Diagnostic Stewardship in Community-Acquired Pneumonia With Syndromic Molecular Testing

This randomized clinical trial evaluates a rapid syndromic polymerase chain reaction–based diagnostic test for lower respiratory tract infections in patients presenting to the emergency department.


Introduction
2][3] Despite their substantial value, microbiological diagnosis and targeted treatment are not received by most patients. 4Although culture-based methods are considered to be a standard approach in the bacteriological diagnosis of CAP, they are labor intensive, detect a pathogen in only 20% to 40% of patients, and are insufficient to influence early decisions on antimicrobial therapy. 4,5pid syndromic polymerase chain reaction (PCR)-based panels have improved pathogen detection, potentially facilitating pathogen-directed treatment, reducing unnecessary use of antibiotics, and shortening hospital length of stay (LOS). 6,7Although the potential benefits of rapid PCR panels in CAP are clear, limited evidence currently supports their routine use.][10][11] A recent randomized clinical trial examined patients with pneumonia using a comprehensive mPOCT and found an increase in result-directed therapy and de-escalation of antibiotics in the mPOCT group. 12However, the study included a mixture of patients from intensive care units with hospital-acquired pneumonia, ventilator-associated pneumonia, and CAP; thus, the findings may not be specifically applicable to patients with CAP.
In this trial, we aimed to determine whether the judicious use of a syndromic PCR-based panel for rapid testing of CAP in the emergency department (ED) leads to faster, more accurate microbiological test result-based treatment.Findings from this trial can potentially inform future guidelines on the management of CAP.

Study Design and Participants
This pragmatic, parallel-arm, single-blinded, single-center, randomized clinical superiority trial was conducted in the ED of Haukeland University Hospital, a large tertiary care institution in Bergen, Norway, that serves as a local hospital for approximately 470 000 residents and a referral hospital for 1 million people.The Regional Committee for Medical and Health Research Ethics in Norway approved the study protocol 13 (Supplement 1).Written informed consent was obtained from all patients or their legal guardian or close relatives.We followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.
Patients were recruited from September 25, 2020, to June 1, 2021, and again from August 15, 2021, to June 21, 2022.Patients were eligible for inclusion if they were 18 years or older; presented to the ED with suspected CAP; and met at least 2 of the following criteria: new or worsening cough, new or worsening expectoration, new or worsening dyspnoea, hemoptysis, pleuritic chest pain, radiological evidence of pneumonia, abnormalities on chest auscultation and/or percussion, or fever

JAMA Network Open | Infectious Diseases
Diagnostic Stewardship in Community-Acquired Pneumonia With Molecular Testing (Ն38.0 °C).Patients were ineligible if they had cystic fibrosis, had severe bronchiectasis, were hospitalized within the past 14 days prior to admission, were under a palliative approach (ie, life expectancy of <2 weeks), or were not willing to provide an LRT sample. 13

Randomization and Masking
Patients were randomized 1:1 to either the intervention arm (receiving rapid syndromic PCR testing in addition to standard-of-care microbiological diagnostics) or to the standard-of-care arm (receiving standard microbiological diagnostics alone) (Figure 1).Block randomization with varying block sizes (4, 6, or 8 patients per block) was applied using the R package blockrand (R Core Team).The allocation sequence was implemented in an electronic data capture system (Viedoc; Viedoc Technologies).Trial participants and ED clinicians were blinded to treatment group allocation.

Procedures
Eligible patients were included shortly after presenting to the ED.Study nurses and physicians collected baseline information.
Rapid testing was performed using the BioFire FilmArray Pneumonia plus panel (FAP plus; bioMérieux).The FAP plus PCR test detects 27 bacterial and viral respiratory pathogens with 7 antimicrobial resistance genes, with the specific targets detailed in eTable 1 in Supplement 2.   At admission, all patients were tested for SARS-CoV-2 infection using oropharyngeal or nasopharyngeal swabs tested on the GeneXpert system (Cepheid).The LRT samples were collected in the ED, primarily through sputum induction, using nebulized isotonic (0.9%) or hypertonic (5.8%) saline.Endotracheal aspiration was performed in case of an unsuccessful sputum induction and in patients with SARS-CoV-2 infection.

