Evaluating the concurrent validity of body mass index (BMI) in the identification of malnutrition in older hospital inpatients.

BACKGROUND: Nutrition screening and assessment tools often include body mass index (BMI) as a component in identifying malnutrition risk. However, rising obesity levels will impact on the relevancy and applicability of BMI cut-off points which may require re-evaluation. This study aimed to explore the relationship between commonly applied BMI cut-offs and diagnosed malnutrition.

METHODS: Data (age, gender, BMI and Subjective Global Assessment (SGA) ratings) were analysed for 1152 inpatients aged ≥65 years across annual malnutrition audits (2011-2015). The receiver operation characteristic (ROC) curve analysed the optimal BMI cut-off for malnutrition and concurrent validity of commonly applied BMI cut-offs in nutritional screening and assessment tools.

RESULTS: Malnutrition prevalence was 36.0% (n = 372) using SGA criteria (not malnourished, moderate or severe malnutrition). Median age was 78.7 (IQR 72-85) years, median BMI 25.4 (IQR 21.8-29.7) kg/m2 ; 52.1% male and 51.2% overweight/obese. ROC analysis identified an optimal BMI cut-off of <26 kg/m2 , 80.8% sensitivity and 61.5% specificity (AUC 0.802, 95% CI 0.773, 0.830; p < 0.0001). Commonly applied BMI cut-offs (between 18.5 and 23 kg/m2 ) failed to meet the alpha-priori requirement of 80% sensitivity and 60% specificity. However, BMI <23 kg/m2 had the highest agreement (κ = 0.458) with malnutrition diagnosed using the SGA.

CONCLUSIONS: Both malnutrition and overweight/obesity are common in older inpatients. Continuing increases in the prevalence of overweight and obesity will impact on the sensitivity of BMI as a screening component for malnutrition risk. The current study suggests tools developed over a decade ago may need to be revisited in future.