Clinical Practice of Surgical Treatment for Left-Sided Infective Endocarditis: Nationwide Data from the NIDUS Registry

Abstract

BACKGROUND:

Surgery is an essential treatment for selected patients with infective endocarditis (IE). Despite indications for surgery, not all patients undergo surgery. Most previous IE cohorts have examined a selected group of patients from primary tertiary centers. Thus, the aims of this study were to describe the use of surgery by indication and to assess mortality in a nationwide cohort of patients with left-sided IE.

METHODS:

This study was an observational cohort study including patients with left-sided IE from the National Danish Endocarditis Studies (2016-2021). Patients were categorized according to the presence of surgical indications defined as a Class I indication for surgery according to the 2015 European guidelines or a vegetation ≥10 mm on the diagnostic echocardiography and whether surgery was performed. One-year mortality was assessed with the 1-Kaplan-Meier estimator and multivariable Cox regression models.

RESULTS:

Among 3017 patients, 662 (21.9%; median age, 66.9 years) underwent surgery, 655 (21.6%; median age, 75.9 years) had surgical indications but received conservative treatment, and 1700 (56.5%; median age, 76.0 years) had no surgical indications. Patients who underwent surgery had fewer comorbidities and more streptococci, whereas conservatively treated patients had more comorbidities and Staphylococcus aureus. Patients with surgical indications who did not undergo surgery had the highest in-hospital (31.8% versus 12.5% versus 15.7%; P<0.001) and 1-year (50.5% versus 17.0% versus 33.5%, P<0.001) mortality compared with patients who underwent surgery and those without indications for surgery. Multivariable Cox models confirmed these findings. High surgical risk (22.8%) was the most reported reason for withholding surgery.

CONCLUSIONS:

In a nationwide consecutive cohort of patients with left-sided IE, 40% had indications for surgery according to guidelines, but only half of these underwent surgery. Surgically treated patients had the highest survival but were a decade younger, had fewer comorbidities, and had less frequent IE caused by S aureus. Thus, confounding by indication remains a limitation, and more studies are warranted to explore the management of patients at high surgical risk.

Clinical Perspective.

What Is New?

• This study assessed surgical treatment for left-sided infective endocarditis in a nationwide all-comer cohort of patients with infective endocarditis.

• Two of five patients with left-sided infective endocarditis had a guideline indication for surgery, whereas only half underwent surgical treatment. The most prevalent known reason for withdrawing surgical treatment was too high surgical risk.

• Patients with surgical indications who did not undergo surgery were older, had a higher comorbidity burden and more Staphylococcus aureus infective endocarditis, and were associated with a higher 1-year mortality.

What Are the Clinical Implications?

• Clinicians should use refined risk‐stratification tools—integrating comorbidity indices, S aureus cause, and echocardiographic findings—to identify candidates who might benefit from intervention despite perceived surgical risk.

• More studies are warranted to assess high-surgical-risk patients and the efficacy of surgical treatment for these patients.

Infective endocarditis (IE) is a critical illness that may necessitate surgical intervention for some patients. However, there is considerable variation in the use of surgery, with rates ranging from 20% to 50%.^1–4 Observational studies have indicated that, according to current guidelines, up to 70% of patients may be eligible for surgery.^3,5 The current 2023 European Society of Cardiology (ESC) guidelines for the management of IE⁶ and expert opinions suggest that surgery is underused, with observational studies showing reduced mortality in surgically treated patients compared with those receiving only medical therapy.^7–11 Furthermore, the 2023 ESC guidelines¹² advocate for more frequent use of surgery compared with the 2015 guidelines,⁶ lowering the threshold for vegetation size from 15 to 30 mm to ≥10 mm as a criterion for surgery. The 2020 American College of Cardiology/American Heart Association guidelines for IE consider vegetation >10 mm as an indication for early surgical treatment.¹³ The evidence supporting the indications for surgical treatment is derived primarily from retrospective case series conducted at tertiary centers. In these series, patients are generally significantly younger and exhibit different characteristics from those in unselected population-based cohorts.^{2,3,5,14–19} Thus, extrapolating findings from case series prone to referral bias may lead to misleading conclusions. A more aggressive surgical approach must be carefully balanced against the associated surgical risks.

