Comparison of culture, confocal microscopy and PCR in routine hospital use for microbial keratitis diagnosis

Aims To evaluate the sensitivity and specificity of polymerase chain reaction (PCR), in vivo confocal microscopy (IVCM) and culture for microbial keratitis (MK) diagnosis. Methods Retrospective review of PCR, IVCM and culture results for MK diagnosis at Moorfields Eye Hospital between August 2013 and December 2014. Results PCR results were available for 259 MK patients with concurrent culture for 203/259 and IVCM for 149/259. Sensitivities and specificities with 95% confidence intervals [95% CI] were calculated for Acanthamoeba keratitis (AK) and fungal keratitis (FK), by comparison with culture, for both IVCM and PCR. For AK, FK and bacterial keratitis (BK) sensitivities were calculated, for each diagnostic method, by comparison with a composite reference standard (a positive result for one or more of culture, PCR or IVCM having a specificity of 100% by definition). For the latter, sensitivities with [95% CI] were: for AK, IVCM 77.1% [62.7–88.0%], PCR 63.3% [48.3–76.6%], culture 35.6 [21.9–51.2]; for FK, IVCM 81.8% [48.2–97.7%], PCR 30.8% [9.09–61.4%], culture 41.7% [15.2–72.3%]; for BK, PCR 25.0% [14.7–37.9%], culture 95.6% [87.6–99.1%]. Conclusion IVCM was the most sensitive technique for AK and FK diagnosis but culture remains our gold standard for BK. These findings reflect results to be expected from service providers to UK ophthalmology units and demonstrates the need at our centre for ongoing diagnostic result audit leading to the potential to improve PCR diagnosis. Both FK and AK are now common in the UK; ophthalmology units need to have all these techniques available to optimise their MK management.