Twenty-Two Months of Syndromic Multiplex-PCR Testing for Acute Infections in a Southern Italian Hospital: Pathogen Epidemiology, Diagnostic Appropriateness and the Cost of Negative Results

Syndromic multiplex-PCR panels deliver rapid, comprehensive detection of pathogens and resistance determinants in acute infections, but their unrestricted use can generate a high proportion of negative results with substantial economic and stewardship implications. We retrospectively analysed all BioFire FilmArray meningitis/encephalitis (ME), upper-respiratory (RP2.1), pneumonia (PN) and gastrointestinal (GI) determinations performed at the UOSD Microbiology and Virology of P.O. “Vito Fazzi”, ASL Lecce (Apulia, Italy) between 1 July 2024 and 30 April 2026. Repeat determinations from the same patient with the same panel were identified in the laboratory information system and removed before any analysis, so that each determination analysed corresponds to a distinct diagnostic episode. Across 5381 determinations (CNS, n = 761; upper respiratory, n = 2601; lower respiratory, n = 399; gastrointestinal, n = 1620), 59.3% were negative for every target, with a steep appropriateness gradient: 90.0% negativity for the ME panel and 77.1% for the gastrointestinal panel versus 44.2% for the upper-respiratory and 26.6% for the pneumonia panel. CNS positives were predominantly viral (75%), led by enterovirus; Streptococcus pneumoniae was the only consistent bacterial agent. Human rhinovirus/enterovirus dominated the respiratory ecology; influenza A was almost entirely H3 and H1N1pdm09, but 8.1% of influenza-A–positive specimens were equivocal or non-subtypeable and were never referred for sequencing, an avoidable surveillance blind spot for novel/zoonotic (avian) influenza. Pneumonia-panel resistance markers (mecA/C–MREJ, CTX-M, KPC, NDM) clustered in Enterobacterales co-infections. The gastrointestinal panel was dominated by diarrhoeagenic Escherichia coli pathotypes (chiefly EPEC and EAEC) and Clostridioides difficile toxin, its 77% negativity identifying a second over-utilised stream. We argue for CSF-pleocytosis gating, tiered/reflex respiratory algorithms, gastrointestinal-panel gating to community-onset diarrhoea, and mandatory reflex sequencing of unsubtypeable influenza A.