What is a failure?
We often define failure of an arthroplasty as a revision. It is evident, however, that the lack of a revision does not indicate that the patient has had a successful outcome. The lack of a revision simply indicates that the surgeon was unwilling to do another operation on the patient — or that the patient did not want more surgery, even though they were unhappy about the result.
If we equate revision and failure, many clinical failures go unseen. In the UK National Joint Registry, among 21,918 patients with a recorded postoperative Oxford Shoulder Score, 26.99% of those having a reverse total shoulder arthroplasty (RSA) had an unsatisfactory result, defined as an Oxford Shoulder Score below 29. Fewer than 1 in 20 of them (4.87%) were revised. Among patients with an unsatisfactory score, the proportion revised was 10.58% after anatomic total shoulder arthroplasty (aTSA) and 13.86% after hemiarthroplasty [1]. The authors concluded that the lower revision rate after RSA may not indicate better outcomes than anatomic arthroplasty, but rather a higher threshold for revising an RSA [1]. When an RSA fails, the prospect for improving the patient’s comfort and function is often poor, so the patient keeps the implant and keeps the disability.
Furthermore, achieving the minimal clinically important difference (MCID), the substantial clinical benefit, or the patient acceptable symptom state on the ASES, SANE, SST, or VAS after shoulder arthroplasty — the common indicators of a “successful arthroplasty” — did not correlate with patient satisfaction, with willingness to undergo the operation again, or with willingness to recommend it to a friend or family member [3]. In a prospective cohort of 1,559 RSAs, the 134 patients (8.6%) who rated themselves unchanged or worse had nevertheless improved on every measured outcome [4]. Their scores went up. They were not satisfied.
On one hand, a change in a patient-reported outcome that does not reach the MCID establishes that the patient has not improved. On the other hand, a change that exceeds the MCID does not establish that the patient is satisfied. MCID thresholds give strong evidence of failure and weak evidence of success.
Thus, the usual methods do not identify the patients who find their outcomes unsatisfactory.
Inspired by the wonderful book, “All the Light We Cannot See,” we wondered about “All the Failures We Cannot See”.
We are realizing that failure needs to be defined more broadly: the patient is dissatisfied or in pain, has sustained a complication, has been revised, or is simply not better than before surgery. The 1 in 10 patients who report themselves unchanged or worse after an anatomic total shoulder arthroplasty [5], and the roughly 1 in 5 who are dissatisfied after total knee arthroplasty [6], are where we can look for opportunities to improve our method by asking, as Codman would have us do, what could we have done differently that might have prevented the failure? [2, 8, 9]. Over five years Codman tracked his patients by sending them End Result Cards at a year after surgery, asking whether they were better. For those who were not better, he tried to determine why not [10]. What deserves emphasis is his method: he asked each patient for follow-up, and he did it with a simple card. The modern version of Codman’s card is a text message or an email, sent on the anniversary of the operation. The medium has changed; the method has not. Asking is what makes the failures visible.
When we identify a patient with an unsatisfactory result, we have a unique opportunity to ask questions such as “If I had chosen a different implant, if my implant fixation had been better, might this failure have been avoided? Should I have operated on this patient at all? What is different about my care of this patient in contrast to comparable patients in my practice who did well?” As Pearl has argued, these counterfactual questions are the language of causal reasoning [11].
It is important to avoid naming the mode of failure and calling it the cause. “Glenoid component loosening” is not a cause; it is what happened. The question is what led to it — the bone quality, the deformity, the technique of bone preparation, or the seating of the component.
By analogy, consider a report that the battle was lost because the general did not arrive. The general’s absence was not the cause; it was what happened. The battle was lost because the farrier failed to place the horseshoe nails properly, the shoe came loose, the horse stumbled, the general broke his leg, and he could not lead the charge. Once recognized, the nail placement is the thing that must be fixed before the next battle. So the question for us is: what might we do to reduce the risk of glenoid component failure in the next case?
How can we see our failures?
It starts with a secure log of our own cases.
(1) For each surgery, enter the following
Name | Medical Record No. | Date of Birth | Mobile | Email
Diagnosis | Procedure | Surgery Date
(2) Prepare a short message — a text or an email, sent through the institution’s patient portal or another secure channel — to go to the patient at 1 year after surgery.
