Impact of an antifungal stewardship intervention on optimization of candidemia management

Abstract

Background:

Candidemia represents a leading cause of healthcare-associated bloodstream infections with significant morbidity and mortality. Previous studies have demonstrated that comprehensive care bundles improve candidemia management but are time-consuming.

Objective:

To determine the impact of a one-time targeted candidemia intervention on time to initiation of adequate therapy compared to standard of care.

Methods:

This Institutional Review Board (IRB)-approved, quasi-experiment evaluated a targeted candidemia intervention involving a single phone call to the primary team providing recommendations for care. Daily follow-up was provided by the infectious diseases (ID) consult service. Two time periods were evaluated: pre-intervention (01 August 2012 to 31 July 2014) and post-intervention (01 October 2014 to 30 September 2016). The primary endpoint was time to adequate antifungal therapy (TTx) in the business hours (6 a.m. to 6 p.m. Monday through Friday) population (BHP). Secondary endpoints were TTx in the total population as well as infection-related length of stay (IF-LOS) and compliance with quality indicators (composite endpoint: ophthalmology (OPH) consult, repeat cultures, and ⩾14 days of adequate therapy).

Results:

In all, 117 patients were included (pre-intervention = 50, post-intervention = 67, BHP = 51). TTx decreased from 2 h 57 m to 2 h 12 m (p = 0.094) in the BHP and 3 h 30 m to 2 h 9 m (p = 0.021) in the total population. There was no difference in IF-LOS (p = 0.797), compliance with quality indicators (p = 0.343), or in-hospital mortality (p = 0.761). Post-intervention, there were more ID and OPH consults (p < 0.001).

Conclusions:

Our one-time candidemia intervention did not statistically decrease time to adequate therapy in the BHP, but did in the total population. No differences were found for other clinical outcomes, except increases in ID and OPH consults. Further studies are needed to examine whether a one-time intervention is non-inferior to a more comprehensive care bundle.

Background

Antimicrobial stewardship (AMS) has received much attention recently because of its focus on improving patient care and health outcomes. The Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America recommend that AMS programs implement interventions to improve antimicrobial use and clinical outcomes that target patients with specific infectious diseases (ID) syndromes.¹ Because candidemia represents a leading cause of healthcare-associated bloodstream infections and carries an attributable mortality of 29–60%, it has been a target for AMS at many institutions.^2–5 Early initiation of effective antifungal therapy has been shown to improve clinical outcomes including in-hospital mortality and 30-day mortality.^6–9 Until rapid diagnostic strategies for fungal pathogens are widely utilized, the major target for antifungal stewardship programs should encompass improving time to initiation of effective antifungal therapy. This is accomplished through timely notification, antifungal order placement, and antifungal administration.

Previous studies have demonstrated that comprehensive care bundles require daily follow-up to improve the optimization of candidemia management.^10,11 With AMS programs being pressed for time and resources, it is important to be efficient while still achieving optimal outcomes. In 2014, our institution implemented an antifungal stewardship intervention, consisting of a single phone call by the AMS pharmacist with follow-up by the ID consult service. The purpose of this study was to determine the impact of the implemented intervention on time to initiation of adequate treatment and other clinical outcomes.