The NIDUS Registry (National Danish Endocarditis Studies) offers a unique opportunity to examine the full spectrum of patients with left-sided IE. This study was based on the hypothesis that a significant proportion of patients with surgical indications did not undergo surgery because of a high surgical risk and that these patients experienced higher mortality rates. Hence, the primary aims of this study were to evaluate surgical treatment patterns in patients with left-sided IE based on established surgical indications and to evaluate mortality rates in relation to whether surgery was performed.

Methods

Study Design

This study was an observational cohort study using data from the NIDUS Registry, which included all possible and definite cases of IE in Denmark between 2016 and 2021. Trained health care professionals gathered data in the NIDUS Registry by reviewing electronic medical records of hospitalized patients with a diagnosis code of IE (International Classification of Diseases, 10th Revision codes DI33x, DI38x, and DI398) across Denmark. Patients residing in Greenland or the Faroe Islands were excluded because of limited access to their medical records, which made it impossible for us to follow up the patients. All patients were assessed with the ESC 2015 diagnostic criteria for IE,¹² and rejected IE cases were not included in the NIDUS Registry. Details on baseline characteristics, echocardiographic findings, microbiology, treatment during admission (including indications for surgery and surgical procedures), infection focus, discharge information, and outcomes are available in the NIDUS Registry and have been comprehensively described previously.²⁰

Study Population

The study population in the NIDUS Registry comprised 3557 patients with first-time possible or definite IE. This study included all Danish patients with the first left-sided IE episode in the study period, including native valve endocarditis and prosthetic valve endocarditis (PVE). Patients could have prior IE episodes before the study period. Non-Danish citizens, individuals lacking a recorded diagnosis date, and patients with IE related to cardiac implantable electronic devices alone were excluded. Indications for surgery were defined as a Class I indication according to the 2015 ESC guidelines for IE, including heart failure (cardiogenic shock or pulmonary edema), uncontrolled infection (abscess, fistula, resistant bacteria, fungi, or enlarging vegetations after onset of appropriate antibiotic treatment), and persistent vegetations >10 mm after ≥1 embolic events despite appropriate antibiotic therapy.¹² Vegetations ≥10 mm without evidence of embolism were also defined as an indication for surgery. PVE was not considered an isolated surgical indication per se. Subsequently, patients were grouped into 3 categories: (1) patients with an indication for surgery who underwent surgery, (2) patients with an indication for surgery who did not undergo surgery, and (3) patients without an indication for surgery. Class I indication and vegetation size were evaluated at the time of the diagnostic echocardiography. The flowchart of the study is presented in Figure 1.

Figure 1.

Flow diagram of the study population. CIED indicates cardiac implantable electronic device; IE, infective endocarditis; and NIDUS, National Danish Endocarditis Studies.

Clinical Presentation (Symptoms, Microbiology, and Imaging Findings)

Patient characteristics and clinical presentations were compared across the study groups. Infections were classified into 3 categories: (1) community-acquired infections; (2) nosocomial healthcare–associated infections, defined as infections occurring after any hospital admission within 3 months of the first symptoms of IE; and (3) nonnosocomial health care–associated infections, defined as infections occurring after any outpatient visits or medical procedures within 3 months of the first symptoms. Symptoms included the following: fever (temperature >38° C), muscle pain located in proximal limb musculature, back, or neck; dyspnea; and unintentional weight loss of >5 kg. Clinical signs of immunological and vascular phenomena, including splinters, petechiae, Osler nodes, Janeway lesions, and Roth spots, were recorded if they were described in the medical records. Complications related to IE evident on admission were also documented, including sepsis, systemic embolism, severe valve regurgitation, or heart block, all based on the description of them in the medical records or the conclusions of imaging modalities such as positron emission tomography/computed tomography. The most likely microbiological cause was grouped according to the type of bacteria Staphylococcus aureus, Streptococcus spp, Enterococcus spp, coagulase-negative staphylococci, and others (including Haemophilus spp, Aggregatibacter actinomycetemcomitans, Cardiobacterium hominis, Eikenella corrodens, and Kingella kingae [HACEK], as well as fungal infections). Significant echocardiographic findings were documented from the echocardiography reports. This included localization of vegetation, vegetation size, valve insufficiency of the affected valve/valves, signs of other valve diseases, and left ventricular ejection fraction. Positron emission tomography/computed tomography was deemed diagnostic if the echocardiography was inconclusive and the reported results of the positron emission tomography/computed tomography verified the suspicion of IE (eg, enhanced fluorodeoxyglucose uptake in PVE).

