Clinical and Financial Impact of a Diagnostic Stewardship Program for Children with Suspected Central Nervous System Infection

Abstract

Objective

This study sought to investigate the optimal implementation, clinical and financial impacts of FilmArray Meningitis Encephalitis Panel (MEP) multiplex PCR testing of cerebrospinal fluid (CSF) in children with suspected (CNS) infection.

Study Design

A pre-post quasi-experimental cohort study to investigate the impact of implementing MEP using a rapid CSF diagnostic stewardship program was conducted at Children’s Hospital Colorado (CHCO). MEP was implemented with EMR indication selection to guide testing to children meeting approved use criteria: i. infants < 2mo, ii. immunocompromised, iii. encephalitis, iv. ≥5 WBCs/μL of CSF. Positive results were communicated with antimicrobial stewardship real-time decision support. All cases with CSF obtained by lumbar puncture sent to the CHCO microbiology laboratory meeting any of the four criteria above were included with pre-implementation controls (2015–2016) compared to post-implementation cases (2017–2018). Primary outcome was time-to-optimal antimicrobials compared using log-rank test with Kaplan-Meier analysis.

Results

Time-to-optimal antimicrobials decreased from 28 hours amongst 1124 pre-implementation controls to 18 hours (p<0.0001) amongst 1127 post-implementation cases (72% with MEP testing conducted). Post-implementation, time-to-positive CSF results was faster (4.8 vs. 9.6 hours, p<0.0001), IV antimicrobial duration was shorter (24 vs 36 hours, p=0.004) with infectious neurologic diagnoses more frequently identified (15% vs. 10%, p=0.03). There were no differences in time-to-effective antimicrobials, hospital admissions, antimicrobial starts or length of stay. Costs of microbiologic testing increased, but total hospital costs were unchanged.

Conclusions

Implementation of MEP with a rapid CNS diagnostic stewardship program improved antimicrobial use with faster results shortening empiric therapy. Routine MEP testing for high-yield indications enables antimicrobial optimization with unchanged overall costs.

Introduction

The diagnosis of central nervous system (CNS) infections in children is challenging due to nonspecific presenting clinical features of meningitis and encephalitis caused by a wide spectrum of viruses, bacteria, fungi, and parasites(1, 2). Due to potential severe morbidity and mortality of untreated CNS infection, many children undergo lumbar puncture (LP) with multiple pathogen-specific tests conducted. Despite this, less than half of these children have an etiologic agent identified in cerebrospinal fluid (CSF)(3). Broad empiric, often nephrotoxic, antimicrobial therapy for potential bacterial meningitis or herpes simplex virus (HSV) infection is administered while microbiologic testing is pending(4, 5). The conventional pathogen-specific diagnostic approach to suspected CNS infections is inconsistent, costly, and can lead to delayed or missed diagnoses(3).

In 2015, the FilmArray Meningitis Encephalitis Panel (MEP), a multiplex RT-PCR panel was FDA approved for testing of CSF obtained by LP. MEP permits syndromic testing using 200μL of CSF to detect nucleic acid sequences of 14 pathogens (Escherichia Coli K1 strains, Streptococcus agalactiae, Listeria monocytogenes, Streptococcus pneumoniae, Haemophilus influenzae, Neisseria meningitidis, enterovirus [EV], human parechovirus [HPeV], HSV-1 and 2, varicella virus zoster virus [VZV], human herpesvirus 6 [HHV-6], cytomegalovirus [CMV], Cryptococcus neoformans/gattii) causing CNS infection within 1 hour of run-time on machine with a list price of $193/cartridge(6). Laboratory and clinical validation studies have demonstrated high overall sensitivity and specificity but identified potential issues with S. pneumoniae false positives(6). Decreased sensitivity for HSV-1 and cryptococcus targets have been more recently noted by the manufacturer(7, 8).

