ld-archer
commented
1 year ago Possibly also a COVID special? Hypothesis that the number of ageing people who are now widowed/widowers increase due to increased mortality in the 50+ from COVID, so index of social isolation increase by at least 1 point. If we can find some numbers for the amount/proportion of people by demographic group that lost spouse we can create a decent counterfactual to model this.
^^ For the above idea, maybe it would be best to have an index of social isolation separate from marital status. This is because some literature has found strong links between widowhood and things like hypertension and certain inflammatory biomarkers, which accounts for most of the relationship between isolation indexes and the health outcomes. If we want to intervene on this and simulate an increased level of widowhood from COVID, we might have to split them up and do the intervention directly. Would complicate the rest of the work, but only slightly I think.
Would be good to find some government plans on tackling social isolation and loneliness in the ageing population, but not a necessity if nothing perfect exists.
One thing we could do is to find the burden of these risk factors in terms of Life Years and associated outcomes. We can do that by generating an input population where each variable is set to the best possible value (i.e. loneliness avg. set to 1, social isolation index set to 0), and making it impossible for these values to change (either relying on mechanisms inside the FEM to make it impossible to transition to different states, or by switching off the prediction of these variables entirely (i.e. taking them out of the csv file).
We can do this separately and together, and might be really interesting to see how the two risk factors differ in terms of outcomes. This is potentially where we could look at dementia / alzheimers incidence, as I hypothesise that social isolation would have stronger links to this outcome than loneliness.