Indications for Surgery and Treatment for IE

Surgical treatment during admission was defined as the repair or replacement of the affected valve(s) or surgery on intrathoracic arteries. Isolated cardiac implantable electronic device extractions were not defined as surgical treatment and thus were excluded. Timing of surgery was defined as the number of days from the date of diagnosis until surgery and categorized as follows: (1) nonurgent if the patient was stable, surgery was performed after >5 days, and surgery could be deferred without increased risk of cardiac outcomes; (2) urgent if the surgery was performed within 2 to 5 days; and (3) emergency if surgery was performed within 24 hours because of acute severe valve regurgitation or obstruction causing refractory pulmonary edema or cardiogenic shock. If surgery was not performed despite indications, the reasons for refraining from surgical treatment were documented as follows: patient refusal; patient death before the onset of surgical treatment; hemodynamic instability caused by sepsis or heart failure; neurological complications (stroke or intracranial hemorrhage); high surgical risk, including patient frailty and poor prognosis before surgery; good prognosis without surgery; and other. If patients had vegetation ≥10 mm and no stated reasons for not undergoing surgery, they were categorized as no clinical indications/unknown.

Outcomes and Follow-Up

We assessed in-hospital and 1-year all-cause mortality. The start of follow-up (index) depended on treatment choice. Those who underwent surgery were followed up from the surgery date, and the nonsurgically treated patients were followed up from the date of IE diagnosis to avoid patients who underwent surgical treatment being “artificially” kept alive until the surgery date (immortal time bias). Follow-up continued until the earliest of the following: 1 year after the IE diagnosis, date of death, date of recurrent IE episode, or date of last follow-up (censoring).

Sensitivity Analyses

Patients were grouped according to class I indication and treatment choice to evaluate the use of surgery according to class I indications using the 2015 ESC guidelines. To account for the absence of surgical risk scores in the NIDUS Registry, we performed an additional time-dependent Cox regression analysis incorporating risk factors included in surgical risk scores, specifically left ventricular ejection fraction and chronic obstructive pulmonary disease, as categorical variables. Last, we assessed the crude proportion of relapse or new IE events by comparing patients with native valve endocarditis with those with PVE.

