This prospective, observational consecutive cohort study was performed at Hospital Germans Trias i Pujol, a 650-bed tertiary-care hospital that serves a population of ∼850 000 inhabitants. The institution has a multidisciplinary endocarditis team (created in 2009) that includes specialists in cardiology, infectious diseases, internal medicine, neurology, microbiology, cardiac imaging, and cardiac surgery. This multidisciplinary IE team holds regular meetings to discuss and evaluate the therapeutic strategies for IE patients. We enrolled all consecutive adult patients (≥ 18 years of age) with a definite diagnosis of LSIE (Modified Duke criteria/ESC 2015 modified criteria since 2015) admitted to our center from January 2003 to March 2018. We excluded patients with right-sided and cardiac device-related IE. For patients who were admitted more than once for LSIE during the study period, we recorded only the first admission episode.

Demographic, clinical, diagnostic (imaging, microbiological, and analytical parameters) and follow-up data were obtained from the patients’ medical charts using standardized reporting forms. Health care-associated endocarditis was defined as either IE that developed more than 48hours following hospital admission, or IE acquired as a result of an invasive procedure within the previous 8 weeks before the development of signs and symptoms or diagnosed within the first 48hours of admission in patients with extensive out-of-hospital contact with the health care system (intravenous therapy or hemodialysis).22 The Charlson index23 was used at admission to stratify overall comorbidity. The estimated glomerular filtration rate was measured using the Chronic Kidney Disease-Epidemiology Collaboration method. Moderate or severe renal disease is defined by an estimated glomerular filtration rate less than 60mL/min/1.73 m2 at admission. Surgery was defined as replacement or repair of the affected valve during the index hospitalization. Indications for surgery were validated prospectively during the index episode by the multidisciplinary endocarditis team and included the following: heart failure, embolic event, persistent bacteremia, local/paravalvular complications, severe acute valvular regurgitation, vegetation size, early prosthetic valve IE, and multiresistant microorganisms. The reasons for the lack of surgery were also collected. Patients discharged with a plan for readmission for surgery later during the course of treatment were included in the surgery group.

The primary outcome of this study was all-cause mortality by day 60. The secondary outcome was all-cause death from day 61 to the 3-year follow-up (follow-up begins from the day of the start of antibiotic treatment for endocarditis). After discharge, survivors were prospectively followed up with regular predefined visits (at 1, 3, and 6 months, 1 year, and 3 years) by the IE team. An echocardiogram was performed at each regular visit. Patients not attending a regular visit were contacted by telephone. Otherwise, information was obtained from the patient's clinical record and by consulting the regional government death registry. This study was performed in compliance with the Helsinki Declaration and was approved by the local Ethics Committee of Hospital Universitario Germans Trias i Pujol (Badalona, Barcelona).

Categorical variables are expressed as the number and percentage. Continuous variables are expressed as the mean and standard deviation (SD) or median and 25–75 percentiles, according to their distribution (normal or nonnormal). Normal distribution was assessed using normal Q-Q plots. Statistical significance for comparisons between groups was determined with the chi-square test or Fisher's exact test for categorical variables and Student t tests or Wilcoxon rank sum test for continuous variables. The characteristics of surgically-treated patients were compared with those of SINUS-LSIE patients. Univariate Cox regression analyses were performed including all database variables potentially associated with primary and secondary endpoints.

A multivariable Cox regression model was performed with adjustment for variables that were significantly associated with the outcome (P <.05) and other confounders. Covariates in the multivariable model for 60-day mortality were age, sex, health care-associated IE, unknown etiology, leukocyte count, platelet count, estimated glomerular filtration rate, heart failure, atrioventricular block, shock, Staphylococcus aureus infection, Enterococcus infection, and EuroSCORE. Covariates in the multivariable model for the secondary endpoint were age, sex, previous IE, coronary artery disease, diabetes mellitus, Charlson index, hemoglobin, estimated glomerular filtration rate, shock, Enterococcus infection, and EuroSCORE. Considering the difficulty of quantifying the prognostic impact of surgery in an unbiased manner, a propensity score analysis was also performed with a logistic regression model. The dependent variable was the performance of cardiac surgery for IE during the index hospitalization and a total of 24 covariates were selected (table 1 of the supplementary data). The prediction accuracy of the logistic model was assessed with an area under the receiver-operating characteristic curve (c-statistic), which was 0.82.

According to the propensity score, patients were selected by 1:1 matching without replacement using the nearest neighbor method. A caliper width of 0.2 standardized differences was used for matching. Furthermore, we assessed the balance of the distribution of the baseline characteristics between the 2 groups by evaluating the absolute standardized differences. Among SINUS-LSIE patients, Cox regression analyses were used to identify variables associated with the 60-day mortality. Cumulative survival probability plots were generated. All comparisons of statistical significance were 2-sided, and a P value <.05 was considered significant. STATA V.13.0 (College Station, Texas, United States) was used for all analyses.

