Between 2008-2017, using the Optum longitudinal database, a patient-level database that integrates multiple US-based electronic health and claim records from several health care providers, we identified 435 679 patients with new HF diagnosis and both an assessment of the left ventricular ejection fraction and at least 1 year of history. TR was graded as mild, moderate or severe and classified as prevalent (at the time of the initial HF diagnosis) or incident (subsequent new cases thereafter). For prevalent TR, the analysis was performed using a Cox proportional hazards model with adjustment for patient covariates. Incident TR was modelled as a time-updated covariate, as were other non-fatal events during follow-up. Prevalence of mild, moderate and severe TR at baseline was 10.1%, 5.1% and 1.4%, respectively. Over a median follow-up of 1.5 years, 121 273 patients (27.8%) died and prevalent TR was independently associated with survival. Compared to patients with no TR at baseline, the adjusted hazard ratios for mortality were 0.99 [95% confidence interval (CI) 0.97-1.01], 1.17 (95% CI 1.14-1.20) and 1.34 (95% CI 1.28-1.39) for mild, moderate and severe TR, respectively. In the 363 270 patients free from TR at baseline, incident TR (at least mild, at least moderate, or severe) developed during follow-up in 12.1%, 5.1% and 1.1%, respectively. Adjusted mortality hazard ratios for such new cases were 1.48 (95% CI 1.44-1.52), 1.92 (95% CI 1.86-1.99) and 2.44 (95% CI 2.32-2.57), respectively. Findings were consistent across all patient subgroups based on age, gender, rhythm, associated comorbidities, prior cardiac surgery, B-type natriuretic peptide/N-terminal pro-B-type natriuretic peptide, and left ventricular ejection fraction.