The impact of infectious disease consultation in candidemia in a tertiary care hospital in Japan over 12 years

Background Candidemia is one of the major causes of morbidity and mortality as a hospital acquired infection. Infectious diseases consultation (IDC) might be beneficial to improve candidemia outcomes; however, only limited data from short periods of time are available thus far. Methods An observational study of all candidemia patients at a large tertiary care hospital between 2002 and 2013 was conducted. A candidemia episode was defined as ≥ 1 positive result for Candida spp. in blood culture. Patients who died or transferred to another hospital within two days after their first positive blood culture were excluded. Independent risk factors for 30-day mortality were determined. Results Among 275 patients with 283 episodes of candidemia, 194 (68.6%) were male, and the mean age was 70.0 ± 15.8 years. Central line-associated bloodstream infections, peripheral line-associated bloodstream infections, intra-abdominal infection, and unknown source comprised 220 (77.7%), 35 (12.4%), 13 (4.7%), and 15 (5.3%) episodes, respectively. A total of 126 patients (44.5%) received IDC. Factors independently associated with 30-day mortality in patients with candidemia were urinary catheters use (adjusted hazard ratio [HR] = 2.94; 95% confidence interval [CI] = 1.48–5.87; P = 0.002) and severe sepsis/septic shock (adjusted HR = 2.10; 95% CI = 1.20–3.65; P = 0.009). IDC was associated with a 46% reduction in 30-day mortality (adjusted HR = 0.54; 95% CI = 0.32–0.90; P = 0.017). Conclusion IDC was independently associated with a reduction in 30-day mortality. Only 44.5% of patients with candidemia in this cohort received IDC. Routine IDC should be actively considered for patients with candidemia.

However, previous studies are limited by small patient sample sizes and short time periods of patient inclusion and/or follow up [13][14][15][16]. Moreover, studies of the survival outcome effects of IDC on candidemia in Asian countries, especially in Japan, are few.
In the past study [17], we reported that peripheral-line associated candidemia (PLAC) was an important cause of candidemia in the healthcare settings. Although IDC was associated with the predictors of PLAC, we did not evaluate the survival outcome of IDC on candidemia [17]. Thereby, we conducted a retrospective cohort study covering a 12-year period to evaluate the relationship between survival outcomes of candidemia and IDC in a tertiary care hospital in Japan as advances on previous work [17].

Hospital setting and study design
A retrospective cohort study of all candidemia was conducted between January 2002 and December 2013 at the National Centre for Global Health and Medicine (NCGM), which has approximately 780 inpatient beds and serves as a tertiary referral hospital for metropolitan Tokyo. This study was approved by the ethics committee of the NCGM (approval no: NCGM-G-001589-00) and was implemented in accordance with the Declaration of Helsinki. Patient information was anonymized and deidentified prior to analysis. Due to the retrospective nature of the study, patient consent was waived.

Data source
We identified all cases of candidemia using the microbiological laboratory database. The parameters were collected from patient charts included the following: (i) demographics including time period variable which was divided 2002-2009 and 2010-2013 due to the establishment of the official infectious disease consultation service with five infectious disease specialists in 2010; (ii) immunosuppressive status; (iii) background and comorbid conditions; (iv) recent healthcare-associated exposures; (v) recent exposure to antibiotic and antifungal therapy; (vi) infection-related characteristics; (vii) the severity of illness (sepsis, severe sepsis, and septic shock) and haematogenous dissemination; (viii) antifungal therapy against candidemia; (ix) outcome (clinical failure, persistent candidemia, in-hospital and 30-day/180-day mortality, discharge to a long term care facility (LTCF), re-admission, length of hospital stay after candidemia (excluding those who died), and duration of candidemia, as well as past our study [17][18][19]. Persistent candidemia means the case with follow up-blood culture positive of Candida. spp., after 72hr with empirical therapy [17]. Duration of candidemia was calculated from the date when the initial blood culture positive of Candida. spp. was drawn, to the date when the follow-up blood culture negative of Candida. spp was drawn [17]. Additionally, we reviewed IDC for management of candidemia and whether or not recommended candidemia treatment including examination for endophthalmitis and endocarditis were performed.

