Comparison of Pharmacist-Directed Management of Multiplex PCR Blood Culture Results with Conventional Microbiology Methods on Effective and Optimal Therapy within a Community Hospital

Multiplex PCR combined with a pharmacist-driven reporting protocol was compared to the standard of care within a community hospital to evaluate initial changes after notification of a positive blood culture. The intervention group demonstrated decreased times to changes in antimicrobial therapy (P = 0.0081), increased changes to optimal antimicrobial therapy (P = 0.013), and decreased vancomycin use for coagulase-negative staphylococcus contaminants (P < 0.01) with multiplex PCR implementation and pharmacist intervention.

ABSTRACT

Multiplex PCR combined with a pharmacist-driven reporting protocol was compared to the standard of care within a community hospital to evaluate initial changes after notification of a positive blood culture. The intervention group demonstrated decreased times to changes in antimicrobial therapy (P = 0.0081), increased changes to optimal antimicrobial therapy (P = 0.013), and decreased vancomycin use for coagulase-negative staphylococcus contaminants (P < 0.01) with multiplex PCR implementation and pharmacist intervention.

INTRODUCTION

It is critical to quickly identify pathogens and to provide effective antimicrobial therapy for bloodstream infections, to limit morbidity and deaths (1–3). Rapid optimization of therapy should limit the use of broad-spectrum antimicrobials and maximize clinical cures (4). Conventional microorganism identification typically takes 48 to 72 h. However, recent advancements in diagnostic technology have led to the availability of rapid microorganism identification from blood cultures. Early studies, primarily performed in academic centers, involving molecular rapid diagnostic tests (mRDTs) demonstrated improved outcomes, especially in conjunction with antimicrobial stewardship programs (ASPs) (2, 5). We report findings regarding effective and optimal therapies with the use of mRDTs with a pharmacist-driven model within a community hospital, compared to standard microbiology reporting methodology.

St. Joseph’s Hospital, an anchor hospital in St. Joseph’s/Candler Health System, is a 330-bed community hospital in Savannah, Georgia. All admitted nonpregnant adult patients who had a positive blood culture with an organism that was identifiable by multiplex PCR (BioFire FilmArray blood culture identification panel), with results reported between 7:00 a.m. and 3:30 p.m. on Monday through Friday, were included. Patients were excluded if they were receiving palliative or hospice care, had been transferred from an outside hospital with an already positive blood culture, or had positive blood culture results reported after discharge or death. The St. Joseph’s/Candler Health System institutional review board approved this investigation.

This single-center, retrospective, before-and-after study was conducted over two 12-month periods. The control group consisted of patients with positive blood cultures between December 2014 and November 2015, who were treated under the conventional microorganism identification protocol, in which the initial Gram stain results were communicated to a nurse, who then notified the provider. The intervention group consisted of patients with multiplex PCR identification in conjunction with a pharmacist-driven reporting protocol; patients were identified through records from microbiology calls received between April 2016 and March 2017. The microbiology laboratory called the pharmacist with a positive multiplex PCR result; the pharmacist notified the provider and nurse, gave recommendations, and entered accepted orders for antimicrobial changes directly into the electronic health record (Meditech 6.15). Pharmacists utilized an algorithm approved by the antimicrobial subcommittee (consisting of antimicrobial stewardship pharmacists and infectious diseases [ID] physicians) to make recommendations in response to positive multiplex PCR results (Table 1).

TABLE 1.

Antimicrobial recommendation algorithm utilized by pharmacists after reported positive multiplex PCR result

