Listen to Sarah Gill, Editor-in-Chief Professor Ben Ollivere and Deputy Editor Brett Rocos discuss the August issue of Bone & Joint 360 in the second episode of BJ360 Podcasts. This includes a brief insight into a selection of six papers and a deeper discussion of two papers.
Listen to Sarah Gill, Editor-in-Chief Professor Ben Ollivere and Deputy Editor Brett Rocos discuss the August issue of Bone & Joint 360 in the second episode of BJ360 Podcasts. This includes a brief insight into a selection of six papers and a deeper discussion of two papers.
[00:00:00] Welcome back to part two of the September 2020 BJ360 Podcasts. I'm continued to be joined by Professor Ben Oliverre from Nottingham and Mr Brett Rocos from Toronto. So picking up where we left off guys, I'm going to start now with my three papers to highlight from the Roundup section.
Really the theme that sort of joins them is I found them interesting because I can immediately relate them to clinical practice.
You know they were pretty simple papers with simple messages. And I thought this actually makes me think about the job we do on a day-to-day basis. The first one was from the trauma section, which was a paper from Adelaide and it was looking at the prevalence of rotational malalignment in tibial fractures, following intramedullary nailing.
And it's something that I think we all sort of try and think about at the time and certainly worry about afterwards. But this paper highlighted that actually, they noticed in up to 25% of their tibial nailings, there was a 15 degree or greater difference between the injured and nailed [00:01:00] compared to the contralateral side, which is certainly quite stark. I would say that 15 degrees, you know, we would probably say that was important and we'd like to think we're better than that. And I suspect this paper suggests we're not.
But the other interesting thing this paper mentioned is that they noticed this was far more likely to be identified in the left side compared to the right. And that's because in the normal anatomical limbs that they compared, there was a four degree external rotation difference between the right and left with the right being external rotated. So that was an interesting observation as well as part of that paper. The second one was from the Children's section, and this was a paper in San Francisco, looking at the popular topic of the contralateral slip in *inaudible*.
And they were basically trying to find predictive factors as is common theme in this area of the literature. And they identified age less than 11 and Oxford score less than 20 and an epiphyseal angle of less than 21 degrees, [00:02:00] were all independent factors associated with subsequent slipping the other side.
But interestingly, when all three of those factors are present the risk of subsequent slip was 70% and that might be of interest to our colleagues who do children's in particular pediatric trauma in trying to predict when and if the contralateral pinning is relevant.
And finally from the research section, I thought this was a real gem of a paper from China, from Hao, looked at the bacteria found at the initial debridement of open fractures, and then compared it to subsequent causative organisms in infections that they're treated and they found, as it's probably predictable that the most prevalent causative organisms later on were gram negative or more polymicrobial, but in 60% of cases, it varied from the initial cultured.
And I thought that had immediate implications to our practice in A how often are we achieving samples and what treatments, [00:03:00] are we sorry, what samples rather are we basing treatment on and B, where are these other bugs coming from? Should we be looking at lots of literature? Obviously coming out about negative pressure dressings or occlusive dressings, et cetera. But really if we're having that many differences in terms of the bacteria at various stages, should we be really honing in on how we manage our open fractures?
So they were my three to highlight. If I come to Professor Oliverre. You're going to give us a look forward, is that right, into what will be the October edition of the 360?
Yeah, a couple of papers to wet people's appetite, Sarah. I'm quite interested in, and I remember editing it as Brett submitted it, his edited summary of the, of the Spine paper and I think it really is relevant just across the whole of practice. You know and I remember reading, and thinking I wonder if that's cause they've had lunch or not, whether it's because they're later on, whether it's because the nursing staff haven't got them, right. Whether it's because they get to recovery and everybody wants to go home and just put someone to the ward. And I think, you know, it is just worth bearing in mind one of the joys of [00:04:00] 360 is you get easy access to the institute from outside of your specialty, because I never would have read that had I not been involved with 360.
And I think, I think that that's really important. And I've tried to pick my papers around the theme and maybe I shouldn't have done having heard Bretts little picks and your little picks, but I've chosen to focus on that I think is, is highly topical actually. And it's relevant to everybody who's got a trauma practice and probably everybody who's got a spine practice or, or anything by a pediatric practice actually. And that that sort of catastrophic bone failure and how we, um, how we manage that.
So I've chosen three papers to do with supracondylar fractures of the femur and, you know, As you know, um, the problem here is that as you get older, your bone changes shape and not only does your bone change shape, but it becomes osteopenic. And the, the constructs that we put in are becoming increasingly ambitious in order to try and prevent preiprosthetic fracture, try and prevent fatigue failure. And, um, on the one hand, you know, if you, if you sort of heal with steel, there's [00:05:00] the advantage that it doesn't, um, it doesn't result in failure of the metalwork quite so often. On the other hand, you've got the biological issue.
