Rehabilitation after rotator cuff repair

Multiple factors interplay to determine whether a repaired rotator cuff will heal. These include patient age, size of the tear, chronicity of the tear, the biologic milieu, tissue quality, the ease of tissue mobilization, the repair construct, and the postoperative mechanical environment. Many of these factors are nonmodifiable, and with retear rates reported at greater than 20%, ^, including more than 94% for large or massive tears, optimizing postoperative rehabilitation takes on added importance. Multiple basic science and clinical studies have analyzed the various postoperative rehabilitation protocols that give the repaired rotator cuff the best opportunity for healing.

Basic science: Tendon to bone healing

Tendon to bone healing is a delicate balance between providing enough motion to prevent stiffness and encourage collagen fiber alignment versus allowing the requisite immobilization to prevent injury and retears. A variety of animal studies have examined the impact of activity on tendon to bone healing, indicating an important role for immobilization in creating the proper environment for healing. Studies with rats with surgically repaired supraspinatus tears have compared those that were immobilized postoperatively with those that were exercised. These studies have shown superior outcomes in the immobilized group with respect to structural, compositional, and viscoelastic properties, and improved tendon to bone healing, indicating an important role for immobilization. ^, Indeed, in one study, after a 2-week period of immobilization, regular cage activity was compared with exercise. Exercise was shown to worsen tendon mechanical properties and shoulder joint mechanics compared with regular cage activity, indicating the shoulder is still at a vulnerable time of healing at this early period. Peltz et al. compared rats immobilized following supraspinatus repair with those treated with two different passive motion protocols. Although there was no impact on collagen organization or mechanical properties of the repaired rotator cuff tendon, paradoxically the rats that were immobilized were less stiff than those treated with either of the passive motion protocols. Indeed, Sarver et al. found that, in a rat model, rotator cuff repair transiently decreased range of motion (ROM) and increased joint stiffness and that the effect was exacerbated by immobilization, although it was transient. By 8 weeks, rotational stiffness was not significantly different in the immobilized group compared with the nonimmobilized group, suggesting that the rotator cuff repair itself causes stiffness and not immobilization, and the benefits of immobilization for healing outweigh and the risks of loss of motion, which appear to be temporary. In another study, Zhang et al. compared the effect of motion on rotator cuff healing. Rabbits were randomly separated into three groups following rotator cuff repair: nonimmobilization, early passive motion, and continuous immobilization. Continuous immobilization and early passive motion had similar histologic findings at the tendon-bone junction that were both superior to the nonimmobilization group. In addition, the failure load was significantly higher in the immobilized and the early passive motion groups than in the nonimmobilized group. These studies all suggest that immobilization may promote rotator cuff healing through a protective mechanical environment.

Perhaps surprisingly, complete removal of load through paralysis and immobilization is detrimental to rotator cuff healing. ^, Galatz et al. used botulinum toxin A to paralyze the supraspinatus in rats following repair, with some of these rats being casted and others being allowed free motion. Saline-injected, casted rats were used as a control following repair. The paralyzed and casted group had the worst biomechanical properties, including a decreased ultimate load to failure. Cross-sectional area and scar volume were decreased in the paralyzed groups, but material properties were not significantly different among the three groups, indicating the changes were quantitative and not qualitative in nature. In a similar study, Hettrich et al. paralyzed the supraspinatus of rats following repair and compared these with rats undergoing repair without paralysis. Early on, there was superior collagen organization of the insertion site in the paralyzed group. However, the benefits of this are outweighed by disuse atrophy of the muscle and tendon, leading to a weaker insertion site by the 8-week time point. By 24 weeks, the strength of the insertion site had equilibrated between both groups. These studies indicate that complete removal of load is detrimental to rotator cuff healing. For other tendons, a rabbit model has shown immobilization leads to decreased remodeling in patellar tendons and decreased load to failure in the repaired Achilles tendons of mice. Presumably this is due to the fact that some level of load is required as a stimulus for robust, parallel collagen fiber alignment, and development of a mechanically sound, healed rotator cuff. Thus there is a mixture of results in animal studies that both complete removal of load and excessive mobilization and tension are detrimental to tendon to bone healing.

