Clostridium difficile Infection in Trauma,Surgery,and Medical Patients Admitted to the Intensive Care Unit

Abstract

Background:

Clostridium difficile infection (CDI) causes significant morbidity and mortality rates, especially for patients in the intensive care unit (ICU). Data comparing trauma and surgery patients with CDI in the ICU with medical patients with CDI in the ICU are limited.

Methods:

In a single-center study, we analyzed retrospective data from 25 trauma patients and 13 surgery patients aged 18 years or older who had CDI and had been admitted to the ICU. A comparison group of 156 medical patients aged 18 years or greater who had CDI and were admitted to the ICU also was identified.

Results:

The trauma/surgery patients had a significantly higher mean number of ventilator days (13.5 ± 9.3 vs. 7.3 ± 7.2; p < 0.0004), Foley catheter days (11.9 ± 6.8 vs. 8.0 ± 7.9; p = 0.005), mean ICU length of stay (LOS) (12.34 ± 9.7 vs. 5.9 ± 5.9 days; p < 0.0003), and mean total LOS (16 ± 9.3 vs. 10.7 ± 8.4 days; p = 0.0008). However, the medical group had a significantly higher mean number of vasopressor days (2.07 ± 3.51) than the trauma/surgery group (0.58 ± 1.55; p < 0.0001). The overall survival rate was significantly higher in the trauma/surgery group than in the medical group (100% vs. 81%, respectively; p = 0.003). A higher percentage of patients in the trauma/surgery group received piperacillin/tazobactam before the diagnosis of CDI than the medical patients (58% vs. 37%, respectively; p = 0.02). The number of days that antibiotics were given prior to the development of CDI was greater in the trauma/surgery group than in the medical group (10.3 ± 6.7 vs. 7.6 ± 7.3 days; p = 0.04). Multiple logistic regression models determined ICU LOS (adjusted odds ratio [aOR] 1.27 days; 95% confidence interval [CI] 1.13–1.41), the presence of chronic obstructive pulmonary disease (COPD) (aOR 3.44; 95% CI 1.19–9.95), and piperacillin/tazobactam use (aOR 3.27; 95% CI 1.24–8.65) to be positively associated with CDI in the trauma/surgery group compared with the medical patients.

Conclusions:

Longer ICU stay, receipt of piperacillin/tazobactam, and having COPD were positively associated with CDI in trauma/surgery patients compared with medical patients. These findings suggest further consideration of the possibility of CDI should be given to patients admitted the surgical ICU for an extended period of time, receiving piperacillin/tazobactam, or having COPD. Additional evaluation of these factors in a larger patient sample is warranted.

The epidemiology of Clostridium difficle infection (CDI) has been evolving in recent years. Historically considered a nosocomial infection associated with antibiotic use, CDI is seen increasingly in community settings in populations previously considered low risk [1]. Such infections can lead to severe disease and death, with one study reporting an observed 30-day crude case-fatality rate of 9.3% for patients with health care–associated CDI [2]. Since 2003, large increases in the incidence and mortality rate have been observed in North America, mostly caused by C. difficile PCR ribotypes 027 and 078 [3]. The economic consequences of CDI are substantial, including a higher mean cost and longer length of stay (LOS) than in matched controls [4].

There are limited data on CDI in patients admitted to trauma and surgical services compared with those admitted to medical services. One small study of trauma patients (n = 19) found the incidence of CDI to be similar to the rate observed in other hospitalized patients [5]. Another showed that some trauma patients may be more susceptible to CDI because of their leukocyte genomic expression profile [6]. Messick et al. reported that of approximately 342,000 operations performed over ten years at their institution, the incidence of post-operative CDI was 0.6% (2,188 cases) [7]. Furthermore, Maltenfort et al. reported that the mean LOS was increased by one week and hospital charges were higher by USD $40,000 if surgery was complicated by CDI [8].

