Nasal swab-guided deescalation of empirical linezolid in respiratory infections at a secondary-level hospital: quantification of quality gaps and antimicrobial stewardship opportunities.

BACKGROUND

Empirical linezolid for respiratory infections risks overtreatment and therapeutic inertia. Methicillin-resistant Staphylococcus aureus (MRSA) nasal swabs have high negative predictive value for ruling out MRSA pneumonia, yet their systematic integration into clinical practice remains poorly documented in Spain. The aim of the study was to quantify three sequential quality gaps in the empirical linezolid prescribing process and estimate the potential for improvement through an antimicrobial stewardship programme (ASP).

METHODS

A retrospective observational study of 83 patients receiving empirical linezolid for respiratory infections at a secondary-level hospital during five non-consecutive months of 2025. Three quality gaps were defined: Gap 0 (initiation without MRSA risk criteria); Gap 1 (failure to request a nasal swab in eligible patients); and Gap 2 (therapeutic inertia following a negative result).

RESULTS

Gap 0 affected 30.1% of patients, predominantly CAP (72%), discontinuation was more frequent in Gap 0 patients than in eligible candidates. Gap 1 affected 32.5%, with marked inter-unit variability: swabs were requested in 78% of Internal Medicine patients versus 41.7% in Respiratory Medicine. Among 56 patients with results, 85.7% were negative; Gap 2 affected 20.8%, generating 56 days of linezolid administered after a negative result was available. Patients with positive swabs had significantly higher cumulative risk scores than negative patients. Overall, 62.7% of patients presented ≥1 gap, accounting for 69.5% of 377 total linezolid-days.

CONCLUSIONS

Quality gaps in empirical linezolid use affect two-thirds of patients and are associated with nearly 70% of treatment days. Targeted two-level ASP interventions, explicit indication criteria, and systematic de-escalation prompts after negative swabs are warranted.