Despite widespread uptake of MEP, optimal implementation strategies, as well as the clinical impact and cost-effectiveness of standardized MEP testing remain unknown. Variability in access, who to test, when to use MEP in the diagnostic evaluation, and how to interpret results were issues raised in a recent survey of pediatric infectious disease providers(9). Pediatric MEP studies have been limited to potential clinical impact evaluations using banked samples(10, 11) and small retrospective observational case series of unrestricted MEP testing ordered at clinician discretion(12–15). Some adult MEP studies applied a more standardized approach to MEP implementation targeting use based on CSF indices and clinical suspicion(16–18). The manufacturer’s economic analysis of MEP reported theoretical cost savings of ~$3500/case using a model based on assumptions about clinical decision-making in a population with confirmed, not suspected, CNS infection(19, 20). There is a clear need for real-world clinical and financial data using a standardized approach to MEP testing in children with suspected CNS infection to determine the true impact in the pediatric clinical setting. Our study sought to fulfill this need by investigating the impacts of a rapid diagnostic stewardship program for standardized implementation of MEP in children with suspected CNS infection at a quaternary care children’s hospital.

Methods

Study Design, Setting, and Inclusion Criteria

A pre-post quasi-experimental cohort study compared pre-implementation controls (2015–2016) to post-implementation cases (2017–2018) following the January 2017 implementation of the rapid CSF diagnostic stewardship program at Children’s Hospital Colorado (CHCO). CHCO is a 479 bed quaternary care children’s hospital in Aurora, CO serving a catchment area of over 750,000 children. On average, the CHCO microbiology laboratory tests 883 non-shunt CSF samples annually.

All cases with CSF obtained by lumbar puncture sent to the CHCO microbiology laboratory for testing from a child meeting ≥1 of the following four criteria during the study period were included: i. <2 months of age, ii. concern for encephalitis (seizures, altered mental status, focal neurologic examination abnormalities, MRI/EEG changes), iii. immunocompromised, iv. ≥5 WBCs/μL in CSF. Data was collected via chart review and input into a study-specific standardized REDCap database by a member of the research team with duplicate review by a study MD for clinical determinations. The study was approved by the Colorado Multiple Institutional Review Board (#16-1526).

Intervention Design

The rapid CSF diagnostic program to implement MEP was designed using pre-analytic, analytic, and post-analytic core components of diagnostic stewardship interventions for molecular diagnostic tests (Figure 1)(21, 22). A simulated decision analysis identified populations in whom MEP was likely to impact clinical care or testing which guided selection of approved use criteria with end-user clinician stakeholder input(11, 23). Electronic medical record (EMR) decision support was implemented which required providers to select an indication for MEP from amongst the three identified approved use criteria selected: i. <2 months of age, ii. concern for encephalitis, or iii. immunocompromised, otherwise the MEP could only be ordered conditionally and was run only if ≥5 WBCs/μL were present on CSF cell counts. MEP was only permitted to be ordered in addition to CSF Gram stain and culture.

Figure 1:

Rapid Cerebrospinal Fluid Diagnostic Stewardship Program Intervention Design

Laboratory Methods

Throughout both study periods, the CHCO microbiology laboratory offered singleplex PCR testing for HSV-1 and 2 using the MultiCode-RTx HSV 1&2 Kit (Luminex, Austin, TX), EV using the Xpert EV kit on GeneXpert (EV; Cepheid, Sunnyvale, CA), and CLIA approved lab developed RT-PCR tests for HHV-6(24), CMV, VZV, and HPeV(25) following nucleic acid extraction and processing with Virus Minikits 2.0 (Qiagen, Valencia, CA) on the BioRobot EZ1. CSF was cultured on blood and chocolate agar plates, as well as in blood culture broth bottles (Plus Aerobic/F and PedsPlus/F [Becton Dickinson and Co, Sparks, MD]) incubated in a BacTec 9120/9240 automated system (Becton Dickinson and Co). MEP testing was performed in real-time on-demand 24 hours/day, 7 days/week with conditional testing occurring following CSF cell counts if WBC≥5uL. MEP was performed in a dedicated biosafety cabinet to reduce the risk of contamination. Positive results on MEP were repeated in duplicate before reporting to providers with discrepant results repeated a third time to adjudicate.

