Name:
Statistics for Orthopaedics
Description:
Statistics for Orthopaedics
Thumbnail URL:
https://cadmoremediastorage.blob.core.windows.net/32cfef48-e2a4-48bf-ade1-df74e3aafa5e/videoscrubberimages/Scrubber_1.jpg
Duration:
T00H39M13S
Embed URL:
https://stream.cadmore.media/player/32cfef48-e2a4-48bf-ade1-df74e3aafa5e
Content URL:
https://cadmoreoriginalmedia.blob.core.windows.net/32cfef48-e2a4-48bf-ade1-df74e3aafa5e/Statistics for Orthopaedics.mp4?sv=2019-02-02&sr=c&sig=FZjCBhs8oIK3zcP%2B4J1UUPsMdQ7A9HYUgfH6DC2qGV0%3D&st=2024-12-08T19%3A30%3A44Z&se=2024-12-08T21%3A35%3A44Z&sp=r
Upload Date:
2024-05-31T00:00:00.0000000
Transcript:
Language: EN.
Segment:0 .
Good evening, everybody. Welcome again to another session of the flexor-pronator webinar, a teaching session on Bible session, which will be at the end of this talk. Naturally, the talk will be recorded and put on our website. And on our YouTube channel. The Viva session will not be as you're aware of the farc, as militant group that is always providing webinar sessions on Wednesday evening, but we also do courses.
If you go to our website, you'll find Viva practice courses there. There's quite a few courses coming up again, so keep an eye out. We plan to keep always maintained, catch up, keep up with demand. We have the concise orthopedic notes textbook. We believe it's very good. We think it's worthwhile purchasing any proceeds from this go to help maintain our ability to provide this service free.
Also, if you're visiting our YouTube channel, please do subscribe because again, the advertising on that is actually used to maintain the cost of providing these webinar sessions. Of course, all our mentors provide their time free of hands, and today we're lucky to have a or vertice who is going to give us a talk on statistics for the fastest part two. Part one is on the YouTube channel.
Everyone knows how he used one of our fastest mentors who has an interest in, of course, basic science. But his main area of interest in surgery is the surgery. Without further ado, honey, if you want to go ahead with the evening, everyone. So today is the part two of our statistics for emphasis. So I just want to remind everyone about our work, which is I think it's very helpful for preparation of efforts in general.
So what is objectives today? So we will discuss the bias. Other types of studies, like court studies, case control studies, case series and expert opinions, critical appraisal, phases of clinical trials, how to conduct trial screening, test box whisker plot survival analysis, measuring performance like the plot and outcome scores. So bias, what is bias, so bias like a flaw in impartiality, like any systematic error while in the methodology of fewer?
Like study or trial, whatever it's considered, biospace has different types. So the first one is a selection bias. So you are doing a trial or a study and you comparing like one type of knee replacement to another one. And you selected your patients straightforward cases when in that arm and elderly patients with these high BMI and other arm.
That means that selection bias because what that would affect the result, because we are expecting like from elderly people like obese patients, more complications. So that would affect us considered bias. So it is a systematic error due to non-random sample. So how you can deal with that is randomization. So we shouldn't know, like if you are doing a trial comparing between two things.
You have randomly allocated like the patient in each arm, you don't know that will be blinded, so that is to avoid selection bias. So another type of bias is ascertainment bias. So knowledge of intervention, if the surgeon or the patient's nose, the like, is this patient's going to this treatment or the patient's nose that is going to this treatment that is asserting bias, especially if there is a belief that is like one treatment is better than the other.
So if I'm doing like comparing like a knee replacement with another prosthesis, so I don't know, I will do the surgery, I don't know the patient in each arm. I will do, except inside the surgery, but not before that. And overcome again by randomization. So and this is a publication bias as well. So you finish your study. And you analyze your data, but you found some complication, some like side effects and you hide that.
You change your result is publication bias. So randomization, we discussed that in the last lecture, but we will just a quick recap how you can do randomization. You can do randomization by computer like computer-generated or stratified a stratified if you are going to do like. Small size sample of population like, for example, in each arm or 20 each arm, you will try to.
Each group should have the same numbers of population like age, gender, BMI. Like if! The same, the same thing. Knee replacement. So I have in each arm, I should have the same number of patients over 65 and the same number of young patients, the same number of females and the same number of male.
And so on. So that's called stratification. But if you are doing like very big study, multicenter study like and the population size maybe 1000, it's very hard to use a stratified. So we have to do blocking block. You will do like each block, like for an example, has about four or six patients and you will try to include in every block the all categories of population.
