Background: The Multidimensional Prognostic Index (MPI) has been found to predict mortality in patients with a variety of clinical conditions. We aimed to assess the association of the MPI with future mortality and number of in-hospital days for the first time in a population-based cohort.
Methods: The study population consisted of 2472 persons, aged 66-99 years, from the Swedish National Study on Aging and Care in Kungsholmen, Sweden, who underwent the baseline visit 2001-4, and were followed up >10 years for in-hospital days and >12 years for mortality. The MPI was a modified version of the original and aggregated seven domains (personal and instrumental activities of daily living, cognitive function, illness severity and comorbidity, number of medications, co-habitation status, and nutritional status). The MPI score was divided into risk groups: low, medium and high. Number of in-hospital days (within 1, 3 and 10 years) and mortality data were derived from official registries. All analyses were age-stratified (sexagenarians, septuagenarians, octogenarians, nonagenarians).
Results: During the follow-up 1331 persons (53.8%) died. Laplace regression models, suggested that median survival in medium risk groups varied by age from 2.2-3.6 years earlier than for those in the corresponding low risk groups (p = 0.002-p<0.001), and median survival in high risk groups varied by age from 3.8-9.0 years earlier than for corresponding low risk groups (p<0.001). For nonagenarians, the median age at death was 3.8 years earlier in the high risk group than for the low risk group (p<0.001). The mean number of in-hospital days increased significantly with higher MPI risk score within 1 and 3 years for people of each age group.
Conclusion: For the first time, the effectiveness of MPI has been verified in a population-based cohort. Higher MPI risk scores associated with more days in hospital and with fewer years of survival, across a broad and stratified age range.