Patient characteristics
The baseline clinical characteristics of the 140 patients included in the study are summarized in Table 1. The mean patient age was 65.98±14.63 years and no significant difference in age was observed between the two groups. Following matching, there was no significant difference in patient locations at the time of SIRS occurrence between the case and control groups. At SIRS onset in both groups, 27 (38.6%) and 43 (61.4%) patients were in the ICU and general ward, respectively.
Among all patients included in this study, the most common underlying disease was solid cancer, followed by neurological disorders and diabetes. Assessment of co-morbidity severity using the McCabe classification showed a greater number of rapidly fatal cases in the case group than in the control group (5.7% vs. 0%, p=0.007). There was also a clear difference in the cause of infection between the two groups (Table 1). The most common cause of candidemia in the case group was catheter-related infection (CRI), which accounted for 48 of the 70 cases (68.6%). CRIs were followed by primary bacteremia (28.6%, 20/70 cases). In contrast, the major cause of SIRS in the control group was pneumonia, which accounted for 42.9% (31/70) of the cases. However, the severity of infection assessed by the Pitt bacteremia score, severe sepsis, or septic shock did not differ significantly between the two groups (Table 1). To measure the CS, the presence of Candida colonization, total parenteral nutrition (TPN), and receipt of surgery were examined. The proportion of patients with Candida colonization and the proportion of patients who underwent surgery while hospitalized were significantly higher in the case group than in the control group (Table 1). All patients were administered antibiotics at the time of participation in this study. The overall 14-day mortality was 10.7% (15/140) and was significantly higher in the case group than in the control group (18.6% vs. 2.9%, p=0.005).
Comparison of DNI and other indicators as predictive markers of candidemia
To evaluate DNI as a predictive factor for candidemia, DNI, CRP level, procalcitonin level, leukocyte count, and CS>3 were compared between the case and control groups (Table 2). The DNI_D1 in the case group was significantly higher than that in the control group (3.5%, [0.5–3.3] vs. 1.3% [0.1–2.4], p<0.001). The DNI_48 value was also significantly higher in the case group (2.0%) than in the control group (1.0%) (p<0.001). The procalcitonin level was also higher in the case group than in the control group (Table 2). However, CS>3, a known predictive factor for candidemia, did not differ significantly between the two groups.
As our control group included five non-infectious SIRS patients, we also compared the DNI values between patients with clinically documented infections and non-infectious SIRS in the control group. The DNI_D1 values were 1.3% (0.2-2.4) in patients with clinically documented infection and 0.0% (0.0-1.0) in those with non-infectious SIRS. While the patients with clinically documented infection tended to show higher DNI_D1 value, the difference was not statistically significant (p=0.304). A similar trend was observed for DNI_48 values (1.0 % vs. 0.2%, p=0.178). Comparison of the case and control groups after excluding patients with non-infectious SIRS showed that the clinical characteristics did not differ from those of the entire control group. The DNI_D1 value in the case group (3.5%, 0.5-3.3) was significantly higher than that in the control group after excluding patients with non-infectious SIRS (1.3%, 0.2-2.4) (p<0.001). The DNI_48 value was also significantly higher in the case group (2.0%) than in the control group after excluding patients with non-infectious SIRS (1.0%) (p=0.010).
The results of multivariate analyses to determine the independent predictive factors for candidemia in the case and control groups are summarized in Table 3. The factors that differed significantly in univariate analysis, CS >3 used for predicting candidemia, and clinically important factors exhibiting insignificant differences in univariate analysis were included in the multivariate analysis.
The multivariate analyses identified DNI_D1 (OR, 2.183, 95% CI, 1.421–3.217, p<0.001) and Candida colonization (OR, 7.361, 95% CI, 1.717–31.553, p=0.007) as useful indices for predicting candidemia (Table 3). Multivariate analyses of the control group, including only patients with clinically documented infection, also identified DNI_D1 (OR, 2.139, 95% CI, 1.424-3.213, p<0.001) and Candida colonization (OR, 7.638, 95% CI, 1.732-33.676, p=0.007) as useful indicators for predicting candidemia.