So like elderly patients, obese patients, low-high BMI, low BMI, male, female, young and elderly. So that's called blocking. And that is because stratified and block are similar. But usually this block in large studies. And that is why so you do like it, should local be small and it has everything, all the categories. OK, so we'll go next to the other types of studies, like a cohort study, what's cohort study cohort study is this it's all the same like randomized controlled study, but it's no randomization and it is like retrospective.
So randomized controlled study is a prospective. So you will start now you collect the patients and you will compare the outcomes in the future. The court is the reverse like you were doing. You had like your compare like between like a 2 knee and triathlon knee for the last five years. And you you will assess the measurements bit like Oxford knee score, Harris knee score and whatever, and they will compare outcomes between the two arms.
That is a cohort study. So there is no control here, because you compare something in the past, not in the future. And there is like case control study, it is almost always the same like cohort study. I actually got confused between differentiation between them, especially in the MCU. Actually, the case control, you have a case and you have a control group, you have these studies where individuals with a certain outcome case are compared to the individual without the outcome that is a control.
The good thing is, this study is it's quick and cheap to perform, and you can give you relevant information. The other issue for an example, if you have 100 patient got exposed to a new vaccine like COVID vaccine and 10 patients got like. DVT, so the risk ratio here is 10 percent, and the other issue here is 10 by 90. So you can calculate that from the case control study.
And that's particularly useful in trying to identify the cause of uncommon disease and how it has a lot of methodological biases because there is no randomization and it is like it is the retrospective. So it has a lot of methodological biases. And the other type is a case series, which is observational study like example, you are doing.
Like knee replacement in vulgar deformity, and you'll do a 50 cases will follow up this patient for four or five years. There is no comparison, no randomization. But that we usually use this case series as a basement for trial, like a pre-trial stage, like, for example, if there is like a new processes like procedures for the tools for an example, like first empty, empty arthroplasty, it's the new or new things, not it was used before.
Like we lose the case series first three cases of the new arthroplasty in the tool and the will full of that and you'll get your outcomes and you lose that like a base for sitting a randomized controlled trials. And the last thing is the expert opinions like elite surgeons give his experience and his opinion, according to his long experience.
We'll go then to a critical appraisal, actually. I don't think it's common in the Air Force exam, but maybe like may come as a surprise question. So what is the critical appraisal is assessment of the research study like, is this a study worth to clinical practice? A conclusion from this study was the clinical practice. Yes or no? It it has four phases.
No one rabbit critical appraisal like use a vehicle technique like population intervention comparison outcome, and that is the rabbit critical appraisal, then you evaluate you do evaluation of these studies together and put the results in a table and then compared as a body of evidence after that census, you will like certain data, will provide a snapshot and use this particular clinical issue pulled out and put it like a showcase from the table.
After that, you will come to the recommendations. About the best practice. OK, so evidence based practice process, what is that like, how you can get the guidelines recommendations? So number one, you will identify a separate inquiry to notice the internal data that indicate opportunity for positive change. Then you have the clinical coach and using the Bickert.
So we'll use like the papers, all the studies about this topic and you'll assess that to leading to population intervention with a bigger technique. And now is the added the tea, which is a time of the outcome, like five years outcome. Three years outcome. Then step two conduct a systematic research. To find out what's already known about this clinical issue, you have to go like go like PubMed or whatever and you find about what's written about this topic in the literature.
Step three conduct critical appraisal again, so we'll do the four steps, then implement best practice, blending external evidence with clinician expert experience and the patient preference and values. Then evaluate that. So after implantation, you will evaluate evidence implementation to see if the study outcomes happen in practice and if the implementation went well or not.
OK, and the last thing was step six. Share your results! OK, so I think it will be like a surprising question if the comments emphasise, but it's not, it's not so difficult, you can easily answer that question. OK, so this is a common one. How to conduct a trial. You are doing ask me how you conduct that. No one identifies a problem for an example.
Patient with ankle osteoarthritis will treat them with the art through disease or total ankle replacement. That's a problem. Then identify the gold standard. I will go to search about the topics. What what is the outcome? What is the benefits of that research? And it is like, can I do that or not?
Is it safe or not? Then I would write my protocol design study population. I will do part analysis how many patients needed to be involved in each group, how to do a power analysis, such as a program you will enter like you study data that the patient reported outcomes that are going to the complications risk. And after that, he will give you. The exact number used in each arm and usually like if it's like that, it tells you like you need to include about 88 patients in your trial.