Human clinical basic science investigations can also shed light on the topic of rotator cuff rehabilitation. Dockery et al. studied electromyograms (EMGs) of healthy participants doing a variety of passive exercises. For the supraspinatus, pulley exercises and straight-arm raise using contralateral arm assist used significantly more muscle power than a continuous passive motion (CPM) machine, Codman pendulums, and therapist-assisted motion, indicating that exercises routinely considered to be “passive” are anything but, and that the mode of early rehabilitation may have a significant effect on the early postoperative mechanical environment.

Clinical: Early versus delayed motion

Physical therapy following rotator cuff repair is believed to be important to regain strength and combat stiffness. Various postoperative protocols have been trialed in an attempt to discover the optimal regimen.

No long-term benefit has been found for CPM devices. Lastayo et al. prospectively compared outcomes in 31 patients who underwent open repair of small-, medium-, and large-sized rotator cuff tears using transosseous tunnels, based upon whether they used CPM or underwent supervised passive ROM in the first 4 weeks following surgery. They found no significant difference between the two groups in pain, validated outcome scores, ROM, or strength.

Lee et al. corroborated the lack of utility for CPM when they compared aggressive therapy (unlimited self-passive stretching exercises and manual therapy twice daily) with limited early passive motion (limited CPM exercise and limited manual therapy) following medium- and large-size rotator cuff arthroscopic single-row repair in 64 shoulders and found that by the 3-month postoperative mark there were no significant differences in recovery of ROM, whereas the aggressive motion group had a higher (23.3% vs. 8.8%) but statistically nonsignificant amount of repair failures. Both groups had similar pain, muscle strength, and functional scores.

In another study evaluating the results of two postarthroscopic rotator cuff repair rehabilitation protocols, Raab et al. found increased ROM and decreased pain in female patients and patients older than 60 years who used physical therapy and CPM compared with those who used physical therapy alone. Their study population included 26 patients with a variety of tear sizes treated with a range of repair techniques. However, follow-up was only for 3 months, and other studies have shown a convergence between early and delayed ROM groups after this time point.

Garofalo et al. studied 100 patients after arthroscopic rotator cuff repair who were randomized to either passive self-assisted ROM or passive self-assisted ROM with the use of CPM for 2 hours a day for 4 weeks. They found some early utility in ROM improvement and pain relief in a group of patients who used CPM. However, all of these advantages were lost by the 1-year follow-up because the outcomes of the two groups was statistically equivalent.

Cuff and Pupello compared 68 patients who were randomized to early passive external rotation and elevation following arthroscopic suture-bridge repair of full-thickness supraspinatus on postoperative day 2 with those who began the same protocol at week 6. Although the early motion group regained ROM more quickly, there was no difference at 1 year. Both groups achieved similar functional, pain, ROM, and satisfaction scores, and the delayed motion group had a nonsignificant amount of higher rotator cuff healing.

Similar findings were discovered by Kim et al., who compared 105 patients with arthroscopic repair of small- to medium-sized full-thickness rotator cuff tears who were randomized to either an early passive motion group (starting postoperative day 1) versus a delayed motion group that started passive motion at 4 weeks for small tears or 5 weeks for medium-sized tears. After 1 year, functional scores, ROM, and rotator cuff healing rates were not significantly different between the two groups.

Keener et al. also compared outcomes based on a traditional rehabilitation of early mobilization or immobilization in which ROM was delayed to 6 weeks. They randomized 124 patients who were treated with double-row repairs for small- and medium-sized full-thickness rotator cuff tears to immediate pendulums and passive ROM to start 1 week after surgery and a group that delayed motion to 6 weeks. Much like prior studies, active elevation and external rotation were improved in the early motion group at early follow-up, yet these differences equilibrated by 6 months’ follow-up, as did functional scores and shoulder strength. There were no differences in retear rates between the two protocols.

