Combination of Sequential Organ Failure Assessment (SOFA) score and Charlson Comorbidity Index (CCI) could predict the severity and prognosis of candidemia more accurately than the Acute Physiology, Age, Chronic Health Evaluation II (APACHE II) score

Background Candidemia has emerged as an important nosocomial infection, with a mortality rate of 30–50%. It is the fourth most common nosocomial bloodstream infection (BSI) in the United States and the seventh most common nosocomial BSI in Europe and Japan. The aim of this study was to assess the performance of the Sequential Organ Failure Assessment (SOFA) score for determining the severity and prognosis of candidemia. Methods We performed a retrospective study of patients admitted to hospital with candidemia between September 2014 and May 2018. The severity of candidemia was evaluated using the SOFA score and the Acute Physiology, Age, Chronic Health Evaluation II (APACHE II) score. Patients’ underlying diseases were assessed by the Charlson Comorbidity Index (CCI). Results Of 70 patients enrolled, 41 (59%) were males, and 29 (41%) were females. Their median age was 73 years (range: 36–93 years). The most common infection site was catheter-related bloodstream infection (n=36, 51%).The 30-day, and in-hospital mortality rates were 36 and 43%, respectively. Univariate analysis showed that SOFA score ≥5, APACHE II score ≥13, initial antifungal treatment with echinocandin, albumin < 2.3, C-reactive protein > 6, disturbance of consciousness, and CCI ≥3 were related with 30-day mortality. Of these 7, multivariate analysis showed that the combination of SOFA score ≥5 and CCI ≥3 was the best independent prognostic indicator for 30-day and in-hospital mortality. Conclusions The combined SOFA score and CCI was a better predictor of the 30-day mortality and in-hospital mortality than the APACHE II score alone. Supplementary Information The online version contains supplementary material available at 10.1186/s12879-020-05719-8.


Background
Candidemia has emerged as an important nosocomial infection, with a 30-50% mortality rate [1][2][3]. It is the fourth most common nosocomial bloodstream infection (BSI) in the United States, and the seventh most common nosocomial BSI in Europe and Japan [4][5][6]. Previously reported risk factors for candidemia include central venous catheterization (CVC), neutropenia, malignancy, abdominal surgery within the previous 30 days, immunosuppressant use and admission to an intensive care unit (ICU). The ubiquity of these risk factors means that most physicians may encounter patients with candidemia. However, there is no established tool or method to evaluate the severity and prognosis of patients with candidemia. The Acute Physiology, Age, Chronic Health Evaluation II (APACHE II) score has been reported to be able to predict the mortality among patients with candidemia [7]. Candidemia patients commonly have severe comorbidity which is responsible for the severity of candidemia. It is reasonable that APACHE II score is useful to predict the prognosis of candidemia as the score includes the evaluation of comorbidity [1][2][3]. However, in general practice, APACHE II score is very complicated to administer, and its utility in clinical practice is limited. Recently, it has been reported that quick Sequential Organ Failure Assessment (qSOFA) and Sequential Organ Failure Assessment (SOFA) scores are reliable as prognostic tools in the management of sepsis and other infections [8][9][10]. qSOFA and SOFA consist of 3 to 5 items and are simpler to administer than APACHE II, and are suitable for use by all physicians. Besides, we hypothesize that the Charlson Comorbidity Index (CCI) which is commonly used for the evaluation of comorbidity in general wards, would be appropriate for evaluating the patients' condition. Therefore, we conducted this retrospective study to assess whether qSOFA and SOFA scores plus CCI could predict mortality in patients with candidemia.

Study design and patient selection
In order to investigate whether SOFA score and the Charlson Comorbidity Index (CCI) could predict the severity and prognosis of patients with candidemia, we performed this retrospective study at the Aichi Medical University hospital, a 900-bed tertiary care center located in Aichi prefecture in central Japan. The study was conducted among patients hospitalized with candidiasis between September 2014 and May 2018. We included patients aged ≥16 years who had hospital-acquired candidemia, which was defined as at least with one positive blood culture of Candida species in patients hospitalized for more than 48 h, with clinical signs and symptoms of infection. Patients who did not have enough information about the disease or those we could not locate because of a transfer to other hospitals were excluded. This study was approved by the Institutional Review Board of Aichi Medical University Hospital (16-H105).

Severity of candidemia
In medical practice, the severity of the initial presentation of candidemia has been assessed using the APAC HE II score [7], qSOFA score, and SOFA score [8][9][10].

Definition of variables
Disseminated intravascular coagulation (DIC) was diagnosed according to the diagnostic criteria developed by the Japanese Association for Acute Medicine (JAAM DIC diagnostic criteria) [6]. An altered state of consciousness was defined as Glasgow coma scale (GCS) < 15. Neutropenia was defined as an absolute neutrophil count < 500 × 10 6 /μl. Antifungal treatment was classified as appropriate or inappropriate depending on whether the identified pathogens were sensitive to the initially prescribed antifungal drugs.

