Introduction

One of the most frequent reasons for visiting a shoulder specialist is a rotator cuff lesion [18]. A rotator cuff tear can be caused by trauma or occur due to intrinsic [5, 6] and extrinsic [23] factors with increasing age [26]. Traumatic tears tend to be larger than non-traumatic lesions [12]. Partial-thickness tears can progress into full-thickness lesions, whilst large full-thickness tears, especially if involving two or more tendons, tend to increase in size and can become irreparable over time [8, 10, 16, 28].

If a degenerative rotator cuff lesion becomes symptomatic, conservative treatment with anti-inflammatory medication, steroid injections and physical therapy is often recommended as first-line treatment [9, 11]. However, non-operative management fails to restore the structural tissue damage since rotator cuff tears show no tendency toward spontaneous healing [28]. In contrast, surgical rotator cuff repair offers the possibility of tendon-to-bone healing. Healing rates following rotator cuff repair are subject to wide variation. Differences in structural healing are observed to be associated with tear size, degree of fatty infiltration, and patient age [3, 7, 17, 25]. In many clinical studies, rotator cuff reconstruction has been proven to restore function, reduce pain, and retard muscular deterioration such as atrophy and fatty degeneration [4]. But do these structural advantages of surgical tissue repair really lead to better clinical results than those achieved by conservative treatment? From the patient’s point of view: Is surgical rotator cuff repair worth the effort? These questions are discussed with animation in the medical literature. Patientsʼ demands as well as the growing socioeconomic impact of health care warrant judicious decision-making by health care providers especially with respect to surgical indications.

Randomized controlled trials (RCTs) should be the main source of evidence to answer these questions. Moreover, such trials should look for long-term follow up periods, since stable clinical results after repair of degenerative musculoskeletal tissue will not occur within the first years. A handful of studies compare operative versus conservative management of rotator cuff tears [13, 15, 21]. Piper et al. [24] identified only three RCTs that met the inclusion criteria for a systematic review in a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) analysis. Piper et al. found a small superiority of clinical results in the surgical arm, with only a few sub-elements reaching statistical significance. Since publication of that review paper in 2018, only one further RCT on this topic has been published. However, that particular paper by Moosmayer et al. can be considered as a hallmark paper, since it offers 10-year follow-up in 95% of patients; 1‑ and 5‑year data on this study group have been reported in earlier publications [20,21,22]. Such an undertaking in orthopedic clinical research is comparatively rare and could prove to be game-changing in a field where evidence from RCTs has been relatively scant to date. Therefore, although adding only one more paper, the authors consider an update of Piperʼs systematic review to be crucial, aiming to expand knowledge of the correct therapeutic approach in patients suffering from symptomatic degenerative rotator cuff tears.

Materials and methods

Piper et al. conducted a systematic review in accordance with the PRISMA statement [19]. Three RCTs with a minimum follow-up of 12 months were selected for analysis [13, 15, 20]. One study included a subgroup of patients that underwent physical therapy and subacromial decompression only, without tendon reconstruction [13]. This subgroup was excluded in Piperʼs analysis. This same study [13] displayed pain measurements as a subscale of the Constant and Murley score, which could not be accurately compared to the visual analog scale (VAS) pain score. Therefore, pain values were also not included. Since Piperʼs literature search, only one RCT has been published, namely the 10-year results of Moosmayerʼs cohort [21]. Thus, the analysis performed by Piper et al. was extended by the 10-year results.

In detail, fixed and random effect meta-analyses were performed according to the review by Piper et al. [24]. Briefly, differences in the intervention effect (surgery or physical therapy) regarding constant and pain scores were investigated. Heterogeneity was assessed by the Q test of heterogeneity and the I2 statistic. For the presentation of the meta-analyses, forest plots were generated showing the confidence intervals of the individual studies along with the pooled mean difference. Analyses were carried out using R Software for Statistical Computing (Version 4.0.3) with the add-on meta-package (Version 4.15-1) [2].

Results

Meta-analyses for the outcomes of the Constant and Murley score as well as VAS pain score were conducted. Heterogeneity was estimated to be low for both models (I2 = 0% and I2 = 7% for the Constant–Murley score and the VAS pain score, respectively). The authors therefore report their results from the fixed-effects models. This meta-analysis shows superior results for the Constant and Murley score in the operatively treated cohort (Fig. 1). The mean difference between operatively and non-operatively treated patients was 6.2 points (95% confidence interval, 2.6; 9.7, P < 0.001).