Similarly, Mazzocca and colleagues randomized 73 patients undergoing arthroscopic double-row repair for single-tendon tears to a group that underwent active-assisted ROM on postoperative day 2 to 3, to those who were held from this motion until postoperative day 28 and followed them for 1 year. At 6 months, there were no significant difference in Western Ontario Rotator Cuff scores, although the early motion group maintained better scores throughout the postoperative period, with both groups exhibiting a similar trajectory of improvement. The early motion group had a nonsignificant amount of retears (31%) compared with the delayed motion group (34%).

Koh et al. compared outcomes of 100 patients who underwent arthroscopic single-row repair of posterosuperior rotator cuff tears based on length of immobilization, randomizing patients to either 4 or 8 weeks of immobilization following rotator cuff repair. Both groups did well, with similar retear rates, functional scores, and ROM. A higher proportion of patients in the 8-week group were considered stiff, as defined by forward elevation less than 120 degrees, internal rotation less than L3, and external rotation less than 20 degrees with the arm at the side.

Indeed, Arndt et al. randomized 100 patients who underwent arthroscopic repair of full- and partial-thickness supraspinatus tears and found better functional results and ROM in their early mobilization group—those who had immediate passive ROM, CPM use, and pendulums—compared with their strict immobilization group, which was allowed only pendulums, with a nonsignificant amount of better healing in the immobilization group.

Düzgün et al. compared 29 patients who underwent arthroscopic repair for medium and large rotator cuff tears who were randomized to an accelerated protocol and likewise found better results in an their accelerated rehabilitation group, which included a preoperative rehabilitation program for 4 to 6 weeks and active ROM at 3 weeks, compared with the slow protocol, which involved active ROM delayed to 6 weeks. They found the accelerated protocol was associated with less pain with activity and improved functional activity level following rotator cuff repair, but by the sixth month the two groups were statistically equivalent.

The case for delaying motion was made by Parsons et al., who immobilized 43 patients with the same postoperative protocol after full-thickness rotator cuff repair of full-time immobilization for 6 weeks. They then reviewed the results of the “stiff” group versus the “nonstiff” group. Although 23% were considered clinically stiff, as defined by less than 100-degree forward flexion and less than 30 degrees of external rotation passively at the 6-week mark, by 1 year there was no difference in the final ROM regained or functional scores between the stiff and nonstiff group. Interestingly, there was a nonsignificant ( P = .079) trend toward fewer retears in the stiff group, indicating that perhaps the same mechanical environment or biologic propensity toward collagen formation may cause both stiffness and promote cuff healing.

These results were challenged by Tirefort et al. in a study that randomized 80 patients to sling or no sling following double-row arthroscopic repair of small- and medium-sized rotator cuff tears for 4 weeks. Both groups engaged in passive ROM and were restricted from active ROM during this period and were followed for 6 months total. At 6 months, the no-sling group had greater ROM and functional scores, and there was no significant difference in retears between the two groups. Although these findings are statistically significant, they may not be clinically significant. For example, the no-sling group had a 6-month Single Assessment Numeric Evaluation score of 85.8, while the sling group had a Single Assessment Numeric Evaluation score of 79.4. In addition, as noted previously, many studies have shown a convergence of gains from early mobilization after 6 months.

Similarly, Sheps et al. also performed a randomized controlled trial in which 206 patients with a variety of rotator cuff tear sizes were allocated to either a standard rehabilitation group (instructed to wear a sling at all times for 6 weeks except during passive or self-assisted exercises) or to the early mobilization group (instructed to wear a sling for comfort only and could perform pain-free activities with the exception of resisted activities). At 6 weeks, the early mobilization group had better ROM, but the groups equalized over 24 months. Pain, strength, quality of life, and retear rates were not different between the two groups.

A number of systematic reviews exist that summarize and analyze the effect of early motion on rotator cuff repair. The consensus between the studies is a trend toward better early outcomes in the early mobilization groups with a difference that is not sustained with longer follow-up. However, it should be noted that regarding the question of whether early mobilization leads to a higher retear rate remains controversial.

