What is the incidence of late detection of developmental dysplasia of the hip in England?

Listen to Mr Andrew Duckworth interviewing Mr Alexander Aarvold and Mr Daniel Perry about their paper "What is the incidence of late detection of developmental dysplasia of the hip in England? A 26-year national study of children diagnosed after the age of one", published in the March 2019 issue of The Bone and Joint Journal.

Click here to read the article

Transcript

[00:00:00] Welcome everyone to this month's BJJ Podcast. I am Andrew Duckworth and a warm welcome to now our fourth podcast from your team here at The Bone & Joint Journal. First I'd like to thank all of our readers and listeners for the feedback we have received so far regarding our podcast series to date as well as to the authors who've already agreed to take part in our series over the past few months and in the upcoming months. 

We are aiming applying to discuss a range of topics throughout the year where there's already having covered trauma with open fractures, arthoplasty looking at robotic knees and most recently tumour looking at the role of Denosumab in giant cell tumours and moving on to paediatrics this month, which we'll come on to in just a moment.

We do have the podcasts are improving the accessibility and visibility of the studies we publish for both you as our readers, as well as for many authors out there. As you know, we hope that during the next 20 minutes or so, we'll cover a range of aspects of the chosen study, emphasizing the important points of how the work was designed, the methodology used, as well as the key findings of the study and how these potentially fit into your day-to-day clinical [00:01:00] practices. We also hope that you get a behind the scenes insight into how the study was designed and how the authors developed the study and give them an opportunity to put forward their key findings of their work.

So today I had the pleasure of being joined by two authors for our upcoming paper in the March edition of the journal entitled What is the incidence of late detection of developmental dysplasia of the hip in England, a 26 year national study of children diagnosed after the age of one. 

So firstly, I'd like to introduce the senior author to the paper, Alex Aarvold, who's an associate professor and consultant paediatric orthopaedic surgeon at Southampton Children's hospital. Alex, thank you very much for joining us. 

Hi.

I also have the pleasure as well of being joined by your coauthor on the paper, Dan Perry, who is also our subspecialty editor here for children's orthopaedics at the BJJ. Dan again, many thanks for taking the time out of your day to join us.

You're welcome. Hi there. 

So Dan and Alex, moving on to your paper, obviously the aim of your study was to establish the instance of DDH diagnosed after one year of age in England, looking at how this was affected by age, gender, as [00:02:00] well as the region and year of diagnosis. And as you very nicely stated in your, in your opening part of your paper, DDH is a very substantial public health risk. And, it's the most common,  largest cause for total hip arthroplasty in young adults with the early diagnosis detection, absolutely essential, as it gives better outcomes for the patient. 

So Alex, if we start with you, can you give us some detail on the background of the paper, including the argument you discussed very nicely about universal versus selective ultrasound screening?

Well the universal vs selective screening debate has been going on for many decades. It was introduced back in 1969 in the UK as a selective screening programme, which has been the UK version, as opposed to some of the European countries, Austria, Switzerland, Germany, which have a universal ultrasound  screening programme. And the debates and the pros and cons of each has been raging.

And there've been many papers in the BJJ over those [00:03:00] decades and papers in the Launcet and editorials in the BJJ all about this debate and it's gone round and round. So this was really trying to add to the basic data of just how often these late detected cases occur. 

So giving us some baseline data was the real aim of it more than anything to move forward? Okay. And Dan, where would you say the current state of play is in the UK? What our current sort of... where is the argument at the moment? 

The argument, as Alex says, has been going around in circle for years, and we're not.... surgeons have got a belief that we need to be doing something better, because we know that there's about 500 kids or so every year that present late with hip dysplasia and we also know that if we treat it early it's possible to be treated with really simple  measures,be it a Pavlik harness or be it some other sort of [00:04:00] harness or perhaps simply a closed reduction. But the later we see it that obviously the more invasive, the more difficult things become. So we know we need to do something. It's just about getting it right early. 

Okay. And in terms of, you know, I mean currently, if you could you just run down, what's our current sort of what's the general screening protocol we have currently in the UK? 

Sure. So, the current screen protocol is a selective ultrasound screening programme. And that means that before every child is discharged from hospital, they're required to have an examination. And that examination is what we all know to be *inaudible* along with looking at limb lengths and anyone who's got an abnormal examination or anyone that's born breach or anyone with a significant first degree family history. So that's mom, dad, brother, sister,  with a known history of DDH treated with a harness or otherwise, needs to [00:05:00] have ultrasound screening in a formalized way. So that's our current state of play. So the question is whether we need to, we need to do something better than that. 

