B (2:59)

Amal thank you. Chase. Bisphosphonates are the most widely used osteoporosis treatment worldwide. They are effective at reducing fracture risk with a risk reduction of up to 70% at the spine, up to 50% at the hip, and up to 25% at non vertebral sites. However, there have been increased concerns about some long term side effects of these medications, namely atypical femoral fractures, which we will refer to as AFF. These fractures were first described in year 2000 and have a specific morphology and diagnostic criteria.

A (3:42)

So we're going to stop here and get David Berger, who is a fellow at Columbia, to ask our first question of GATA this morning. Well, I have a two part question. First of all, what are the diagnostic criteria used for aff? Second, what is the current understanding of why bisphosphonates are prone to cause aff?

C (4:02)

Good morning. The diagnostic criteria for AFF were first generated in 2010 Elizabeth Jane JBMR paper and then revised in 2014. So the current paper is using the revised criteria and to have the diagnosis of aff, one must fulfill four out of five major diagnostic criteria and I'll go over those in a minute. Importantly, the fracture has to be below the lesser trochanter and above the supracondylar flare. If the fracture is located on that side, it needs to happen with minimal or no trauma, a fall from a standing height or less. It usually starts in the lateral cortex and goes interiorly, medially, either in a transverse way or sometimes may become oblique. Important it could be complete through both cortices or incomplete number four. It is minimally comminuted. Basically it's a clean cut sledgehammer fracture and finally you may have some cortical thickening. So four out of these five major criteria fulfill the diagnosis. There are minor criteria that were part of the original definition and deleted, such as the prodromal syndrome, pain in the groin, and if you have a patient on bisphosphonate with such symptoms, you need to be aware and probably do an X ray of the femur or a single energy in the DEXA machine. Others are delayed fracture healing and in addition, some other criteria include incompleteness of ephemeral diastasis fracture. But these are not counted as major. They may be additional. I think the most important thing to recall are the major criteria. 4 out of 5. And if you have a patient presenting with a prodromal pain in the groin, just push the alarm bell, get the X ray done. Now, with regards to current thinking of why bisphosphonate calls aff. Of course, the skeleton is a very dynamic organ with constant modeling, formation and remodeling, resorption. When there is any impact of loading on a bone, they may be micro cracks that are immediately repaired with the formation and resorption. Now, what bisphosphonates do is suppress this remodeling and therefore suppress formation and interfere with this ability to fill in the gaps where the micrografts are happening. These may extend to the point where you get your aff. Of course, bisphosphonates are a major risk factor, but there are patients who get AFF who are not on bisphosphonates. So there is more to the story than that. Potentially genetic factors, potentially other risk factors, femur geometry, bowing of the femora, et cetera.

B (6:38)

Thank you, Radha. Concerns regarding AFF have led to difficulties in decision making among patients and providers regarding the use of bisphosphonates. Such decisions could be informed by data on the absolute benefit versus harm with patterns of bisphosphonate use and about clinical factors that can predict AFF in specific patients. However, such data are not readily available. This study was developed to address some of these questions. It consists of a nationwide population based longitudinal assessment to determine which factors beyond bisphosphonate dose and duration are predictive of aff.

C (7:22)

Right.

A (7:22)

