Sara G. Murray, MD*;Joanne Yim, PhD;Rhiannon Croci, BSN, RN;Alvin R. Rajkomar, MD;Maria E. Otto;Raman Khanna, MD, MAS and Russ Cucina, MD, MS, UCSF, San Francisco, CA

Meeting: Hospital Medicine 2017

Categories: Oral Presentations, Research Abstracts

Keywords: , , ,

Background: Hospital-acquired C. difficile colitis is associated with increased length of stay (LOS) and significant morbidity and mortality. During hospitalization, patients visit many procedural, diagnostic, and treatment areas throughout the hospital, presenting opportunities for spore contamination of surfaces and nosocomial disease transmission. We developed a novel methodology using electronic health record (EHR) data to map potential C. difficile transmission and identified an opportunity for improvement in our hospital’s infection control practices.

Methods:  We analyzed all patient encounters in our health system from 2013 through 2015, which included 86,648 hospitalizations. During hospitalizations, there were 434,745 patient location changes within the hospital (see figure). If a patient with a diagnosis of C. difficile spent time in any space (e.g. a bed, a MRI machine, endoscopy) that space was considered “contaminated” for the following 24 hours, regardless of real-world cleaning practices. All patients who passed through that “contaminated” space during the subsequent 24 hours were considered “exposed” to C. difficile. Both “exposed” and unexposed hospitalized patients were followed for 60 days for the development of disease, as measured by a positive laboratory test for C. difficile (GDH and toxin immunoassay +/- confirmatory PCR) obtained either as an inpatient or outpatient. For every area of the hospital, we calculated the odds ratio for developing the disease if “exposed” in comparison to unexposed individuals who passed through the same location in the hospital when it hadn’t been occupied by a patient with C. difficile in the preceding 24 hours.

Results: Being “exposed” to C. difficile in the emergency department (ED) was significantly associated with the development of C. difficile within the next 60 days (OR 2.7, p=0.01).This effect remained significant (OR 2.3, p=0.03) after adjustment for age, gender, recent antibiotics, number of location changes, and LOS. Heat maps (not shown) showed potential concentration of the effect in the ED’s CT scanner.  We performed adjusted analyses for the ED’s CT scanner which showed a significant association between being “exposed” in this location and disease acquisition (OR 2.9, p<0.01).

Conclusions:  In our hospital, getting a CT in the ED within 24 hours after a patient with C. difficile used the scanner was associated with increased odds of developing the disease in comparison to getting a CT when the scanner is not in the “contaminated” state defined here. This identified an opportunity to improve cleaning practices in that location. Our novel data analytic strategy may be widely applicable for infection control quality improvement at other institutions and for other infectious diseases.

To cite this abstract:

Murray, SG;

Yim, J;

Croci, R;

Rajkomar, AR;

Otto, ME;

Khanna, R;

Cucina, R
. SPATIAL AND TEMPORAL MAPPING OF C. DIFFICILE: AN EXPLORATORY BIG DATA ANALYSIS [abstract]. Journal of Hospital Medicine. 2017; 12 (suppl 2). Accessed October 19, 2017.

« Back to Hospital Medicine 2017