The epidemiology, antifungal use and risk factors of death in elderly patients with candidemia: a multicentre retrospective study

Background The elderly patients affected by candidemia are growing in proportion to inpatients, but available data are limited. We aimed to determine the epidemiology, antifungal management and clinical risk factors of death in the elderly population with candidemia in China. Methods This retrospective study included 63 elderly (≥65 years) and 84 younger patients (16–60 years) at 4 tertiary hospitals. Multivariable logistic regression model was used to identify independent risk factors of death in elderly patients. Results The distribution of Candida species did not differ between elderly and younger patients (p >0.05). Resistance to fluconazole and voriconazole for non-Candida albicans species in elderly patients was approximately double that in younger patients. Host-related risk factors (e.g., underlying solid tumour, diabetes mellitus and chronic renal failure) and hospital-related factors (e.g., prior stay in an intensive care unit, mechanical ventilation, central vascular and urethral catheters placement) were identified more common in elderly patients. Elderly patients less often received triazoles and were less likely to receive antifungal therapies mostly because elderly or their guardians quit antifungal therapies. APACHE II scores and 30-day mortality were higher for elderly than younger patients (31.7% vs. 16.7%, p =0.032). For elderly patients, antifungal therapy administered before microbiological documentation was the only protective factor for death, whereas absence of antifungal therapies, receipt of mechanical ventilation and APACHE II score ≥20 were independent predictors of death. Conclusions Elderly patients with candidemia had poor prognoses characterized by certain host and hospital-related risk factors and special pathogen resistance features. More awareness of the burden of this disease is required, and the absence of antifungal therapies should be avoided to improve the prognoses of elderly patients with this severe infection.


Background
Candidemia has emerged as an important nosocomial infection associated with significant morbidity and mortality [1][2][3]. It is the fourth most common nosocomial bloodstream infection in the United States and the seventh in Europe [4,5]. It prolongs hospital stays and increases the costs associated with patient management [6]. The epidemiology of candidemia has been studied extensively worldwide, and data are available from a large series of laboratory-based and population-based surveillance studies, as well as studies focusing on specific patient populations such as neonates and those with cancer, undergoing surgery, and staying in intensive care units (ICUs) patients [7][8][9][10][11].
The elderly population is large and is growing in proportion to the general hospitalized population. Because of comorbidities, aging and age-associated physiological changes, increased rates of oropharyngeal colonization with Candida species, and concomitant drug use, elderly patients are more vulnerable to Candida infections [12,13]. The features of the pathogen, risk factors and severity of candidemia in elderly patients may differ from that in younger adults and might be associated with worse prognosis [14]. Moreover, physicians may choose different antifungal management for elderly and younger patients, depending on patient' compliance, individual experiences and different guidelines [15]. However, available data for elderly patients with candidemia are limited [16,17].
Here, we hypothesized that candidemia in elderly patients would present several peculiarities in epidemiology, clinical risk factors for death and antifungal management. To characterize these peculiarities, we evaluated a database of patients with candidemia from 4 multicentre, retrospective surveillance studies coordinated by our group, and compared the above indicators in elderly versus younger patients. Furthermore, to improve the clinical management and outcome of elderly patients, we used multiple logistic analyses to identify the risk factors for death.

Study design and patient selection
We performed a multicentre retrospective study of candidemia in 4 tertiary general hospitals in Shandong, China: Qilu Hospital of Shandong University (Jinan, 2000 beds), Qianfoshan Hospital affiliated with Shandong University (Jinan, 1500 beds), Jinan Center Hospital affiliated with Shandong University (Jinan, 1000 beds), and Liaocheng People's Hospital affiliated with Taishan Medical College (Liaocheng, 2000 beds). The study was conducted from June 2008 to June 2010. We included patients ≥16 years who had hospital acquired candidemia, which was defined as at least one positive blood culture for Candida species in patients hospitalized for more than 48 hrs with clinical signs and symptoms of sepsis [18]. Elderly patients were defined as the patients 65 years of age or older, whereas patients with the age between 16 and 60 were defined as younger patients. All episodes of candidemia were identified via the laboratory computer system. For patients with multiple candidemic episodes, only the first episode was included. Patients with candidemia caused simultaneously by different species of Candida were excluded. The study was approved by the Ethics Committee of Qilu Hospital of Shandong University and was carried out according to the ethical standards set forth in the Declaration of Helsinki of 1964.

