Multiplex real-time PCR in non-invasive respiratory samples to reduce antibiotic use in community-acquired pneumonia: a randomised trial

Community-acquired pneumonia (CAP) is a leading cause of morbidity, mortality, and antibiotic consumption among hospitalized adults. Causative pathogens often remain unidentified, contributing to antibiotic overuse, antimicrobial resistance, and antibiotic-related adverse events. Current CAP guidelines do not recommend multiplex PCR pneumonia panels for initial diagnostic testing, and guidance on conventional tests (sputum culture, blood cultures, urinary antigens) is inconsistent and based on low-quality evidence. The study’s research question was whether adding multiplex real-time PCR on non-invasive respiratory samples to conventional microbiological testing reduces antibiotic use, measured by days of antibiotic therapy (DOT), compared with conventional testing alone in hospitalized adults with CAP. The trial also aimed to evaluate safety and secondary clinical outcomes.

Guidelines (e.g., ATS/IDSA, BTS) have not incorporated multiplex PCR panels for initial CAP diagnostics, citing limited evidence and variability in conventional testing utility. Prior work shows rapid syndromic molecular testing can increase microbial yield and reduce time to results, yet its impact on antibiotic stewardship endpoints in CAP remains uncertain. Existing studies have highlighted both the potential diagnostic advantages of multiplex PCR and the discordance between molecular quantification and culture-based bacteriology, underscoring the need for randomized trials to assess clinical impact on antibiotic exposure and outcomes.

Design: Randomized, controlled, open-label, parallel-group trial (RADICAP) across four Spanish reference hospitals. Recruitment: February 20, 2020–April 24, 2023. Ethics/registration: Approved by AEMPS (PRX-0385) and Elche University Hospital Ethics Committee (PRHL/418); SPIRIT- and CONSORT-compliant; ClinicalTrials.gov NCT01458492 (text elsewhere also cites NCT04158492) and EudraCT 2018-004580-29. Participants: Adults (≥18 years) with CAP diagnosed in the emergency department; screened within 24 h. Key exclusions included immunosuppression, anticipated death <24 h, participation in another trial, pregnancy/breastfeeding, and acute SARS-CoV-2 infection. Randomization/Groups: 1:1 allocation to (a) multiplex real-time PCR on non-invasive respiratory samples plus conventional microbiological testing or (b) conventional microbiological testing alone. PCR samples targeted sputum (including induced) or nasopharyngeal swabs. Interventions/Diagnostics: Multiplex PCR used the BioFire FilmArray Pneumonia Plus panel detecting typical bacteria and respiratory viruses, with semi-quantitative bacterial load reporting; S. pneumoniae positivity threshold ≥10² CFU/mL. Atypical bacteria reported as detected/not detected. Conventional testing included Gram stain and culture of high-quality sputum (criteria: <10 squamous cells and >25 leukocytes per low-power field), two sets of blood cultures, pleural fluid culture when available, S. pneumoniae urinary antigen (Binax), and L. pneumophila serogroup 1 urinary antigen (immunoenzymatic assay). Results were returned via the electronic medical record to treating clinicians; clinical management decisions were made by attending physicians, not investigators. Endpoints: Primary—days of antibiotic therapy (DOT) for the CAP episode (inpatient and outpatient), summing DOT across agents; new treatment counted if antibiotic-free interval >48 h. Secondary—length of antibiotic therapy (LOT), time to IV-to-oral switch, time to etiological diagnosis from admission, antimicrobial de-escalation to narrower spectrum, time to clinical stability, ICU admission, mechanical ventilation days, 30-day readmission, and mortality (48 h and 30 days), plus antibiotic-related adverse events. Follow-up/Data collection: Daily inpatient assessments; 30-day post-discharge clinic visit or phone interview; adverse events recorded per CTCAE; data captured in REDCap; DSMB blinded to group. Sample size/statistics: Assumed DOT ~8 days when etiology known, ~11 when unknown; targeted difference of 2 DOT; planned n=440 (220/group), α=0.05, power >80% allowing 10% dropout. Interim analysis (March 27, 2023) showed observed difference ~1 DOT with conditional power ~40%; DSMB recommended stopping for futility; recruitment halted April 24, 2023. Analyses by intention-to-treat and per-protocol; statistical tests included t-test or Wilcoxon rank-sum for continuous variables and chi-squared for categorical; adjusted analyses considered age, sex, Charlson score, and PSI score; analyses conducted in R (v4.0.5).

