Cancer inpatient malnutrition risk, documentation, and ICD-10 coding in an academic medical center.

Abstract

12107 Background: Malnutrition (MN) is common in hospitalized cancer patients but often underdiagnosed. We evaluated the prevalence of MN risk, dietitian documented MN (DDMN), and physician coded malnutrition (PCMN) in a consecutive cohort of cancer inpatients in an academic, community-based medical center. Methods: Electronic medical records (EMR) were reviewed for inpatients with a solid tumor diagnosis staged I-IV and admitted to Atrium Health Carolinas Medical Center at least once between 1/1/2016 to 5/21/2019. All data were collected from the first admission EMR encounter closest to the cancer diagnosis date. High MN risk was a score ≥2 on the Malnutrition Screening Tool (MST) completed by an RN at admission. Registered Dietitian (RD) assessments were reviewed for DDMN and grade (mild, moderate, severe). PCMN diagnosis was based on MN ICD-10 codes extracted from the medical coder’s discharge summary. Multivariate logistic regression models identified associations between clinic-demographic factors and the prevalence of DDMN and PCMN with stepwise selection. Results: N=5,143; 48% females. Median age 63 (range 18-102) years. 70% White; 24% Black, 3% Latino. Most common cancers: thoracic 19% and digestive system (14% other, 11% colorectal). 28% had known stage IV disease. The MST was completed in 79%. Among those with MST ≥2 (N=1,005; 25%), DDMN and PCMN prevalence was 30% and 22%, respectively. In the entire cohort, 8% had DDMN; 7% PCMN; 4% both. Prevalence of MN risk, DDMN, and PCMN by cancer site are in the Table. DDMN (N=420) was mild 2%, moderate 16%; severe 66%; unspecified 16%. On discharge, PCMN (N=360) was mild 10%; moderate 0%; severe 69%; unspecified 21%. Male gender (OR 1.27 [1.01, 1.59]), Black race (OR 1.57 [1.25, 1.98]), stage IV disease (v. I-III) (OR 3.08 [2.49, 3.82]), and primary site were all independent predictors of DDMN (all p<0.05); Black race (OR 1.46 [1.14, 1.87]), stage IV disease (OR 2.70 [2.15, 3.39]), and primary site were independent predictors of PCMN (all p<0.05). Conclusions: 25% of cancer inpatients were at high risk for MN. Primary site, disease stage, and race were independent predictors of a greater risk. MN appears to be under-diagnosed compared to population studies. This is the first study to report the prevalence of MN in a large cancer inpatient database with a representative population.[Table: see text]