This whole series has done justice to statistics. Best part, it seems so easy and interesting. An experience in itself. Please make more videos for undergraduates.
Thank you for your suggestions. We have made note of them for discussion should future videos be developed. In the meantime, here is another video that explains the concept: ruclips.net/video/1jzZe3ORdd8/видео.html.
Odds ratios and relative risks measure the association between a dichotomous outcome and a predictor variable. A relative risk tends to be calculated when the study design is prospective (participants are followed forward in time during the study), whereas an odds ratio is generally used when a retrospective study design is used (data is collected back over time). A relative risk is calculated by dividing the rate of an event (outcome) in the treatment group by the rate of an event in the control group. The rate of an event, in either group, is calculated by dividing the number with the event by the total number of participants in that group. An odds ratio, on the other hand, compares the number of events to the number of non-events in each group. So in both the treatment and control groups, the odds of an event is calculated by dividing the number of participants with the event by the number of participants without the event. Then the odds of the event in the treatment group is divided by the odds of the event in the control group to determine the odds ratio.
@@nccmt I understand how RR and OR are calculated. And I hear you that in general RR is used prospectively while OR is used retrospectively. I'd be grateful for any explanation of 𝘄𝗵𝘆 this pattern developed. In both cases we have two pairs of numbers. Why would it be useful to perform different arithmetic operations in the two cases? In fact why introduce the concept of "odds" at all? According to my always suspect arithmetic: If we normalize each sample size to 100, and call the resulting numerators a and b: RR = a / b OR = a / b x (100 - b) / (100 - a) On the off chance I got that right, what's the benefit of the extra multiplier in the OR calculation? Thanks for any light you can shed on this.
@@axolotl5327 ORs are typically used for retrospective studies because in many retrospective studies, such as case-control designs, the investigator actually determines the number with the event and without the event in each group. For example, they identify 100 participants with a particular disease and 100 without a disease and identify the number within each group that has the predictor variable of interest. This means you can’t actually determine the incidence of the event (the number with the event divided by the total number of participants who may be at risk) in the general population, so you can’t calculate probabilities for an RR. You can calculate the OR for associations of the event with given exposures instead.
Yes - from the same data you can calculate both an OR and an RR. However, there are differences in when you use one versus the other. “Calculation of risk requires the use of “people at risk” as the denominator. In retrospective (case-control) studies, where the total number of exposed people is not available, RR cannot be calculated and OR is used as a measure of the strength of association between exposure and outcome. By contrast, in prospective studies (cohort studies), where the number at risk (number exposed) is available, either RR or OR can be calculated.” Citation: Ranganathan, P., Aggarwal, R., & Pramesh, C. S. (2015). Common pitfalls in statistical analysis: Odds versus risk. Perspectives in clinical research, 6(4), 222.
Still don't get why you'd choose either RR or OR over the other. The examples in both videos are kind of similar. There's a group exposed to an intervention and a group that isn't, in both examples..
The NCCMT has done a great job! I hope the NCCMT can use ONE example, and put them side by side, to see how relative risk and how odds ratio look compared to each other, and then use plain English to explain what the differences are between the OR and RR. Thank you for a great series!
Hi Chad. The expression “4 fold” means 4x the risk, which would equate to the outcome being 400 times more likely or 400 times less likely in the intervention group versus the control group, depending on the outcome of interest (i.e. whether more or less of the outcome is what is desired). If we turn this into a percentage, it would be 4.0 - 1.0 = 3.0 x 100= 300% more likely or less likely. For example, an RR of 1.0 means there is no difference between the two groups. An RR of 1.5 means the outcome is 1.5 times more/less likely in the intervention group vs. the control group, or 50% more/less likely. An RR of 2 means the outcome is 2 times more/less likely in the intervention group vs. control group, or 100% more/less likely.