Methods

This study was a quasi-experimental, Institutional Review Board–approved study with a pre-test and post-test design conducted at an urban, tertiary academic medical center. Prior to the implementation of the antifungal stewardship intervention, the standard of care was that a nurse or member of the primary team would be notified of a positive yeast Gram stain on a blood culture with subsequent therapeutic management left to the discretion of the primary team. In August 2014, antifungal susceptibility testing along with an antifungal stewardship candidemia intervention was implemented to improve the management of patients with candidemia. This intervention consisted of real-time culture alerts (electronic notification as cultures turned positive for yeast) to the AMS pharmacist using Theradoc (Premier, Inc., Charlotte, NC, USA), which was added to the standard phone call from the microbiology lab to the nurse or member of the primary team of the positive yeast Gram stain from a blood culture. This pharmacist would then perform a comprehensive review of the patient’s chart and contact the primary team, if needed, to provide recommendations regarding initiation of appropriate antifungal therapy with an echinocandin, ID consultation, and, if needed, other candidemia management interventions [removal of central lines, ophthalmology (OPH) consult, and repeated blood cultures]. All recommendations were voluntarily implemented at the discretion of the primary team. This intervention did not require daily follow-up from the AMS pharmacist as further recommendations were expected to be done by the ID consult service. The AMS pharmacist intervened around the clock; however, this occurred most consistently immediately after notification during business hours Monday through Friday from 6 a.m. to 6 p.m.

Patients were eligible for inclusion if they were at least 18 years of age and had a positive blood culture with Candida albicans, C. glabrata, C. parapsilosis, C. tropicalis, or C. krusei during one of the study periods: pre-intervention from 1 August 2012 to 31 July 2014 or post-intervention from 1 October 2014 to 30 September 2016. Only the first positive blood culture per patient per admission, not including relapses within 30 days, was included. Patients were excluded if they received a systemic antifungal agent other than an echinocandin or fluconazole, the Candida isolate was resistant to both echinocandins and fluconazole, or they died, were transitioned to hospice, or were transferred to an outside hospital within 24 h of the positive yeast Gram stain.

Adequate antifungal therapy was defined as an agent with documented or expected in vitro susceptibility to the pathogen. In the pre-intervention group, this included micafungin 100 mg daily for all Candida spp. or fluconazole ⩾ 400 mg daily (adjusted for end-organ damage) for all Candida spp., except C. glabrata and C. krusei. Fluconazole ⩾ 400 mg daily was also considered adequate in patients with C. glabrata if it was empirically started and the patient clinically improved with negative follow-up cultures. In the post-intervention group, adequate antifungal therapy was micafungin 100 mg daily unless resistant to echinocandins; fluconazole ⩾ 400 mg daily (adjusted for end-organ damage) for C. albicans, C. parapsilosis, and C. tropicalis with a fluconazole minimum inhibitory concentration (MIC) ⩽ 4 mg/liter; fluconazole ⩾ 400 mg daily (adjusted for end-organ damage) for C. glabrata with a fluconazole MIC ⩽ 8 mg/liter; or fluconazole ⩾ 800 mg daily (adjusted for end-organ damage) for C. glabrata with a fluconazole MIC 16–32 mg/liter. Appropriate antifungal therapy was defined as the narrowest agent a patient could receive (accounting for contraindications including allergies, liver function tests, QTc, and drug–drug interactions). In the preintervention group, this was micafungin 100 mg daily for patients with C. glabrata or C. krusei or if patient with contraindication to fluconazole, fluconazole ⩾ 400 mg daily (adjusted for end-organ damage) in patients with C. glabrata if it was empirically started and patient clinically improved with negative follow-up cultures, and fluconazole ⩾ 400 mg daily (adjusted for end-organ damage) for C. albicans, C. parapsilosis, and C. tropicalis. In the post-intervention group, appropriate antifungal therapy was considered to be micafungin 100 mg daily for C. krusei or any fluconazole-resistant Candida isolate or if the patient had a contraindication to fluconazole; fluconazole ⩾ 400 mg daily (adjusted for end-organ damage) for C. albicans, C. parapsilosis, and C. tropicalis with a fluconazole MIC ⩽ 4 mg/liter; fluconazole ⩾ 400 mg daily (adjusted for end-organ damage) for C. glabrata with a fluconazole MIC ⩽ 8 mg/liter; or fluconazole ⩾ 800 mg daily (adjusted for end-organ damage) or micafungin 100 mg daily for C. glabrata with a fluconazole MIC 16–32 mg/liter. Time to therapy calculations were defined as time from the initial positive yeast Gram stain to medication administration documentation of the antifungal. Infection-related length of stay (IF-LOS) was defined as days from index Candida spp. blood culture draw to completion of antifungal therapy, discharge, or death, whichever occurred first.