Definitions of variables including candidemia episode
We defined an episode of candidemia as isolation of Candida spp. from at least one peripherally-taken blood culture in a patient with clinical signs and symptoms of infection [17,20]. When caused by different Candida spp. or occurring at least 30 days apart, with improvement of clinical features of infection and at least one negative blood culture in the period [17,21], we considered that episodes of candidemia were separated. We excluded the episodes identified within 48 hours of hospital admission, because these episodes were thought not to be hospital acquired, and determining important candidemia parameters, such as duration, would be difficult [17]. Additionally, episodes who died or transferred to another hospital within 2 days after their first positive blood culture were also excluded, due to the limited opportunity for evaluating IDC effects.
Central line-associated bloodstream infections (CLABSI) and intra-abdominal infection (IAI) were defined according to the National Healthcare Safety Network Surveillance definition and the guidelines of the Infectious Diseases Society of America (IDSA) [17,20,22,23]. We defined peripheral line-associated bloodstream infections (PLABSI) as the presence of at least one of the following conditions: (i) the presence of phlebitis, and/or (ii) resolution of clinical symptoms after short-term peripheral line withdrawal with a careful exclusion of another focus of bacteraemia [17,24]. We defined empiric and definitive therapy as administration of systemic antifungal drugs within 72 hours of the onset of candidemia, and based on the guideline of IDSA [17,20]. The variables related therapy (the time to antifungal therapy, adequate source control, time to central or peripheral vein catheter removal, and clinical failure) were defined as past studies [17,25] Infectious disease consultation and appropriateness of an antifungal therapy or duration IDC was recommended for patients with candidemia as per hospital policy, and was performed when requested by the primary physician in charge. Request for consultation was not mandatory. IDC comprised chart review, physical examination of the patient, a follow-up visit, and written recommendations for therapy based on published IDSA guidelines [20]. Individual case discussion was performed with the primary physician in charge. We evaluated whether antifungal therapy, including the duration, was in accordance with the published IDSA guidelines [20].

Microbiological data
Candida spp. from positive blood culture were identified using API 20 C AUX (Biomerieux Japan Co., Ltd., Japan), and ID 32 C (Biomerieux Japan Co., Ltd., Japan). Antifungal susceptibility testing was performed using the commercially prepared colorimetric microdilution panel (ASTY; Kyokuto Pharmaceutical Industrial Co., Ltd.). which was developed according to the CLSI recommendation. During the study period, there were no changes to the microbiological identification and susceptibility testing process.

Statistical analysis
Continuous variables were shown as the mean ± standard deviation (SD) or the median with interquartile range (IQR), and compared using Student's t-test or Mann-Whitney U test. Categorical variables were shown as absolute and relative frequencies, and compared using the χ 2 test or Fisher's exact test.
We compared demographic characteristics, clinical factors, and outcomes between episodes with and without IDC, using logistic regression univariate analysis with odds ratios (OR) and 95% confidence intervals (CI). Multivariable survival analyses were performed and predictive models for 30-day and 31-180-day all-cause mortality were built.
For the 30-day and 31-180-day mortality models, Cox proportional hazards models were applied. We considered the potential predictive factors with a P-value of < 0.10 in the univariate analysis, or that were hypothesized a priori to be clinically or epidemiologically important, for inclusion in the multivariate model for predictive factors. The relationship of the variable of IDC and mortality during the 180-day follow up period was illustrated using a Kaplan-Meier estimator. Survival characteristics were compared between groups with and without IDC using the log-rank test. We defined statistical significance as a 2-sided P-value of < 0.05, and all statistical analyses were performed with SPSS Version 24 (IBM Corp., Armonk, NY, USA).

Description of candidemia from 2002 to 2013
The 12-year study period included a total of 283 episodes of candidemia from 275 patients.

