1 Introduction

To restore shoulder function after a tendon tear, rotator cuff repair is typically performed in as many as 30% of the United States population older than 60 years and about 60% of the population older than 80 years [1, 2]. This common surgery incurs a cost greater than $7 billion annually in the US alone [1-3]. Moreover, the post-surgery period involves a lengthy rehabilitation program to completely restore shoulder function [4], but the most effective time to start physical therapy (PT) is unclear [4-7]. Lack of sufficient shoulder function can lead to impaired shoulder use for daily activities.

Impaired shoulder use can lead to a tendon retear in 20%–60% of individuals [1-3] and require a revision rotator cuff surgery [8]. Rotator cuff tendon retears usually occur within the first 19 weeks after surgery [3, 8-10], which is the period that frequently coincides with patients’ PT. However, there are limited evidence-based guidelines for when to initiate PT post-surgery or protocols for specific treatment modalities to protect the healing shoulder tendons [3, 6, 7, 11].

Several clinical factors related to the patient, surgery, and the specific PT treatments can contribute to a rotator cuff repair success or failure. Some patient-related factors include a patient's physical function such as presurgery shoulder motion [12, 13], muscle quality [14], and preoperative fatty tissue presence [8, 12], and smoking [15, 16]. Patient co-morbidities such as diabetes [17, 18] or osteoporosis [19, 20] may negatively impact rotator cuff repair outcomes. Other patient-related factors could be social determinants of health (e.g., payment source [21] and clinic access). Additionally, surgery-related factors can include complicated repair methods [8, 9, 12] and the need for major reconstructive procedures [12]. Specific exercises and timing of the initiation of shoulder motion [11, 22, 23] may play a role, which is important since these patients spend months in PT as compared to 1 day of surgery. However, scientific evidence is inconclusive about the effectiveness of conservative or accelerated rehabilitative care [4, 11, 24].

To improve patients’ clinical outcomes and to efficiently use rehabilitative care after rotator cuff repair, clinical factors and protocols should be investigated. However, there is a paucity of evidence about the most effective PT protocol after rotator cuff repair [5, 11]. We investigated the optimal time to initiate PT after rotator cuff repair surgery and whether the timing influenced patients’ shoulder function. It was hypothesized that PT should start within 0 to 4 weeks post-surgery to gain the most effective shoulder function.

2 Materials and Methods

A retrospective observational analysis was completed with data collected from Focus on Therapeutic Outcomes Inc. (FOTO) and patient electronic health records. Data was provided by a national outpatient PT clinic group of 52 clinics. The study was approved by the lead author's university Institutional Review Board for ethics and given exempt status. This study conforms to all STROBE guidelines and reports the required information accordingly. For this study, 1111 electronic medical records were extracted of patients who had rotator cuff repair.

The study included patients who had rotator cuff repair surgery and completed a series of PT sessions postoperatively. A series of PT included a case with an intake FOTO Upper Extremity Functional Status (FS) score, at least one follow-up FS score, and a closed electronic chart. Patients who started PT later than 98 days were excluded because rehabilitation is typically started before 98 days post-surgery. This exclusion criteria reduced the data set to 1002 patients. Incomplete data sets of the primary variable reduced the data to a range of 768–790. The data set included patients who had rotator cuff surgery between October 2016 to June 2021.

The primary variables included: (1) timings between surgery and the start of PT, (2) the change in functional shoulder scores [25] (final score minus baseline score), (3) the number of PT visits completed, and (4) the change in functional scores per visit. To understand the change in functional scores per visit, the following calculation was utilized: final shoulder functional score divided by the total number of PT visits. Additionally, timings between surgery and the start of PT were categorized in (0–2 weeks [0–14 days], 2–4 weeks [15–28 days], 4–6 weeks [29–42 days], and 6–14 weeks [43–100 days]) to provide a specific guideline for starting PT. Timing between surgery and the start of PT was categorized according to the typical weekly timeframes for PT progression [8, 10].

Confounding (independent) variables were collected for each patient: demographics (age, sex), payor source, and tear size. There were a variety of 12 payor sources and only the primary insurance of each patient was provided in the data set. The payor sources were organized into four categories: Workman's Compensation, indemnity, or automobile insurance, government insurance (Medicare or Medicaid), other insurance or patient pays, and commercial insurance. Tear size was collected from the electronic chart by research clinicians. Tear size was organized into the three categories: tendon tear < 3 cm, tendon tear > 3 cm or unknown tear size. Tear sizes < 3 cm are considered small to medium, and tear sizes > 3 cm are considered large [26-29]. Tear sizes were categorized as stated because patients with larger tendon tears [14] or major reconstructive repairs [12] are not expected to have the best postsurgical outcomes. The number of different surgeons and therapists involved in each patients’ care was unknown. The patients’ cigarette uses and co-morbidities such as diabetes and osteoporosis were unavailable to the authors.

