Development and validation of a predictive model for the diagnosis of rheumatic heart disease in low-income countries based on two cross-sectional studies

Abstract

Objectives: We developed a questionnaire-based risk-scoring system to identify children at risk for rheumatic heart disease (RHD) in rural India. The resulting predictive model was validated in Nepal, in a population with a similar demographic profile to rural India. Methods: The study involved 8646 students (mean age 13.0 years, 46% boys) from 20 middle and high schools in the West Midnapore district of India. The survey asked questions about the presence of different signs and symptoms of RHD. Students with possible RHD who experienced sore throat and joint pain were offered an echocardiogram to screen for RHD. Their findings were compared with randomly selected students without these symptoms. The data were analyzed to develop a predictive model for identifying RHD. Results: Based on our univariate analyses, seven variables were used for building a predictive model. A four-variable model (joint pain plus sore throat, female sex, shortness of breath, and palpitations) best predicted the risk of RHD with a C-statistic of 0.854. A six-point scoring system developed from the model was validated among similarly aged children in Nepal. Conclusions: A simple questionnaire-based predictive instrument could identify children at higher risk for this disease in low-income countries where RHD remains prevalent. Echocardiography could then be used in these high-risk children to detect RHD in its early stages. This may support a strategy for more effective secondary prophylaxis of RHD.

Keywords

• Low-income countries
• Early detection
• Educational intervention
• Rheumatic heart disease
• Secondary prophylaxis