Thank you, Amal and Ghada for giving us an overview. And even though these are incredibly rare occurrences, which we'll see those numbers shortly, the frequency that this comes up in our conversations with patients and our patients worry about this, I think makes it very clear that we need to understand this as well as we possibly can so that we can hopefully reassure our patients on this. So we'll be thinking about that, particularly as we get into the discussion and try to understand how we can apply these results. The first thing that we're going to do in the methodology here is think about the study design that's utilized. And one of the reasons for doing that is this is a very uncommon study design, but one that I really like. So the authors describe what they do as a population based case cohort study. And you can think of this as a hybrid of two things that show up in the name of this study design. That it is a cohort study, but it's also a case control study for those of you all who are into this sort of thing. If you're familiar with a nested case control study, this operates in a very similar fashion. And when you do this, you can take advantage of the strength of both of those study designs, but also avoid the weaknesses that they have. And the reason in a study like this why you would want to do a cohort is the cohort is going to give you important information about how frequently something is occurring. So the incidence and when we're looking at fracture reduction and also this risk of atypical femoral fracture, incidence and understanding that will be very, very key. So that's why you need the cohort component of that. But the weakness that goes along with cohort studies is that they are not very good about understanding rare outcomes. Because if it's a rare outcome in your cohort study, you have to enroll many, many, many people and it often outstrips your ability to find that many patients to be able to adequately power your study. That's the nice thing about case control studies is they're designed for that. So case control studies are designed by building it around the outcome. So you find everybody that you can who had the outcome of interest, and then you go back and try to understand the exposures that might have driven that outcome. And that's why it's helpful to have that component, because we're talking about a rare outcome. And so we need to understand what's driving those risks as much as possible. So it's a hybrid here. And if you're familiar with those two types of studies, the results will will vary as they come up. And we'll be switching back and forth as Amal goes through those results to try to understand where that's coming from. So keep all those things in. Now back to this study in particular, the data sources. So the importantly the cases, that's the main thing that we're going to talk about here, and we'll define those in a second. But the cases come from the Danish National Patient Registry. And as we were analyzing this article, I tasked Gada with trying to figure out exactly what that is. I think some of us may be familiar with it, but that's probably not something that a lot of us are familiar. So Gadot Just give us an overview. What is this registry? How does it work? How can we utilize it in biomedical research?

C (10:09)

Thank you, Chase. First, your review of the nature of the design case cohort studies is very important. And what the Danish National Patient Registry does is it actually provides the database to extract the outcomes of interest in the cases and the controls. This is one of the finest and oldest registries. And Northern European countries are notorious for having very robust registry data linking vital statistics, health outcomes, hospitalization, outpatient medication use, etc. Making it possible to basically reconstruct life and health trajectories for an entire population in a large country. This registry was established in 1977 for inpatient data, but outpatient data was established in 1995. And you can see later that the exposure to the bisphosphonate starts in 1995 and is linked to actually medication intake, prescription databases, civic registers. If people migrate in and out of the country, that is captured, randomized trial population surveys, so on and so forth. So it's a very robust source of data to extract the outcomes of interest. And we'll be getting to those in a minute.

A (11:19)

Few other words on where data comes from, because it's not just from this registry, but there's also migration dates that were pulled from the National Civic Register. The medical history, in particular medication use, comes from the Danish national prescriptions database. And then finally, image files from all Danish public hospitals with orthopedic imaging services were available to these researchers. Gotta mention the study periods already, but if we think about it in terms of exposures and outcomes, the exposure data came from the years 1995 up to 2014, with outcomes focused on the time period of 2010 to 2014. A little bit more on the study population. So we talked about the cases already, and Those come nearly 5,000 individuals, 4,973 individuals who are 50 years of age and older who had a sub trochanteric femoral fracture or a fracture of the femoral shaft. And one of the things I liked about this, just to editorialize on that is sometimes, particularly in case control aspects, studies that utilize that study design, there is a concern for a survival bias. And with that, what you do is when you're finding these patients, depending upon how you build, that is if you were having to talk to patients and get information from them directly, you are going to only be talking to individuals who are still alive. And so because of that, you can miss the sickest individuals and not have a good understanding of your population. So a really nice thing about this is that they are pulling from a database and they can find essentially everybody or nearly everybody that had this event. And so you don't have this issue with survival bias. So I think that's one of the many strengths of this article. As far as the cohort overall, so what these authors did is they did age and gender matching to collect an otherwise random population. And this was nearly 40,000 individuals. So it was 37,021 individuals made up the cohort here. As far as outcomes, another strength of the study is how it's defined is that all plain films from these cases were reviewed by a blinded skeletal radiologist. And you can insert your joke about a blinded radiologist. And this individual did that to determine if the affs, the atypical femoral fractures, if they met criteria for that, and if the reviewer thought that they in fact did meet criteria for aff, they were again reviewed by another skeletal radiologist and an orthopedic surgeon to make sure that everybody agreed that, yes, this met criteria for that. Hip fractures, so not the subtrochanteric fractures, but regular hip fractures were identified based on ICD10 codes. And in terms of the covariates and the confounders that were assessed here, the main one, the one of interest, as you would expect, was duration of bisphosphonate use. The authors grouped this together by duration of use and tried to put them into categories that fit with what we might try to do clinically. So they had individuals who were on a bisphosphonate for less than one year. So more on that in just a second. If you took a bisphosphonate for one to three years, if you did it for three to five years, or if you did it for more than five years. As far as other comorbidities, there was fracture history was incorporated in this. The Charleston Comorbidity Index was utilized, as well as medications. And that'll turn out to be important. One other thing that I would like for our group to discuss is what other information might have been helpful to include here. The authors in their discussion point out a couple that they would have liked to include, but that didn't have access to. So DEXA information would be the obvious one. And then also smoking, that was just not something that was in these databases. So they didn't have access to that I would include in that list socioeconomic status, because that is just something that is a driver for nearly every health outcome. And so I might have liked to see that. But again, probably not something that the authors had Access to. But Amal and God. Anything else that you would like to have seen included, particularly for trying to account for confounders in this group.

