Of successful mammoth hunting and glenoid component failure - modeling causation

 

The goal of the Shoulder Arthroplasty Failure Research initiative is to reduce the risk of future patients having a surgical revision for failure of their shoulder arthroplasty.

It is apparent that this goal cannot be accomplished simply by classifying the failure (glenoid component failure, instability, infection, etc).  Such a classification does not generate actionable intelligence that can be used to reduce the rate of failure. Rather we need to identify and seek to address the factors contributing to each type of failure.

In his "Book of Why"

Pearl has set out a compelling approach to the study of causation based on a model that identifies the factors that may affect the outcome of interest. 

He gives the example of a mammoth hunt taking place 10,000 years ago. Here's my modification of the figure in his book.

Note that the design of the model does not attempt to predict the relative importance of each of the factors on a given day, (1) because cumulative data (i.e. experience) from past hunts progressively modifies the weight attached to each factor and (2) because these factors vary from hunt to hunt (e.g. rain vs sunshine).

Note also that the items listed on the left are not under the hunters' control, but they do affect the chances of a successful hunt and may modify the hunter-controlled factors on the right (a bigger mammoth may indicate the need for more and more experienced hunters). In other words, this causal diagram helps the hunters imagine a plan for a hunt that has yet to take place.

Let's now do some time travel to today where we confront the fact that about one in ten shoulder arthroplasties requires a surgical revision. Published articles indicate that a leading indication for surgical revision is glenoid component failure  Reducing the risk of future glenoid failures requires a causal model analogous to that for the mammoth hunt, which identifies factors that could contribute to revision for a failed glenoid component.

This causal model provides a framework enabling the investigating surgeon to (1) evaluate the role that each factor (both modifiable and non-modifiable) may have played in each case of glenoid component failure and (2) asses the counterfactual argument: "if this factor had not been present, would the chances of glenoid component failure have been reduced?" As in the case of the mammoth hunt, cumulative data from past failures progressively modifies the weight that can be attached to each factor.

The bad news is that there are a lot of factors to consider in each case (my current count is 25, but I'd appreciate the reader adding to the list). The good (?) news is that there are lots of cases of glenoid failure out there from which we can collect the data that will inform future practice.

While the opportunity to improve the chances of a successful mammoth hunt has passed

The opportunity to reduce the risk of patients having surgical revision for a failed glenoid component is clear and present.

Follow on twitter/X: https://x.com/RickMatsen
Follow on facebook: https://www.facebook.com/shoulder.arthritis
Follow on LinkedIn: https://www.linkedin.com/in/rick-matsen-88b1a8133/

Here are some videos that are of shoulder interest
Shoulder arthritis - what you need to know (see this link).
How to x-ray the shoulder (see this link).
The ream and run procedure (see this link).
The total shoulder arthroplasty (see this link).
The cuff tear arthropathy arthroplasty (see this link).
The reverse total shoulder arthroplasty (see this link).
The smooth and move procedure for irreparable rotator cuff tears (see this link)
Shoulder rehabilitation exercises (see this link).

Complications after total shoulder arthroplasty - the Surgeon is the Method

 

The quest for ways to make shoulder arthroplasty safer for future patients continues through Shoulder Arthroplasty Research. Here are some things we know: 

(1) most shoulder arthroplasties turn out well for the patient, thus our greatest opportunities to learn come from studying failures

(2) it is insufficient to focus on the type of failure (e.g. glenoid component loosening, rotator cuff failure); rather we need to hone in on what could have been done differently at the primary arthroplasty to lower the risk of component or cuff failure - this can be thought of as actionable intelligence.

(3) we want to avoid the assumption that technologies such as 3D CT based planning, patient specific instrumentation, robotics, virtual reality, augmented reality will lower failure risk until their effectiveness in vivo has been rigorously demonstrated

(4) our attention falls on the surgeon and the elements of care that are under her/his control

A recent article, The effect of surgeon volume on complications after total shoulder arthroplasty: a nation-wide assessment, provides some actionable intelligence. The authors retrospectively queried the Pearl Diver Mariner database for the years 2010 to 2022. Their analysis included 155,560 patients having primary anatomic total shoulder arthroplasty, excluding those younger than 40 years, those who underwent revision arthroplasty, cases of bilateral arthroplasty, and cases with a history of fracture, infection, or malignancy.

