Diffuse lamellar keratitis: incidence, associations, outcomes, and a new classification system

To evaluate the incidence, associations, and visual outcomes in patients with diffuse lamellar keratitis (DLK) after laser in situ keratomileusis (LASIK).

University-based refractive surgery center, Boston, Massachusetts, USA.

This retrospective review comprised 2711 eyes that had LASIK between September 1996 and September 1999. All eyes that developed DLK after LASIK were included. They were divided into type I DLK (center sparing) or type II DLK (center involved) and then subdivided into A (sporadic-DLK not diagnosed in other patients treated on the same day) or B (cluster-other patients identified with DLK). Type IA corresponded to center sparing, sporadic; type IB, center sparing, cluster; type IIA, center involved, sporadic; and type IIB, center involved, cluster. The main outcome measures were incidence of DLK after LASIK, time to diagnosis, time to resolution, and changes in best spectacle-corrected visual acuity (BSCVA). Unpaired t tests were used for statistical analyses.

Thirty-six eyes (1.3%) developed DLK. Type I occurred in 58.3% of cases (type IA, n = 18; type IB, n = 3) and type II, in 41.7% (type IIA, n = 10; type IIB, n = 5). The mean time to diagnosis was not statistically significantly different between type I (1.8 days) and type II (1.1 days). Fourteen eyes (38.9%) developed DLK after an epithelial defect, representing an odds ratio of 13 times. The association with an epithelial defect was statistically significantly greater with type I (11/21 eyes, 52.4%) than with type II (3/15 eyes, 20.0%; P =.05). The mean time to resolution was 3.5 days in type I (type IA = 3.6 days; type IB = 2.7 days). This was significantly shorter than in type II, which had a mean time to resolution of 12.1 days (type IIA = 9.3 days; type IIB = 10.2 days) (P =.001). Loss of 2 or more lines of BSCVA occurred in 2 of 5 patients with type IIB and in no patients with types IA, IB, or IIA.

Epithelial defects after LASIK increased the risk of DLK occurrence, especially type I. Type II DLK was associated with a prolonged time to resolution and carried a significantly higher risk of BSCVA loss than type I.