Organisma	Antibiotic(s)	No. of casesb
		Control group (n = 123)	Intervention group (n = 85)
Gram-positive organisms
Enterococcus
VSE (VanA/B negative)	Ampicillin i.v., 2 g q4h	5	6
PCN allergy	Vancomycin i.v.
VRE (VanA/B positive)	Linezolid i.v., 600 mg q12	1
Alternative	Daptomycin i.v., ≥8 mg/kg q24h
Listeria	Ampicillin i.v., 2 g q4h
PCN allergy	Sulfamethoxazole-trimethoprim i.v., 10–15 mg/kg/day divided q6–12h
Staphylococcus
MSSA (MecA negative)	Cefazolin i.v., 2 g q8h	6	5
Alternative	Nafcillin i.v., 2 g q4h
PCN allergy	Vancomycin i.v.
MRSA (MecA positive)	Vancomycin i.v.	11	6
CoNS		58c	41
MSCoNS (MecA negative)			10
1/2 BC sets	Cefazolin, 2 g q8h, or consider discontinuing antibiotics
2/2 BC sets	Cefazolin, 2 g q8h, or consider discontinuing antibiotics
MRCoNS (MecA positive)			31
1/2 BC sets	Vancomycin i.v. or consider discontinuing antibiotics
2/2 BC sets	Vancomycin i.v.
Streptococcus agalactiae (group B)	Penicillin G i.v., 3 million units q4h	3	2
Alternative	Ceftriaxone i.v., 1–2 g q24h
PCN allergy	Vancomycin i.v.
Streptococcus pyogenes (group A)	Penicillin G i.v., 3 million units q4h	1
Alternative	Ceftriaxone i.v., 1–2 g q24h
PCN allergy	Vancomycin i.v.
Streptococcus pneumoniae/Streptococcus mitis		5	2
CNS source	Ceftriaxone i.v., 2 g q12h, + vancomycin i.v.
Non-CNS source	Ceftriaxone i.v., 1–2 g q24h
PCN allergy	Levofloxacin i.v., 500 mg q24h, for lung source or vancomycin otherwise
Streptococcus species	Ceftriaxone i.v., 1–2 g q24h	8	3
PCN allergy	Vancomycin i.v.
Gram-negative organisms
Acinetobacter baumannii	Piperacillin-tazobactam i.v., 4.5 g q8h, + tobramycin i.v.		1
Alternative	Meropenem i.v., 2 g q8h, + tobramycin i.v.
Haemophilus influenzae	Ceftriaxone i.v., 1–2 g q24h	2
Alternative	Ampicillin-sulbactam i.v., 3 g q6h
PCN allergy	Levofloxacin i.v., 750 mg q24h
Neisseria meningitidis	Ceftriaxone i.v., 2 g q24h
Alternative	Penicillin G i.v., 4 million units q4h
PCN allergy	Levofloxacin i.v., 750 mg q24h
Pseudomonas aeruginosa	Piperacillin-tazobactam i.v., 4.5 g q8h, + tobramycin i.v.	3	2
PCN allergy	Meropenem i.v., 2 g q8h, + tobramycin i.v.
Enterobacteriaceae
Enterobacter cloacae complex	Cefepime i.v., 2 g q8h or 2 g q12h for non-CNS		5
PCN allergy	Meropenem i.v., 1 g q8h
Escherichiacoli	Piperacillin-tazobactam i.v., 3.375 g q8h	9	8
PCN allergy	Meropenem i.v., 1 g q8h
Klebsiella oxytoca	Cefepime i.v., 2 g q8h
	Meropenem i.v., 1 g q8h
Klebsiella pneumoniae	Meropenem i.v., 1 g q8h	6	2
Proteus species	Ceftriaxone i.v., 1–2 g q24h	4	2
PCN allergy	Meropenem i.v., 1 g q8h
Serratia marcescens	Ceftriaxone i.v., 1–2 g q24h	1
PCN allergy	Meropenem i.v., 1 g q8h
Enterobacteriaceae species	Cefepime i.v., 2 g q8h
PCN allergy	Meropenem i.v., 1 g q8h
KPC resistance gene positive	Consider ID consult

^aBC, blood culture; VSE, vancomycin-susceptible enterococcus; i.v., intravenously; q, every; PCN, penicillin; VRE, vancomycin-resistant enterococcus; MSSA, methicillin-sensitive Staphylococcus aureus; MRSA, methicillin-resistant Staphylococcus aureus; MSCoNS, methicillin-sensitive CoNS; MRCoNS, methicillin-resistant CoNS; CNS, central nervous system; KPC, Klebsiella pneumoniae carbapenemase.

^bValues indicate the total number of organisms identified, because some patients grew multiple organisms.

^cUnable to determine whether the case was MecA positive.

The following data were collected: age, gender, drug allergies, comorbidities, all blood and other-source cultures and susceptibilities from that admission, and all antimicrobials, with the time of initiation and/or discontinuation, from that admission. Progress notes for each patient were read to determine whether antimicrobials were being used for anything aside from the specified cultures and to obtain additional treatment information.

The primary outcome was the time to change in antimicrobial therapy, measured from the time of the call from the microbiology laboratory to the time at which the antimicrobial change was verified in the electronic health record. If the change was made ≥24 h after the phone call or after a subsequent call, then the patient was categorized as “no change,” so that only initial changes were captured. Secondary outcomes further delineated the primary outcome as the time to change from suboptimal to optimal antimicrobial therapy or from ineffective to effective antimicrobial therapy. Optimal antimicrobial therapy was defined by the treatment algorithm described previously. Effective antimicrobial therapy was defined as the organism being susceptible to the antimicrobial regimen prescribed but with further modification being necessary to obtain optimal classification. Other secondary outcomes included the number of patients changed to optimal or effective antimicrobial therapy, the presence of effective antimicrobial therapy, and vancomycin use for coagulase-negative staphylococcus (CoNS)-contaminated cultures.

Descriptive statistics were calculated to characterize patients in the two groups and the study outcomes. Nonparametric statistics were used to examine differences between the groups, due to the nonnormal distributions of the outcome variables. Bivariate analyses, including chi-square, Fisher’s exact, and Wilcoxon rank-sum tests, were conducted to examine differences in outcomes between the two groups. Outcome variables were dichotomized at the median scores. Multivariate logistic regression analyses were used to examine differences in outcomes between the two groups, with adjustment for potentially confounding variables. Stata MP 13.1 was used to analyze the data.

