Lianne Kurina: How controlling confounders makes better epidemiology
Public health studies are among the most challenging to get right, says this expert in epidemiology. But better design can yield greater confidence.
As the world has learned through the recent pandemic, epidemiological studies can be complicated by many unanticipated factors. Lianne Kurina is an expert in the design of epidemiological studies who says that the key to greater confidence is better design.
The gold standard, she says, is the randomized controlled trial—a study that compares groups that are essentially identical by every apparent factor but one—the vaccinated vs. the unvaccinated, for instance. In the case of COVID-19 vaccinations, Kurina stresses that investigators did an exemplary job of this.
In situations where we can't use a randomized controlled trial, achieving a similar balance and specificity is far harder. Kurina says that researchers working with observational data, rather than trial data, must always be attuned to the overlooked factors—“confounders” she calls them—that can muddy the data and render a study moot.
However, Kurina notes, the better one controls the confounders in these observational studies via better design and data collection, the greater confidence we can have in the end results, as she tells listeners to this episode of Stanford Engineering’s The Future of Everything podcast with host Russ Altman. Listen and subscribe here.
Subscribe here: https://the-future-of-everything-stanford-engineering.simplecast.com/
Видео Lianne Kurina: How controlling confounders makes better epidemiology канала Stanford University School of Engineering
As the world has learned through the recent pandemic, epidemiological studies can be complicated by many unanticipated factors. Lianne Kurina is an expert in the design of epidemiological studies who says that the key to greater confidence is better design.
The gold standard, she says, is the randomized controlled trial—a study that compares groups that are essentially identical by every apparent factor but one—the vaccinated vs. the unvaccinated, for instance. In the case of COVID-19 vaccinations, Kurina stresses that investigators did an exemplary job of this.
In situations where we can't use a randomized controlled trial, achieving a similar balance and specificity is far harder. Kurina says that researchers working with observational data, rather than trial data, must always be attuned to the overlooked factors—“confounders” she calls them—that can muddy the data and render a study moot.
However, Kurina notes, the better one controls the confounders in these observational studies via better design and data collection, the greater confidence we can have in the end results, as she tells listeners to this episode of Stanford Engineering’s The Future of Everything podcast with host Russ Altman. Listen and subscribe here.
Subscribe here: https://the-future-of-everything-stanford-engineering.simplecast.com/
Видео Lianne Kurina: How controlling confounders makes better epidemiology канала Stanford University School of Engineering
Показать
Комментарии отсутствуют
Информация о видео
1 октября 2021 г. 21:00:36
00:28:25
Другие видео канала
Stanford Engineering Graduate Fellow Melina MathurBreakthroughs: Engineering Solutions to the COVID-19 PandemicMargot Gerritsen: A diverse team leads to higher revenue and better solutionsData science and entrepreneurship panel: Personalized data is the key opportunityStraw Into Gold: Engineers Changing the WorldEngineering Health Through 3D PrintingStanford Homecoming: Allison Okamura and Debbie SeneskyA Q&A with the host of Stanford Engineering's The Future of Everything podcastStanford Engineering EDay 2011: Keynote - Judy Estrin part 1 of 4Jennifer Chayes: Being an engineer is not a lonely, geeky job; it’s collaborative and creativeTransposition: Drew Endy talks about engineering biologyFiorenza Micheli: The race to save the oceanHow to get involved in undergraduate researchIngmar Reidel-Kruse and Biotic GamesI'm an engineer and...Celeste Matarazzo: Using quantitative data to make informed cybersecurity decisionsBest of: The future of proteinsHow to develop ever better computer chipsDoug James: Computer-generated sound catches its graphical siblingSimone D’Amico: “The Swarm” is coming to an orbit near you