The primary outcome was time to adequate antifungal therapy in the business hours population (BHP). Secondary outcomes included IF-LOS and compliance with quality indicators, which was a composite outcome of OPH consultation, repeat blood cultures every 24–48 h until culture clearance starting on the day of the initial positive yeast Gram stain, and receiving ⩾14 days of adequate therapy from culture clearance. Additional outcomes recorded include time to appropriate antifungal therapy in the BHP, time to adequate and appropriate antifungal therapy in the total population, time to adequate antifungal therapy order in the BHP and total population, in-hospital mortality, hospital length of stay (H-LOS), and conversion to adequate oral antifungal.

Statistical analysis

Results were reported as proportions or medians [interquartile range (IQR)]. Comparisons between the pre-intervention and post-intervention group were performed using Mann–Whitney U tests for continuous variables and Chi-square test or Fisher’s exact test for categorical variables. A p value of <0.05 was considered statistically significant. Statistical analyses were performed using SPSS (v. 24).

Results

A total of 574 total candidemia cultures were reviewed, and 117 patients met inclusion criteria (Figure 1). There was no significant difference in baseline characteristics between study groups with exception to significantly more neutropenia in the pre-intervention group (p = 0.045) and a higher alanine aminotransferase (ALT) in the pre-intervention group (p = 0.033; Table 1).

Figure 1.

Study enrollment.

Table 1.

Baseline demographics for total population.

Variable Presented as n (%) or median (IQR)	Pre (n = 50)	Post (n = 67)	p value
Age (years)	53 (38–64)	56 (43–66)	0.401
Sex (male)	30 (60)	37 (55.2)	0.605
Race [Caucasian (versus AA)]	17 (34)	32 (47.8)	0.136
Antifungal allergies (micafungin)	1 (2)	0 (0)	0.427
Height (in; n = 108; pre = 46; post = 62)	67 (63.8–71)	67 (64–71)	0.774
Actual body weight (kg; n = 116; pre = 46; post = 67)	68.7 (58.8–99.9)	80.8 (61.7–103.9)	0.313
Ideal body weight (kg; n = 108; pre = 46; post = 62)	66.1 (54.1–75.3)	62.7 (54.1–75.3)	0.928
Renal replacement therapy
HD	3 (6)	10 (14.9)	0.129
CRRT	3 (6)	3 (4.5)	1.000
None	44 (88)	54 (80.6)	0.283
Serum creatinine (mg/dL; n = 98; pre = 44; post = 54)	1.01 (0.8–1.4)	1.14 (0.7–1.5)	0.666
Neutropenia*	10 (20)	5 (7.5)	0.045
AST (units/liter; n = 108; pre = 45; post = 63)	29 (20.5–58)	27 (15–45)	0.199
ALT (units/liter; n = 108; pre = 45; post = 63)	31 (18.5–60)	21 (13–36)	0.033
QTc (ms; n = 99; pre = 39; post = 60)	440 (408–471)	453 (420–468)	0.647
Charlson comorbidity index	2 (1–3)	2 (1–4)	0.104
Pitt bacteremia score$	2 (0–4)	2 (1–4)	0.891
Presence in ICU	15 (30)	29 (43.3)	0.142
Business hours	22 (44)	29 (43.3)	0.938

AA, African American; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CRRT, continuous renal replacement therapy; HD, hemodialysis; ICU, intensive care unit; IQR, interquartile range.

^*Absolute Neutrophil Count ⩽500 ⩽48 h prior to initial positive fungal blood culture draw.

^$Highest point score within 48 h before or on the day of first positive blood culture.