C (15:04)

I think for a registry database, they probably had as much information as possible. As we'll see going through these oaths, some of the information they had turned out to be significantly associated with a health outcome with not much biological plausibility, such as hypertension. I think that vitamin D would have been interesting, being a vitamin D passionate person. But vitamin D may be interesting because very low levels of vitamin D may cause really abnormal bone remodeling which could precipitate these factors. Dietary intake of calcium, probably less. And I think they pretty much had all the other risk factors in terms of medication intakes that are deleterious to bones. So I'm pretty satisfied with what we have. What about you, Amal? Would you think of anything else?

B (15:45)

I agree. Vitamin D would have been top on my list. Another thing that would have been interesting is ethnic origins of these patients, particularly that they look at migration records. And we do know there might be differences in the risk of affecting based on ethnic background. So that would have been an interesting piece of information to look at.

C (16:07)

Yeah. And I think that brings a very good point. The other one that comes to mind because of the effect of bone geometry. If these people had X rays done, and obviously they did, to have the diagnosis would have been nice to look at the bowing of their femoral shafts, because bowing is felt to be a risk factor, therefore the Asians being at high risk.

A (16:24)

And we'll wrap up the methodology by thinking about the data analysis. The authors have a lot, as you would expect for an observational study like this, we're going to hit just some of the highlights. So first of all, the authors assess this risk of atypical femoral fracture and they based it on cumulative bisphosphonate exposure. And one of the things that the authors did, the way they did this assessment is that they compared the longer term users with short term users and not never users. And their hope was, is that this would help to avoid confounding by indication. I think this was a great strength of this study. It's often a concern that's raised when we're thinking about atypical femoral fractures is this recognition that, as GAD had mentioned earlier, this can occur in people who have never used bisphosphonates. And some people think that it may just be something that happens when you have severe osteoporosis. And the challenge is, well, if that is something that occurs with people who have severe osteoporosis, then if you ask yourself, well, who is likely to end up on an osteoporosis medication, Particularly take one for long term, it's people with severe osteoporosis. And so the question has come up over and over again, is this really a side effect of bisphosphonates? Or at least as part of it, just explained by the reason that these people were on bisphosphonates is because they had such severe osteoporosis, and that's what caused the atypical femoral fracture. And it wasn't the bisphosphonate that was driving that. The way that the authors designed this study to try to get around that is by comparing to people who were on bisphosphonates for only a short period of time. The idea behind that being is, is that clinicians recognized that these people had osteoporosis, or at least osteopenia that was bad enough to warrant treatment, and at least recommended a bisphosphonate therapy for whatever reason, they didn't end up staying on it very long. But that would be better than just taking a relatively random population of people who had never been on a bisphosphonate, because then you would likely find many, many individuals who had very healthy bones and so were not at a risk for this at baseline. So I think that's another strength of what the authors did here. They estimated number needed to treat and number needed to harm. That was one of their main outcomes that they're going to look at. And Amal will tell us about that here in just a second. And then finally, another thing that I really like that the authors did here is they included a power calculation. Power calculations are typically utilized in clinical trials to figure out how many individuals you need in your study, so that if you do not find a difference between your treated and your untreated group, you can be fairly confident that there is, in fact, no difference between those two groups. In observational studies, you're not usually doing that to try to figure out how many individuals you need in your study. You're usually just including all the subjects that you have access to. But why I really like that is when the authors do that, then if they do not find a difference between their two groups, if they've done a power calculation, then they can reassure you that we had enough subjects in this study that if there was going to be a difference, we would have found it. So it's an optional step in observational studies, but I really like it when authors do it. It's uncommon, but I think that's yet another strength of this paper. So enough on the methodology. We'll turn things back over to Amal to go through the results for us.

