Infective endocarditis (IE) refers to the inflammation of the heart valve or the inner lining of the ventricular wall caused by the invasion of bacteria, fungi or other microorganisms. It is a common inflammatory disease worldwide, with a global incidence of 1.8/100,000–9.7/100,000 (1). As the aging trend of the global population increases, the number of patients with degenerative valvular disease is increasing, and IE is the most serious and devastating complication of heart valve disease, with a high clinical mortality rate. The 2016 American Association for Thoracic Surgery (AATS) consensus reported that even with appropriate antimicrobial therapy and surgical interventions in current medical conditions, the in-hospital mortality rate of IE is still as high as 15–20% (2). The prognosis of IE is influenced by four main factors: the clinical characteristics of the patient, the presence of cardiac and non-cardiac complications, the type of microorganisms infected and the echocardiographic findings (3). Applying early clinical characteristics and laboratory testing to identify high-risk patients with IE for targeted treatment is an effective strategy to reduce mortality during hospitalization in patients with IE and is a significant problem for doctors.

In recent years, statistical prediction models have been widely used in the prognostic studies of clinical diseases. One of the prognostic prediction models is the nomogram. Its working method is to build a statistical prediction model to generate numerical probabilities of clinical events, then use the nomogram to display the outcomes of the statistical prediction model in simple graphics, and finally to make clinical decisions based on them. Nomogram is frequently utilized in the prognostic research of tumor patients due to its simplicity and convenience of usage. The effectiveness of the nomogram to predict prognosis in many tumors is much better than that of traditional TNM staging (4). The Nomogram prediction model has been shown in a number of studies to accurately quantify and evaluate the probability of clinical events in patients, with the predicted results matching the real world (5–7).

We hope that by constructing a Nomogram prediction model based on patients' early clinical data, we will be able to use the probabilities generated by the model predictions to risk-stratify patients with IE in the clinic, allowing clinicians to identify patients with a poor prognosis earlier and take more aggressive and targeted treatment measures to reduce mortality during hospitalization in patients with IE.

IE is an uncommon infectious disease in clinical practice, but with the development of society and medical advances, the number of patients suffering from valve disease, prosthetic valve implantation, cardiac devices, hemodialysis, and receiving immunosuppressive therapy is increasing, eventually leading to a yearly increase in the incidence of IE, which is now the fourth most common life-threatening infectious syndrome after sepsis, pneumonia, and intra-abdominal abscess (2, 11). The incidence of IE continues to increase and the mortality rate of IE has also remained at a high level (14–22% in hospital and 40% at 1 year after discharge) (12, 13). In addition to the clinical characteristics of the patient and the presence of underlying cardiac disease, studies have shown that the prognosis of IE is also influenced by the type of causative microorganism, the location of the cardiac vegetation, and the type of valve involved (native valve or prosthetic valve) (9, 14). It was found that among the pathogenic bacteria in IE, Staphylococcus aureus, Streptococcus griseus and Enterococcus are the most frequently seen in IE, accounting for about 70% of all (15). The mitral and aortic valves were reported to be the most commonly involved in IE patients with valve involvement (12). When compared to native valve endocarditis (NVE), prosthetic valve endocarditis (PVE) has a higher clinical death rate. PVE is also an independent risk factor for the prognosis of IE (14). The key to reducing the clinical mortality of IE patients is to immediately identify high-risk IE patients through comprehensive examination of early clinical characteristics, and then to use mature anti-infective medication therapy and suitable surgical intervention. To achieve this goal, physicians urgently need an adequate statistical prediction model to quantify the likelihood of clinical death in IE patients.

Anti-infective medication therapy is the foundation of the treatment of IE in clinical practice, and mature drug treatment programs can enhance the effectiveness of anti-infective therapy (16). According to the traditional approach, the strategy of using IE anti-infective medications is to start treatment early, with a large dose, and for a long period of time, with anti-infective treatments normally taking 4–6 weeks to work. Surgical treatment for IE patients began in the early 1960s and has since evolved into a standard treatment for IE (17). The proportion of patients treated surgically in our study was 74.1%. In today's IE treatment, the duration of anti-infective medicine and whether or not to treat surgically have become key variables influencing IE patients' in-hospital mortality (18). Therefore, in this study, the duration of anti-infective medication use during the IE hospitalization and the adoption of surgical treatment were also included together as two key variables in the construction of the statistical prediction model.

