Long-term shoulder-related secondary health-care contact among people with subacromial pain: a secondary analysis of a randomized controlled trial of home-based vs. supervised exercise regimens Authors: Nina Dyrberg Lorenzen, Cecilie Ruud Budtz, David Høyrup Christiansen Published: 2026 ABSTRACT BackgroundSubacromial pain is associated with substantial individual burden and health-care costs. Exercise rehabilitation can be delivered either at home or under physiotherapist supervision. This study investigated whether home-based exercise rehabilitation was associated with increased long-term secondary health-care utilization compared with supervised exercise rehabilitation.MethodsA total of 208 participants were randomized to home-based, group-based, or individually supervised exercise rehabilitation. For this secondary analysis, the 2 supervised groups were pooled and compared with the home-based group. The outcomes were shoulder-related secondary health-care contacts and shoulder surgery during the 5-year follow-up. Associations were analyzed using logistic regression and reported as adjusted odds ratios (ORs) with 95% confidence intervals (CIs).ResultsDuring the 5-year follow-up, 22/71 participants (31%) in the home-based group and 45/137 (33%) in the supervised group had a shoulder-related secondary health-care contact (adjusted OR 0.91; 95% CI 0.52–1.61). Shoulder surgery was recorded in 14/71 participants (20%) in the home-based group and 24/137 (18%) in the supervised group (adjusted OR 1.16; 95% CI 0.42–3.22).ConclusionIn this 5-year register-based follow-up of a randomized controlled trial, we found no evidence of increased shoulder-related secondary health-care contacts or surgery following home-based compared with supervised exercise rehabilitation for subacromial pain. However, the estimates were imprecise and should be interpreted with caution. These findings support home-based exercise as a potentially scalable rehabilitation strategy, although larger studies are needed to confirm long-term health-care utilization outcomes. Subacromial pain is common in the general population and is associated with substantial individual and societal cost.9,14 Most patients are managed in primary care using conservative approaches, including analgesics, corticosteroid injections, and physiotherapy.11,15 In more persistent cases, referral to further evaluation and treatment in specialized orthopedic clinics in hospitals may be needed. Following hospital assessment, both non-surgically and surgically treated patients are often referred to physiotherapy for shoulder-specific exercise rehabilitation to restore shoulder function.3 Physiotherapy exercise rehabilitation may effectively reduce pain and improve function.12 However, there is no consensus on the optimal content, setting, or mode of delivery. In a recent multicentre 3-arm trial, we found equal effectiveness on pain, function, and exercise adherence for exercises performed at home and more supervised exercise regimes.4 These findings suggest that the level of supervision may not influence the effect of exercise therapy and that home-based exercise rehabilitation represents a scalable alternative to more resource-intensive supervised models. However, reduced supervision could theoretically increase the risk of subsequent referral and surgery. Given the high prevalence of subacromial pain and increasing pressure on health-care resources, identifying safe and scalable rehabilitation strategies is of substantial clinical and societal importance. This study provides long-term evidence on health-care utilization associated with different exercise delivery models. Thus, this secondary analysis aimed to compare and evaluate the probability of shoulder-related secondary health-care contact or surgery over a 5-year period in patients with subacromial pain having received supervised or home-based exercise rehabilitation. Materials and methods Design, setting, participants, and intervention This is a register-based 5-year follow-up of a randomized controlled trial. Details of the study design, interventions, and primary outcomes have been reported previously.4 The study was approved by the Central Denmark Region Committees on Biomedical Research Ethics (identification number: 1-10-72-178-16) and was performed in accordance with the Helsinki Declaration. In brief, 208 patients referred to physiotherapy rehabilitation after hospital conservative or surgical treatment under the diagnosis of rotator cuff, shoulder impingement, acromioclavicular osteoarthritis (International Classification of Diseases 10th revision group M75.1-M75.8) were included. The participants were recruited between February 1, 2017, and March 1, 2018, and were randomly allocated to 3 interventions: group-based exercise rehabilitation (GE) or individual exercise rehabilitation (IE), both supervised by a physiotherapist. Finally, the third group performed the exercise program independently at home (HE). Participants included both conservatively and surgically treated individuals prior to trial inclusion. The supervised interventions were delivered by physiotherapists in municipal settings; all physiotherapists were experienced in shoulder rehabilitation and followed a standardized protocol with progression based on individual performance. The standardized exercise program included mobility, scapular control, and strengthening exercises. Each exercise had 3 progression levels. Participants advanced when able to complete 3 × 10 repetitions with good control and without exacerbation of pain. The supervised groups received guidance from physiotherapists, while the home-based group performed the same program independently. Outcomes were collected using questionnaires and register data at baseline, 3, and 6 months. For the present secondary analysis, the group-based (GE) and individually supervised (IE) exercise groups were pooled because both represented supervised physiotherapy and differed only in the mode of delivery (group versus individual supervision). The original randomized controlled trial demonstrated similar clinical