Data collection
Trained study team members collected demographic and clinical data by chart review. Demographic and microbiological data, underlying diseases, predisposing factors, laboratory data, concurrent infections, antifungal agents exposure and outcome were recorded on standardized case report forms. Candida species were identified by use of the VITEK-32 system (BioMerieux Vitek, St. Louis, MO, USA). We used the recently updated species-specific antifungal drug susceptibility thresholds for fluconazole, voriconazole and caspofungin by the Clinical and Laboratory Standards Institute (CLSI) [19][20][21]. Isolates of C. krusei were considered intrinsically resistant to fluconazole. For voriconazole, isolates of C. glabrata and C. krusei isolates with minimal inhibitory concentration (MIC) of ≥2 μg/ml were considered resistant. Predisposing factors (including immunosuppressive drugs and severe hypoalbuminemia) and underlying diseases (including solid organ tumour, hematologic malignancy, neutropenia, diabetes mellitus and chronic renal failure) were recorded as host-related risk factors [22]. Other predisposing factors, including surgeries, abdominal surgical operations, ICU stay ≥5 days, multiple blood product transfusion, parenteral nutrition, receipt of mechanical ventilation, placement of central venous catheter (CVC) or urethral catheter, and prior antibacterial exposure were recorded as hospital-related risk factors [22]. All predisposing factors had to occur within 30 days before the onset of candidemia. CVC-related candidemia required the isolation of the same Candida spp. from both blood and catheter tip.
The severity of the initial presentation of candidemia was assessed by the Acute Physiology and Chronic Health Evaluation (APACHE) II score [23]. Because of the specific purpose of the study, we additionally reported an age-adjusted APACHE II score in which points attributed to older age were subtracted from the total score. We recorded the antifungal agents used for more than 2 days within 2 weeks before and 2 weeks after the microbiological documentation of candidemia.
Neutropenia was defined as absolute neutrophil count <0.5 × 10 9 /l. Severe hypoalbuminemia was defined as serum albumin levels <23 g/L. Immunosuppressive drugs received included glucocorticoids (≥20 mg/day of prednisone or equivalent doses of other corticosteroids for >1 week), chemotherapy drugs or other immunosuppressive agents. Multiple blood product transfusion was defined as transfusion of ≥6 units consisting of at least 2 units of erythrocytes and 2 units of fresh frozen plasma. Septic shock was defined as systolic blood pressure <90 mmHg, diastolic blood pressure <60 mmHg, or fluid/inotrope required to maintain blood pressure above these levels. Concurrent bloodstream infections were defined as an isolation of a positive culture for bacterium in patients with signs or symptoms of infection that occurred within 2 weeks of the onset of candidemia.

Statistical analyses
The data for categorical variables are expressed as percentages and continuous variables as mean ± SD or median with inter-quartile range (IQR). Chi-square or Fisher's exact test (two-tailed) was used to compare categorical variables and unpaired Student's t test or Mann-Whitney U test to compare continuous variables. Multivariable, backwards, stepwise, logistic regression analyses was used to identify independent risk factors associated with day-30 mortality of elderly patients with candidemia. Variables with p ≤0.10 on univariate analyses were entered into the multivariable model. Statistical analyses involved use of SPSS v15.0 for Windows (SPSS Inc., Chicago, IL, USA). A p <0.05 was considered statistically significant.