Reporting Methods

Pre-implementation, a microbiology laboratory scientist communicated positive CSF results directly to the responsible medical provider. Post-implementation, real-time decision support was provided by a team of antimicrobial stewardship and infectious disease providers. This team received positive results via text page, then communicated the interpretation of the results with antimicrobial recommendations based on consensus guidelines to the provider, as previously described(26). This service was provided in real-time weekdays 8am–5pm with off-hour results called directly from the microbiology laboratory scientist to the provider and followed up by the stewardship team the next weekday morning.

Clinical Outcomes Analysis

The primary outcome of the study was time-to-optimal antimicrobials, defined as the time from LP to the first dose of antimicrobials targeted to the identified pathogen (as defined by pre-determined consensus recommendations), or to the time of antimicrobial cessation if no treatable pathogen was identified. Time-to-effective antimicrobials for treatable etiologies was defined as the time from LP to the first dose of antimicrobials to which the bacteria identified was found to be susceptible according to Clinical Laboratory Standards Institute criteria or first dose of acyclovir for HSV and VZV.

Time-to-outcome variables were compared using the Kaplan-Meier method and log-rank tests. Adjusted analysis, considering inclusion criteria variables, was conducted using Cox proportional hazards. Categorical variables were compared using chi-square or Fisher’s exact test. Continuous variables were compared using two-sample t-tests. Statistical analysis was conducted using R 3.1.1 software with significance set at 0.05. An estimated sample size of 1766 children compared 1:1 with pre-intervention controls was found to provide 94% power to detect an 8-hour difference in time-to-optimal therapy.

Cost Analysis

CSF molecular microbiology testing costs were directly calculated from units of singleplex PCR and MEP testing performed multiplied by actual cost per unit of testing. Total microbiology testing costs and total hospital costs were derived by applying the CHCO 2019 overall cost to charge ratio to hospital charges adjusted to 2019 prices based on service date using 5% discount rate with sensitivity analysis at 4% and 6%.

Results

A total of 1124 pre-implementation controls and 1127 post-implementation cases were included with groups sharing similar subject demographics and CSF indices (Table 1). The average age was 3.5 years with one-third of the study population having an underlying medical condition. The post-implementation group had fewer infants <2 months (49% vs. 56%, p=0.0019) and more cases with concern for encephalitis (53% vs. 46%, p=0.0006).

Table 1:

Characteristics of Included Subjects and Cerebrospinal Fluid Specimens

Characteristic	Pre-Implementation Controls (n=1124)	Post-Implementation Cases (n=1127)	P Value
Demographics
Age (median months)	42±67	43±68	0.71
Female Sex	506 (45%)	531 (47%)	0.3389
Race			0.0598
American Indian/Alaska Native/Native Hawaiian/Other Pacific
Islander	13 (1%)	17 (2%)
Asian	27 (3%)	38 (4%)
Black or African-American	77 (8%)	68 (7%)
White	766 (81%)	757 (78%)
More than one race	63 (7%)	95 (10%)
Hispanic of Latino Ethnicity	315 (29%)	325 (31%)	0.5426
Underlying medical condition	343 (31%)	345 (31%)	1
Inclusion Criteria Met
Age <2 months	626 (56%)	552 (49%)	0.0019
CSF WBC>5	439 (39%)	474 (42%)	0.1542
Immunocompromised	31 (3%)	40 (4%)	0.3404
Concern for encephalitis	518 (46%)	602 (53%)	0.0006
Cerebrospinal Fluid Indices
CSF WBC	3 (1, 13)	4 (1, 18)	0.2263
CSF RBC	7 (1, 280)	11 (1, 386)	0.5129
CSF glucose	51 (46, 59)	51 (44, 60)	0.4959
CSF protein	52 (26, 83)	49 (26, 84)	0.8455

Testing practices changed from over half of cases with CSF tested by pathogen-specific singleplex PCR pre-implementation to 10% (p<0.0001) post-implementation with 72% undergoing syndromic testing using MEP (Table 2). Time-to-positive CSF testing results decreased from 8.6 to 4.6 hours (p<0.0001) post-implementation.