Enrollment and populations: 353 assessed; 242 randomized (119 PCR+conventional; 123 conventional only); ITT n=242; per-protocol n=230. In the PCR arm, PCR was performed in 118/119 (99.2%): sputum 77 (65.2%), nasopharyngeal swab 41 (34.7%). Primary endpoint (DOT): Median DOT 10.04 days (IQR 7.98–12.94) with PCR+conventional vs 11.33 (IQR 8.15–16.16) with conventional; difference −1.04 days (95% CI −2.42 to 0.17); p=0.093 (not statistically significant). Secondary endpoints: - LOT: 9.00 days (7.42–11.00) vs 8.76 (6.92–12.73); difference 0.12 (95% CI −0.79 to 0.96); p=0.75 (NS). - Time to IV-to-oral switch: 1.17 days (0.58–3.17) vs 2.67 (0.85–6.49); difference −0.87; p=0.011 (faster with PCR). - Time to etiological diagnosis from admission: 0.17 days (0.13–0.25) vs 1.57 (0.94–2.90); difference −1.32; p<0.0001 (faster with PCR). - De-escalation to narrower spectrum: 19.3% vs 24.4%; p=0.425 (NS). - Time to clinical stability: 2.50 days (1.00–6.00) vs 2.00 (1.00–5.00); p=0.408 (NS). - ICU admission: 1.68% vs 8.94%; p=0.026 (fewer in PCR group). - Mechanical ventilation days: 11 vs 11; p=1.00 (NS). - Antibiotic-related side effects: 5.04% vs 4.07%; p=0.955 (NS). - 30-day readmission: 19.24% vs 5.69%; p=0.419 (NS; reporting note indicates investigators not involved in readmission decisions). - Mortality: 48 h 0% vs 1.63% (p=0.498); 30-day 1.68% vs 3.25% (p=0.683); no significant differences. Microbiological yield: Etiology established in 63.9% (76/119) with PCR+conventional vs 26.2% (32/123) with conventional; difference 27.85 percentage points (95% CI 25.42–30.28); p<0.001. Common pathogens: Streptococcus pneumoniae, Legionella pneumophila, Haemophilus influenzae. Polymicrobial and viral etiologies more frequent with PCR. Gram-negative bacilli, including Pseudomonas aeruginosa, were uncommon; 3 S. aureus cases (all non-MRSA) occurred in the PCR arm.

The trial tested whether adding multiplex real-time PCR on non-invasive respiratory samples to conventional testing reduces antibiotic exposure in hospitalized CAP. Despite substantially improving etiologic yield and shortening time to pathogen identification and IV-to-oral switch, multiplex PCR did not significantly reduce days of antibiotic therapy or length of therapy. Safety outcomes, adverse events, and 30-day mortality were similar between groups. Fewer ICU admissions were observed in the PCR arm, but other clinical outcomes showed no significant differences. These results indicate that, within routine care where antibiotic decisions remain at clinicians’ discretion, enhanced and expedited diagnostics alone did not translate into shorter antibiotic courses. Accordingly, the findings do not support routine implementation of multiplex PCR panels for initial microbiological testing in hospitalized CAP patients.

In hospitalized adults with CAP, multiplex real-time PCR added to conventional microbiological testing markedly increased pathogen detection and accelerated etiologic diagnosis and IV-to-oral transitions but did not significantly reduce antibiotic days or improve most clinical outcomes. The study does not support routine incorporation of multiplex PCR into initial diagnostic testing for CAP. Future research should evaluate targeted use in specific subgroups (e.g., high-severity or atypical presentations), integrate rapid diagnostics with structured antibiotic stewardship and decision-support pathways to influence prescribing, and assess cost-effectiveness and impact on resistance and downstream outcomes.

The trial was open-label and stopped early for futility following interim analysis, yielding a sample size smaller than planned and potentially underpowering detection of modest effects. Recruitment and hospital operations were affected by the COVID-19 pandemic, including restricted investigator access to COVID-specific areas. The study was conducted at four Spanish centers, which may limit generalizability. Non-invasive samples (sputum or nasopharyngeal swabs) were used for PCR, with variable yield across sample types. Immunosuppressed patients and selected high-risk groups were largely excluded. Clinical decisions were made by attending physicians without a mandated stewardship protocol linked to PCR results, which may have diluted potential impacts on antibiotic duration.