So I'm just gonna run you through three very quick papers, which I think put into perspective this issue with distal femur. The first one is a paper from Wright et al, from California. And what they were asking was, you know, what is the best kind of supplemental fixation?
So it's become pretty popular in the States now and it's becoming more popular in the UK, not just to put a nail in our plate, but to go well, you know ones good so maybe two's better. Um, and they looked at a nice little paper looking at dual plate and plate nail constructs - a plate nail construct being essentially a nail on a plate, where you lock the screws through the nail. Those who haven't done them. And obviously you've got a, you've got vastly different biomechanics there.
So they did a whole lot of initial experiments on 24 synthetic osteoporotic femurs. And they then went on to refine that in paired cadaveric studies, cadaveric specimens, and what they essentially were able to do [00:06:00] was to quantify the increased stiffness that you get from these constructs and essentially demonstrated that the dual plate constructs were much stiffer than the plate nail constructs which they concluded was it was a good thing. And obviously is a good thing if you want to make it stiffer, but if you don't want to make it stiffer, then perhaps it's not as such a good thing, but it's good to understand the mechanics behind this and, um, Which is why essentially I've, I've chosen three papers on the same topic cause that.
That then brings us onto the second little snippet, um, which is about, double plating of the, of the femur. Um, and it's from Rolec et al, based in New York. And what they were looking at is what happens to the blood supply. And it's a really fascinating paper, its a cadaveric paper, but they did contrast CTs and MRI scans on their cadaveric specimens and how on earth they did that, I'm not entirely sure because it sounds sort of implausible but they did and they were able to validate it. And they looked at some. They looked at, um, filling, the vascular supply then undertook their plating and were able to look at what happens as far as the blood [00:07:00] supply is concerned, which is one of those things that's always, always worried me about this sort of dual plate constructs.
And they found that actually there was only 4% difference in terms of damage to the small vessels and from dual-plating versus single plating. So. That kind of, you know, is potentially a practice changing thing for me. Cause I wouldn't have considered dual plating because of that kind of strippage, *inaudible* kind of thing.
And it is becoming more popular and maybe it is because you can get away with it. And that, that paper sort of, um highlights that potentially as a, um, as a thing. Um, and just as an aside, they established that plating the distal femur full stop reduces your blood supply by about 20%. So it's 20% versus 24% reduction in their, in the blood supply.
Of course they haven't, um, They haven't told you how safe it is to dabble around on the distal middle end of the femur. And then, you know, not for the faint hearted, but I do do it occasionally.
And that brings me to the third paper, which kind of explains perhaps the reason that even looking at putting plate nail constructs or dual, uh, dual plating, which is a [00:08:00] paper from Greenville, South Carolina.
Uh, it's Begad et al and they just asked the question, um, You know, really do the distal femur plates fit. And so they took CT scans of intact femurs. It's pretty straightforward, you know, kind of resident registrar project, they use trauma CAD. They had, um, uh, I can't remember how many exactly it was. Uh, did a 32, uh, distal femurs and looked at the common brands of implants and amazingly on average, the best fitting plate, which was a striker plate, stood off by five millimeters half a centimeter. So it's not surprising if you put something on like that on a canter lever or you don't put it on its canter lever and you over reduce the fracture. So you're giving the, giving the patient a varus or valgus deformity. They're going to have a problem.
Next up was the Smith and nephew previous seven millimeters, the Synthes is a whopping eight millimeters off the femur and Zimmer a whole centimeter 9.3 millimeters off the femur, on average at the widest point. So maybe the solution here is to make [00:09:00] implants that fit rather than taking the implants that don't quite fit and adding a second one to make it stiffer.
So I those two papers just kind of sat together really nicely and told a story about a problem and maybe the solution for it. And maybe my own view is that we should be thinking about either accurately contouring our plates or maybe coming up with some better ones that do actually fit the average femur. So those are my three little picks.
Yeah, I really enjoyed those Ben actually I liked the theme between them. They do sort of lend themselves to discussing as a group. Um, do you know, it's something that comes up quite often in discussion and I, you know, being North of the border and we've sort of slightly spared from this, but in terms of best practice.
Is that sunshine, youre talking about?
Listen that, you know, I wouldn't tell you the figures about Glasgow and the number of days of rainfall. It's too depressing to talk about right now.
