Predictors of Clostridium Difficile Infection and Mortality in Adult Patients in the Intensive Care Unit
Teja, Bijan1; O’Gara, Brian1, Mueller, Ariel1, Shaefi, Shahzad1
1Department of Anesthesia, Critical Care and Pain Medicine, Beth Israel Deaconess Medical Center, Boston, MA, USA
INTRODUCTION
Several factors including leukocytosis, antibiotic use, prior surgery, and advanced age have been associated with Clostridium difficile (C. diff) infection in hospitalized patients1, 2, 3. The critically ill may suffer from additional conditions increasing susceptibility to develop C. diff. While risk factors for C. diff infection in the intensive care unit (ICU) have been identified4, several possible predictors have not yet been evaluated. Predictors of mortality in ICU patients with C. diff have also not been evaluated to our knowledge.
OBJECTIVES
To evaluate predictors of C. diff infection in the ICU and identify risk factors for mortality among those patients with C. diff using a large retrospective database.
METHODS
We performed a retrospective cohort study using the MetaVision database at Beth Israel Deaconess Medical Center, comprising data from over 40,000 ICU admissions between 2009 and 2017. Patients with at least one recording of white blood cell count (WBC) while in the ICU were included in the analysis. We obtained data on age, gender, peak WBC count, peak HbA1c, ICU type, days of antibiotic and antifungal use (excluding vancomycin and metronidazole), incidence of mechanical ventilation, central line use, C. diff testing results and stool characteristics.
Logistic regression was performed using a forward stepwise variable selection algorithm with a minimum p-value of 0.05 to identify potential risk factors for C. diff. Predictors of in-hospital mortality were identified in a similar fashion in the subgroup of critically ill patients with known C. diff infection.
RESULTS
Data from 42,139 ICU patients was included in the study (Table 1). The incidence of C. diff in our cohort was 4.8%. The following variables were predictive of C. diff infection (p<0.001 for all values): loose or liquid stool (OR 4.0), central line use (OR 1.4), male gender (OR 1.2), peak WBC (0.8% increase per 1K/µL elevation), days receiving antibiotics (1.0% increase per day), and age (9% increase per decade). Mechanical ventilation was a negative predictor for C. diff infection (OR 0.73).
The risk of C. diff infection as a function of peak WBC value is shown in figure 2. Patients with persistent leukopenia (peak WBC <3K/µL) had a 9.6% incidence of C. diff. The risk of C. diff increased steadily as patients’ peak WBC values increased above 10K/µL. The highest incidence of C. diff (15.3%) was seen in those with peak WBC > 40K/µL.
Among the subgroup of 2,021 patients with a C. diff infection, in-hospital mortality was 21.9%. Mechanical ventilation (OR 2.0), loose or liquid stool (OR 1.46), central line use (1.34), elevated WBC (2% increase per 1 K/µL elevation), days receiving antifungals (4% increase per day) and age (20% increase per decade) were found to be predictors of mortality (Table 2). Surgical ICU patients had a significantly lower risk of mortality (OR 0.6) as compared to medical admissions.
CONCLUSION
The overall incidence of C. diff in our cohort (4.8%) was similar to the 4% incidence found in another study of ICU patients4. We identified several risk factors for C. diff infection, including peak WBC count, age and antibiotic use. These predictors may have useful implications in determining which patients should receive empiric treatment for C. diff while awaiting stool testing results, and which critically ill patients with C. diff have the highest risk of mortality.
Predictors of Clostridium Difficile Infection and Mortality in Adult Patients in the Intensive Care Unit
Teja, Bijan1; O’Gara, Brian1, Mueller, Ariel1, Shaefi, Shahzad1
1Department of Anesthesia, Critical Care and Pain Medicine, Beth Israel Deaconess Medical Center, Boston, MA, USA
INTRODUCTION
Several factors including leukocytosis, antibiotic use, prior surgery, and advanced age have been associated with Clostridium difficile (C. diff) infection in hospitalized patients1, 2, 3. The critically ill may suffer from additional conditions increasing susceptibility to develop C. diff. While risk factors for C. diff infection in the intensive care unit (ICU) have been identified4, several possible predictors have not yet been evaluated. Predictors of mortality in ICU patients with C. diff have also not been evaluated to our knowledge.
OBJECTIVES
To evaluate predictors of C. diff infection in the ICU and identify risk factors for mortality among those patients with C. diff using a large retrospective database.
METHODS
We performed a retrospective cohort study using the MetaVision database at Beth Israel Deaconess Medical Center, comprising data from over 40,000 ICU admissions between 2009 and 2017. Patients with at least one recording of white blood cell count (WBC) while in the ICU were included in the analysis. We obtained data on age, gender, peak WBC count, peak HbA1c, ICU type, days of antibiotic and antifungal use (excluding vancomycin and metronidazole), incidence of mechanical ventilation, central line use, C. diff testing results and stool characteristics.
Logistic regression was performed using a forward stepwise variable selection algorithm with a minimum p-value of 0.05 to identify potential risk factors for C. diff. Predictors of in-hospital mortality were identified in a similar fashion in the subgroup of critically ill patients with known C. diff infection.
RESULTS
Data from 42,139 ICU patients was included in the study (Table 1). The incidence of C. diff in our cohort was 4.8%. The following variables were predictive of C. diff infection (p<0.001 for all values): loose or liquid stool (OR 4.0), central line use (OR 1.4), male gender (OR 1.2), peak WBC (0.8% increase per 1K/µL elevation), days receiving antibiotics (1.0% increase per day), and age (9% increase per decade). Mechanical ventilation was a negative predictor for C. diff infection (OR 0.73).
The risk of C. diff infection as a function of peak WBC value is shown in figure 2. Patients with persistent leukopenia (peak WBC <3K/µL) had a 9.6% incidence of C. diff. The risk of C. diff increased steadily as patients’ peak WBC values increased above 10K/µL. The highest incidence of C. diff (15.3%) was seen in those with peak WBC > 40K/µL.
Among the subgroup of 2,021 patients with a C. diff infection, in-hospital mortality was 21.9%. Mechanical ventilation (OR 2.0), loose or liquid stool (OR 1.46), central line use (1.34), elevated WBC (2% increase per 1 K/µL elevation), days receiving antifungals (4% increase per day) and age (20% increase per decade) were found to be predictors of mortality (Table 2). Surgical ICU patients had a significantly lower risk of mortality (OR 0.6) as compared to medical admissions.
CONCLUSION
The overall incidence of C. diff in our cohort (4.8%) was similar to the 4% incidence found in another study of ICU patients4. We identified several risk factors for C. diff infection, including peak WBC count, age and antibiotic use. These predictors may have useful implications in determining which patients should receive empiric treatment for C. diff while awaiting stool testing results, and which critically ill patients with C. diff have the highest risk of mortality.