Impact of QuickFISH in addition to antimicrobial stewardship on vancomycin use and resource utilization in cancer patients with coagulase-negative staphylococcal blood cultures

Abstract

Objective

To evaluate the impact of rapidly identifying coagulase-negative staphylococci (CoNS) from positive blood cultures (BC) combined with an established antimicrobial stewardship (AS) program at a tertiary cancer center.

Methods

Cancer patients >=18 years with >=1 positive CoNS BC identified by Staphylococcus QuickFISH®, a peptide nucleic acid fluorescence in situ hybridization assay, between 01/1/13 and 12/31/13 (QuickFISH) were compared to cancer patients >=18 years with CoNS identified via standard microbiologic techniques between 01/01/11 and 12/31/11 (baseline). Positive BC results were reported to the clinician by microbiology staff; restricted antibiotics (e.g., vancomycin) required approval by the AS team.

Results

There were 196 baseline and 103 QuickFISH patients. Faster median time-to-organism identification (33 (IQR 27–46) vs 49 (IQR 39–63) hours, P<0.001), more vancomycin avoidance (51/103 (50%) vs 60/196 (31%), P=0.002), shorter median antibiotic duration (1 (IQR 0–3) vs 2 (IQR 0–6) days, P=0.019), fewer central venous catheter (CVC) removals (14/78 (18%) vs 57/160 (36%), P=0.004), and reduced vancomycin level monitoring (16/52 (31%) vs 71/136 (52%), P=0.009) were observed in the QuickFISH group. QuickFISH implementation was predictive of being less likely to be prescribed antibiotic therapy (OR 0.35, 95% CI 0.20–0.62, P<0.001). Prior transplant (RR 1.47, 95% CI 1.13–1.92, P=0.004), neutropenia (RR 1.47, 95% CI 1.09–1.99, P=0.012), multiple positive BC (RR 4.23, 95% CI 3.23–5.54, P<0.001), and CVC (RR 1.60, 95% CI 1.02–2.53, P=0.043) were independent factors for antibiotic duration.

Conclusions

QuickFISH implementation plus AS support leads to greater vancomycin avoidance and improved resource utilization in cancer patients with CoNS blood cultures.

INTRODUCTION

Coagulase-negative staphylococci (CoNS) are the most frequently isolated organisms from blood in both neutropenic and non-neutropenic cancer patients [1]. The presence of a central venous catheter (CVC) is an important risk factor [1, 2]. Because CoNS are part of normal skin flora, these organisms are also common contaminants in clinical specimens. In studies assessing the significance of CoNS recovered from blood cultures in the general patient population, the majority are deemed contaminants with only 12–25% constituting true bacteremia [3, 4].

Despite established criteria for bloodstream infection (BSI) due to a skin commensal [5], prescribers may overtreat since BSI remain a significant cause of morbidity and mortality in immunocompromised patients. At our oncology center, we found that 14% received unnecessary vancomycin courses and that unneeded treatment of CoNS blood culture contaminants led to an excess expenditure of $5,246 per episode [6].

Standard microbiologic techniques for identification of organisms implicated in BSI are based on phenotypic methods that require 48–72 hours to provide final results although some rapid phenotypic methods (e.g., tube coagulase test) may provide information within a few hours from the time that a blood culture turns positive [7]. Peptide nucleic acid fluorescence in situ hybridization (PNA FISH) assays are commercially available to quickly identify select pathogens directly from positive blood culture bottles [8]. A faster and less labor-intensive Staphylococcus QuickFISH® test (AdvanDx, Woburn, MA) has been on hand to differentiate between Staphylococcus aureus and CoNS since 2013 with a turnaround time <30 minutes and excellent sensitivity and specificity results that are commensurate to those seen with the original PNA FISH kits [9].

Rapid detection of BSI combined with delivery of that information to healthcare providers can lead to optimization of antimicrobial therapy in general patient populations [10–12]. In one retrospective study, use of PNA FISH to distinguish between S. aureus and CoNS combined with an antimicrobial stewardship program (ASP) led to a significant reduction in median hospital length-of-stay (LOS) by two days and a trend towards less intravenous (IV) vancomycin usage [10]. The exclusion of patients with cancer from this study however makes it difficult to draw meaningful conclusions about the utility of PNA FISH in special patient populations. Similarly, there are no data on the role of the second-generation PNA FISH assay in oncology patients.

