The risk of microbial keratitis with overnight corneal reshaping lenses

To estimate the incidence of microbial keratitis (MK) associated with overnight corneal reshaping contact lenses and to compare rates in children and adults.

A retrospective study of randomly selected practitioners, stratified by order volume and lens company, was conducted. Practitioners were invited to participate and those agreeing were asked to provide deidentified patient information for up to 50 lens orders and to complete a comprehensive event form for any of these patients who have attended an unscheduled visit for a painful red eye. Duration of contact lens wear was calculated from the original fitting date or January 2005 (whichever was later) to when the patient was last seen by the practitioner wearing the lenses on a regular basis. Cases of MK were classified by majority decision of a 5-member expert panel.

For the 191 practitioners who could be contacted, 119 (62%) agreed to participate. Subsequently, 11 withdrew, 22 did not respond, and 86 (43%) returned completed forms corresponding to 2202 lens orders and 1494 patients. Limiting the sample to those patients with at least 3 months of documented contact lens wear since 2005 resulted in a sample of 1317 patients; 640 adults (49%) and 677 children (51%) representing 2599 patient-years of wear (adults = 1164; children = 1435). Eight events of corneal infiltrates associated with a painful red eye were reported (six in children and two in adults). Two were classified as MK. Both occurred in children but neither resulted in a loss of visual acuity. The overall estimated incidence of MK is 7.7 per 10,000 years of wear (95% confidence interval [CI] = 0.9 to 27.8). For children, the estimated incidence of MK is 13.9 per 10,000 patient-years (95% CI = 1.7 to 50.4). For adults, the estimated incidence of MK is 0 per 10,000 patient-years (95% CI = 0 to 31.7).

The risk of MK with overnight corneal reshaping contact lenses is similar to that with other overnight modalities. The fact that the CIs for the rates estimated overlap should not be interpreted as evidence of no difference. True differences fewer than 50 cases per 10,000 patient-years were beyond the study's power of detection.