Performance status
Patients' general conditions were evaluated by using the Eastern Cooperative Oncology Group (ECOG) performance status (PS) [11] and Karnofsky Performance Status (KPS) [12]. In medical practice, determining PS is an attempt to quantify cancer patients' general wellbeing and activities of daily living. Recently, this measurement is used to determine whether patients can receive anticancer treatment as well as a tool to evaluate conditions such as interstitial lung disease or emphysema [13][14][15].

Evaluation of comorbidities
We used the Charlson Comorbidity index (CCI) to evaluate the patients'comorbidities [16,17]. This index could forecast ten-year mortality for 22 different underlying disease and medical conditions, including cardiac disease, AIDS, and malignancy. Each condition is assigned a score of 1,2,3 or 6 depending on the risk of death, and the sum of these scores is used as the total score to predict mortality.
Identification of candida spp. and susceptibility testing Candida species were identified using the VITEK-MS system (bioMérieux, Marie l'Étoile, France).
Susceptibility to amphotericin B, caspofungin, fluconazole, itraconazole, and voriconazole was detected using the AST-YS07 card of VITEK-2 (bioMérieux, Marie l'Étoile, France). The susceptibility of antifungal drugs was assessed with minimum inhibitory concentration (MIC) testing according to the guidelines of the Clinical and Laboratory Standards Institute (CLSI) [24,25]. MIC values were interpreted according to species-specific clinical breakpoints as established by CLSI for caspofungin (CPFG), fluconazole (FLCZ), itraconazole (ITCZ) (only for C. albicans), and voriconazole (VRCZ) [24]. Susceptibility to amphotericin B (AMB) and liposomal amphotericin B (L-AMB) were interpreted according to species-specific clinical breakpoints as established by European Committee on Antimicrobial Susceptibility Testing (EUCAST) [25].

Statistical analyses
The data for categorical variables were reported as percentages, and continuous variables were reported as the mean ± the standard deviation (SD), or the median with the interquartile range (IQR). Chi-square tests or Fisher's exact tests (two-tailed) were used to compare categorical variables, and unpaired Student's t-tests or Mann-Whitney U-test were used to compare continuous variables. Logistic regression analysis was used to identify independent risk factors associated with 30-day or in-hospital mortality. Variables with p≤0.10 on univariate analyses were entered into the multivariable model. The Hosmer-Lemeshow test was performed to assess the calibration of the model. Receiver-operating characteristic (ROC) curves were evaluated for the predictive values for 30-day or in-hospital mortality. Statistical analyses were performed using SPSS Version 23 for Windows (SPSS Inc., Chicago, IL, USA). A p-value< 0.05 was considered statistically significant.

Incidence of candidemia
During the study period (941,990 patient-days), the incidence of candidemia was 7.4/100,000 persons. Candida spp. was the fourth most common pathogen among patients with bloodstream infections in our institute, after coagulase-negative Staphylococcus, Escherichia coli, and Staphylococcus aureus, accounting for 7% of positive blood cultures.

Demographic data
The demographic data and clinical characteristics of the patients, clinical outcomes, pathogens isolated were summarized in Table 1. A total of 70 patients with candidemia were enrolled in this study, of whom 41 (59%) were males and 29 (41%) were females. Their median age was 73 years (range 36-93 years). The patients had a 30-day mortality of 36%, and an inhospital mortality of 43%. Six of the 10 patients who received an antifungal treatment based on a positive blood culture, died. Of the 68 patients, 27 (40%) received antifungal treatment within 24 h from the onset. The difference in the time to starting antifungal treatment did not differ significantly between those who survived, and those who died (45.6 h vs 36 h, p=0.4).

Microbiological data and antifungal drug selection
Detection of Candida spp.
C. albicans was the most common of the Candida species identified, accounting for 39% of the cases, followed by C. parapsilosis (28%), C. glabrata (20%), C. tropicalis (10%) and others (9%). The 30-day mortality according Fig. 2 ROC curves of SOFA score, CCI, APACHE II, combined APACHE II and CCI and combined SOFA score and CCI for 30-day mortality to species was highest in patients with Patients with C. albicans candidemia had the highest mortality (44%). C. albicans candidemia (vs. non-C. albicans) was not associated with an increased 30-day mortality ( The most frequently used initial antifungal treatments were echinocandin (63%), and L-AMB (26%). All candida isolates were susceptible to initial antifungal agents based on CLSI and EUCAST breakpoints. Of the 70 patients, 74% received antifungal treatment within 3 days of the onset, which reflects the period taken for fungus to grow from the blood cultures., and only three patients (4%) did not receive antifungal treatment.
Relationship between qSOFA, SOFA, APACHE II eastern cooperative oncology group performance status and Karnofsky performance status The SOFA, APACHE II score and ECOG-PS were higher patients with qSOFA ≥2 than in those with qSOFA < 2. KPS was lower in patients with qSOFA ≥2 than in those with qSOFA< 2 (Table S1).

Associations between qSOFA, SOFA score and 30-day mortality
The associations between qSOFA, SOFA, APACHE II score, Candida spp. and 30-day mortality among are shown in Fig. 1. Higher qSOFA and SOFA scores were associated with a higher 30-day mortality. The 30-day mortality of Candida albicans was the highest among these patients.