B (19:27)

Thank you, Chase. So, as Chase mentioned, this is a case cohort study and Figure one shows the case and control selection process. Administrative data were searched to identify all inpatient and outpatient reports of surgically treated subtrochanteric and femoral shaft fractures. There were a total of 4,973 such fractures. As Chase previously mentioned. Of those, 4,769 cases had radiographs available for review. A blind review was conducted of These images and 189 fractures were classified as atypical femoral fractures or affect. These are our cases now. The remaining 4580 fractures did not meet the criteria for AFF and were excluded for the sake of time. We will not go over the reasons for exclusion, but those are detailed in Figure 1. Now, the comparison cohort or our control was a random sample of all Danish adults 50 years of age or older, and it consisted of 37,021 individuals. Within this cohort, 699 traditional hip fractures were identified by ICD 10 code. So ultimately we had three groups of subjects that were created. Those with AFF, their number is 189. The comparison cohort, that's the 37,021 individuals and within this cohort those with a hip fracture. And the number for this cohort was 691 individuals. Table 1 describes the baseline characteristics of the study subjects, which will later be used in the predictive analyses. I will not spend time discussing Table one for the sake of time. We will move on to Table 2 which describes osteoporosis medication use and the three groups that were identified. 68.8% of the AFF patients had used bisphosphonate, compared to only 8.8% of the comparison cohort and 19.2% of the traditional hip fracture sub cohort. Alendronate was the most used bisphosphonate in all three groups of individuals. 59 of the 189 patients with AFF were not exposed to any bisphosphonates.

A (22:08)

And we're now ready for our second fellow question. So, Camilla Villavencencio from the Mayo Clinic has a question for gada how applicable these results are to a population with a more frequent use of intravenous bisphosphonates.

C (22:22)

Great. I'm not aware of any data that is very reliable for the groups of aff. In terms of proportion, I've seen some numbers, but very small studies. I think the most important number we can anchor our answer to is the hip fractures. And for the hip fractures, 20% were taking bisphosphonate. We do not know whether that was IV. IV was only 1.3%. You know the practice of giving IV bisphosphonate upon experiencing a hip fracture is convenient but very sketchy. Depending on the center you belong to, you may have a protocol to give IV bisphosphonate or not. So I'm not able to give a clear answer for IV bisphosphonate. If I lump all bisphosphonate together, I would say 20% is within the range we see. We know there is a large care gap. The majority of hip fracture patients are not treated. I've seen numbers anywhere from 10% to maybe 30 or 35%. So it falls within, but it's very hard to be more precise than that.

