Figure 3. Time-Varying Hazard Ratios Comparing Surgical and Matched Nonsurgical Patients.

Time-varying hazard ratios comparing the risk of incident macrovascular disease and death following bariatric surgery vs matched nonsurgical patients. Separate estimates for coronary artery disease events (B) and cerebrovascular events (C) are shown, as well as a composite estimate for incident macrovascular disease due to either of these event classes (A) and all-cause mortality (D). The median follow-up time among surgical patients was 4.7 years (interquartile range, 3.2-6.2), and among matched nonsurgical patients was 4.6 years (interquartile range, 3.1-6.1). All analyses used the full cohort of patients who underwent surgery (n = 5301) and matched controls (n = 14 934). Shaded areas represent the 95% CIs.