@@nccmt Thanks for your response. I'm confused. That seems like a huge number; like it will happen. I thought 4x or 4 fold would look like 1.4 meaning it would only be 4%. The research I'm looking at is someone with a relative (parent, grandparent, sibling, etc.) that has alcoholism. They are stating one is 4x more likely to have alcoholism if there is a genetic disposition. Wouldn't that mean one would have alcoholism if your parent does? I'm confused that it goes beyond 100% and the meaning of that number. Cause even if I have a parent with alcoholism and I drink I may or may not get alcoholism. Which seems to defy 400%. I do respect your feedback and would greatly appreciate more explanation on this subject. So I can grasp the concept of 4x, 4 fold, and 400%. I only did one semester of research methods in college and we did not cover this subject. So I'm missing some basic information.
Got a message back from someone who actually did the research. They stated that they found 5% of people with Alcoholism that had no alcoholism in family (biological). Alcoholism increased to 20% for individuals that did have alcoholism in their families. So 5 x 4 equals 20% and would be considered 4 times increase. Just wanted to clear that up. It can be really confusing.
@@bananasinpajamas9499 I asked someone and found the answer. I embarrassed to say how easy it. In order to now what 4 fold means one has to know the base number. For example: If I'm at 4 fold risk of becoming an alcoholic because I have a family history. The formula is 4 x y=X or we can say everyone is at a 5 percent risk of becoming an alcoholic and if I have a 4 fold increase risk because of genetic factors. The formula would be 4 x 5=20. So someone is 20 percent more likely to be come an alcoholic. I hope that makes since. It's a play on statics. One needs the base number to find the answer and that is not always given.
Great video about RR! Sooo many doctors DO NOT understand RR, and should watch this video.
This whole series has done justice to statistics. Best part, it seems so easy and interesting. An experience in itself. Please make more videos for undergraduates.
Best channel for basics of research 👍♥️
Amazing explanation
Thank you very much. Such an effective and helpful video. Thank you! Thank you!
Thanks for the great videos
Is it possible please to make videos about prevalence (point and period ) and incidence
Regards
Thank you for your suggestions. We have made note of them for discussion should future videos be developed. In the meantime, here is another video that explains the concept: ruclips.net/video/1jzZe3ORdd8/видео.html.
@@nccmt thank you
Thankyou so much
Thanks a lot very nice and informative video.
amazing fact to you doc
Best video 👍
Thank you!
Explicit and helpful
Great video!
How is RR any different to an odds ratio?
Odds ratios and relative risks measure the association between a dichotomous outcome and a predictor variable. A relative risk tends to be calculated when the study design is prospective (participants are followed forward in time during the study), whereas an odds ratio is generally used when a retrospective study design is used (data is collected back over time). A relative risk is calculated by dividing the rate of an event (outcome) in the treatment group by the rate of an event in the control group. The rate of an event, in either group, is calculated by dividing the number with the event by the total number of participants in that group.
An odds ratio, on the other hand, compares the number of events to the number of non-events in each group. So in both the treatment and control groups, the odds of an event is calculated by dividing the number of participants with the event by the number of participants without the event. Then the odds of the event in the treatment group is divided by the odds of the event in the control group to determine the odds ratio.
@@nccmt I understand how RR and OR are calculated. And I hear you that in general RR is used prospectively while OR is used retrospectively.
I'd be grateful for any explanation of 𝘄𝗵𝘆 this pattern developed. In both cases we have two pairs of numbers. Why would it be useful to perform different arithmetic operations in the two cases? In fact why introduce the concept of "odds" at all?
According to my always suspect arithmetic: If we normalize each sample size to 100, and call the resulting numerators a and b:
RR = a / b
OR = a / b x (100 - b) / (100 - a)
On the off chance I got that right, what's the benefit of the extra multiplier in the OR calculation?
Thanks for any light you can shed on this.
@@axolotl5327 ORs are typically used for retrospective studies because in many retrospective studies, such as case-control designs, the investigator actually determines the number with the event and without the event in each group. For example, they identify 100 participants with a particular disease and 100 without a disease and identify the number within each group that has the predictor variable of interest. This means you can’t actually determine the incidence of the event (the number with the event divided by the total number of participants who may be at risk) in the general population, so you can’t calculate probabilities for an RR. You can calculate the OR for associations of the event with given exposures instead.
