Microbial Keratitis: A Workup Approach

Last updated: October 2026

Microbial keratitis is a true ophthalmic emergency where the diagnostic workup and empirical treatment happen in parallel, not sequentially — waiting for culture results before starting treatment risks irreversible corneal scarring and vision loss.

For organisms, stains and culture media across all eye infections, see ocular microbiology in depth.

Epidemiology: bacteria dominate, but know the outliers

Recent large series consistently show bacteria as the leading cause (roughly 52-90% of culture-positive cases depending on region and population), followed by fungi and Acanthamoeba. Pseudomonas aeruginosa and Staphylococcus species are frequently implicated bacterial isolates; Fusarium is the leading fungal cause in many warm, humid, agricultural regions. Ocular trauma increases microbial keratitis risk roughly 5-fold, and agricultural activity roughly 1.3-fold — both worth asking about specifically in history-taking.

The workup: corneal scraping remains central

• Do not delay antibiotics to "clear" the culture first — the older practice of withholding antibiotics for 12 hours before scraping to improve culture yield is no longer recommended by most practitioners; obtain the scraping, then start empirical treatment immediately
• Smears: Gram stain (bacteria), Giemsa, acid-fast stain (mycobacteria/Nocardia), and KOH with calcofluor white (fungi and Acanthamoeba) — KOH/calcofluor white has notably higher sensitivity for fungi (~90%) and Acanthamoeba (~84%) than Gram stain has for bacteria (~57%), a useful fact when interpreting a negative smear
• For mild, small peripheral infiltrates (<1-2mm), many practitioners reasonably defer scraping and treat empirically, particularly where lab resources are limited
• Confocal microscopy is an increasingly valuable adjunct, allowing direct in-vivo visualization of fungal filaments and Acanthamoeba cysts in the stroma without waiting for culture
• Impression cytology (a cellulose acetate filter applied to the cornea rather than a blade) is an emerging alternative sample-collection method, with comparable or superior isolation rates to traditional scraping in recent large comparative studies

Clinical clues that should raise suspicion for Acanthamoeba

Acanthamoeba is rarely culture-positive on routine scraping, so recognising its distinct clinical pattern matters more than lab confirmation early on:

• Pseudodendrites (can be mistaken for HSV)
• Perineural infiltrates (radial keratoneuritis) — a relatively specific finding
• Ring-shaped infiltrate, particularly if it develops later in a case initially presumed bacterial or fungal
• Contact lens wear, especially with tap water exposure or poor lens hygiene
• Disproportionate pain relative to clinical signs early in the disease
• Longer symptom duration before presentation, and prior topical antibiotic use without improvement
• A non-healing, worsening ulcer despite appropriate antibacterial/antiviral therapy should always prompt reconsidering Acanthamoeba rather than simply escalating the presumed-bacterial or presumed-viral treatment

Empirical treatment paradigm

• Presumed bacterial: broad-spectrum fortified antibiotics covering both Gram-positive and Gram-negative organisms — fortified cefazolin plus a fluoroquinolone (e.g., moxifloxacin) is a commonly used combination
• Presumed fungal: topical natamycin 5% (particularly effective for filamentous fungi like Fusarium) or voriconazole; oral voriconazole may be added for deeper/severe disease
• Presumed Acanthamoeba: topical PHMB (polyhexamethylene biguanide) and/or propamidine, often combined with a fluoroquinolone for bacterial co-coverage; prolonged treatment (often continued 2+ weeks beyond symptom resolution) is standard given the cyst stage's resistance to treatment
• Therapy is refined once culture and sensitivity results return, narrowing from broad empirical coverage to targeted treatment
• Fungal keratitis carries the worst prognosis of the three in comparative outcome data — a meaningfully higher proportion require surgical intervention (roughly 51% vs. 43% for bacterial) and achieve a healed scar less often, reinforcing why accurate early fungal diagnosis matters so much

2026 update: what the big trials showed

Sources for this update: Srinivasan M, et al. Corticosteroids for bacterial keratitis (SCUT). Arch Ophthalmol. 2012;130:143-150 | Prajna NV, et al. Natamycin vs voriconazole for fungal keratitis (MUTT I). JAMA Ophthalmol. 2013;131:422-429.

Key references: Five-Year Analysis of Microbial Keratitis Incidence and Sensitivity, South West England. 2025.  |  Review of epidemiological features, diagnosis and treatment outcome of microbial keratitis.  |  Emerging diagnostic modalities in microbial keratitis: beyond culture and smear. 2025.