We analyzed the data from the cured and dead groups to see whether there were any risk factors that could have influenced the study's outcomes. After using univariate and multivariate logistic regression, we discovered that age (OR = 2.28, 95%CI 1.01–5.15, P = 0.048), abnormal leukocyte count (OR = 2.46, 95%CI 1.12–5.36, P = 0.024), left-sided IE (OR = 2.92, 95%CI 1.20–7.13, P = 0.019), right-sided IE (OR = 6.58, 95%CI 1.63–26.5, P = 0.008) and without surgery (OR = 18.94, 95% CI 8.18–43.86, P < 0.001) were independent risk factors for mortality during hospitalization in patients with IE. We created a statistical prediction model based on the five variables listed above and displayed it as a nomogram (Figure 2).

In order for the model to be used in the clinic, the accuracy of the model prediction is our primary concern. Therefore, after the nomogram prediction model was established, we evaluated the diagnostic performance of the model. The evaluated C-index of the model was found to be 0.878 (95% CI: 0.824–0.931). Internal validation of the prediction model using bootstrap validation method revealed that the prediction accuracy of the model was 0.852 and the kappa value was 0.53, which showed that the prediction accuracy of the model was high. In essence, the nomogram prediction model represents a scientific statistical method that can theoretically predict the probability of occurrence of any positive event in the clinic. Many oncologists currently use the nomogram prediction model to predict the prognosis of cancer patients and are also increasingly being used to predict the occurrence of clinical adverse events in non-oncology patients (4, 5, 10). One study established a Nomogram prediction model to predict the incidence of sternal incision problems in patients with median thoracotomy. The C-index of the model was 0.705, indicating that the Nomogram prediction model could accurately predict the occurrence of postoperative sternal incision problems (5). The findings of these studies suggest that the nomogram prediction model can accurately predict the incidence of positive clinical events in patients.

Among the five independent risk factors obtained from multifactorial logistic regression analysis, the age factor was a common clinical variable affecting survival outcome, and patients with IE ≥50 years of age will have a higher risk of in-hospital death compared to those <50 years of age (OR = 2.28). As an infectious condition, IE is frequently accompanied by a rise in white blood cell counts, and individuals with severe infections will develop leukopenia as a result of bone marrow suppression. Blood screening parameters could predict adverse outcomes and in-hospital mortality in IE, according to a study by Li et al. The neutrophil-to-platelet ratio (NPR) from routine blood tests at IE patients' admission allowed risk stratification and prediction of in-hospital mortality in IE patients, and the results showed that the prediction model using NPR as the main variable performed well in predicting in-hospital mortality in IE patients. Evaluation with the ROC revealed an AUC of 0.832 (19). In our study, patients with abnormal leukocyte count had a higher risk of death than those with normal leukocyte count (OR = 2.46).

TTE was performed on all of the individuals, while TEE was performed on some of the patients. TTE was our major method of locating cardiac vegetation in our study. When there is cardiac vegetation on the valve, the valve function will be seriously affected, resulting in valve insufficiency and movement disorders. Furthermore, pathogens wrapped in cardiac vegetation can produce toxins and enzymes, and these toxins and enzymes can cause tissue disintegration and invasion, which can lead to serious complications, such as heart failure caused by valve reflux or fistula, and the damaged tissue cannot regenerate, resulting in irreversible heart damage (20). Exfoliation of cardiac vegetation can cause embolism and other significant consequences in addition to direct injury to the attached myocardium. Embolism events produced by cardiac vegetation detachment have been revealed to be the main cause of poor prognosis and increased mortality in IE (21). Systemic embolism can result from the loss of left-sided IE cardiac vegetation, and all arteries may be affected. The most prevalent clinical embolism event in IE patients is embolic stroke, which is often associated with a poor prognosis (22). In our study, 193 (65.6%) patients had left cardiac vegetation (OR = 2.92) and 19 (6.5%) patients had right cardiac vegetation (OR = 6.58). Both left-sided IE and right-sided IE were associated with higher in-hospital mortality compared with patients without cardiac vegetation. During the treatment of IE, there are cases in which the infection is not controlled despite the identification of the pathogen by blood culture and the prolonged and continuous application of sensitive antimicrobial drugs. Recent research has provided a plausible explanation for this phenomenon, the concept of biofilms, it refers to bacterial populations living embedded in a self-produced extracellular polysaccharide slime-like matrix, and it will protect bacteria from host immune defenses and impedes antimicrobial efficacy. Biofilms protect bacteria from the immune system of the host and reduce the efficacy of antimicrobial medications, making it difficult to eradicate infection with drug therapy alone (23). In order to reverse this scenario, surgical treatment to remove infected lesions from heart tissue is required.