Intensive care units (ICUs) have patients with CDI in three categories: Those with CDI that is incidental to the reason for ICU need, those who are transferred because of serious CDI (such as toxic megacolon), and those who develop CDI while in the ICU [9]. Trauma and surgery patients admitted to the ICU tend to be different from other critically ill patients in terms of age, co-morbidities, and illness severity [10].They may be particularly vulnerable to CDI given their relative immune suppression caused by injury [11, 12] and the high likelihood of receiving broad-spectrum antibiotics [13]. Indeed, it has been estimated that a 30% reduction in the use of broad-spectrum antibiotics could result in a 26% reduction in CDI cases [14]. Moreover, trauma and surgery patients frequently receive stress-ulcer prophylaxis with proton-pump inhibitors, which also may increase the risk of CDI [15, 16]. A meta-analysis that included 22 reports of CDI among 80,835 ICU patients showed the prevalence of CDI to be 2% and the mortality rate of those with CDI to be significantly higher (32% vs. 24% without CDI; p = 0.03), and the mean length of ICU stay was greater [17]. Furthermore, a retrospective study that included 67 patients with blunt trauma and CDI found that these patients had more days on the ventilator and in the hospital as well as higher serum creatinine concentrations, white blood cell counts, and organ dysfunction scores than trauma patients without CDI [18].

To our knowledge, there has been only one study that evaluated trauma patients with CDI specifically in the ICU, which was reported nearly ten years ago and did not include surgery or medical patients [5]. Therefore, because of the changing demographics and epidemiology of CDI, this study was undertaken to evaluate this infection in trauma and surgical patients admitted to the ICU and compare their clinical characteristics and outcomes with those of medical patients in the ICU with CDI. We also aimed to further elucidate the association between antibiotics administered prior to infection and CDI in trauma and surgery patients in the ICU, including the time frame for infection to develop.

Patients and Methods

A single-center, retrospective study was conducted at Cleveland Clinic Akron General, a 532-bed community teaching hospital and Level One trauma center in which 17,461 surgical procedures were performed in 2016. The Institutional Review Board approved the study. All surgical, trauma, and medical patients aged 18 years or greater who were admitted to one of the ICUs from January 2010 to December 2016 and found to have CDI were included in the analysis. The electronic medical record (EMR) was used to identify general demographic data; patient-specific data; laboratory data including white blood count (WBC) at the time of CDI diagnosis, serum creatinine concentration, and positive C. difficile assay; antibiotics prescribed prior to the diagnosis of CDI; and data from the initial day of infection. Co-morbidities analyzed included diabetes mellitus (DM), malignant disease, smoking, coronary artery disease, hypertension (HTN), congestive heart failure (CHF), peripheral vascular disease (PVD), cerebrovascular disease (CVD), dementia, chronic obstructive pulmonary disease (COPD), peptic or gastric ulcer disease, renal disease, liver disease, hemiplegia, quadriplegia, and human immunodeficiency infection/acquired immunodeficiency syndrome (HIV/AIDS).

CDI was diagnosed by treating physicians on the basis of clinical and laboratory data including fever, abdominal pain, three or more watery stools daily, elevated WBC count, and a positive stool assay for C. difficile. From January 2010 until January 2012, the C. difficile Tox A/B QuikChek (Inverness Medical, Princeton, NJ) was used to confirm CDI. After January 2012, a polymerase chain reaction assay for C. difficile (Xpert C. difficile, Cepheid, Sunnyvale, CA) was used. The initiation and duration of therapy for CDI was at the discretion of the treating physicians, which was not the focus of this study and is not further described. The first day of infection was defined as the date on which a positive test for C. difficile was obtained.

The patients with CDI were divided into three groups that were admitted to the ICU: (1) Trauma patients, (2) surgery patients, and (3) medical patients. The primary objective of the study was to discover any differences in outcomes between the trauma and surgery patients compared with the medical patients. Outcome data included hospital length of stay (LOS), ICU LOS, days on the ventilator, number of days on vasopressors, and survival. The secondary objective was to identify how pre-infection antibiotics differed between the groups with respect to the development of CDI.

