Evaluation of the effect of antibiotics used during parenteral nutrition treatment on Candidemia

SUMMARY

OBJECTIVE:

Parenteral nutrition is an important risk factor for candidemia. In this risk analysis study, the effect of previous antibiotic administration apart from the length of hospital stay, duration of Parenteral nutrition treatment, and Candida score parameters on developing candidemia was evaluated in the non-neutropenic patients receiving Parenteral nutrition treatment.

METHODS:

In this double center, retrospective, and cross-sectional study, the data of patients who received Parenteral nutrition treatment were collected. Patients with or without candidemia after the initiation of Parenteral nutrition treatment were compared in terms of demographic features, Candida score, length of hospital stay, duration of Parenteral nutrition treatment, and previous use of antibiotics. Then, predictor factors affecting the probability of candidemia during Candida growth time were determined by the Cox regression analysis.

RESULTS:

A total of 148 patients (59.5% males) were included and 16 (10.81%) of these had candidemia after initiation of parenteral nutrition treatment. The median (min–max) duration of parenteral nutrition treatment was 11 (4–72) days and the Candida growth time was 13 (7–29) days. Statistically significant differences were found between patients with or without candidemia groups in terms of length of hospital stay (p<0.001), duration of parenteral nutrition treatment (p<0.001), and Candida score (p<0.001). To determine the effect of these variables and antibiotics on candidemia, length of hospital stay [Hazard Ratio 1.030; p=0.021] and piperacillin–tazobactam (Hazard Ratio 5.626; p=0.030) were found significant and independent risk factors on the development of candidemia.

CONCLUSION:

There are some well-known risk factors including length of hospital stay, duration of Parenteral nutrition treatment, and Candida score; the potential impact of piperacillin–tazobactam administration should also be considered since they may be effective on the development of candidemia.

KEYWORDS:
Parenteral nutrition; Candidemia; Risk factors; Antibiotic; Piperacillin-Tazobactam

INTRODUCTION

Hospital-acquired Candida and bloodstream infections (BSI) represent approximately 9% of all nosocomial BSI¹. In a multicentered, point prevalence study conducted on the sepsis cases with causative agents in the intensive care unit (ICU), the rate of Candida was determined to be 4.7%². While the candidemia-related mortality rate was 83%, invasive candidiasis was found to be an independent risk factor for mortality³. Parenteral nutrition (PN) is consistently identified as an independent risk factor for candidemia in both neutropenic and non-neutropenic patients. The mortality rate of Candida catheter-related BSI in patients on PN treatment was found 30%⁴. Potential mechanisms that may be responsible for the increase in the candidemia risk include intestinal mucosal atrophy and subsequent translocation of microorganisms or endotoxins, hyperglycemia, nutrient-rich components supporting bacterial and fungal growth, and indwelling parenteral access devices⁵. According to a retrospective, case–control study, both hospital and ICU length of hospital stay (LOS) are time-dependent risk factors for candidemia (p<0.001)⁶. PN exposure time is a risk factor for the development of candidemia, especially in critically ill patients. According to Chow et al., PN duration was a significant risk factor for the development of candidemia in all patients with or without Candida albicans (p<0.01)⁴. PN is lifesaving when it is needed but besides the benefit of PN, determining the risk of developing candidemia in patients and initiation of the most appropriate antifungal treatment at the most propitious time can minimize the occurrence of PN-related candidemia⁵. PN treatment indications, safe administration techniques, and duration are reported in both the American Society for Parenteral and Enteral Nutrition (ASPEN) and European Society for Clinical Nutrition and Metabolism (ESPEN) guidelines⁷,⁸. León et al. developed a bedside “Candida score” to decide early antifungal treatment in non-neutropenic critically ill patients with Candida colonization. The “Candida score” for a cutoff value of 2.5 points were as follows for deciding early antifungal treatment: PN, surgery, multifocal colonization (1 one point each), and severe sepsis (2 two points)⁹.

In this risk analysis study, the effect of antibiotic administration apart from LOS, duration of PN treatment, and Candida score (i.e., clinical sepsis, PN administration, surgery, and multifocal colonization) parameters on developing candidemia was evaluated in the non-neutropenic patients receiving PN treatment.

