WEBVTT
Kind: captions
Language: en
00:00:14.060 --> 00:00:18.060
So as I mentioned, let's look at a power example to get a little bit better of idea of
00:00:18.060 --> 00:00:20.260
what I was talking about before.
00:00:20.260 --> 00:00:22.040
So you can see what I've highlighted here
00:00:22.040 --> 00:00:26.100
they set a sample size of 205 patients per treatment group
00:00:26.100 --> 00:00:30.960
was needed to achieve 92 percent power to detecting difference of 0.4%
00:00:30.960 --> 00:00:33.700
in change in HbA1c from baseline.
00:00:33.700 --> 00:00:36.920
So you can see I listed the four components of power on the right.
00:00:36.920 --> 00:00:39.220
So it's the first one is sample size.
00:00:39.220 --> 00:00:40.940
Well, from this you can tell
00:00:40.940 --> 00:00:44.240
because remember I told you it's a sample size in the final analysis.
00:00:44.240 --> 00:00:48.560
So you need to go look at the results to see how many patients they actually had.
00:00:48.560 --> 00:00:50.440
And so we really would have to look
00:00:50.440 --> 00:00:53.060
and so you can see I want to look for us and it showed that
00:00:53.060 --> 00:00:58.020
the results had 233 and 223 patients respectively in each group,
00:00:58.020 --> 00:01:01.000
so that criteria for sample size was meant
00:01:01.060 --> 00:01:03.020
needed to be linked to the primary outcome,
00:01:03.020 --> 00:01:07.600
well you'd have to look somewhere else and change in HbA1c was their primary outcome.
00:01:07.780 --> 00:01:08.880
So that was correct.
00:01:08.880 --> 00:01:10.940
And the power of 80-90%
00:01:10.940 --> 00:01:14.040
but then linked to a clinically appropriate clinical marker.
00:01:14.040 --> 00:01:15.180
If you look in the guidelines,
00:01:15.180 --> 00:01:20.340
usually 0.5 is the the least amount of change that you see is clinically relevant.
00:01:20.340 --> 00:01:22.320
So this one is considered a little bit low.
00:01:22.320 --> 00:01:27.180
So it's easier to find a significant difference the lower they affect sizes.
00:01:27.180 --> 00:01:31.940
So I know there's a lot on this slide and I'll go over each part separately.
00:01:31.940 --> 00:01:34.540
But I wanted to put it all in place for you.
00:01:34.540 --> 00:01:37.860
So when we're looking to determine if the correct statistical test was used,
00:01:37.860 --> 00:01:40.560
there're three main things that you need to look at
00:01:40.560 --> 00:01:45.100
and that's study design, data types and the number of groups.
00:01:45.100 --> 00:01:49.200
So the first one is looking at is the study independent or dependent?
00:01:49.200 --> 00:01:50.680
And you'll say well why does that matter?
00:01:50.680 --> 00:01:53.160
Well, when they're calculating the statistics,
00:01:53.160 --> 00:01:56.540
an independent test will have a smaller N.
00:01:56.540 --> 00:02:00.740
Then you will end a dependent test because you have double the data when it's dependent.
00:02:00.744 --> 00:02:07.560
So you just need to make sure you can determine whether it was baseline to endpoint in each group, it was looking at two groups separately,
00:02:07.560 --> 00:02:10.000
to make sure you're picking the right statistical tests.
00:02:10.000 --> 00:02:16.720
And that'll become obvious when I show you the lists of the drugs that or the statistical test that you'll use.
00:02:16.720 --> 00:02:21.560
The next one is determining the data types and this will determine this statistical tests that are used.
00:02:21.560 --> 00:02:23.640
And again I'll go over those a little bit more.
00:02:23.640 --> 00:02:26.020
And you can see that there're some other little caveats, too.
00:02:26.020 --> 00:02:32.320
To that, besides determining, the datatype that I'll also go over as we get to those sections.
00:02:32.320 --> 00:02:35.500
And the last one is the number of groups.
00:02:35.500 --> 00:02:40.467
Now this is important because you'll have some statistical tests can handle two groups
00:02:40.467 --> 00:02:42.720
and some can handle two or more groups.
00:02:42.720 --> 00:02:48.700
So if you can use a study for for two or more for three or more groups, you can also use it for the two.
00:02:48.700 --> 00:02:51.520
But if it can handle two groups it can't handle three.
00:02:51.520 --> 00:02:57.840
So you'll see that as we go along as we list the options for each statistical test that could be used.