Categorical variables were analyzed using the Pearson χ² and Fisher exact tests. Normally distributed data were analyzed with the Student t-test and skewed data with the Mann-Whitney U test. Quantitative variables were analyzed using one-way analysis of variance. Multiple logistic regression analysis was performed to identify risk factors for the development of CDI while controlling for other predictors in the model. A p value of <0.05 was considered significant. Statistical analysis was performed using SAS version 9.4.

Results

A total of 194 patients with CDI were included in the study. Of these, 25 (13%) were trauma patients, 13 (7%) were surgical patients, and 156 (80%) were medical patients. Because of the similar demographics and the small numbers (Table 1), the surgery and trauma patients were combined into a single group.

Table 1.

Demographic Characteristics of Patients with Clostridium Difficile Infection Admitted to Intensive Care Units

Trauma and surgical patients
Variable	Trauma (n = 25)	Surgical (n = 13)	P Value^a
Mean age (y) ± standard deviation	56.58 (22.01)	65.42 (10.96)	0.11
Sex (n [%])
Male	13 (52.00)	5 (38.46)	0.42
Female	12 (48.00)	8 (61.54)	0.42
Race (n [%])
Caucasian	21 (84.00)	10 (76.92)	0.60
Black	4 (16.00)	3 (23.08)
Other	0	0
Trauma/surgical and medical patients
Variable	Trauma/Surgical (n = 38)	Medical (n = 156)	P Value
Mean age (y) ± standard deviation	59.61 (19.27)	63.25 (15.27)	0.28
Sex (n [%])
Male	18 (47.37)	81 (51.92)	0.61
Female	20 (52.63)	75 (48.08)	0.61
Race (n [%])
Caucasian	31 (81.58)	110 (70.51)	0.17
Black	7 (18.42)	46 (29.49)	0.17

P values obtained by Student t-test, χ² test, or Fisher exact test as applicable.

As illustrated in Table 2, the medical group had a significantly higher percentage of patients with DM (48%) than the trauma/surgery group (16%; p = 0.003), whereas the trauma/surgery group had a significantly higher percentage of patients with malignant disease (21%) than the medical group (7%; p = 0.02). There were no significant differences between the two groups with regard to smoking status, HTN, CHF, PVD, CVD, COPD, dementia, history of peptic or gastric ulcer disease, renal disease, liver disease, or hemiplegia. The mean creatinine concentration on the first day of CDI infection was higher in the medical group (2.8 ± 4.5 mg/dL) than did the trauma/surgery group (1.2 ± 1.3 mg/dL; p = 0.0001). Only one patient in the trauma/surgery group was HIV positive, and there were none in the medical group. Significantly more patients in the medical group were admitted from a nursing home (38% vs. 11%; p = 0.002).

Table 2.

Co-Morbidities of Patients Infected with Clostridium difficile Admitted to Intensive Care Units (n = 194)

