The association between the timing of valve surgery and 6-month mortality in left-sided infective endocarditis

OBJECTIVE: The optimal timing of valve surgery in left-sided infective endocarditis (IE) is undefined. We aimed to examine the association between the timing of valve surgery after IE diagnosis and 6-month mortality among patients with left-sided IE. METHODS: We analysed data from a retrospective cohort of patients with left-sided IE who underwent valve surgery within 30 days of diagnosis at a tertiary centre. The association between time from IE diagnosis to surgery and all-cause 6-month mortality was assessed using Cox proportional hazards modelling after adjusting for the propensity score (to undergo surgery 0-11 days vs >11 days, median time, after IE diagnosis). RESULTS: Of 546 left-sided IE cases seen between 1980 and 1998, 129 (23.6%) underwent valve surgery within 30 days of diagnosis. The median time between IE diagnosis and surgery was 11 days (range 1-30). There were 35/129 (27.2%) deaths in the surgical group. Using Cox proportional hazards modelling, propensity score and longer time to surgery (in days) were associated with unadjusted HRs of (1.15, 95% CI 1.04 to 1.28, per 0.10 unit change, p = 0.009) and (0.93; 95% CI 0.88 to 0.99, per day, p = 0.03), respectively. In multivariate analysis, a longer time to surgery was associated with an adjusted HR (0.97; 95% CI 0.90 to 1.03). The propensity score and time from diagnosis to surgery had a correlation coefficient of r = -0.63, making multicollinearity an issue in the multivariable model. CONCLUSION: On univariate analysis, a longer time to surgery showed a significant protective effect for the outcome of mortality. After adjusting for the propensity to undergo surgery early versus late, a longer time to surgery was no longer significant but remained in the protective direction. Multicollinearity between the time to surgery and the propensity score may have hindered our ability to detect the independent effect of time to surgery.