There has been a longstanding system of prompt reporting of blood culture results to clinicians who then page the ASP for guidance and antibiotic approval at our institution. Our clinical microbiology laboratory adopted QuickFISH in January 2013. We aimed to see whether earlier identification of CoNS in adult cancer patients further influenced vancomycin initiation and duration of therapy.

METHODS

Study setting and patients

This historical cohort study was reviewed and approved by the Institutional Review Board at Memorial Sloan Kettering Cancer Center (MSKCC), a 471-bed tertiary care cancer center in New York, NY. All patients 18 years and older with one or more positive blood cultures for gram-positive cocci in clusters (GPCC) and had CoNS identified by routine microbiologic methods between January 1, 2011 and December 31, 2011 comprised the baseline group. The post-implementation group (QuickFISH) consisted of patients 18 years and older with one or more positive blood cultures for CoNS identified by Staphylococcus QuickFISH® between January 1, 2013 and December 31, 2013. Patients were excluded when vancomycin was started prior to the identification of GPCC from blood cultures, the QuickFISH assay was not performed, or there were incomplete records. Medical records were abstracted for demographics, cancer diagnosis, symptoms and signs temporal to blood culture collection, complete blood counts, blood culture specifications (e.g., collection date and time, date and time of GPCC positivity, date and time of GPCC identification), number of positive blood cultures for CoNS, presence of CVC and its management, other utilized resources relevant to CoNS management (vancomycin levels, echocardiogram), antimicrobial use, hospital LOS, and inhospital mortality.

Laboratory methods

During baseline, all positive blood cultures with Gram stain showing GPCC were immediately called and verbally reported as a critical value to the ordering clinician. The positive blood culture was sub-cultured following standard protocols and once enough growth was detected (24–48 hours), identification was made based on biochemical (catalase and coagulase) reactions. Following identification of CoNS and S. aureus, antimicrobial susceptibility testing was performed by microbroth dilutions, using an automated platform (Dried Gram positive panels, Microscan Walkaway plus System, Beckman Coulter, Jersey City, NJ) with results available after 24 hours.

With QuickFISH implementation, the identification of CoNS and S. aureus was performed immediately on all positive blood cultures with a Gram stain showing GPCC. For patients with multiple positive blood cultures for GPCC in a 24-hour period, QuickFISH was performed only on the first positive culture. Only results of CoNS and/or S. aureus (not GPCC) were verbally reported to the ordering clinician. Susceptibility testing was performed as described in the baseline period.

Result management

MSKCC utilizes a system of prior approval for initiation of vancomycin and other restricted anti-infectives 7 days a week between the hours of 9 am and 10 pm. After 10 pm, therapy can be initiated pending review and approval the following morning by antimicrobial stewardship (AS) personnel. When primary teams are notified of positive blood culture results by clinical microbiology, they contact the hospital’s ASP for antibiotic approval. The ASP verifies microbiologic results by accessing the patient’s electronic medical record or calling microbiology directly. The ASP also flags patients who are initiated on therapy for prospective audit & feedback unless the patient is followed by the Infectious Disease consultation team.

Definitions

Laboratory-confirmed BSI due to CoNS was defined according to established criteria by the Centers for Disease Control & Prevention (CDC) [5]. A contaminant was considered if one of at least two sets of blood cultures was positive for CoNS and/or there was absence of supporting symptoms (e.g., fever, chills, hypotension). Time-to-organism identification was defined as time from blood culture collection to organism identification.

Outcomes

Primary outcomes were initiation of therapy in which only antibiotics with coverage against resistant GPCC (e.g., vancomycin) were considered as therapy for CoNS and duration of therapy. A day of therapy was counted when a patient received at least one dose of an antibiotic directed against GPCC on that day [13]. Secondary outcomes included time-to-organism identification, CVC removal, echocardiogram performed, vancomycin levels, median hospital LOS, and in-hospital mortality.

