Award: Presidential Poster Award
Umer E. Malik, MD1, Karthik Gnanapandithan, MD, MS1, Prabin Sharma, MD2, Muhammad Sohail Mansoor, MD3, Paul Feuerstadt, MD, FACG4
1Yale New Haven Hospital, New Haven, CT; 2Yale New Haven Health Bridgeport Hospital, Bridgeport, CT; 3Albany Medical Center, Albany, NY; 4Yale New Haven Hospital, Hamden, CT
Introduction: CDI is the most common healthcare-associated infection with significant morbidity, mortality and cost. Various scoring systems have been formulated to aide disease prognostication. Our goal was to compare the ATLAS and CDSS predictive scores, initially to validate them and subsequently establishing which score was more accurately associated with short-term mortality and colectomy.
Methods: We performed a retrospective cohort analysis on patients admitted to Yale-New Haven Hospital between 9/13 and 4/16. For each patient with CDI (≥3 liquid stools in a 24-hour period with a positive stool assay) we gathered various data points including demographics, diagnostics, risk factors, treatments and outcomes. The CDSS and ATLAS scoring systems were independently used to grade the severity in each patient and association with outcome (e.g. any ICU admission or 30-day mortality). Using Logistic regression (SAS 9.4), we used area under the ROCs to compare the ability of CDSS and ATLAS to predict adverse outcomes in these patients. Following initial analysis, we performed a sub-group analysis considering a cohort > 65 years old and those
Results: 488 patients were identified with CDI; 105 were excluded due to insufficient retrospective data to complete the ATLAS score. The mean age was 65.1 ±16.1 years with females accounting for 52.7%. Mean Charlson co-morbidity score was 5.8 ±3.1. 20.1% had prior CDI and 81.5% received antibiotics in the prior 12 weeks. Of 383 pts, 40 (10.4%) died and 122 (31.8%) required ICU admission during their stay. Both ATLAS and CDSS independently predicted adverse outcomes in these pts. In our overall cohort (all ages), there were no significant differences between the two scores in predicting 30-day mortality (Fig 1, p=0.37), however ATLAS performed significantly better predicting ICU requirement (Fig 1, p=0.0002). When considering only those > 65 years, ATLAS performed significantly better than CDSS for 30-day mortality (Fig 2, p< 0.05) and any ICU requirement (Fig 2, p< 0.01). For those ≤ 65, ATLAS was significantly more accurate at predicting ICU admission (p< 0.001) but not mortality (p=0.92).
Discussion: CDI is associated with severe outcomes and prediction of these outcomes could help triage therapeutic decisions. ATLAS is superior to CDSS in prediction of ICU stay in our patient population with CDI and ICU/mortality in age >65. We recommend strong consideration for usage of ATLAS in clinical practice.
Citation: Umer E. Malik, MD; Karthik Gnanapandithan, MD, MS; Prabin Sharma, MD; Muhammad Sohail Mansoor, MD; Paul Feuerstadt, MD, FACG. P0122 - ATLAS SCORES ARE BETTER THAN CDSS IN PREDICTING ANY ICU REQUIREMENT AND 30-DAY MORTALITY IN PATIENTS WITH C. DIFFICILE INFECTION (CDI) >65 YEARS OLD. Program No. P0122. ACG 2019 Annual Scientific Meeting Abstracts. San Antonio, Texas: American College of Gastroenterology.