Table 6.

Associations Between Pre-Existing HF and Cause-Specific Mortality

	HR (95% CI)
	Lymphoma Mortality	Cardiovascular Mortality	Nonlymphoma Cancer Mortality	Non-CV, Noncancer Mortality
Cohort sample size	3,331	3,331	3,331	3,331
Number of events	1,300	368	257	464
Unadjusteda	1.50 (1.29-1.75)	3.36 (2.61-4.31)	1.42 (0.97-2.07)	1.97 (1.53-2.53)
Model A (adjusted for clinical variables)b	1.23 (1.05-1.44)	2.61 (2.00-3.39)	1.22 (0.82-1.81)	1.45 (1.11-1.89)
Model B (adjusted for clinical variables, SDOH, and hospital variables)c	1.25 (1.06-1.46)	2.57 (1.96-3.36)	1.21 (0.82-1.80)	1.42 (1.09-1.86)
Model C (clinical and treatment variables)d	1.12 (0.95-1.31)	2.40 (1.83-3.16)	1.12 (0.74-1.69)	1.33 (1.02-1.73)

Cox proportional hazards model using competing risks for 4 different cause-specific mortalities. Comparison is between patients with pre-existing HF and those without pre-existing HF.

Abbreviations as in Tables 2 and 5.

^aUnadjusted includes pre-existing HF as only independent variable.

^bAdjusted for age (including an age-time interaction term), sex, race, Hispanic ethnicity, diabetes, any prior cancer diagnosis, chronic bronchitis or emphysema, dementia, and moderate or severe renal dysfunction and stratified by advanced cancer stage (III or IV vs I or II).

^cAdjusted for model A variables plus Surveillance, Epidemiology, and End Results region; metropolitan, nonurban metropolitan, or rural; Medicaid dual eligibility; National Cancer Institute cancer center designation; hospital medical school affiliation; number of beds; and stratification for census tract poverty indicator.

^dAdjusted for model B variables plus number of anthracycline claims (time varying), radiation (time varying), and cardioprotective medications (dexrazoxane or liposomal formulations, time varying).