Basic science: Arm position

Proper positioning of the arm following surgery is crucial to the success of rotator cuff repair. This can be achieved by a variety of immobilization methods. The position of the arm has implications for strain on the repaired rotator cuff. Hatakeyama et al. performed a cadaver study in which strain on the repaired rotator cuff was studied in a variety of positions; they found the optimal position with the least amount of strain to be above 30 degrees elevation in the coronal or scapular plane and between 0 and 60 degrees of external rotation. This finding was corroborated by Jackson et al. in a computer simulation of a musculoskeletal model. They demonstrated an optimal position of elevation and external rotation to decrease stresses on the repaired tendons. Howe et al. performed a cadaver study and found that rotation caused substantial tension imbalances, with external rotation causing increased tension in the anterior suture and internal rotation causing increased tension in the posterior suture that was independent of the amount of abduction.

The role of abduction on decreasing tissue tension was studied intraoperatively by Reilly and colleagues. They studied tendon tension during rotator cuff repair surgery as the supraspinatus was advanced into the bone trough and found, unsurprisingly, that tension increased as the tendon advanced into the trough. Moving the arm from 0 to 30 degrees of abduction resulted in a mean decrease of 34 N of passive tension. This 34 N load was statically applied over 24 hours to cadaveric rotator cuff repairs and resulted in a gap formation of 9 mm, indicating the significance of the difference in joint positioning. Rathbun and Macnab found a unique avascular zone in the supraspinatus of cadavers near its insertion, which they postulated was from compression forces on the vascular supply by the arm being held in a resting anatomic position of adduction, because this avascular zone resolved when the shoulder was injected in abduction. This study indicates that holding the shoulder in abduction following rotator cuff repair may be better for the blood supply to the supraspinatus and thus promote tendon to bone healing. Hawthorne and colleagues compared three types of slings with increasing amount of abduction in an imaging and cadaver study. They found, not surprisingly, that increasing the size of the abduction pillows increased the abduction angle as measured by radiographs. Furthermore, increasing the abduction significantly reduced the tension on repaired supraspinatus tendons in cadavers. An additional cadaveric study by Andarawis-Puri et al. studied principal strain of the infraspinatus and supraspinatus at 0, 30, and 60 degrees of glenohumeral abduction. They demonstrated that the interaction between the supraspinatus and infraspinatus, which may have a stress-shielding effect, decreased at increased abduction angles. They found that 30 degrees of abduction to be the angle at which both supraspinatus and infraspinatus tendon strain are decreased, indicating that, although some amount of abduction may be helpful in decreasing strain, too much abduction can be detrimental. In an ovine model, increasing the abduction after rotator cuff repair decreased the contact pressure for a variety of repair techniques. Hersche and Gerber obtained data on the intraoperative passive tension of the supraspinatus musculotendinous unit at the time of repair and found that passive tension increases with long-standing rupture. Abduction decreases the passive tension compared with adduction, with a fourfold difference in longstanding ruptures and a twofold difference for the control of a partial-thickness tear that was resected to a complete tear at the time of surgery.

Not all studies are in agreement about the impact of abduction on tension on the rotator cuff. Bey and colleagues performed an imaging-based cadaver study that measured strain in intact supraspinatus tendons across varying degrees of abduction by examining texture correlation on the imaging. They found that intratendinous strain increased with increasing glenohumeral abduction joint angles. Footprint contact is another important concern. Park et al. performed a cadaver study of various types of rotator cuff repair techniques and found that contact area decreased as abduction increased, whereas contact area increased at the extremes of rotation, supporting immobilization in internal rotation within the lower range of abduction. Thus the position in which the arm is placed after surgery (in a sling with or without an abduction pillow) can play an important role in the forces and contact that occurs at the repair site. It must be noted that devices with an increasing amount of abduction and/or external rotation put the limb farther from the torso, and this increased lever arm may require more shoulder muscle activation to stabilize these on the chest during movement. Cadaver studies cannot assess this.