The majority of patients in both groups had an intravenous (IV) catheter as the source of infection and received early source control with removal of the IV catheter (Table 2). The most commonly occurring Candida spp. were C. glabrata (34% versus 22.4%, p = 0.163) and C. albicans (28% versus 53.7%, p = 0.005), pre- and post-intervention, respectively. There were significantly more patients in the post-intervention group with ID consultations overall (36% versus 74.6%, p < 0.001). There were also significantly more ID consultations post-Gram stain and a shorter duration of time to ID consultation from Gram stain in the post-intervention group.

Table 2.

Microbiological characteristics for total population.

Variable Presented as n (%) or median (IQR)	Pre (n = 50)	Post (n = 67)	p value
Source of infection
IV catheter	30 (60)	41 (61.2)	0.896
Intra-abdominal	5 (10)	9 (13.4)	0.571
Genitourinary	4 (8)	5 (7.5)	1.000
Other	0 (0)	3 (4.5)	0.260
Unknown	11 (22)	9 (13.4)	0.223
Source control*
Early$	23 (76.7)	31 (75.6)	0.918
Late‡	6 (20)	7 (17.1)	0.753
Not performed	1 (3.3)	3 (7.3)	0.633
Cultures drawn 24–48 h until clearance	25 (50)	37 (55.2)	0.575
Isolated Candida Species
C. albicans	14 (28)	36 (53.7)	0.005
C. glabrata	17 (34)	15 (22.4)	0.163
C. parapsilosis	8 (16)	10 (14.9)	0.873
C. tropicalis	9 (18)	5 (7.5)	0.082
C. krusei	3 (6)	1 (1.5)	0.312
ID consult	18 (36)	50 (74.6)	<0.001
ID consult, post-stain	15 (30)	33 (49.3)	0.036
Time to ID consult, days (post-stain)	3 (1–8)	1 (0–1)	0.002

ID, infectious diseases; IQR, interquartile range; IV, intravenous.

^*Source control only accessed with IV catheters.

^$⩽48 h from positive Gram stain.

^‡>48 h from positive Gram stain.

There was no difference in initial antifungal administered, the percent of patients who obtained adequate or appropriate therapy, the patients who received at least 14 days of adequate antifungals from documented culture clearance, or those with proof of candidemia clearance (Table 3). There were significantly more patients in the post-intervention group (18% versus 38.8%, p = 0.015) who were converted to adequate oral antifungals.

Table 3.

Therapy characteristics and outcomes for total population.

Variable Presented as n (%) or median (IQR)	Pre (n = 50)	Post (n = 67)	p value
Therapy characteristics
Initial antifungal administered
Fluconazole	21 (42)	33 (49.3)	0.436
Micafungin	26 (52)	33 (49.3)	0.769
None	3 (6)	1 (1.5)	0.312
Adequate antifungal therapy	46 (92)	64 (95.5)	0.459
Duration of adequate therapy from culture clearance (days)*	14 (11–16)	14 (10.25–16)	0.810
Minimum of 14 days of adequate therapy from culture clearance$	21 (44.7)	36 (54.5)	0.301
Switched to adequate oral antifungal	9 (18)	26 (38.8)	0.015
Appropriate antifungal therapy	42 (84)	62 (93)	0.146
Proof of candidemia clearance	38 (76)	57 (85.1)	0.214
Outcomes
IF-LOS (days)	10 (5.75–16)	10 (5–17)	0.797
H-LOS (days)	18 (6.75–37.5)	27 (13–44)	0.070
In-hospital mortality	17 (34)	21 (31.3)	0.761
Ophthalmology consult	18 (36)	46 (68.7)	<0.001
Compliance	9 (18)	17 (25.4)	0.343

H-LOS, hospital length of stay; IF-LOS, infection-related length of stay; IQR, interquartile range.

^*Of those who received it and had culture clearance; pre = 33; post = 56.

^$Three unknown in pre-group and one in post-group.