Um, but yeah, so the, um, best practice of tariff in England, my understanding is that some of the standards that apply to Neck of femur are going to be rolled out and [00:10:00] applied to femoral fractures. Um, potentially later this year or next year. And, you know, you're going to need, basically, we're going to need to be fixing these, uh, femoral fractures in osteoporotic patients in a way that allows us to mobilize them day one, which is what we expect of our Neck of femur fractures. And for me, that means that there's going to be a, you know, a significant increase in interest in these either nail plate combinations or dual-plating, as you say. So definitely an area of big interest.
I think there is, I think I mean, talking about that from a philosophical perspective and it is interesting, you can't prioritize everything because you end up back where you started. So how much do you prioritize in order to get things to the top of the list?
One of the things that worries me a little bit, although, you know, it is, it is, it is my friend and colleague, Chris Moran who pushes these, these things forwards. And you know, he's a wise man, but one of the things that we failed to do in the NHS and in hospitals in general is increase capacity. So if we make too much stuff an urgent target and you've got to do ankles within a certain period of time. We've got to do hip fracture in certain period of time, femoral [00:11:00] fractures, open fractures. What happens to the poor little old lady with a wrist fracture whos there, you know, a number of days later still sitting on the ward?
So there is a, you know, I think he's absolutely right. Don't get me wrong that we, um, that we, uh, can strive to do better and particularly for the frail patients, but I think it's really important on a local level that these things are used as levers to increase capacity rather than, than just chasing the target and artificially making other patients wait longer. It's a, it's something that we've really got to do as a as a system change lever rather than simply become target driven like a lot of family health in general practice doctors have become, you know, where you go in and you go and because you sprained your ankle, you get your cessation and smoking advice and they tick two boxes.
Yeah. Yeah. I think they're very valid points. And on the second paper there we had a discussion recently, actually actually to do with proximal tibial fractures, um, to do with the anatomic plating system they were using. And the designers of these plates tell me that the proximal tibia is the most anatomically [00:12:00] varied area in the body and that's why none of the plates have a perfect fit and things. So actually, It's, you know, it's very interesting to hear about the studies of the distal femoral plating uh, fitting. Cause it's sort of the problem with anatomic, I suppose they're just composites aren't they of everyone's anatomy and therefore they actually anatomically fit no one.
Yeah, it is. I mean, it's not rocket science, is it? We've all done knee replacements, knee replacement systems come in, you know, come in two sizes, sometimes male and female, and they get two sides, sometimes male and female, and a bunch of sizes. There's normally six. So, you know, that gives you at least 12 different geometries. And yet we're surprised when one average plate doesn't fit every single femur.
Yeah, well, surprise. I don't know if I'm ever surprised that mine don't fit, but you know, disappointed maybe yeah. So now I take your point, but we haven't said anything about spine or paeds for about six minutes now. So if you can wake Brett up, um, we'll move on.
Yeah Im feeling left out.
Just chime in, you know, if you're waiting for Ben nothing would ever happen would it?
Well, you know, yeah. You know, our implants, aren't [00:13:00] typically anatomically fitting, you know, we contour and rods, our cages are fairly uniform in size and shapes. It's not something we encounter too often. Though you do see it with condyle plates and, um, occipital, cervical fusions and so on so I'm sure it will come to us eventually, but it's interesting that, you know, going back to Ben's point about the knees, you know, every single patient of ours, the rods are a different shape. You know, there is an infinite variation and we don't think too much about it.
You know, in the old days, when they did use plates, they were using one-third tubulars. And what we now call the recon plates to fix pedicle screws and we're contouring them as well. There was never an attempt to anatomical implants in any shape or form. So yeah, I think it's... this personalized thing is coming.
What we are seeing on this side of the Atlantic now is more and more sort of customized implants and so on using 3D printing. So I'm sure that's the future, even in trauma.
Even in trauma.
Im sure it's a great idea, Brett. [00:14:00] Hang on there's an anatomical plate coming. Don't worry. The printer's just on it.
Long traction for six days and anatomical plates. Perfect. Um, right. We're going to move on to the sort of the main course as it were.
The paper that I had highlighted, um, for discussion guys was, um, a paper, uh, from Florida, um, which was in relation to the modern, uh, results of functional bracing of humeral, shaft fractures, a multicenter spreads, respective analysis.
And this is from Serrano and sharp in, in Tampa. And I picked this one really because humeral shaft, fractures - when do you leave them alone? When do you intervene, which ones need fixing? I think is just a question that comes up in every fracture clinic, you know, is this okay in a brace? Is this brace alright? This brace is rubbing, do we need to do anything with this? And I think it's an area that, you know, um, we [00:15:00] could never be too well-informed about basically.