Comparison of candidemia patients with and without IDC
Comparison of candidemia patients with and without IDC is summarized in Table 1.
The IDC group (55.6% [70/126]) had significantly more patients with candidemia during 2010-2013 than non-IDC group (21.0% [33/157]) (P < 0.001). Although there was a similar profile of chronic conditions in both groups, patients in the IDC group were significantly more associated with HIV infection (P = 0.006) and connective tissue disease (P = 0.018). There were no differences of healthcare-associated exposures between the two groups. The IDC group had more previous exposure to carbapenems (P = 0.014), metronidazole (P = 0.023), micafungin (P = 0.020), and liposomal amphotericin b (P = 0.038). Although there was no difference of isolated Candida spp. between groups, the IDC group had significantly more episodes of polymicrobial bacteraemia/fungaemia than the non-IDC group (P = 0.013). The proportion of PLABSI was significantly higher in the IDC group (P = 0.020). No differences of severe sepsis/septic shock and chorioretinitis were observed between groups, but patients in the IDC group developed significantly more acute renal failure than did non-IDC group (P = 0.034). Regarding treatment, the IDC group more frequently received micafungin (P < 0.001) as empiric therapy. In contrast, the number of patients who received fluconazole or who did not receive any empiric therapy was higher in the non-IDC group (P = 0.001 for both). Patients with IDC were more likely to receive the appropriate definitive antifungal therapy (P < 0.001) and appropriate duration of antifungal therapy (P < 0.001). Although there c Other was no difference in the overall adequate source control between groups, early CVC removal and early peripheral-line removal were more frequent in patients with IDC than patients without (P = 0.023, and P = 0.013, respectively). Patients with IDC received more consultations from an ophthalmologist to evaluate chorioretinitis (P < 0.001) ( Table 2). The overall 30-day and 180-day all-cause mortality were 23.7% (67/283) and 47.0% (133/ 283), respectively. The median length of time from diagnosis to death was 26 (IQR, 11-55) days. No difference in hospital mortality between two groups, but the IDC group had more episodes of persistent candidemia (P < 0.001) with longer total LOS (P < 0.001) and longer duration candidemia (P <0.001) ( Table 1). The Kaplan-Meier survival curve for patients with candidemia stratified by IDC is shown in Fig 1.

Predictive factors for mortality of candidemia
Predictive factors for mortality of candidemia are summarized in Table 3.
Patients with candidemia who died within 30 days of the first positive blood culture were more likely to be older (P = 0.035), to have chronic heart disease (P = 0.047), to have urinary catheters placed (P < 0.001), to have exposure to glycopeptide (P = 0.044), to have polymicrobial bacteremia/fungemia (P = 0.028), to have CLABSI (P = 0.050), and to be in severe sepsis/ septic shock (P < 0.001). They were less likely to have candidemia due to C. parapsilosis (P = 0.040) and to have PLABSI (P = 0.041) than were candidemia patients who survived longer than 30 days after candidemia onset. In the multivariate model, factors independently associated with 30-day mortality among candidemia were urinary catheters use at onset (adjusted hazard ratio [HR] = 2.94; 95% CI = 1.48-5.87; P = 0.002) and severe sepsis/septic shock (adjusted HR = 2.10; 95% CI = 1.20-3.65; P = 0.009). IDC was associated with a decreased risk of 30-day mortality (adjusted HR = 0.54; 95% CI = 0.32-0.90; P = 0.017). For the 216 patients who survived 30 days after initial positive culture, the effect of IDC from 31-to 180-day of the diagnosis of candidemia was also evaluated. There was no statistical difference in 31-180-day mortality after the diagnosis of candidemia between patients who did and did not receive IDC (crude HR = 0.98; 95% CI = 0.60-1.58; P = 0.92).