Overall, there was a statistically significant decrease in time to an order for an adequate antifungal by 1 h 11 m in the post-intervention group compared to pre-intervention (p = 0.017), as well as receipt of adequate antifungal therapy in the post-intervention group by 1 h 29 m in the total population (p = 0.026; Table 4). There was no difference in either population with time to appropriate therapy. With regard to the secondary outcomes of IF-LOS and the composite outcome of compliance with quality indicators, there were no statistically significant differences between the two groups (Table 3). There was also no difference found in in-hospital mortality or H-LOS. There were, however, significantly more OPH consultations in the post-intervention group (35% versus 68.7%, p < 0.001).

Table 4.

Time to therapy outcomes.

Variable Presented as median (IQR)	Business hours		p value	Total population		p value
	Pre	Post		Pre	Post
Time to adequate therapy*	2 h 57 m (1 h 59 m–17 h 36 m)	2 h 12 m (0 h 0 m–4 h 27 m)	0.094	3 h 30 m (1 h 3 m–18 h 36 m)	2 h 9 m (0 h 0 m–4 h 56 m)	0.021
Time to adequate therapy order*	1 h 40 m (0 h 17 m–16 h 51 m)	0 h 34 m (0 h 0 m–2 h 13 m)	0.068	1 h 35 m (0 h 4 m–17 h 34 m)	0 h 24 m (0 h 0 m–2 h 46 m)	0.017
Time to appropriate therapy$	17 h 36 m (2 h 0 m–58 h 5 m)	27 h 55 m (1 h 3 m–70 h 42 m)	0.778	11 h 15 m (1 h 40 m–52 h 24 m)	12 h 1 m (0 h 0 m–65 h 6 m)	0.944

^*Of those who received it; business hours: pre (n = 19) and post (n = 27); total population: pre (n = 46) and post (n = 64).

^$Of those who received it; business hours: pre (n = 17) and post (n = 25); total population: pre (n = 42) and post (n = 62).

With regard to the BHP, the time to adequate antifungal therapy was not significantly different between the two groups. In addition, there was no difference in the therapy characteristics or clinical outcomes listed in Table 3, with exception to significantly more patients being converted to adequate oral antifungals (9.1% versus 34.5%, p = 0.034) and receiving OPH consultations (27.3% versus 79.3%, p < 0.001) in the post-intervention group.

Discussion

Based on the results, our one-time antifungal stewardship intervention did not significantly reduce time to an order for appropriate therapy or receipt of adequate therapy in our BHP. The intervention, which included follow-up by the ID team when consulted, was associated with almost twice as many patients receiving OPH consultations, which is recommended for all patients with candidemia.⁶ While many other factors such as documented clearance, appropriate duration, and percent of patients getting adequate and appropriate therapy seemed to trend favoring our post-intervention group, these results were not statistically significant. There were, however, more patients transitioned to oral therapy in our post-intervention group, which may highlight the role of pharmacy involvement in this intervention and on the ID consult team.

With respect to time to adequate antifungal therapy, while our results show a modest reduction in median time in our total population, our BHP post-intervention group had times similar to those seen in a study by Reed et al.¹¹ (2 h 12 m versus 1 h 18 m). In addition, we were able to see a decrease in our IQR from 1 h 59 m–17 h 36 m to 0 h 0 m–4 h 27 m in the BHP and 1 h 3 m–18 h 36 m to 0 h 0 m–4 h 56 m in the total population, which represents a larger percentage of patients getting adequate therapy in a shorter amount of time. In the BHP, our pre-intervention group had significantly better time to adequate therapy, with time to order placement of 9 h 12 m and a hang time of 13 h 30 m in the Reed study as compared to 1 h 40 m and 2 h 57 m, respectively, in our study. This large gap in pre-intervention groups between studies may account for the difference in statistically significant results. While the study by Reed et al. also involved real-time culture alerts during business hours, the antifungal intervention was more comprehensive and involved daily follow-up by AMS. They did not see a difference in clinical outcomes such as mortality, H-LOS, or IF-LOS, but they were able to show more transitions to oral therapy and OPH consultations which was similar to the results of this study.

