Randomized Control Trials and Confounding
HTML-код
- Опубликовано: 6 июл 2024
- In this video, Dr Greg Martin describes how randomized control trials work and how they deal with confounding variables. This is episode three in a series of videos looking at epidemiology, research methods and study design in the context of global health.
Global health (and public health) is truly multidisciplinary and leans on epidemiology, health economics, health policy, statistics, ethics, demography.... the list goes on and on. This RUclips channel is here to provide you with some teaching and information on these topics. I've also posted some videos on how to find work in the global health space and how to raise money or get a grant for your projects.
Please feel free to leave comments and questions - I'll respond to all of them (we'll, I'll try to at least). Feel free to make suggestions as to future content for the channel.
SUPPORT:
------
This channel has a crowd-funding campaign (please support if you find these videos useful).
Here is the link: bit.ly/GH_support
OTHER USEFUL LINKS:
--------
Channel page: bit.ly/GH_channel
Subscribe: bit.ly/GH_subscribe
Google+: bit.ly/GH_Google
Twitter: @drgregmartin
Facebook: bit.ly/GH_facebook
HERE ARE SOME PLAYLISTS
---------------
Finding work in Global Health: bit.ly/GH_working
Epidemiology: bit.ly/GH_epi
Global Health Ethics: bit.ly/GH_ethics
Global Health Facts: bit.ly/GH_facts
WANT CAREER ADVICE?
------------
You can book time with Dr Greg Martin via Google Helpouts to get advice about finding work in the global health space.
Here is the link: bit.ly/GH_career
~-~~-~~~-~~-~
Please watch: "Know how interpret an epidemic curve?"
• Know how to interpret ...
~-~~-~~~-~~-~ 
Наука
I have a research methodology exam tomorrow and you've given me the confidence I never thought I'd have, thank you so much
So happy to hear that, Rose. Thank you for your feedback. You can do it!
Excellent explanation. Background music is very distracting.
Thank you, you are much better than my professor at school. Saving me for my Research Methods Final! Keep it up.
I have been struggling to understand confounding but this 4-minute video has helped me understand it so well! Thank you for this!
thank you so much for explaining it in short. really like your video on research as it explains perfectly without taking so much time. keep it up.
+Komal Patel thanks for the comment! :)
Fantastic explanation sir..a very useful video indeed!
Thank you Mr. Martin for these dynamic videos, 'confounding' point is interesting
Thanks for the comment Ayse. I'm going to create some more videos on confounding at some point (and talk about how to control for confounding if you're doing a case control or cohort study)
Thank you very much. Very helpful for currently writing my research methods paper
Here is a short video on Randomized Control Trials and "confounding"... please feel free to comment... share with others... etc...
Thank you, Sir. It is helpful.
I appreciate your videos, makes studying much easier!
Glad to be able to help Rebecca - thanks for the feedback! :)
Hallo Greg, am a mid career epidemiologist starting an online MPH soon and the short to the point character of your videos is an excellent refresher on key topics. Bravo!
Thanks! Glad you liked it. I'm going to be creating more shortly (so watch this space). Thanks for the feedback Dataman!
Excellent explanation
Here is a short video on Randomized Control Trials and "confounding".. please comment, share with others etc....
Thank you sir for this wonderful video.
sir how in cohort study confounding factors can be nullify
This is great, thank you
Really liked your confounding example - very well explained
Thanks very much for the comment Bernadette Feeney and thanks for sharing the video on Google+ (much appreciated).
Thank you very much
Nicely done
I like your examples, simple and fun. Helps it stick :)
Glad to hear it!
Thank You!
hi sir..
thank u so much for the explanation...its really useful n very easy to understand...
+surya kumasi Glad you liked it !
Love love .i wish my professors were great like u in explaining
Thanks for the feedback (much appreciated). Have a great 2019!!!
Excellent explanation of confounding factor.. U mean that confounding factor is associated with risk factor and outcome but not related to causation of the outcome. It remains same in both groups. It doesn't change.
Thank you!!
Thank You
Thanks Greg !
You're welcome!
good explanation very easy to fathom
+Peter James thanks - glad you found it easy to follow / fathom... please send any thoughts or suggestions re future videos. And keep the feedback coming...
Background music wasn't necessary and became rather distracting. Solid information and explanations provided in the video.
Thank you very much!
You're welcome!
धन्यवाद जी।
👏👏
Life Saver
Thank you for the feedback. Glad you enjoyed it!
What do you think about matching to eliminate confounding?
Love from india 🙌🙌 preparing for my MPH entrance
Thank you, Bipul. Wishing you the best of luck!
sir sir you rock. I love your video, the way you explain is superb. can you tell me where can I find the full video if possible. on Research methods?
+Mohsinul Haq Thanks for the comment - you can watch all five videos that I've created on research and epidemiology here: ruclips.net/p/PLujS9ooBebKWlbmIQOtYaJBjKE4VSimXJ
awesome
+DrAhadmz thanks :)
I'm not sure I fully understand for the confounding variables, in a trial there no perfect randomization, I understand that it improves with larger populations but is there some thresholds in the results for us to be sure that we can safely ignore confounding variables?
i love you already! i get it now
Glad that my video could help!!
If the music was just slightly lower, it would be even better. Liked and subscribed!
Thanks for the feedback
So in a cohort study, the individuals in the 2 groups aren't randomly assigned from a group?
reardelt thats right - participants are not randomly assigned. For this reason, you need to think about "confounding" variables that might distort the outcomes of the study.
ok, this is helpful. which biases are then associated with this type of a study. what if the alternative group knows they were treated with the placebo, wont they give negative results that will then affect the analyses.
Hi Sandisiwe Mlotshwa the trick is to do a "blinded" study which basically means that the participants don't know which group they are in (the intervention or the placebo).