Statistics

Baseline characteristics are reported as frequencies and percentages for categorical variables and as medians with corresponding 25th to 75th percentiles for continuous variables. Missing values for clinically significant variables are reported but are not imputed. Baseline patient characteristics are presented in the order of the study groups. The 1-Kaplan-Meier estimator is used to evaluate 1-year mortality rates stratified by indications for surgery and treatment choice. In a sensitivity analysis, we assessed 1-year mortality stratified by study groups from the same index date (date of IE diagnosis). However, this Kaplan-Meier curve should be interpreted cautiously because of immortal time bias for patients who underwent surgery (the median time to surgery was 3 days). The log-rank test was performed to test the overall difference between the survival curves for each group. To account for immortal time bias in the adjusted survival analysis, we applied a time-dependent Cox regression model. Follow-up began at the time of IE diagnosis for all patients. Patients with an indication for surgery who eventually underwent surgery contributed follow-up time to the group of patients with indications for surgery who did not undergo surgical treatment until the date of their surgery, after which their follow-up time was attributed to the surgical group. Patients who died did not contribute more follow-up time. The multivariable Cox model was adjusted for the following confounders based on previous clinical knowledge from existing IE literature: age as a continuous variable, sex, PVE or native valve endocarditis, microbiological cause, history of heart failure, chronic kidney disease, diabetes, congenital heart disease, and prior IE episode. The proportional hazards assumption was tested and found to be valid using cumulative sums of Martingale residuals for all variables included in the final Cox model except for the patients who had indications for surgery and underwent surgery. Hence, we conducted a sensitivity analysis assessing the mortality rate within the first 45 days of IE diagnosis to account for the perioperative mortality and a landmark after 45 days, in which 602 patients were excluded who either died or were censored. Values of P<0.05 were considered statistically significant. Data management and statistical analyses were performed with SAS Enterprise 8.4 (SAS Institute, Inc, Cary, NC) and R software version 3 (R Foundation for Statistical Computing, Vienna, Austria).

Ethics

The study adheres to all national ethical guidelines for register-based research established by the Danish National Committee on Health Research Ethics. Informed consent was waived, and data acquisition was approved by the Danish Data Protection Agency (P-2020-92).

Results

Baseline Demographics

We included 3017 patients with left-sided IE. Indications for surgery were present in 1317 patients (43.7%), of whom 662 (50.3%) underwent surgery and 655 (49.7%) did not. The remaining 1700 patients (56.5%) had no indications for surgery and did not undergo surgery (Figure 1). The median time to surgery was 3 days, ranging from 0 to 191 days, and 90% had surgery performed within 21 days (Figure S1). The timing of surgery was reported as nonurgent in 40.0%, urgent in 40.5%, and emergency in 19.5% of cases. Patients who underwent surgery were hospitalized for a median of 45 days compared with 33 median days for those with an indication for surgery who did not undergo surgery and patients without indications for surgery. In addition, patients who underwent surgery received antibiotic treatment for IE for a median duration of 43 days compared with 32 days for those with a surgical indication who did not undergo surgery and 36 days for those without a surgical indication. Patients with surgical indications who did not undergo surgery were more often treated with lifelong suppressive antibiotic treatment (8.3%) compared with those with indications who underwent surgery (0.9%) and those without surgical indications (2.8%). The type of surgical intervention is presented in Table S1. Patients with indications for surgery who underwent surgical treatment were primarily younger male patients (median age, 66.9 years; 77.6% male). In comparison, patients with an indication for surgery who did not undergo surgery and patients with no indications for surgery were older (median age, 75.9 and 76.0 years, respectively) with a lower proportion of men (61.4% and 65.0%, respectively). In patients who did not undergo surgery, 75.1% of those with an indication for surgery and 61.9% without an indication for surgery were admitted to a tertiary hospital. (Table 1).

Table 1.

Baseline Characteristics Stratified by Indications for Surgery and Treatment Choice

Patient Characteristics, Lifestyle, and Functional Status

Patients with an indication for surgery who underwent surgery had a low burden of comorbidities, with diabetes as the most frequent (18.2%); 9.6% of patients had congenital heart disease, and most (92.4%) were self-reliant in activities of daily living. Patients with an indication for surgery who did not undergo surgery and those without indications for surgery had a notable burden of comorbidities, including diabetes (24.3% and 24.0%, respectively), chronic kidney disease (20.1% and 19.2%, respectively), and stroke (18.3% and 15.8%, respectively), before IE admission. They were less self-reliant in activities of daily living (61.2% and 65.8%, respectively; Table 1).

