Factors Affecting Mortality in Patients with Blood-Culture Negative Infective Endocarditis

Abstract

Infective endocarditis retains high morbidity and mortality rates despite recent advances in diagnostics, pharmacotherapy, and surgical intervention. Risk stratification in endocarditis patients, including blood-culture negative endocarditis, is crucial in deciding the optimal management strategy; however, the studies investigating risk stratification in these patients were lacking despite the difference with blood-culture positive endocarditis. The aim of this study is to identify risk factors associated with in-hospital mortality in blood-culture negative infective endocarditis patients. A retrospective cohort study was conducted at National Cardiovascular Center Harapan Kita, Jakarta in blood-culture negative infective endocarditis patients from 2013 to 2015. Patient characteristics, clinical parameters, echocardiographic parameters, and clinical complications were collected from medical records and hospital information systems. There were 146 patients that satisfy the inclusion and exclusion criteria out of 162 patients with blood-culture infective endocarditis. The in-hospital mortality rate was 13.5%. On bivariate analyses, factors that were related to in-hospital mortality include New York Heart Association (NYHA) class III and IV heart failure ( p  = 0.007), history of hypertension ( p  = 0.021), stroke during hospitalization ( p  < 0.001), the decline in renal function ( p  < 0.001), and surgery ( p  = 0.028). Variables that were independently associated with mortality upon multivariate analysis were heart failure NYHA functional class III and IV (OR 7.56, p  = 0.011), worsening kidney function (OR 10.23, p  < 0.001), and stroke during hospitalization (OR 8.92, p  = 0.001). Presence of heart failure with NYHA functional class III and IV, worsening kidney function, and stroke during hospitalization were independently associated with in-hospital mortality in blood-culture infective endocarditis patients.

Infective endocarditis (IE) has high morbidity and mortality rates (20–30%) despite recent advances in diagnostics, pharmacotherapy, and surgical intervention. There are several aspects on the treatment of IE that are not well-defined, some of them are the management of blood-culture negative endocarditis (BCNIE), choice of empiric antibiotics, management of stroke and systemic embolism, and optimal timing of surgery. The prevalence of BCNIE in developing countries was <10% of all IE patients and 21 to 67% in the developing countries. ¹ These variations may have resulted from the differences between diagnostic criteria, sampling, the difference in the epidemiology of zoonosis microorganism from different regions, administration of antibiotics before collection of blood culture, and unknown pathogens. The mortality rate in BCNIE patients ranged from 2.5 to 31.4%. ^{1 2} There were no national data on the epidemiology of both IE and BCNIE at the time this article was written; in National Cardiovascular Center Harapan Kita (NCCHK) the total of IE patients in the last 5 years (2013–2017) were 283 patients.

IE with negative blood culture using standardized method is called BCNIE. ^{3 4} This may result from previous antibiotic use, endocarditis caused by a slow-growing microorganism, or noninfective endocarditis. ⁵ Patients with BCNIE were associated with worse prognosis, including a higher rate of heart failure and valve perforation compared to those with positive blood culture. ^{2 6}

Risk stratification in IE patients, including BCNIE, is crucial in deciding the optimal management strategy to lower the rate of mortality and morbidity in these patients. However, the studies investigating risk stratification in BCNIE patients were lacking. Most of the prognostic studies selected mortality as their primary outcome. These studies usually include both positive or negative culture IE, and in most of these studies, the negative blood-culture IE is only a small proportion of the study sample despite the different characteristics between these two groups. ^{7 8 9} The results also vary widely between countries presumably due to the difference in patient characteristics and health care quality; hence, the factors reported to be associated with the worse outcome in one study may not be representative in a different population. The aim of this study is to identify risk factors associated with in-hospital mortality in BCNIE patients.

Methods

This study is a cohort retrospective study that included all patients with blood-culture negative infective endocarditis that were hospitalized in NCCHK, Jakarta, Indonesia with or without surgical intervention between January 2013 and March 2018. The inclusion criterion was patients above 18 years old with definite and possible IE diagnosis. The exclusion criterion was patients with incomplete medical record. This study was approved by NCCHK Research Ethics Committee.

Patient characteristics, clinical parameters, echocardiographic parameters, and clinical complications were collected retrospectively from medical records and hospital information systems. Diagnosis of IE was defined as definite and possible based on modified Duke Criteria. Negative blood culture was defined as the absence of a microorganism in blood culture within 5 to 7 days.