And what the guys in Florida did was a large, retrospective cohort essentially. And they took data from nine centres, nine level one centers in the US so we need to sort of, I think just acknowledge the fact that these are all big trauma centres in the US over a long period period between 2005 and 2015, so 10 years worth of data from these nine centres.
And they based their analysis and conclusions on the followup of nearly 1,200 patients. And the big headline of this paper was that only 71% of those 1,200 patients went on to heal without some sort of surgical intervention. So the big headline is, you know, almost 30% of patients ended up with some sort of intervention for humeral shaft fractures.
Now they only looked at shaft fractures, nothing with an extension into the shoulder or elbow or elbow joints. [00:16:00] They excluded any patients that had an ipsilateral additional upper limb injury. So whether that be elbow wrist, forearm, whatever they all went. So it was all humeral fractures. And basically there was a quarter of proximal fractures, quarter of distal, diaphysis and half of them were in the middle.
So that was sort of the spread basically. This paper was interesting, I thought, cause the lead author had already published their results of their single centre, which again was quite pro you know, pro-intervention or rather, certainly highlighted deficiencies in purely conservative management of these fractures. And I think they then went and made this much larger collaborative paper.
So in terms of their treatment algorithm, so that we can compare it to our practice. They said that these patients were seen in the emergency department and they were put into a coaptation or a splint or backslab. So sugar tong splint or a backslab [00:17:00] for two weeks and then they were brought back to the fracture clinic and about two weeks, and they were assessed for the adequacy of the position and they defined adequacy of position as less than 30 degrees of varus valgus, less than in the coronal plane, less than 20 degrees of AP angulation in the lateral and less than 30 millimeters of shortening.
So if you passed all of those tests, you're put into a functional brace. They're assessed at four weeks, eight weeks, three months and six months, or until they were radiologically united.
The analysis they've done, they've described the stats that they've employed, which looked like, you know, robust stats, no big alarm bells there. One of the downsides I saw when I initially read the paper was there was no power calculation with this paper. So it's just large numbers. But in fact, they retrospectively did a post hoc power analysis and said that the numbers were sufficient to detect a small to [00:18:00] moderate effect size, which I thought lended credence to the results.
In terms of just before we go into the analysis, one of the limitations of the paper is that they initially actually found about 1500 patients that were suitable for this study, but lost 290 of them to follow up basically to either non-attendance or migration out of region. And those patients weren't followed up. So that's about 20% actually. So it's really a conclusion based on an 80% follow up.
And so they followed these patients up and they said, okay, well, what were the reasons for intervention? And they basically defined, they've identified four reasons why surgery would become advised. And that was clinical, which is movement of pain or radiological signs of a nonunion, the pain or noncompliance with the brace because of rubbing or inability to wear it for some other reason, that the fracture reduction had been lost and wasn't regained on reapplication of the brace, or a persistent radial nerve palsy, [00:19:00] which required exploration. And they found that union was achieved at an average of about 15 weeks for these patients.
So in terms of those four reasons, the big, the main one for intervening was non-union which of their 344 patients, 29% of the patients that went on to get surgery, 60% of that, or 17% of your overall population were judged to have a nonunion and had a fixation because of that. 24% of the surgical population, so 7% of the whole group had a malalignment unsuitable for continuing conservative management. And 12% of the revisions of the surgical candidates or 3.5% again, of the overall 1200 patients were unable to tolerate the brace and a very small number were explored because of persistent radial nerve palsy. So they were the sort of interventions.
[00:20:00] So putting out to you guys just to start there, they seem like reasonable reasons for intervening. Any other things that you see in a clinical practice?
Yeah, I'm, I'm going to be very pedantic about post hoc power calculations. Cause it's a really important thing. And people kind of misunderstand it because we've ended up with, mal-statistics practice. If that such thing exist. Mr M would be upset about the statistics litigation and it going up because this is one of those things that we do and we don't really think about, and we shouldn't do because the point of power calculation is to work out how many people you need to work out if you've got a type two error, i.e. not finding a difference when actually there's one there. So if you found a difference and then you do a post hoc power calculation to say there's enough patients to not find the difference it adds nothing to the study, because you've taken an effect size you've already seen.
So you think basically it's a north confirming exercise?
Yeah it's a south confirming [00:21:00] thing and people will, you know, it is published and you can kind of do it, and if you read the statistics literature which I'm really sad to say that on occasions, when nobody is looking, I do, you'll find that the world is split slightly on post-hoc calculations of any of any variety, you know, but post-hoc power to me doesn't add anything to a retrospective study where you've basically collected the numbers that you have and you found the difference. It doesn't say anything. It just makes people think that you've conducted it with a higher level of rigor than you actually have, because there is a difference so there can never be no difference. So the post hoc power doesn't add anything at all.
