Diagnosing endocarditis relies on a combination of blood cultures, heart imaging, and clinical judgment rather than any single definitive test. The standard framework, known as the Duke criteria, was substantially updated in 2023 to incorporate newer imaging and molecular tools, but the core logic remains the same: find the bug in the blood, find the damage on the heart, and piece together the clinical picture when neither alone is conclusive. Because the infection can mimic dozens of other conditions, missed or delayed diagnosis is common and significantly worsens outcomes.
The Duke Criteria Framework
Since the 1990s, the modified Duke criteria have served as the go-to checklist for classifying suspected endocarditis as “definite,” “possible,” or “rejected.” The system divides evidence into major criteria (positive blood cultures with typical organisms, imaging findings of valve involvement) and minor criteria (predisposing heart conditions, fever, vascular or immunologic phenomena, and supportive microbiologic findings). Meeting certain combinations of these criteria determines the classification.
In 2023, the Duke-International Society for Cardiovascular Infectious Diseases published a major update. The revised criteria now recognize newer microbiological diagnostics such as PCR, metagenomic sequencing, and enzyme immunoassays for organisms like Bartonella. Imaging criteria were expanded to include PET/CT with a radioactive sugar tracer and cardiac CT, not just echocardiography. Intraoperative inspection of the valve during surgery was added as a new major clinical criterion. The list of “typical” organisms that count toward a major criterion was broadened, and for patients with prosthetic valves, additional pathogens were added. Predisposing conditions now explicitly include transcatheter valve implants, certain cardiac devices, and a history of prior endocarditis.1PubMed Central. The 2023 Duke-International Society for Cardiovascular Infectious Diseases Criteria for Infective Endocarditis: Updating the Modified Duke Criteria One practical change worth noting: the old requirement for specific timing and separate venipuncture sites when drawing blood cultures was dropped, reflecting how modern microbiology labs actually operate.
Blood Cultures Remain the Starting Point
Blood cultures are still the single most important laboratory test when endocarditis is suspected.2PubMed Central. Laboratory Diagnosis of Infective Endocarditis The goal is to draw enough sets to catch the organism circulating in the bloodstream. Two large observational studies suggest that three sets of blood cultures recover the pathogen roughly 96 to 98 percent of the time, and a fourth set pushes that to nearly 100 percent. Each set should consist of two bottles filled with about 10 mL of blood apiece, and ideally all sets are drawn before antibiotics are started, because prior antibiotic use substantially reduces the chance of growing anything. Modern incubation techniques mean that a standard five-day culture period catches virtually all organisms that will grow.3JAMA Network Open. Guidelines for Diagnosis and Management of Infective Endocarditis in Adults: A WikiGuidelines Group Consensus Statement
That said, even with perfect technique, not every case produces positive cultures. When cultures come back negative, clinicians turn to serological tests for organisms like Coxiella burnetii (the cause of Q fever) and Bartonella species, which do not grow well under standard conditions.4PubMed Central. Laboratory Diagnosis of Infective Endocarditis
When Cultures Come Back Negative
Culture-negative endocarditis is not rare and can be genuinely frustrating for both the patient and the clinical team. The causes fall into a few broad categories: blood cultures that were sterilized by antibiotics given before the cultures were drawn; infections with fastidious organisms that need special growth conditions (such as nutritionally variant streptococci); and truly intracellular bugs like Coxiella burnetii, Bartonella species, and Tropheryma whipplei that simply cannot be grown with standard culture techniques.5PubMed Central. Updates in Culture-Negative Endocarditis6PubMed. Laboratory Approach to the Diagnosis of Culture-Negative Infective Endocarditis
The diagnostic approach to these cases has expanded considerably. Beyond serology, molecular techniques now play a growing role. Metagenomic next-generation sequencing, which reads all the DNA in a blood or tissue sample and matches fragments to known pathogens, has shown diagnostic yields in the range of 80 to over 90 percent in culture-negative cases, particularly when patients have already received antibiotics.7PubMed Central. Unveiling the Future of Infective Endocarditis Diagnosis: The Transformative Role of Metagenomic Next-Generation Sequencing in Culture-Negative Cases These tools are expensive and not universally available, but they represent a genuine shift in how stubborn cases get cracked. The 2023 Duke criteria formally recognize PCR, amplicon sequencing, metagenomic sequencing, and in situ hybridization as acceptable pathways to a microbiological diagnosis.8PubMed Central. The 2023 Duke-International Society for Cardiovascular Infectious Diseases Criteria for Infective Endocarditis: Updating the Modified Duke Criteria
Echocardiography and Its Limits
Echocardiography is the imaging workhorse for endocarditis. Clinicians are looking for vegetations (clumps of bacteria and blood clot material stuck to a valve), abscesses, new holes in the valve leaflets, or new regurgitation (blood leaking backward). The two main approaches are transthoracic echocardiography (TTE), where the ultrasound probe sits on the chest wall, and transesophageal echocardiography (TEE), where a specialized probe goes down the esophagus to get much closer to the heart.