And Alex, as you mentioned, your paper, is it right to say that the incidence of lady DDH before your study is presumed to be higher than in the rest of Europe or not?

The incidence in the countries that have universal ultrasound screening is very low, late detection rates. 

Okay. And that is from recent  literature is it as well?

That is and there's three different papers referenced in our paper with the incidence rates there. 

Okay, fine. So if we move on to how your study was designed, it's obviously a descriptive observational study. You've used data from two large national UK databases. The first being the clinical practice research data link. And then the HES data or the hospital episodes statistics. So obviously having looked at this data, can you sort of, for the readers comment on the accuracy and the robustness of those datasets? How complete was it? And did it sort of change over [00:06:00] that? You know, there's a large time period you've used, did the data become better should we say in inverted commas commrce as the time went on? 

So going through the  questions the CPRD is representative of the national UK population. It is repeatedly validated, and it is considered to be the benchmark database for epidemiological research. So in that case, it is the most robust database to be using. However, it is intirely reliant on the codes they put in at the start. And that also applies for the HES, the hospital episode statistics, which may not be quite as robust as CPRD then therefore for this study we used CPRD for the baseline diagnostic code. And HES was used simply as a supportive code. So just increased validation and help us be more reassured that that CPRD code was true. 

And in terms of the information in the actual database, is it just the diagnosis or is there other bits of information, so such a risk factors and things like that?

So, [00:07:00] no, it doesn't record risk factors reliably, so we couldn't get that information. That would have been lovely. 

Yeah. 

The HES database has more of the operation records, which would be part and parcel of the late detected hip dysplasia. And your question about, did the data change along the way, well it has variably incorporated slightly  different percentages of the population. Naturally there were about 8% of the UK population involved in CPRD. Currently it's about 4% and right back at the start, it was very small in the first couple of years. And on one of the figures, you'll see a figure four, I think it is where the first couple of years there are very wide error bars, which rapidly come down. Yeah and that's the incidents over the years, which we believe is unchanged. It's just related to the smaller population in the infancy of CPRB. 

Okay. And so looking [00:08:00] at your inclusion, exclusion criteria could just for the listeners as a brief overview of how are patients included? Obviously you've gone from 15 million records down to just over 700. How was that sort of process done? 

So 15 million is the total number of records that are in the CPRD. And then you want to look at the population that might have the late detected hips, which is up to, we put a cut off of eight years of age, thinking that after eight years of age it would be unlikely to do any operative intervention of social reductions. And that brings you down to the sort of at risk population of children. And then within that, you're looking at how many kids have these codes. 

Okay. 

Or DDH. And the most important thing is that  anyone with a code of DDH before the age of one was excluded because that's the vast majority obviously. Treated in  public, picked up in screening so none of those are in that. There's any code existing [00:09:00] before the age of one. They're not in. So it is only new codes after one year of age. 

Okay. And did you get any feel for... did you have the data for under one at all? Did you look at that at all? And I know not with this study, but as a 

No, we didn't no. We could send it through, but that's a vast amount of data, selected over one. It would have been lovely to be able to look at greater than six months today, because that also would be a detected cohort, probably looking at the surgery. You miss the chance of *inaudible*. But the CPRD only does annual increments. 

Okay. Okay. 

We can only do annual cohort, and that's the reason it's late detected in our definition now as over one. 

And, obviously you excluded neuromuscular disorders. What was the reasoning behind that? 

Well, so a kid with cerebral palsy or neuromuscular will have a hip migration where the hip can come out. We didn't want any confusion or [00:10:00] overestimation of the number. 

Yeah.

Neuromuscular hip is very different from a DDH. So Charlotte, who's the first author, went with a fine tooth comb through all of the codes of every single patient, not 15 million, but every stage that was potential late detected DDH and anyone that had a hint of a neuromuscular diagnosis was excluded. We really don't want to be overestimating the number. 

No, just get it as accurate as possible. Yeah, sure. And obviously we're just moving on to how you presented the data. You said that you've looked at the study-specific incidence rates sort of age, gender, year and region, calculated using specifically calculates at risk populations. What sort of do you mean by that? A brief explanation. 