Antifungal agent exposure
The 2 age groups did not differ in the selection of antifungal agents before or after the identification of Candida species  (all p >0.05) (Table 3). Overall, elderly patients were less often administered triazoles (57.1% vs. 78.6%; p =0.005) and were more often lacking antifungal therapies (25.4% vs. 11.9%; p =0.034) compared with younger patients ( Table 3). The possible reasons for the absence of antifungal therapy in the elderly were determined from the medical records: 11 patients or their guardians decided to quit antifungal therapy, mainly because of hopelessness in the recovery, high hospitalization expenses, or non-coverage of antifungal agents by medical insurance; 2 patients died before microbiologicy documentation of Candida in the bloodstream; and 3 patients with undetermined reasons. On univariate analyses, compared with non-survivors, survivors showed higher, although not significantly proportion of prophylactic and empiric treatment (all p >0.10) ( Table 5); the rate of CVC removal was higher for survivors than non-survivors (p <0.10) and was included in the multivariable regression analyses. On multivariable analyses, antifungal therapy administered before microbiological documentation was the only protective factor for death (odds ratio [OR]0.8, 95% confidence interval [95% CI] 0.7-0.9, p =0.046). Independent predictors of death were absence of antifungal therapies (OR 2.1; 95% CI 1.2-23.8; p =0.039), receipt of mechanical ventilation (OR 3.5; 95% CI 1.6-12.4; p =0.042) and APACHE II score ≥20 (OR 4.0; 95% CI 2.5-8.6; p =0.018).