Table 2:

Microbiologic Testing and Discharge Diagnoses

Category	Pre-Implementation Controls (n=1124)	Post-Implementation Cases (n=1127)	P Value
Overall Testing
Number of Singleplex PCR Tests per CSF Specimen			<0.0001
0	539 (50%)	969 (90%)
1	289 (27%)	97 (9%)
>2	255 (24%)	7 (1%)
CSF Singleplex PCR Positive	72 (6%)	4 (0%)
MEP Testing Performed	N/A	805 (72%)
MEP Positive	N/A	102 (13%)
Time to Positive CSF Test Result (hrs)	8.6 (5, 17)	4.6 (3.1, 6)	<0.0001
Bacterial Testing
Positive Gram Stain	11 (1%)	16 (1%)	0.4243
Positive CSF culture	17 (2%)	20 (2%)	0.7109
Positive MEP for bacteria	N/A	20/805 (2%)
Viral Testing (positive/tested)
Enterovirus detected in CSF by singleplex PCR	63/348 (18%)	1/25 (4%)
Enterovirus detected in CSF by MEP	N/A	55/805 (7%)
Enterovirus neurologic cases (any site)	68 (6%)	88 (8%)	0.1189
HSV detected in CSF by singleplex PCR	3/442 (1%)	3/80 (4%)
HSV detected in CSF by MEP	N/A	3/805 (0%)
HSV neurologic cases (any site)	13 (1%)	15 (1%)	0.8547
Parechovirus detected in CSF by singleplex PCR	5/72 (0%)	0/4 (0%)
Parechovirus detected in CSF by MEP	N/A	11/805 (1%)
Parechovirus neurologic cases (any site)	5 (0%)	12 (1%)	0.1456
HHV-6 detected in CSF by singleplex PCR	1/8 (13%)	0/3 (0%)
HHV-6 detected in CSF by MEP	N/A	13/805 (2%)
VZV detected in CSF by singleplex PCR	0/41 (0%)	0/5 (0%)
VZV detected in CSF by MEP	N/A	2/805 (0%)
CMV detected in CSF by singleplex PCR	0/16 (0%)	0/2 (0%)
CMV detected in CSF by MEP	N/A	0/805 (0%)
Fungal Testing (positives/tested)
Cryptococcus Ag detected in CSF	0/2 (0%)	0/2 (0%)
Cryptococcus detected in CSF by MEP	N/A	0/805 (0%)
Discharge Diagnosis
Neurologic Discharge Diagnosis	430 (38%)	499 (44%)	0.0017
Infectious Neurologic Disease Discharge Diagnosis	115 (10%)	167 (15%)	0.0298
Proportion due to Viral Cause	86 (75%)	124 (74%)
Proportion due to Bacterial Cause	27 (23%)	35 (21%)
Non-CNS Bacterial Infection	175 (16%)	178 (16%)	0.9149
Respiratory Viral Infection	253 (23%)	284 (25%)	0.1347

There were no significant differences in pathogen-specific yield between study periods. A pathogen was identified in 103 (12.8%) cases with MEP testing sent (n=805), most commonly EV (n=54), followed by HHV-6 (n=13) and HPeV (n=11). Only one child with HHV-6 detected was immunocompromised, six were healthy febrile children diagnosed with roseola, and three cases had serious bacterial infections identified outside the CNS. Bacteria were detected by MEP in 2% (n=20) of cases tested, including five cases with negative cultures, four of which had positive Gram stains and were pre-treated with antibiotics, three had positive blood cultures for the pathogen identified on MEP, and all were treated as true bacterial meningitis. Two cases had negative MEP testing and positive bacterial CSF cultures, both for Staphylococcus aureus, which is not a target on the panel. More cases received neurologic diagnoses (44% vs. 38%, p=0.0017), and more specifically infectious neurologic diagnoses (15% vs. 10%, p=0.0298), post-implementation.