B (23:20)

Thank you, Rada. So next, the researchers analyzed the risk of AFF with respect to bisphosphonate use. And here I would like to direct you to figure two. Among those not exposed to bisphosphonates, the incidence of AFF was very small at 0.2 per 10,000 person year. Among current bisphosphonate users, the incidence of AFF was higher and increased with duration of use. The incidence was less than 0.9 per 10,000 person year. For those with less than 1 year of exposure to bisphosphonates, the incidence increased to 1.4 per 10,000 patient year with 1 to 3 years of use. It then increased to 2 per 10,000 person year for 3 to 5 years of use. And finally, the incidence was 4.9 per 10,000 patient year for those with more than 5 years of bisphosphonate use. Conversely, the incidence of AFF declined after bisphosphonate discontinuation, regardless of how long the patient is patient had been on the drug. We will revisit this concept later when we look at the multivariable analysis. Figure 2 also shows incidence of traditional hip fracture. Here we see a declining trend with increased exposure to bisphosphonates. That's the bottom part of Figure 2. For example, the rate of hip fracture is 102.4 per 10,000 patient years in people with less than one year exposure to bisphosphonates and goes down to 45 per 10,000 patient year in individuals with more than five year exposure to bisphosphonate, but even then the rate of hip fracture remains much, much higher than that of an AFF, which as a reminder is around 4.9 per 10,000 patient year in people with more than 5 years use of bisphosphate. The next set of analyses that the researchers performed was a multivariable risk modeling of AFF and of hip fracture. These analyses were adjusted for age, gender, prior fracture, duration of bisphosphonate use, comorbid conditions, and certain medications such as hormone replacement therapy, glucocorticoids, and proton pump inhibitors. The results of these analyses are shown in Table 3. Regarding the risk of AFF, hypertension predicted a higher risk of AFF, as did the cumulative glucocorticoid exposure and the cumulative PPI exposure. Regarding bisphosphonate exposure, there was a gradual increase in the risk of AFF with longer exposure to bisphosphonates. As we discussed previously, when compared with those exposed to abis phosphonate for zero to one year, there was a 2.6 fold increase in risk for one to three years of use, and this gradually went up to an almost nine fold increase in risk for more than seven years of use. Conversely, the adjusted risk of AFF after bisphosphonate discontinuation gradually decreased with time. Being off bisphosphonates for one to three years decreased the AFF risk by 47%, and being off the bisphosphonate for more than three years decreased the risk of AFF by 77%. Similar analyses were performed, but this time looking at the risk of a hip fracture rather than AFF as an outcome. Not surprisingly, older age was a risk predictor with a gradual increase per decade of life over the age of 60. Hypertension, COPD, and previous fractures were all associated with higher risk of hip fracture, and male gender was protective compared to individuals who never received a bisphosphonate or who received them for less than one year. Those exposed to bisphosphonates for more than one year had a 44 to 55% decrease in the risk of a traditional hip fracture, which is consistent with results we have seen from other trials now. The last component of the data analyses looked at the number needed to treat and the number needed to harm in association with bisphosphonate use. These results can be visualized in Figure 3. With bisphosphonate exposure for three years, one hip fracture will be prevented for every 94 patients treated, and one additional AFF will occur for every 2,970 patients treated as the duration of exposure to bisphosphonate increases. The number needed to treat in order to prevent a hip fracture decreases, which is a good thing, but also the number needed to harm decreases, which is not a good thing. The two graphs were projected to converge at 10 years of treatment with abysphosphonate.

A (28:46)

At this point, we'll move into the discussion and we'll start with thinking about how the authors describe their work and then we'll give you our own thoughts along the way. So where we'll start is where the authors start and I'll quote them as they summarize their findings. They say that atypical femoral fracture risk with bisphosphonate treatment is very low in Danish men and women, even with greater than 5 years of bisphosphonate use. And they go on further to say that that atypical femoral fracture risk declines rapidly when bisphosphonate treatment is discontinued. And they point out, as Amal has told us already, that that five year number needed to treat a 56 and number needed a harm of 1,424 overwhelmingly favor bisphosphonate treatment. And the implication that the authors draw from that, I think is one that we'll often think of ourselves, is that these results should be used to counsel patients and material. So our third fellow question is coming from Mazen Almagrabi, who is at McGill University, and he has a question for Ghada.