Yeah. So within each year, we'd look at how many one-year-olds had a late detected diagnosis or diagnosis at one year of age. And then you look for that year how many one-year-olds there were in CPRD. Your incidence of [00:11:00] one-year-olds in that year. And we do that every year or for 26 years that the CPRD was running. And then you repeat that again for every kid that was detected between two and three years of age and every kid between three and four years of age. And you repeat that for all 26 years. You build up the overall incidence

Yeah. 

You get a cumulative incidence and that is in the final  column of the table. 

Okay, great. Yeah. 

Within CPRD, they also get the break down of all the ages, obviously, but also the region. So that's saying it could be broken down, not just nationally, but region by region. 

Region by region. Okay. So, that sort of moves on to the results nicely. So you had 754 patients that were identified as having a delayed presentation of DDH or detected DDH from one to eight years. Over eight to ten were female, four to one ratio as you sort of expect. And that gave you an overall cumulative incidence of late diagnostic DDH of 1.28 per [00:12:00] thousand live births. 

 So just for a bit more detail for the key results. How did the incidence of DDH vary with age and what you found when you looked at the year and regional variations as well? 

 So regional variation was much the same across the country, so let's say a national national issue if you like. And then age as expected, the vast majority of the detection is over one year of age were between one and two years of age. And it's very rare that you'd get a seven or eight year old coming in with a late detected dislocation. 

And in terms of that, the figure four, which you've mentioned already, which is, I think it's a nice figure looking at, you know, how, you know, there's a sudden drop isn't there at one stage, you know, there's sort of the first couple years have wide confidence intervals it feels, like you say, and then it sort of, you know, 93, 94, there's a sudden drop off in the incidents. And do we have any feel about why that was or...?

 That we feel is just because that was the infancy of CPRD. So there were very [00:13:00] few patients in the first couple of years and it rapidly built up and gained traction. And you can see for the whole rest of the time, the confidence intervals get narrower and narrower. And we think it is actually an unchanged incidents across the whole time. 

Yeah. It's fairly consistent. Isn't it? 

Dan. I was wondering if you could say was there much of a change in terms of the screening programme we had over that 26 year period roughly? 

Not really. So there's been a standing committee that this existed for, ultrasound screening for a long time, since the 1960s. And there has been small changes throughout the period. Obviously there's some areas, in particularly Coventry, did universal screening for a while, but as a whole, as a country, there's been no significant change in screening over that period.

Excellent. So then nothing would really explain that sort of change other than that. Yeah. Okay. 

I could argue with Alex that I think the change that's there is very much due to the uncertainty of CPRD in the early years, because we've seen that in other CPOD studies. 

Yes. Okay. Yeah. So just [00:14:00] the tightness of the data as it goes on with time. Yeah. Okay, great. 

So just moving on. So obviously the strengths of the study are, you know, without question, great use of large big data and the number of patients included and, you know, it really has given, I would say, a detailed upstate presentation of the instance of late presenting DDH now in the literature, which, you know, can not only be compared with previous work as we can go on to discuss, but also I suppose future instance rates as you move forward. And without a doubt it clearly provides important information for the debate that we've already discussed about DDH screening in the UK, but sort of putting it into context what do you feel the primary limitations of the study and what other, if you could not change anything, but what other data would you have liked in there to potentially improve it if you could?

The age detection, if it was a monthly or six month intervals, then capturing any kids detected between six and 12 months of age would also technically be a [00:15:00] late detected dislocation with likely surgical treatment so that is an important group that is not captured in this study. The risk factors you mentioned before. That'd be great to know what the risk factors were. All these kids with late detection dislocations, we think that probably they are late detected because they didn't have risk factors and therefore didn't have ultrasound screening as well as the clinical exam. 

Okay. 

So I think the biggest problem with any CPRD or any big data studies is always the codes. And we, you know, we tried our hardest, or rather Charlotte tried her hardest, to really clean that data as much as possible. But we only know what's in there from what the GPs recorded or what's made its way into the hospital coding systems. So it's very easy for cases to perhaps not quite reach that dataset in the right way. And that's always a problem with big data, but the advantage of big data is that it's so big that we can ignore some of the [00:16:00] minor, you know, the minor miscodings, because we've got such a big data set. 

Absolutely. It's just that the pure size should control for it almost  yeah.

When you're looking at incidents though, you have to assume that this is a minimal incidents, because this is just the ones we're seeing. But comparing over a time period or comparing within regions, we can confidently say that we don't expect coding to be significantly different over the period or between regions and that therefore those comparisons are fair to make.