Data analysis

Descriptive comparisons between years were analyzed with Wilcoxon rank sum test for continuous variables and Fisher exact test for categorical variables. Logistic regression was used to determine which factors were associated with a patient being put on anti-infective therapy for CoNS. Poisson models were used to determine which factors were associated with the number of days a patient was on anti-infective therapy for CoNS. Factors that were significant in the univariate analysis were tested in the multivariate analysis and backward selection was used to determine the final logistic and Poisson models. All statistical tests were two sided, and P values less than 0.05 were considered statistically significant. Statistical analyses were performed in SAS 9.4 (SAS Institute, Inc, Cary, NC).

RESULTS

Overall, there were a total of 684 patients with one or more positive blood culture for CoNS of which 385 patients were excluded (Figure 1). Of 299 evaluable patients, 223 (75%) were assessed to have contaminant blood cultures.

Figure 1.

Patient flow chart

Comparison across groups

Demographic and other patient characteristics are shown in Table 1. There were 196 baseline patients and 103 in the QuickFISH group. In contrast to baseline, there were significantly more men, patients with a hematologic malignancy, and neutropenic patients in the QuickFISH group (Table 1). The time-to-organism identification was significantly shortened to a median of 33 hours (interquartile range (IQR), 27–46) in the QuickFISH group compared to a median of 49 hours (IQR 39–63) in the baseline group (P< 0.001) (Table 1). While not statistically significant, median hospital LOS was 3 days shorter in the QuickFISH group (12 (IQR 5–27) vs 15 (IQR 7–29) days, P=0.19).

Table 1.

Demographic and other characteristics of patients with one or more positive blood cultures for coagulase-negative staphylococci (CoNS)

	Baseline (N=196)	QuickFISH (N=103)	Total (N=299)	P value
Patient characteristics
Median age (range)	61 (21–93)	62 (22–84)	61 (20–93)	0.88
Male, N (%)	91 (46)	64 (62)	155 (52)	0.011
Cancer diagnosis, N (%)
- Hematologic malignancy	63 (32)	55 (53)	118 (40)	0.001
- Solid tumor	132 (67)	48 (47)	180 (60)
- None	1 (1)	0 (0)	1 (0)
Prior HSCT, N (%)	40 (20)	28 (27)	68 (23)	0.19
Presence of CVC, N (%)	160 (82)	78 (76)	238 (80)	0.23
Neutropenia at blood culture collection, N (%)	18 (9)	20 (19)	38 (13)	0.017
Microbiologic characteristics
Single positive blood culture, N (%)	129 (66)	77 (75)	206 (69)	0.12
Median time-to-organism identification, hours (IQR)	49 (39–63)	33 (27–46)	44.6 (33–61)	<0.001
Management of blood culture results
Clinical impression of contaminant result, N (%)	141 (72)	82 (80)	223 (75)	0.16
CVC removal, N (%)^	57 (36)	14 (18)	71 (31)	0.004
Echocardiogram, N (%)	7 (4)	3 (3)	10 (3)	0.99
No vancomycin doses given, N (%)	60 (31)	51 (50)	111 (37)	0.002
Median antibiotic duration, days (IQR)	2 (0–6)	1 (0–3)	1 (0–5)	0.019
Vancomycin level monitoring, N (%)*	71 (52)	16 (31)	87 (46)	0.009
Clinical outcomes
Median hospital duration, days (IQR)	15 (7–29)	12 (5–27)	14 (6–29)	0.19
In-hospital mortality, N (%)	22 (11)	15 (15)	14 (6–29)	0.46

Abbreviations: CVC, central venous catheter; HSCT, hematopoietic stem cell transplantation; IQR, interquartile range; N, number

^{^}The denominator consisted of patients who had a CVC.

^*The denominator consisted of patients who received vancomycin therapy.

Management of positive blood cultures

Although the proportion of patients who had a CVC was similar between the two groups, there were significantly fewer CVC removals in the QuickFISH group (14/78, 18%) compared to baseline (57/160, 36%) (P=0.004). There were no differences in the proportion of patients who underwent echocardiogram to rule out valvular involvement (baseline 7/196 (4%) vs QuickFISH 3/103 (3%), P=0.99). No doses of vancomycin were given in 51 of 103 (50%) patients in the QuickFISH group compared to 60 of 196 (31%) baseline patients (P=0.002). Antibiotic duration was shortened to a median of 1 day (IQR 0–3) in the QuickFISH group versus a median of 2 days (IQR 0–6) in the baseline group (P=0.019). There was also less vancomycin level monitoring in the QuickFISH group (16/52 (31%) vs 71/136 (52%), P=0.009).

