From structural heart lesions to host burden: changing epidemiology and outcomes of native valve infective endocarditis, a nationwide study, 2006–2023

Abstract

Background

The epidemiology of native valve infective endocarditis (IE) has shifted toward older adults with substantial comorbidity burdens, yet contemporary nationwide data on outcomes and surgical impact remain limited.

Methods

We conducted an 18-year nationwide cohort study of adults hospitalised with native valve IE in Korea (2006–2023). Outcomes included in-hospital and 5-year all-cause mortality, IE relapse and a composite of death or relapse. Temporal trends, mortality predictors and surgical associations across age strata were evaluated using multivariable Cox models and stratified survival analyses.

Results

Among 18 402 patients (mean age 63.7 years), incidence declined in individuals <45 years but increased in those ≥65 years. In-hospital mortality was 25.5%, and 5-year mortality exceeded 50% overall. Advanced age, dialysis dependence, cancer and major complications predicted mortality. Valve surgery, performed in 29.1% of patients, was consistently associated with lower short- and long-term mortality across age groups, with no evidence of age-by-treatment interaction. Both early (≤7 days) and late (>7 days) surgery showed reduced mortality versus medical therapy. IE relapse was more frequent in older adults, and surgery was associated with a lower relapse risk. In the composite outcome of death or relapse, older adults had a higher event burden, whereas surgery remained associated with fewer composite events.

Conclusions

Native valve IE in Korea has shifted toward an elderly, multimorbid population with persistently high mortality. Despite declining utilisation, the survival benefit of surgery was preserved across the age spectrum, supporting operative consideration in appropriately selected older adults.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Native valve infective endocarditis (IE) remains a high-mortality condition despite modern care. Its epidemiology has shifted from structural heart disease to disease occurring predominantly in older adults with multiple comorbidities. The survival benefit of surgery in elderly patients has been uncertain.

WHAT THIS STUDY ADDS

• In this 18-year nationwide cohort of 18 402 patients, incidence declined in younger adults but increased in those ≥65 years. Mortality remained high—especially in individuals ≥75 years—with dialysis, cancer and major in-hospital complications as key predictors. Surgical treatment was consistently associated with reduced short- and long-term mortality across all age groups, with benefits preserved even in elderly patients. Antibiotic exposure increased markedly over time despite stable incidence and mortality.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings emphasise the need for age-tailored and comorbidity-tailored management strategies, support surgical consideration irrespective of chronological age and inform health-system planning for an ageing population with IE.

Introduction

Infective endocarditis (IE) remains a life-threatening disease with substantial morbidity and mortality, despite advances in diagnostic and therapeutic strategies.¹,⁴ Short- and long-term mortality remains high, and surgery is often the only intervention associated with improved survival.⁵,⁷ At the same time, the epidemiology of IE has shifted markedly, driven by population ageing, increased survival among patients with complex cardiac disease and wider use of cardiovascular devices.⁸,¹² Whereas early studies primarily involved rheumatic or community-acquired infections, contemporary IE increasingly affects older adults with multiple comorbidities and healthcare-associated pathogens—particularly Staphylococcus aureus and Enterococcus.⁸,¹⁰ These changes underscore the need for updated, population-based evidence describing current patient characteristics and outcomes in the modern era. Accordingly, this study aimed to provide contemporary nationwide data on epidemiology, clinical features and both short- and long-term outcomes of native valve IE in Korea over 18 years.

Methods

Data source and study population

We conducted a retrospective cohort study using the Korean National Health Insurance Service (NHIS) claims database.^{11 12} Adult patients (≥18 years) hospitalised with a primary diagnosis of native valve IE between January 2006 and December 2023 were identified using International Classification of Diseases, 10th Revision (ICD-10) codes. To enhance diagnostic specificity, we restricted the cohort to admissions at general hospitals with more than 100 beds. Patients with a history of prosthetic valve replacement or transcatheter aortic valve replacement were excluded. Hospital admission records were linked to longitudinal claims data to assess comorbidities, treatments and outcomes.