During the study period, 353 patients with LSIE were admitted to our institution and 271 had surgical indications and were included in the analysis. The flowchart of the study is depicted in figure 1. Median follow-up among survivors was 4.3 (2.3-8.4) years. The overall cohort included 201 (74.2%) males and the mean age was 64.3 years. Health care-associated IE was noted in 37.3% of patients and 96 (35.4%) had prosthetic valve IE. About 50% of our LSIE patients were transferred from other community hospitals. There was evidence of previous antibiotic administration in 28% of patients and in 65% of subjects with unknown etiology. The baseline characteristics of the surgical and nonsurgical patients (unmatched and propensity score matched cohorts) are detailed in table 1. Surgery was performed in 188 patients (53%), while 83 patients did not undergo surgical treatment despite having indications for it. SINUS-LSIE patients more frequently had a history of IE and health care-associated IE and had higher comorbidities as expressed by the Charlson index than patients who underwent surgical treatment. They also had greater deterioration of renal function and higher levels of C-reactive protein at admission. There were no differences between the groups regarding cardiac valvular involvement of IE, such as new moderate or severe aortic regurgitation, valve perforation, abscess, or paravalvular complications. Patients who underwent surgery were more likely to have infection caused by Streptococcus species in contrast to SINUS-LSIE patients who were more frequently infected with S. aureus. During admission, stroke and systemic embolization were more frequent in SINUS-LSIE patients. Severe valvular regurgitation, local complications, and heart failure were the main indications for cardiac surgery in both groups (table 2 of the supplementary data). In SINUS-LSIE patients, the main reasons for the lack of surgery, despite indication, were high surgical risk (66.7%), death before surgery (13.1%), stroke (10.7%), and patient refusal (2.4%).

Figure 1.

Flow diagram of the study population. IE, infective endocarditis; SINUS-LSIE, surgery-indicated not undergoing surgery patients with left-sided infective endocarditis.

(0.06MB).

At the 60-day follow-up, 40 patients (21.3%) who underwent surgical treatment and 53 SINUS-LSIE patients (63.9%) died (P <.001, figure 2). From day 61 to the 3-year follow-up, 21 patients (11.1%) who underwent surgical treatment and 11 SINUS-LSIE patients (13.3%) died (P=.624). Among all surgery-indicated patients, multivariable Cox regression analyses showed that unknown etiology (hazard ratio [HR], 2.19; 95% confidence interval [95%CI], 1.07–4.50; P=.033), SINUS-LSIE (HR, 3.59; 95%CI, 2.16-5.96; P <.001), heart failure (HR, 1.73; 95%CI, 1.07-2.80; P=.027), atrioventricular block (HR, 2.04; 95%CI, 1.14-3.62; P=.016), and shock (HR, 1.64; 95%CI, 1.01–2.69; P=.048) were independent predictors of the occurrence of the primary endpoint (table 2).

Figure 2.

Survival in surgically-treated and SINUS-LSIE. SINUS-LSIE, surgery-indicated not undergoing surgery patients with left-sided infective endocarditis.

(0.09MB).

Similarly, and in multivariable model, the independent variables associated with all-cause death from day 61 to the 3-year follow-up were previous IE (HR, 15.54; 95%CI, 4.44-54.4; P <.001), diabetes mellitus (HR, 3.17; 95%CI, 1.13-8.90; P=.028), and the Charlson index (HR, 1.27; 95%CI, 1.07-1.51; P=.007), as shown in table 3 of the supplementary data. Of note, mortality rates from day 61 to the 3-year follow-up were similar in surgically-treated and SINUS-LSIE patients (figure 2) and no significant differences in the risk of death between surgically-treated and SINUS-LSIE patients were found (HR, 1.89; 95%CI, 0.68-5.19; P=.220) (table 3). The characteristics of patients included in the long-term analysis are shown in table 4 of the supplementary data.

Propensity score matching yielded 66 matched pairs of patients from the 2 groups. The distribution of created propensity score and the absolute standardized differences in the unmatched and matched patients are presented in figure 1 of the supplementary data. Risk of 60-day mortality was higher in SINUS-LSIE patients (HR, 3.17; 95%CI, 1.77-5.67; P <.001), but there were no significant differences in the risk of death from day 61 to the 3-year follow-up (HR, 1.25; 95%CI, 0.42-3.76; P=.679) (table 3).

In the subgroup of SINUS-LSIE patients and after multivariable Cox regression analysis, the variables independently associated with 60-day mortality were shock, S. aureus infection, new moderate or severe valve regurgitation, and leukocyte count (table 4). Exploratory analyses of SINUS-LSIE patients infected by S. aureus showed that this infection was more frequently related to health care-associated IE (table 5 supplementary data).