OBJECTIVES: Since about ten years ago, various coordinated strategies for people with chronic diseases care have been implemented. Stratification, based on predictive models, has become the most important axis on which these strategies gravitate. So far, predictive models do not take socioeconomic and environmental inequalities into account. As a consequence, some patients may be poorly stratified, which could have important implications regarding the main strategy of care, the use of specialized hospital care and, therefore, in the allocation of available resources.

Two are our objectives. First, the use of Real World Data for development of an algorithm to evaluate the complexity of chronic patients controlling for socioeconomic and environmental inequalities. Second, we estimate the probability of survival in the next 12 months of people with chronic diseases, controlling for socioeconomic and environmental inequalities (exposure and differential susceptibility).

METHODS: We present the results of an analysis in which we use Real World Data combining two types of data sources: i) population-based cohort consisting of 21546 individuals aged 15 and over, living in a region of Girona, Catalonia, Spain and followed from 2005 to 2012; ii) contextual indicators: deprivation index (and an environmental quality index. Contextual variables are linked to the individual data from the census section in which the individual resides. We use an Andersen-Gill survival model in which we control for observed confounders (sex, age, occurrence of cancer, smoking and enol habit) and unobserved confounders.

RESULTS: The hazard ratio of dying (HR) having 3 chronic diseases goes from 1.21 (95% CI : 1.07-1.70) to 1.26 (95% CI: 0.66-2.38) when we control for socioeconomic and environmental inequalities; and the HR having 4 or more chronic diseases from 1.49 (95% CI: 1.06-2.11) to 1.10 (95% CI: 1.01-2.18).

CONCLUSIONS: Taking socioeconomic and environmental inequalities into account greatly varies both the stratification of the individual.