Moxifloxacin susceptibility mediates the relationship between causative organism and clinical outcome in bacterial keratitis

Abstract

Purpose: Bacterial keratitis is a sight-threatening infection of the cornea that is one of the leading causes of blindness globally. In this report, we analyze the role of moxifloxacin susceptibility in the relationship between causative organisms and clinical outcome in bacteria keratitis.

Methods: A mediation analysis is used to assess the role of moxifloxacin susceptibility in the relationship between causative organisms and clinical outcome in bacterial keratitis using data collected in a randomized, controlled trial.

Results: In the Steroids for Corneal Ulcers Trial (SCUT), 500 corneal infections were treated with topical moxifloxacin. The outcome of 3-week best spectacle-corrected visual acuity was significantly associated with an organism (Streptococcus pneumoniae, Pseudomonas aeruginosa, etc., P = 0.008). An indirect effects mediation model suggests that MIC accounted for approximately 13% (95% confidence interval, 3%-24%, P = 0.015) of the effect of the organism on 3-week visual acuity.

Conclusions: Moxifloxacin mediates the relationship between causative organisms and clinical outcome in bacterial keratitis, and is likely on the causal pathway between the organism and outcome. (ClinicalTrials.gov number, NCT00324168.).

Conflict of interest statement

Disclosure: C.E. Oldenburg, None; P. Lalitha, None; M. Srinivasan, None; P. Manikandan, None; M.J. Bharathi, None; R. Rajaraman, None; M. Ravindran, None; J. Mascarenhas, None; N. Nardone, None; K.J. Ray, None; D.V. Glidden, None; N.R. Acharya, Alcon (F); T.M. Lietman, Alcon (F)

Figures

Figure

Mediation diagram depicting relationship between causative organism, MIC, and BSCVA at 3 weeks from enrollment. Causative organism is the independent variable (IV); BSCVA at 3 weeks is the dependent variable (DV); and MIC is the mediator (M). P values and corresponding statistical tests associated with each arrow show each individual relationship, and the overall relationship of the mediator on the relationship between the independent and dependent variables is shown in bold in the middle of the diagram. The mediation analysis can be represented by the following equations: Where Equation 1 is the total effect model (effect of organism on outcome without adjusting for MIC where i1 is the Y-intercept, e1 is the amount of 3-week BSCVA that is not explained by organism, and c is the effect of organism on 3-week BSCVA; Equations 2 and 3 consist of the mediation model, with c' representing the effect of organism on 3-week BSCVA adjusted for MIC, a is the effect of organism on MIC, b is the effect of MIC on 3-week visual acuity, e2 represents the amount of 3-week BSCVA that is not due to organism, e3 represents the amount of MIC that is not explained by organism, and i2 and i3 are the Y-intercepts for Equations 2 and 3, respectively.