Definitions

IE relapse was defined by the occurrence of IE after discharge from the index hospitalisation. Comorbidities were identified using ICD-10 codes in combination with procedure codes and material fee claims. Hypertension was defined as at least one claim with ICD-10 codes I10–I15 accompanied by a prescription for antihypertensive medication. Diabetes mellitus was defined as ICD-10 codes E10–E14 with at least one prescription for glucose-lowering medication. Dialysis status was determined as follows: maintenance dialysis was identified using procedure codes for haemodialysis or peritoneal dialysis. In contrast, new-onset dialysis was defined as the initiation of haemodialysis or continuous renal replacement therapy during the index hospitalisation in patients without prior maintenance dialysis. Cancer was defined as the presence of a registered cancer diagnosis (C00–C97) within the preceding year. Congenital heart disease was identified using ICD-10 codes Q20–Q26.

In-hospital events were also defined systematically. Cardiac arrest was identified using cardiopulmonary resuscitation procedure codes, shock was defined as the use of inotropes or a mechanical left ventricular support system and intracranial events were identified using ICD-10 codes I60–I64. Pyogenic spondylitis was determined using the ICD-10 code M46.

Outcomes

The primary outcomes were in-hospital all-cause mortality and long-term all-cause mortality up to 5 years after the index admission. Secondary outcomes included 30-day and 1-year all-cause mortality, the composite outcome of all-cause mortality and IE relapse and in-hospital complications (shock, intracranial events, new dialysis initiation). Additional analyses examined temporal trends in patient characteristics, antibiotic use as a surrogate for microbiology and the association between surgical intervention and survival across age groups.

Statistical analysis

Continuous variables are presented as mean±SD or median with IQR, and categorical variables as counts (percentages). Between-group comparisons were performed using Student’s t-test, χ² test or Mantel-Haenszel χ² test where stratified categorical comparisons (eg, relapse or composite events by surgical treatment) were applicable. Survival analyses were conducted using Kaplan-Meier estimates, with differences assessed by log-rank tests. Multivariable Cox proportional hazards models were used to evaluate predictors of mortality and to assess the association between surgical timing and long-term outcomes. Adjusted HRs with 95% CIs were reported after adjustment for age, sex and significant comorbidities.

To address confounding by indication for surgical treatment, we applied propensity score (PS) methods. PS were estimated using logistic regression models including age, sex, atrial fibrillation, prior stroke, cancer, dialysis and other clinically relevant covariates. One-to-one nearest-neighbour matching without replacement was performed, and covariate balance was assessed using standardised mean differences. We also attempted inverse probability of treatment weighting (IPTW) to estimate the average treatment effect in the overall cohort; however, due to poor overlap in covariate distributions and unstable extreme weights, IPTW results were not retained.

All statistical analyses were conducted using R V.4.0.3 (R Foundation for Statistical Computing, Vienna, Austria). A two-sided p value <0.05 was considered statistically significant.

Ethical statement

Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.

Results

Baseline characteristics and temporal trends

The cohort included 18 402 patients with native valve IE (figure 1). The mean age was 63.7±17.2 years, and 51.8% were male. Over the study period, incident cases declined among individuals <45 years but increased steadily in those ≥65 years (figure 2A). 90-day mortality decreased slightly in younger adults yet remained highest in those ≥75 years, with no meaningful improvement over time (figure 2B).

Figure 1. Flow chart of patient selection for native valve IE between 2006 and 2023.

Figure 2. Annual incidence of IE cases by age groups from 2006 to 2023 (A) and corresponding 90-day mortality rates (B). The incidence of IE remained relatively stable, whereas early mortality consistently increased with advancing age. IE, infective endocarditis.

Shift from structural heart disease to comorbidity-driven IE

Younger patients more often had congenital heart disease, whereas older adults demonstrated a markedly greater prevalence of hypertension, diabetes, atrial fibrillation, prior stroke and cancer (table 1), highlighting a transition towards comorbidity-driven IE.

Table 1. Baseline characteristics of the study population.

