Credentialing & privileging an important factor for diagnostic accuracy of diabetic retinopathy screening in the health clinics of Penang.

Introduction: Early detection and management of diabetic retinopathy (DR) is crucial in preventing blindness. Screening is recommended at diagnosis and yearly for Type 2 diabetes patients. DR screening using non-mydriatic fundus cameras (NMFC) has been extended to Health Clinics since 1997, but competency and experience of the medical officers (MOs) remain an issue in Primary Care. This study aims to assess the accuracy of the eye examination using NMFC for DR screening done by MOs and identify the factors associated with the accuracy.

Materials and methods: This is a cross-sectional study using secondary data obtained from the Penang State Health Department, which conducted clinical audits of fundus images from health clinics in 2019. The audit involved two consultant ophthalmologists to comment on the accuracy of the interpretation of retinal images and the quality of the images sampled from all health clinics with NFMC. Sampling was performed on the audited data set to include diagnosis by the MOs and diabetic retinopathy spectrum of disease. The subject of the study was the images with the corresponding reports. The outcome of this study was the accurate interpretation of the images, as commented by the ophthalmologists. The independent variables studied were the demographic of the MOs who interpreted the images, their training background and the quality of the images.

Results: The Universal sampling method was used, and the final 1129 images fulfilled the eligible criteria. The sensitivity, specificity, PPV and NPV were found to be 80.6%, 92.7%, 76.4% and 94.2%, respectively. Overall accuracy was 83.8%. After missing values were managed, 997 samples were analysed using logistic regression. The final model shows that significant factors associated with accuracy are foreign graduates MOs (Adjusted OR 1.98, 95% CI: 1.35-3.07), MOs with Credentialing & Privileging (Adjusted OR 2.32, 95% CI:1.32-2.88) and Good Image Quality (Adjusted OR 3.62,95% CI:2.37-5.71).

Conclusion: MO with C&P showed better accuracy than MO without C&P. This study suggests that MOH should emphasise the C&P when performing this procedure in health clinics. As image quality showed the highest association with accuracy, strengthening the C&P among the paramedics who perform the procedure using NMFC to get the retinal images is also necessary. This study also indicates that evaluating DR screening programs in health clinics is necessary nationwide.