The primary outcome of time-to-optimal antimicrobials decreased by 10 hours from a median 28 hours (95% CI: 25–32) to 18 hours (95% CI: 13–21), p<0.0001, following implementation of the rapid CSF diagnostic stewardship program (Figure 2; Table 3). The probability of receiving optimal antibiotics was 1.13 (95% CI: 1.04–1.23) times higher post-implementation compared to pre-implementation after adjusting for age < 2 months and concern for encephalitis (p=0.004). The proportion of subjects started on IV antibiotics and IV acyclovir, and number of antibiotics, did not differ between groups. The antimicrobial duration, however, decreased from 36 to 24 hours (p=0.0037) and total hours of antimicrobials received decreased from 60 to 38 hours (p=0.0007). EV cases were started on antibiotics more often than non-EV cases (80% vs. 69%, p=0.0044), however had significantly shorter duration of antibiotics (14.4 vs. 31.9 hours, p=0.0006). Time-to-effective antimicrobials did not differ pre- and post-implementation amongst cases with treatable etiology (p=0.471) or treatable organism in CSF (p=0.5817). There were no differences in the proportion of subjects hospitalized (p=0.6929) or the length of inpatient hospital stay (p=0.9635). Death due to CNS infection was uncommon (<1%) and did not differ between groups.

Figure 2:

Kaplan-Meier Analysis of Time-to-Optimal Antimicrobial Therapy in Pre-Implementation Controls (n=1124) vs. Post-Implementation Cases (n=1127)

Table 3:

Clinical and Financial Outcomes

Category	Pre-Implementation Controls (n=1124)	Post-Implementation Cases (n=1127)	P Value
Clinical Outcomes
Hospitalized	1024 (91%)	1033 (92%)	0.6929
Length of Inpatient Stay (median days)	4 (3, 9)	4 (3, 9)	0.9635
Death During Hospitalization	21 (2%)	13 (1%)	0.2264
Death Due to CNS Infection	6 (1%)	7 (1%)	1
Started on IV Antibiotics	799 (71%)	768 (68%)	0.1685
Number of IV Antibiotics Received	2 (2, 3)	2 (2, 3)	0.5326
IV Antimicrobial Duration (hrs)	36 (0, 60)	24 (0, 50.4)	0.0037
IV Antimicrobial Hours (hrs)	60 (0, 190)	38 (0, 98)	0.0007
Received IV Acyclovir	301 (27%)	284 (25%)	0.42
Duration of IV Acyclovir Amongst Those Started (hrs)	24 (8, 40)	17 (8, 40)	0.206
Treatable Etiology Identified	205 (18%)	217 (19%)	0.5799
Time to Effective Antimicrobials for Children with Treatable Etiology (hrs)	0.68 (0.48, 0.93)	0.75 (0.43, 1.018)	0.471
Time to Effective Antimicrobials for Children with Treatable Organism in CSF (hrs)	0.43 (0, 1.7)	0.65 (0, 1.58)	0.5817
Time to Optimal Antimicrobial Regimen Initiation (hrs) a	28 (25, 32)	18 (13, 21)	<0.0001
Cost Outcomes (median, IQR)
CSF Molecular Microbiology Testing Costs	$95 (0, 99)	$193 (99, 193)	<0.0001
Total Microbiology Testing Costsb	$910 (631, 1267)	$1010 (684, 1321)	0.0001
Total Hospital Costsb	$15614 (9168, 34603)	$15507 (9408, 34602)	0.9634

^aCox Proportional Hazards Adjusted Analysis: The probability of receiving optimal antibiotics was 1.13 (95% CI: 1.04, 1.23) times higher in cases compared to controls after adjusting for age (<2 mo vs. >2 mo) and concern for encephalitis; p=0.004

^bEstimated costs derived by applying CHCO 2019 overall cost to charge ratio to hospital charges inflated to 2019 prices based on service date using 5% discount rate. Total hospital charges included all room & board, laboratory & radiology diagnostic, pharmacy & surgery treatment charges.