That means 44 in each trials. So you have to put extra patients will use a 50, 50, 50 versus 50 because to avoid like definitely you have a patient loss of follow up. So to get a significant result, add extra number of patients above that after doing the power analysis, then decide what kind of randomization are going to use. Like do they like computed randomization, a ruling slap stratified blocking technique, whatever.
And you have to mention that what's your inclusion and exclusion criteria? So I will exclude patients below 50 years. I will exclude patients above certain limits. I will. I will exclude patients with diabetes, Charcot disease, whatever. And you have to decide what is the study timeline, so I will do a follow up of pools like arms for five years, 10 years.
OK, now we have the protocol. You will send it to for ethical approval. So you have ethical committee in each University. You have to send and you have to get ethical approval before starting your research. After that, collect data, analyze your data, write up your paper and publish. OK, that's me.
It might come as a separate question. What is the ethical committee so ethical committees professors from the university? And, OK, how you can get ethical approval for your trials. OK the application must include what is the study details? Number one, number two is a consent you have to attach the consent form in your application. Plus the protocol.
Who is the investigators like, who is the main investigators? Who is like who is going to help him? Then this is any funding, any sponsor, so which company are going to use for the annual arthroplasty is are you going? Are you going to receive any funding from that company? Yes or no? That's very important to highlight in your application. And started insurance, each city has like has to have insurance like.
If you are working in the NHS, in Ireland or NHS, I think all the doctors have insurance related to the HSC or NHS. But but you have to mention that. Five, what is the details of the procedure to which humans will be subjected like you have it will be part of the descriptions of like your study and what is the benefits. So you have to mention the benefit of this trial.
OK, for an example, all the patients that got. At anchor tenodesis, they are young, not happy. There is an all-star status of the all-Star status of the stars aligned for an example. They lose the function, but there is a paper published about the ankle. Arthroplasty mentioned is a patient. Satisfaction is satisfactory. The range of motion is better and there is.
There is a complication is not that bad. So we have to convince the ethical committee about your trial. And you have to mention the potential risk. OK it is not the best thing in the world. It has some complications like infection as aseptic closing, but is like any others are so blast. But it's not successful like a knee or hip replacement, but in some part, or some studies mentioned that the risk or complication is not that bad.
And alternative policy there as well, like Ankit arthroplasty or insoles, bas status treatment. Are you going to receive any payment or reward from this study, you have to mention that as well. And information off of previous ethical applications. Usually if you are going to apply for our city, if there is another city with the same title mentioned that will like make your application more strong. And will easily be approved.
OK this one might come as a separate question, OK, consent me for your trial, how you can consent a patient for our city. What's the criteria of that concern, it is different from consent from knee replacement or surgery? It's a different one. You have to mention at the beginning, what is the title of that study?
It's very clear who is the researcher, the name and the detailed phone number, and the email address is not necessarily to be like the. Is there a consultant or the registrar phone number, but at least the hospital or the orthopedic department secretary? Phone number and the email official email. And you have to mention the purpose of the study, detail of the procedures like are going to do like, for example.
We are running a trial now in our hospital about this 3D structure like four weeks versus six weeks. If you are going to do a concert treatment, four weeks, six weeks in cast. So what you have to mention that to the patient. Risk, what is the risk after that, you might get stiffness or pain, like loss of reduction, for an example. You have to mention that and the benefits.
And you have to be. To tell the nation, like less time for immobilization, it might have some benefits for earlier range of motion, less stiffness, early rehab, less risk of complex regional pain syndrome, for example. And you have to mention the randomization you have to tell the patients that he will be randomized to one of each arm and that the researcher will not know that is maybe like independent research nurse who will be responsible for that.
And the computer will be responsible for that. Will you have to mention that in the consent as well? And the very good things you have to mention that the patient has the right to withdraw from this research at any time? OK that's very important point in your consent. And confidentiality, so you have to mention that the patient name, patient detail will not be involved in your study.
So anonymous everything could be anonymous. OK, so next. Screening this is very common if asked this question, how you can do screening, test no one identify unrecognized disease in people without sign and/or symptoms must have very high sensitivity and very specific. What is the WHO guidelines for screening test? You have to have an own effective treatment for the condition, and the pathology will be understood.
And the facilities of diagnosis and the treatment should be available. And it should be a latent stage of the disease, and there is the test itself, non-invasive and acceptable to the population. And natural history of disease is to be understood, should be agreed policy to whom to treat as well, and of course, it should be as cost effective as well. And if the treatment starts early, is more benefits from started later.