They included cases performed by surgeons with a minimum of 10 cases. 

The 90th percentile for surgeon volume was determined to be 112 cases during the study period. Surgeons above the 90th percentile (n 340) operated on 68,531 patients, whereas surgeons below the 90th percentile (n 3038) operated on 87,029 patients. Surgeons in the high-volume group were significantly more likely to have completed a Shoulder and Elbow fellowship and less likely to have no fellowship training or fellowship training outside of Shoulder and Elbow or Sports Medicine. 

Low-volume surgeons operated on patients with higher baseline comorbidities. Here's my summary of their data.

After adjusting for age, gender, CCI, obesity, and tobacco use, high-volume surgeons experienced lower rates of medical complications including renal failure, anemia, and urinary tract infection. All-cause readmission, reoperation at 90 days, and reoperation at 1 year were significantly lower among high-volume surgeons. Cases performed by high-volume surgeons exhibited lower rates of all complications including prosthetic joint infection and periprosthetic fracture. Here's my summary of their data. NB an odds ratio <1 means that cases operated by high volume surgeons had a lower rate of the complication than low volume surgeons.

Finally, the authors found that the proportion of shoulder arthroplaties performed by high volume surgeons has been decreasing with time.

Comment: This study appropriately puts the focus on the surgeon - the individual that decides which treatment is best suited for each patient, carries out the surgery, and manages the aftercare. In other words, the surgeon controls the modifiable variables for each patient. The surgeon is the method.

The authors characterize the surgeon in three dimensions: (1) case volume, (2) fellowship, and (3) the comorbidites of the patients the surgeon selects to have total shoulder arthroplasty. They then go on to compare complications for surgeons performing ≥ 112 arthroplasties to those performing < 112. 

Thus the data available are ripe for a multivariable analysis (MVA) characterizing the relationship among these variables - individually or in combination - to the occurrence of medical and surgical complications. Without such an analysis we cannot know the relative importance of each of these dimensions.

Let's look at each of these characteristics:

Surgeon case volume: One of the big questions in orthopaedics is whether more is more, i.e do we continue to get a bit better with each case, or is there a threshold above which we are "good"? In this light it might be more informative to characterize surgeon case volume as the number of cases rather an whether they exceeded a threshold for qualification as "high volume"?  This would get around the problem of having a surgeon performing 111 cases designated as "low volume" whereas if the surgeon had done one more case he/she would suddenly become "high volume". On reading this paper, a patient might ask "should I travel four hours to have an arthroplasty by a surgeon who has done 120 cases rather than sticking with my local surgeon who has done 110?  Numbers may be better than categories. An MVA should be able to sort this out.

Fellowship: The additional year of specialized training afforded by fellowship exposes trainees to a greater case volume and breadth. High volume surgeons were more than twice as likely to have taken a shoulder fellowship; however, fewer than 30% of high volume surgeons took a shoulder fellowship. As a result we do not know from the data presented whether taking a shoulder fellowship results in a significantly greater arthroplasty practice volume or whether taking a shoulder fellowship reduces the surgeon's complication rate. An MVA should be able to sort this out.

Comorbidities: The patient population of high-volume surgeons was significantly healthier, i.e., comorbidities as reflected by the Charlson Comorbidity Index were lower in patients operated by higher volume surgeons (perhaps because experience teaches to think carefully before offereing elective surgery to patients who are ill or perhaps high volume surgeons operate in outpatient centers that exclude sick patients). The question is whether a shoulder fellowship or being a high volume surgeon enables safer surgery on patients with comorbidites. An MVA should be able to sort this out.

Complications: This article presents data on medical and surgical complications in terms of odds ratios, but does not present data on the rate of each complication. In an MVA it may be easier to characterize complications in terms of their rates.

