The prevalence of patient-GP communication about medical EoL treatment preferences and patient appointment of a surrogate decision-maker (n = 4,396)a.

a% of missing observations ranged from 0.3–1.2%.bmultivariable logistic regressions (forced enter). Dependent variables were ‘Patient did not discuss a medical EoL preference with GP or appoint a surrogate decision-maker’; ‘Patient discussed a medical EoL preference’; ‘Patient appointed a surrogate decision-maker’; ‘Patient appointed a surrogate decision-maker in writing’; and ‘Patient only appointed a surrogate decision-maker verbally’. Independent variables included country (OR and p-value shown), age, cause of death, dementia diagnosis; place of death; the number of contacts with the GP in the last week and in the second and third months before death; GP palliative care provision; the importance of curative, life-prolonging and palliative care as treatment aims and if the GP had discussed the primary diagnosis. The results of the multivariate logistic regressions were compared with equivalent univariate analyses (not shown) to check for any major differences in the magnitude or direction of associations.cToo few patients in this category to conduct a logistic regression.