OK it's another one. Sometimes it might come like a separate question. What is that? That is box whisker plot. OK, tell me about that. What what you see here. So why we use it number one? So analyze numerical data. So information about range, median interquartile center density and spread for numerical data.
That's very important. OK, what is the line in the center is the median. What is the other like each one like the green is a 2 green line, the first and second quarantine. Quarantine interquartile range indicated by the box. And there is a whisker, which is the line up and the line, the vertical line up and down, which is upper and lower value beyond considered outliers.
OK, so the survival analysis, how you can do survival analysis, there is a different methods of analyzing the survival outcome of intervention. It's very common for hip arthroplasty like hip and knee arthroplasty doing survival analysis. A like intervention is plotted over time. So, which allows the valuable dates of entry and different lengths of follow up.
You can do is like table. How you can do that. Like life table, like this table, you will enter all the patients by numbers and you will mention the complication and what is in the bones should in the bone should be defined like failure. What is the failure here? Like revision of losses, that is the end point.
And you mentioned that in the table. OK, so how you construct life table for joint replacements in the bone as well, it should be defined and the number of joint being followed and the number of failure determined for each year after operation should be mentioned as well. And for each time period, what is the number of lots of flower?
The number of complications and failure for them? Another way is a kaplan-meier survival analysis. So we'll put in front of you that and you'll ask you, what is that? So I usually start with the X and y. So what is the y is the proportion of failure and what is the X is the number with a period of time. And you have three lines, you have two dotted lines, which is the standard deviation.
And there is a line in the center, which is the medium. So outcome of the intervention plotted over time X-axis shows time y-axis percentage of study population surviving dotted line are the confidence intervals. This is two standard deviations. So you, as you see here with the time, the distance between the two dotted lines become widen. So that is the conflict because you lost follow up observations.
So now we have a wide confidence interval. So the. Proportion or the ratio of uncertainty become high. OK so and you have this line, this is more lines here. Uh, this is represent like a loss of full or patient death or revisions. So you have a steps here and this is line and you have here like not all like loss of up or if the patient refuse to participate in the after that, that's not a failure, but we call that like a positive.
Or censor the most patient or positive outcomes. So the sense of lost patient here like an infant was the little mark. This is small mark. OK, so some patients refused to do a follow up. OK, so. Measuring performance, so it is very common as well. Very common emphasis. So what is that?
The same technique. What is in the x? What is y? What is the line, the center? What is the two blue lines? What is the two red lines? What is these dots? So that is your technique of answering this question. So, so you have the White lines with the proportion of failure and what is the X line is the numbers over time.
OK, so what is the line in the center, which is the median? Sorry, the mean and you have two blue lines, which are the one standard deviation and you have two red lines, which are two standard deviations. You have dots above here. This area code posted negative outliers. And this area here under the red line is called positive outliers.
That means it's less complex. Here is more complication. OK, so how you interpret that? So this area here. So they are not a bad surgeon, but need justification. So the buzzword here is justification. So it might this surgeon, they do like the most complex cases or they did like a small number of patients, like if you did like a 10 total hip replacement and there is to a dislocation or 2 infection that is 20% is very high like percentage.
But if you did a 100 it might only have only two or three dislocation. So the area here is just need justification. And the area, he says, is a positive this is very good, but as well it might did like five hip replacement and there is no complication at all. Your your complication rate is zero. That that doesn't mean that your best surgeons award, but because that you did only a small number of operations.
But here, when you go into the right, that's another thing. This is the negative outliers here. You have to do something about that. So that's on like unaccepted complications with a big number of operations, so that one need like investigation. So you will if you've said that, well, let me ask you what you were going to do with this. Like this surgeon did like unexpected rate of infection, what we're going to do.
So it's the same like my duty towards the patient, my duty towards my colleague and my duty toward my institution. So for infection, I will stop him doing surgery. For patient safety and for him, I will go through his case, I will discuss his technique. I will see what he did, why he has this big number of infection and for the institution, I will speak to the lead clinician to discuss the higher rate of complication for that surgeon.
OK, so outcome scores. What is outcome course, so is the question here or given to the patients pre and post procedure like pre before the knee replacement and after the knee replacement in relation to the outcome relating to the intervention? So what's called patient reported outcome scores? Why you need it.
We we need it in the research. If you are going to like to use a new knee replacement, new hip replacement or a new surgery, how you can know that like your surgery is successful or not, you have to use outcome score. So you will use that for research, quality improvement, audit and economic evaluation. How to choose that? The outcome score have to be reliable, valid, and it's used in a similar patient demographics, so don't use like if you are doing like a knee replacement and you are working in the UK, it's better to use Oxford knee score, another one using like outcomes from North America or from the Middle East.