	Trauma/Surgical (n = 38)	Medical (n = 156)
Co-Morbidity	N (%)	N (%)	P Value^a
Coronary artery disease
No	35 (92.11)	122 (78.21)	0.05
Yes	3 (7.89)	34 (21.79)	0.05
Malignant disease
No	30 (78.95)	145 (92.95)	0.02
Yes	8 (21.05)	11 (7.05)	0.02
Congestive heart failure
No	34 (89.47)	128 (82.05)	0.27
Yes	4 (10.53)	28 (17.95)	0.27
Chronic obstructive pulmonary disease
No	27 (71.05)	117 (75.00)	0.62
Yes	11 (28.95)	39 (25.00)	0.62
Cardiovascular disease
No	36 (94.74)	154 (98.72)	0.17
Yes	2 (5.26)	2 (1.28)	0.17
Dementia
No	36 (94.74)	143 (91.67)	0.74
Yes	2 (5.26)	13 (8.33)	0.74
Diabetes mellitus
No	32 (84.21)	81 (51.92)	0.0003
Yes	6 (15.79)	75 (48.08)	0.0003
Human immunodeficiency virus infection
No	37 (97.37)	156 (100)	0.19
Yes	1 (2.63)	0	0.19
Hypertension
No	14 (36.84)	53 (33.97)	0.74
Yes	24 (63.16)	103 (66.03)	0.74
Liver disease
No	38 (100)	153 (98.08)	1.00
Yes	0	39 (1.92)	1.00
Peripheral vascular disease
No	36 (94.74)	140 (89.74)	0.53
Yes	2 (5.26)	16 (10.26)	0.53
Renal disease
No	33 (86.84)	113 (72.44)	0.07
Yes	5 (13.16)	43 (27.56)	0.07
Smoking
No	29 (76.32)	136 (87.18)	0.10
Yes	9 (23.68)	20 (12.82)	0.10
Ulcer disease
No	38 (100)	137 (87.82)	0.0279
Yes	0	19 (12.18)	0.0279

P values obtained by χ2 or Fisher exact test as appropriate. Percentages omit missing values.

As noted in Table 3, the mean Acute Physiology and Chronic Health Evaluation (APACHE) II scores differed between the trauma/surgery group (13.03) and the medical group (17.79; p = 0.0006). Compared with the medical patients, the trauma/surgery patients had a significantly higher mean number of ventilator days (13.5 ± 9.3 vs. 7.3 ± 7.2; p < 0.0004), Foley catheter days (11.9 ± 6.8 vs. 8.0 ± 7.9; p = 0.005), mean ICU LOS (12.34 ± 9.7 vs. 5.9 ± 5.9 days; p < 0.0003), and mean total LOS (16 ± 9.3 vs. 10.7 ± 8.4 days; p = 0.0008). However, the medical group had a significantly higher mean number of vasopressor days (2.07 ± 3.51) than the trauma/surgery group (0.58 ± 1.55; p < 0.0001). There were no significant differences between the groups in the number of central venous catheter days or WBC count on the first day of CDI. The overall survival rate was significantly higher in the trauma/surgery group (100%) than in the medical group (81%; p = 0.003).

Table 3.

Clinical Characteristics of Patients Infected with Clostridium difficile Admitted to Intensive Care Units (n = 194)

Mean ± standard deviation
Variable	Trauma/Surgical (n = 38)	Medical (n = 156)	P Value^a
Antibiotic days	10.32 (6.71)	7.62 (7.34)	0.04
APACHE II scores	13.03 (6.02)	17.79 (7.20)	0.0006
Central catheter days	10.79 (8.37)	8.04 (7.91)	0.06
Serum creatinine concentration	1.19 (1.26)	2.83 (4.51)	0.0001
Foley days	11.95 (6.82)	8.01 (7.90)	0.005
ICU length of stay	12.34 (9.76)	5.88 (5.85)	0.0003
Length of stay	16.00 (9.33)	10.67 (8.42)	0.0008
Vasopressor days	0.58 (1.55)	2.07 (3.51)	0.0001
Ventilator days	13.45 (9.25)	7.34 (7.16)	0.0004
WBC count	14.68 (7.32)	15.29 (13.51)	0.71
Antibiotics for CDI	18.20 (9.44)	12.18 (8.32)	0.007
ICU for antibiotics	6.53 (4.45)	4.24 (3.49)	0.005
Admitted for infection	6.37 (4.06)	4.38 (3.55)	0.0003
N (%)
Admitted from:
Home	33 (86.84)	91 (59.09)	0.002
Nursing home	4 (10.53)	58 (37.66)
Other hospital	1 (2.63)	5 (3.25)
Survival
No	0	30 (19.23)	0.003
Yes	38 (100)	126 (80.77)	0.003

P values obtained by Student t-test, χ² test, or Fisher exact test as appropriate. Percentages omit missing values.