To determine shoulder function, the FOTO Upper Extremity Functional Status (FS) test was utilized [30]. This 53-item list is a computerized adaptive test that provides an understanding of a patient's perceived physical abilities [25]. Patients self-report their abilities on a scale of 0 (low functioning) to 100 (high functioning) [25]. The test has an internal consistency reliability of α = 0.99 [25]. The test includes five stages which indicate a patient's level of disability: “stage 1 (scores 0–24) has exceedingly limited shoulder function, stage 2 (scores 25–43) has poor shoulder function, stage 3 (scores 44–59) has fair shoulder function, stage 4 (scores 60–80) has good shoulder function, and stage 5 (scores 81–100) has excellent shoulder function” [30]. Examples of shoulder function include the ability to dress or bathe, to perform household chores, to perform work or sports, and so forth.

All statistical analyses were performed using the Statistical Package for the Social Sciences software (SPSS) version 27.0 (IBM Corp). The pre-determined level of significance was p < 0.05. Data analysis investigation included correlations and differences among shoulder function scores, the number of PT visits completed, and the timing to initiation PT post-surgery with a 95% confidence interval for means. Confounding (independent) variables were investigated as well (age, sex, payor source, and tear size). Levene's tests for homogeneity of variances were not significant for the following variables among the four PT start times: (1) change in functional score (F [3,786] = 0.347, p = 0.791); (2) the number of PT visits completed (F [3,764] = 0.7, p = 0.533); and (3) the change in functional scores per visit (F [3,639] = 0.994, p = 0.395). Therefore, parametric analyses were used. Baseline functional scores were controlled in the analysis.

An ANOVA test was completed to compare differences in the timing categories (0–2 weeks [0–14 days], 2–4 weeks [15–28 days], 4–6 weeks [29–42 days], and 6–14 weeks [43–100 days]). A linear regression analysis was used to understand the influence of age, sex, payor source, and tear size had on change in functional shoulder scores, number of PT visits utilized, and functional scores changed per visit.

3 Results

Descriptive statistics are available in Table 1 for three primary variables of shoulder functional scores (baseline, final, and change in score), the number of PT visits used, functional shoulder score change per PT visit (final score divided by visit), and secondary variables. The number of patients for each rehabilitation phases initially was 1002 patients, however, missing data for the primary variables reduced the data analyzed for each primary variable. Completed data set numbers were: 790 patients for the change in shoulder functional scores, 768 patients for the number of PT visits used, 643 patients for the functional shoulder score change per PT visit.

Primary analyses included a one-way ANOVA and baseline functional shoulder scores were controlled. There were no significant differences among the PT start times for the change in functional scores (p = 0.302) and change per visit scores (p = 0.910) when baseline functional scores were controlled. See Table 2 for the detailed results. Baseline scores were controlled because those outcomes were significantly different (F = 9.54, p < 0.001) among the PT start times. However, final functional scores were not significant different (F = 1.335, p = 0.262) among PT start times. The number of PT visits utilized between the four timings was statistically significant (p < 0.001) between 0 and 2 weeks and 2–4 weeks, and 4–6 weeks, and 6–14 weeks (Table 3). The following tables provide details about the test statistic, effect size, and mean difference.

Regression analyses of the following independent variables were completed: sex, age, payor source, and tear size (Table 4). Sex did not significantly impact the change in functional shoulder scores, the number of PT visits, and the timing of PT post-surgery, and functional shoulder score change per PT visit. Age did not impact the change in functional shoulder scores (p = 0.275), and the number of PT visits post-surgery (p = 0.292), and functional shoulder score change per PT visit (p = 0.200). However, age may have influenced the timing of PT post-surgery (p = 0.007). Specifically, time to initiation of PT was higher by 0.008 weeks for a 1-year age difference. However, this effect only explained a small amount of the variance in time to initiation of PT (R² = 0.010, p < 0.007). Payor source did not affect the change in functional shoulder scores (p = 0.427). However, payor source may have impacted the number of visits (p < 0.001), the timing of PT post-surgery (p = 0.001), and the functional shoulder score change per PT visit (p = 0.003). Similar to the effect of age on initiation of PT, payor source explained only a small amount of the variance in PT visits utilized (R² = 0.060, p < 0.001), the timing of PT (R² = 0.016, p = 0.001), and the functional shoulder score change per PT visit (R² = 0.021, p = 0.003). Tear size did not impact the change in functional shoulder scores (p = 0.812), the number of PT visits (p = 0.520), and the functional shoulder score change per PT visit (p = 0.196). However, there was an effect on the timing of PT (p = 0.026). Tear size explained only a small amount of the variance in the timing of PT (R² = 0.007, p < 0.026). The numbers reduced from initial collection to analyses due to missing data. There were 12 different payor sources for the data set but reduced to four categories as mentioned in Section 2.

