‘cupping’ parameters had the highest sensitivities andspecificities but no single cut-off criterion for these param-eters had both high sensitivity and specificity. Multiplelogistic regression modeling indicated that two disc param-eters in combination, narrowest rim width and peripapil-lary atrophy, were independently associated with glaucomaand achieved best prediction. Combining informationfrom these parameters achieved a sensitivity of 81% and aspecificity of 90%, a level of discrimination that is notsignificantly better than narrowest rim width alone interms of area under the receiver operator characteristiccurve. The authors conclude that disc assessment duringclinical examination with direct ophthalmoscopy canachieve comparable sensitivities and specificities withthose previously reported for stereophotographic assess-ment or visual fields screening. However, since the diag-nostic accuracy of disc assessment in isolation isinadequate for screening, they suggest a combined teststrategy is necessary.—Thomas J. Liesegang

*Academic Department of Ophthalmology, Manchester Royal Eye Hos-pital, Oxford Road, Manchester, M13 9WH, United Kingdom. E-mail:robert.harper@man.ac.uk

● Recurrent erosions of the cornea—Epidemiology andtreatment. Reidy JJ,* Paulus MP, Gona S. Cornea. 2000;19:767–771.

THROUGH A RETROSPECTIVE CHART REVIEW THE AU-

thors evaluated the epidemiologic characteristics of alarge clinical population of patients with recurrent erosionsof the cornea and the efficacy of different modalities oftreatment. All patients with the diagnosis of recurrentcorneal erosion treated between January 1990 and Decem-ber 1998 were reviewed with clinically confirmed recurrenterosions identified in 104 patients. There were 36 malesand 68 females. A history of trauma was present in 47patients (45%), 30 patients (29%) had epithelial basementmembrane dystrophy (EBMD), and 18 patients (17%) hadboth a history of trauma and evidence of EBMD. Morethan 87% of all erosions occurred on the inferior third ofthe cornea. Conservative therapy was used as the primarytreatment in 52 patients with a recurrence rate of 6%.Corneal stromal micropuncture was performed on 38patients with a recurrence rate of 40%. Eleven patients hadepithelial debridement with a recurrence rate of 18%. Fourpatients had a superficial keratectomy with a diamond burrwith a recurrence rate of 25%. Only one patient had anexcimer phototherapeutic keratectomy, and she had aminor recurrence posttreatment. In this series, the distri-bution of trauma and EBMD in patients with recurrenterosions of the cornea was roughly equivalent. The authorsconclude that conservative therapy was effective in ap-proximately one half of the patients. All surgical treatmentmodalities were associated with recurrences. Those pa-

tients with both EBMD and trauma were more likely tohave a recurrence after treatment.—Thomas J. Liesegang

*Department of Ophthalmology, The State University of New York atBuffalo, Erie County Medical Center, 462 Grider St, Room CC133,Buffalo, NY 14215.

● The repeatability of corneal thickness measures. Mar-sich MM, Bullimore MA.* Cornea. 2000;19:792–795.

THE AUTHORS COMPARED THE REPEATABILITY OF THREE

measures of corneal thickness: Orbscan Slitscanpachymetry, ultrasound pachymetry, and optical pachym-etry. Twenty normal subjects were tested on three occa-sions. Two occurred on the same day and the third was ona different day at approximately the same time of day asone of the first two visits. Central corneal thickness of theright eye was measured with a Haag-Streit opticalpachymeter, a Humphrey Model 855 ultrasoundpachymeter, and the Orbscan system. Day-to-day andsame-day repeatability was assessed by calculating thedifference between the values from two visits and deter-mining the mean difference, the SD, and the 95% Limitsof agreement (LoA). Mean central corneal thickness asmeasured by each instrument was as follows: 539 6 33mmm for optical pachymetry, 542 6 33 mmm for ultra-sound pachymetry, and 596 6 30 mmm for Orbscanpachymetry. For day-to-day comparisons, optical pachym-etry showed the poorest repeatability with 95% LoA of261 to 132 mum. Ultrasound pachymetry showed betterrepeatability with 95% LoA of 222 to 124 mum. TheOrbscan showed the best repeatability centrally with 95%LoA of 210 to 117 mum. Peripheral Orbscan pachymetrywas less repeatable than that measured centrally but stillmore repeatable than central optical pachymetry. Similarresults were found with same-day comparisons. The au-thors conclude that the Orbscan system is the mostrepeatable technique for measuring corneal thickness butshows a significant bias toward greater corneal thicknessmeasures than both ultrasound and optical pachymetry.—Thomas J. Liesegang

*Ohio State University, College of Optometry, 338 W 10th Ave,Columbus, OH 43210

● The role of glaucoma therapy in the need for repeatpenetrating keratoplasty. Aldave AJ, Rudd JC, CohenEJ,* Rapuano CJ, Laibson PR. Cornea. 2000;19:772–776.

THE AUTHORS SOUGHT TO EVALUATE THE EFFECTS OF

medical and surgical therapy for glaucoma in patientsrequiring repeat penetrating keratoplasty (PK) for endo-thelial failure. They performed a retrospective review ofthe charts of all patients undergoing repeat PK at theCornea Service at Wills Eye Hospital between January 1,

ABSTRACTSVOL. 131, NO. 3 413