We then analyzed factors to examine their association with antibiotic administration via logistic regression. QuickFISH implementation and increasing age were predictive of being less likely to be prescribed antibiotic therapy, whereas male sex, prior HSCT, neutropenia, multiple positive blood cultures, and presence of CVC were predictive of being more likely to be prescribed antibiotic therapy on univariate analysis (Table 2). QuickFISH implementation, age, male sex, neutropenia, and multiple positive blood cultures were independently associated with prescription of antibiotic therapy in multivariate analysis (Table 2). We also analyzed factors to assess association with duration of antibiotic therapy via Poisson regression. Age, hematologic malignancy, prior HSCT, neutropenia, multiple positive blood cultures, presence of CVC, and echocardiogram were significantly associated with antibiotic duration on univariate analysis (Table 3). HSCT, neutropenia, multiple positive blood cultures, and presence of CVC were independently associated with duration of therapy in multivariate analysis (Table 3).

Table 2.

Univariate and multivariate logistic regression for predictors of antibiotic administration

Variable	Univariate OR (95% CI)	P	Multivariate OR (95% CI)	P
Period
Baseline	Ref		Ref
QuickFISH	0.45 (0.28–0.74)	0.001	0.35 (0.20–0.62)	<0.001
Age	0.81* (0.96–0.99)	0.015	0.82* (0.96–0.99)	0.040
Male	1.84 (1.14–2.96)	0.012	2.26 (1.31–3.87)	0.003
Cancer diagnosis
Solid tumor	Ref
Hematologic malignancy	1.32 (0.81–2.15)	0.264
Prior HSCT	2.06 (1.12–3.78)	0.020
Neutropenia	2.93 (1.25–6.91)	0.014	2.76 (1.06–7.18)	0.037
Number of positive blood cultures
Lone	Ref		Ref
Multiple	5.58 (2.93–10.66)	<0.001	5.58 (2.84–10.94)	<0.001
Presence of central venous catheter	2.42 (1.37–4.28)	0.002

Abbreviations: CI, confidence interval; HSCT, hematopoietic stem cell transplantation; OR, odds ratio

^*OR is per a 10 unit increase.

Table 3.

Univariate and multivariate Poisson regression for predictors of antibiotic duration

Variable	Univariate RR (95% CI)	P	Multivariate RR (95% CI)	P
Period
Baseline	Ref
QuickFISH	0.82 (0.59–1.15)	0.256
Age	0.99 (0.98–1.00)	0.040
Male	1.27 (0.93–1.73)	0.136
Cancer diagnosis
Solid tumor	Ref
Hematologic malignancy	1.70 (1.26–2.30)	<0.001
Prior HSCT	1.70 (1.24–2.35)	0.001	1.47 (1.13–1.92)	0.004
Neutropenia	1.98 (1.38–2.84)	<0.001	1.47 (1.09–1.99)	0.012
Number of positive blood cultures
Lone	Ref		Ref
Multiple	4.71 (3.59–6.19)	<0.001	4.23 (3.23–5.54)	<0.001
Presence of central venous catheter	2.77 (1.62–4.73)	<0.001	1.60 (1.02–2.53)	0.043
Echocardiogram performed	4.55 (2.98–6.96)	<0.001

Abbreviations: CI, confidence interval; HSCT, hematopoietic stem cell transplantation; RR, rate ratio

DISCUSSION

We describe the impact of QuickFISH in the management of adult cancer patients with CoNS recovered from blood. This second-generation assay has demonstrated excellent sensitivity and specificity for the detection of both S. aureus and CoNS. The ease and speed of this test has the potential to improve therapeutic intervention [9, 14]. Following QuickFISH implementation, we found significantly faster median time-to-organism identification, a higher proportion of patients with vancomycin avoidance, shorter median antibiotic duration, fewer CVC removals, and less vancomycin level monitoring.

Earlier organism identification is comparable to other reports employing PNA FISH technology in the management of enterococcal bacteremia or candidemia [15, 16]. A proactive approach however is needed to take advantage of quickly recognizing the organism and has been demonstrated in prior studies of the PNA FISH assay distinguishing between S. aureus and CoNS [10–12]. Otherwise, no benefit with respect to antibiotic optimization or other resource utilization is seen [17]. Depending on available personnel and resources, implementation strategies should be customized to the institution when considering introduction of rapid diagnostics.