Arthroplasty choice: The authors point out that "distinctions between anatomic and reverse shoulder arthroplasty were not made because of limitations associated with CPT coding". This is an important shortcoming of the analysis, because experienced (and perhaps fellowship-trained shoulder surgeons), may be better at deciding which patients are the best candidates for each procedure in terms of avoidance of medical and surgical complications.  

Incremental value of each case: Numbers are not the only important thing. How much the surgeon learns from each case depends in large part on whether the sugeon conducts an After Action Report (AAR) after each case. An AAR is a structured process used to review the case to identify what happened, what went well, and what could be improved in future cases. We can assume that a 100 case surgeon who routinely conducts AARs will have better outcomes than a120 case surgeon who goes on to the next case without introspection. This is important because most shoulder arthroplasties are not operated on by high volume surgeons.

We can do a better job of helping our patients avoid problems.

Fresh Grizzly Bear Footprint

Devil's Gap, Alberta

Photo by Laura Matsen, M.D

8/23/2025

Follow on twitter/X: https://x.com/RickMatsen
Follow on facebook: https://www.facebook.com/shoulder.arthritis
Follow on LinkedIn: https://www.linkedin.com/in/rick-matsen-88b1a8133/

Here are some videos that are of shoulder interest
Shoulder arthritis - what you need to know (see this link).
How to x-ray the shoulder (see this link).
The ream and run procedure (see this link).
The total shoulder arthroplasty (see this link).
The cuff tear arthropathy arthroplasty (see this link).
The reverse total shoulder arthroplasty (see this link).
The smooth and move procedure for irreparable rotator cuff tears (see this link)
Shoulder rehabilitation exercises (see this link).

Shoulder Arthroplasty Failure Research in action

Most patients having primary arthroplasty do well, 

thus we learn most from studying the failures.

Publications of shoulder arthroplasty outcomes usually report the diagnostic categories associated with failure, such as glenoid component loosening, rotator cuff tear, infection, or instability. While such classifications are descriptively useful, they provide little insight into how failures might be prevented for future patients.

As Judea Pearl has emphasized in his must-read book

actionable knowledge is more likely to arise from asking the counterfactual question: “What might have been done differently that could have prevented this complication?” This principle underlies the Shoulder Arthroplasty Failure Research initiative, which seeks to move beyond descriptive epidemiology toward identification of surgeon-controlled, modifiable factors that govern arthroplasty failure.

A recent study exemplifies this approach: “Humeral and glenoid component malposition in patients requiring revision shoulder arthroplasty: a retrospective, cross-sectional study.” In this investigation, failure of a primary arthroplasty was defined as the occurrence of revision. The authors reviewed 234 revision shoulder arthroplasties performed at 3 institutions.

They reported demographic characteristics and frequencies of revision types following hemiarthroplasty, anatomic total shoulder arthroplasty (TSA), and reverse total shoulder arthroplasty (RSA)

While such descriptive information is important, it does not inform strategies for prevention of arthroplasty failure for patients having arthroplasty in the future. 

The study focused on some of the factors under the surgeon’s control, specifically the position of glenoid and humeral components. The findings were striking: glenoid malposition was identified in 51% of anatomic TSA revisions and 93% of RSA revisions. Humeral component malposition was also frequent, present in 57% of anatomic TSA, 62% of RSA, and 54% of hemiarthroplasty cases. These observations support the counterfactual inference that had the components been positioned appropriately, the likelihood of failure requiring revision may well have been substantially reduced. 

Here are a few examples from the article.

Placement of the RSA baseplate in superior tilt.

Superior placement of the RSA baseplate.

Superior placement of the humeral component in anatomic TSA

Inadequate humeral neck cut in hemiarthoplasty resulting in

superior-medial placement of the humeral component and overstuffing of the joint.

In that the surgeon is the method, each of these malpositions could have been avoided by better surgical technique.  The institutions conducting the revisions were usually not involved in the majority of the primary procedures and thus medical records for many of these patients could not be fully reviewed. As a result, the characteristics of the surgeon performing the primary arthroplasty that was revised (age, training, years in practice, arthroplasty experience, etc) were not available, but would be of great interest.

