The impact of methodology and confounding variables on the association between major depression and coronary heart disease: Review and recommendations