BackgroundThe purpose of this study is to identify which variables –among those commonly available and used in the primary care setting– best predict mortality in a cohort of elderly dependent patients living at home (EDPLH) that were included in a home care program provided by Primary Care Teams (PCT). Additionally, we explored the risk of death among a sub-group of these patients that were admitted to hospital the year before they entered the home care program.MethodsA one-year longitudinal cohort study of a sample of EDPLH patients included in a home care programme provided by 72 PCTs. Variables collected from each individual patient included health and social status, carer’s characteristics, carer’s burden of care, health and social services received.Results1,001 patients completed the study (91.5%), 226 were admitted to hospital the year before inclusion. 290 (28.9%) died during the one-year follow-up period. In the logistic regression analysis women show a lower risk of death [OR= 0.67 (0.50-0.91)]. The risk of death increases with comorbidity [Charlson index OR= 1.14 (1,06-1.23)], the number of previous hospital admissions [OR= 1,16 (1.03-1.33)], and with the degree of pressure ulcers [ulcers degree 1–2 OR = 2.94 (1.92-4.52); ulcers degree 3–4 OR = 4.45 (1.90-10.92)]. The logistic predictive model of mortality for patients previously admitted to hospital identified male sex, comorbidity, degree of pressure ulcers, and having received home care rehabilitation as independent variables that predict death.ConclusionsComorbidity, hospital admissions and pressure ulcers predict mortality in the following year in EDPLH patients. The subgroup of patients that entered home care programs with a previous record of hospital admission and a high score in our predictive model might be considered as candidates for palliative care.