Relationship between removal of central venous catheters and outcomes
External devices such as CVCs or CV ports were removed in 43 of the 53 (81%) patients. CVCs or CV ports were removed within 24 h from the onset in 29 of the 43 patients (67%). In 37 of the 43 patients (86%), CVCs or CV ports were removed within 3 days of the onset of candidemia. Seven of the 53 patients (13%) did not have their CVCs or CV ports removed. Not removing CVCs or CV ports was a poor prognostic factor (OR 12.5, 95% In terms of the patients with cancer, there were no differences of ECOG-PS and KPS between the patients with and without CVCs or CV ports (data not shown). As for the 28 cancer patients with CVCs or CV ports, not removing CVCs or CV port was a poor prognostic factor (OR 7.4, 95% CI: 1.5-152.3, p=0.008). Removing CVCs or CV ports within 24 h of the onset was not a poor prognostic factor (OR 0.6, 95% CI: 0.1-2.5, p=0.445).

Receiver-operating characteristic (ROC) curves of predictive values for 30-day or in-hospital mortality (Figs. 2 and 3)
With respect to the diagnostic value of predictive values for 30day and in-hospital mortality among candidemia patients, the area under the ROC curve for SOFA score, CCI, APACHE II score, combined SOFA score with CCI and combined APAC HE II score with CCI were 0.77 ( Prognostic accuracy of SOFA and APACHE II score for 30day mortality Table 2 shows the prognostic accuracy of SOFA and APACHE II for predicting 30-day mortality. The cut-point for the SOFA score was 5. and was selected based on the Youden Index.

Analysis of prognostic factors
The prognostic factors for 30-day mortality are shown in Table 3. Seven of 35 parameters (Table S2) were found to be associated with 30-day mortality in the univariate analysis. These were: SOFA score ≥5, APACHE II score ≥13, initial antifungal treatment with echinocandin, albumin < 2.3, C-reactive protein > 6, disturbance of consciousness, and CCI ≥3. The Hosmer Lemeshow statistic suggested a good fit (χ 2 =11.4, p=0.184) in the cohort study.
Of these seven parameters, logistic regression analysis showed that the combination of SOFA score ≥5 and CCI ≥3 were independent prognostic factors for 30-day mortality and in-hospital mortality (Table 4).

Discussion
In contrast to the complicated APACHE II scoring system, which consists of ASP points, age and immunocompromised state, both the SOFA score and CCI are simple to calculate. SOFA is a tool for evaluating the severity of failure of organs such as the kidney or the liver [26]. In addition, while platelet count is included in SOFA score, it is not included in APACHE II. This difference might contribute to the more precise evaluation of patients' conditions by SOFA compared to APACHE II in seriously ill patients with conditions such as disseminated intravascular coagulation or multiple organ failure. In this study, 23 of the 70 patients were diagnosed as having candidemia in ICU, while the remaining   Table S1 showed that the patients with qSOFA≧2 could reflect the disease severity of SOFA score and APACHE II score. In a general ward, patients' condition should be evaluated by qSOFA [27] (γ=0.505 with Spearman's rank correlation test). The results showed the validity of qSOFA among candidemia patients in a general ward. The severity and prognosis of candidemia could be evaluated by SOFA score more reliably than by APACHE II. CCI is a useful tool for evaluating comorbid conditions in patients with underlying diseases [13,14]. Furthermore, the outcome of candidemia can be affected by patients' general condition and underlying diseases such as malignancy. Thus, it is reasonable that combined SOFA score and CCI could more precisely predict the severity and prognosis of patients with candidemia. Removal of CVCs is considered to be a standard procedure among patients with candidemia [27,28]. Although it has been reported that removing CVCs within 24 h of the onset of candidemia is associated with a reduced mortality rate [29], our study results did not confirm this finding. These discrepancies may be attributable to the lack of uniformity of variables and differences in both previous studies and ours. We found that patients whose CVCs were removed, showed better PSs and had lower SOFA scores than those whose CVCs were not removed (Table 5).
We can assume that physicians tended to remove CVCs of patients in good general condition, whose life expectancy is considered favorable. Conversely, physicians might be hesitant to remove CVCs of patients with poor general condition.
Candidemia is almost always primarily of gastrointestinal origin in patients with cancer who have severe neutropenia and mucositis, and removal of CVCs is less likely to have an impact on the outcome in this setting. However, we believe that removal of CVCs is appropriate in the treatment of patients with candidemia considering that CVCs are foreign bodies.
On insertion of CVCs, physicians should take into account the ease and safety of CVC removal when candidemia is suspected, and select catheter device. Peripherally inserted central venous catheters could be removed easier and safer than CV port. This study has several limitations. Firstly, it is a retrospective study with a relatively low sample size. Thus, there might be a bias in the data selection and analysis. Secondly, we enrolled only patients with candidemia diagnosed by a blood culture. Generally, 50% of individuals with candidemia have negative blood cultures. Thus, the patients in our study may not have been representative of all individuals with candidemia.

Conclusions
Combined SOFA score and CCI could possibly be a more accurate predictor of severity and prognosis among patients with candidemia than the APACHE II score for 30-day, or in-hospital death.