Comment: This is an imporant study in that it identifies actions that shoulder surgeons can take to reduce the risk of arthroplasty failure for their future patients. It provides a model of how clinically meaningful shoulder arthroplasty failure research can be conducted.

This is an uncommon bird, but well worth investigating.

Elegant Trogan

Madera Canyon, Tucson AZ

May, 2022

Follow on twitter/X: https://x.com/RickMatsen
Follow on facebook: https://www.facebook.com/shoulder.arthritis
Follow on LinkedIn: https://www.linkedin.com/in/rick-matsen-88b1a8133/

Here are some videos that are of shoulder interest
Shoulder arthritis - what you need to know (see this link).
How to x-ray the shoulder (see this link).
The ream and run procedure (see this link).
The total shoulder arthroplasty (see this link).
The cuff tear arthropathy arthroplasty (see this link).
The reverse total shoulder arthroplasty (see this link).
The smooth and move procedure for irreparable rotator cuff tears (see this link)
Shoulder rehabilitation exercises (see this link).

Despite advancements in implant design and surgical expertise, the rate of failures due to glenoid component loosening remains unchanged

 We continue to probe the question - "what could have been done differently to reduce the risk of each patient experiencing shoulder arthroplasty failure" (see Recent JBJS article on failures (51%) of anatomic total shoulder arthroplasties)

A recent article, Why do primary anatomic total shoulderarthroplasties fail today? A systematic review and meta-analysis. reviewed 44 studies involving 35,168 aTSA procedures. 2,744 (one out of every 13) patients experienced failure of their arthroplasty. 21.7% of the failures were attributed to glenoid component loosening. 

In their discussion, the authors state, "Glenoid loosening is a known late-term complication with radiolucency of the glenoid component commonly occurring at 5 years and progressing continually over time.

Despite advancements in implant design and surgical expertise, the rate of failures due to loosening remains unchanged, especially with the rise in augmented polyethylene glenoid components for advanced posterior glenoid erosion.

Failure of the glenoid component is likely due to multiple factors, some of which may be related to surgical technique or implant design, and others are likely patient-specific factors such as rotator cuff insufficiency or glenoid morphology."

This study defines the problem: too many patients are experiencing glenoid component failure, but does not provide a path forward.

Recognizing that when we look at a large series of failures, such as the one presented here, we are befuddled by the many factors that may contribute to failure. However, each failure occurs because of factors specific to that patient: the surgeon performing the procedure, the patient having the arthroplasty, the characteristics of the shoulder prior to surgery, the details of the procedure  (surgical technique, implant selection), and the failure mode.

So, there is an opportunity to reduce the risk of failure in future arthroplasties by conducting case by case analyses of past failures. The goal of Shoulder Arthroplasty Failure Research is to implement a system for conducting these analyses and progressively adding analyzed cases to a surgeon-accessible de-identified, HIPAA-compliant library. As the library grows it has the potential for becoming an increasing valuable resource to inform the future care of patients with shoulder arthritis. 

Learning from each case of failure

On 22 October 1895, the No.56 train arriving from Grandville hurtled into the Gare de l'Ouest in Paris at a speed of 60 kilometres an hour, and, unable to stop, plunged through the buffers at the end of the platform. Its engine crashed through the façade of the station building and fell down on to the Place de Rennes

Follow on twitter/X: https://x.com/RickMatsen
Follow on facebook: https://www.facebook.com/shoulder.arthritis
Follow on LinkedIn: https://www.linkedin.com/in/rick-matsen-88b1a8133/

Here are some videos that are of shoulder interest
Shoulder arthritis - what you need to know (see this link).
How to x-ray the shoulder (see this link).
The ream and run procedure (see this link).
The total shoulder arthroplasty (see this link).
The cuff tear arthropathy arthroplasty (see this link).
The reverse total shoulder arthroplasty (see this link).
The smooth and move procedure for irreparable rotator cuff tears (see this link)
Shoulder rehabilitation exercises (see this link).