APACHE = Acute Physiology and Chronic Health Evaluation; CDI = Clostridium difficile infection; ICU = Intensive care unit, WBC = white blood cells.

All antibiotics received by the trauma/surgery patients and medical patients prior to the onset of CDI were recorded. As described in Table 4, most of the patients received at least one antibiotic (79% in both groups). A higher percentage of patients in the trauma/surgery group received piperacillin/tazobactam before the diagnosis of CDI than did the medical patients (58% vs. 37%; p = 0.02). There were no significant differences between the two groups regarding receipt of ciprofloxacin, ceftriaxone, cefepime, or ampicillin/sulbactam. The time from admission to the ICU to receipt of the first dose of antibiotic was significantly less in the medical group than in the trauma/surgery patients (4.24 ± 3.49 vs. 6.53 ± 4.5 days; p < 0.005). Finally, the number of days that antibiotics were received prior to the development of CDI was greater in the trauma/surgery group than in the medical group (10.3 ± 6.7 vs. 7.6 ± 7.3 days; p = 0.04).

Table 4.

Pre-Infection Antibiotics Given to Patients (n [%]) Admitted to Intensive Care Unit with Clostridium Difficile Infection (n = 194)

Antibiotic	Trauma/Surgical (n = 38)	Medical (n = 156)	P Value^a
PI Cefepime
No	38 (100)	153 (98.08)	0.90
Yes	0	3 (1.92)	0.90
PI Ceftriaxone
No	33 (86.84)	139 (89.10)	0.77
Yes	5 (13.16)	17 (10.90)	0.77
PI Ciprofloxacin
No	33 (86.84)	117 (75.00)	0.12
Yes	5 (13.16)	39 (25.00)	0.12
PI Pip/tazo
No	16 (42.11)	98 (62.82)	0.02
Yes	22 (57.89)	58 (37.18)	0.02
Pre-infection antibiotic
No	8 (21.05)	33 (21.15)	0.99
Yes	30 (78.95)	123 (78.85)	0.99
PI Amp/sulb
No	34 (89.47)	147 (94.23)	0.29
Yes	4 (10.53)	9 (5.77)	0.29

P values obtained by χ² or Fisher exact test as appropriate. Percentages omit missing values.

Abbreviations: Amp/sulbactam = ampicillin/sulbactam; PI = pre-infection, Pip/tazo = piperacillin/tazobactam.

Table 5 shows the results of the multiple logistic regression analysis. After controlling for other predictors in the model, ICU LOS, exposure to pre-infection piperacillin/tazobactam, and having COPD were positively associated with being in the trauma/surgery group, whereas admission from a nursing home, having DM, or being on vasopressors were negatively associated with being in the trauma/surgery group. For every one-day increase in the ICU LOS, the odds of being in the trauma/surgical ICU were 30% higher than the odds of being in the medical group (adjusted odds ratio [aOR] 1.27; p < 0.0001). Patients who received pre-infection piperacillin/tazobactam were 3.3 times more likely to have been in the trauma/surgery group than the medical groups compared with patients who did not receive pre-infection piperacillin/tazobactam (aOR 3.27; p = 0.02). Patients who had COPD were 3.4 times more likely to have been in the trauma/surgery group than in the medical group compared with patients who did not have COPD (aOR 3.44; p = 0.02). Conversely, patients who were admitted from a nursing home were 72% less likely to be in the trauma/surgery group than those who were admitted from home. Those who had DM were 82% less likely to be in the trauma/surgery group vs. the medical group compared with those who did not have DM. Overall, this indicates that COPD status, receiving pre-infection piperacillin/tazobactam, and ICU LOS may play a larger role in trauma/surgical ICU patients than in medical ICU patients developing CDI. Conversely, DM status, admission from a nursing home, vasopressor days, and central venous catheter days likely play a larger role in medical ICU patients than in trauma/surgical ICU patients developing CDI.