It's one of those things that we kind of read and think, Oh yeah, you know they did it right. And actually...
No, well I thought that the purpose of that was that it essentially prevented you just overpowering studies saying, well, I'm going to collect data on a huge, huge number of patients today.
Well youve already collected it. When you do your post hoc, you've already collected it. So, when you do a sample size [00:22:00] calculation, which is what it actually is, what you do is you look at the look at what you'd expect to see in terms of difference, you look at the populations and you say, well, how many patients would I need to see to avoid a type two error? You know, there being a difference in not finding one. But this paper can't have a type two error because they found a difference. So they can't have a type two error. And again, it's one those things that we do just kind of slightly misunderstand. And so we think it adds rigour and believability to the paper, whereas to my mind it actually does something slightly different to that. It suggests that they haven't really sought statistical advice and they've done something that is kind of valid, but it doesn't, it doesn't add what they think it does to the paper.
So that would be my two minutes on post-hoc power calculations. I'm sure I'll get lots of angry emails from cleverer people than me, like Professor Costa saying there is some reason, but in fact, I'm pretty certain that Matt would agree with me on this, there's no valid reason for doing a post hoc sample size calculation, which is essentially what they've done, because you've already found your sample. So that was my first thing.
My second thing was, [00:23:00] you know, I kind of want to believe this paper cause it sort of reflects what I see in practice, and there's nothing better than self-affirming your own preexisting beliefs by reading in a paper and saying, yeah, I'm absolutely right. Which is, you know, we all know the fall out rate was low, we know that he convinced patients to, to stay in their splints. But we do know that if you persevere with it, splintings a reasonable thing for most patients.
But the thing that concerns me about this is, you know, nine level one centres, so 118 fractures per year, and you kind of alluded to it. And that means that you've got one per month per centre. And that actually suddenly doesn't sound so believable when you say one patient in, because you know, they would have had a lot more than one patient that was suitable for bracing. So, what have they done? Have they not reported them? So that just adds a little bit of, you know...
Well, that's the thing I think that's an interesting design thing and it's certainly something that I wondered because you know that, um, I'm not sure that the level one trauma centres in America are necessarily the place to [00:24:00] pick up all of your community fractures. So I think it doesn't diminish the ones that they have collected, but it does, as you say, by the time you do the maths, how many were getting and how often over what period of time it just starts to make you think, okay, well maybe *inaudible*.
Nonunion is more common than we traditionally taught that it is. And we know that because we all do a few. I think that the really interesting paper, and it's a couple of years old now, I had to get the reference up while you were talking, is the paper from Ken Egles group.
Is that the New York paper?
That's the New York paper, which is 2017, which is entitled fracture site mobility six weeks after the English exactly on union. And that is a real practice changing thing for me. So you put the two together, you go there's an easy way to see whether it's going to heal or not, waggle about about six weeks to see whether it moves and moving it, you know, the nonunion is more common than we expect. And suddenly that [00:25:00] changes my practice that says to me, okay, six weeks, if it doesn't seem to be progressing, bailout and you probably end up at about the right number then.
Yeah, the next thing I was going to go into with this paper is that they have, and I think in a fairly reasonable way, they've tried to say, well, can we predict who's going to go on to nonunions or failure bracings et cetera?
And the, the groups that come out consistently in terms of patient demographics, they identified were white patients, female patients, and interestingly employed patients. Now in a North American system, whether confers any information about followup or likelihood to intervene would be an interesting topic I think. Another one when they looked at nonunion specifically, alcoholics had an increased risk of going on to, to non-union as did, high energy injuries, you know, basically cars versus pedestrians. You're more likely to go on to a nonunion than if you had a low [00:26:00] energy mechanism of injury,
No, I agree. And there's also things about brace compliance and just, just to make Brett feel involved. Cause obviously, you know, spinal bracing is one of those things that one reads about rather than does to one's patients, I think.
Yeah, absolutely.
And the thing is it appears to me that if you actually talk to the patient about their brace tell them how to use it, how to tighten it, how to do that kind of thing, what the purpose is and persuade them to wear it and engage with the treatment its probably quite successful. Similar to scoliosis bracing.
Yeah, absolutely. A brace on the side doesn't do anything at all. You know, they have to wear it and engage with it and so on. So yeah.
I think that's one of the things about the sort of the unsexy side of humeral fracture management is, is actually the time invested in clinic and also the support for patients in between fracture clinics in terms of being able to come in, have a look at their brace, it's rubbing there, you know, because again, one brace does not fit all patients. And they do need a bit of customization.
