A total of 867 cases with a definite diagnosis of IE were included in the study. Among these, 292 (33.7%) were female and the median age was 59.5 (16–96) years. A total of 711 cases had 2 major criteria and 136 cases had 1 major and ≥ 3 minor criteria.
Microbiological data
Blood cultures were negative in a total of 101 (11.65%) patients. Among the remaining 766 patients (88.4%) with identifiable microorganisms from blood cultures, the predominant pathogens were Staphylococcus aureus (n = 267, 33.6%), Streptococcus viridans (n = 149, 18.7%), enterococci (n = 128, 16.1%), and coagulase-negative staphylococci (n = 92, 11.6%). The details can be found elsewhere [7].
Univariate analyses
Demographic characteristics and concurrent cardiac disorders/conditions based on blood culture results were shown in Table 1, along with univariate comparisons. BCNE cases were younger than BCPE cases [median age 55 (18–86) versus 63 (16–96), p < 0.001]. The percentage of BCNE was found to be statistically significantly higher in the presences of rheumatic heart disease, congenital heart disease, prosthetic valves, mitral regurgitation, aortic stenosis, mitral stenosis, and was lower in the absence of a pacemaker (Table 1).
Table 1
Demographic characteristics and coexistent cardiac disorders/conditions according to blood culture results
| Culture Negative Endocarditis (n = 101, 11.65%) | Total (N = 867) | p value |
Age Median (Min – Max) | 55 (18–86) | 59,5 (16–96) | < 0.001 |
Gender (n,%) | Female | 27 (9.2%) | 292 | 0.116 |
Male | 74 (12.9%) | 575 |
Ischemic Heart Disease (n,%) | Absent | 92 (12.2%) | 756 | 0.213 |
Present | 9 (8.1%) | 111 |
Congestive Heart Failure (n,%) | Absent | 90 (11.7%) | 766 | 0.800 |
Present | 11 (10.9%) | 101 |
Degenerative Cardiac Lesions (n,%) | Absent | 94 (11.7%) | 802 | 0.818 |
Present | 7 (10.8%) | 65 |
Cardiomyopathy (n,%) | Absent | 96 (12%) | 800 | 0.266 |
Present | 5 (7.5%) | 67 |
Rheumatic Heart Disease (n,%) | Absent | 89 (10.9%) | 817 | 0.005 |
Present | 12 (24%) | 50 |
Congenital Heart Disease (n,%) | Absent | 94 (11.2%) | 838 | 0.033 |
Present | 7 (24.1%) | 29 |
Cardiac Implants (Subtotal) (n,%) | Absent | 59 (10.1%) | 584 | 0.041 |
Present | 42 (14.8%) | 283 |
1. Prosthetic valves (n,%) | Absent | 62 (9.8%) | 635 | 0.004 |
Present | 39 (16.8%) | 232 |
2. Pacemaker use (n,%) | Absent | 97 (12.4%) | 785 | 0.045 |
Present | 4 (4.9%) | 82 |
Valvular Problems (Subtotal) (n,%) | Absent | 61 (10.3%) | 592 | 0.070 |
Present | 40 (14.5%) | 275 |
1. Mitral regurgitation (n,%) | Absent | 75 (10.6%) | 708 | 0.041 |
Present | 26 (16.4%) | 159 |
2. Aortic regurgitation (n,%) | Absent | 88 (11.3%) | 782 | 0.270 |
Present | 13 (15.3%) | 85 |
3. Aortic stenosis (n,%) | Absent | 87 (10.8%) | 805 | 0.005 |
Present | 14 (22.6%) | 62 |
4. Tricuspid regurgitation (n,%) | Absent | 93 (11.5%) | 812 | 0.489 |
Present | 8 (14.5%) | 55 |
5. Mitral stenosis (n,%) | Absent | 92 (11.1%) | 829 | 0.018 |
Present | 9 (23.7%) | 38 |
Antibiotic use prior to blood culture sampling (n = 377) (n, %) | Yes | 53 (36.1%) | 147 | < 0.001 |
No | 31 (13.5%) | 230 |
Final model
The outcomes of our ultimate model indicate that the likelihood of a negative blood culture decreases significantly by 0.975 times (OR 0.975, 95% CI 0.963–0.987, p < 0.001) with each year increase in age. Furthermore, it was noted to increase by 2.036 times (OR 2.036, 95% CI 0.970–4.276, p = 0.049) in patients with rheumatic heart disease, 3.066 times (OR 3.066, 95% CI 1.564–6.010, p = 0.001) in those with aortic stenosis, 1.693 times (OR 1.693, 95% CI 1.012–2.833, p = 0.045) in individuals with mitral regurgitation, and 2.539 times (OR 2.539, 95% CI 1.599–4.031, p < 0.001) in patients with prosthetic valves (Table 2).
Table 2
B | S.E. | OR | 95% C.I.for OR | | |
Lower | Upper | p | |
Age | -0.025 | 0.006 | 0.975 | 0.963 | 0.987 | < 0.001 |
Rheumatic Heart Disease (Present versus Absent) | 0.711 | 0.378 | 2.036 | 0.970 | 4.276 | 0.049 |
Aortic Stenosis (Present versus Absent) | 1.120 | 0.343 | 3.066 | 1.564 | 6.010 | 0.001 |
Mitral Regurgitation (Present versus Absent) | 0.527 | 0.263 | 1.693 | 1.012 | 2.833 | 0.045 |
Prosthetic Valves (Present versus Absent) | 0.932 | 0.236 | 2.539 | 1.599 | 4.031 | < 0.001 |
Constant | -1.157 | 0.374 | 0.314 | | | 0.002 |
B: Regression coefficient, S.E.: Standard error, C.I.: Confidence interval, OR: Odds ratio |
The power of the model
The predictions of our model indicate the probabilities of patients belonging to culture-negative or culture-positive groups. When the cut-off value for these probabilities was calculated using the ROC analysis, the cut-off value corresponding to the best diagnostic success was found to be 0.104. The area under curve (AUC ± SE) was calculated to be 0.707 ± 0.027 (Fig. 1). The classification of patients based on the cut-off value determined according to the model is shown in Table 3 (p < 0.001). With a cut-off value of 0.104, the sensitivity of the final model was 70.3%, the specificity 57.0%, the positive predictive value 93.6% and the negative predictive value 17.8%.
Table 3
The classification of patients based on the cut-off value determined according to the nomogram
| Blood Culture Negative Endocarditis | Blood Culture Positive Endocarditis | Total (N, %) |
n | % within Model predictions | % within Blood Culture | n | % within Model predictions | % within Blood Culture |
Model Predictions | Negative | 71 | 17.8% (NPV) | 70.3% (Sensitivity) | 329 | --- | 43.0% (FN) | 400 |
Positive | 30 | --- | 29.7% (FP) | 437 | 93.6% (PPV) | 57.0% (Specificity) | 467 |
Total (N, %) | 101 | | | 766 | | | 867 |
NPV: Negative predictive value, PPV: Positive predictive value, FP: False positive, FN: False negative, ---: The % in these cells were not written because they do not have any meaning. |
The nomogram, which facilitates the interpretation of the final model, is shown in Fig. 2. The value "Prob" at the bottom of the figure corresponds to the probability of a patient's blood culture result being negative. The higher this value is, the greater the probability of a negative blood culture.