707 Platelet to Albumin Ratio: An Early Predictor of Colectomy in Acute Ulcerative Colitis

Abstract

INTRODUCTION: Acute ulcerative colitis (UC) is associated with significant morbidity and comes with a high risk of surgery. Prior clinical predictors have often included Day 3 laboratory values to assess risk of colectomy. Utilization of readily available data on admission may expedite medical decisions allowing for prompt initiation of therapies. Our aim was to identify early laboratory predictors of colectomy at the time of admission. METHODS: We designed a retrospective study including patients >18 years old with an admission for acute UC at a single academic center between 1/1/2013 and 4/1/2018. Cases were identified using the ICD-9 code 556.X and ICD-10 code K51.X and separately manually verified. Clinical variables of interest and laboratory values were obtained via chart review or extracted from the electronic medical record. Statistical analysis was conducted using JMP ® 13.1.0. Data was analyzed using Wilcoxon rank-sum test for continuous variables, Fischer's exact test for categorical variables, followed by multivariate logistic regression for predictive modeling of in-hospital colectomy. RESULTS: 262 patients admitted with acute UC were reviewed. 59 (22.5%) required inpatient colectomy. 188 of these patients had complete platelet and albumin laboratory data. Basic demographic variables and significant predictors of colectomy in univariate analysis are described in Table 1. Platelet to Albumin ratio (PAR) was found to be the strongest assessed univariate predictor of colectomy (AUC = 0.72, P-value < 0.0001). The optimal PAR cutoff to predict colectomy was 129 (Sensitivity = 70%, Specificity = 72%). Final multivariate logistic regression model with only significant covariates included PAR score (categorical), outside hospital transfer, and number of prior biologics with an area under curve (AUC) = 0.84. PAR remained statistically significant in this model (OR: 6.48, 95% CI: 2.73-15.34). Admission platelet count and admission albumin were also predictors of colectomy in both univariate and separate multivariate analyses, whereas CRP on admission, CRP on Day 3, and CRP to Albumin ratio were not predictive of colectomy (Table 2). CONCLUSION: PAR ratio was found to be a robust predictor of in-hospital colectomy in acute UC. This simple laboratory tool may provide a means to quantify early risk of in-hospital colectomy. As such, an elevated PAR identified on admission can help guide therapeutic strategies aimed at mitigating risk for colectomy in high risk groups.