You know what Sarah, if I was at [00:27:00] six weeks with the mobile fracture site, you know, and I tried this brace and it was a pain I couldn't wash properly and all that kind of stuff and it didn't appear to be progressing towards healing, I'd get it fixed.
You see patients come in, you know, they're doing okay, they're back at work, things are progressing, you know, avoid the radial nerve complication.
Well, I think that's the thing. So *inaudible* work really said that essentially you can brace all of these, they'll all heal, but equally union at nine months was considered a victory, whereas I think for a large proportion of our patient population and I think all three of us union at nine months would not feel like a victory. It would be a significant life changing event to have a wobbly arm for nine months.
And I think, I think for me, this paper brings up a really interesting sort of philosophical thing, which is that any paper that promotes or any, if you, if you pursue a conservative [00:28:00] management of an injury, all of the risk is assumed by the patient. You know, it might not go on to heal. It might not go on to heal in a good position, et cetera. It might have functional implications. It might have implications in delay for your job, et cetera, all these things, but they're all assumed by the patient. Whereas any time that we do an operative intervention, we start to assume certain risks. We feel liable for infection or nerve injury, non-union et cetera. And I think therefore any paper that is trying to move us from where a practice for an injury is largely conservatively managed to saying let's have an increasingly interventional, input to these, to these fractures is always going to come up against that barrier of this who is assuming the risk for this going wrong?
So based on that, you talked about the paper from New York, which same for me, that's the one I always quote and I always give it a wobble at [00:29:00] six weeks, but should this give us increased confidence in intervening earlier for some injuries?
So I think we probably already are. I think it's just a bit emperor's new clothes. So I do, I think, I mean, you know, Nottingham is a very conservative unit. You've worked with us, but actually if a month and a half later, the patient has no progress towards healing, I give them an operation. But I think its one of those things, we still teach our residents everybody gets a brace, but actually we probably already know the answer. We're just not doing it. We're quite good at picking the ones early that needs surgery. And they tell you they don't tolerate a brace, it's really sore and painful. They come back to clinic and basically, you know, if you don't give them an operation, then they bring their husband or wife to berate you into doing an operation. And that's normally the right thing to do.
What we're bad at is recognizing that we waste patients time for eight weeks when we know it really isn't going to heal. You know, [00:30:00] we know we see the patient, there's no progress, it's all a bit wobbly and you say keep pumping your bicep well see you soon, make sure your elbow doesn't get stiff so when I have to do a non union surgery at six months, like. I think we already know it. We should just do it, you know?
Well, so yeah, so I think exactly, I think this paper can actually give us a bit of increasing, you know, Gusto to say actually, you know, we know some of these don't do well and it's not all of them because actually 70% of them healed absolutely fine. There's a significant portion here and certainly doing something that hasn't united, trying to plate it at six weeks is very different from trying to place it at six months in terms of the technical demand of that procedure, the risks involved in it and all the rest of it. Professor Ollivere is going to go on and say, yeah, we know this know, we should be embracing this.
Absolutely embracing, embracing. Give it a try, doesnt work? Do something else.
Excellent.
So I wanted to discuss the IMPACT Scott [00:31:00] paper, just to come from my dear friend, Andrew Duckworth and colleagues in Edinburgh. So that's part of the reason that I feel fuzzy and warm towards this paper. And I know that the whole of Scotland is a happy place so you probably feel fuzzy and warm towards this paper as well.
Absolutely it's a family up here.
But COVID has changed all of our lives and those of us interested in research, you know, you kind of, you're not entirely sure what to do. And the reason I picked this out is actually, because I think it's really well done within the constraints of what they had available to them.
So, you know, the background, everybody knows like COVID, you know, it's a bad thing, affects frail people, high mortality. But the things that people might not know having read the IMPACT paper, which is basically about the effects of Neck of femur or COVID on Neck of femur patients in Scotland is that it came out of the Scottish hip fracture audit, stopping. Cause the Scottish Hip Fracture audit stops and a group in Edinburgh think well, what can we do about this? It's really important for us to know what's going on [00:32:00] in COVID. So that's the first thing is that for me, that's a really productive, like we were trying to rearrange our service. We're doing all sorts of other bits and pieces, and we're actually going to look after our patients, we're going to audit like we should.
So they get on the phone to their various friends around Scotland and they identify nine units that are able to recruit patients to a sort of rapid sprint audit. And they do, you know, 23 days pre lock down and 23 days post lock down. And it's a little bit of a messy paper. You know, it's a paper that, that does have multiple end points, but it's cleaning up and well thought through because how can you think through something that you do on, you know, on the hops?
