REFERENCES 1. Patel CK, Chen SDM, Farmery AD. Optical coherence tomography under general anesthesia in a child with nystag- mus. Am J Ophthalmol 2004;137:1127–1129. 2. Hartnett ME, McColm JR. Retinal features predictive of progressive stage 4 retinopathy of prematurity. Retina 2004; 24:237–241. 3. Murata T, Nakagawa K, Khalil A, et al. The temporal and spatial vascular endothelial growth factor expression in retinal vasculogenesis of rat neonates. Lab Invest 1996;74:68 –77. 4. Young TL, Anthony DC, Pierce E, et al. Histopathology and vascular endothelial growth factor in untreated and diode laser-treated retinopathy of prematurity. J AAPOS 1997;1:105–110. Correlation Between Clinical Suspicion and Polymerase Chain Reaction Verification of Infectious Vitritis Nisha Acharya, MS, MD, Thomas Lietman, MD, Vicki Cevallos, BS, MT, John P. Whitcher, MD, MPH, Michael Saidel, MD, Donald Stone, MD, Jacque Duncan, MD, and Todd P. Margolis, MD, PhD PURPOSE: To compare polymerase chain reaction (PCR) results to presumptive clinical diagnosis in patients with vitritis. DESIGN: Retrospective review of PCR laboratory records from vitreous samples. METHODS: Fifty consecutive laboratory records of vitre- ous samples sent for PCR testing were reviewed. Three reviewers with uveitis training ranked the clinical suspi- cion of a specific diagnosis using a classification system (scale of 1 to 4) and were masked to the PCR results. RESULTS: The degree of clinical suspicion of a particular diagnosis was significantly associated with a positive PCR result (P  .048). Higher clinical suspicion was signifi- cantly more associated with a positive PCR result com- pared with cases with lower clinical suspicion (P  .01). CONCLUSIONS: If the clinical suspicion of a specific diag- nosis is low, the PCR for any infectious etiology is unlikely to be positive. (Am J Ophthalmol 2006;141: 584 –585. © 2006 by Elsevier Inc. All rights reserved.) P OLYMERASE CHAIN REACTION (PCR) BASED ASSAYS OF aqueous and vitreous specimens are used in the diag- nostic evaluation of patients with infectious retinitis and uveitis. 1–5 The Proctor Foundation laboratory conducts PCR testing for herpes simplex virus (HSV), varicella zoster virus (VZV), cytomegalovirus (CMV), and toxoplas- mosis. Prior reports have described the PCR techniques and validity of results from this laboratory. 1,2,5 Given that PCR testing is both expensive and involves some risk, our aim was to determine whether particular clinical situations are associated with a higher rate of PCR positivity. Specifically, is clinical suspicion of a specific diagnosis correlated with the rate of PCR positivity? The Institutional Review Board at UCSF approved this study. Proctor laboratory records from all vitreous samples sent for PCR testing from January 1, 2002 until May 1, 2005 were collected (n = 50). Each laboratory record consisted of a requisition sheet that described the patient’s history and PCR tests being requested, and a separate sheet with the PCR results. Before reviewing the laboratory records, a classification system to rank the requesting clinician’s suspicion of a specific diagnosis was created. A scale from 1 (highest suspicion) to 4 (lowest suspicion) was used (Table 1). Requisition sheets were given to three uveitis-trained reviewers who ranked the clinical suspicion of a specific diagnosis based on the classification system. Reviewers were masked to the PCR results. The median ranking for each patient was then determined, and a 4  2 contingency table was created. PCR positivity was significantly associated with clinical suspicion (P = .048, 2  4 Fisher’s exact test). A 2  2 contingency table was created by combining groups 1 and 2 (higher suspicion) together and 3 and 4 (lower suspicion) together (Table 2). Since the 2  4 contingency table can be partitioned in numerous ways, we considered a Bonferroni-adjusted P value of .05/3 to be significant, given 3 degrees of freedom in the original table. This revealed that a higher clinical suspicion was significantly associated with a positive PCR result (P = .01, 2  2 Fisher’s exact test). We had 14 positive PCR results out of 50 vitreous specimens. Of these, seven had toxoplasmosis, three had CMV, three had VZV, and one had HSV. In 11 of the 14 cases, PCR results matched the requesting physician’s presumed diagnosis. Of the three remaining PCR-positive cases, two had a history of ongoing vitritis with no presumed diagnosis, and the third had a clinical diagnosis of ARN but the PCR was positive for toxoplasmosis DNA. The median age in the PCR-negative group was 53 years with a range of 7 to 83 years. In the PCR-positive group, the median age was 62 years with a range of 32 to 90 years. In the PCR-negative group, 47% (n = 17) of the patients Accepted for publication Oct 6, 2005. From the Proctor Foundation, Department of Ophthalmology (J.D.), University of California, San Francisco, San Francisco, California (N.A., T.L., V.C., J.P.W., M.S., D.S., T.P.M.). Grant support for this project: EY10008, EY02162 grants are from the National Institutes of Health, National Eye Institute, Bethesda, Mary- land. Research to Prevent Blindness Senior Scientist Award (Todd P. Margolis, MD, PHD). Research to Prevent Blindness Scientist Award (Jacque Duncan, MD). Both Awards are from the Research to Prevent Blindness Foundation in New York, New York. Inquiries to Nisha Acharya, MS, MD, Proctor Foundation, Department of Ophthalmology, University of California, San Francisco, 95 Kirkham St, San Francisco, CA 94143; e-mail: nisha@stanfordalumni.org AMERICAN JOURNAL OF OPHTHALMOLOGY 584 MARCH 2006