| Abstrakt: |
Background:Guidelines advise relaxing glycemic control goals in diabetes patients with coronary artery disease (CAD). However, the hemoglobin A1c (HbA1c) trajectories that occur in routine clinical care in diabetes patients with objectively assessed CAD have not been described, and their association with mortality is unknown.Methods:We studied 7780 individuals with diabetes using data from the Veterans Affairs Clinical Assessment, Reporting, and Tracking Program, a registry of US Veterans who have undergone coronary angiography since 2005. We used HbA1c values from the measurement closest in time preceding catheterization and all values during two years of follow-up to fit longitudinal latent class models. We determined optimal trajectory model fit by Bayes Information Criteria, varying the functional form of time and the numbers of HbA1c trajectory classes. We then fit a joint latent class longitudinal mixed model to estimate associations between HbA1c trajectory class and two-year mortality, adjusting for clinical and demographic covariates, including an interaction term between HbA1c trajectory and CAD burden (no, non-obstructive, or obstructive), and with an autoregressive correlation structure for repeated HbA1c measurements.Results:Three trajectory classes best fit the data: individuals with stable glycemia after catheterization (class 1; 89%, 6934 of 7780), those with a decline in HbA1c after catheterization (class 2; 4.7%, 364 of 7780), and those with an increase in HbA1c after catheterization (class 3; 6.2%, 482 of 7780). Class 1 participants were older, more likely to be white, less likely to have congestive heart failure, and more likely to be adherent to cardioprotective medications. In multivariable joint mixed models, two-year mortality was 4.3% in class 1, 4.7% in class 2, and 5.0% in class 3 and differed significantly across HbA1c trajectory classes (p=0.047). In pairwise comparisons, two-year mortality differed significantly between individuals in classes 1 and 3 (p=0.03), but not between those in classes 1 and 2 (p=0.9) or between those in classes 2 and 3 (p=0.5). The interaction between trajectory class and CAD burden was non-significant (p=0.1), but limited by the number of participants with no or non-obstructive CAD. In individuals with obstructive CAD, we observed a significant association between HbA1c trajectory and mortality (p=0.04), driven by a difference between classes 1 and 3 (p=0.04) as in the full cohort.Conclusions:Distinct HbA1c trajectories were evident within 6 months after cardiac catheterization in patients with diabetes and were associated with two-year mortality. Serial HbA1c measurements in outpatient follow-up after cardiac catheterization could classify individuals with diabetes based on HbA1c trajectory; these trajectories may inform mortality risk stratification, especially in those with obstructive CAD. |