In a similar study, Antworth et al.¹⁰ reviewed a comprehensive care bundle that also involved real-time culture alerts during business hours and daily follow-up by AMS. While they did not measure the time to adequate antifungal therapy, they did not find a difference in patients initiated on antifungals within 24 h of the positive yeast Gram stain. Similar to our study, they found no difference in H-LOS or mortality, but they did show significantly more OPH consultations. They were, however, also able to show an increase in their post-intervention group in overall compliance with their bundle which included appropriate therapy, catheter removal, repeat blood cultures every 48 h until clearance, appropriate duration of therapy, and OPH consultation (40.5% versus 78%, p = 0.0016). Their higher percentages seen in bundle compliance and the individual components as compared to our study may be explained by their exclusion of patients who died prior to possible completion of all bundle elements, which this study included.

In addition to the above, part of our intervention was a recommendation for ID consultation. Because candidemia management can be complex, ID physician involvement can play a critical role. Early ID consultation has been associated with improved outcomes in some populations of patients with candidemia and has also clinically shown benefit with the management of other infections.^12,13 We recognize that formal ID teams may not be available to all patients at all institutions, notably smaller or rural centers, but those that do should quickly and comprehensively consult ID.

It is worth noting that the Clinical Practice Guideline for the Management of Candidiasis was updated by the Infectious Diseases Society of America in February 2016.⁶ The most significant change in these guidelines was the recommendation of initiating echinocandins as empiric therapy for the majority of patients, as this has been associated with decreased mortality.^6,14 Due to our patient population and significant percentage of patients with C. glabrata, our recommendation with our antifungal stewardship intervention was to initiate echinocandins as initial therapy for all patients throughout the post-intervention period starting in August 2014. Also, there was no difference between our pre- and post-intervention groups with regard to the initial antifungal used, which may have been started prior to the AMS pharmacist intervention.

The quasi-experimental pretest-posttest design has some limitations. First, there is a lack of randomization which does not allow one to assign causality to the results. Second, regression to the mean can influence time to adequate therapy, but this was controlled using different study periods and examining both the business hours and total population. Third, maturation of physicians may play a role as there were different groups of medical residents in each group. We attempted to control for this using two full years for our study periods, allowing for a similar maturation of residents throughout their residency years.

Another limitation with this study is that in many cases, adequate antifungals were given prior to positive Gram stain results (20% and 39% in the pre- and post-intervention groups, respectively). This may have contributed to finding a statistical difference in time to adequate therapy in our total population. Also, during our study periods, there was a shift from a predominance of C. glabrata and other non-albicans species in the pre-intervention group to a predominance of C. albicans in the post-intervention group. This shift to the more virulent C. albicans may have played a role in some of the clinical outcomes such as mortality or length of stay. Although we were able to show differences in some outcomes such as OPH consultations, ID consultations, and conversion to oral antifungals, we were unable to show a difference in time to adequate therapy in our BHP or other clinical outcomes, likely due to the small study population. It is possible significant differences may have been present if more patients were included in the study. Finally, it is important to note that although OPH consults increased, data on the rate of endophthalmitis detected before and after the intervention are not available. It is possible, then, that the increased number of consults may have been a function of the shift of mycology toward C. albicans, which is often responsible for cases of endophthalmitis.

Conclusion

Our one-time candidemia intervention did not statistically decrease time to initiation of adequate therapy in the BHP at our institution. There was, however, a statistically significant decrease in time to adequate antifungal order and administration in our total population. No differences were found for other clinical outcomes, except increases in ID and OPH consults. Further studies are needed to conclusively determine whether a one-time intervention is non-inferior to a more comprehensive care bundle.