Data Analysis

Numerical data with normal distribution are presented with mean ± standard deviation. Numerical data with abnormal distribution are presented with a median (minimum–maximum value). Categorical data are presented with the total number and percentage (%). The numerical variable was converted into categorical data according to the cut-off value. The association of risk factors and in-hospital mortality was analyzed using Chi-square test and p  < 0.25 was included in multivariate analysis using logistic regression. Data were analyzed using SPSS version 21.

Results

A total of 146 out of 162 BCNIE patients were included in this study. The median age was 38 years old, with 18 patients (12.3%) >60 years old. Women comprised 34.9% of the patients, and most of the patients were referred from other hospitals (75.3%). History of fever was present in 83.6% of the cases and prior antibiotic use in 28.1% of the patients. Diagnosis of definite IE was established in 55.5%, with the remaining 44.5% given diagnosis of possible IE. Coronary artery disease (CAD) was present in 12% patients, and conventional CAD risk factors, such as diabetes and hypertension, were found in 9.6% of the patients, respectively. Degenerative valve disease, rheumatic heart disease, and congenital heart disease were the most prevalent IE predisposing factors in this study. Sixty percent of the patients suffered from heart failure with New York Heart Association (NYHA) functional class III and IV. Prior history of cardiac surgery was found in 24 patients (16.4%), of these patients 19 of them had prosthetic valves. Echocardiography results showed that most of the patients (85.6%) had good left ventricular (LV) systolic function. Vegetation was most frequently found on mitral valve followed by aortic valve, with only a small proportion right-sided vegetation found on echocardiography. Large vegetation (>10 mm) was found in 41.5% of the patients ( Table 1 ).

Table 1. Baseline characteristics of the research subjects ( n  = 146) .

Variable	Value a
Age (y)	38 (18–79)
Gender
Male	95 (65.1)
Female	51 (34.9)
Clinical characteristics
Fever	122 (83.6)
Functional class
NYHA I-II	62 (42.5)
NYHA III-IV	84 (57.5)
Arrhythmia (in-hospital)	27 (18.5)
Prior antibiotic use	41 (28.1)
Prior stroke	6 (4.4)
History of cardiac surgery	24 (16.4)
BMI (kg/m 2 )	20.1 ± 3.6
Leukocyte >8,150/mm 3	115 (78.8%)
Predisposing factors
Degenerative valve disease	68 (46.6)
Prosthetic valve	19 (13.0)
Congenital heart disease	26 (19.3)
Rheumatic heart disease	26 (17.8)
Unknown	7 (4.8)
Comorbidities
Hypertension	13 (9.6)
Diabetes	13 (9.6)
Coronary artery disease	16 (11.9)
Diagnosis
Definite EI	81 (55.5)
Possible EI	65 (44.5)
Echocardiography
Left ventricle ejection fraction <50%	21 (14.4)
TAPSE <17 mm	28 (19.2)
Size of vegetation
≤ 10 mm	85 (58.2)
> 10 mm	61 (41.8)
Side of vegetation
Left-sided b	124 (84.9)
Right-sided	22 (15.1)
Location of vegetation
Aortic valve	33 (22.6)
Mitral valve	77 (52.7)
Aortic and mitral valve	7 (4.8)
Tricuspid valve	6 (4.1)
Pulmonary valve	10 (6.8)
Others	13 (8.9)
Perivalvular complications	7 (4.8)
Management
Surgery during active IE phase	52 (35.6)
Ceftriaxone and gentamycin antibiotics	135 (92.5)

Abbreviations: BMI, body-mass index; IE, infective endocarditis; NYHA, New York Heart Association; TAPSE, tricuspid annular plane systolic excursion.

^aNumerical data with normal distribution is presented with mean ± standard deviation. Numerical data with abnormal distribution is presented with median (minimum–maximum value). Categorical data are presented with total number and percentage (%).

^bIncluding patients with both left and right-sided vegetations.

The in-hospital mortality rate of BCNIE in this study was 13.5% (22 out of 162 patients). Adverse events during hospitalization include stroke, worsening renal function, arrhythmia, sepsis, acute pulmonary edema, and requiring mechanical ventilation that was unrelated to perioperative surgical requirements ( Table 2 ). Worsening renal function was most frequently observed. Analysis between the independent variables and in-hospital mortality is presented in Table 3 . There was a significant association between heart failure NYHA functional class, history of hypertension, worsening renal function, stroke during hospitalization, and surgery during the active phase of endocarditis with in-hospital mortality. This result of bivariate analysis was correlated with multivariate analysis showing that heart failure NYHA class III and IV, worsening renal function, and stroke during hospitalization were independent predictors of in-hospital mortality in BCNIE patients ( Table 4 ).