Optimal cutoff value of DNI for predicting candidemia
ROC curve and AUC analyses were conducted to determine the DNI_D1 cutoff value for predicting candidemia (Fig. 1). The optimal cutoff value was 2.75% and the AUC of DNI_D1 was 0.804 (95% CI, 0.719-0.890, p<0.001). The sensitivity and specificity in predicting candidemia for a cutoff DNI value of 2.75% were 72.9% and 78.6%, respectively, and the positive and negative predictive values were 77.3% and 74.3%, respectively. For DNI_D1 >2.75%, the OR for the presence of candidemia was 9.842 (95% CI, 4.562-21.402, p<0.001).
In comparison with the control group, including only patients with clinically documented infection, the cutoff value was 2.75%, and the AUC of DNI_D1 was 0.796 (95% CI, 0.721-0.871, p<0.001). The sensitivity, specificity, positive predictive, and negative predictive value were 72.9%, 76.9%, 77.3%, and 72.5%, respectively. The OR for the predicting candidemia was 8.947 (95% CI, 4.096-19.544, p<0.001).
Factors associated with 14-day mortality in patients with candidemia
To determine the prognostic factors for mortality in patients with candidemia, various factors were comparatively analyzed according to 14-day mortality. Among 70 patients with candidemia, 13 died within 14 days of candidemia onset. There were no differences in age, sex, or patient location at the time of candidemia onset (p=0.210) between the survivor and non-survivor groups. The underlying diseases and site of infection also did not differ significantly between the survivor and non-survivor groups. The non-survivor group showed a significantly higher Pitt bacteremia score (4.0 [1.0-5.0] vs. 0.0 [0.0-2.0], p<0.001) and septic shock rate (61.5% vs. 7.5%, p=0.003) than those in the survivor group. The percentage of patients with CS ≥3 was higher in the non-survivor group than in the survivor group (46.2% vs. 19.3%, p=0.042). DNI_D1 and DNI_48 values were also significantly higher in the non-survivor group than in the survivor group (7.4% [4.0–22.0] and 6.1% [2.1–14.4], respectively. The TTP in all patients with candidemia was 48 hours (32.0-73.0, range 10-120 hours). Among patients with candidemia, the most common Candida species was C. albicans (33/70, 47.1%) followed by C. tropicalis (16/70, 22.9%), C. parasilopsis (10/70, 14.3%), and C. glabrata (9/70, 12.7%). The TTPs for the four most frequently identified Candida species were 68 hours (47.0-78.0), 30.5 hours (26.5-33.5), 47.0 hours (34.0-58.0), and 72.0 hours (50.5-93.0), respectively
The TTP in the non-survivor group was shorter than that in the survivor group, but the difference was not statistically significant (31.0 hours [26.5-53.0] vs. 48.0 hours [34.0-72.0], p=0.066).
The non-survivor group showed a significantly shorter TAT than did the survivor group (36 hours [12.5-42.0] vs. 60 hours [42.0-96.0] p=0.013), showing that antifungal therapies were administered earlier to the non-survivor group. As shown in Table 4, no significant difference was observed in TAT between the survivor and non-survivor groups in the multivariate analysis. However, antifungal therapy tended to be administered faster, with earlier Candida detection from blood cultures (r=0.56 p=0.044). In addition, TAT and 14-day mortality showed an inverse correlation (r=-0.33, p=0.011).
The duration until the negative conversion of candidemia also did not differ between the survivor and non-survivor groups (Table 4).
Multivariate analysis also confirmed that DNI_D1 (OR, 1.156, 95% CI, 1.039-1.287, p=0.008) and the occurrence of septic shock were reliable prognosis factors for 14-day mortality. The DNI_D1 cutoff value for predicting 14-day mortality was 3.95%, and the AUC was 0.769 (95% CI, 0.624-0.914, p=0.001).