Table 5.

Predictors of Clostridium Difficile Infection among Trauma/Surgical Patients vs. Medical Patients

Variable	AOR (95% CI)	P
ICU Length Of Stay	1.27 (1.13–1.41)	<.0001
Central Venous Catheter days	0.89 (0.81–0.98)	0.0155
Vasopressor days	0.57 (0.42–0.78)	0.0005
COPD Yes vs. No	3.44 (1.19–9.95)	0.0224
Pre-infection Pip/tazo Yes vs. No	3.27 (1.24–8.65)	0.0170
Admit from Nursing Home Yes vs. No	0.28 (0.08–0.96)	0.0434
Diabetes Mellitus Yes vs. No	0.18 (0.06–0.56)	0.0030

AOR = Adjusted Odds Ratio, CI = Confidence Intervals, Pip/tazo = piperacillin/tazobactam.

Discussion

CDI is a serious threat to all hospitalized patients. Our study is the largest to evaluate CDI specifically in trauma and surgery patients admitted to the ICU. It is also the first to compare trauma and surgery patients with CDI in the ICU with medical patients. In contrast to a previous report on trauma patients in the ICU with CDI, those in our cohort were older (mean 48.7 vs. 56.6 years) and included fewer males (84.2% vs. 52%) [4]. The surgery patients in our cohort were similar to those in another study of nosocomial infections (CDI was not included) in the ICU with regard to age (65.4 vs. 62 years, respectively), male sex (38% vs. 46%, respectively), and white race (both 77%) [19].

We found that both overall LOS and ICU LOS were significantly longer in the trauma/surgery group, yet their survival rate was higher. This might be explained either by variations in management by the primary services or, more likely, by differences in patient characteristics. For example, even though the ages of the two groups were similar, the medical patients had more co-morbidities and were more likely to be admitted from a nursing home. Our study did not take into account a previous episode(s) of CDI, and having one or more previous episodes is a strong risk factor for recurrent CDI in ICU patients [20]. Moreover, we did not determine whether any of the medical patients who did not survive died as a result of CDI.

Several studies have shown gastric acid suppression to be a risk factor for CDI [15,21, 22]. None of the trauma and surgery patients reported a history of peptic or gastric ulcer disease compared with 19% of the medical patients. It is reasonable to assume that many of the medical patients with a history of ulcer disease would have been taking a medication to suppress gastric acid at the time of their admission to the ICU and might therefore have been at higher risk for CDI. Moreover, gastric acid-suppressing agents such as H2 blockers and proton-pump inhibitors often are prescribed for patients on ventilators for stress-ulcer prophylaxis [23]. This practice might also increase the risk for CDI in ventilator-dependent patients [24]. In our study, the number of ventilator days was higher in the trauma/surgery than in the medical group, presumably exposing these trauma/surgery patients to more days of gastric acid suppression. However, we did not measure the use of gastric acid-suppressing agents in either group, and antibiotic use may be a more salient risk factor for CDI in the ICU [25]. Further investigation is needed to elucidate the association between mechanical ventilation, gastric acid suppression, antibiotic use, and the development of CDI in patients in the ICU.

Of all the antibiotics given to both groups of patients, piperacillin/tazobactam was the only one to be significantly associated with CDI in the trauma/surgery group compared with the medical group. Also, the multiple logistic regression analysis found an aOR of 3.27 for receiving piperacillin/tazobactam prior to the onset of CDI in the trauma/surgery group. This result is somewhat surprising, as it generally is perceived that other classes of antibiotics, such as quinolones and cephalosporins, are more likely to be associated with the development of CDI [26]. Furthermore, piperacillin/tazobactam achieves sufficient concentrations in the intestinal tract to inhibit C. difficile colonization [27]. One possible explanation may be that piperacillin/tazobactam was prescribed more frequently to the trauma/surgery patients than to the medical patients (57% vs. 37%, respectively), and the greater antibiotic heterogeneity in the medical group diluted the risk associated with the other antibiotics. Another is the inherent difficulty in identifying “time zero” in surgical patients, which could be a reason for the differences in observed times from ICU admission to first antibiotic.