IE Characteristics, Clinical Presentations, and Microbiology

Community-acquired infections accounted for two-thirds across study groups. Fever on admission was the most predominant symptom across all study groups (range, 59.6%–67.1%), followed by dyspnea (range, 31.1%–37.7%) and myalgias (24.7%–32.0%). Of all patients with left-sided IE, 286 (9.5%) had an embolism to the central nervous system at admission. Among patients who underwent surgery, 19% presented with signs of embolism at admission, with the central nervous system being the most frequently affected (68.3%). In comparison, that rate was 15.4% (embolism to central nervous system, 73.5%) and 7.8% (embolism to central nervous system, 76.3%) for patients with an indication for surgery who did not undergo surgery and those without indications for surgery, respectively (Table 1). A detailed overview of clinical presentations is presented in Table S2. The frequency of microbiological causes stratified by indications for surgery and treatment choice is depicted in Figure 2. Streptococcus spp was the most frequent primary microbiological cause among patients who underwent surgery compared with S aureus for patients who did not undergo surgery, regardless of indications for surgery.

Figure 2.

Microbiological causes stratified by study groups. Bar chart displays the proportion of the primary microbiological causes stratified by study groups. BCN indicates blood culture-negative; and CoNS, coagulase-negative staphylococci.

Diagnostics, Indications for Surgery, and Reasons for Withholding From Surgery Despite Indications for Surgery

Among patients with an indication for surgery who underwent surgery, 47.4% had a class I indication for surgery compared with 28.5% in those with an indication for surgery who did not undergo surgery. Furthermore, other indications for surgery accounted for 17.2% of patients who underwent surgery, in whom the most frequent indications were severe or worsening valve regurgitation (68.4%) and secondary prevention for embolism (12.3%). Patients with indications for surgery who did not undergo surgery had more frequent mitral valve involvement and larger vegetations (Table 2 and Figure 3). Among patients with vegetations measuring 10 to15 mm (n=528), 33.9% underwent surgery. For those with vegetation >15 mm (n=362), 51.9% received surgical treatment. In the remaining patients with vegetation between 10 and 25 mm not further specified (n=76), 34.2% underwent surgery compared with patients treated conservatively. Other concomitant indications for surgery in patients with vegetation size as the primary indication included severe valve regurgitation (22.8%). In more than half of the cases (56.6%), the reasons for withholding surgery were unknown. The unknown reason for withholding surgical treatment was observed predominantly in patients with vegetation ≥10 mm as the primary indication for surgery (96.0% of the unknown group). Among the documented reasons, high surgical risk, frailty, and hemodynamic instability were the most common (Table 3).

Table 2.

Imaging Findings Stratified by Study Group

Figure 3.

Bar chart showing an overview of the indications for surgery. Bar chart displays the proportion of specific indications for surgery according to study groups 1 and 2. Veg. indicates vegetation.

Table 3.

Overview of Indications for Surgical Treatment and Reasons for Refraining From Surgery Despite Indications for Surgery

Mortality and Surgical Treatment

Patients with an indication for surgery who did not undergo surgery were associated with higher in-hospital mortality (31.3%) compared with those who underwent surgery (12.5%) and patients without indications for surgery (15.7%; P<0.001). Evaluating 1-year mortality, we followed up patients for a total of 2164.7 person-years with a median follow-up time of 365 days (89–365 days). During follow-up, 982 patients (32.6%) died. One-year mortality rates exceeded 50% for patients with indications for surgery who did not undergo surgery compared with 17% for those who underwent surgery. The mortality rate was highest during the first month across all study groups (Figure 4). In adjusted analysis, surgically treated patients were associated with mortality rates similar to those of patients without indications for surgery. In contrast, patients with indications for surgery who did not undergo surgery were associated with higher 1-year mortality rates (Table 4).

Figure 4.

Crude 1-year mortality (1-Kaplan-Meier estimator) stratified by indications for surgery and treatment choice. The 1-Kaplan-Meier (KM) estimator (est) for 1-year crude all-cause mortality rates stratified by study group. In the top right-hand corner, the 1-KM estimates with 95% CIs are presented at 1 year. Note that patients who underwent surgery were followed up from the date of surgery compared with the other 2 groups who were followed up from the date of infective endocarditis diagnosis.

Table 4.

