Skip to 0 minutes and 14 seconds So we look at the chart in the article to figure out where they got that information from to be able to do our own calculations and you can see they looked at each one separately but you could also see that the confidence intervals for many of these are extremely wide again that indicates that the data is not very precise. So what we would try to estimate in the population would not be very precise. And also note that the percentages you can see at the top it’s listed is as number in percent most of the percentages are very very small.
Skip to 0 minutes and 48 seconds So what they did was they took the composite of all of them together and at least this one has a better confidence interval and the numbers that are listed there. Now one thing if I haven’t taught you already to look at the legends because you can find some information there that may not be listed in other places. This word that’s in here always scares me a little bit and it’s called pooled. This means they fudged the data and not necessarily in an inappropriate way but they’re going to remove some patients maybe because of conflicting or confounding information. So it’s going to be very difficult to reproduce their numbers and you’ll see this as we try to calculate them.
Skip to 1 minute and 32 seconds So I set this up just like you would normally have your control in your NSAIDs hospitalized hospitalized no come up with your numbers. And this is where you get these sticky wickets because when we calculate an odds ratio we get 0.999 they got one point into one nine in the study. Again which means that they pooled that they they didn’t get provide us with the data, so we cannot produce that result. So then obviously we can’t reproduce the RRR, and you see we got a number needed to treat of ten thousand or in this case harm because it’s causing a problem.
Skip to 2 minutes and 7 seconds Now remember I told you you don’t want to list these without having some kind of a context but I just wanted to show you looking at benefits versus risk. So I took four or three of the NSAIDs that were in there that I could also find and NNT and the literature for the treatment of pain and you can see the NNT’s are very small that most NSAIDs work pretty well in removing pain and then you can see how large the NNH are and especially for the trial without the pooled information that they had. We got it an NNH of 10,000 which is a considerably high for all of them.
Skip to 2 minutes and 46 seconds So from this trial even though they tried to state that there are some risks with using NSAIDs and the incidence of heart failure admissions really doesn’t look like there’s that much of a risk there for at least the pooled data. So in conclusion I hope you’ve observed through what with the terminology that have given you in some of the examples that you really need to analyze a studying rather than just reporting what’s there. You need to verify all the numbers that are provided in a study because you want to make sure that you can reproduce those results and interpret them correctly.
Skip to 3 minutes and 23 seconds You also want to ensure that you provide even because most of the time they want all the components of an NNT and NNH statement to see if you can get a complete picture and decide on whether you can extrapolate that to your patients. And then you what you need to do is when you’re looking at applying this to individual patients that you need to compare your patients baseline risk to the studies and adjust upward or downward depending on if the patient’s risk is higher or lower than what you observed.
Skip to 3 minutes and 51 seconds And then again you have to make sure you look at both the NNT and the NNH together when you’re looking at treatments because you want to make sure that the benefit is worth the potential risks that are going to occur in when patients receive these and again with all of these we want to make sure that you put this in context of a full literature evaluation and not just focus on the numbers because we can we know those are easy to look at. We know that it’s easy to calculate and NNT but but determining what it means and explaining it to a healthcare professional to be able to come up with the right decisions is a little bit more difficult.
Skip to 4 minutes and 30 seconds And this is the slide that I mentioned before where there are a few websites where you can try to find some NNT comparative numbers. Like I said there really isn’t one place. You’re gonna have to really search to see if you can find them and there’s not some for each of the different types of data so you can see if there’s some summary data. These if not you’re gonna have to look at similar trials see see if they’re NNT’s and NNH is available that you can compare these two. Because again you can’t look at them in isolation in a single trial to be able to clearly understand what the data is telling you
Conclusion: NNT/NNH Comparative Numbers
Prof. Mary Ferrill gives a summary of this lesson.
Continuing from the previous video, we can find the confidence intervals in this example are extremely wide, which indicates the data is not very precise. Also, we should notice that the percentages are very low in both case and controls.
Besides, she takes three of the NSAIDs to make an example. We can find the NNTs are very small, which means most NSAIDs work well in removing pain. However, the NNHs are much bigger.
In conclusion, we need to analyze a study instead of just reporting what is there. We need to verify all the numbers provided in a study to reproduce those results and interpret them correctly.
It is easy to calculate NNT, but it is difficult to determine what it means and explain it to a healthcare professional. Thus, remember to look at both the NNT and the NNH together to make sure the benefit is worth the potential risks. Put this in the context of a full literature evaluation and not just focus on the numbers.
This is the final video in this week. If you have any questions or thoughts, please share them below.