METHODS

Patients

In this double-center, retrospective, and cross-sectional study, the data of patients who received PN treatment between January 2019 and December 2019 were collected. Patients aged 18 years or older, who were non-neutropenic (neutropenia: neutrophil count <0.1×10⁹ cells/L) and did not receive chemotherapy during PN treatment, who had available culture test results, and who did not have candidemia before PN treatment were included. PN treatment was evaluated by the clinical nutrition team, and only the patients who indicated PN treatment were able to receive the PN treatment. Multi-chamber bag PN was administered to all patients in this study (OliClinomel® N4-550E; Baxter Healthcare Corporation). The patients with and without candidemia after PN were compared in terms of demographic features (i.e., age, gender, and admitted department), LOS, duration of PN treatment, leukocyte and platelet counts, and concomitantly administration of antibiotics.

The time from the start of PN treatment until the development of candidemia was determined as the Candida growth time. According to blood culture tests, Candida species was reported as C. albicans for all study patients by the microbiology laboratory. The patients who were prescribed antibiotics before PN treatment were excluded from this study. Patients who were prescribed at least one antibiotic and only their antibiotic treatments that have been initiated after the initiation of PN treatment and before the development of candidemia were included in this study. The duration of antibiotic treatment is at least 7 days for all study patients. Also, the patients who received fluconazole prophylaxis before the development of candidemia were excluded from this study. This study protocol was approved by the Çukurova University Ethics Committee (Decision No. 2020/56-105).

Statistical analysis

Chi-square test, Fisher's exact test, Student's t-test, Mann–Whitney U test, Poisson regression, and Cox regression analysis, whichever appropriate, were performed. Our collective data met the criteria of a Cox distribution (multivariate model), and the appropriate model had been adjusted to determine the independent predictors of the patient outcome. According to the literature and clinical experience, we created two different models, namely, a model with Candida score, LOS, and duration of PN treatment and another model with narrow and broad-spectrum antibiotics administered during PN treatment. For all tests, p<0.05 was considered statistically significant. According to the Omnibus test in Cox regression analysis, having all the independent variables in our example models, we have p-values for first (−2 Log-Likelihood: 51.84; p=0.025) and second models (-2 Log-Likelihood: 38.80; p=0.017), indicating statistically significant overall model. IBM SPSS Statistics 23.0 software was used to analyze and evaluate the data. Since the medical literature does not contain similar studies, the sample size could not be calculated. However, at the end of this study, the power analysis result was determined as 98.06% (G*Power 3.1 Statistical Power Analysis).

RESULTS

Patients’ characteristics

A total of 148 patients [88 (59.5%) males] with the mean (standard deviation, SD) age of 63.92±18.85 years were included. Half of the patients were admitted in ICU (50.0%) wards. The median (min–max) LOS was 22 days (5–206 days). The most commonly prescribed antibiotics in these patients were cephalosporins (n=60) (Table 1).

Table 1
Distribution of demographic characteristics and antibiotic treatments of the study population (n=148).

Gender, male, n (%)	88 (59.5)
Age, mean±SD	63.92±18.85
Leukocyte, median (min–max)	10170 (1300–37000)
Platelet, median (min–max)	242000 (18000–821000)
Number of patients in ICU, n (%)	74 (50.0)
LOS, median (min–max) days	22 (5–206)
Duration of PN treatment, median (min–max) days	11 (4–72)
Candida growth time, median (min–max) days	13 (7–29)
Candida score, median (min–max)	2 (1–5)
Candidemia, n (%)	16 (10.8)
Use of antibiotics (narrow spectrum), n (%)	71 (48.0)
Use of antibiotics (broad spectrum), n (%)	131 (88.5)
Antibiotic combinations (2 or more), n (%)	91 (61.5)
Carbapenems	59 (39.9)
Tigecycline	27 (18.2)
Piperacillin–tazobactam	22 (14.9)
Fluoroquinolones	15 (10.1)
Macrolides	5 (3.4)
Cephalosporins	50 (40.3)
Cefazolin	12 (8.1)
Glycopeptides	20 (13.5)
Aminoglycosides	15 (10.1)
Metronidazole	29 (19.6)
Colistin	8 (5.4)

• SD: standard deviation; ICU: intensive care unit; LOS: length of hospital stay; PN: parenteral nutrition.