Discussion
A major strength of this study is that our data are representative of 4 centres and included patients from different hospital settings in China. The resistance to fluconazole and vorizonazole of non-C. albicans species in elderly patients being double that in younger patients should be addressed when applying empirical or prophylactic antifungal therapy. As well, risk factors related to host characteristics (e.g., underlying solid tumour, diabetes mellitus and chronic renal failure) and hospital exposure (e.g., prior ICU stay, mechanical ventilation, central venous catheters and urethral catheter placement) were identified more commonly in elderly patients than younger patients. Thirdly, as a independent risk factor for death in elderly, the absence of antifungal therapies was more common than younger patients, mostly because the patients or their guardians decided to quit antifungal therapy. The rapidly growing elderly population has specific physiological characteristics, which makes it susceptible to colonization and subsequent infection due to Candida species [24]. Not surprisingly, in this study, the prevalence of host-related risk factors, including solid tumour, diabetes mellitus and chronic renal failure, was greater for elderly than younger patients with candidemia. Cancer was a frequent underlying disease in both younger and older patients, but younger patients incurred a higher incidence of haematological malignancies, whereas elderly patients were more likely to present solid tumours, which is consistent with a previous report [25].
Another main feature in the diagnosis of candidemia is the evaluation of hospital-related risk factors [26]. Luzzati et al. [27] reported that candidemia in elderly patients was strongly associated with duration of total and peripheral parenteral nutrition, other central vascular catheters and glycopeptide antibiotics. Here, we identified more healthcare-related factors, including prior ICU stay, mechanical ventilation and urethral catheter placement in elderly patients. To decrease the incidence of candidemia in elderly patients, methods aimed to reduce unnecessary medical procedures should be encouraged whenever feasible.
The elderly patients are particularly susceptible to various infections and exposed to numerous antibiotic and antifungal treatments. Exposure to antibiotics and antifungal agents induces antifungal resistance and is an important cause of increased azoles-resistant Candida isolates [28][29][30], which might explain our finding of high resistance to azoles of non-C. albicans species in elderly patients. Non-C. albicans species such as C. glabrata, C. parapsilosis and C. tropicalis are especially vulnerable to acquiring resistance after a period of exposure to antifungal agent [31]. The emergence of fluconazole resistance in C. parapsilosis occurred after more than 10 years of fluconazole prophylaxis, which suggests that the use of fluconazole prophylaxis contributed to the emergence of C. parapsilosis with decreased susceptibility among the isolates responsible for bloodstream infections [32]. C. glabrata may be intermediately resistant to all azoles, and about 20% of strains develop resistance during therapy and prophylaxis with fluconazole [13]; previous fluconazole use is a significant risk factor for health care-associated fluconazole-resistant C. glabrata [33]. Considering the low resistance to caspofungin in the Candida isolates in elderly patients, caspofungin might be a better alternative to treat candidemia in this population.
In this study, elderly patients were more likely to experience septic shock and have poor outcomes than younger patients. The difference in outcome may be explained by higher severity of illness, as evidenced by increased APACHE II score, a widely recognized scoring system used to evaluate the severity of illness in critically ill patients [23]. The APACHE II score is calculated from 12 routine physiological measurements, including age, mean arterial pressure, heart rate, creatinine level, arterial pH, serum potassium and sodium levels, hematocrit value, Glasgow Coma Scale. However, after calculating an ageadjusted APACHE II score, we found no significant difference in illness severity between elderly and younger groups. Thus, advanced age rather than other indicators involved in the APACHE II score was the crucial factor determining the worse prognosis of elderly patients. The elderly patients were more likely to not have antifungal treatment than younger patients, and in agreement with the previous report [34], the absence of antifungal agents was independently associated with worse prognosis in elderly. Therefore, prompt initiation of early antifungal therapy is warranted in high-risk elderly patients. Elderly patients or their guardians were likely to quit therapy when the patient's conditions worsened, mainly because of lack of hope for the patient's recovery, high hospitalization expenses, or non-coverage of antifungal agents by medical insurance. The poor care for elderly patients from family members and society, as well as the lag in the medical insurance industry, are challenges to the health of geriatric populations in China [35]. To decrease the high mortality of elderly patients with candidemia, the absence of antifungal therapy should be avoided by taking measures to correct the above causes.
One of main points regarding candidemia is that delaying antifungal treatment significantly increases mortality [36]. Early treatment strategies, including prophylactic and empiric treatment is beneficial for patients with candidemia and decrease mortality [34,37]. Here, early antifungal treatment administered before microbiology documentation was a protective factor for death and could improve the outcome of elderly patients. However, we found no role for prophylactic and empiric treatment in the outcome of candidemia episodes: prophylactic and empiric treatment was used more often, although not significantly, for survivors than non-survivors, mostly because of the small number of older patients. A larger investigation with more cases is warranted to disclose the potential role of different treatment regimens on outcome in older population. Our observations have several limitations. First the retrospective design of the study limited our ability to obtain exact variables such as prior antifungal exposures, which could be important for the emergence of antifungal resistance of non-C. albicans species. Second, compared to results from previous reports [38][39][40], the removal rate in the elderly patients was relatively low (43.8%), so the real proportion of CVC-related candidemia as well as the effect of CVC removal on prognosis might have been underestimated. Considering that CVC retention has a negative impact on outcome in patients with candidemia [38][39][40], the awareness of the risk of CVC retention needs to be strengthened in the management of candidemia in Chinese hospitals. Third, we considered only the presence or absence of risk factor exposure, not the duration of exposure. Because the study was not designed to quantify the length of exposure, this variable was not available for analyses, and its associated bias could not be determined. Furthermore, some of our conclusions may not be generalizable to other countries because of differences in antifungal usage and the epidemiology of candidemia. Further studies are necessary in different geographical areas.

Conclusion
In conclusion, the elderly patients account for a substantial proportion of patients with candidemia and have higher mortality than younger patients. Such patients are characterized by certain host and hospital-related risk factors as well as special pathogen resistance features. More awareness of the burden of this disease is required and the absence of antifungal therapies should be avoided to improve the prognoses of elderly patients with this severe infection.

Consent
Written informed consent was obtained from the patient for the publication of this report and any accompanying images.