Recent JBJS article on failures (51%) of anatomic total shoulder arthroplasties

Patients and surgeons are interested in the causes and prevention of shoulder arthroplasty failures. Failures can be viewed in terms of the characteristics of the surgeon, the patient, the shoulder, and type of failure. Regardless of the cause of the failure, it is the patient (not the surgeon or the implant company) that bears the consequences. Thus, in each case we should do our best to determine "what could have been done to prevent the failure experienced by the patient?" In this way, each failure - whether ours or someone else's - becomes a learning opportunity. At an early morning meeting in August 2025, a small group of shoulder surgeons took the first steps in developing a Shoulder Arthroplasty Failure Research (SAFR) program with the goal of learning from individual cases of arthroplasty failure.

The June, 2025 JBJS article, High Failure Rates of Polyethylene Glenoid Components in Stemless Anatomic Total Shoulder Arthroplasty for Primary and Secondary OA, is interesting to consider in this regard.  These components were used in this series

At a mean followup of 72 months, out of 197 patients, over half (101) had failures necessitating surgical revision, 86 because of glenoid component loosening. What might be done to prevent these failures? 

Left:  Intraoperative view following explantation of the polyethylene glenoid with a 4 x 2-cm bone defect. Right: Explanted components with severe glenoid wear and superior glenoid abrasion, with the pegs completely separated from the body of the glenoid component.

Is the problem (as the title might suggest) the use of a stemless humeral component with a polyethylene glenoid component (in which case failures could be avoided by using a stemmed humeral component)? 

Or is it the type of polyethylene glenoid component being used (in which case a different glenoid component design could be used)? 

Or is it the technique by which the glenoid component was inserted (in which case greater attention could be directed at the quality of glenoid component preparation, cementing and seating?). 

These three elements are surgeon-controlled variables (in contrast to patient age, sex, diagnosis, BMI, critical shoulder angle and lateral acromial angle, which were measured in this study but which are of lesser interest in that they are not modifiable by the surgeon). 

So...is the stemless humeral component or the glenoid component the problem? Prior studies of this glenoid component (with either stemless or stemmed humeral components) reported glenoid component loosening rates between 25% and 100% after 5 years when used with either stemless or stemmed humeral components: Univers II shoulder prosthesis: a multicenter, prospective randomized controlled trial and Radiologic midterm results of cemented and uncemented glenoid components in primary osteoarthritis of the shoulder: a matched pair analysis. Perhaps failure could be avoided with a different glenoid component.

Or...might the technique of glenoid component insertion be an issue?

In this 6 week postoperative x-ray from the article there is cement between the back of the glenoid component and the glenoid bone (yellow arrows).  Cement in this location can crack and displace leaving the component unsupported. 

This problem can be avoided by placing the component directly on the properly reamed glenoid bone without interposed cement as shown below.

The other issue shown on this x-ray is the large amount of bone in the glenoid vault surrounding the pegs (red arrow) that prevents bone ingrowth around them.

This problem can be avoided by careful preparation of the glenoid bone with good carpentry so that bone ingrowth is enabled.

Perhaps the risk of failure could be reduced using a different insertion technique

As shown below, this patient experienced glenoid component failure

How could this failure have been prevented?

Comment: This is an example of the type of analysis that we hope to carry out in the Shoulder Arthroplasty Failure Research program. Stay tuned!

Looking to make wise choices

Barred Owl

Union Bay Natural Area

May 2025

Follow on twitter/X: https://x.com/RickMatsen
Follow on facebook: https://www.facebook.com/shoulder.arthritis
Follow on LinkedIn: https://www.linkedin.com/in/rick-matsen-88b1a8133/

Here are some videos that are of shoulder interest
Shoulder arthritis - what you need to know (see this link).
How to x-ray the shoulder (see this link).
The ream and run procedure (see this link).
The total shoulder arthroplasty (see this link).
The cuff tear arthropathy arthroplasty (see this link).
The reverse total shoulder arthroplasty (see this link).
The smooth and move procedure for irreparable rotator cuff tears (see this link)
Shoulder rehabilitation exercises (see this link).