It was found that 16 (10.81%) patients had candidemia diagnosis after the initiation of PN treatment. In these patients, the median (min–max) duration of PN treatment was 11 (4–72) days, and the median (min–max) duration of Candida growth time after the initiation of PN treatment was 13 (7–29) days. In addition, antibiotic polypharmacy (2 or more) was determined in 91 (61.5%) patients (Table 2).

Table 2
Distribution of demographic characteristics and antibiotic treatments by candidemia.

	Candidemia (+) group (n=16)	Candidemia (-) group (n=132)	p-value
Gender, male, n (%)	9 (56.2)	79 (59.8)	0.78
Age, mean±SD	62.94±13.92	64.04±19.40	0.53
Leukocyte, median (min–max)	10170 (4800–17100)	10150 (1300–37000)	0.42
Platelet, median (min–max)	247500 (113000–821000)	241000 (18000–734000)	0.62
LOS, median (min–max) days	52 (17–206)	21 (5–127)	<0.001*
Duration of PN treatment, median (min–max) days	21 (9–69)	10.5 (4–72)	<0.001*
Candida growth time, median (min–max) days	13 (7–29)	0	–
Candida score, median (min–max)	3 (1–5)	2 (1–4)	<0.001*
Use of antibiotics(narrow spectrum), n (%)	8 (50.0)	66 (51.5)	0.24
Use of antibiotics(broad spectrum), n (%)	16 (100)	120 (90.9)	0.21
Antibiotic combinations(2 or more), n (%)	9 (56.3)	82 (62.1)	0.85
Carbapenems	9 (56.2)	50 (37.8)	0.16
Tigecycline	5 (31.2)	22 (16.6)	0.15
Piperacillin–tazobactam	4 (25.0)	18 (13.6)	0.22
Fluoroquinolones	0 (0)	15 (11.3)	0.15
Macrolides	1 (6.2)	4 (3.0)	0.54
Cephalosporins	4 (25.0)	56 (42.4)	0.16
Cefazolin	0 (0)	12 (9.0)	0.09
Glycopeptides	3 (18.7)	17 (12.8)	0.53
Aminoglycosides	0 (0)	15 (11.3)	0.32
Metronidazole	1 (6.2)	28 (21.2)	0.27
Colistin	2 (12.5)	6 (4.5)	0.45

• ^*
  p<0.05. SD: standard deviation; LOS: length of hospital stay; PN: parenteral nutrition.

Correlation and regression analyses

Statistically significant differences were found between the with and without candidemia groups in terms of LOS (p<0.001), duration of PN treatment (p<0.001), and Candida score (p<0.001) (Table 2). Thirty (20.3%) patients were identified as high risk (≥3 points) according to the Candida score. There was a significant relationship between the patients with high risk according to the Candida score and the candidemia diagnosis (p<0.001). In addition, according to the Poisson regression analysis, the Candida score was a significant and independent predictor of the candidemia diagnosis (p<0.001). For every extra one point in Candida score, 1.169 (95%CI 1.110–1.231) times more candidemia was diagnosed (16.9% higher risk). However, the cutoff value (high risk as three or more points) for Candida score was not a significant predictor of the candidemia diagnosis (p=0.224).

A Cox regression was run to predict whether a patient treating with PN has a diagnosis of candidemia based on the Candida score, LOS, duration of PN treatment, narrow and broad-spectrum antibiotics (i.e.., carbapenems, tigecycline, piperacillin–tazobactam, cephalosporins, glycopeptides, and colistin) prescribed after the initiation of PN treatment. Since the number of patients who were prescribed fluoroquinolones, macrolides, cefazolin, aminoglycosides, and metronidazole in the candidemia (+) group was <2, these drugs were not included in the Cox regression analysis to ensure the validity of our model (Table 2). For every extra day in LOS, 1.030 (95%CI 1.004–1.057) times more candidemia risk was determined (p=0.021). Also, the hazard ratio (HR) for piperacillin–tazobactam (HR=5.626) indicates that patients who prescribed piperacillin–tazobactam treatment had a higher risk of candidemia than patients who do not (p=0.030). In contrast, the duration of PN treatment, Candida score, and other antibiotic treatments were not significant risk factors (p>0.05) (Table 3).

