Apr 09, 2021 9:00 AM
The study, titled “Validation of a mortality composite score in the real-world setting: overcoming source specific disparities and biases,” explored the Syapse Learning Health Network platform’s ability to accurately assess oncology mortality rates, ultimately demonstrating that a composite mortality score was capable of achieving sensitivity of 94.9% and specificity of 92.8%, as compared to the National Death Index (NDI), with concordance within one day of 98.7%.
“As an oncologist and investigator, being able to provide accurate mortality prediction is essential. In order to do so, we need to have access to high quality, trustworthy real-world data,” said Dr.
Mortality is an important data element in oncology research. Complete and accurate mortality data are critical for calculating accurate overall survival. In the real-world setting, there isn’t a single complete source for mortality that can be used broadly. In addition, potential disparities in which patients are included based on sex, race, age, and socioeconomic status are not well studied. The Syapse Learning Health Network™ draws upon a combination of six individual real-world data sources to develop a high-quality composite mortality score. This study observed that each of the six individual sources had disparities in coverage across subgroups of sex, race, age, and socioeconomic status, but that the composite mortality score was able to overcome these biases, maintaining high sensitivity (84.6% - 96.1%), specificity (77.9% - 99.2%), and concordance within one day (95.7% - 100.0%) across subgroups.
“This study showed that social determinants of health can impact which patients are represented in data collection. However, by rolling up data across six real-world sources we were able to achieve a complete and accurate picture of mortality data.” added
About the Syapse Learning Health Network™
This international network of healthcare providers enables improved cancer care by sharing real-world data to support clinical decisions and by fostering collaborations among participants. Healthcare providers, including doctors and nurses, share and learn which cancer treatments produced the best real-world outcomes in clinically and molecularly similar patients. Researchers learn from real-world clinical, molecular, treatment and outcomes data. These collaborations are enabled by a secure platform of shared de-identified data that is standardized and normalized across the Learning Health Network and complies with all applicable federal and state data privacy protection regulations.