A recent study by Bernatz et al. examined risk factors for CDI in four surgical services (orthopedic surgery, neurosurgery, general surgery, and trauma) and found that the rates were four to five times higher in the general surgery and trauma patients [28]. Those investigators hypothesized that patient factors, antibiotic use, and surgical procedures such as intra-abdominal trauma operations may be the reasons for the difference. The specific risk factors for CDI in the study by Bernatz et al. were continuing antibiotics for more than 24 hours after surgery, a higher American Society of Anesthesiologists (ASA) score, more hospital admissions over the preceding 12 months, and longer LOS [28]. Our study also found that a longer ICU LOS was associated with the trauma/surgical group and could be a significant predictor of CDI for this group.

There are several limitations to our study. It was conducted at a single institution, which limits the generalizability of the results to other settings. Because of the retrospective design, the results may have been influenced by unmeasured confounding variables. The number of cases in the trauma/surgery group was relatively small, which limits the power of our statistical analysis. The APACHE II scores were significantly different in the trauma/surgery and medicine groups, which may be attributable to the disproportionate number in each group, whereby a few severely ill patients could have a larger impact on the trauma/surgery group. Another limitation was that not every antibiotic and antibiotic combination prescribed to the study patients was included in the analysis, only the most common ones. Finally, missing data from the electronic medical record may have led to ascertainment bias.

In conclusion, we found that longer ICU LOS, receipt of piperacillin/tazobactam, and having COPD were positively associated with the trauma/surgical group and could be significant predictors of CDI in trauma/surgery patients admitted to the ICU compared with medical patients admitted to the ICU. Survival was significantly higher among in the trauma/surgery patients than the medical patients, although this may have been secondary to the relatively small sample in the former group. Further studies are needed to determine how other factors contribute to the risk of CDI in trauma and surgery patients.

Footnotes

Author Disclosure Statement

R.R.W. has received research grant support and serves on an advisory board for Allergan. The other authors report no competing financial interests.

References

Depestel

, Aronoff

. Epidemiology of Clostridium difficile infection. J Pharm Pract, 2013; 26:464–475.

Lessa

, Mu

, Bamberg

, et al. Burden of Clostridium difficile infection in the United States. N Engl J Med, 2015; 372:825–834.

Smits

, Lyras

, Lacy

, et al. Clostridium difficile infection. Nat Rev Dis Primers, 2016; 2:16020.

Pakyz

, Carroll

, Harpe

, et al. Economic impact of Clostridium difficile infection in a multihospital cohort of academic health centers. Pharmacotherapy, 2011; 31:546–551.

Lumpkins

, Bochicchio

, Joshi

, et al. Clostridium difficile infection in critically injured trauma patients. Surg Infect, 2008; 9:497–501.

Efron

, Liu

, Lottenberg

, et al. A genomic analysis of Clostridium difficile infections in blunt trauma patients. J Trauma Acute Care Surg, 2013; 74:334–338.

Messick

, Hammel

, Hull

. Risk factors that predict recurrent Clostridium difficile infections in surgical patients. Am Surg, 2017; 83:653–659.

Maltenfort

, Rasouli

, Morrison

, et al. Clostridium difficile colitis in patients undergoing lower-extremity arthroplasty: Rare infection with major impact. Clin Orthop Relat Res, 2013; 471:3178–3185.

Bartlett

. Clostridium difficile infection. Infect Dis Clin North Am. 2017; 31:489–495.

10.

Kaufman

, Haas

, Edinger

, et al. Antibiotic susceptibility in the surgical intensive care unit compared with the hospital-wide antibiogram. Arch Surg, 1998; 133:1041–1045.

