CARDIOVASCULAR JOURNAL OF AFRICA • Volume 34, No 5, November/December 2023 270 AFRICA 1.00 indicating no agreement, poor, fair, moderate, substantial and almost perfect agreement were used. Interpretation of LV structure, LV function, RV structure, RV function, and pericardial effusion was compared to expert interpretation by a paediatric cardiology fellow. Results Forty studies from patients aged seven to 75 years, of whom 72% were males, were interpreted and included in the analyses. Twentyeight echocardiograms (70%) had at least one abnormality in the five evaluated categories. LV abnormalities were more common than RV abnormalities (Table 4). Inter-rater reliability was generally moderate to strong and improved over successive evaluation periods, as measured by the kappa coefficient (Table 5). One notable exception was LV structure, for which kappa slightly decreased over successive timepoints, beginning at 0.67 during the first evaluation and ending at 0.52 during the final evaluation. Similarly, percentage agreement was high throughout the study period and reached perfect agreement for some measurements. Although 70% agreement was found for pericardial effusion at the second timepoint, kappa was negative because 74% agreement was expected. Of note, only trivial and small pericardial effusions were missed. In order to continue supporting the residents, a gateway for image uploading was provided. This process allowed the cardiology fellow to continue to observe echocardiograms obtained in Ghana and to help the imagers with any questions or concerns, particularly images that were out of their scope of practice. Discussion A short training course based on ASE recommendations was sufficient to adequately train non-cardiologist residents to perform focused limited echocardiography in a low-resource setting in Ghana over a 10-week period. There was generally high percentage agreement and inter-rater reliability compared to a paediatric cardiology fellow with echocardiographic experience. This study provides a potential model that could be expanded, both at the Cape Coast Teaching Hospital and in other similar settings. More than 90% of children with heart disease are born in parts of the world where adequate cardiac care is not available.36 The expansion of echocardiography beyond specialists can begin to bridge at least part of the gap across the continuum of cardiac care. This study adds to the growing literature that non-echocardiographers can demonstrate acceptable accuracy using focused cardiac ultrasound.20-28,34 Although studies in the global setting have traditionally utilised experienced cardiologists or sonographers, more recent studies have incorporated other providers, such as nurses and healthcare workers.18,29-33 However, these studies have been limited to the evaluation of RHD by increasingly limited protocols. To our knowledge, this is the first study, albeit in a very small group, to show that non-experts in LMIC settings can successfully perform and interpret general echocardiograms in both children and adults. This is an important expansion of studies conducted in high-income countries, as LMICs provide a specific set of logistics challenges, including new equipment, infrastructure and educational barriers. There were some notable exceptions to the generally high level of inter-rater reliability in the study. Most notably, kappa was 0.00 for RV function at the first timepoint and –0.15 for effusion at the second timepoint despite high percentage agreement. This is a known paradox for the kappa statistic when events, such as RV dysfunction or effusion, are rare in the data.37,38 The high percentage agreement shows that the trainees demonstrated sufficient accuracy despite the low kappa values in these isolated circumstances. As the application of handheld echocardiography evolves, the training of personnel not originally intended to perform focused cardiac studies is becoming increasingly important. Skills in acquiring and interpreting focused handheld cardiac images should be taught to novices via both didactic and proctored training prior to implementation in resource-limited settings. Expectations of trainees and subjects should be carefully managed, including the limitations of handheld echocardiogram. The curriculum developed for this study was created to quickly train providers in basic echocardiographic assessment, but this curriculum can be duplicated in other underserved regions and modified to emphasise other areas of focus (for example, more focused assessment of valve disease for RHD and more focused assessment of wall-motion abnormalities for myocardial infarction). The didactic lectures have been recorded and can be distributed. The hands-on training is relatively short when compared with the potential improvement in access to care. Table 4. Characteristics of echocardiograms read by paediatric cardiologist Variables Normal Mild dilation At least moderate dilation LV structure, n (%) 18 (45) 15 (37.5) 7 (17.5) RV structure, n (%) 36 (90) 1 (2.5) 3 (7.5) Normal Mild depression At least moderate depression LV function, n (%) 21 (52.5) 5 (12.5) 14 (35) RV function, n (%) 32 (80) 3 (7.5) 5 (12.5) Absent Present Pericardial effusion, n (%) 33 (82.5) 7 (17.5) Table 5. Agreement in interpretation Evaluation 1 (n = 20) Evaluation 2 (n = 10) Evaluation 3 (n = 10) Agreement (%) Kappa p-value Agreement (%) Kappa p-value Agreement (%) Kappa p-value LV structure 80.0 0.67 < 0.001 80.0 0.64 0.002 70.0 0.52 0.01 LV function 60.0 0.23 0.07 80.0 0.55 0.03 80.0 0.65 0.004 RV structure 75.0 0.22 0.06 90.0 0.47 < 0.001 90.0 0.71 0.002 RV function 75.0 0.00 – 80.0 0.41 0.04 100.0 1.00 < 0.001 Effusion 100.0 1.00 < 0.001 70.0 –0.15 0.70 100.0 1.00 < 0.001
RkJQdWJsaXNoZXIy NDIzNzc=