There's lots of secondary outcomes. And I think people feel a little bit funny about that, but actually it's a really good paper. They managed to get the 317 patients in very rapidly. They demonstrated that the patients had normal demographics and then they dichotemised the series into pre and post lockdown basically.
They looked at what were the [00:33:00] mortality predictors, and they found the usual suspects you'd expect. They had a nice multi-variable analysis. So they found the mortality rate being older or having male sex. Cause, you know, men are programmed to die when they have an injury. We know about that at all levels. Coming from a care home increases multiple hip fractures. Another reason I like this paper, they use our outcomes Score. And whether or not the patient's COVID positive. And they were able to quantify that being COVID positive gives you a relative risk of 5.5%, 5.5 times of death.
Now, obviously the death rate is relatively low anyway, so it's not, you know, it's not every patient dies and you're clearly not going to leave a hip fracture patient on the ward without surgery. That's not what I'm suggesting. It just informs, you know, and I think informing and knowing what the risks are, is super important. So it increases their risk of death.
They then went on and they said, okay, well we've collected all this cohort together. That was basically, I think the primary aim of the paper was to look at the impact of COVID. They then went on and [00:34:00] did a bunch of other slightly useful things, some more useful than others based around the data they've collected.
And they went on to look at whether or not they were able to predict the chances of having a positive COVID test. Cause all their patients have COVID tests - some of them had symptoms, some didn't. And they sort of unlocked, you know, again, in a sort of a fishing trip really, they established that having a high paid account was predictive of COVID.
We know there's some good reasons why that might be actually, because you do become *inaudible* with COVID for various reasons. I only did rock analysis, which is a way of looking at continuous variable with a known outcome and working out whether or not its a good or bad predictor and found that it was kind of a mild to moderate predictor, but definitely a predictor that was there.
They then went on again to look at what was going on with the patient populations and found there's no difference in the patient population, which I think again is to me the second big message of this paper, which is, you know, a lot of us, who've done a lot of hip fracture research and I've done a lot of hip fracture [00:35:00] research now recognize that the hip fracture is really the population is the same in, in normal time.
Whichever hospital you look at, whether you look at Toronto where Brett is, whether you look in Nottingham or in Glasgow, Edinburgh, you know, anywhere in the world where you've got a westernized society where people live to a moderate age and have frailty fractures, there's a very, very classic distribution of comorbidities of mortality associated with hip fracture of dependency of all those things. It's a very static population. It's a really good weather vein of how well a system's coping.
And interestingly, they found that the pre- and post-lockdown populations were the same. And one of the things I think we all really worried about was, you know, what is happening to our patients and our younger patients? I saw patients who presented, you know, four months or three months after their pilon fracture because they didn't want to present and had to do some terrible things to people's legs that you wouldn't normally have to do. I havent done [00:36:00] *inaudible* osteotomies in westernized populations before. I get a fair trickle coming in for injuries elsewhere. But you know, but with the hip fractures, you know, we did see the same thing in Nottingham. We saw the hip fractures coming in and we saw a little increase in them as time went on, as the elderly became a little bit more isolated, but it's reassuring to find that they were the same population. They were still being presented. People were not neglecting them, delaying them. They weren't dying at home. And so that was my second kind of positive take home from this.
And the third positive take home really is that just really impressed they got it out. You know, really difficult thing to do in such a short time. And it's still the best paper in terms of a consistent population, but they've audited all of them. So they're able to make a proper estimate of risk of mortality. If you look at the, you know, the infamous COVID surge paper, it's terrible. And it was in The Lancet. Absolutely terrible. And the orthopedic world has done better than that. And I think that's [00:37:00] great.
I think as you say, I think there were several, you know, really good things about this paper in terms of the topic. And one of them as you say is trying to genuinely quantify the risk. You know, if you're a Neck of femur patient, your family member is a Neck of femur patient, during COVID and then they get COVID what is their increased risk, you know, should you be more worried? And I think that was a very Valiant thing, a question to address and try and answer. I thought, again it was reassuring that the numbers weren't significantly different, you know, despite the sort of social isolation of these elderly patients, that's probably more of a social care... what's the word ....topic or advent, but I thought that was, that was interesting.
I thought the way that they identified in their description of how they identified their COVID patients. So in this paper, there were of these, of these 317 patients. There were what, 20, 26, 27 patients who were COVID [00:38:00] positive. And in fact, the minority of those, 6 were COVID positive at admission. And then of that, there was then a sub further 11 who were positive within the first two weeks of admission and a further nine who were positive 14 days after admission. And they touch on that in the discussion. Well first of all, in the analysis, which is that if you test positive for COVID at any time, not just at admission, you've got an increased mortality risk. And then I thought they very sensibly discussed, and I think this is one of the key things of the paper, is if we see a winter, you know, um, a secondary spike in COVID numbers, how are we going to look after these patients as inpatients? Because they are obviously the most vulnerable in terms of COVID and COVID has a negative effect on their mortality. So are we going to care for these patients in six bed wards? Are we going to have cohorted space available for these patients? I think these are really important things that we need to look at as a community and prepare [00:39:00] ourselves for in the next few months.
