Table 3.

Quantitative bias analysis accounting for possible unmeasured confounding due to presence of comorbid conditions at diagnosis

Breast cancer therapy	Observed association, HR (95% CI)	Bias parameters			HRc	Bias-adjusted estimate
		HRCD	HRCE	p
Chemotherapy	1.10 (0.62, 1.96)	4.2	0.8	0.29	0.93	1.19
		4.2	0.8	0.4	0.93	1.19
		4.2	0.7	0.29	0.89	1.24
		4.2	0.7	0.4	0.89	1.24
		7.5	0.8	0.29	0.90	1.22
		7.5	0.7	0.4	0.85	1.29
Hormone therapy	0.94 (0.59, 1.50)	4.2	0.8	0.29	0.93	1.01
		4.2	0.8	0.4	0.93	1.01
		4.2	0.7	0.29	0.89	1.06
		4.2	0.7	0.4	0.89	1.06
		7.5	0.8	0.29	0.90	1.04
		7.5	0.7	0.4	0.85	1.10
Trastuzumab	2.05 (0.76, 5.52)	4.2	0.8	0.29	0.93	2.21
		4.2	0.8	0.4	0.93	2.21
		4.2	0.7	0.29	0.89	2.31
		4.2	0.7	0.4	0.89	2.32
		7.5	0.8	0.29	0.90	2.27
		7.5	0.7	0.4	0.85	2.41

HR_CD is the estimated hazard ratio for the association between comorbid conditions at breast cancer diagnosis and cardiovascular disease mortality. HR_CE is the estimated hazard ratio for the association between comorbid conditions at breast cancer diagnosis and receipt of breast cancer therapies. P is the estimated prevalence of comorbid conditions present at breast cancer diagnosis among breast cancer survivors. HR_C is the estimated hazard ratio due to confounding by comorbid conditions.