Table 3
Cox proportional hazard regression analysis.

	B	SE	Wald	Df	p-value	Hazard ratio(HR)	95%CI for HR
Length of hospital stay*	0.030	0.013	5.306	1	0.021	1.030	1.004–1.057
Duration of PN treatment	0.002	0.082	8.171	1	0.364	1.002	0.998–1.006
Candida score	-0.975	0.639	2.332	1	0.127	0.377	0.108–1.318
Carbapenems	1.214	0.902	1.812	1	0.178	3.366	0.575–9.706
Tigecycline	1.414	0.903	2.453	1	0.117	4.112	0.701–14.123
Piperacillin–tazobactam*	1.727	0.795	4.724	1	0.030	5.626	1.185–16.712
Cephalosporins	-0.180	1.264	0.020	1	0.887	0.835	0.070–9.941
Glycopeptides	1.775	1.034	2.946	1	0.086	5.902	0.777–14.800
Colistin	-0.186	1.013	0.034	1	0.854	0.830	0.114–6.043

• Dependent variable: control group=0, study group = 1. Bold values indicate statistically significant variables. PN: parenteral nutrition; CI: confidence interval.
• ^*
  p<0.05.

DISCUSSION

Candidemia was found in 10.81% of the patients who received PN in this study, which appears higher than the findings of (2–6%) other studies in the literature¹⁰,¹¹. However, since all the patients who received PN treatment in the hospital were not included, our result is not reflecting the actual incidence due to the design of this study. There were significant differences between groups with and without candidemia in terms of LOS, duration of PN treatment, and Candida score. According to the Cox regression analysis, LOS and piperacillin–tazobactam (broad-spectrum antibiotic) administration were found to be independent risk factors for increased development of candidemia. However, the duration of PN treatment was not found to be a prediction risk factor in our model.

Although LOS and piperacillin–tazobactam administration were found as risk factors in this study, it is known that the factors affecting the development of candidemia may differ between patient groups depending on the chosen study design. Recent guidelines recommend the use of risk prediction tools to facilitate earlier recognition and initiation of antifungal treatment¹²,¹³. According to the study that used “Candida score,” surgery, multifocal colonization, PN treatment, and severe sepsis were independent predictors of proven Candida infection⁹. In contrast, the sensitivity of this scoring system may differ in each patient group¹³. In this study, a significant relationship was found between Candida score and candidemia using the Poisson regression analysis. However, the cutoff points (high risk) for the score were not found as a significant predictor. This result means that the cutoff value of the Candida score should be reconsidered, especially in patients receiving PN treatment.

While broad-spectrum antibiotic was presented as a risk factor in some studies¹⁰, the others were declared that the risk factor differs between the antibiotic groups¹¹,¹⁴. In a case-control study, the previous use of antibiotics (OR 2.61; p=0.03) was found as an independent risk factor for the development of candidemia¹⁵. However, according to a case-comparator and 10-year study, piperacillin–tazobactam was not independently associated with BSI due to non-C. albicans species¹⁴. In contrast, in this study, the administration of piperacillin–tazobactam was determined as a significant and independent risk factor for the development of candidemia due to C. albicans. This result stated that the increasing use of broad-spectrum antibiotics, such as piperacillin–tazobactam, is an important risk factor especially for patients susceptible to candidemia due to PN treatment. Beyond our results, the influence of the presence of vancomycin and metronidazole with the risk of candidemia was reported¹¹.

According to a retrospective study, candidemia was developed after an average of 17.2 (5–74) days of administration of PN treatment. A similar result was determined in this study similar to 13 (7–29) days of the PN treatment¹⁰. Moreover, Luzzatti et al. reported that the elderly patients who received PN treatment had a significantly higher risk of candidemia even on the seventh day of PN treatment¹¹. The fact that the mean age of our patient population (63.92 years) was close to the elderly people may explain the shorter Candida growth time in this study.

