A Cytomegalovirus Retinitis Screening Program: Evaluation of Enrollment Criteria for HIV Patients in Singapore.

PURPOSE: Identification of optimal enrollment criteria for a CMVR screening program suitable for a resource-limited environment. METHODS: A prospective audit was performed on newly diagnosed HIV patients referred for CMVR screening with any of the following four criteria: (1) visual symptoms, (2) low CD4(+) counts (<50 cells/µL), (3) AIDS-defining illnesses (ADI), and/or (4) opportunistic infections (OI). Odds ratios for each of the demographic factors and enrollment criteria were calculated. Sensitivities, specificities, and workload reduction for the various combinations were determined. RESULTS: A total of 348 screening visits for 176 HIV patients were performed. While individually only ADI was statistically significant for increased CMVR risk, the combination of CD4(+) counts <50 cells/µL with either ADI or visual symptoms or all 3 criteria were also statistically significant. Two enrollment criteria, ADI and ADI with CD4(+) <50 cells/µL, demonstrated good sensitivities, specificities, and workload reduction. CONCLUSION: We propose ADI and possibly CD4(+) counts <50 cells/µL as enrollment criteria for CMVR screening.

Telemedicine screening for cytomegalovirus retinitis using digital fundus photography.

BACKGROUND: Screening for cytomegalovirus retinitis (CMVR) is important in patients with acquired immune deficiency syndrome and low CD4(+) counts. However, many human immunodeficiency virus (HIV) patients lack access to specialist ophthalmological care. Telemedicine screening is a cost-effective method for screening these patients. We aimed to report the use of composite nine-field digital fundus photography (DFP) images for CMVR screening. We report its sensitivity and specificity in detecting CMVR and the level of agreement with gold-standard binocular indirect ophthalmoscopy. MATERIALS AND METHODS: An audit was performed on our national CMVR screening program that screened all HIV patients referred to the Ophthalmology Department at Tan Tock Seng Hospital, Singapore. All patients underwent retinal screening with DFP. Images were categorized as CMVR-positive, CMVR-negative, suspicious, or unreadable by blinded retinal specialists. Patients subsequently underwent dilated gold-standard indirect ophthalmoscopy by a different retinal specialist. Diagnoses were categorized as CMVR-positive, CMVR-negative, or unreadable. Sensitivity and specificity of retinal findings on DFP and kappa values for level of agreement between the two screening methods were calculated. RESULTS: Three hundred seventy screenings on 188 patients were performed. Twenty-three eyes diagnosed with CMVR on indirect ophthalmoscopy were also identified on DFP (100% sensitivity). A 99.9% specificity was achieved. The fundus photograph of one eye without CMVR was read as CMVR-positive because of an artifact, accounting for a false-positive. Kappa values ranged from 0.739 to 0.987. CONCLUSIONS: DFP is a sensitive and specific method of screening HIV patients for CMVR and has a high level of agreement with indirect ophthalmoscopy.