Prevalence of malnutrition and 12-month incidence of mortality in two Sydney teaching hospitals

Aims: The objectives of the present study were to determine: (i) the preval ence of malnutrition in two Sydney teaching hospitals using Subjective Glob al Assessment (SGA), (ii) the effect of malnutrition on 12-month mortality and (iii) the proportion of patients previously identified to be at nutriti onal risk. Methods: A prospective study using SGA to assess nutritional status of elig ible inpatients, from April to September 1997, with a 12-month follow-up to assess mortality. A total of 819 patients was systematically selected from 2194 eligible patients. Patients were excluded if they were under the age of 18, had dementia or communication difficulties, or were under obstetric or critical care. The main outcome measures were prevalence of malnutrition , 12-month incidence of mortality, proportion of patients identified with m alnutrition, and hospital length of stay (LOS). Results: The prevalence rate of malnutrition was 36%. The proportion of mal nourished patients was not significantly different between the two hospital s (P = 0.4). The actuarial incidence of mortality at 12 months after assess ment was 29.7% in malnourished subjects compared with 10.1% in well-nourish ed subjects (P < 0.0005). Malnourished subjects had a significantly longer median LOS (17 days vs 11 days, P < 0.0005) and were significantly older (m edian 71 years vs 63 years, P < 0.0005) than well-nourished subjects. Only 36% of the malnourished patients had been previously identified as being at nutritional risk. Conclusions: Malnutrition in Australian hospitals is a continuing health co ncern and is associated with increased LOS and decreased survival after 12 months. The present study revealed that malnourished patients were not regu larly identified. Further studies are required to determine whether routine identification of malnutrition and subsequent nutritional intervention are effective in improving clinical outcomes in these individuals.