Approximately half of the patients with IE require surgical treatment due to severe complications (20). Currently, surgery is a common treatment for IE, and the indications for surgery are well agreed upon. Patients with signs of heart failure, severe valve dysfunction, PVE, paravalvular abscess or myocardial fistula invasion, recurrent systemic embolism, large mobile cardiac vegetation, and persistent sepsis after more than 5–7 days of adequate antibiotic therapy should be considered for surgical treatment (2). The majority of patients with IE are accompanied by the development of cardiac vegetation. In our study, 202 individuals (68.7%) were discovered to have cardiac vegetation during the early examination by TTE and TEE (193 left-sided vegetation, 19 right-sided vegetation, and 10 bilateral vegetation). In 262 (89.1%) of the 294 research participants, surgical exploration or recurring echocardiographic examination during hospitalization revealed cardiac vegetation. Surgery is beneficial for a variety of reasons. First, infected lesions in the heart or valve tissue can be entirely removed with direct-vision cardiac surgery. Second, while antimicrobial medications can prevent future structural damage to the heart, they cannot restore the integrity of injured tissue or valves, whereas surgery can restore valve function as well as the heart's structural integrity. Third, microbiological investigation of surgically excised tissue gives clinicians another chance to identify bacteria and assess their antibiotic susceptibility, and microbiological examination of surgical materials is especially useful in IE with negative blood cultures. Fourth, the most common preventable complication of IE is embolic stroke, and surgical cardiac vegetation removal can effectively prevent embolic occurrences (2, 24, 25). The left-sided IE and the right-sided IE serve as key variables in the nomogram prediction model we developed, and we can simulate different situations that may occur in IE patients in the clinic by assigning them different values, and the impact of these different conditions on the in-hospital mortality of IE is presented directly by the probabilities predicted by the model.

In terms of the timing of surgery, the traditional view is that patients with IE need 4–6 weeks of regular anti-infection treatment with normal temperature and negative blood cultures before surgery to reduce the probability of postoperative recurrence. However, during this time, patients may experience severe structural damage such as valve regurgitation and perforation, and may also develop heart failure and other manifestations as a result. Furthermore, some patients may develop embolic events as a result of cardiac vegetation dislodgement during therapy, as well as an increase in-hospital mortality. As a result, a growing number of experienced groups are actively advocating for early surgery rather than waiting for complications to appear (2, 3, 26). Although there is no clear definition of early surgery in terms of the timing of surgery, a growing number of studies have confirmed that early surgery reduces in-hospital mortality, follow-up mortality, and IE-related mortality in patients and does not increase the probability of recurrent infection (27–29).

In our study, 218 (74.1%) of the 294 study participants had surgical treatment. Our cardiac surgeons all used a median sternotomy surgical technique under general anesthesia, created extracorporeal circulation on a regular basis, and chose different surgical procedures depending on the patient's situation. On the basis of entirely removing the infected tissue in the heart, the principle of surgery is to reconstruct the heart structure to the greatest extent possible and restore normal heart function. One hundred and sixty five patients (75.7%) had valve replacement surgery, 36 (16.5%) had valve repair surgery, and 17 (7.8%) had congenital heart disease correction surgery. In our statistical prediction model, the absence of surgical therapy (OR = 18.94) was an independent risk factor for in-hospital mortality in IE. Patients receiving surgery had a 24.1% in-hospital mortality rate, while patients without surgery had a 75.9% in-hospital mortality rate. We can make a dynamic nomogram (Figure 5) based on a statistical prediction model, which can calculate the in-hospital mortality rate of IE by inputting the patient's early clinical features and surgical option into the model. The probability predicted by the model can intuitively tell doctors and patients the harm caused by IE and the benefits of surgical treatment.

In conclusion, the nomogram prediction model we developed performed well, with a C-index of 0.878 (95%CI: 0.824–0.931) and the internal validation of the model by bootstrap validation method showed a prediction accuracy of 0.852 and a bootstrapped-concordance index of 0.53. The development of dynamic nomogram could facilitate the use of statistical prediction models in clinical practice. The algorithm can accurately forecast IE patients' in-hospital death rate, and doctors can use the anticipated likelihood to make treatment decisions.

Comprehensive assessment of early clinical characteristics of patients can help doctors identify high-risk IE patients early. In this study, we found that age, abnormal leukocyte count, cardiac vegetations and the absence of surgical treatment were independent predictors of in-hospital mortality in IE patients. This Nomogram prediction model, which is based on IE patients' early clinical features and surgical options, has good diagnostic performance and clinical application potential.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee of Fujian Provincial Hospital. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

Z-JY and ZD drafted the manuscript. JL and Z-JN participated in the data collection. All authors were involved in the design of the study and approved the final version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.