00:02:57.840 --> 00:03:02.920
So the data types as I mentioned after determining this study design is very important.
00:03:02.920 --> 00:03:05.660
And even though there're four types of data,
00:03:05.660 --> 00:03:09.420
it's really convenient that you can list them is interval/ratio together.
00:03:09.420 --> 00:03:12.420
Because the same statistical test will be used for them.
00:03:12.420 --> 00:03:16.180
And I know people always want to know what's the difference between interval and ratio?
00:03:16.180 --> 00:03:18.500
Well, interval and ratio are both continuous data,
00:03:18.500 --> 00:03:23.780
so it's most of the things that you think about that we see in medicine, blood pressure, height, weight.
00:03:23.780 --> 00:03:25.840
Ratio is just an absolute zero,
00:03:25.840 --> 00:03:30.840
so it's using Fahrenheit versus Celsius type of a thing for a zero scale.
00:03:30.840 --> 00:03:33.220
So there's no reason to differentiate between them.
00:03:33.220 --> 00:03:38.040
So you just need to know is it a continuous scale or not and that's listed as interval/ratio data.
00:03:38.040 --> 00:03:40.460
And this is considered the most specific.
00:03:40.460 --> 00:03:44.460
The middle data type is ordinal which is rankings and ratings,
00:03:44.460 --> 00:03:47.280
a lot of the subjective data with that we have
00:03:47.280 --> 00:03:53.220
with Likert scales and psychiatric scales and things like that.
00:03:53.220 --> 00:03:56.000
And then the least specific type of data is nominal
00:03:56.000 --> 00:03:59.420
and you'll see that that's primarily yes/no,
00:03:59.420 --> 00:04:04.240
although it could be yes no no no if you've got more than two groups.
00:04:04.240 --> 00:04:07.080
So it's very important that we be able to differentiate between these
00:04:07.080 --> 00:04:09.620
and we'll go over some examples of them
00:04:09.620 --> 00:04:12.880
to help you to get a better feel for what they are.
00:04:13.220 --> 00:04:14.600
Now converting data types.
00:04:14.600 --> 00:04:16.340
Now you're not going to be doing this,
00:04:16.340 --> 00:04:20.780
but it's something that the statisticians in the article might do
00:04:20.780 --> 00:04:25.120
and there're appropriate conversions and there're not appropriate conversions.
00:04:25.120 --> 00:04:26.960
Now interval/ratio data,
00:04:26.960 --> 00:04:30.160
and we'll get to this in a minute with the previous slide,
00:04:30.160 --> 00:04:34.380
in order to be able to use a test on interval/ratio data,
00:04:34.380 --> 00:04:36.500
the data must be normally distributed.
00:04:36.500 --> 00:04:39.180
And sometimes and you've got very small sample sizes
00:04:39.180 --> 00:04:42.600
or even sometimes in larger sample sizes it may not be.
00:04:42.600 --> 00:04:45.600
So you can't use a certain statistical test on that.
00:04:45.600 --> 00:04:50.860
So it's appropriate for the statistician to convert that interval/ratio data to ordinal.
00:04:50.860 --> 00:04:54.680
You're not doing it, they will be, and use an ordinal test.
00:04:54.680 --> 00:04:58.260
So we'll see that when we go through finding things that
00:04:58.260 --> 00:05:00.000
if you can use an interval/ratio test,
00:05:00.000 --> 00:05:03.100
you can also use an ordinal test on that interval/ratio data,
00:05:03.100 --> 00:05:04.880
and that's not a violation.
00:05:04.880 --> 00:05:07.100
Now again that's the only conversion that's acceptable.
00:05:07.100 --> 00:05:10.940
And the other way, you're losing too much of your data specifics
00:05:10.940 --> 00:05:16.760
so you can't convert ordinal to nominal that again loses too much of your data types.
00:05:16.760 --> 00:05:19.400
Now one conversion that they'll sometimes make
00:05:19.400 --> 00:05:22.780
which is okay as long as they present it both ways.
00:05:22.780 --> 00:05:25.380
If they take their interval/ratio data
00:05:25.380 --> 00:05:28.340
and change it into response versus nonresponse,
00:05:28.340 --> 00:05:30.320
and that would be nominal nominal data.
00:05:30.330 --> 00:05:34.320
That's all right as long as they provide you with the original specifics of the data.
00:05:34.320 --> 00:05:41.320
Because otherwise, you're not getting as robust of results if they only report it as a response and nonresponse.