Routinely collected data and comparative effectiveness evidence: promises and limitations

Innovation to build learning health systems

“…The limitations of routinely collected data for research purposes are not always recognised 9. Debate on this topic at a workshop held at Medical Informatics Europe 2017 provides the basis for our leading article.…”

Once Again About the Hierarchy of Evidences in Medicine or Whether It Is Possible to Choose the Most Effective and Safe Drug With the Help of Observational Studies

“…такой дизайн помогает умень-шить систематическую ошибку. Результаты ретроспек-тивных наблюдательных исследований следует оцени-вать с большой осторожностью [30].…”

“…Учитывая большое количество ограничений, прису-щих ретроспективным наблюдательным исследова-ниям (особенно если речь идет об административных ресурсах при составлении баз данных), вызывают удив-ление попытки статистико-математического объедине-ния результатов таких исследований с формулированием соответствующих выводов [30]. Так, в журнале STROKE в марте 2017 г. был опубликован системный обзор и ме-таанализ наблюдательных исследований НОАК и вар-фарина [31].…”

Big data’s big bias: bringing noise and conflicts to US drug regulation

“…4 But claims data and routinely collected electronic health record data are not structured for research purposes, and researchers report that recognition, collection, reporting, and reproducibility of adverse events are unreliable at best. [5][6][7][8][9][10] In 2015, drug safety experts reviewed various big data projects, including the FDA's Sentinel. They reported that the projects have "proved largely unable to provide credible evidence of new, unsuspected drug adverse effects" and that six years after Sentinel was launched it "has not yet been the primary data source in identifying a single new drug risk that led to a significant regulatory action such as a drug withdrawal, boxed warning, restriction or contraindication."…”