Costs of molecular microbiology tests conducted on CSF increased from a median $95 to $193 (p>0.0001) and total microbiology testing costs increased from $910 to $1010 (p=0.0001) following MEP implementation. There were no significant differences in the total inflation-adjusted hospital costs between the pre- and post-implementation eras (p=0.9634). Sensitivity analysis using a 4–6% discounting rate did not change this finding (Table 4).

Discussion

Implementation of rapid molecular syndromic testing of CSF in conjunction with a diagnostic stewardship program was associated with improved antimicrobial use in children with suspected CNS infections at a quaternary care children’s hospital. Key diagnostic stewardship components to the successful implementation of MEP included the identification of high-impact patient populations to target for testing using EMR decision support and ensuring rapid, appropriate response to positive test results with real-time antimicrobial stewardship decision support. More rapid results from targeted testing with MEP likely led to decreased duration of empiric antimicrobial therapy which improved the time-to-optimal therapy. Costs of microbiologic testing increased with MEP but total hospital costs were unchanged. The clinical impact observed without increased hospital costs suggests this rapid diagnostic stewardship program can be a value-added implementation strategy for MEP.

EVs were the most frequently identified cause of neurologic disease in this pediatric population, accounting for 55% of cases with an infectious neurologic discharge diagnosis. Identification of EV in CSF was associated with shorter empiric antimicrobial duration as shown in previous studies(27–31). Only 2% of children with suspected CNS infection in the study had a treatable etiology identified in CSF with most receiving broad empiric antimicrobials shortly following LP, therefore differences in time-to-effective therapies were not expected. However, MEP detected a bacterial cause of meningitis in five cases with negative CSF cultures, suggesting a potential beneficial role for MEP in cases with antibiotic pre-treatment prior to LP.

The widespread uptake of MEP testing following FDA approval preceded knowledge of effective implementation strategies for use, particularly in the pediatric realm. Without clear indications for use, there is potential for under-utilization in high-risk populations and clinically impactful situations, such as infants, immunocompromised children, those with pleocytosis or concern for encephalitis, as well as over-utilization in those unlike to benefit, such as healthy older children without concern for encephalitis and without pleocytosis to suggest meningitis. Wide variability in clinical outcomes reported amongst the mostly retrospective observational pediatric MEP studies reflects the lack of standardized implementation and need for diagnostic stewardship of this new technology. Controversy has surrounded the use of cell counts to target molecular CSF testing(32–34). The nuanced approach taken in this study of targeting MEP testing to those with CSF pleocytosis, but with caveats for populations in whom CSF pleocytosis may be absent in the presence of CNS infection (such as infants and immunocompromised patients) may alleviate some concerns of missed diagnoses. Testing all CSFs collected from a pediatric population, regardless of cell count, inherently includes a large proportion of patients in whom the only concern is meningitis where normal CSF indices make the probability of treatable CNS infection unlikely. Indiscriminate use in low pre-test probability cases leads to unnecessary resource utilization and may confuse clinical decision-making with incidental detection of HHV-6(35, 36) or false positive detection of pneumococcus(6).

