Abstract
Purpose::
To know the corneal endothelial specular microscopy sample size with three different statistical power.
Methods::
Retrospective study performed with right eye of 103 contact lens users who had undergone corneal specular microscopy in 12 consecutive months. The Bio-Optics specular microscopy was used with Bambi 2000 Plus software. The sample size was computed with CELLS ANALYZER - Corneal Endothelial Statistical LabPat Rec with patient age, endothelial density, endothelial cell area, variation coefficient and hexagonal cell percentual. The choose statistical powers were confidence reliability (CR) of 90% and relative error (RE) of 0.1 for triage finality, CR of 95% and RE 0.05 for endothelial diagnosis and follow-up and CR 99% and RE 0.01 for endothelial diagnosis, follow-up and research finalities.
Results::
The sex distribution was 72 (69.24%) of the eyes of female patients and 32 (30.76%) of male patients. Minimum age was 13 years, maximum 76 years and the average 34±11.5 years. Minimum counted cells was 116 cells, maximum 715 cells and the average 330±119 cells. Computed sample size with CELLS ANALYZER - Corneal Endothelial Statistical Lab - CR of 90% and RE of 0.1: Minimum sample size was 12 cells, maximum 71 cells and the average 27±11 cells. The minimum endothelial sample error was 2.1%, maximum was 4.9% and average 3.38±0.58, all exams had planned relative error (0.1 or 10%) largest than calculated what determine statically correct sample size for all exams. CR of 95% and RE 0.05: Minimum sample size was 65 cells, maximum 389 cells and the average 148±60 cells. The minimum endothelial sample error was 1.79%, maximum was 4.18% and average 2.18±0.48, all exams had planned relative error (0.05 or 5%) largest than calculated what determine statically correct sample size for all exams. CR 99% and RE 0.01: Minimum sample size was 2775 cells, maximum 15897 cells and the average 6240±2427 cells. The minimum endothelial sample error was 2.77%, maximum was 6.46% and average 4.43±0.77, all exams had planned relative error (0.01 or 1%) shortest than calculated what determine statically incorrect sample size for all exams.
Conclusions::
All sample size was correct with 90% CR and 0.1 RE with small sample size only for triage finality, all sample size was correct with 95% CR and 0.05 RE with possible sample size with some different endothelial images and with 99% CR and 0.01 RE all exams counted small number of the cells for the actual corneal specular microscopy technology.
Clinical Trial::
BAMPESQ - UFPR
Keywords: cornea: endothelium • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology • imaging/image analysis: clinical