A total of 118 patients were identified for the control group and 77 for the intervention group, based on inclusion and exclusion criteria (Table 2). The intervention group demonstrated decreased median time-to-change values for effective therapy, as well as increased numbers of patients changed to optimal therapy (Table 3). Multivariate logistic regression analyses indicated that the intervention group was less likely to have a greater time-to-change value (P < 0.01) and more likely to be changed to optimal therapy (P < 0.01), in comparison to the control group, after accounting for differences between the groups. Changes to optimal therapy were split between escalation and de-escalation. The most common changes in the intervention group included the addition of vancomycin for methicillin-resistant Staphylococcus aureus and the discontinuation of vancomycin when it was determined to be clinically unnecessary (Table 4). Five patients, all in the control group, continued to receive ineffective antimicrobial therapy after the initial microbiology call; all others ultimately received either effective or optimal therapy, with or without a change. Vancomycin use for CoNS contaminants decreased considerably in the intervention group (p < 0.01).

TABLE 2.

Demographic characteristics

Characteristica	Control group (n = 118)	Intervention group (n = 77)	P
Age (mean ± SD) (y)	68.78 ± 15.26	62.69 ± 16.01	0.0082b
Gender (no. [%])			0.70
Male	58 (49.15)	40 (51.95)
Female	60 (50.85)	37 (48.05)
CCI (mean ± SD)	5.67 ± 2.58	4.62 ± 2.77	0.0078b

^aSD, standard deviation; CCI, Charlson comorbidity index.

^bBy t test.

TABLE 3.

Results

Outcome	Control group	Intervention group	P	Multivariate logistic regression analysisa
				OR (95% CI)	P
Patients with change in therapyb
Time to change (median) (min)	160	50	0.0081	0.28 (0.10–0.77)	0.014
Time to optimal therapy (median) (min)	178	76.5	0.083	0.19 (0.22–1.55)	0.12
Time to effective therapy (median) (min)	152	50	0.015	Outcome predicted perfectly
No. (%) changed to optimal therapy	7 (15.6)	12 (41.4)	0.013	4.28 (1.38–13.31)	0.012
No. (%) changed to effective therapy	11 (24.4)	5 (17.2)	0.462	0.53 (0.15–1.83)	0.318
Patients with CoNS contaminantsc
No. (%) with vancomycin use for CoNS contaminants after call	27 (69.23)	3 (10)	<0.01

^aOR, odds ratio; CI, confidence interval.

^bControl group, n = 45; intervention group, n = 29.

^cCoNS contaminants all had 1/4 blood culture sets positive or 1/2 blood culture sets positive with adjudication by an ID pharmacist or ID physician. Control group, n = 39; intervention group, n = 30.

TABLE 4.

Optimal therapy interventions

Change to optimal therapy in intervention group (n = 12)	Organism(s)a
Escalation (n = 5)
No treatment to vancomycin	MRSA (n = 2), CoNS (n = 2)
No treatment to cefazolin	MSSA (n = 1)
De-escalation (n = 7)
Discontinuation of vancomycin	CoNS contaminant (n = 2),b Gram-negative organism (n = 2), Streptococcus species (n = 1)
Piperacillin-tazobactam to cefepime	KPC-negative Enterobacter (n = 1)
Meropenem to cefazolin	MSSA (n = 1)

^aMRSA, methicillin-resistant Staphylococcus aureus; MSSA, methicillin-sensitive Staphylococcus aureus; KPC, Klebsiella pneumoniae carbapenemase.

^bCoNS contaminants all had 1/4 blood culture sets positive or 1/2 blood culture sets positive with adjudication by an ID pharmacist or ID physician.

The majority of studies demonstrating benefits with mRDT and ASPs have been conducted in academic medical centers, with significantly more resources than community hospitals (2–5). One study found that mRDT in the absence of an ASP did not improve clinical outcomes, demonstrating the necessity of ASP resources (6). Our study focused on combining multiplex PCR results with pharmacist-driven management within a community hospital that employed 10 clinical pharmacists, 7 of whom participated in this study. Four clinical pharmacists providing recommendations had no formal specialty ID postgraduate training but had access to formally trained ID pharmacists if needed. Additionally, non-ID pharmacy residents were involved in the communications and recommendations, supervised directly by the clinical pharmacist. The outcome data were encouraging, as this study best represents most clinical practice settings in the United States where mRDT and ASPs could potentially be implemented. Our study design also differs from that of many published studies, because we evaluated only immediate changes after identification, rather than evaluating the final regimen. Limitations of our study included communication issues, prescriber unfamiliarity with the new technology, documentation inconsistencies, and the implementation of two interventions at once.

In conclusion, integration of multiplex PCR with pharmacist-directed management of blood culture results resulted in shorter times to effective therapy and increased numbers of patients receiving optimal therapy after initial interventions within a community hospital. More studies are needed to further delineate the role of mRDT and pharmacists in these settings.