welcome back to another video this time around we'll be talking about how to quantify risk now risk generally is seen as a bad term however I just want to quickly explain to you that risk isn't always bad risk being high is bad worse being low is bad when the outcome is bad risk being high can sometimes be good if the outcome is good risk is generally seen as somewhat of a hostile word because you know you can always hear people say oh that was a risky play oh that was a risky move but in in this idea risk pretty much just means your chance of developing the disease before we start with this video I just want to quickly highlight to you that it's going to be using a very similar two by two table to the one that you see in the previous page about evaluation of diagnostic tests I actually have a video on that if you'd like to go see it I urge you to go see it because we're gonna be using the same sort of paradigm that we're gonna be doing here if you're ready familiar with with this sort of table then this table is no different the only difference here is that instead of test here we're not using test we're using exposure or intervention so pretty much we're putting people let's say for example we're doing a course sorry we're putting people into two different groups one that is exposed one that is not exposed and then we're seeing each groups risk of developing the disease and then we're reporting them sometimes the exposure can actually benefit the group sometimes the exposure can worsen the chance of them developing disease commonly tested ways to quantify risk involve the odds ratio and relative risk they are very similar to each other and that's that's actually because they sort of tell the same story but in different perspectives in different ways now what I mean by that let's imagine you're a detective and if you asked for example the mother what happened at the crime scene and she tells you a story and then you go ask the Father and he gives you a different story but he's somewhat describing the same events but in his perspective that's exactly what odds ratio and relative risk are they're sort of the same thing they're sort of trying to describe the same thing but in their own different way now they have a small distinction between them which will be clear when we talk about their formulas when we talk about odds ratio and we talk about relative risk the only difference is that one uses proportions and one uses ratios now you can probably guess which one uses ratios it's going to be odds ratio so other issue the formula goes as such a over B over C over D very very simple very very easy to remember relative risk it's going to use proportions a over a plus B over C over C plus D now I'm going to assume that you already understand the difference between case control study and cohort studies if not I have actually a video talking about the study designs so feel free to go watch that if you'd like to see more detail but in that video I was talking about how the formula actually you can relate the formula to this why they pick this study design so I'm actually going to go back to this whiteboard and explain this to you really quickly just a small little notes this will hopefully never be tested on on your step or possibly anybody exam I'm not quite sure on any board exam part there but on stuff I don't think they really care about this they care more about how can you calculate relative risk can you calculate odds ratio and that should be that but if you're interested in finding out why they use cohorts with relative risk and case control with odds ratio then I'll just stand spend a minute on it so why is this the case well the answer is actually here the reason that they are actually using you know relative risk for record studies and case control for odds ratio is because relative risk uses proportions and Osby she uses ratios now proportions they tend to mean something in this case they actually are depicting the prevalence among the exposed and the prevalence among the unexposed oh that is to say that that this should pretty much be very very close to the natural prevalence and since cohorts by design doesn't really care about distinguishing disease as it is distinguished the exposure it generally depicts the natural problems within its study however with case control studies that is thrown out the window you are handpicking people with disease and trying to make one group with for example 50 people with diabetes and 50 people without diabetes the prevalence is already out the window it is nothing like the natural prevalence therefore we can't really use proportions proportions are sort of meaningless that is to say if you actually apply relative risk to case control and if you apply odds ratio to to a court study then you're going to figure out that they they do give you an answer but that answer is not as meaningful if you were to use the proper way to quantify that risk and the answer is is pretty much because one user proportions which depict the natural prevalence and one uses ratios because the study design by buyeth design uses a biased sample with more people that have the disease therefore not depicting not depicting the natural prevalence okay so if we go back here I just have a quick little mnemonic to show you odds ratio Oh are two different letters you can remember that with case control eh oh so uh two different letters they are two different letters a Oh for relative risk just remember cohort oh oh remember our our LT first so that's just an easy way to remember and again if you if you need to remember which one uses the ratios which one uses the proportions I always remember odds ratio so you then you focus that you pretty much remember that Oh odds ratio uses the ratio so the other one you just proportions so that's the quick way to not mix you up we have two more notes here one don't one of the notes is that Ferrer diseases odds ratio approximates relative risk now how can we explain this if we actually go to a whiteboard it'll make a lot more sense there is actually a way to do this through hard numbers and deriving the formula however no one even bothers asking you this so I'd rather use just a logical way of of making it remember the whole paradigm I was talking about where when I was discussing how a relative risk depicts an actual prevalence but this one does not just imagine it like this if you have odds ratio and you have relative risk if you have these two and the prevalence day of the disease you're interested in is already low already low then you cannot get a biased sample here anyway you can't get a biased sample because the prevalence is already low you're forced into a low prevalence you can't raise it regardless of how hard you try the prevalence is already it's already low for that reason pretty much these two are gonna coincide these two are gonna be very very very similar once again recall that the really on the only difference between these two is that this one was depicting the natural problems but this one wasn't the second you get rid of that pretty much they're gonna be the same the last point we have here is about relative risk it's it's this part right here and this part is really important basically telling you that if the relative risk is greater than one that means that the exposure is actually associated with an increase disease occurrence the exposure actually increases the risk of you developing this disease with with relative risk less than one than the exposure decreases the chance of you operating the disease if it was equal to 1 then the exposure pretty much does not increase or decrease your chance of getting the disease and these are really important because sometimes they just tell you all the relative risk for example equals 2 what is this to mean and you have to explain to them exactly what this two means now I'm