Bacterial keratitis is a potentially sight-threatening corneal infection that is commonly treated with intensive topical antibiotics. Despite appropriate therapy, some cases show inadequate clinical response, particularly when the infection involves the deep corneal stroma. Limited penetration of topical antibiotics into deeper corneal layers may contribute to treatment failure in these recalcitrant cases.

Intrastromal antibiotic injection is a targeted drug-delivery approach that allows high local antimicrobial concentrations directly at the site of infection. Moxifloxacin is a broad-spectrum fluoroquinolone with proven efficacy in bacterial keratitis and favorable corneal tissue penetration. However, evidence regarding the clinical benefit and safety of intrastromal moxifloxacin as an adjunctive treatment remains limited.

This randomized controlled trial aims to evaluate the efficacy and safety of intrastromal moxifloxacin injection as an adjunct to standard topical moxifloxacin therapy compared with topical therapy alone in patients with recalcitrant bacterial keratitis. The primary outcome is time to complete clinical resolution of infection. Secondary outcomes include visual acuity improvement, ulcer healing rate, need for additional interventions, and treatment-related complications.

Inclusion Criteria:

• • Age ≥ 18 years

  • Clinical diagnosis of bacterial keratitis confirmed by corneal scraping and microbiology
  • Recalcitrant keratitis defined as no significant clinical improvement after 48-72 hours of intensive topical antibiotic therapy
  • Ability to provide written informed consent

Exclusion Criteria

• Fungal, viral, or acanthamoeba keratitis
• Corneal perforation or impending perforation
• Known hypersensitivity to fluoroquinolones
• Pregnancy or lactation
• Immunocompromised state or current systemic immunosuppressive therapy
• Previous intrastromal or intracameral antibiotic injection for the same episode

(IPD) will not be shared because the study was not designed with a data-sharing framework, and participants did not provide explicit consent for secondary use or external sharing of their individual-level data. In addition, sharing IPD may pose potential risks to participant privacy, even after de-identification, particularly in single-center studies with limited sample sizes. Current clinical trial registration requirements mandate transparency regarding data-sharing intentions but do not obligate investigators to share IPD. Therefore, the decision not to share IPD is made in accordance with ethical standards, privacy regulations, and applicable registry policies.

Randomization will be performed using a computer-generated random sequence. Allocation will be concealed using sealed opaque envelopes.