11.

Girardot

, Rimmelé

, Venet

, et al. Apoptosis-induced lymphopenia in sepsis and other severe injuries. Apoptosis, 2017; 22:295–305.

12.

Marik

, Flemmer

. The immune response to surgery and trauma: Implications for treatment. J Trauma Acute Care Surg, 2012; 73:801–808.

13.

Yun

, Murray

, Nelson

, et al. Infection after orthopaedic trauma: Prevention and treatment. J Orthop Trauma, 2016; 30(Suppl 3):S21–S26.

14.

Fridkin

, Baggs

, Fagan

, et al. Vital signs: Improving antibiotic use among hospitalized patients. MMWR Morb Mortal Wkly Rep, 2014; 63:194–200.

15.

Abdelfatah

, Nayfe

, Nijim

et al. Factors predicting recurrence of Clostridium difficile infection (CDI) in hospitalized patients: Retrospective study of more than 2000 patients. J Invest Med, 2015; 63:747–751.

16.

, Eun

, Kim

, et al. Risk of Clostridium difficile infection with the use of a proton pump inhibitor for stress ulcer prophylaxis in critically ill patients. Gut Liver, 2016; 10:581–586.

17.

Karanika

, Paudel

, Zervou

, et al. Prevalence and clinical outcomes of Clostridium difficile infection in the intensive care unit: A systematic review and meta-analysis. Open Forum Infect Dis, 2015; 3:ofv186.

18.

Vanzant

, Ozrazgat-Baslanti

, Liu

, et al. Clostridium difficile infections after blunt trauma: A different patient population?. Surg Infect, 2015; 16:421–427.

19.

Kelly

, Banerjee

, Golob

, et al. Where's the difference? Presentation of nosocomial infection in critically ill trauma versus general surgery patients. Surg Infect, 2014; 15:377–381.

20.

Jasiak

, Alaniz

, Rao

, et al. Recurrent Clostridium difficile infection in intensive care unit patients. Am J Infect Control, 2016; 44:36–40.

21.

Cao

, Chen

, Wang

, et al. An updated meta-analysis of controlled observational studies: Proton-pump inhibitors and risk of Clostridium difficile infection. J Hosp Infect, 2017 Aug 22. doi: 10.1016/j.jhin.2017.08.017 [Epub ahead of print].

22.

Shivashankar

, Khanna

, Kammer

, et al. Clinical factors associated with development of severe–complicated Clostridium difficile infection. Clin Gastroenterol Hepatol, 2013; 11:1466–1471.

23.

Krag

, Perner

, Wetterslev

, et al. Stress ulcer prophylaxis in the intensive care unit: An international survey of 97 units in 11 countries. Acta Anaesthesiol Scand, 2015; 59:576–585.

24.

Micek

, Schramm

, Morrow

, et al. Clostridium difficile infection: A multicenter study of epidemiology and outcomes in mechanically ventilated patients. Crit Care Med, 2013; 41:1968–1975.

25.

Faleck

, Salmasian

, Furuya

, et al. Proton pump inhibitors do not increase risk for Clostridium difficile infection in the intensive care unit. Am J Gastroenterol, 2016; 111:1641–1648.

26.

Brown

, Khanafer

, Daneman

, et al. Meta-analysis of antibiotics and the risk of community-associated Clostridium difficile infection. Antimicrob Agents Chemother, 2013; 57:2326–2332.

27.

Kundrapu

, Sunkesula

, Jury

, et al. Do piperacillin/tazobactam and other antibiotics with inhibitory activity against Clostridium difficile reduce the risk for acquisition of C. difficile colonization?. BMC Infect Dis, 2016; 16:159.

28.

Bernatz

, Safdar

, Hetzel

, et al. Antibiotic overuse is a major risk factor for Clostridium difficile infection in surgical patients. Infect Control Hosp Epidemiol, 2017; 38:1254–1257.