TTE is usually done first because it is noninvasive, quick, and widely available. But its sensitivity is limited. A systematic review and meta-analysis found that TTE detected vegetations with a sensitivity of about 61 percent and a specificity of 94 percent, meaning it catches just over half of the vegetations that TEE would find.9Journal of the American Society of Echocardiography. Diagnostic Accuracy of Transthoracic Echocardiography for Infective Endocarditis: A Systematic Review and Meta-Analysis A separate meta-analysis looking specifically at native valve endocarditis reported a somewhat higher TTE sensitivity of about 71 percent but broader confidence intervals.10PubMed. Diagnostic accuracy of transthoracic echocardiography to identify native valve infective endocarditis: a systematic review and meta-analysis Either way, a negative TTE does not rule out the disease. If clinical suspicion is moderate to high and the TTE is negative or unclear, TEE is the next step.
TEE has a sensitivity exceeding 90 percent for most valve vegetations. It is especially valuable for prosthetic valve infections, where the artificial valve hardware creates acoustic shadowing that makes TTE even less reliable. The trade-off is that TEE is semi-invasive, requires sedation, and is less readily available in some settings.
CT, PET/CT, and Other Advanced Imaging
Cardiac CT has carved out a specific niche in endocarditis diagnosis, particularly for detecting complications around the valve such as abscesses and pseudoaneurysms. A systematic review comparing cardiac CT with TEE found CT sensitivity of about 78 percent for these perivalvular complications, rising to 87 percent when multiphase scanning protocols were used, compared with about 69 percent for TEE.11PubMed Central. Comparative Value of Cardiac CT and Transesophageal Echocardiography in Infective Endocarditis: A Systematic Review and Meta-Analysis When CT and echocardiography are used together, one study found they collectively identified all abscesses and pseudoaneurysms confirmed at surgery.12PubMed Central. The usefulness of cardiac CT in the diagnosis of perivalvular complications in patients with infective endocarditis CT also provides a roadmap for surgeons planning an operation and can reveal embolic complications in other organs.
PET/CT using a radioactive sugar tracer works differently. Instead of showing anatomy, it highlights areas of increased metabolic activity, which active infection produces. Its performance varies dramatically by clinical scenario. For prosthetic valve endocarditis, a meta-analysis reported sensitivity around 86 percent and specificity around 84 percent. For native valve endocarditis, however, sensitivity dropped to roughly 31 percent while specificity remained very high at about 98 percent.13PubMed. Diagnosis of Infective Endocarditis by Subtype Using (18)F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography: A Contemporary Meta-Analysis A separate meta-analysis reported somewhat higher overall pooled sensitivity for PET/CT at around 77 percent, again with better performance in prosthetic valve cases.14PubMed. Meta-analysis of 18F-FDG PET/CT in the diagnosis of infective endocarditis The takeaway is that PET/CT is most useful in prosthetic valve and cardiac device infections, where echocardiography often struggles, and less helpful for native valve disease.
