Time of CDI acquisition

The average time to CDI acquisition in all patients was16.7 � 15 days, with the diagnosis made at a minimum offour days and a maximum of 60 days after admission. How-ever, patients with minimal or no antibiotic exposure hadtheir infections diagnosed an average of 13.1 days earlierthan those who had therapeutic antibiotic exposure (7 daysafter admission vs. 20 days; p � 0.013).

Discussion

Clostridium difficile infection continues to be a major threatto hospitalized patients worldwide. The increasing virulenceof some strains of C. difficile, in combination with recent dev-astating nosocomial outbreaks, have highlighted the impor-

C. DIFFICILE IN CRITICALLY INJURED PATIENTS 499

TABLE 3. OUTCOME BY CLOSTRIDIUM DIFFICILE INFECTION (CDI) STATUS

Hospital LOSa ICU LOSa Ventilator daysa Mortality rate

Non-CDI 19.0 � 13.5 13.4 � 10.9 13.2 � 12.0 21.0% (82/558)CDI 34.9 � 20.1 26.3 � 20.0 30.1 � 20.9 14.7% ( 4/ 19)

aP � 0.05.LOS � length of stay; ICU � intensive care unit.

TABLE 4. AGE AND OUTCOME BY ANTIBIOTIC EXPOSURE IN PATIENTS WITH AND WITHOUT CLOSTRIDIUM DIFFICILE INFECTION

Hospital LOSa ICU LOSa Ventilator daysa Agea

Minimal antibiotic 16.4 � 4.1 11.4 � 5.4 10.8 � 5.9 29.6 � 11.8Therapeutic antibiotics 41.5 � 19.3 31.6 � 20.6 37 � 20 55.5 � 19.0

aP � 0.05.LOS � length of stay; ICU � intensive care unit.

tance of continued vigilance [3,6]. During a recent outbreak,there was a 53% case fatality rate and a 23% rate of colec-tomy in a series of patients admitted to the ICU for severeCDI [14].

Less is known about the incidental development of CDIin the general ICU population. In a cohort study of 2,859 pa-tients admitted to a community hospital, Chang and Nelsonreported that 27 of the 554 (4.8%) ICU and CCU patients werefound to have CDI [7]. Lawrence et al. reported a similar in-cidence rate of 4% among a sample of 1,872 ICU patients [8].Thirty-six (54%) of these patients acquired C. difficile whilein the ICU. Antimicrobial treatment, longer HLOS, enteralfeeding, mechanical ventilation, colonization with vanco-mycin-resistant enterococci, and a high in-hospital preva-lence of C. difficile were risk factors for CDI. Fernandez et al.found that previous or concurrent admission to the ICU wasassociated with a higher incidence of metronidazole treat-ment failure [15]. Death in the ICU CDI population has beenassociated with greater age and increasing degrees of organdysfunction [13].

To our knowledge, no studies have focused on CDI in thecritically injured trauma population. At some point, CDI wasdiagnosed during the hospital stay in 3.3% of patients ad-mitted initially to the trauma intensive care unit at our in-stitution. This figure compares favorably with the previouslypublished incidence rate of CDI in ICU populations. Thelonger hospital and ICU LOS in patients with CDI are con-sistent with the well-known finding that the risk of CDI in-creases with longer hospitalization. Indeed, in this cohort,HLOS was the only significant risk factor for CDI (OR 1.04;p � 0.002) in a multiple logistic regression model controllingfor age, sex, ISS, mechanism of injury, and laparotomy sta-tus.

A small population of patients in this cohort developedCDI after exposure only to prophylactic antibiotics duringoperative procedures, or, in one instance, after no antibioticexposure at all. This population was significantly younger(mean age 29 years) than the patients who had received ther-apeutic antibiotics (mean age 55 years). These patients alsodeveloped CDI rapidly—as early as four days and no laterthan 11 days after admission—in comparison to an averageof 20 days after admission in the patients treated with ther-apeutic antibiotics. Finally, the minimal exposure group hadsignificantly shorter HLOS and ICU LOS than the therapeu-tic exposure group, even after controlling for the differencein age. One limitation of this study is that antibiotic expo-sure prior to injury could not be ascertained. It is possiblethat the patients who developed CDI rapidly had been tak-ing antibiotics on an outpatient basis prior to injury. On theother hand, this information often is not readily available toclinicians, potentially rendering it of limited value to thetrauma care team.