Clinical: Arm position

Although it is difficult in cadaver studies to determine the impact of active muscle activated in various arm positions, clinical studies have investigated the impact of arm position on outcomes. Clinically, arm position was studied by Hollman et al., who randomized patients to antirotation slings (ProCare Shoulder Sling; DJO Global) in neutral adduction and internal rotation or abduction brace (Össur Smartsling) in 30 to 40 degrees of abduction and neutral rotation following arthroscopic rotator cuff repair. Both groups did well, and there was no significant difference in pain, function, range motion, patient satisfaction, or quality of life. This finding was replicated by Ghandour et al., who found no difference in patient-report outcomes, postoperative pain, or isokinetic strength between patients immobilized in a pouch arm sling—an antirotation brace—and those immobilized in an abduction brace. Jenssen and colleagues randomized 120 patients with small- or medium-sized rotator cuff tears that were treated arthroscopically with a single-row repair to a simple sling for 3 weeks, after which they began active ROM or to a brace with a small abduction pillow for 6 weeks, after which point they began active ROM. They found the simple sling group to be not inferior to the abduction brace as measured by the Western Ontario Rotator Cuff index. There were no significant differences in radiographic healing between the two groups. In contrast, Conti et al. compared bracing in 15 degrees of internal rotation and 15 degrees of external rotation and found benefits in pain and passive ROM that extended to 6 months for patients immobilized in external rotation. McColl and colleagues studied repair integrity in 1600 consecutive arthroscopic rotator cuff repair cases. Over time, the retear rate decreased. Although it is not possible to attribute the decrease to any one factor, it occurred concurrently with the implementation of a less aggressive physical therapy regimen, which included more emphasis on the use of a postoperative abduction sling. Thus the impact of the position of immobilization on clinical outcomes is an area that requires more research. Furthermore, it should be noted that increasing amounts of abduction and external rotation put the hand farther from a functional position in front of the body. For this reason, patients may tolerate these devices only to an extent.

Physical therapy modalities

Cryotherapy is a common practice following rotator cuff surgery. Speer et al. studied the efficacy of cryotherapy by randomizing 50 consecutive patients who were admitted to the hospital following open surgery of either shoulder arthroplasty, rotator cuff repair, or anterior shoulder stabilization. They were randomized either to receive cryotherapy or no cryotherapy via Cyro/Cuff (AirCast), and their discomfort was measured through postoperative day 10. They found that patients who received cryotherapy on the night following surgery reported fewer and less severe pain episodes, slept better, and reported a lower use of narcotic pain medications. In addition, by postoperative day 10, they reported less frequent and less severe pain, with less swelling, and less shoulder pain with movement—potentially facilitating their rehabilitation. Some of the same authors followed this study up with a prospectively randomized study of 70 patients evaluating a variety of open (rotator cuff repair, anterior shoulder stabilization, biceps tenodesis) and arthroscopic procedures (subacromial decompression, biceps tenotomy, capsulorrhaphy, and labral repair) carried out to a longer follow-up period. Cryotherapy was effective in reducing the frequency and severity of pain and facilitating rehabilitation through day 21, with the impact being larger in arthroscopic cases than in open cases.

Although it appears that cryotherapy is effective in managing pain, the need to spend a great deal of money on specialized devices was challenged in a study by Kraeutler et al. In this study, 25 patients were randomized to cryotherapy using the Game Ready (CoolSystems Inc.) commercial device to 21 patients who were randomized to cryotherapy using a Ziploc bag full of ice and an Ace wrap to secure it in place for the first 7 postoperative days following arthroscopic rotator cuff repair or subacromial decompression. There was no significant difference at any point in pain scores between the two groups, including the day and night of surgery and average or worst pain scores. In fact, patients using the commercial device, which cost the patient $190 to rent for 9 days at the institution where the study was performed, actually used higher morphine-equivalent doses of narcotics for the first week after surgery. This was a nonsignificant finding ( P = .28), although in days 5 to 7, it was found to be clinically significant.