Table 2. Complications during hospitalization.

IE complications	Frequency (%)
Acute pulmonary edema	7 (4.8)
Stroke during hospitalization	22 (15.1)
Worsening renal function	36 (24.7)
Mechanical ventilation a	15 (11.5)
New onset arrhythmias	9 (6.2)
Sepsis	14 (9.6)

Abbreviation: IE, infective endocarditis.

^aMechanical ventilation unrelated to perioperative surgical requirements.

Table 3. Bivariate analysis of the independent variable with in-hospital mortality.

	p -Value	OR	Confidence interval 95%
			Minimum	Maximum
Age  ≥60 y  <60 y	0.618	0.67	0.14	3.17
History of fever  Yes  No	0.144	0.45	0.16	1.31
Functional class  NYHA III and IV  NYHA I and II	0.007	5.75	1.62	20.42
History of hypertension  Yes  No	0.021	4.26	1.25	14.56
History of DM  Yes  No	0.448	0.44	0.05	3.60
History of previous stroke  Yes  No	0.911	1.13	0.13	10.19
History of coronary artery disease  Yes  No	0.317	0.35	0.043	2.763
History of previous IE  Yes  No	0.393	2.90	0.25	33.48
History of previous cardiac surgery  Yes  No	0.144	2.21	0.76	6.39
Previous antibiotics use  Yes  No	0.927	0.95	0.34	2.63
Arrhythmia  Yes  No	0.527	0.66	0.18	2.41
Leukocyte  >8,150/mm 3  <8,150/mm 3	0.351	1.85	0.51	6.70
Prosthetic valve  Yes  No	0.069	2.72	0.92	8.04
TAPSE <17 mm  Yes  No	0.647	1.29	0.43	3.86
Ejection fraction <50%  Yes  No	0.448	0.55	0.12	2.56
Vegetation >10 mm  Yes  No	0.398	1.48	0.59	3.67
Left-sided vegetation  Yes  No	0.167	4.28	0.55	33.60
Perivalvular complication  Yes  No	0.319	2.38	0.43	13.12
Stroke during hospitalization  Yes  No	0.000	7.78	2.78	21.76
Worsening renal function  Yes  No	0.000	8.11	3.03	21.69
Surgical intervention  No  Yes	0.028	4.14	1.16	14.73

Abbreviations: DM, diabetes mellitus; IE, infective endocarditis; NYHA, New York Heart Association; OR, odds ratio; TAPSE, tricuspid annular plane systolic excursion.

Table 4. Multivariate analysis of the independent variable with in-hospital mortality.

	p -Value	OR	Confidence interval 95%
			Minimum	Maximum
Heart failure NYHA III-IV	0.011	7.56	1.59	35.96
Worsening renal function	0.000	10.23	3.07	34.04
Stroke during hospitalization	0.001	8.92	2.38	33.46

Abbreviations: NYHA, New York Heart Association; OR, odds ratio.

Discussion

Analyses of factors related to mortality of infective endocarditis cases are crucial due to the high mortality in this condition. Even with numerous studies on factors related to mortality in this condition, caveats exist due to different patients characteristics between the centers of study. ^{2 3 9} Furthermore, factors that are reported to be related to worse outcomes in one study often are not statistically significant across other studies.

This study evaluates factors that are related to in-hospital mortality BCNIE patients at Harapan Kita National Cardiovascular Center, a tertiary hospital for heart and vascular disorders. Data collected from this institution represent the characteristics of IE patients with negative culture results in Indonesia. This study was also the first study in Indonesia which specifically observed BCNIE patients. A previous study exists by Sastroasmoro et al in 1989 which is limited to the pediatric patient population. ¹⁰ Several studies have assessed mortality predictor in IE patients; however, these studies do not differentiate between positive and negative culture patients. ^{3 9} A study by Siciliano et al specifically reported mortality predictor in IE negative-culture patients, albeit with relatively fewer samples (51 cases in 9 years [1991–1999]) compared with 146 cases in 5 years of this study. ²

This study differs from the other study in several aspects, namely risk factor definition population characteristics, sample size, and the novel inclusion of tricuspid annular plane systolic excursion as a mortality predictor. This new risk factor definition includes evaluation of acute kidney injury within the course of treatment, using criteria defined by the acute kidney injury network. ¹¹