not gonna go over the examples but they're really helpful for explaining to you what exactly this two means so I urge you to go through the middle I'm not gonna go through them here because they're gonna take too much time all right then we move on to this whole page which is sort of new for a lot of people I can tell you for myself when I learned these I learned them myself because personally my in-house exams never focused on them they are more focused instead with with hypothesis testing and and alpha errors and beta errors so on and so forth and for that reason we didn't really have enough time to really go over these but they're very important they sometimes could test it on on step one so you need to be familiar with them so let's start with the first one relative risk reduction the relative risk reduction it has to come with a relative risk that is less than one that is to say that the exposure is lessening your chance of disease that's why they're giving the example of a vaccine here now what they're basically trying to say is that instead of reporting that that in this example going from 8 percent to 2 percent is is basically a of 6% instead of saying that instead of saying 6% they're giving it another name they're giving it something called relative risk reduction which basically puts the 6% in a different format it puts it into this relative risk reduction format the only way you can reach that is by doing 1 minus relative risk so I don't know the relative they give it to you here 2 out of 8 you just do 2 out of 8 minus 1 and then you're gonna get basically 6 out of 8 which is 3/4 so that's how they got this number again this number is basically trying to be this number it says that they there is trying to say the same thing but in different ways and one of the ways you can do that is by reporting it as a relative risk reduction it's more focused on the relative risk rather than the actual percentage when I skip attribute adverse because what if we want to report this 6 percent well how do we do that we actually just take this number and we - affirm this number and how did they originally get this number it was just a proportion it was the proportion of those that are unexposed unexposed that developed the disease in this case 8 percent - those who are exposed in this case vaccinated who developed a disease in this case you actually go down 6 percent so we can say oh the ACT the absolute risk reduction to the flu vaccine is 6 percent again the format is very clear there are a portion of those who who are unexposed who develop disease - those who are exposed to develop the disease now let's go back to attributable risk it's actually the same as absolute risk reduction the only difference there is that instead of instead of it being protective it's actually being it's increasing your chances building the disease with the exposure so again this one your your relative risk has to be greater than 1 this one your relative risk has to be has to be less than 1 because this has to be protected that's the only way it's gonna reduce the risk this one has to be more because that's the only way it will increase the risk so so you'd imagine the only thing you need to do is just to flip these 2 around and that's exactly what they do here a over a plus B - C over C plus D basically B be exposed the unexposed because the exposed will be the larger number in this case in this example we give you them as 21% and the the unexposed are 1% so the attributable risk is 20% another way you can reach that I'm sorry if you're not seeing it but it's basically they might just give you the relative risk they might not give you all of these numbers and in the table they might not give you this all all this table so instead these might give you the relative risk and the only way you can answer it is by using this formula and this formula you can reach the same exact answer is using this formula but they might not give you this given they might instead just give you the relative risk you just do a relative risk minus one over relative risk and then to make it a percentage you just multiply it by a hundred so this should be good for relative risk reduction as you build a risk an absolute risk reduction so this leaves us with the last three number needed to treat number needed to harm I'm gonna go pieces one and case fatality rate case fatality rate is actually very simple I'm gonna start with it first basically you take the number of deaths and you put it on top of all of the cases the the example really really makes it simple if you have four patients that died from meningitis among ten cases in the hospital for example then it'll be for over ten number of deaths number of cases times 100 to make it a percentage and that will give you basically forty percent this is how the report case fatality rate for four different cases but now let's let's close this video off with number needed to treat the number needed to harm so you could just memorize it the mnemonic here is very simple harm AR so basically harmed AR and you always put it one over a R and then the S pretty much me mark the other one number to treat will be the double double are it's a very very simple way of handling it but let's try it I'll starting taking an extra step and see why they make this formula so let's just make a quick little example let's say after doing our our two by two table yadda-yadda-yadda let's say one groups chances of developing a disease is 25 and let's say our exposed group was 15 percent 15 percent we - them from each other and that gives us our absolute risk reduction that's going to be 10 percent sorry 10 percent so equally so now we have an absolute action of 10% now they tell us well how many do we need to treat for this 10% to actually help one person alright so the way the formula goes I'll just show you how this formula goes it goes 1 over 1 or 10 sorry over 100 I was gonna simplify but 10 over 100 now if we take this formula and we got an answer I'll show you how they got it we just put this over 1 we crisscross and then 100 over 10 equals 10 what this 10 means is that we need to we need to treat 10 patients in order to basically benefit 1 every time we treat 10 patients we will be benefiting 1 person as opposed to be using using the old method with with the 25% chance of people so so how did they how did they come up with this formula why does this formula work so well it's actually very simple with what they're doing is extremely simple they just put this equal sign in the middle and then they put 10 over 100 which is which is this percentage this absolute risk reduction right here and then what they do is very simply put 1 over question mark or X or whatever and and solve further for the missing number you can see 10 over 100 equals 1 over how many how many do you need to treat in order to save one person and you can already see the answer it's it's gonna be basically 1 out of 10 because 1 out of 10 is going to equal 10 out of 100 and that's where they got that that's where they got answer this question mark this variable if you solve for it is gonna be the exact same value as this so that's how they do it they just put absolute reduction on this side and it equals 1 over X and then just solve for X the same exact logic goes to number needed to harm the only difference is instead of using the absolute risk reduction you're using attributable risk and this makes sense because when you when you're treating patients you want the exposure to lessen the chance of disease but would never need it to harm you need the exposure to increase the chance of disease so it sort of makes sense why you need to use a AR AR for nominate to treat and why you need to use AR for a number needed to harm and with that said hopefully you benefited from this video consider liking and subscribing as always thanks for watching