I am interested in knowing how you are doing after your surgery. Please reply to this message. Are you better than before? Have you had any problems? If so, please let me know about them.
(3) Trigger the message from the surgery date. A calendar reminder and a delayed send will do it; an automated text service will do it without your having to remember. This trigger is the whole reason the log exists.
(4) When the patient replies, add the response to the log. When the patient does not reply, consider asking the office to telephone them.
Follow-up Date | Improved? | Additional Surgery?
(5) For each failure (not improved, additional surgery), compare the case and its treatment with similar cases that did not fail.
(6) Ask yourself what might have been done differently — patient selection, characterization of anatomy, procedure and implant choice, technique, perioperative management, team and system factors — to avoid the failure. Ask, as Codman did, “why not?” Enter this information into the log.
(7) Keep the log where you will read it. Before a comparable case, look back at what you wrote about the last failure of that kind.
(8) Recognize that each patient who finds their outcome unsatisfactory is an opportunity to refine your method. Two failures of the same kind are not two instances of one thing. A dislocation after reverse total shoulder arthroplasty in one patient and a dislocation in another are different events — a different patient, a different anatomy, a different set of decisions, a different day in the operating room — and a unique opportunity to learn.
Final thoughts
The measure of success is the steady refinement of our method across a career. Each of us has the opportunity to see whether our patients are better, and to learn from those who are not. Each patient who reports that they are no better is one case we can learn from.
Some of us might prefer to avoid asking a question that carries the risk of getting a “bad news” response. Others might want to avoid calling the patient’s attention to a suboptimal outcome. However, the opposite may be true — by asking, we show the patient that we care. Furthermore, in seeing the failure we may identify a chance to remedy it.
The opportunity to identify, learn from and care for these patients is there for each of us.
Seeing by looking
Short eared owl
Skagit
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//
References
[1] O’Malley O, Davies A, Rangan A, Sabharwal S, Reilly P. Is there a difference in thresholds for revision between shoulder arthroplasty types? A National Joint Registry study. PLoS One. 2025;20(8):e0330975. doi:10.1371/journal.pone.0330975
[2] Menendez ME, Matsen FA 3rd. Learning from surgical failures. J Bone Joint Surg Am. 2026;108(8):547-548. doi:10.2106/JBJS.25.01110
[3] Khan AZ, Vaughan A, Aman ZS, Lazarus MD, Williams GR, Namdari S. Reaching MCID, SCB, and PASS for ASES, SANE, SST, and VAS following shoulder arthroplasty does not correlate with patient satisfaction. Semin Arthroplasty JSES. 2024;34(4):819-826. doi:10.1053/j.sart.2024.03.017
[4] Parsons M, Routman HD, Roche CP, Friedman RJ. Patient-reported outcomes of reverse total shoulder arthroplasty: a comparative risk factor analysis of improved versus unimproved cases. JSES Open Access. 2019;3(3):174-178. doi:10.1016/j.jses.2019.07.004. PMID:31709358
[5] Hao KA, Hones KM, O’Keefe DS, Elwell J, Simovitch RW, Wright TW, King JJ, Schoch BS. Does the relationship between preoperative function and achievement of clinically important benchmarks of success after total shoulder arthroplasty depend on outcome assessment design? Clin Orthop Relat Res. 2025;483(3):377-395. doi:10.1097/CORR.0000000000003347. PMID:39778205
[6] Bourne RB, Chesworth BM, Davis AM, Mahomed NN, Charron KDJ. Patient satisfaction after total knee arthroplasty: who is satisfied and who is not? Clin Orthop Relat Res. 2010;468(1):57-63. doi:10.1007/s11999-009-1119-9. PMID:19844772
[8] Codman EA. The product of a hospital. Surg Gynecol Obstet. 1914;18:491-496.
[9] Reverby S. Stealing the golden eggs: Ernest Amory Codman and the science and management of medicine. Bull Hist Med. 1981;55(2):156-171. PMID:7020802
[10] Codman EA. A Study in Hospital Efficiency: As Demonstrated by the Case Report of the First Five Years of a Private Hospital. Boston, MA: Thomas Todd Co.; 1918.
[11] Pearl J, Mackenzie D. The Book of Why: The New Science of Cause and Effect. New York, NY: Basic Books; 2018.