C (29:44)

Dr. Ghada, how would you recommend us, as osteoporosis providers, address patient concerns about rare bisphosphonate side effects? This is a very important question. I think patient education and insight and understanding into actually absolute risk increase versus relative risk is crucial. And I believe that I'm finding myself spending more time in the clinic educating patients about side effects of bisphosphonate, notoriously aff, but also onj, and really trying to develop written material going over it, explaining it and handing them that information to take home. Patients we have to remember, are getting mixed messages from multiple sources, media, neighbors, friends, other doctors, some of which is very contradictory and confusing. So I think what we see in the clinic has to be solid documented with something to take home to reflect upon. Now, whenever I want to explain risk, I go over the following equation. I use FRAX in Lebanese patients, but many other countries worldwide do that to look at the absolute risk increase in fractures over 10 years. I put it this way. If the MOF major osteoporotic fracture risk is 20%, I say if I took 100 women of your age, height, weight and risk factor profile and followed them over 10 years. Of these, 80% would not fracture and 20% would. We consider 20% to be high enough. Countries have different criteria. Right. And I won't go into that now. If I were to follow 10,000 women over three years of bisphosphonate use, that risk would be in single digits 5 over 10,000. So the magnitude of the absolute risk increase and the risk reduction are amazing. And this is when I say this is why you need to take the medication. You need to be aware of the side effect. You know it's longer with longer duration. We go over that. But it is almost infinitesimal compared to the benefits you have. If the patient risk of fracture is very low, there is no reason to expose that patient to the minor risk or very low risk of aff. Actually we did that exercise when Bob Adler, myself and several worked on the task force for long term bisphosphonate use in JBMR 2010. And we actually showed a beautiful graph and the data was as follows. We actually estimated benefits risk equation for bisphosphonate intake for five years. So this is a slightly different way of looking at it. For up to five years of bisphosphonate therapy we would approximately prevent 175 hip fracture, 1,470 vertebral fracture and 945 risk fracture per 100,416 aff. So that equation mean we would reduce the incidence of 162 fracture spine, hip or wrist compared to potentially causing one aft. This is almost a 200 to 1 ratio and they need to understand that. They need to take that home.

A (32:52)

Other important findings that the authors point out is one that that AFF risk is similar between women and men. Next is that there is a dose dependent relationship between systemic steroids and atypical femoral fractures. Next, that there is also a connection between PPI use and that AFF risk. And finally that age is only weakly associated with the risk for atypical femoral fracture. An important negative finding, and we alluded to this earlier, is that the authors failed to find a clinical scenario with five years of bisphosphonate use in which the risk of an atypical femoral fracture offset the hip fracture reduction. And we now have our final fellow question. So Lippy Marion from Mass General will ask a question.

B (33:38)

I've got a around 30% of the.

C (33:40)

Cases of AFFs did not have any previous exposure to bisphosphonates in the study population.

B (33:47)

What do we think is driving these numbers?

C (33:50)

And what does it ultimately tell us about the relationship between AFF and bisphosphonates, if there actually is any. So 2/3 of the cases were on bisphosphonate and one third were not. We already went over the mechanism of bisphosphonate interfering with a healing mechanism for micro cracks, allowing cracks to extend and cause a fracture. But there are other factors and we touched upon them. One of them is mechanical loading. And if you notice, Dr. Amal has mentioned that age was not a predictor of aff. If anything, the mean age of AFF patients was almost eight years younger than a regular hip fracture patients. So these are relatively healthier, more ambulatory subjects who may be going around and making more activities. Second, AFF happen at the lateral cortex, which is where most of the tensile strength impact happens, again pointing to activity and loading. And last but not least, we touched upon femoral geometry and there have been some studies suggesting that we need to understand more. And that is why Asians may be at high risk for eff. So. So as Dennis Black said it, it could be a combination of a storm of risk factors. Some of it is bisphosphonate related. It's like the fire on already, a fire going on. But it's not the only causative factor. It precipitates matter because it's altering the healing process that may have occurred in any bone. And the last thing I want to say is AFF are not osteoporotic fractures par excellence. Bone density has never been found to be a predictor of these factors. And actually these people are younger, and I think it's bone quality that we need to look into that we don't have a good grasp upon. So these are all unknowns that need to be elucidated with more investigations.