4 Discussion

When compared to other timings, the 0–4 weeks’ timing category revealed the highest significant functional shoulder score change per PT visit. The change in functional scores of the 0–2 weeks’ timing category were 12–20 points higher than the scores of 4–6 week and 6–14 week timing categories. The change in functional scores of the 2–4 weeks’ timing category were 11 points higher than the scores of 6–14 week timing category. These scores exceeded the minimal clinically important improvement score for the shoulder, which is 8 [30, 31]. This result indicates that starting PT early leads to the best shoulder function improvement. Similar findings are reported by other researchers, who found that early range of motion (ROM) therapy provided accelerated recovery of postoperative stiffness with small to medium rotator cuff tears [11].

The number of PT visits completed pertaining to each of the four timing categories was significant when the early timing categories were compared to the later timing categories. Patients had an average of 24 visits among the four PT start times. The outcomes showed a small effect size; therefore, the results do not clearly imply that starting PT early will cause a patient to use more PT after rotator cuff surgery. These results suggest that patients will use a similar number of visits whether PT is started early or later post-surgery.

Sex and age had little influence on the primary outcomes. Sex did not significantly impact the change in functional shoulder scores, the number of PT visits, and the timing of PT post-surgery, and functional shoulder score change per PT visit. These results indicate that sex [32] did not play a role in either the patients’ functional recovery or their ability to utilize PT. This outcome contrasts with what has been reported in previous literature [8, 12]. Post-surgery, age did not significantly impact the change in functional shoulder scores, the number of PT visits, and functional shoulder score change per PT visit. However, age may have mildly influenced the timing of PT post-surgery. Younger patients participated in PT earlier in this study, but this result had a small effect size.

Previous literature has generally demonstrated that increased age [12, 33], sex (males) [12], and larger tear sizes [8, 33, 34] lead to repeat rotator cuff tendon tears of the same arm. Moreover, surgeons often refer older patients to earlier rehabilitation to prevent loss of motion, but delay rehabilitation referral in younger patients because of re-tear concerns. This study revealed that when younger patients started PT sooner, they appeared to achieve better shoulder function. This finding may suggest that clinicians should consider holistically whether younger patients should start PT sooner. Additionally, tear size was not a predictor of shoulder function in our population. However, tear size demonstrated a small effect on the timing of PT. This finding is in line with previous reports that a larger cuff tear could cause a higher retear rate [11]. Our assessment of tear size may have been limited by our dichotomized variable of greater than or less than 3 cm as opposed to more specific categories of tear sizes.

Payor source may have influenced the number of PT visits, the timing of PT, and the functional shoulder score change per PT visit. As the effect size was small, this result is likely of minimal clinical importance [31, 35]. There were 12 primary payor sources (Medicare to Commercial Insurances), and this variance does not indicate a specific insurance impacted the timing of PT post-surgery.

The study outcomes cannot reflect all factors that can influence shoulder function post rotator cuff surgery and timing to initiation of PT may not be the only factor. All possible variables were not available to the authors for this study's data set. Patient social situations could impact some patients’ ability to attend PT promptly after surgery (e.g., needing to be a caregiver, other work commitments, and/or the patient not being able to afford transportation to their clinic of choice). Co-morbidities or other patient lifestyle factors (e.g., osteoporosis [19, 20], diabetes [17, 18], and nicotine use [15, 16]) could have been linked as possible factors limiting these patients’ shoulder function and not the timing of PT post-surgery. It is possible there was sample selection bias [36, 37] because one PT organization owned the studied 52 outpatient clinics which could be a geographical limitation. Although the diversity of care from multiple licensed PT providers may reflect the diversity of care provided across the United States (US). The licensed PTs likely adjusted care as needed depending on the patient's needs which is an uncontrolled variable, but a diversity with care across the United States. Unfortunately, there are many factors, and the relative influence of such factors is unknown because very few studies describe the intersection of how these factors and especially rehabilitation factors predict shoulder function. Even with these limitations, the large sample size demonstrates rehabilitation still may play an important role since the recovery time from a rotator cuff repair includes a lengthy PT regime and patients typically retear during the rehabilitation period [8, 10].

5 Conclusion

The timing of post-rotator cuff surgery PT may not affect functional shoulder score outcomes. The study results may help surgeons and physical therapists consider rehabilitation to improve shoulder function, but PT timing may not be a limiting factor on function post-surgery. Not obtaining adequate shoulder function could lead to retear and the need for a revision rotator cuff repair. Future studies will consider the intersection between PT timing and co-morbidities associated with patients who have repeat rotator cuff tears. In conclusion, this study provided healthcare providers a guideline to not consider the timing to initiate PT post-surgery to gain adequate patient shoulder function.

Author Contributions

Berrios Barillas analyzed the data and prepared the first draft of the manuscript, participated in the conception and design of the study, participated in data organization and supervised the study throughout. Both authors constructively revised the manuscript and share authorship. All authors commented on previous versions of the manuscript and approved the final version.

Ethics Statement

The study protocol was approved by the Ethics Committee of the University of Pittsburgh (STUDY21010199), and it was compliant with the Helsinki Declaration of 1975, as revised in 2008.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.