Inequalities in the use of secondary prevention of cardiovascular disease by socioeconomic status: evidence from the PURE observational study