	Overall (N=18 402)	<65 years (n=8432)	≥65 years (n=9970)	P value
Age, years	63.7±17.2	48.3±12.1	76.7±7.3	<0.001
Male, n (%)	9534 (51.8)	5352 (63.5)	4182 (42.0)	<0.001
Comorbidities
Hypertension, n (%)	12 463 (67.7)	4032 (47.8)	8431 (84.6)	<0.001
Diabetes, n (%)	5262 (28.6)	1816 (21.5)	3446 (34.6)	<0.001
Atrial fibrillation, n (%)	3114 (16.9)	775 (9.2)	2339 (23.5)	<0.001
Stroke, n (%)	5041 (27.4)	1245 (14.8)	3796 (38.1)	<0.001
Dialysis, n (%)	1228 (6.7)	384 (6.9)	644 (6.5)	0.206
Cancer, n (%)	2423 (13.2)	931 (11.0)	1492 (15.0)	<0.001
Congenital heart disease, n (%)	535 (2.9)	410 (4.9)	125 (1.25)	<0.001
Antibiotics used
Cephalosporins, n (%)	15 165 (82.4)	7235 (85.8)	7930 (79.5)	<0.001
Penicillins, n (%)	10 629 (57.8)	4994 (59.2)	5635 (56.5)	<0.001
Anti-MRSAs, n (%)	9174 (49.9)	4549 (54.0)	4625 (46.4)	<0.001
Aminoglycosides, n (%)	8821 (47.9)	4778 (56.7)	4043 (40.6)	<0.001
Carbapenems, n (%)	4197 (22.8)	1566 (18.6)	2631 (26.4)	<0.001
Hospital stay, days	42 (20–62)	34 (25–59)	50 (17–67)	<0.001
ICU stay
ICU stay ≥2 days, n (%)	11 073 (60.2)	5709 (67.7)	5364 (53.8)	<0.001
ICU stay, days	22 (17–39)	17 (19–36)	26 (16–42)	0.004

Data are presented as means±SD, n (per cent) or medians (IQR).

ICU, intensive care unit; MRSA, methicillin-resistant Staphylococcus aureus.

In-hospital outcome

Overall in-hospital mortality was 25.5%. Shock, cardiac arrest and intracranial events occurred more often in older adults, who also underwent surgery less frequently (figure 3). New dialysis initiation and intracranial events were associated with higher 1-year mortality (online supplemental figure S1).

Figure 3. In-hospital adverse events among patients with infective endocarditis, stratified by age group. Older patients experienced significantly higher rates of cardiogenic shock, cardiac arrest and in-hospital death compared with younger patients. *p<0.05, **p<0.001.

Antibiotic use

Cephalosporins were the most commonly prescribed antibiotics (82.4%), followed by penicillins (57.8%), anti-methicillin-resistant S. aureus (MRSA) agents (49.9%), aminoglycosides (47.9%) and carbapenems (22.8%, figure 4). Older adults were more frequently prescribed carbapenems, whereas aminoglycosides were more commonly prescribed to younger patients. Across the 18-year study period, total antibiotic exposure increased markedly—including for broad-spectrum classes—despite relatively stable incidence and mortality, indicating a sustained escalation in antimicrobial intensity over time.

Figure 4. Temporal trends in antibiotic utilisation for infective endocarditis from 2006 to 2023. Cephalosporins were most frequently prescribed, followed by penicillins, anti-MRSA agents, aminoglycosides and carbapenems. Notably, the cumulative days of antibiotic therapy steadily increased over time, despite relatively stable incidence and mortality rates. MRSA, methicillin-resistant Staphylococcus aureus.

Outcomes: all-cause mortality and IE relapse

The median follow-up was 2.7 years (IQR 1.2–7.7). Survival at 30 days, 1 year and 5 years was 91.0%, 79.2% and 70.5% in patients <65 years, compared with 82.1%, 55.7% and 34.4% in those ≥65 years. Advanced age, dialysis dependence, cancer and major in-hospital complications were independent predictors of mortality, while valve surgery was strongly associated with improved survival (online supplemental figure S1).

Both early (≤7 days) and late (>7 days) surgery were associated with significantly lower mortality than medical therapy alone (all p<0.001, online supplemental figure S2), indicating that the survival benefit of surgery persisted regardless of timing.

IE relapse occurred more frequently in older adults, and surgery was associated with a lower relapse risk across age strata (online supplemental figure S3). Although relapse was uncommon, the age-related pattern paralleled that observed for mortality.

When death and relapse were combined as a composite outcome, older adults consistently exhibited a higher event burden. Surgical treatment was associated with significantly fewer composite events in both age groups (Mantel-Haenszel χ²=80.9, p<0.001, online supplemental figure S4), reinforcing the overall direction of the primary survival analyses.

