Purchase this article with an account.
ER de la Rua, JC Pastor, A Mayo, J Aragon, RM Sanabria, A Giraldo, C Bailez, I Miranda, V Martinez, J Garcia-Arumi; Clinical Risk Factors of Proliferative Vitreoretinopathy (PVR). A Case- Control and Multicentric Study . Invest. Ophthalmol. Vis. Sci. 2002;43(13):4401.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: Identification of clinical risk factors for development of postoperative PVR. To develop a mathematical model for predicting this complication with acceptable sensitivity and specificity rates. Methods: A multicentric, case-control study of 335 patients with RD (201 controls and 134 cases) was conducted. Patients were divided in two groups, according to the surgical procedure performed as a primary option (group 1: extrascleral surgery; group 2: vitrectomy). A logistic regression analysis was used to identify risk factors for PVR among 83 variables related to preoperative, intraoperative and postoperative characteristics in both groups. Then, results were also employed to predict the development of PVR. Evaluation of the sensitivity and specificity of this prediction was performed by cross validation. Results: Several variables showed association with PVR. In both groups, risk of PVR was increased in patients older than 70 years (odds ratio (OR): 1.66 Confidence interval 95% (CI): 1.05-2.52) and in cases with retinal breaks larger than "1 clock hour" (OR: 5.15; CI: 2.42-10.95). In group 1, patients with aphakia or pseudophakia had also a higher risk of PVR (OR: 2.59; CI: 1.08-6.19) and those receiving cryotherapy (OR: 22.2; CI: 7.82-63.7). However, in group 1, reattachment of the retina in the first postoperative day (OR: 0.33; CI: 0.14-0.78) and a preoperative intraocular pressure higher than 14 mmHg (OR: 0.53; CI: 0.34-0.82), were protective factors. Sensitivity and specificity of the prediction of the development of PVR by using a formula obtained from these results were 80.5% and 67.0% respectively. This formula has shown to be useful for predicting PVR in both groups of patients. Conclusion: It is possible to predict the risk of developing PVR according to the clinical characteristics of patients, with acceptable sensitivity and specificity rates.
This PDF is available to Subscribers Only