Absolutely. No, I agree. And in terms of sort of moving on nicely. So you comment in the paper about how the incidence you've reported is certainly higher than other studies, which were obviously published several decades ago. Do you think there's a potential reason for that? Is it to do with the inclusion criteria or just we're picking it up better?

Probably picking up better. And those studies, are quite big studies, but they were done in specific regions and specifically Bristol [00:17:00] areas and specifically Southampton area. So there's population migration in and out whereas this has captured the whole of the UK. So this is probably far more accurate.

Yeah. And in terms of now, so I'll ask both of you, Alex if you want to go first. So where do we go now? We're using this data and the way the current  debate is what's our next step forward in terms of screening for DDH do you think? 

So we know that there are this pretty high incidence of kids who are not picked up in the screening programme. So how can we optimize that? And I think probably the one thing that is not at all contentious is that there should be maximum education of the people doing the screening programme and be it the perinatal screening within 72 hours of birth, be that the midwives or the neonatologists, paediatricians, albeit those at the six week check, which is maybe the GPs or the health visitors, and it seems that education on this is now [00:18:00] not a part of the GP training, which is a fairly recent change. So that, I think, is a non-contentious thing and it should be improved. Other things that could be looked into for more information is if you're thinking of widening an ultrasound screening programme, and you are probably gonna have some pretty tough questions about cost analysis. And that's certainly work that can be done on the back of this, of as simply as looking at the cost of surgically treating this number of kids versus the cost of ultrasound screening everybody and whether it's practical and over-treatment implications. So then they're going around again with the ever circular debate. 

Yeah. And, Dan what would you add to that? 

So, Im Mr Evidence and so, as I see it, from before we started screening, the rate of late presenting DDH was about one in a thousand and now often we've started screening and throughout the whole screening period, the rate of [00:19:00] DDH is about one in a thousand for late presenting.

So I'm not sure that the screen programme we've got is doing much. The national screening committee on which I sit says that the current screening misses two thirds of true cases of DDH and they actually say that the screening should be stopped. So the clinical examination should be stopped, but it's impossible to stop because it's so ingrained in clinical practice, it's impossible to stop.

So we're in a kind of crazy situation where we're doing something where we've got no really good evidence for doing it. And there are different models of care. There is a universal ultrasound programme, which is one model of care which might work. Or the other thing is just to say, okay, we won't bother doing anything. And the argument for saying that is, well, you know, it's perhaps not making a difference to the ones we actually need to detect and what it is doing is it means we're over treating DDH by, some people say seven times, some of the studies say 20 times. We'll be treating very mild dysplasia [00:20:00] causing upset to families and to new new mums, which, you know, we can't underestimate.

And I know that's a controversial statement and I know we shot down at BSCOS so we do have a situation where we're not quite sure what's going on. There are some opportunities and one of the opportunities is there's a new system called NIPE SMART. So every time that a child is born now,  all of their details or all of their routine details are entered on a computer system with all the maternity hospitals, including details of the examination. And if we could start linking that NIPE SMART data to our hip finding data, so the ultrasound data, then we could actually have a massive cohort in which we could start embedding trials really efficiently. So we can start saying, what is the best system of screening? And we can do that really efficiently because we've got this big data set, which is going to be routine. And it's kind of virtually there already and the national screening committee have put that forward. So we just [00:21:00] need to, as surgeons, as orthopedic surgeons, we need to just get tight with the national screening committee, because we've got an amazing opportunity to change this and to answer these questions once and for all, rather than this debate going on for another 20, 30 years and back, you know. 

It sounds like that would produce a huge amount of data going forward that you could really, you could really try and answer the question. 

And just to finish off, in terms of, you know, that was very interesting point you made about the cost-effectiveness of the system, you know, whether you go for universal screening or not, and the balance with surgery, do we have any idea from, you know, places on the continent, in Europe? Has anybody looked at that saying, you know, actually universal screening does save money in the longer term or not?

There's not good health economic analyses of ultrasound screening. 

Fine. Very good. Okay. 

And that's in different reviews as well. Sorry. 

No, that's fine. Thank you so much for joining us for our podcast. And that was a really interesting discussion. And I think [00:22:00] for all our listeners and congratulations, both of you on a really, really interesting and excellent study. 

To our listeners we do that you've enjoyed joining us and we encourage you to share your thoughts and comments about our podcast today and our previous ones through Twitter,  Facebook, and a like, and feel free to post and tweet about anything we've discussed here today. And thanks again for joining us.