Effect of cataract surgery on the corneal endothelium: modern phacoemulsification compared with extracapsular cataract surgery.

PURPOSE: To investigate whether modern phacoemulsification surgery results in more damage to the corneal endothelium than extracapsular cataract extraction (ECCE), and to examine which preoperative, operative, and postoperative factors influence the effect of cataract surgery on the endothelium. DESIGN: Randomized controlled trial. PARTICIPANTS: Five hundred patients 40 years or older were randomized into 2 groups (ECCE, 249; phacoemulsification, 251). METHODS: Central corneal endothelial cell counts, coefficient of variation of cell size, and hexagonality were assessed before surgery and up to 1 year postoperatively. MAIN OUTCOME MEASURE: Endothelial cell count. RESULTS: Four hundred thirty-three patients completed the trial. The initial preoperative mean cell count for the entire sample was 2481 (standard error [SE]: 18.6), reduced at 1 year postoperatively to 2239 (SE: 23.5). An average 10% reduction in cell count was recorded by 1 year postoperatively. There was no such change in hexagonality or in the coefficient of variation. There was no significant difference in overall percentage cell loss between the 2 treatment groups. Factors associated with excessive cell loss (> or =15% by 1 year) were a hard cataract (odds ratio [OR]: 2.1, 95% confidence limits: 1.1-4.1; P = 0.036), age (OR: 1.04, P = 0.005), and capsule or vitreous loss at surgery (OR: 2.38, P = 0.106). Phacoemulsification carried a significantly higher risk (OR: 3.7, P = 0.045) of severe cell loss in the 45 patients with hard cataracts relative to ECCE (52.6% vs. 23.1%; chi-square test, P = 0.041), with both procedures achieving similar postoperative visual acuity outcomes. CONCLUSIONS: No significant difference in overall corneal endothelial cell loss was found between these 2 operative techniques. The increased risk of severe cell loss with phacoemulsification in patients with hard cataracts suggests that phacoemulsification may not be the optimal procedure in these cases, and that ECCE should be preferred.