The LOS, duration of PN treatment, and Candida score were higher in patients with candidemia compared with patients without candidemia. Similar to our results, higher LOS and duration of PN treatment were also reported in patients who developed candidemia¹¹. In another study whose primary outcome was the relationship between time-dependent risk factors and candidemia, LOS was greater in patients with candidemia versus control patients (36 days versus 13 days; p<0.001)⁶. In this study, with a similar result, statistically significant differences were found between the patients with and without candidemia groups in terms of LOS (52 days versus 21 days; p<0.001).

According to the study by Tsai et al., thrombocytopenia was associated with survival time in non-neutropenic patients requiring PN treatment after the onset of candidemia (p=0.006)¹⁶. In contrast, in this study, thrombocytopenia (<150000 cells/μL) was not determined, and it was not associated with the development of candidemia (p=0.62).

Particularly, since this is a retrospective study, it harbors some conspicuous limitations. Limited study period, small sample size, and lack of duration of antibiotic administration to evaluate the effect on the development of candidemia were the other limitations of this study.

CONCLUSIONS

This is the first study showing that piperacillin–tazobactam, a broad-spectrum antibiotic, is a significant and independent risk factor for candidemia in non-neutropenic patients receiving PN treatment. There are some well-known risk factors including the longer hospitalization and duration of PN treatment and higher Candida risk score; piperacillin–tazobactam treatment should also be considered as an important risk factor. Our data suggest that randomized controlled studies should be performed to evaluate the effectiveness of candidemia prevention by restricting the use of broad-spectrum antibiotics, such as piperacillin–tazobactam, and by managing PN treatment consistent with the current guidelines.

• Funding: none.