Labeled white blood cell SPECT/CT is another nuclear imaging option, though less widely studied. In one series, it showed 90 percent sensitivity and 100 percent specificity, with no false positives even in patients evaluated within two months of surgery.15Journal of Nuclear Medicine. Added Value of 99mTc-HMPAO–Labeled Leukocyte SPECT/CT in the Characterization and Management of Patients with Infectious Endocarditis European guidelines recommend considering it when other tests are inconclusive, particularly for prosthetic valves and vascular grafts.16Journal of Nuclear Medicine. Labeled White Blood Cell SPECT-CT in the Diagnostic Workup of patients with suspected Infective Endocarditis in the Real-World
Brain MRI and Silent Embolization
One of the more striking findings in endocarditis research over the past two decades is how often infection throws tiny clots to the brain without causing obvious symptoms. A pilot study using MRI found subclinical brain embolization in a substantial number of patients with left-sided endocarditis, suggesting the true rate of brain involvement is considerably higher than what clinical exams and CT scans detect.17PubMed. Subclinical brain embolization in left-sided infective endocarditis: results from the evaluation by MRI of the brains of patients with left-sided intracardiac solid masses (EMBOLISM) pilot study
Brain MRI does more than just detect complications. It can actually help clinch the diagnosis. One large study found that cerebral MRI performed within a week of admission changed the diagnostic classification in about a third of cases. When tiny microbleeds were included, the reclassification rate climbed to about half of patients who initially did not have a definite diagnosis.18PubMed Central. Value of brain MRI in infective endocarditis: a narrative literature review The pattern of silent ischemic lesions and microbleeds on MRI may eventually serve as a distinct diagnostic marker, though additional studies are needed before it can be formally adopted as a minor criterion in the Duke framework.
Can Blood Markers Help with Triage?
Clinicians have long searched for a blood test that could quickly flag endocarditis the way troponin flags a heart attack. Procalcitonin, a marker that rises during bacterial infections, has received the most attention. One study found that procalcitonin levels were significantly higher in patients with endocarditis compared to those with other diagnoses, and at a cutoff of 2.3 ng/mL it showed about 81 percent sensitivity and 85 percent specificity.19PubMed. Procalcitonin and the early diagnosis of infective endocarditis
Those numbers sound promising in isolation, but a meta-analysis pooling multiple studies painted a less rosy picture, reporting lower pooled sensitivity of about 64 percent and specificity of 73 percent for procalcitonin. Interestingly, C-reactive protein, a much older and cheaper test, performed at least as well across the pooled data. The meta-analysis concluded that neither marker is reliable enough to rule endocarditis in or out on its own.20PubMed. Role of procalcitonin in the diagnosis of infective endocarditis: a meta-analysis Where procalcitonin may be more useful is in predicting which organism is responsible: one study found that patients with elevated procalcitonin, white blood cell count, and CRP above specific thresholds had almost a 20-fold higher risk of Staphylococcus aureus infection, which matters because S. aureus endocarditis is more aggressive and often requires urgent surgery.21PubMed. Procalcitonin, white blood cell count and C-reactive protein as predictors of S. aureus infection and mortality in infective endocarditis
Why the Diagnosis Gets Missed
Endocarditis is one of those conditions that textbooks describe with a classic presentation (fever, new heart murmur, embolic events) but that real patients often fail to follow. One retrospective study from a teaching hospital found that a wrong initial diagnosis was made in over half of episodes. Patients with a discrepant diagnosis on admission had more than twice the odds of a poor outcome compared with those who were diagnosed correctly from the start.22PubMed Central. Errors in diagnosis of infective endocarditis
Several factors independently predicted delayed diagnosis in another study: use of antibiotics before the diagnosis was considered (which suppresses fever and sterilizes cultures), absence of high fever, and relatively low inflammatory markers. In other words, the patients who look the least sick on paper are paradoxically the ones most likely to be missed.23PubMed Central. Factors associated with delayed diagnosis of infective endocarditis: A retrospective cohort study in a teaching hospital in Japan The classic physical exam findings people associate with endocarditis, like Osler nodes (painful bumps on the fingers), Janeway lesions (painless spots on the palms or soles), splinter hemorrhages under the nails, and Roth spots in the eyes, are well described but actually uncommon. Case reports where a patient presents with all four at once are notable enough to be published specifically because they are rare.24PubMed Central. Classic Peripheral Signs of Subacute Bacterial Endocarditis
Cardiac Device and Right-Sided Infections