Nice videos. But in given examples, whether OR can be calculated for example quoted for RR and vice versa?
Yes - from the same data you can calculate both an OR and an RR. However, there are differences in when you use one versus the other.
“Calculation of risk requires the use of “people at risk” as the denominator. In retrospective (case-control) studies, where the total number of exposed people is not available, RR cannot be calculated and OR is used as a measure of the strength of association between exposure and outcome. By contrast, in prospective studies (cohort studies), where the number at risk (number exposed) is available, either RR or OR can be calculated.”
Citation: Ranganathan, P., Aggarwal, R., & Pramesh, C. S. (2015). Common pitfalls in statistical analysis: Odds versus risk. Perspectives in clinical research, 6(4), 222.
Still don't get why you'd choose either RR or OR over the other. The examples in both videos are kind of similar. There's a group exposed to an intervention and a group that isn't, in both examples..
Thank you for your comment, we are looking into the matter.
The NCCMT has done a great job! I hope the NCCMT can use ONE example, and put them side by side, to see how relative risk and how odds ratio look compared to each other, and then use plain English to explain what the differences are between the OR and RR. Thank you for a great series!
Question: What does the term "4 fold" mean? Or 4x greater risk? Would that be 40%?
Hi Chad. The expression “4 fold” means 4x the risk, which would equate to the outcome being 400 times more likely or 400 times less likely in the intervention group versus the control group, depending on the outcome of interest (i.e. whether more or less of the outcome is what is desired). If we turn this into a percentage, it would be 4.0 - 1.0 = 3.0 x 100= 300% more likely or less likely.
For example, an RR of 1.0 means there is no difference between the two groups. An RR of 1.5 means the outcome is 1.5 times more/less likely in the intervention group vs. the control group, or 50% more/less likely. An RR of 2 means the outcome is 2 times more/less likely in the intervention group vs. control group, or 100% more/less likely.
@@nccmt Thanks for your response. I'm confused. That seems like a huge number; like it will happen. I thought 4x or 4 fold would look like 1.4 meaning it would only be 4%. The research I'm looking at is someone with a relative (parent, grandparent, sibling, etc.) that has alcoholism. They are stating one is 4x more likely to have alcoholism if there is a genetic disposition. Wouldn't that mean one would have alcoholism if your parent does? I'm confused that it goes beyond 100% and the meaning of that number. Cause even if I have a parent with alcoholism and I drink I may or may not get alcoholism. Which seems to defy 400%. I do respect your feedback and would greatly appreciate more explanation on this subject. So I can grasp the concept of 4x, 4 fold, and 400%. I only did one semester of research methods in college and we did not cover this subject. So I'm missing some basic information.
Got a message back from someone who actually did the research. They stated that they found 5% of people with Alcoholism that had no alcoholism in family (biological). Alcoholism increased to 20% for individuals that did have alcoholism in their families. So 5 x 4 equals 20% and would be considered 4 times increase. Just wanted to clear that up. It can be really confusing.
@@chadwiley6917 Yes I agree it is very confusing I don't have the mind for it but I'm trying,thank you for your comments.
@@bananasinpajamas9499 I asked someone and found the answer. I embarrassed to say how easy it. In order to now what 4 fold means one has to know the base number.
For example: If I'm at 4 fold risk of becoming an alcoholic because I have a family history. The formula is 4 x y=X or we can say everyone is at a 5 percent risk of becoming an alcoholic and if I have a 4 fold increase risk because of genetic factors. The formula would be 4 x 5=20. So someone is 20 percent more likely to be come an alcoholic. I hope that makes since. It's a play on statics. One needs the base number to find the answer and that is not always given.
اللي جاي عشان اختبار الوبائيات لايك 🤣💔
لايكات والله 😂💔💔💔.
ya t il un video en francais?
Bonjour Fanie, merci pour votre commentaire. Oui, je vous invite à regarder la vidéo en français ici : ruclips.net/video/z6gW2-7Gjs0/видео.html