The proportion of BCNIE in this study was 57%, with rates in developed countries ranging between 2.5 and 31%. ⁴ The proportion found in this study is similar across southeast Asia, including Thailand and Laos. These proportions are still relatively high compared with that of Brazil which ranges at 23%. ^{2 12 13} Mirabel et al at Laos reported a proportion of BCNIE patients at 61.1%, ¹² while Watt et al in Thailand reported an incidence of BCNIE at 69%. ^{13 14}

Further microbiologic studies using serology and polymerase chain reaction (PCR) have not been routinely implemented in Harapan Kita National Cardiovascular Center, which poses a challenge in investigating causal microorganism in BCNIE at our institution. The high proportion of patients with negative blood culture in developed countries is thought to be caused by preculture antibiotic administration. Watt et al reported that the amount of patients with antimicrobial administration prior to sample collection for culture reaches 71% in Thailand, compared with 48% in France. ¹³ In this study, the incidence of antimicrobial agent administration prior to culture can only be traced in 28% of the patients, and we postulated that the slow-growing or fastidious organism played a significant role in this aspect. The high incidence of preculture administration of antimicrobial agents in developing countries is suspected of having arisen from the lack of strict regulations regarding the administration of antimicrobial agents.

Culture-negative patients at our institution tend to be younger compared with other studies, and the mean age was found to be 38 years old compared with 52 years old in England. ¹² In this study, only 18 (12.3%) of the patients aged >60 years. These relatively younger patients can be explained by the higher incidence of congenital and rheumatic heart diseases in the developing countries. ^{12 13 15} With the similar proportion of congenital and rheumatic heart diseases in this study (17.8%), it is to be noted that valvular degeneration remains the most commonly encountered predisposing factors in this study (46.6%). This differs with several other studies in southeast Asia which stated that IE is most commonly found in rheumatic heart disease patients. ¹⁴

History of fever is seen in more than 80% of the patients in this study, a rate which is similar to culture-positive IE (86%). ² Patients with BCNIE experienced fever less frequently due to the presence of slow-growing causal organisms which lead to less inflammation. However, due to the lack of follow-up microbiologic studies in this study, this cannot be ascertained. Comorbidities such as coronary heart disease, hypertension, and diabetes are seen in only 10% of the patients, which are lower in developed countries (13–15.9%). ^{6 9} This difference can be caused by the fact that the population in this study were younger compared with studies done in developed countries. History of prior heart surgery and the use of prosthetic valve are only seen in 15% of this study population, compared with 40% in Spain in a study done by Zamorano et al. ⁶

Definite IE diagnosis in this study was established in 55.5% of the subjects, with patients fulfilling one major criterion and a minimum of three minor criteria based on Duke criteria. There are disparities between studies regarding the diagnosis in inclusion criteria in the study, with some studies including patients with possible IE diagnosis while others do not. Studies done by Wallace et al, Siciliano et al, and Watt et al only included patients with definite IE diagnosis. ^{2 9 13} While a study by Mouly et al included patients with possible IE. ¹⁶ In this study done at Harapan Kita National Cardiovascular Center, patients with possible IE are also included, due to the fact that these group of patients were commonly encountered and were given the same therapeutic regiments as patients with definite IE diagnoses.

In this study, vegetations are more commonly seen on the left side of the heart compared with the right side. Mitral valve is most commonly affected followed by the aortic valve. In a minority of patients, vegetations are seen on both sides of the heart. This result is in accordance with several other studies with a population of both culture positive and negative IE. ^{2 9 13 15} Majority of patients in this study showed an excellent LV function ranging between 80 and 85%. This finding is in accordance with another study which reported that the majority of patients with IE maintain a good ejection fraction. ² The decrease in ventricular function; however, can occur in the face of IE, or due to the preexisting heart condition that commonly acts as a risk factor for IE. There are no other studies regarding right ventricle function in relation to infective endocarditis to this date.

The combination of ceftriaxone and gentamycin is given in almost all culture-negative IE in this study (92.5% of cases), either as empirical treatment prior to culture results or as a definitive treatment. This actually differs from both European Society of Cardiology (ESC) and American College of Cardiology (ACC)/American Heart Association (AHA) guidelines which stated that there is no specific recommendation for antimicrobial agents in culture-negative IE. ^{3 5} Both ESC and ACC/AHA recommended consultation with infectious medicine specialist regarding the choice of antimicrobial agents to be administered to BCNIE cases and the use of serology and PCR to identify the causal organism. Implementation of specific antimicrobial approach necessitates the determination of the causal organism ( Table 2 ). ^{3 5} Several studies reported that the implementation of serology and PCR could increase the identification rate of microorganism in IE cases. ^{4 17 18}