Univariable and Multivariable Time-Dependent Cox Regression Model Assessing Association With 1-Year Mortality

Sensitivity Analyses

In Figure S2, the 1- Kaplan-Meier estimator using the same IE index (date of IE diagnosis) yielded similar results for all study groups. Assessing 45-day mortality yielded identical findings (Table S3). In the landmark analysis with an index 45 days after IE diagnosis, patients with an indication for surgery who underwent surgical treatment were associated with reduced mortality rates compared with those without an indication for surgery (Table S4). We also assessed patients with class I indication for surgical treatment, and patients with class I indication and no surgery had the highest crude 1-year mortality rates, followed by patients without class I indication and no surgery (Figure S3). Similar results were observed in univariable and multivariable Cox models (Table S5). An extended Cox regression model including additional surgical risk factors (left ventricular ejection fraction and chronic obstructive pulmonary disease) yielded similar results (Table S6). Among patients with native valve endocarditis (n=2259, 74.9%), the crude proportions of relapse and new IE events were 1.3% and 3.5%, respectively, compared with 3.3% and 7.1% among patients with PVE (n=758, 25.1%).

Discussion

This is the first study to describe surgical treatment for left-sided IE in an all-comer IE cohort with comprehensive clinical data. The significant findings were that half of the patients with an indication for surgery underwent surgery and that medical therapy only for this group was associated with significantly higher in-hospital and one-year mortality rates. However, patients with indications for surgery who did not undergo surgery were a decade older with a higher burden of comorbidities. In addition, S aureus was the most prevalent primary microbiological cause in patients with an indication for surgery who did not undergo surgical treatment.

The surgical rate for left-sided IE in this nationwide study was lower than that of other international IE cohorts, in whom surgical treatment has been described to be as high as 50%,^3,5,17,18,21 but is consistent with findings from other population-based studies.^1,2,16 In addition, in previous IE cohort studies, indications for surgery were found in 70% to 75% of patients, based primarily on class I indications,^3,5,21,22 significantly higher than in our study. In addition, half of those who had indications of surgery underwent surgical treatment, in alignment with previous studies^3,22 but lower compared with other IE series in which 3 of 4 patients with indications for surgery underwent surgical treatment.^5,23–25 In previous IE cohorts, patients who underwent surgery were younger but comparable in the prevalence of IE-associated risk factors. Thus, the differences in this findings of this study may partly be explained by referral bias with the primary inclusion of younger patients with complicated left-sided IE from tertiary centers in previous IE cohorts.²⁶ Moreover, different definitions of surgical indications have been used, and severe valve regurgitation or early prosthetic valve IE was considered a surgical indication in some of the studies.^5,22 This study considered large vegetations (≥10 mm) as a surgical indication, which contributed to a lower proportion of patients receiving surgical treatment among those who had surgical indications because more than two-thirds of those who did not undergo surgery had vegetation size as their indication. Hence, this could also contribute to the higher proportion of patients with surgical indications in the other IE cohorts. Similar results in terms of baseline demographics and patient characteristics in a previous population-based study support our findings.¹⁶