So what is the patient reported outcome, what is the promise? It may be patient specific objectives like you are like there is a common cold McMaster turned to arthritis. Why are doing to the hip replacements as the patient specific? And maybe diseases specific like for hep, there is a very simple score, Oxford type score.
Uh, well, my arthritis. And generic, like the most to comments like generic promises F12 and SF 36. They are very common use because the general quality of life. Maybe a regional specific Oxford school Oxford knee score. That is 12 question. It's so severity of pain right now that pain in the Southern veins, you may ask is the examiners may ask you, OK, so you have a patient specific, disease specific.
What's your tell me about a school use in your practice? So you have to pick one. And you have to know it. You have to discuss it to explain it to the examiner. So for the most two common example, Oxford hibiscus only score and only 12 questions, it's easy to be to recall that in the exam. Thank you.
One question from Gitmo. Abizaid and he says the central line in funnel plot represent the mean or represent the effect size of combined estimate of meta analysis. I couldn't. Would you repeat that again? Yeah, it's he's put the central line in a funnel plots. I think he means, does it represent the mean or does it represent the effect size of the combined estimator of meta analysis?
OK, I think he got confused between the funnel plot and forest plot. We don't use the funnel plot. The forest plots usually use that in the meta analysis. So there is a central line, which represents any difference between the tools like the control group and, Uh, Uh, investigating its other groups, the two groups, for example, if you're going to totally placement but tell from a joint arthroplasty for isolated Bertelli femoral osteoarthritis.
So this is one about one, but there is a diamond if this diamond like touching the central line. That means there is no difference. There is no significant difference. We use that in the meta analysis, but in the funnel plot is a different thing. That's a performance. So it's not like a study will assess the performance. It's a different thing.
OK, thanks, honey. I hope that made sense, so I've also got a question. Just going of see who's asked it from jabez? Is the cohort study always retrospective? Yes OK. And a question from Norman. What's the difference between a case control and cohort study? Yes, it's very confused.
One word. The cohort study is analytical study. The key is the case, the other type as a case control study is an observational study. So this is analytic and this is observational. OK that's all the questions that I have. I'll hand back to Sian. Excellent, OK.
Really strong parts 2 to the accent part one, which I'm sure everyone is going to go back in with you again, as well as this video as always, guys, if you have any questions, please do ask. But if there isn't any overwhelming question, does any of the members have any questions that would be pertinent for the candidates?
Honey, I want to ask if you were selecting your which study you want to do. How would you answer that as a question? Actually in the exam or in the exam, OK, I would like our big the studies that I really understand it we can speak about in the exam, for example, the city for me, I will visit the city because it has a lot of things to discuss and to the page, to the examiner about it.
Well, my approach to the question is specifically what is the question that needs to be answered by the study? So I wouldn't immediately pick asked because need to know, what is the question, what is the topic? So, for example, something extremely rare. That has very few patients and/or events you may not be able to do in our city.
Yes special criteria to be done. Yeah, Yeah. So my approach would always be to answer with what is the question? And I would design, I would design my study and select the type of study based on the condition. I'm studying the population, I'm studying eating and the event that I'm trying to study. So events which are very common, which every day and we can recruit large numbers of patients.
The ideal study would be a prospective randomized control, triple blinded trial. While the condition, which is extremely rare, which is very hard to find patients to recruit to the event is really rare. And maybe only a case series may be appropriate in these situations because we don't have enough patients to even begin a prospective, randomized controlled trial.
But a collection of data on those patients may be appropriate, but saying it this way, you recognize that. You studies the superiority of our society is not always available to us. Sometimes the case series is just enough or is all you can do and is the best possible thing you can do with. Does that make sense?
Candidates go everyone. OK um, answering this way will also demonstrate that you have thought about this and also that you understand the principles behind your research. OK, well, the. Any thing else? OK if there's no other questions? Excellent talk.
Thank you. So topic to get your head around and you did that really well. Thank you very much. We will end the recording part of this session. As always, we're going to have Bible sessions afterwards and just a reminder to everybody that we do this every Wednesday. You are very welcome to join the FARC. Farc preparation group.
Just send one of the mentors request to join, and you can always go to the FARC web FARC mental page and email us. We will add you to the FARC preparation group on arteriogram. Everything we do is given free, free of time for my mentor. So thank you again to all entries that participated today, that includes me. KneeKG syrie, who's joined us.
I'm David Hughes as well. Thank you, everybody.