Frailty Indices and Nutritional Screening Tools as Predictors of Adverse Outcomes in Hospitalised Older People

Frailty and malnutrition are two major medical issues influencing the health of older people. This doctoral thesis investigated the predictive ability and discriminatory power of clinically applicable frailty instruments and their malnutrition counterparts - nutritional screening tools (NSTs). The study was prospective and observational by design, and included patients aged ≤ 70 years consecutively admitted to the Geriatric Evaluation and Management Unit (GEMU) at The Queen Elizabeth Hospital, South Australia. Thesis aims were to: (i) identify the prevalence rates of malnutrition and frailty in hospitalised older people and (ii) determine the predictive ability and accuracy of these measurements. The mean (standard deviation) age of patients was 85.2 (6.4) years; 123 (72 %) were female, n = 172. Malnutrition and frailty prevalence rates were high: malnutrition was found in 53 (31 %) of patients using the Mini Nutritional Assessment (MNA) for classification; and frailty was found in 107 patients (62 %) by the Cardiovascular Health Study (CHS) frailty index. When looking at nutritional screening tools as predictors of hospital discharge outcomes: the MNA and the MNA-short form (MNA-SF) were associated with length of stay (LOS); the Geriatric Nutritional Risk Index (GNRI) and calf circumference (CC) were associated with functional decline; and mid arm circumference (MAC) was associated with a higher level of care on discharge. At six months post-hospitalisation, malnutrition by the MNA (OR = 3.29) and GNRI (OR = 2.84) was predictive of poor outcome (defined as mortality or admission to high level care). However the discriminative ability of this prediction was inadequate (area under Receiver Operating Characteristic curve (auROC) values were < 0.7). iii Regarding frailty, almost all frailty and functional decline indices were predictive of poor outcome (mortality or high level care admission) at both hospital discharge and at six month post-hospitalisation. However when discriminative ability was considered, only the Frailty Index of Cumulative Deficits (FI-CD) and the adapted Katz score of Activities of Daily Living showed adequate values (auROC values of 0.735 and 0.704 respectively). The FI-CD was the only instrument to show adequate discriminatory power in predicting poor six month outcome (auROC = 0.702, P < 0.001). Malnutrition shares many characteristics with frailty; however the overlap between these two conditions lacks a quantitative foundation. Therefore, this doctoral project also looked at the efficacy of nutritional screening tools as frailty indices in hospitalised older people. An additional focus of this thesis was the association between appetite, body composition and inflammation in healthy people of all ages. This thesis illustrated the high prevalence rate of both malnutrition and frailty in hospitalised older people. Results highlight the importance of research into the predictive ability of both NSTs and frailty instruments in hospitalised older people. Such knowledge will be of assistance in the areas of gerontology research, clinical practice and public health policy, particularly in the wake of the global expansion of the number of older people. Thesis results may also assist in standardising definitions for both frailty and malnutrition, definitions which are greatly needed in clinical practice and research.