1. Prediction of probability of rubella based on eye outcomes (PORBEO Nomogram)—a cross-sectional sentinel surveillance of 1134 infants
- Author
-
Ravinder Kaur Sachdeva, Devika Shanmugasundaram, Ashish Bavdekar, Sanjay Verma, Mahantesh Sangappa, Parul Chawla Gupta, Jagat Ram, R Sabrinathan, Kuldeep Singh, Sanjay Shah, Deepa John, Sridhar Santhanam, Manoj V Murhekar, Praveen Kumar-M, and Krishna R Murthy
- Subjects
Pediatrics ,medicine.medical_specialty ,Antibodies, Viral ,Logistic regression ,Rubella ,Odds ,Congenital Rubella ,03 medical and health sciences ,Cellular and Molecular Neuroscience ,0302 clinical medicine ,medicine ,Humans ,030212 general & internal medicine ,Child ,Probability ,Congenital rubella syndrome ,Univariate analysis ,business.industry ,Infant ,Nomogram ,medicine.disease ,Sensory Systems ,Nomograms ,Ophthalmology ,Cross-Sectional Studies ,030221 ophthalmology & optometry ,Etiology ,business ,Sentinel Surveillance - Abstract
Rubella is an important infectious, vaccine-preventable etiology of congenital defects. The aim of the study was to develop a prediction nomogram to assess the probability of an infant being at risk for congenital rubella based on demographics and ophthalmological findings. This was a cross-sectional sentinel surveillance study conducted at 5 centers spanning pan-India and involved 1134 infants. The diagnosis of rubella was made using standard guidelines. For the construction of the prediction model, laboratory-confirmed cases were grouped as “at-risk” (AR) infants and the discarded cases into “not at risk” (NAR) infants. Univariate analysis (p value cut-off < 0.05) followed by multivariate binary logistic regression model development was performed. The average (median) age of the suspected CRS infants was 3 (IQR 1–6) months, and the average (mean) age of their mothers was 25.8 ± 4.1 years. Out of the total infants, 81 (7.3%) died, 975 (88%) were alive, and 55 (5.0%) were lost to follow-up. The final model showed that the odds of cataract, retinopathy, glaucoma, microcornea, and age of the infant at presentation were 3.1 (2.2–4.3), 4.9(2.3–10.4), 2.7(1.1–5.9), 2.3(1.1–4.7), and 1.1 (1–1.1), respectively, for the AR infant as compared to NAR infant. AUC of final model was 0.68 (95% CI Delong, 0.64–0.72). Bootstrapping for calibration of the model showed satisfactory results. Nomogram, along with a web version, was developed. The developed nomogram would have a wide community-based utilization and will help in prioritizing attention to high-risk children, thereby avoiding loss to follow-up.
- Published
- 2020