Outcomes
Two primary outcomes were the (1) provision of pathogen-directed treatment based on a relevant microbiological test result and (2) time to provision of pathogen-directed treatment (within 48 hours after randomization).The first was a binary outcome, whereas the second was an event-time outcome wherein right censoring was present (ie, patients may cease participation due to death, discharge, or reaching 48 hours after randomization without receiving pathogen-directed treatment).Primary outcomes were defined for all patients who were randomized.
Two of us (D.L.M and S.S.) assessed whether and when a patient received pathogen-directed antimicrobial treatment based on a microbiological test result.In case of disagreement, a third physician (S.T.K.) arbitrated.To be considered as pathogen-directed treatment, documentation in the patient's journal by the treating physician was required that described a change in antimicrobial treatment, continuation of an already correctly initiated antimicrobial treatment, or discontinuation of an antimicrobial treatment.The final diagnosis of CAP was established retrospectively through clinical adjudication using prespecified criteria (eTable 3 in Supplement 2).
Secondary outcomes included the binary outcomes: provision of any antibiotics, provision of narrow-spectrum antibiotics within 48 hours, provision of a single dose of antibiotics, antibiotics not used for more than 48 hours, treatment with intravenous antibiotics, de-escalation from broadspectrum to narrow-spectrum antibiotics, and escalation from narrow-spectrum to broad-spectrum antibiotics.Continuous secondary outcomes included the duration of provision of antibiotics (days), intravenous antibiotics (days), broad-spectrum antibiotics (days), time from admission to the administration of antibiotics (hours), and turnaround time (hours; ie, time from admission to receiving a microbiological report (result of FAP plus PCR test and/or sputum culture).Broadspectrum antibiotics were defined as penicillin with enzyme inhibitors, second-and third-generation cephalosporins, carbapenems, and quinolones. 14Additionally, LOS in days, mortality (30 days and 90 days), readmission within 30 days after discharge, and adverse outcomes were reported.

Statistical Analysis
Because 2 primary outcomes were used, separate sample size calculations were performed for each outcome at a 2-sided significance level of .05/2= .025(instead of .05),assuming a power of 80%.
To detect an increase in the provision of pathogen-directed treatment from .40 to .50, we required the sample size to be 470 per arm.Similarly, to detect a reduction of 0.2 SD in the time to provision of pathogen-directed treatment, we established the sample size to be 477 per arm (ie, 954 in total).
Allowing for a 10% dropout rate resulted in a total sample size of 1060 patients.
Baseline patient characteristics were summarized using counts, percentages, and totals for categorical variables and medians and IQRs for continuous variables.Missing baseline values were imputed by means of a single imputation using chained equations. 15e 2 primary outcomes were analyzed according to the intention-to-treat principle, and Bonferroni adjustment was applied.Available-case analyses were used for the secondary outcomes.For binary outcomes, logistic regression models with logit and identity link functions were used to estimate odds ratios (ORs) and absolute risk differences, respectively.Two models were fitted for the event-time primary outcome: Cox proportional hazards regression model and restricted mean survival time model. 16The proportional hazards assumption for the Cox regression model was assessed visually using cumulative log-log plots.The restricted mean survival time model is a flexible survival analysis model that does not require proportional hazards because it is based on survival curves, which may be estimated parametrically or nonparametrically. 17Kaplan-Meier survival curves and log-rank tests were also reported.For continuous secondary outcomes, linear regression with logarithm-transformed outcomes was used to estimate differences in medians and ratios of medians; differences in medians were approximated using a Taylor expansion. 18For the 2 primary outcomes, a post hoc analysis that was adjusted for season was also carried out through inclusion of an indicator of whether recruitment took place from September 25, 2020, to June 1, 2021, or from August 15, 2021, to June 21, 2022.

JAMA Network Open | Infectious Diseases
A significance level of P = .05was applied, and the statistical software R version 4.1.2(R Core Team) was used. 19Imputation through chained equations and fitting of restricted mean survival time models were carried out using the R packages mice and SURVRM2, respectively.

Results
The trial was stopped earlier than planned because an ad hoc interim analysis that we conducted due to slow recruitment, carried out June 16, 2022, showed substantial differences between the intervention and standard-of-care arms for both primary outcomes.A total of 2265 patients were assessed for eligibility, of whom 374 participated, with 187 patients randomized to each arm (Figure 1).Both arms showed similar distributions of patient characteristics (