Surgical benefit across age strata

Valve surgery was performed in 29.1% of patients, with utilisation declining markedly with advancing age (figure 5A). Nevertheless, surgery was consistently associated with improved 30-day and 1-year survival across all age groups, with no evidence of a surgery-by-age interaction in time-dependent Cox or PS-matched landmark analyses (figure 5B).

Figure 5. Kaplan-Meier survival curves stratified by age group and surgical treatment. (A–B) Survival rates were higher in surgically treated patients across age groups. (C) Restricted cubic spline curves showing the association between age and 1-year mortality risk, using age 55 years as the reference. Surgical (blue) and non-surgical (red) groups demonstrated parallel trajectories with increasing age. Dashed lines represent 95% CIs.

Restricted cubic splines showed a monotonic rise in mortality risk with age; however, surgical and non-surgical groups demonstrated parallel trajectories, indicating that older age increased absolute mortality but did not attenuate the relative survival benefit of surgery (figure 5C). Importantly, no excess hazard was observed among elderly surgical patients, suggesting that the benefit of surgery persisted even at advanced ages.

Discussion

In this 18-year nationwide cohort study of 18 402 native patients with IE, the incidence declined in the young but rose in the elderly with significant comorbidities. Mortality remained high, especially in those aged ≥75 years, with dialysis, cancer and in-hospital complications as key predictors. Surgical treatment was underused, particularly among elderly patients. Nevertheless, it consistently reduced both short- and long-term mortality across all age groups, with the survival benefit preserved even in the very old adults. The number of antibiotic exposure days increased over time despite stable overall incidence and mortality. These findings highlight the need for age-tailored and comorbidity-tailored management strategies, support surgical consideration irrespective of chronological age and inform health policy to optimise care for an ageing population with IE.

Comparison with international trends

The crude incidence of IE in Korea (1.8–2.2 per 100 000 person-years; age-standardised incidence rate ∼2.0 per 100 000 person-years) paralleled recent Japanese data but remained lower than in many Western countries.¹³,¹⁵ This discrepancy likely reflects differences in case ascertainment, inpatient-based identification and healthcare exposure patterns. Short-term mortality (25.5%) was higher than in Japan, Europe and the USA.^{910 14},¹⁷ Several factors may contribute: more conservative surgical utilisation in Korea compared with surgical-rich European registries; an older, multimorbid patient mix; and the inclusive nature of the NHIS system, which captures severely ill inpatients at general hospitals rather than referral-centre populations. Despite these differences, the age-stratified long-term survival curves in our cohort closely aligned with those from Japan, the European Infective Endocarditis registry and recent US studies (figure 6). Overall 5-year mortality exceeded 50%, rising to 66% in older adults, consistent with international data demonstrating that age, comorbidity and healthcare-associated infection increasingly shape long-term prognosis in contemporary IE.^{9 16 18 19}

Figure 6. Long-term survival after IE in Korea compared with international cohorts. Red lines indicate age-stratified survival in the present Korean nationwide cohort (<65 years, dashed; ≥65 years, solid). Black squares represent survival estimates from prior studies: Spain,⁷ EURO-ENDO,⁹ USA,¹³ Japan¹⁵ and the US surgical cohort.¹⁹ EURO-ENDO, European Infective Endocarditis; IE, infective endocarditis.

Our findings mirror global shifts: declining structural heart disease-related IE, rising elderly/comorbidity-driven cases, frequent S. aureus or Enterococci and longer antibiotic courses were not associated with improved survival.

Pathophysiological drivers in elderly IE

The disproportionate burden of IE in older adults likely reflects a convergence of biological and healthcare-related vulnerabilities.^{2 20 21} Frailty-associated immune dysregulation reduces physiological reserve and increases susceptibility to bacteraemia. At the same time, healthcare-associated exposures—such as recurrent hospitalisations, intravascular devices, haemodialysis access and long-term care residence—further heighten risks of MRSA and Enterococcus infections. These pathogens, predominant in contemporary elderly or multimorbid IE, are linked to severe presentations, metastatic complications and high early mortality. Reduced surgical accessibility among frail or medically complex older patients may also contribute to adverse outcomes, as surgery remains the only intervention consistently associated with improved survival.

