Abstract 16989: Use of Screening Echocardiography to Improve Healthcare in Underserved Areas of Brazil: Data From the PROVAR+ Study

Abstract

Introduction: Access to public health care is limited in Brazil, especially in remote areas and many patients remain on long waiting lists for echocardiography (echo). Advances in handheld echo and telemedicine provide an opportunity for integration of screening for cardiovascular disease into the existing primary care system. This may improve early detection of heart disease (HD) and guide prioritization of referrals. We aimed to test the accuracy of screening echo to identify significant HD in this setting. Methods: Patients on waiting lists for the echos were eligible and answered a standardized clinical questionnaire. All subjects underwent focused 7-view screening echo (GE VSCAN), acquired by non-physicians in primary care centers and interpreted remotely by experts through cloud-storage and telemedicine. All patients underwent standard echo (GE Vivid Q), performed by an expert, blinded for the screening echo images and diagnoses. Significant HD was defined as any ventricular dysfunction, moderate to severe valve disease, moderate to severe left ventricle hypertrophy, aortic root dilatation or congenital heart disease. Statistical analysis was performed using classification and regression tree analysis based on 70% of the sample for training and 30% for test. Results: Over 9 months we evaluated 569 patients. Median age was 58.7 years, 61.4% were females, 71.6% had hypertension, 22.9% diabetes, 16.3% Chagas disease, 9.3% coronary disease, 4.8% valve disease. The most frequent symptoms were chest pain, dyspnea and palpitations, isolated or in combination, and 29.5% were asymptomatic. The presence of symptoms was not a good predictor for abnormal echo findings. Significant HD was observed in 147 (25.8%) patients on the standard echo. The model utilizing clinical features alone didn′t show a good accuracy on detecting significant HD. The screening echo alone had a negative predictive value of 0.94, with an overall accuracy of 78%. The integration of the screening echo variables with clinical features on the decision tree generated an accuracy of 82.4%, using the variables with statistical significance (Chagas disease, valve disease, coronary disease, body mass index, gender and age). Conclusions: Integration of screening echo into primary care is feasible in Brazil and should ensure that patients with higher needs have priority access to standard echocardiography and referrals. The burden of HD observed and the accuracy of screening echo performed by non-physicians suggests this tool may help deliver health care to underserved areas.