A (35:37)

The authors then move on to go through their strengths and weaknesses. And to start with their strengths, they point out, one of them is that their plain films are read centrally by blinded experts. They also mentioned that they had long medication records available. And finally, that bisphosphonate use that was documented by pharmacy records avoided a recall bias. And to expand on that a little bit more, I would actually add to that is that it's a common issue with case control studies is that when you are finding individuals who have had an outcome and you are talking to those individuals or in some way doing something at that point to collect that information, it's very common for those subjects, for those individuals who know that they have had an outcome to recall their exposures differently. This can also happen on the investigator side. So if you know that the person you are talking to, the subject you are speaking with, had an outcome, you may be looking differently for an exposure as opposed to somebody who did not have that outcome. And that's called a detection bias. So with this strength, if they're pulling records, that avoids the recall bias and the detection bias. So I would definitely include that in a strength that the authors list there. Finally, as far as limitations go, they point out, as we've talked about already, that they do lack some lifestyle and, and clinical data that they would have liked to have had. They point out that atypical femoral fractures were too rare to determine a difference based on the type of bisphosphonate that was used by the patient. They also mentioned that they could not understand the impact of sequential treatment or ending a drug holiday and that the findings may only be applicable to a north European population. So more on that in a second. And finally, it was also been pointed out they didn't have any data on hip geometric measurements. I would like us to hang out here for just a second and think a little bit more about what the authors point out is, is whether these findings are going to be applicable to a broader population. So how much concern do we have? We don't know all the data, as Amal pointed out, exactly what the population is, but fairly safe to assume it's an overwhelming majority of a North European population. So are we going to be able to extrapolate these results to other populations around the world, or are we going to have to hold on that until we get more representative data?

C (37:49)

This is a very important point and it gets back to the issue of relative risk and absolute risk in general for health outcomes. Relative risks for any health outcome are usually applicable to other population, but where we cannot go and make that leap to externally generalizing is the absolute risk. So we could say that the relative risk of five folds over five years is correct, probably in other population. But what this means a number needed to treat and number needed to harm. And the balance of risk benefits is going to be different based on the actual risk in that population. And the actual risk in that population, as we know, is definitely higher in Asians. So the numbers are not totally transposed from this population. The relative risk may be as in any health outcome, cardiovascular cancer, et cetera, but not the average absolute risk. So I think this is one of the largest studies that gives us some good insights and confirms what we knew from previous publications on the Fact that the risk is very small, the risk increases with time, decreases with discontinuation, but the numbers for number needed to treat and number needed to harm only applies to the population where they were generated and would have to be validated in other population. Amal, what do you think?

B (39:05)

I agree you practice in Lebanon where your population is very different. I practice in Iowa, so my population is rather homogeneous ethnically or racially. But yes, it is definitely a concern, particularly in areas where the population is drastically different or there's much more heterogeneity and background. What I like about this study is that it's the longest to date and with a solid design. So it's one of the most solid sources of data on this topic to this point. But the generalizability will need to be a factor when we apply it.

A (39:44)

So we'll wrap up by quoting the authors in their conclusion and they say that bisphosphonate treatment for over three to five years was associated with a very small increase in the absolute risk for atypical femoral fractures and confirms that such treatment indeed substantially reduces hip fracture risk in a real world setting. Setting. So in a minute we're going to think about how we might change our practice based on this or at least have our practice confirmed. I would like us to think a little bit more about the quality of this report overall. Amal, we'll start with you. You've mentioned some of your thoughts already, but expand on those. What are your thoughts on the quality of this study as reported by the authors?

B (40:21)

I think given that this type of question can only be answered in a population study setting, the design of the study combines two very strong methodological approaches, which is the cohort approach to look at the occurrence of the outcome over time, but by definition, because of the rarity of the outcome, the need to mix it with a case control approach. So I think the design for the question is a very solid design. It incorporated a long period of follow up. The adjudication of the study of the fractures makes it more trustworthy that we are talking about what we think we are talking about. Then what I really appreciate about this study is not just the quantification or the assessment of the changes in risk during the period of use, but also the quantification of the decline in risk after discontinuation of the bisphosphonates, which there were some data on this topic, but rather limited data and smaller numbers. So now we have a much bigger assessment of that question.