This study highlights several areas where continued diagnostic stewardship work is needed to further optimize the use of MEP. The widespread empiric use of acyclovir for low-risk children outside of the neonatal age range(37) and vancomycin in the post-pneumococcal conjugate vaccine era(38) could potentially be decreased, along with nephrotoxicity risks, through rapid negative testing on MEP(39). Given the lack of significant impact on the use of these therapies in this study, additional focused post-analytic diagnostic stewardship work will be necessary to achieve this goal. In the emergency department setting, algorithms incorporating the more rapid 2-hour potential turnaround time of MEP for detection of EVs and HPeVs in children outside the neonatal period at low risk for HSV and bacterial meningitis has potential to decrease unnecessary antimicrobial starts and potentially hospital admissions, in addition to the decreased antimicrobial duration seen in this study. While HHV-6 was included as a target on MEP to detect as a cause of encephalitis in immunocompromised patients, this study demonstrates that when MEP is used in healthy children, it is more likely to detect HHV-6 from febrile young children with non-CNS roseola presentations or may represent chromosomal integration in older children, which is difficult to differentiate with qualitative CSF testing(40). The removal or suppression of this target for non-immunocompromised, older populations could be considered to minimize confusion caused by this result when it is unlikely the cause of disease, such as in the three healthy older children in this study with HHV-6 detected in CSF found to have serious bacterial infections.

It is important to note that this was a study of the rapid CSF diagnostic stewardship program implementation strategy including MEP, not a study of the impact of MEP testing alone, as we included all cases meeting inclusion criteria, not just those who received MEP testing in the post-implementation group. This was essential to having a comparable historical pre-implementation control group defined by the same standardized criteria as the post-implementation group, as it is impossible to assess who would have received MEP testing before it was available. The previously noted financial modeling study by the manufacturer limited their study population to the small minority of children with confirmed CNS infection at discharge by administrative billing codes and concluded based on this data that all children with suspected CNS infection should undergo MEP testing, without the external validity to be applied to this population(19, 20). The strength of our real-world financial analysis was the application of strict inclusion criteria at the time when CNS infection was suspected, lending validity of our findings to the undifferentiated child at the time of clinical decision-making. In this real-world scenario, our study did not confirm the theoretical cost savings previously reported.

The pre-post quasi-experimental design of this single center study has inherent limitations. Though this type of dissemination and implementation study provides practical real-world impact and cost analysis, non-intervention associated changes over time at CHCO had potential to impact outcomes. For instance, CHCO updated febrile neonate guidelines in 2018 to include risk-stratification using procalcitonin to decrease lumbar punctures in low-risk infants 29–60 days, which may account for fewer infants being included in the post-implementation group, though this did not impact outcomes in the adjusted analysis(41). CHCO benefitted from having an onsite dedicated pediatric microbiology laboratory, an existing diagnostic stewardship program(26) and successful handshake antimicrobial stewardship program(42), which made implementation of MEP with on-demand testing and real-time decision support more feasible than it may be at institutions without existing infrastructure. While the cautious approach of repeating all positive MEP results increased confidence in the accuracy of results, the decreased timeliness of result reporting and subsequent antimicrobial optimization may have underestimated clinical impact. Over time with increased experience and trust in MEP results, CHCO discontinued this practice.

Newly available rapid molecular diagnostic technologies are revolutionizing our ability to rapidly and accurately detect infectious pathogens from biologic specimens. However, technology alone does not optimally improve care and may increase costs. Thoughtful, judicious use of these promising tools using diagnostic stewardship principles to guide testing towards impactful use, conduct testing real-time on-demand, and ensure rapid, appropriate medical decision making in the clinical context of the patient is key to optimizing use. This study demonstrates that the approach of implementing advanced molecular diagnostic technologies combined with antimicrobial stewardship support can be successfully applied to CSF testing for suspected CNS infections in children, and likely replicated with future emerging diagnostic technologies.

Abbreviations:

MEP

Meningitis Encephalitis Panel

CSF

Cerebrospinal Fluid

CNS

Central Nervous System

CHCO

Children’s Hospital Colorado

LP

Lumbar Puncture

HSV

Herpes Simplex Virus

EV

Enterovirus

HPeV

Human Parechovirus

VZV

Varicella Zoster Virus

HHV-6

Human Herpesvirus 6

CMV

Cytomegalovirus

MRI

Magnetic Resonance Imaging

EEG

Electroencephalography

WBC

White Blood Cell