Cochrane in CORR®: Surgery for Rotator Cuff Tears

Importance of This Topic Atraumatic rotator cuff tears are common in older patients, and more than 30% of people older than 60 years have asymptomatic rotator cuff tears, a proportion that increases with age [16]. It is not well-understood why some rotator cuff tears become symptomatic, but symptomatic tears are a common reason for patients to seek a consultation from an orthopaedic surgeon [15]. Common treatment modalities include physical therapy with or without a corticosteroid injection and rotator cuff repair, but the efficacy of surgery, what approach and technique to use, and the indications for performing it are topics of ongoing debate [1]. The debate is certainly warranted, given that the number of rotator cuff repair procedures performed annually is increasing over time [3, 17], the cost of rotator cuff repair surgery is high [10], and the patient-reported and functional outcomes after rotator cuff repair may not be superior to those of nonoperative treatment [1]. In this Cochrane review, the primary objective of Karjalainen et al. [5] was to compare operative and nonoperative treatment of symptomatic atraumatic rotator cuff tears. The review found no clinically important differences in pain, function, or health-related quality of life 1 year after either rotator cuff repair or physical therapy. They considered the between-group differences in light of the minimal clinically important difference, which is defined as the smallest difference in a given outcome that patients perceive to be beneficial [4]. A secondary objective of this review was to compare rotator cuff repair with and without acromioplasty. Not surprisingly, there were no observed benefits to acromioplasty in terms of pain, function, or health-related quality of life—a finding that has been known for some time and is based on several high-quality randomized controlled trials [2]. Upon Closer Inspection Looking closely at the data, all three of the studies in the review comparing rotator cuff repair to physical therapy [8, 9, 11] were at a high risk of bias. First, participants were not blinded, which can overestimate the effects of the intervention, and there were no placebo control groups. Second, the confidence intervals are wide, which suggests the estimates of the results are not precise. Third, the Constant score was used as the primary outcome in each study. Given that 65% of the Constant score is based on clinician assessment, this outcome can be subject to assessor bias [14]. Fourth, there was heterogeneity in the patient populations between studies (one trial included traumatic tears [11]), duration of preoperative symptoms (10 to 28 months) and prestudy treatment, and surgical technique and concomitant procedures, all of which can make grouping and comparing data challenging. Interestingly, many of the biases in these studies would tend to favor surgery, yet despite them, the review found no clinical benefit to surgery in terms of pain or function [5]. Specifically, the magnitude of the between-treatment differences were so small that the typical patient would not consider them important. One important consideration of this review is follow-up, whereby most of the included studies were limited to 1 or 2 years of follow-up. The argument against nonoperative treatment has been the unknown risk of tear enlargement, increase in tear retraction, and the development of fatty infiltration of the muscle, all of which could contribute to making a later repair more challenging with the potential for an inferior outcome compared with early surgery [6]. This consideration is certainly relevant given that in one study, up to 24% of patients in the nonoperative group eventually underwent surgery at 5 years [11]. Conversely, retear rates after rotator cuff repair were also reported to be high (up to 31% at 2 years [8]), and although many patients did not undergo repeat surgery or exhibit a decline in clinical outcome, one study [12] noted that the functional outcome (Constant score) of patients who had a retear was inferior to the outcome of those with an intact repair at 10 years of follow-up. Take-home Messages This Cochrane review [5] found no clinically important benefit to surgery in the treatment of symptomatic, atraumatic rotator cuff tears. This is at odds with common surgical practice, and we need to take this difference seriously. This does not mean that surgery is always ineffective; rather, it suggests that we need to refine our indications to see whether there are certain subpopulations in whom surgery is more effective. For instance, the studies included in this Cochrane review focused largely on atraumatic rotator cuff tears in older patients, and the recommendations of this review do not pertain to acute tears in younger patients. The ongoing debate surrounding the role of surgery in the management of symptomatic atraumatic rotator cuff tears is reminiscent of the debate surrounding surgery for the management of degenerative meniscus tears in older patients—another area where reasonable randomized trials have found no benefit to surgery, yet surgeons continue to perform these operations in great numbers [7, 13]. Failing to follow the evidence is a blemish on our specialty and a risk to our credibility with patients. Generally, when there is ambiguity in the evidence between the benefits of surgery and a nonsurgical alternative, we generally should recommend the nonsurgical alternative first, given the risk of surgery. However, given the aforementioned limitations of the available evidence, the story of the role of repair in the management of symptomatic atraumatic rotator cuff tears is far from over, and there remains a clear need for large, prospective, comparative studies to understand whether rotator cuff repair is truly no different from physical therapy in both the short- and long-term (longer than 1 year) and what factors influence the success and failure of either surgical or nonsurgical treatment.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.