A (41:29)

Got us. And question for you, thoughts on the quality of this report overall.

C (41:32)

All right, so I'm not going to repeat what Amal said. I fully agree, but I'd like to add a couple of observations. We look at randomized trials as the highest evidence, but we know that randomized trials with oral bisphosphonate and even IV bisphosphonate almost never reported aff. So we're never going to be able to get the data in the pristine status that we want to get. This is as good as it gets. So I I fully concur with ML. AFF only appeared in the post marketing era when the numbers were in the ten thousand and hundred thousand. So this is very solid and robust. I want to go back to one thought actually I just thought about while reflecting on this we said we would have loved to have vitamin D levels. Having said that, we do know that northern European country, contrary to what we have thought, have some of the best vitamin D levels around because of governmental regulation of vitamin D fortification of food products. So it is possible that they have an nhanes like equivalent and they may have an idea of vitamin D levels in this population. So although we would have loved to have it, this is probably findings in a north European country that is probably vitamin D replete and I want to make sure we understand that. And that may not be necessarily applicable to populations who are Asians, Indians, vitamin D deficients, et cetera and Gada.

A (42:47)

Let's stick with you for final question as we think about how our practice might need to change or perhaps in this case how we can strengthen in our current practice. So how would you recommend that we endocrinologists change what we do or can we make it better in particular as we try to communicate this to patients? You mentioned already how you communicate this in clinic, but do we need to be thinking about other methods? You mentioned that the worry about AFF is out there in many different places from friends, from family and the media. Do we need to be thinking about any non traditional methods, at least non traditional for us and trying to get that word out there. What are your thoughts on how it works?

C (43:23)

Absolutely. So one of my actually topics from my master degree many years ago was patient adherence and compliance with medication. And when I studied that topic we found multiple factors that affect compliance. One of them is contradicting information from health care providers, from media, from friends, et cetera. So I think we as physician have the responsibilities to come and the SBMR did talk about the bone coalition of a common message that all societies that are concerned with bone health outcomes should give the same message, maybe share the development of education material that we all hand to the patients so they hear the same thing over and over again and don't hear many mixed messages. Media is very important and I think media plays a major role. People just search on the surf the Internet or watch the TV episodes and we have our task cut for us over there. And that's another part. Finally, the Endocrine Society is very involved in patient education and there is a lot of patient educational resources that that come out with practice guidelines I think at some point. And we didn't do it for the long term bisphosphonate task force, but it's something to be rethought with additional data concerns, low compliance, major care gap with bisphosphonate, which is a forefront medication to develop patient education material that is on the website, that is pamphlets that we can hand on to them and give them. And then in my clinic when I have a patient coming with an EFF and says doctor so and so told me, I basically pick up the phone or send an email and give my idea. And specifically not only for aff but aonj, I talk to the dentist, I give them the guidelines from the dentist associations. I have a beautiful tabulation for that and I say here's the evidence and when you do that, when you prevent mixed opinions or you engage physician in a common approach to a patient, the patient is happier and believe me, the patient is more compliant. It's not as simple as that. But we have our task cut for us to improve patient care in education.

A (45:24)

And with that I would like to thank Amal Shibli, Rahal Ghada, El Haj Fulahan and our four fellows for joining me for this month's edition of Endocrine Feedback Loop. I hope that you all learned as much as I did and that you'll join us again next month. And now you're in the loop. This has been Endocrine Feedback Loop. Endocrine Feedback Loop is brought to you by the Endocrine Society with production oversight by Brandi Brown and Andrew Harmon. If you want to like and subscribe, you can find us on Apple, Spotify or wherever you get your podcasts. We'd love to hear your feedback on this episode or the podcast itself. Please email us@podcastrent.org.

C (46:12)

Endocrine Feedback Loop is.

A (46:13)

A free service of the Endocrine Society.

C (46:16)

To learn more or to become a member, visit the society's website at www.endocrine.org.