Discussion
We showed that IDC was associated with a 46% reduction in all-cause mortality among candidemia patients within 30 days of candidemia onset. Past studies have described the positive effect of IDC on mortality with candidemia; 18-24% and 39-56% in the IDC and the non-IDC groups, respectively [13][14][15][16]. Compared the data of rates of IDC, patient characteristics at baseline, and clinical outcomes of candidemia in a hospital in North America [16], the rate of IDC was lower (45% vs 77%), the median age and population of male were higher (69 years old vs 53 years old; 72% vs 55%), and 30 days mortality was slightly lower (18% vs 20%) among IDC group in our results. However, these studies were limited due to small sample sizes (50-171 patients) and short study periods (1-3 years). Although our study was a retrospective cohort study, our sample size was larger (283 patients) and study period was longer (12 years) than past studies [13][14][15][16]. Moreover, because previous studies did not use the cox proportional hazard model [13][14][15][16], the evaluation of the effect of IDC on mortality with candidemia was prone to confounding effects, and might thus be incorrect. In our study, the IDC group more frequently received appropriate choice (OR = 6.48; 95% CI = 2.20-19.07; P < 0.001) and appropriate duration (OR = 3.89; 95% CI = 2.19-6.92; P < 0.001) of antifungal therapy than the non-IDC group. The duration of antifungal therapy was significantly longer in the IDC group than the non-IDC group (19 days versus 14 days; P < 0.001). Past research has also revealed that IDC intervention for candidemia led to appropriate antifungal therapy [14][15][16].   Besides antifungal therapy, adequate source control is an important component of appropriate candidemia management. While previous research has reported that removal of CVC was more frequent in the IDC group than in the non-IDC group [13][14][15][16]26], our results showed no statistical difference in numbers of removal of CVC between both groups. In our study, we found that the IDC group had a statistically higher proportion of the removal of peripheral-line than the non-IDC group (OR = 2.91; 95% CI = 1.21-6.97; P = 0.013), which, to the best of our knowledge, has not been reported previously. This might reflect the difficulty of diagnosing PLAC, and need for the consultation of an ID specialist [17]. In fact, the proportion of PLAC was higher in the IDC group than the non-IDC group (OR = 2.34; 95% CI = 1.13-4.87; P = 0.020). Similar to a past study [26], our study revealed that consultation with an ophthalmologist was conducted more often in the IDC group then in the non-IDC group (OR = 6.68; 95% CI = 3.92-11.38; P < 0.001). For candidemia, the evaluation of endophthalmitis is thought to be a predictive factor for outcome improvement of candidemia, and contributes to an appropriate choice and duration of antifungal therapy [20,27]. While not statistically significant, the proportion of endophthalmitis was higher in the IDC group than the non-IDC group in our study (OR = 1.58; 95% CI = 0.76-3.27; P = 0.22), suggesting that IDC led to the detection of endophthalmitis in many cases.
On the other hand, urinary catheters placed (for � 2 days) at onset of candidemia (adjusted HR, 2.94; 95% CI, 1.48-5.87, P = 0.002) and severe sepsis/septic shock (adjusted HR, 2.10; 95% CI, 1.20-3.65, P = 0.009) were independently associated with increased 30-day mortality. These results indicate that IDC might not significantly improve outcomes for patients with very severe conditions. Clinical severity such as severe sepsis/septic due to candidemia was previously reported as an independent risk factor for 30-day mortality [28].
The IDC group was associated with the following factors indicating the clinical severity of candidemia: ICU stay, development of acute renal failure, exposure to carbapenems/micafungin/liposomal amphotericin b, longer duration of candidemia, and longer length of hospital stay after candidemia [28]. In the IDC group, episodes of polymicrobial bacteremia/fungemia were statistically more frequently found than in the non-IDC group (OR = 2.91; 95% CI = 1.21-6.97; P = 0.013). This may be due to heavier contamination of peripheral lines than central lines, and better ability of ID specialists for diagnosing PLAC [17]. IDC as candidemia intervention may not improve the severity of the current disease status, however it did improve the mortality of candidemia, with improved management of the disease.
This study has several limitations. Due to the retrospective nature of the study, we were unable to collect information regarding the duration of time from onset of candidemia to IDC. This is a time-dependent variable, and might have been an unmeasured confounding factor. Similarly, unmeasured confounding factors such as other interventions might have affected the patients' outcomes. Following the recommendation of IDC for candidemia was not mandatory. Although IDC consisted of chart review, physical examination of the patient, follow-up visits, and written recommendations for therapy based on published IDSA guidelines, final decision-making in each case depended on the primary team in charge.
In conclusion, this study was the first epidemiological clinical study with a large sample size and a long study period to evaluate the value of IDC in candidemia in Japan. IDC was associated with a 46% reduction in adjusted all-cause mortality among candidemia patients within 30 days of onset of candidemia. These results suggest that IDC should be actively considered to improve the frequently poor outcome of candidemia patients. Further studies, including evaluations of the outcome effects of time between IDC and onset of candidemia, are needed to further reduce the mortality of candidemia.
Supporting information S1 Table. A data sheet of the present study. (XLSX)