Risk stratification and bedside prognostication in infective endocarditis.

Abstract

BEFORE THE ADVENT OF ANTIBIOTICS, INFECTIVE ENdocarditis was almost always fatal. With the introduction of penicillin in 1941, endocarditis became a treatable disease. However, many patients still died, predominantly of congestive heart failure resulting from valve destruction. In 1960, Kay et al successfully performed tricuspid valve debridement and closure of a ventricular septal defect in a patient with active Candida endocarditis. For patients with endocarditis, cardiac valve replacement surgery was initially performed to replace damaged valves only after the successful eradication of infection. In 1965, Wallace et al described a patient with active endocarditis who failed to respond to antibiotics alone but was cured by the combination of valve replacement and antibiotics. During the next 2 decades, reports from observational studies argued in favor of the effectiveness of valve replacement in patients with active endocarditis complicated by congestive heart failure or uncontrolled infection. Subsequently, guidelines for surgery for patients with active endocarditis were proposed. In 1985, infectious disease specialists and cardiothoracic surgeons used the medical literature, their local data quantifying the risks of valve replacement, and a crude cost-analysis to construct a point scoring system to assist physicians considering surgery for patients with active endocarditis. For native valve endocarditis, these authors assigned numerical values to 12 different complications in an attempt to weigh their relative importance as indications for surgical intervention. They recommended that any patient who accumulated 5 points should be strongly considered for urgent valve replacement. Severe congestive heart failure, persistent bacteremia, and fungal endocarditis were each assigned 5 points. Consequently, it was recommended that patients with any 1 of these 3 complications be considered for urgent valve replacement. Although this model was helpful, it was never formally validated. Valve surgery is now performed in at least 25% of patients with active endocarditis. A compilation of 26 nonrandomized studies from 15 countries examining the results of surgical treatment of active native valve endocarditis found a mean operative mortality rate of 12%. When Alexiou et al reviewed their experience with 83 patients who underwent replacement of an infected native valve, they found a 79% actuarial survival at 10 years and only 2 cases of infection of the prosthetic valve. The rate of significant postoperative paravalvular leakage is less than 5%. Although valve replacement surgery may be life-saving for some patients with active endocarditis, deciding which patients should undergo surgery is often difficult despite expert opinion and unvalidated scoring systems. Numerous large randomized trials have provided cardiac surgeons with data on the risks and benefits of coronary artery bypass grafting. In contrast, there are no randomized trials of surgery for endocarditis. Deciding whether and when to perform valve surgery in a patient with complicated endocarditis is made on a case-by-case basis. Cardiac surgeons and primary care physicians frequently consult infectious disease specialists, cardiologists, and neurologists to help decide whether to recommend surgery to their patients. Although abundant clinical and laboratory data are available on most patients with endocarditis, the inability to use these data to quantify risk reliably limits optimal decision making. The article by Hasbun and colleagues in this issue of THE JOURNAL attempts to address this problem. These authors are the first to both derive and validate a prognostic classification system for patients with active endocarditis. Their derivation cohort consisted of 259 adults with complicated leftsided native valve endocarditis admitted to 5 Connecticut hospitals between January 1990 and January 2000. Complicated endocarditis was defined as the presence of congestive heart failure, new valvular regurgitation, refractory infection, systemic embolization, or a valvular vegetation. A notable aspect of this case definition is its inclusion of any vegetation observed on echocardiography, irrespective of size. Using clinical and microbiology data, chest radiograph and echocardiographic findings, and 6-month mortality data, the authors developed a weighted scoring system with 5 independent predictors of mortality identified through bivariate