One successful strategy has relied on microbiology staff promptly reporting PNA FISH results to providers [11, 12, 18]. In a prospective, randomized, controlled trial, Ly and colleagues demonstrated a shortened median antibiotic duration for CoNS from 2.5 to 0 days (P<0.01), suggesting that enhanced communication between the laboratory and clinicians likely facilitated reduction in unnecessary antibiotic use for false-positive blood cultures [11]. Another tactic is to report PNA FISH results directly to the institutional ASP, which then determines the need for vancomycin. Even though there were only 87 evaluable patients who had CoNS blood cultures, Forrest et al. reported a trend for greater number of patients in the PNA FISH group who did not receive vancomycin compared to the control group (PNA FISH: 17% vs control: 9%, P=0.06) and a trend for reduced vancomycin utilization (PNA FISH: 2.6 defined daily dose (DDD)/patient vs control: 4.8 DDD/patient, P=0.06) [10].

In examining our well-established system of real-time notification of positive blood culture results to primary teams and ASP oversight of restricted antibiotics, approximately one-third of patients avoided vancomycin altogether for contaminant blood cultures, and the median antibiotic duration was 2 days in the baseline group. The addition of QuickFISH to this system further increased vancomycin avoidance from 31% to 50% and shortened median antibiotic duration from 2 to 1 day. The reduction in vancomycin level monitoring in the QuickFISH group probably occurred as a result of increased vancomycin avoidance. Another finding was the significant reduction in CVC removals in the QuickFISH group. Retaining a well-functioning CVC is a boon for the cancer patient because management is contingent on stable venous access for chemotherapy, blood transfusion, drug administration, fluid resuscitation, and clinical monitoring [19]. Cancer patients have many mitigating factors that contribute to duration of hospitalization. Nevertheless, although not significant, median hospital LOS was shortened by 3 days in the QuickFISH group. Prior studies in non-cancer patients have similarly found reductions in median hospital LOS by 2 days using the PNA FISH assay [10, 11].

The value of QuickFISH is primarily in the avoidance of vancomycin initiation for CoNS blood cultures that are determined to be clinically insignificant. This factor was found on both univariate and multivariate analysis to be predictive of a patient less likely to be prescribed antibiotic therapy. Conversely, neutropenia and multiple positive blood cultures were predictors for being prescribed antibiotic therapy. After organism identification and the decision to treat or not are made, clinical factors such as prior HSCT, neutropenia, multiple positive blood cultures, and presence of CVC influence antibiotic duration. Although we use the CDC criteria for BSI definitions, the unique cancer population in our study might lead to differences in blood culture interpretation for some patients. For example, treatment of a single positive blood culture for CoNS may be appropriate in a febrile, neutropenic patient with a CVC [6].

We did not examine the effect of QuickFISH on attributable mortality related to inappropriate treatment of true bacteremia due to CoNS. However, CoNS bacteremia is known to be associated with lower mortality compared to patients with BSI caused by other pathogens, even in febrile, neutropenic patients [20]. Additionally, inappropriate empiric therapy may not be necessarily associated with CoNS bacteremia-related mortality [21]. Besides QuickFISH, there are other available rapid diagnostic technologies such as matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry and multiplexed nucleic acid amplification tests (NAAT) that when integrated with AS intervention have also been useful in the management of CoNS blood cultures [22, 23]. These options allow laboratories to select the method that works best for their individual workflow and antimicrobial stewardship policies.

There are limited data on the role of rapid diagnostics and antimicrobial stewardship in immunocompromised hosts [24]. Our study is the first to describe the impact of the QuickFISH assay in a cancer patient population that included patients with solid tumors or hematologic malignancies, as well as HSCT recipients. While the QuickFISH group had more patients with hematologic malignancies and/or were neutropenic, we still found a significant reduction in vancomycin and other resource utilization.

In summary, QuickFISH implementation combined with active antimicrobial stewardship support led to greater avoidance of vancomycin, shorter antibiotic duration, reduced vancomycin level monitoring, and less CVC removals in cancer patients with positive blood cultures due to CoNS. Additional studies examining the impact of rapid diagnostics in cancer patients should be encouraged.