Fig. 1
figure 1

Forest plot for Constant and Murley score. SD standard deviation, MD mean difference, CI confidence interval

Full size image

VAS pain score also improved more in the operatively treated patients, with a mean difference of −1.4 compared to the non-operatively treated patients (95% confidence interval −2.1; −0.8, P < 0.001; Fig. 2).

Fig. 2
figure 2

Forest plot showing the visual analog scale. SD standard deviation, MD mean difference, CI confidence interval

Full size image

Discussion

In the previous systematic review, Piper et al. found statistically significant superiority in favor of the surgical approach with respect to Constant and Murley Score as well as the VAS pain score. However, the minimally important clinical difference (MICD) remained unsurpassed.

Kukkonen defined the threshold for rotator cuff tears to be 10.4 points in the Constant and Murley score [14]. For the VAS pain score, Tashjian [27] defined an improvement by −1.4 to be the critical difference to make the patient sufficiently “feel better” (MICD) and 3 or less to achieve a patient acceptable symptomatic state (PASS). Including the 10-year results in the analysis, the MICD of 10.4 points in the Constant score is still not met. However, the difference in the VAS pain score at 1.4 corresponds exactly with the defined threshold. In Moosmayerʼs 10-year results, the difference between the operatively treated patients (VAS mean 0.6) and the non-operatively treated cohort (VAS mean 2.3) is even greater at 1.7. The operatively treated patients in both RCTs with monitoring of the VAS score [15, 21] reported a mean VAS value below the PASS of 3 at 0.6 and 2.2, respectively. In the non-operatively treated groups, Moosmayerʼs cohort reached an acceptable pain score of 2.3 after 10 years, while Lambers Heerspinkʼs cohort was above this at 3.2 at 12-months and thus not at an acceptable pain level.

Although this review highlights the superiority of long-term clinical results after surgical rotator cuff repair, one should not underestimate the fact that a non-operative approach has the capacity to lead to long-lasting pain relief and acceptable shoulder function. Thus, decision-making between a conservative and operative approach in the situation of degenerative rotator cuff tear remains an individual and challenging situation.

The RCTs included patients with small to medium rotator cuff tears and low grades of fatty infiltration (I–II), mostly involving the supraspinatus tendon only. The reviews offer no information on treatment strategies with respect to complex, large, multi-tendon tears or tears with high degrees of fatty infiltration or atrophy. The presented results include all patients that were enrolled in the RCTs. However, it should be noted that the groups of patients in the surgical study arms represent a mixture of healed and not healed tendons. While Kukkonen et al. [12] do not report on non-healing, Lambers Heerspink [15] found an unusually high number of reruptures at 73.7% of cases at 1‑year follow-up. Moosmayer [21] et al. report a retear rate of 34% partial and full thickness tears detected by ultrasound. Both authors conducted subgroup analysis and clearly demonstrated that patients with intact tendons at the time of follow-up achieve better clinical results than those with non-healing. The group of healed tendons outweigh the overall result of the surgical group.

In an analysis of retear rates, it is necessary to take the underlying surgical method of rotator cuff repair into account. Lambers Heerspink et al. performed mini-open single row suture anchor fixation, while Moosmayer et al. used a method of suture-based transosseous mini-open repair. Arthroscopic transosseous equivalent double row repair is currently considered to offer the highest biomechanical robustness of the tendon-to-bone repair construct [1]. The literature demonstrates improvements in healing rates due to double-row techniques in comparison to single-row repair, such as applied by Lambers Heerspink [15]. Although the present review does not include subgroup calculation, the authors assume that the high rate of non-healing in one of the three RCTs leads to a bias towards under-emphasis of the clinical superiority of tendon repair in both reviews. Thus, the application of modern and durable techniques might shift the pendulum even further in favor of operative rotator cuff repair.

In summary, this review shows that surgical repair of torn rotator cuff tendons offers clinically relevant superior results regarding pain development in comparison to a conservative approach, especially in the long term. Although there is a statistically significant improvement in clinical and functional results in the operatively treated patients, the clinically relevant difference for the Constant and Murley score was not met. Further research is needed with special emphasis on integrating patient cohorts that have been treated with current standard techniques of arthroscopic double-row repair into systematic reviews of this kind.