Formal outpatient physical therapy versus self-directed programs

In addition to the modalities offered by physical therapy, the importance of supervision has also been the subject of study. Roddey et al. compared 1-year self-reported functional outcomes in patients who received formal physical therapy and those who received video instructions on how to do exercises and were assessed in person only for readiness to progress to the next level of the protocol; they found no differences in outcome scores or compliance between the two groups. Similarly, Hayes et al. compared the efficacy of formal, individualized supervised physical therapy following rotator cuff repair with an unsupervised home regime. Muscle force, ROM, and functional outcomes were similar between the two groups. Lisiński et al. compared an unsupervised program of simple exercises to a formal physical therapy group that was treated with advanced modalities including myofascial release, proprioceptive neuromuscular facilitation, isotonic stretching, and rhythmic stabilization following rotator cuff repair and found the physical therapy group had superior outcomes in terms of ROM, pain, and the activity of muscle motor units at rest and with activity as tested by EMG and electroneurography. Chou et al. also compared a group that did primarily home therapy with in-person formal assessments at 2, 6, and 12 weeks with a group that did formal physical therapy that was significantly different than the home therapy group. The formal physical therapy group performed pendulums using an accelerometer-based device under physical therapist supervision to ensure it was safe and delayed the start of the home therapy regimen to 5 weeks, while moving all therapy to aquatic therapy at 6 weeks. By 12 weeks, there were no significant differences in functional outcomes. However, there was a significantly higher rate of retear in the home therapy group. As Dickinson et al. astutely point out in their systemic review, it is hard to draw conclusions from the disparate findings that emerge from these studies. With small sample sizes and dramatically different protocols used in between home therapy and formal physical therapy in each study, the study designs simply do not allow firm conclusions to be drawn about the role and necessity of supervision in physical therapy following rotator cuff repair.

Authors’ preferred protocol

The vast majority of postoperative protocols are divided into phases, starting with immobilization and followed by progressive passive ROM, active ROM, and strengthening, and are customized to the patients’ pathology and goals. We recommend that patients attend physical therapy within 1 week of surgery. Our preferred protocol starts very slowly, but we do find early involvement of a skilled physical therapist is helpful to remind the patient what not to do during the critical early recovery phase. During phase 1 of the protocol, patients are placed in a sling with a 15-degree abduction pillow (Slingshot; Breg) at all times except for hygiene and for exercises. The repair is protected from undesirable loads during the short time the sling is removed by giving patients very specific instructions on how to perform activities of daily living. We demonstrate and explain how to gently let the arm extend and provide specific instructions not to abduct, but to lean over and forward to, for example, accomplish axillary hygiene or when getting dressed by sliding a shirt sleeve on over the affected limb first. No weight bearing or active ROM is allowed through the operative extremity, although finger, wrist, and elbow ROM is encouraged to prevent swelling and stiffness if no concomitant procedure such as a biceps tenodesis precludes this. Pendulums supported by the nonoperative arm are allowed right away. For small and medium tears, unsupported pendulums are allowed at 3 weeks, whereas for large and massive tears, supported pendulums are maintained for the entire 6 weeks of phase 1 ( Fig. 55.1 ). Passive ROM is begun at 2 weeks for small- and medium-sized tears and 4 weeks for large and massive tears. This allows a bridge of type III collagen to bone to form, eventually to be replaced by type I collagen, which studies have stated can take anywhere from 12 to 26 weeks to reach final tensile strength. Passive motion is allowed within defined ROM limits with limited repetitions to prevent cyclic loading. Weaning from the sling is begun at approximately 4 to 6 weeks—4 weeks for partial-thickness, small-, and medium-sized tears and 6 weeks for large and massive tears. When the patient is off narcotics, is out of the sling, and can operate a vehicle safely, he or she is cleared to drive. Between 3 and 6 weeks, depending on the size of the tear, aquatic therapy is allowed. EMG testing of shoulder muscle activation is significantly less in water than on land, indicating aquatic therapy is safer earlier than land therapy. If the rotator cuff tear involves only the subscapularis, then for the first 3 weeks, the passive motion is limited between 90 degrees of forward flexion and 0 degrees of external rotation. Between 3 and 6 weeks, external rotation may progress to 30 degrees of passive rotation. If a concomitant biceps tenodesis was performed, active elbow flexion is discouraged for 6 weeks.