UW Shoulder and Elbow Academy: More thoughts on the rotator cuff

In followup to the prior post (see this link), we were posed two additional questions. Here they are along with our thoughts.

1. What is the role of partial acromioplasty during surgery for rotator cuff disease?

The rationale for an acromioplasty is not established.

In 1934, Codman pointed out that the proximal humerus normally articulates with the coracoacromial arch, providing stability of the shoulder against superiorly directed loads. He stated that the coracoacromial arch “has an important duty and should not be thoughtlessly divided at any operation”.[1] When this advice is ignored – especially in the presence of cuff deficiency - the shoulder can be destabilized against superiorly directly loads risking anterosuperior escape.

There is a tendency to confuse the association of cuff pathology and acromial changes with causation of the cuff changes by the acromion. Today, many hold the belief that ‘impingement” causes cuff lesions. By contrast, in 1972, Neer reported shoulders with a “characteristic ridge of proliferative spurs and excrescences on the undersurface of the anterior process (of the acromion) apparently caused by repeated impingement of the rotator cuff and the humeral head, with traction of the coracoacromial ligament”[2] , in other words the changes in the acromion were caused by the cuff. Establishment of causation requires the application of criteria originally described by Hill et al in demonstrating the causation of scrotal cancer in chimney sweeps, and recently used to demonstrate the causation of chondrolysis by pain pumps.[3] When we see a certain acromial configuration in patients with cuff lesions, we must ask, “which is the chicken and which is the egg or are they both products of genetics and aging?”

The diagnosis of ‘impingement syndrome’ is difficult to pin down.[4] [5] Authors of a recent article[6] summarizing the published evidence relevant to the diagnosis of ‘impingement syndrome’ of the shoulder were unable to find good evidence to support any of these statements: (1) clinically common forms of rotator cuff abnormality are caused by contact with the coracoacromial arch, (2) spurs on the anterior aspect of the acromion extend beyond the coracoacromial ligament and encroach on the underlying rotator cuff, and (3) successful treatment of the “impingement syndrome” requires surgical alteration of the acromion and/or coracoacromial arch.

Our practice is to preserve the coracoacromial arch (call us ‘Codmanists’ if you will), finding that its surgical compromise is not necessary to perform the necessary cuff surgery and finding no reliable evidence that altering the coracoacromial arch changes the outcome for the patient.

2. Regarding surgery for rotator cuff disease, what do you perceive to be the major gaps in our evidence base, and how do you make treatment decisions in light of these gaps?

The gap is lack of evidence.

For each of our patients we must ask, was our treatment successful and “if not why not?” [7] so that our treatment of the next patient will be better informed.

We are humbled in the knowledge that over ¼ million cuff repairs are done each year[8] while we have such a limited understanding of how much patients benefit from this procedure in comparison to other management approaches for cuff pathology. The recent American Academy of Orthopaedic Surgeons clinical practice guidelines on optimizing the management of rotator cuff problems[9] provide no high-level evidence to guide treatment; 22 of 25 statements were inconclusive or based on weak evidence or opinion. A recent Cochrane review[10] on the rotator cuff concluded that (1) surgery may not lead to any difference in pain compared with different exercise programs, (2) arthroscopic surgery may not lead to any difference in outcome in the long run compared with open surgery, 3) there was not enough information to tell whether surgery would make a difference in the ability to use the shoulder normally, the quality of life, the shoulder's range of motion, the strength, the chance that the symptoms might come back, the time it takes to return to work or sports and whether people are satisfied with surgery.

Thus, rather than asking, “Why is it that despite the reported success of nonoperative treatment for cuff tears, arthroscopic rotator cuff repair is one of the most commonly performed procedures in the field of orthopedic surgery?,” one might ask “why is it that less than 5% of people with cuff tears undergo surgery each year?[11]”

As pointed out recently[12], “a patient's decision to undergo surgery is influenced more by low expectations regarding the effectiveness of physical therapy than by patient symptoms or anatomic features of the rotator cuff tear”. These expectations are, of course, set in large part by the information they get from the health care provider they visit, be it therapist or surgeon. As one wag pointed out, what we get depends on whether we visit the patisserie or the charcuterie.

We have many gaps that need to be bridged before we can come up with evidence-based practice guidelines. It is evident that the evidence we need will only come from a large database that includes basic data on individual patients from the full range of providers (‘consensus panels’ cannot provide evidence). Such a database will be enabled in large part by the answers to these questions:

(1) How can we capture data on the largest and most diverse sample of patients with cuff disease?

(2)What is the most practical way to characterize a patient with a cuff tear, i.e. what is the minimal data set we need to include the patient in a database that will enable us to compare the natural history and treatment outcomes for individual patients[13]?

(3)In such a database, how can one be sure that the breadth of possible treatments (including non-repair surgery[14]) are represented and that the type of treatment is characterized in a manner that enables analysis?

(4)In such a database, how will the preoperative and postoperative condition of the shoulder be documented?

If we are to derive robust practice guidelines for the management of rotator cuff conditions we will need ‘big data’ to avoid the trap of ‘fragility of statistically significant findings’ identified by Bhandari et al[15]. That is the gap we need to fill.