Surgery in the active phase of endocarditis was done in 36% of the patients in this study. Surgeries were done at different time points during their course of treatment. A study of BCNIE in Brazil reported the proportion of patients that underwent surgery at the active phase of endocarditis to be 47%, ² while Zamorano et al reported a higher rate of surgery at 65%. ⁶ This rate of surgery varies according to the available resources at the facility. A study in Laos by Mirabel et al stated that none of their patients undergo surgery; however, the absence of patients undergoing surgery at the study is caused by the limited resource at the facility. ^{12 15} It is to be noted that studies in developed countries reported a much higher rate of surgery at 55%. ¹⁹

The mortality rate of culture-negative IE in this study was 13.5% (22 out of 162 patients), which is higher than the mortality rate in patients with culture-positive IE (10.6%). Several contemporary studies reported a varied mortality rate ranging from 5.1% at a study by Menu et al in France to 31.4% in a study by Siciliano et al in Brazil. ^{2 19 20} This disparity can be caused by different patient characteristics in these studies and the difference in health care delivery between facilities. Hospitals in developed countries are commonly supported by advanced laboratory facilities, while hospitals in underdeveloped countries are often not.

On bivariate analyses, factors that are related to in-hospital mortality were shown to be NYHA class III and IV heart failure, history of hypertension, stroke during hospitalization, decline in renal function, and surgery. On multivariate analysis, factors that are related to in-hospital mortality were shown to be NYHA III and IV heart failure, stroke during hospitalization, and decline in renal function. These findings were different compared with a study done by Siciliano et al which reported the factors related to mortality in BCNIE to be diabetes and severe sepsis. ²

In accordance with a guideline from ESC published in 2015 regarding diagnosis and treatment of infective endocarditis, there are several parameters that can be related to worse outcomes in patients such as: (1) patients' characteristics including age, prior heart surgery, and comorbidities; (2) Clinical complications of IE including functional class, decline in renal function, stroke, and septic shock; (3) Causal organism; and (4) echocardiographic findings such as impaired left ventricular function, pulmonary hypertension, periannular complications, dysfunction of prosthetic valve, and size of vegetation. ³ However, all of these parameters do not always correlate with mortality in IE patients due to different patient characteristics across institution and countries, as shown in a study by Siciliano et al. ² Another study in IE population in general also reported different predictors of mortality. Wallace et al reported that the factors related to mortality are leukocytosis, low serum albumin, disorders of heart rhythm, elevated serum creatinine, and presence of two Duke's major criteria, while according to Mouly et al, these factors ought to be nosocomial IE, heart failure, and prosthetic heart valves. ^{9 16}

Heart failure can be caused by the preexisting comorbidities which act as predisposing factors for IE or due to the endocarditis itself which can worsen preexisting heart failure or cause a new-onset heart failure due to regurgitation of native valves, prosthetic valve leakage, or perforation and obstruction of valves by vegetation. The finding of symptoms and signs of heart failure in IE patients can increase the mortality rate in these patients. ^{2 3}

Stroke in infective endocarditis cases can arise from vegetation embolus, mycotic aneurysm, or due to atrial fibrillation which causes a structural abnormality of the heart. The incidence of stroke in IE patients posed challenges in treatment strategies, especially regarding at which point should patients undergo surgery. While the surgery itself is needed to prevent embolus-related adverse events, candidates with a history of stroke generally have worse outcomes. The decline in renal function can arise from the embolism of vegetation material to renal vessels, administration of aminoglycosides, or due to heart failure (cardiorenal syndrome). The onset of renal impairment will pose further challenges in treatment regimens due to the calculations needed for drug administrations in patients with renal impairment.

The analyses of factors related to the mortality rate in hospitals allow specific approaches to be implemented in high-risk IE patients to prevent adverse events from occurring in these patients.

The authors would like to acknowledge the limitations of this study, the retrospective nature of the study, which implicated that the gathered data relied on the accuracy of medical record documentation, laboratory data, and hospital information system. The lack of advanced microbiologic assessments such as PCR and serology renders the determination of causal organism impossible.

Conclusion

The risk factors that were associated with in-hospital mortality include NYHA class III and IV heart failure, history of hypertension, stroke during hospitalization, the decline in renal function, and surgery. Upon multivariate analysis, variables that were independently associated with mortality are heart failure NYHA functional class III and IV, worsening kidney function, and stroke during hospitalization.