The early diagnosis of CDI after minimal antibiotic ex-posure may reflect existing colonization with C. difficile.Three percent of individuals in the community are be-lieved to be colonized [10]. It is possible that serious injury, in combination with even brief exposure to antibi-otics in the operative setting, disrupts the colonic milieusufficiently to permit the overgrowth of and infection withC. difficile. High concentrations of antibodies to toxin Ahave been correlated with lesser severity of infection [16],and higher antibody titers have been found in asympto-

matic carriers [17]. This suggests that colonization de-creases the severity of CDI, possibly providing part of theexplanation for the short HLOS and ICU LOS experiencedby this population.

It is unknown whether the deaths in the patients with CDIwere a direct result of the infection; however, no patients inthis study underwent colectomy for CDI. No significant dif-ference in the mortality rate was seen in patients with andwithout CDI. It also is unknown whether the patients in thiscohort experienced a later relapse of CDI that might havecontributed to further morbidity or death.

Clostridium difficile infection occurred in our critically in-jured trauma population at a rate consistent with the ratesreported in ICU patients (4–5%) [11,12]. It is important fortrauma surgeons and intensivists to consider C. difficile as asource of diarrhea even in patients who would be consid-ered low risk—those with minimal antibiotic exposure whodevelop diarrhea early in their hospital course.

References

1. McFarland LV. Epidemiology, risk factors and treatmentsfor antibiotic-associated diarrhea. Dig Dis 1998;16:292–307.

2. Bartlett JG. Narrative review: The epidemic of Clostridiumdifficile-associated enteric disease. Ann Intern Med 2006;145:758–764.

3. Muto CA Pokrywka M, Shutt K et al. A large outbreak ofClostridium difficile associated disease with an unexpectedproportion of deaths and colectomies at a teaching hospitalfollowing increased fluoroquinolone use. Infect ControlHosp Epidemiol 2005;26:273–280.

4. Loo VG, Poirier L, Miller MA et al. A predominantly clonalmulti-institutional outbreak of Clostridium difficile-associateddiarrhea with high morbidity and mortality. N Engl J Med2005;353:2442–2449.

5. McDonald LC, Killgore GE, Thompson A, et al. An epidemic,toxin gene–variant strain of Clostridium difficile. N Engl J Med2005;353:2433–2441.

6. Pepin J, Valiquette L, Cossette B. Mortality attributable tonosocomial Clostridium difficile-associated disease during anepidemic caused by a hypervirulent strain in Québec. CMAJ2005;173:1037–1042.

7. Chang VT, Nelson K. The role of physical proximity in noso-comial diarrhea. Clin Infect Dis 2000;31:717–722.

8. Lawrence SJ, Puzniak LA, Shadel BN, et al. Clostridium dif-ficile in the intensive care unit: Epidemiology, costs and col-onization pressure. Infect Control Hosp Epidemiol 2007;28:123–130.

9. Marra AR, Edmond MB, Wenzel RP, et al. Hospital-acquiredClostridium difficile-associated disease in the intensive careunit setting: Epidemiology, clinical course, and outcome.BMC Infect Dis 2007;7:42–48.

10. Wren SM, Ahmed N, Jamal A, et al. Preoperative oral anti-biotics in colorectal surgery increase the rate of C. difficilecolitis. Arch Surg 2005;140:752–756.

11. Hashimoto M, Sugawara Y, Tamura S, et al. Clostridium dif-ficile-associated diarrhea after living donor liver transplan-tation. World J Gastroenterol 2007;13:2072–2076.

12. Crabtree T, Aitchison D, Meyers BF, et al. Clostridium diffi-cile in cardiac surgery: Risk factors and impact on postop-erative outcome. Ann Thorac Surg 2007;83:1396–1402.

13. Harbarth S, Samore MH, Carmeli Y. Antibiotic prophylaxisand the risk of Clostridium difficile-association diarrhea. JHosp Infect 2001;48:93–97.

LUMPKINS ET AL.500

14. Lamontagne F, Labbe AC, Haeck O, et al. Impact of emer-gency colectomy on survival of patients with fulminantClostridium difficile colitis during an epidemic caused by ahypervirulent strain. Ann Surg 2007;245:267–272.

15. Fernandez A, Anand G, Friedenberg F. Factors associatedwith failure of metronidazole in Clostridium difficile-associ-ated disease. J Clin Gastroenterol 2004;38:414–418.

16. Warny M, Vaerman J, Avesani V, et al. Antibody responseto C. difficile toxin A in association with clinical course of in-fection. Infect Immun 1994;62:384–389.

17. Kyne L, Warny M, Qamar A, et al. Asymptomatic carriageof C. difficile and serum levels of IgG antibody against toxinA. N Engl J Med 2000;342:390–397.

Address reprint requests to:Dr. Grant V. Bochicchio22 S. Greene St., T1R59

Baltimore, MD 21201

E-mail: gbochicchio@umm.edu

C. DIFFICILE IN CRITICALLY INJURED PATIENTS 501