Findings for All Randomized Patients
Forty-eight hours after randomization, 66 of 187 (35.3%) patients in the intervention arm and 25 of 187 (13.4%) patients in the standard-of-care arm received pathogen-directed treatment (Figure 2), corresponding to a reduction in absolute risk of 21.9 (95% CI, 13.5-30.3)percentage points and an OR for the intervention arm of 3.53 (95% CI, 2.13-6.02;P < .001)(Table 2).
The median (IQR) time to provision of pathogen-directed treatment within 48 hours was 34.Similarly, empirical antibiotic treatment was de-escalated to more narrow-spectrum treatment for 10 of 97 patients (10.3%) in the intervention arm and 5 of 103 patients (4.9%) in the standard-of-care arm (absolute difference, 5.5 percentage points; 95% CI, −1.9 to 12.8 percentage points; P = .14).
Continuation of appropriate empirical treatment occurred in 16 of 97 patients (16.5%) in the intervention arm and 7 of 103 (6.8%) in the standard-of-care arm.This resulted in an absolute  The dotted line at 48 hours indicates the censoring threshold for the primary outcome of time to pathogen-directed treatment.Abbreviations: HR, hazard ratio; OR, odds ratio.
a Differences were estimated as absolute differences for binary outcomes (on use) and mean differences for event-time outcomes using logistic regression (with the identity link function) and restricted mean survival time methods, respectively.The ORs for binary outcomes and HRs for event-time outcomes were estimated using logistic regression and Cox regression, respectively.P values corresponding to testing ratios equal to 1 were reported.
b Time elapsed from randomization to provision of pathogen-directed treatment within 48 hours.patients in the intervention and standard-of-care arms, respectively, died within 90 days (absolute risk difference, 2.7 percentage points; 95% CI, −2.6 to 7.9 percentage points; P = .32).

Microbiological Detections and Adverse Events
The intervention arm compared with standard-of-care arm had a higher number of bacterial detections (175 vs 72) and viral detections (74 vs 63).When considering only the patients with confirmed CAP, the intervention arm maintained a higher total number of bacterial (113 vs 57) and viral (39 vs 34) detections than the standard-of-care arm.The breakdown of microbiological detections in the 2 arms is shown in eTable 7 in Supplement 2.
No serious adverse events were observed, and the number of adverse events was similar in both arms for saline-induced sputum: 7 were registered in the intervention arm and 8 in the standard-of-care arm.These adverse events were dyspnea (n = 6), rapidly resolved hypoxemia (n = 5), nausea (n = 1), coughing (n = 1), and nonsevere tachycardia (n = 2).One patient in the intervention group experienced coughing during endotracheal aspiration.

Discussion
This randomized clinical trial demonstrated that use of the FAP plus PCR test for LRT pathogens as part of the diagnostic workup for hospitalized patients with suspected CAP increases the provision of and reduces the time to pathogen-directed treatment compared with comprehensive standard-ofcare microbiological testing.The FAP plus PCR test provides ED clinicians with close to real-time information for actionable treatment decisions.
To our knowledge, this trial was the first to examine the effect of a rapid syndromic PCR pneumonia panel applied specifically to patients hospitalized with CAP.Most previous studies did not use a comprehensive syndromic PCR panel or included patients shortly after admission, potentially limiting the advantages of rapid molecular testing.Another aspect of its novelty is the emphasis on pragmatism whereby decisions to continue, switch, or discontinue antimicrobial treatment were at the discretion of the treating physician alone.
The intervention led to a reduction (by 9.4 hours) in the median time without provision of pathogen-directed treatment within the first 48 hours after randomization, compared with standard of care.For patients with CAP, the median turnaround time (from admission to receiving an LRT test result without restriction to 48 hours) was reduced by much more (53.8hours) for the intervention vs standard-of-care group.This result partly reflects hospital practice and is comparable to previous findings. 12A faster microbiological diagnosis allows for directed therapy, which has been shown in previous studies to improve outcomes, limit antibiotic overuse, and prevent antimicrobial resistance. 6,10Despite crowded conditions at the ED, a FAP plus PCR test result was delivered within 4 hours for patients with CAP, a turnaround time comparable to that achieved in other centers. 12 significant differences in clinical outcomes were observed between the intervention and standard-of-care groups.However, the primary objective of this trial was to enhance diagnostic stewardship by determining and then leveraging the rapid, multipathogen detection capabilities of the FAP plus PCR test.
We sought to reduce the time to provision of pathogen-directed treatment, potentially decreasing unnecessary or broad-spectrum antibiotic use and fostering antimicrobial stewardship.
Future research should continue to explore innovative approaches to improving the diagnosis and management of respiratory infections, such as incorporating clinical decision support tools and antimicrobial stewardship programs into routine practice.

