Analysis of the secondary endpoints of the SPR study using competing risk models

Abstract

The Scleral Buckling and Primary Vitrectomy (SPR) study is an European multicenter clinical trial where patients with a type of retinal detachment (caused by a break, tear or hole in the retina) called rhegmatogenous retinal detachment (RRD) were randomized into one of two surgical methods of treatment (scleral buckling or primary vitrectomy method) in each of two subtrials. The subtrials, which consist of two groups of patients according to the state of their eye lens is either the aphakic (artificial lens) or the phakic (natural lens) group. The study recorded the time of the occurrence of three events of interest, the proliferative vitreoretinopathy (a post operation complication in the retina), retina re-detachment and re-operation. The purpose of the study was to determine the more effective of the two methods in the treatment of rhegmatogenous retinal detachment of medium complexity. Survival analysis using one of the three events at a time was carried out to study the differences in the risk of each event between the two treatment groups for each subtrial. This was followed by a competing risk analysis using the log-rank and Gray’s test. The event of re-operation was taken as competing with the other two events. The results of the ‘one-event at a time’ survival analysis and that of the competing risk analysis were similar. The competing risk analysis when the event of interest was proliferative vitreoretinopathy revealed that the aphakic patients treated with either method have similar or same risk of the post-operative incidence of proliferative vitreoretinopathy both in the absence (p=.327) and presence(p=.305) of competing risk of re-operation and they also have same risk of experiencing a reoperation (p=.721). However, for the phakic patients, the risk of having re-operation was found to be higher for patients treated with primary vitrectomy (p=0) than with scleral buckling. There also was no significant difference in the risk of experiencing proliferative vitreoretinopathy both in the absence (p=.67) and presence (p=.675) of the re-operation, the competing event. When the event of interest was retinal re-detachment, the result show that for the aphakic patients, the risk of retinal re-detachment was higher for patients with scleral buckling both in the absence (p=.0054) and presence (p=.0048) of the competing risk of re-operation but patients treated with either method experienced the same risk of re-operation. For the phakic patients, there was no significant difference in the risk of retinal re-detachment both in the absence (p=.078) and presence (p=.081) of the competing risk of re-operation. Conclusively, it was found that primary vitrectomy was more effective for the aphakic patients and scleral vitrectomy was more effective for the phakic patients.