Computer Detection of Diabetic Retinopathy Compared to Clinical Examination (CDDR)

Primary Outcome Measures:

• Sensitivity of detecting moderate non proliferative diabetic retinopathy or more severe DR and/or significant macular edema on a person-specific basis. [ Time Frame: Recruitment over 4 months ] [ Designated as safety issue: No ]
  If the interval excludes differences of 10% or more (IDx worse than clinical examination), then the IDx system will be considered non-inferior to clinical examination by ophthalmologists. If IDx is non-inferior to clinical examination, a two-sided test of the null hypothesis that the difference in sensitivities is 0 will be conducted to determine if the IDX is superior to clinical examination. 24 McNemar's test for paired proportions will be used to test the difference in proportions.
  
  

Secondary Outcome Measures:

• Specificity (true negatives divided by true negatives plus false positives) at detecting more than moderate non proliferative diabetic retinopathy and/or significant macular edema on a person-specific basis. [ Time Frame: Four months ] [ Designated as safety issue: No ]

  Secondary Outcome Variables:

  • Specificity on a person-specific basis
  • Sensitivity on an eye-specific basis [Clinical examination only]
  • Specificity on an eye-specific basis [Clinical examination only]
  
  
• Sensitivity on an eye-specific basis [Clinical examination only] [ Time Frame: Four months ] [ Designated as safety issue: No ]
  Analysis of secondary outcomes will follow the same approach as for the primary outcome analysis of paired proportions as described above for the comparison of person-specific sensitivity.
  
  
• Specificity on an eye-specific basis [Clinical examination only] [ Time Frame: Four months ] [ Designated as safety issue: No ]
  Analysis of secondary outcomes will follow the same approach as for the primary outcome for analysis of paired proportions as described above for the comparison of person-specific sensitivity.
  
  

The project is a multicenter, observational study to compare the accuracy of the IDX-DR computer detection system to board certified ophthalmologists performing dilated eye examination at identifying between no or mild NPDR without clinically significant diabetic macular edema from more than mild NPDR with or without CSDME. Approximately 400-600 persons with diabetes will be recruited through 5 clinical centers. The ophthalmologist who examines the patient will evaluate and record the status of the eye based on his/her clinical examination. The study subjects will then have two fundus photographs taken of each eye. The photographic images will be transferred to a central Reading Center at the University of Pennsylvania and interpreted by professional graders. The Reading Center will then transfer the images to IDx-DR, the computer detection system. IDx-LLC, the Sponsor of this study, will have no access to the images. Data from the Reading Center interpretation will be considered the "gold standard". The results of the clinical examination and IDX-DR will be compared against this gold standard as well as against each other.