REFERENCES

• ¹
  Wisplinghoff H, Bischoff T, Tallent SM, Seifert H, Wenzel RP, Edmond MB. Nosocomial bloodstream infections in US hospitals: analysis of 24,179 cases from a prospective nationwide surveillance study. Clin Infect Dis. 2004;39(3):309-17. https://doi.org/10.1086/421946
  » https://doi.org/10.1086/421946
• ²
  Baykara N, Akalin H, Arslantas MK, Hanci V, Caglayan C, Kahveci F, et al. Epidemiology of sepsis in intensive care units in Turkey: a multicenter, point-prevalence study. Crit Care. 2018;22(1):93. https://doi.org/10.1186/s13054-018-2013-1
  » https://doi.org/10.1186/s13054-018-2013-1
• ³
  Tukenmez Tigen E, Bilgin H, Perk Gurun H, Dogru A, Ozben B, Cerikcioglu N, et al. Risk factors, characteristics, and outcomes of candidemia in an adult intensive care unit in Turkey. Am J Infect Control. 2017;45(6):e61-3. https://doi.org/10.1016/j.ajic.2017.02.022
  » https://doi.org/10.1016/j.ajic.2017.02.022
• ⁴
  Chow JK, Golan Y, Ruthazer R, Karchmer AW, Carmeli Y, Lichtenberg DA, et al. Risk factors for albicans and non-albicans candidemia in the intensive care unit. Crit Care Med. 2008;36(7):1993-8. https://doi.org/10.1097/CCM.0b013e31816fc4cd
  » https://doi.org/10.1097/CCM.0b013e31816fc4cd
• ⁵
  Stratman RC, Martin CA, Rapp RP, Berger R, Magnuson B. Candidemia incidence in recipients of parenteral nutrition. Nutr Clin Pract. 2010;25(3):282-9. https://doi.org/10.1177/0884533610368704
  » https://doi.org/10.1177/0884533610368704
• ⁶
  Aljeboori Z, Gorelik A, Jenkins E, McFarlane T, Darvall J. Risk factors for candidaemia and their cumulative effect over time in a cohort of critically ill, non-neutropenic patients. Crit Care Resusc. 2018;20(4):313-9. PMID: 30482139
• ⁷
  McClave SA, Taylor BE, Martindale RG, Warren MM, Johnson DR, Braunschweig C, et al. Guidelines for the provision and assessment of nutrition support therapy in the adult critically ill patient: Society of Critical Care Medicine (SCCM) and American Society for Parenteral and Enteral Nutrition (A.S.P.E.N.). JPEN J Parenter Enteral Nutr. 2009;33(3):277-316. https://doi.org/10.1177/0148607109335234
  » https://doi.org/10.1177/0148607109335234
• ⁸
  Cederholm T, Barazzoni R, Austin P, Ballmer P, Biolo G, Bischoff SC, et al. ESPEN guidelines on definitions and terminology of clinical nutrition. Clin Nutr. 2017;36(1):49-64. https://doi.org/10.1016/j.clnu.2016.09.004
  » https://doi.org/10.1016/j.clnu.2016.09.004
• ⁹
  Leon C, Ruiz-Santana S, Saavedra P, Almirante B, Nolla-Salas J, Alvarez-Lerma F, et al. A bedside scoring system (“Candida score”) for early antifungal treatment in nonneutropenic critically ill patients with Candida colonization. Crit Care Med. 2006;34(3):730-7. https://doi.org/10.1097/01.CCM.0000202208.37364.7D
  » https://doi.org/10.1097/01.CCM.0000202208.37364.7D
• ¹⁰
  Stratov I, Gottlieb T, Bradbury R, O’Kane GM. Candidaemia in an Australian teaching hospital: relationship to central line and TPN use. J Infect. 1998;36(2):203-7. https://doi.org/10.1016/s0163-4453(98)80014-5
  » https://doi.org/10.1016/s0163-4453(98)80014-5
• ¹¹
  Luzzati R, Cavinato S, Giangreco M, Grana G, Centonze S, Deiana ML, et al. Peripheral and total parenteral nutrition as the strongest risk factors for nosocomial candidemia in elderly patients: a matched case-control study. Mycoses. 2013;56(6):664-71. https://doi.org/10.1111/myc.1209012
  » https://doi.org/10.1111/myc.1209012
• ¹²
  Lamoth F, Lockhart SR, Berkow EL, Calandra T. Changes in the epidemiological landscape of invasive candidiasis. J Antimicrob Chemother. 2018;73(Suppl. 1):i4-13. https://doi.org/10.1093/jac/dkx444
  » https://doi.org/10.1093/jac/dkx444
• ¹³
  Laine ME, Flannery AH, Moody B, Thompson Bastin ML. Need for expanded Candida Score for empiric antifungal use in medically critically ill patients? Crit Care. 2019;23(1):242. https://doi.org/10.1186/s13054-019-2525-3
  » https://doi.org/10.1186/s13054-019-2525-3
• ¹⁴
  Chow JK, Golan Y, Ruthazer R, Karchmer AW, Carmeli Y, Lichtenberg D, et al. Factors associated with candidemia caused by non-albicans Candida species versus Candida albicans in the intensive care unit. Clin Infect Dis. 2008;46(8):1206-13. https://doi.org/10.1086/529435
  » https://doi.org/10.1086/529435
• ¹⁵
  Yapar N, Pullukcu H, Avkan-Oguz V, Sayin-Kutlu S, Ertugrul B, Sacar S, et al. Evaluation of species distribution and risk factors of candidemia: a multicenter case-control study. Med Mycol. 2011;49(1):26-31. https://doi.org/10.3109/13693786.2010.501344
  » https://doi.org/10.3109/13693786.2010.501344
• ¹⁶
  Tsai CC, Lay CJ, Wang CL, Lin ML, Yang SP. Prognostic factors of candidemia among nonneutropenic adults with total parenteral nutrition. J Microbiol Immunol Infect. 2011;44(6):461-6. https://doi.org/10.1016/j.jmii.2011.04.002
  » https://doi.org/10.1016/j.jmii.2011.04.002

Publication Dates

• Publication in this collection
  26 Nov 2021
• Date of issue
  Oct 2021

History

• Received
  12 July 2021
• Accepted
  14 Aug 2021