You were very positive about the design of the paper and things. I had a question about it, which is why 23 days? As you said, they did a very good job of getting good data collected on this population, but why 23 days before and after? And why, what was my other question? You know, the 317, can we do reliable, multiregression analysis on 317 patients, as you said, where there are big differences in terms of there were univariant differences with ASA, age, sex...
Yeah. So there's a couple of things there. So I don't know why they did 23 days. You'd have to have to ask Ducks that I guess. My suspicion is they did 23 days because you know, it's three weeks and a weekend. I suspect that's the reason. Actually, but I mean, I don't know. It may have been that they were collecting data prior to lockdown in the hope of writing a different [00:40:00] paper and then lockdown comes along and they think, well, we'll balance things out and have the same thing.
And you know, that there are some more IMPACT papers coming. We've contributed data to them, and I'm sure you guys have as well, which are, you know, better longitudinal analyses of larger numbers of patients, which, you know, I think they just wanted to get this out.
Answering your question about the Cox regression model. So the correct size of the multivariable of which they've done a number. So the Cox regression, you don't need to worry so much about the, about the number of covariates. Because it's continuous data. So you've got continuous data, you know, when the event happened, i.e. death and you also know what the covariate is that you're looking at.
And so you don't need to worry so much about the Cox models. You definitely do need to worry about the multivariable and regression models in terms of numbers of participants in covariate. So what they've done and I'm just screening through the paper because they've got these [00:41:00] huge tables of the things that they looked at and screened. And what they've done is they've looked for potential statistical links by selecting covariables with a potential reasonable P-value and excluding the others, which is an okay way of doing it. So you
look at everything, you say, okay, what we'll do, because we've got potential interactive variables here. Anything that has a p-value of say less than 0.1 or 0.5. Obviously it's not, it's not a per person 0.1 or less than 0.5. Wouldn't be traditionally considered to be highly statistically significant. Obviously the cutoff is formally 0.05. So we'll put all of those things in. And then what happens is you're testing less variables for the number.
Theres a general rule of thumb, you can have around about 30 participants per covariate that you test. So actually, when you look at most of their analyses theyre alright. They're not massively powered, but theyre alright. And the reason [00:42:00] for that is what you're doing with a multiple regression model is you're essentially describing the population. If you have too many, too many variables, you describe that exact population and then it's not generalizable.
So that's the sort of the logic behind it, but there are no hard and fast rules about how many you use. Again, it's a common thing. It's like the post-hoc power calculation. You commonly see, you know, we had 50 patients and we did a multivariable of these 75 variables and therefore we found the ones that were significant. So yeah, I think they've done the best that they can.
I think that was the point exactly, which is that these things, you know, when they are done and they are done in real time about time critical topics, they cannot be designed with the scientific rigour of having complete free reign on what data you collect and when et cetera, and what patients are presented and admitted, et cetera. So I think that [00:43:00] that was a good explanation there actually of the robustness of the stats. So that was, that was a useful insight. Thanks, Ben.
In terms of future research where we're going from here, as you say, there'll be more studies out about this. I would really like them to explore, you know, the findings in relation to the platelets. You mentioned this platelets of lower than, well, they mentioned 217 of the, you know, 10 to the power of nine per liter as a risk factor. And that certainly seems, in a world where we do not have a perfect COVID test, these sorts of associations would seem very relevant to the entire medical community, not just orthopedics. So I hope that there's more data that comes out in relation to that as well.
After that statistical masterclass from Ben... well, no, why we keep you around Ben, you know, I mean, otherwise we'd have, you know all cut you off years ago. I'm sure.
Absolutely I'm stunned into silence here.
It's not my witty banter. I worked [00:44:00] that out years ago.
But guys, thank you very much. That was very interesting for me. I hope the listeners have enjoyed it. We have a suspicion that this evening's podcast may run slightly longer than the anticipated usual length time, but we hope that you've all sort of stayed engaged and enjoyed that. So guys, thank you very much for joining us. That draws us, I think, to the end of the podcast. So thank you very much there to professor Ben Oliverre, Mr. Brett Roco for joining us. And thank you very much to you listeners for joining us. And I look forward to welcoming you to the next podcast.