This is very helpful, but I have some questions:
1. how does assigning people randomly to a group make the two similar?
2. what's an exposure?
3. explain how the two groups are equal in "every way you can imagine"? Is it accounted for when selecting the participants? Or is that random as well?
ben diaz thanks for the questions1) when two groups are formed under non-random conditions, there are many reasons why the two groups may be different. The person doing the assignment, for example, may place more men than women into one of the groups. Or if the groups are self selecting, people of a certain ethnicity might all join a particular group. If however, the assignment is truly random, and you have a large enough sample, then each group will have as many men as women etc. (there will be no reason why one group would have more).
2) an exposure can be a risk factor for disease or a health intervention. Basically anything that a person can be exposed to that might alter their health.
3) randomness of selection is what causes the groups to be equal in every possible way.
Hope that helps.
Greg
Yes this definitely helps. So they've observed a relationship between shark attacks and ice cream sales. What could possibly be a causal relationship? Are there times when a confounding variable would suffice as a hypothesis for further experiment? I'm just a bit confused. If you did a RCT on two groups and gave one a new diet pill and the other a placebo (or nothing) what would be a causal explanation and what would serve as an alternative explanation, or confounding variable?
hi ben diaz - its hard to get into this in detail over text like this. The point of the shark attacks and ice-cream sales example is that there is no legitimate causal relationship. The association that is observed is completely a function of confounding. In the case of an RCT, there are no confounding variables because both groups (the intervention and control) are the same and so all confounding is automatically controlled for.
Thank you for taking the time
Thanks these are really great. Is this made with a green screen? Or if not what software is used? Thanks
This is so good!
It is made with Apples Final Cut Pro X. www.apple.com/finalcutpro/
You can make something similar in Camtasia for either Windows or Mac www.techsmith.com/camtasia.html.
With Adobe Presenter www.adobe.com/dk/products/presenter.html
If you want this quality in your video, you will need a Good videocamera, but most important is lighting. There is a good guide to Three Point lighting her www.adorama.com/alc/0013795/article/DSLR-Video-Tips-with-Richard-Harrington-Three-Point-Lighting-Adorama-Photography-TV
If you want to learn more about video and Final Cut Pro X look here www.izzyvideo.com
Hi Christopher - yes, I use a green screen and then edit out the green in Final Cut Pro X on my Mac. Glad you like the video. Let me know if you have any suggestions re content for format.
You can also tell because of the way his shadow falls compared to the shadow of the board; even if that board was real, the shadow would go the other way.
You are right Kiki K
Potrebbe accadere che coloro che mangiano il gelato, successivamente, vadano meno in acqua rispetto a coloro che non lo mangiano, perché sentono meno caldo degli altri.
In questo caso, si arriverebbe alla conclusione che il gelato funziona, fa cioè avvenire meno attacchi di squalo (per effetto del minor bisogno di entrare in acqua), rispetto a coloro che non ne fanno uso, ma non si agirebbe sulla causa (in questo caso non controllabile), cioè la temperatura o meglio l'ingresso in acqua.
Questo da un lato produrrebbe meno danneggiati dagli squali, ma dall'altro un non conosciuto effetto deleterio sulla salute di coloro che mangiano tutti i giorni i gelati e che potrebbe portare a danni ben peggiori e per un numero maggiore di soggetti rispetto a coloro che vengono attaccati dagli squali.
Se invece si indagasse per capire la vera causa dell'attacco degli squali si potrebbe arrivare a capire che la causa è l'entrare in acqua in una zona infestata dagli squali. Si potrebbe quindi non entrare in acqua in quella specifica zona.
Se si accettano gli RCT come sistema per giungere alla prevenzione o alla cura, non si arriverà a stabilire la reale causa o si limiterà molto questo processo di conoscenza.
It would been interesting if you used the example of the ice cream at the end of the video to show how that confounding variable would be nullified.
Thanks for the suggestion Somcana - I'll make another video on this subject soon.
Thanks for responding. I applied for masters in epidemiology and community medicine. I am so excited and your videos are just splended!
I am starting my masters in public heath 2018, now taking my BSN with public health. I am trying to I understand research and statistic. can someone else do this work while I become a Nurse at the bed side. GLOBALLY (I know I cannot say Globally without knowing research and Statistic ) I am alway looking for you to explain things to me . Thank you for having the need to share. HA
Thanks for the comment. I'm glad to hear that you're starting an MPH next year. Don't worry too much about the fact that research and statistics seem complicated (during your MPH it will all become much clearer). Good luck!
Global Health with Greg Martin i am starting public health class a 7 week class. My BSN in Nursing. I know i will look for if i get lost. I love how you explain things, good teacher
Thank you very much for the valuable points... am taking master degree in public health and I try to figure out and to find the advatage of cohort on RCT?
I would have liked to see if the cohort study couldnt deal with the confounding variable of the example: shar-ice cream. I would think that yes, so in that example the RCT doesnt provide more power than the Cohort..
what type of theories are produced in rcts?
Hi Johnnie, RCTs generally provide evidence of causation. For example, the fact that a medicine produces (or causes) a certain result can be shown using an RCT
Global Health with Greg Martin would the Stradford prison experiment be considered a rct experiment?
So I can still eat my ice-cream without being afraid of taking a swim afterwards?
Please don't do music in the background of your educational videos. I cannot comprehend what you are saying over the background sounds.
plz turn off annoying music
Hello Sir,
You are too fast with your speech, its real hard to catch your pace and I reduced the speed of the video to 0.75 to hear and understand what you are saying.
Anyways thanks
Thanks for the feedback Mahesh - I will try to talk more slowly in future videos :)
Fantastic explanation sir..a very useful video indeed!
Thanks Sreejita - much apprecaited.