Patients who underwent surgical treatment had a higher proportion of Streptococcus spp.^5,8,16 In contrast, patients who did not undergo surgery had a higher proportion of S aureus, which agrees with previous findings.^7–10,27 Patients with S aureus IE are less likely to undergo surgery, likely because of its high virulence, which often leads to embolization, hemodynamic instability from septic shock, and ultimately multiorgan failure. Therefore, those patients will not be candidates for surgical treatment despite a class I indication for IE according to the current American College of Cardiology/American Heart Association guidelines for IE.¹³ S aureus IE is associated with high mortality rates.^{16,18,28–30} This aligns with findings from our study in which patients with S aureus IE had the highest proportion of in-hospital mortality (26.3%). Moreover, S aureus IE is strongly associated with health care–associated infections¹⁸ often affecting older patients with significant comorbidities, including those undergoing dialysis treatment.^31,32 Thus, patients have higher surgical risk, which could explain the lower rate of surgical treatment for S aureus IE and the higher in-hospital mortality rates. In contrast, health care–associated infections may be discovered early in the course of the disease, and initiating appropriate antibiotic treatment early may reduce the need for surgical treatment for some patients.Patients with indications for surgery who did not undergo surgery had more comorbidities and were less self-reliant in activities of daily living, indicating higher frailty. This likely contributed to an increased surgical risk, explaining the decision to withhold surgical treatment. Frail patients who underwent surgery for IE were associated with poor prognostic outcomes in a previous study.³³ Many surgical risk scores have been developed to enhance risk stratification of patients with IE and to predict postoperative mortality such as EndoSCORE (Early Mortality After Surgical Treatment of Heart Valve or Prosthesis Infective Endocarditis),³⁴ RISK-E (Risk Score for Cardiac Surgery in Active Left-Sided Infective Endocarditis),³⁵ and the Society of Thoracic Surgery IE score.³⁶ However, because of selection bias in the derivation cohort, the discriminative ability of these scores has not been optimal in external validation cohorts.^37–39

The reasons for withholding from surgical treatment differed from findings from the International Collaboration of Endocarditis cohort.⁵ In our study, the predominant known reason for withholding from surgical treatment was high surgical risk or poor prognosis regardless of treatment, in alignment with previous findings.^22,23 However, Chu et al⁵ reported stroke, hemodynamic instability, poor prognosis regardless of treatment, and death before surgical treatment as the most frequent reasons for not undergoing surgery despite indication. The higher proportion of patients with stroke and hemodynamic instability may result from a selected younger population with more complicated IE from tertiary centers compared with an all-comer cohort of patients with left-sided IE. Furthermore, the reasons for not undergoing surgery were not always documented in the medical records, leading to potential information bias in this study and possibly underestimation of the reasons for withholding surgical treatment.

As expected, our study found that patients with an indication for surgery who did not undergo surgery had significantly higher in-hospital and 1-year mortality compared with those who underwent surgery and those without an indication for surgery. These findings align with results from other prospective IE cohorts^3,5,21–25 and a recent meta-analysis.⁴⁰

Clinical Implications

This study presents data from an all-comer IE cohort, revealing that a substantial proportion of patients with a surgical indication were classified as at high surgical risk. These findings represent a more unselected characterization of patients with left-sided IE, which increases the generalizability of the findings to other all-comer IE series. Moreover, patients who did not undergo surgery experienced high mortality rates. Therefore, further research is needed to better characterize this patient group, to improve risk stratification, and to improve clinical outcomes, particularly among those with a high comorbidity burden and S aureus IE. Like other studies, our results suggest that surgery may confer better patient outcomes, but the confounding by indication should be included cautiously in the interpretation. Our results further suggest that clinicians may consider the risk of surgery higher than the potential benefit. Taken as a whole, prospective studies or randomized trials are warranted.

Strengths and Limitations

The strength of this study is that surgical practice patterns are followed in an all-comer IE population with comprehensive clinical data, including almost every case of definite or possible IE between 2016 and 2021. Data validation through review of electronic medical records ensured minimal missing values and complete follow-up. However, documentation gaps for left ventricular ejection fraction and vegetation size led to missing data for these variables. Furthermore, this study was subject to information bias, which may lead to an underestimation of the true prevalence of clinical parameters, including lifestyle factors, symptoms, and clinical findings, due to a lack of reporting in the medical records. In 56.6% of patients, the reasons for withholding surgical treatment were not clearly documented in the medical records and were therefore categorized as unknown. This represents a limitation of the study and may have led to an underestimation of other potential reasons for withholding surgical treatment. Furthermore, surgical risk scores such as Society of Thoracic Surgery IE,³⁶ EuroSCORE II,⁴¹ or other commonly used IE surgical risk scores were unavailable in the NIDUS Registry.^{34,35,42–44} The absence of surgical risk scores remains a challenge because we lack data on the estimated postoperative mortality risk, an important factor that can influence the decision to proceed with or withhold surgery. However, we included several key risk factors used in surgical risk scores such as the EuroSCORE II as covariates in an additional multivariable time-dependent Cox regression analysis, which produced comparable findings. Data on blood tests and serial blood cultures were not available in the NIDUS Registry, limiting our ability to evaluate early response to antibiotic therapy. However, the registry did include information on additional echocardiographic examinations that influenced patient management, including decisions about surgical treatment.

