The mortality prediction represents a key factor in the managing of elderly hospitalized patients. Since in older subjects mortality results from a combination of biological, functional, nutritional, psychological and environmental factors, a Multidimensional Prognostic Index (MPI) that predict short- and long-term mortality based on a standardized comprehensive geriatric assessment (CGA) has recently been developed and validated.

This study compares the accuracy in predicting the mortality of the MPI with a modified version of the MPI (m-MPI) that included the Mini Nutritional Assessment-Short Form (MNA-SF) instead of the standard MNA.

This prospective study with a one-year follow-up included 4088 hospitalized patients aged 65 years and older. A standardized CGA that included information on functional (Activities of Daily Living, ADL and Instrumental-ADL), cognitive (Short Portable Mental Status Questionnaire), risk of pressure sore (Exton-Smith Scale), comorbidities (CIRS Index), medications, living status and nutritional status (MNA and MNA-SF) was used to calculate the MPI using a previously validated algorithm.

Higher MPI values were significantly associated with higher mortality rates with a close agreement between the estimated and the observed mortality both after 1-month (MPI1=2.8% versus m-MPI1=2.8%, p=0.946; MPI2=8.9% versus m-MPI2=9%, p=0.904; MPI3=21.9% versus m-MPI3=21.9, p=0.978) and 1-year of follow-up (MPI1=10.8% versus m-MPI1=10.5%, p=0.686; MPI2=27.3% versus m-MPI2=28%, p=0.495; MPI3=52.8% versus m-MPI3=52.7%, p=0.945). The estimated areas under the receiver operating characteristics (ROC) curves suggested a clinically negligible difference between the two indices.

The m-MPI is as sensitive as the MPI in stratifying hospitalized elderly patients into groups at varying risk of short- and long-term mortality, but with fewer items.