Strengths and Limitations
The strengths of this study include the pragmatic design, primary outcome values obtained from electronic medical records by treating physicians who were not involved in the study, duration of 2 winter seasons, broad inclusion criteria representing typical patients with respiratory tract symptoms admitted to Norwegian hospitals, representative sampling from the LRT, a simple intervention, and comparison of intervention to standard-of-care microbiological testing.These factors suggest that the primary study findings are generalizable to similar hospital settings.The standard-of-care diagnostic testing in this study encompassed a wide battery of tests, including in-house molecular tests, rendering standard of care as competitive as possible compared with the commercial FAP plus PCR test.Moreover, the FAP plus PCR test includes several bacterial and viral pathogens along with targets for selected antimicrobial resistance genes, making this panel applicable to other settings with different microbial etiological profiles and background resistance rates for CAP. 20key study strength is that the study physicians were not involved in the treatment of patients.
The restriction to a narrow time frame (48 hours after randomization), which spanned the period from respiratory sampling to availability of results, supports interventions that can contribute to the timely administration of appropriate antibiotics, a central tenet of care for patients with pneumonia.
The higher rate of continuation of appropriate empirical treatment for patients with CAP in the intervention arm suggests that this diagnostic tool assists in confirming the appropriateness of the initial empirical therapy and prevents unnecessary changes in treatment.Additionally, a larger More patients in the intervention arm than in the standard-of-care arm had an escalation from narrow-spectrum to broad-spectrum antibiotics.This escalation could raise concerns about antibiotic overuse; however, it is pertinent to emphasize that Norway has a low level of antibiotic resistance, and guidelines recommend using narrow-spectrum antibiotics. 21Per the guidelines, empirical treatment is benzylpenicillin for mild to moderate CAP and benzylpenicillin and gentamicin for severe CAP. 21Treatment with narrow-spectrum antibiotics for respiratory tract infections is common practice in Norway, 22 leaving little room for de-escalation, and even change from penicillin G to a more broad-spectrum penicillin (eg, ampicillin) was considered to be an escalation.A background of a low level of antibiotic resistance implies that the differences found between the 2 arms for escalation or de-escalation of an antibiotic are, in a sense, the minimal differences to expect when introducing rapid testing in a Norwegian hospital setting.
This study has some limitations.First, the single-center design limits generalizability.However, the study demonstrated that embedding comprehensive rapid testing in a busy ED setting is possible.Second, the trial was stopped early for efficacy, and there could be a risk of inflated estimates of differences between the intervention and standard of care, although this risk is presumably small as we found highly significant differences.Moreover, inflation is generally small, and continuing a trial to achieve a slight change in estimates would not be rational. 23

Conclusions
In this randomized clinical trial, use of the FAP plus PCR test led to faster and more targeted microbial treatment for hospitalized patients with CAP.The findings align with the broader concept of clinical management or treatment stewardship for LRT infections.Routinely deployed rapid syndromic testing could complement or replace targeted components of the standard laboratory-based diagnostic repertoire for patients who are admitted to the hospital with an acute respiratory illness.
Future studies should examine the effect of comprehensive rapid syndromic testing on clinical outcomes, the cost-effectiveness of this diagnostic tool, and the development of implementation strategies that facilitate the integration of rapid syndromic testing into routine clinical practice.

Figure 1 .
Figure 1.Flow Diagram of Study Participants 2265 Patients assessed for eligibility

Figure 2 .
Figure 2. Kaplan-Meier Curve of the Proportion of Patients Receiving Pathogen-Directed Treatment 1.00

JAMA
Network Open | Infectious Diseases Diagnostic Stewardship in Community-Acquired Pneumonia With Molecular Testing proportion of patients in the intervention arm received only a single dose of antibiotics, indicating the potential for reduced antibiotic exposure.

Table 1
). Patients included 153 females (40.9%) and 221 males (59.1%), with a median (IQR) age of 72 (60-79) years.Among these patients, 208 had a diagnosis of CAP, of whom 200 (97 in the intervention arm and 103 in the standard-of-care arm) provided an LRT sample.Baseline characteristics for CAP patients only (n = 200) are provided in eTable 4 in Supplement 2.

Table 1 .
Baseline Characteristics for All Randomized Patients a .7 percentage points (95% CI, 0.9-18.5 percentage points; P = 0.03) and an OR of 2.66 (95% CI, 1.07-7.33).Additionally, 4 of 97 patients (4.3%) in the intervention arm and none in the standard-of-care arm received only a single dose of antibiotics (absolute difference, 4.3 percentage points (95% CI, 0.2-8.4percentagepoints;P = 0.04).Further details on the comparison of rapid testing using FAP plus vs standard of care for patients with CAP are presented in Table3.A breakdown of pathogen-directed treatment within 48 hours for patients with CAP is provided in eTable 5 in Supplement 2. For the secondary outcomes on the provision of narrow-spectrum a Missing data (intervention: n = 10; standard of care: n = 7) were imputed using multiple imputations through chained equations.bTheCRPand WBC counts were the highest values during hospitalization.JAMA Network Open | Infectious DiseasesDiagnostic Stewardship in Community-Acquired Pneumonia With Molecular Testing JAMA Network Open.2024;7(3):e240830.doi:10.1001/jamanetworkopen.2024.0830(Reprinted) March 6, 2024 6/12 Downloaded from jamanetwork.comby guest on 03/22/2024 difference of 9

Table 3 .
Comparison of Rapid Testing for Patients With Community-Acquired Pneumonia Only by Treatment Arm