This study is limited by selection bias and confounding by indication inherent to its observational design. Although multivariable adjustments were applied, residual confounding may be an issue. Advanced causal inference methods were not conducted because of the limited comparability between the exposure groups. The retrospective observational study design limits the ability to establish causal relationships, allowing only the identification of associations and the generation of hypotheses. Evaluating the efficacy of surgery in IE requires further investigation, ideally through randomized controlled trials or advanced causal inference methods applied to prospective cohort studies.^45,46

Conclusions

In this nationwide study, we find that 40% of patients with left-sided IE had an indication for surgery but only half underwent surgical treatment. Patients who underwent surgical treatment during hospitalization were associated with the highest survival. However, these patients were a decade younger, had a lower burden of comorbidities, and less frequently had S aureus as the primary microbiological cause. These findings highlight the importance of balancing surgical treatment with operative risk to improve decision-making and treatment strategies for high-risk patients with indications for surgery, with the aim of optimizing their clinical outcomes. Prospective cohorts and randomized trials are essential to evaluate the efficacy of surgery and to guide optimal management in these high-risk patients.

Article Information

Acknowledgments

All medical students and research nurses engaged in the NIDUS Registry assisted with the review of medical records, data acquisition, and data entry into the online NIDUS Registry: Alin Iskandar, Andreas Creutzburg, Andreas Dalsgaard Jensen, Anna Parbøl Engelund, Anne Katrine Post Hansen, Annika Filtenborg-Barnkob, Asbjørn Karlebjerg Broholm, Camilla Torpegart, Caroline Ingeborg, Nolfi Frøstrup, Dogukan Jesper Gür, Jeppe Kofoed Petersen, Elisabeth Specht Linde, Frederik Kyhl, Hanne Rasmussen, Hicham Laghmoch, Ida Holm Jessen Knudsen, Isabella Christiane Hansen, Josefine Frieser Lange, Kathrine Jacobsen, Kirsten Thrysøe, Lucas Grove Vejlstrup Bager, Lukas Schak Nielsen, Louise Frandsen, Mathilde Nørlund, Natasja Schytte, Sara Aakre Faradonbeh, and Zaid Ayub.

Sources of Funding

The salary for Dr Graversen was covered by an independent research grant from the Novo Nordisk Foundation (number NNF20OC0060362). The authors are solely responsible for the design and conduct of this study, including all study analyses, the drafting and editing of the paper, and its final contents.

Disclosures

Dr Østergaard received an independent research grant from the Novo Nordisk Foundation for the study of mitral regurgitation. Dr Møller received institutional research grants from Abiomed and the Novo Nordic Foundation, received speakers fees from Abbott and Boehringer Ingelheim, and served on the Advisory Board for Boston Scientific. Dr Bruun received investigator-initiated grants from the Novo Nordisk Foundation, Augustinus Foundation, and Health Insurance Denmark, not related to this study. Dr Modrau received a research grant from the Novo Nordisk Foundation (5.9 Mill DKK) with no conflict of interest for this work. Dr Bundgaard reports lecture fees from Amgen, BMS, MSD, Pfizer, and Sanofi. Dr Køber reports speaker honorarium from AstraZeneca, Boehringer, Novartis, and Novo. Dr Fosbøl reports receiving from the Novo Nordisk Foundation and Danish Heart Association independent research grants related to valvular heart disease and endocarditis. The other authors report no conflicts.

Supplemental Material

Checklist

Tables S1–S6

Figures S1–S3