Pacemakers, implantable defibrillators, and other cardiac electronic devices create a unique diagnostic challenge. The leads that run through veins into the heart can serve as a surface for bacteria to colonize. Transesophageal echocardiography, normally the gold standard, misses some of these lead-associated masses. A study comparing TEE with intracardiac echocardiography (a catheter-based ultrasound placed directly inside the heart) found that TEE detected intracardiac masses in 73 percent of patients with device-related infection, while intracardiac echocardiography caught them in all patients.25PubMed. Usefulness of intracardiac echocardiography for the diagnosis of cardiovascular implantable electronic device-related endocarditis Diagnosis of device-related endocarditis tends to take a long time across all centers, and the prognosis is sobering, with in-hospital mortality around 18 percent in a large European registry.26European Heart Journal Open. Cardiac device-related infective endocarditis need for lead extraction whatever the device according to the ESC EORP EURO-ENDO registry
Right-sided endocarditis, which primarily affects the tricuspid valve, accounts for roughly 5 to 10 percent of all endocarditis cases but is far more common among people who inject drugs, where up to 40 percent of cases involve the right side. Staphylococcus aureus is the causative organism in 60 to 70 percent of right-sided cases in this population. TTE actually performs reasonably well for tricuspid vegetations, with sensitivity around 70 to 75 percent and specificity about 90 percent, and TEE pushes sensitivity above 90 percent.27PubMed Central. Injection drug use and right sided endocarditis
Tissue Diagnosis and Intraoperative Findings
When a patient goes to surgery for valve replacement or repair, the removed tissue offers a chance for definitive diagnosis. Histopathology can reveal organisms, inflammatory infiltrates, and vegetations directly. One study of mechanical heart valves found that a specific measure of neutrophil density in the tissue, at a particular cutoff, was 90 percent sensitive and 98 percent specific for endocarditis, helping distinguish true infection from the noninfectious fibrosis and inflammation that can also accumulate on prosthetic valves.28PubMed. Quantitative histological examination of mechanical heart valves The 2023 Duke criteria formally recognize intraoperative inspection as a major criterion, acknowledging what surgeons have known for years: sometimes you need to look at the valve directly to know what is going on.
Nonbacterial Thrombotic Endocarditis
Not every vegetation on a heart valve is caused by infection. Nonbacterial thrombotic endocarditis involves sterile clumps of fibrin and platelets forming on valve leaflets, typically in the setting of advanced cancer, autoimmune disease, or other conditions that promote abnormal clotting. It can look nearly identical to infective endocarditis on echocardiography, and it carries its own serious risks, including stroke and systemic embolization. The distinction matters enormously because the treatment is completely different: antibiotics and surgery for infective endocarditis versus anticoagulation and treatment of the underlying condition for the nonbacterial form. Diagnosis requires excluding infection through negative blood cultures and clinical context, then confirming the presence of vegetations on imaging.29Cardiology in Review. Nonbacterial Thrombotic Endocarditis: Pathogenesis, Diagnosis, and Management
Pediatric Considerations
Most of the diagnostic evidence for endocarditis comes from adult populations and has been extended to children without much pediatric-specific validation. This is a real gap. Children with congenital heart disease, especially those who have undergone surgical repairs, are at particular risk but may present differently from adults, with vaguer symptoms and less reliable classic findings. A 2022 literature review noted that while diagnosis has improved with multimodal imaging, the evidence base for children remains thin.30PubMed Central. Pediatric Infective Endocarditis: A Literature Review Pediatric cardiologists generally follow the same Duke framework but lean more heavily on echocardiography and clinical judgment given the limited data on newer tools in younger patients.
Artificial Intelligence for Vegetation Detection
Interpreting echocardiography for endocarditis is a skill that takes years to develop, and even experienced readers can disagree. Several research groups are training AI models to spot vegetations on echocardiographic images automatically. One study using TEE images tested two common deep learning architectures and found the better-performing model achieved an area under the curve of 0.91 for identifying patients with vegetations, with a true positive rate of 83 percent.31PubMed Central. Automatic Detection of Vegetations With Transesophageal Echocardiography in Infective Endocarditis Using Artificial Intelligence Another study using a different model architecture reached an area under the curve of about 0.86.32PubMed Central. Detection of Valve Vegetations in Native and Prosthetic Valves using Echocardiographic Radiomics and Deep Learning on Transesophageal Echocardiography Images These tools are not yet ready for independent clinical use, but they could eventually help less experienced sonographers flag suspicious findings and reduce the time to diagnosis, particularly in hospitals without round-the-clock access to expert echocardiographers.