Microbiology signals from antibiotic patterns

Although microbiological data were not directly available, antibiotic utilisation provided indirect insight into pathogen profiles. Increasing use of anti-MRSA agents and carbapenems—particularly among older adults—suggests a growing influence of resistant gram-positive and gram-negative organisms, aligning with prior registry and guideline data indicating that S. aureus and Enterococcus predominate in elderly or healthcare-associated IE.⁸,¹⁰¹⁵ These pathogen profiles may contribute to both the persistently high early mortality and the lack of improvement in long-term outcomes.^{22 23} In such settings, vancomycin is recommended as a core empiric agent because it provides broad coverage against MRSA, streptococci and most enterococci.^{20 24 25} Accordingly, in our claims-based data set—where direct microbiological data were unavailable—the use of vancomycin and other broad-spectrum agents served as a pragmatic operational proxy for underlying resistant gram-positive infections. This approach aligns with prior large cohort studies reporting similar pathogen distributions in elderly or healthcare-associated IE.

Surgical benefit and real-world decision-making

Surgical treatment reduced mortality across all age groups, although utilisation declined sharply with advancing age. Frailty and comorbidities—rather than chronological age alone—remain the primary barriers to operative management.^{19 26 27} Nevertheless, surgery was consistently associated with improved survival across age strata. In this study, restricted cubic spline analyses demonstrated parallel age-related risk trajectories between surgical and non-surgical patients, indicating that older age increased absolute mortality but did not attenuate the relative benefit of surgery. These findings reinforce that age alone should not preclude surgical consideration in appropriately selected patients.

Limitations of this study

This study has several limitations inherent to its administrative claims-based design. First, detailed surgical information was unavailable. Valve position, prosthesis type and procedural characteristics could not be distinguished. Surgical candidacy may also have been influenced by unmeasured factors such as frailty or patient preference, raising the possibility of residual confounding and selection bias that may overestimate the apparent survival benefit of surgery. Second, microbiological and antibiotic-related variables were constrained. Treatment duration relative to guideline recommendations and the use of combination versus monotherapy regimens could not be determined. In addition, direct microbiological data were lacking, and antibiotic utilisation served only as a surrogate for pathogen distribution, potentially leading to misclassification. Third, several clinically meaningful outcomes—such as readmission, reinfection, relapse severity, postdischarge functional status and patient-reported quality of life—were not captured in claims data. As a result, the broader impact of IE on long-term survivorship may be underestimated. Fourth, patients with prosthetic valve endocarditis (PVE) were intentionally excluded to maintain a pathophysiologically homogeneous native valve IE cohort. Because PVE differs substantially from native valve endocarditis in microbiology, structural complications and surgical decision-making, inclusion of both entities would obscure age-related patterns. However, this approach limits external validity, and a dedicated analysis of our PVE population is underway. Fifth, valve-specific involvement (aortic vs mitral) and sex-related differences could not be reliably assessed due to ICD-10 coding ambiguity and inconsistent reporting across institutions. Consequently, surgical utilisation and outcomes could not be stratified by valve location. Sixth, we were unable to determine the specific clinical or anatomical indications that prompted surgical intervention, as echocardiographic anatomy, haemodynamic severity, vegetation size, abscess formation and other structural complications were not available in administrative data. Accordingly, surgery performed for heart failure, uncontrolled infection or embolic prevention could not be differentiated. This limitation is universal to large population-based IE data sets and underscores that the surgical benefit observed here reflects real-world associations rather than causal effects. Despite these limitations, the nationwide scope, large sample size and comprehensive long-term follow-up provide valuable insight into contemporary IE epidemiology and outcomes in routine clinical practice.

Conclusions

In this nationwide 18-year cohort, native valve IE in Korea has shifted from structural heart disease–related cases toward an elderly, multimorbid population with persistently high mortality. Although surgical utilisation declined with advancing age, the survival benefit of surgery was preserved across the age spectrum, indicating that chronological age alone should not preclude operative consideration. These findings underscore the need for age-tailored and comorbidity-tailored management strategies, timely multidisciplinary evaluation and strengthened antimicrobial stewardship to improve outcomes in this increasingly vulnerable patient group.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.