outcomes, including pain, function, exercise adherence, patient satisfaction, and cost-effectiveness, between these 2 supervised interventions.4 Therefore, the pooled supervised group (GE + IE; n = 137) was compared with the home-based exercise group (HE; n = 71), in accordance with the objective of evaluating supervised versus home-based exercise rehabilitation. For this secondary analysis, register data was retrieved from CROSS-TRACKS in Central Denmark Region Midtjylland.13 This included information on diagnosis, treatment, and surgical hospital inpatient and outpatient visits from the Danish National Patient Register.10 The Danish National Patient Register contains information on all contacts to public and private hospitals in Denmark. The study is reported as recommended in the Consolidated Standards of Reporting Trials (CONSORT) statement.2 Outcomes Primary outcome The primary outcome was a shoulder-related secondary health-care contact within 5 years of inclusion, defined as a hospital inpatient or outpatient contact registered with an International Classification of Diseases 10th revision diagnosis code within the M75 group (disorders of the shoulder region). Hospital shoulder-related contact included both outpatient clinic visits and inpatient admissions within the secondary health-care system. Secondary outcome The secondary outcome was any registered shoulder surgical procedure within 5 years of inclusion, identified using the Danish NOMESCO Classification of Surgical Procedures (KNB). All KNB shoulder procedure codes were included irrespective of the accompanying diagnosis to maximize sensitivity for capturing subsequent shoulder surgery. Because registry data do not contain valid information on laterality, it was not possible to determine whether procedures were performed on the ipsilateral shoulder or to distinguish primary procedures from revision procedures. Data analysis Descriptive statistics were calculated for all variables. The sample size in the original study was estimated to be 180 participants.4 No formal sample size calculation was performed for this secondary analysis, as the sample size was determined by the original randomized trial. Analyses were performed in accordance with the intention-to-treat principle. Health-care utilization was analyzed as a binary outcome to reflect clinically relevant events rather than frequency of contacts. We used logistic regression models to compare shoulder-related hospital contacts or surgery between the 2 groups. This analysis included participants treated both conservatively and surgically prior to inclusion, reflecting clinical practice. The regression models were adjusted for study site and pretrial management. Estimates were reported as odds ratios (ORs) with 95% confidence intervals (CIs). Analyses were performed using Stata version 18 (StataCorp, College Station, TX, USA). Results Flow of participants through the study Figure 1 shows the flow of participants throughout the study, and the 2 merged groups at the 5-year follow-up. Overall, no notable differences were found between the 2 groups at baseline due to randomization (Table I).Figure 1A CONSORT flow-diagram, showing the flow of participants through the study at the 5-year follow-up. CONSORT, Consolidated Standards of Reporting Trials.Table IBaseline demographics.CharacteristicsSupervision (n = 137)Home (n = 71)Gender, n female (%)76 (56)37 (52)Age (yr), mean (SD)52 (13)54 (11)Dominant shoulder affected, n (%)89 (65)45 (62)Duration of symptoms (mo), median (IQR)∗15 (7 - 36)24 (11 - 46)Use of analgesics, n (%)† No55 (41)33 (47) Yes80 (59)37 (53)Pretrial managementConservative85 (62)43 (61)Surgical52 (38)28 (39)Education, n (%) Unskilled35 (26)23 (32) Lower level (<3 y)24 (18)7 (10) Vocational and training51 (38)24 (34) Medium or higher level (>3 y)26 (19)17 (24)SD, standard deviation; IQR, interquartile range; GE, group-based exercise rehabilitation; IE, individual exercise rehabilitation; HE, independently at home.Originally, there were 3 randomized groups (GE, IE, and HE). At the 5-y follow-up 2 groups – supervision (GE and IE) and home (HE) were created.∗Supervision 6 missing, home 1 missing.†Supervision 2 missing, home 1 missing. A CONSORT flow-diagram, showing the flow of participants through the study at the 5-year follow-up. CONSORT, Consolidated Standards of Reporting Trials. Table IBaseline demographics.CharacteristicsSupervision (n = 137)Home (n = 71)Gender, n female (%)76 (56)37 (52)Age (yr), mean (SD)52 (13)54 (11)Dominant shoulder affected, n (%)89 (65)45 (62)Duration of symptoms (mo), median (IQR)∗15 (7 - 36)24 (11 - 46)Use of analgesics, n (%)† No55 (41)33 (47) Yes80 (59)37 (53)Pretrial managementConservative85 (62)43 (61)Surgical52 (38)28 (39)Education, n (%) Unskilled35 (26)23 (32) Lower level (<3 y)24 (18)7 (10) Vocational and training51 (38)24 (34) Medium or higher level (>3 y)26 (19)17 (24)SD, standard deviation; IQR, interquartile range; GE, group-based exercise rehabilitation; IE, individual exercise rehabilitation; HE, independently at home.Originally, there were 3 randomized groups (GE, IE, and HE). At the 5-y follow-up 2 groups – supervision (GE and IE) and home (HE) were created.∗Supervision 6 missing, home 1 missing.†Supervision 2 missing, home 1 missing. Baseline demographics. SD, standard deviation; IQR, interquartile range; GE, group-based exercise rehabilitation; IE, individual exercise rehabilitation; HE, independently at home. Originally, there were 3 randomized groups (GE, IE, and HE). At the 5-y follow-up 2 groups – supervision (GE and IE) and home (HE) were created. Supervision 6 missing, home 1 missing. Supervision 2 missing, home 1 missing. Outcomes The primary outcome showed that 45/137 participants (33%) in the supervised group and 22/71 participants (31%) in the home-based group had a shoulder-related secondary health-care contact (adjusted OR 0.91, 95% CI 0.52–1.61) (Table II). Likewise, shoulder surgery occurred in 24/137 participants (18%) in the supervised group and 14/71 participants (20%) in the home-based group (adjusted OR 1.16, 95% CI 0.42–3.22) (Table II).Table IIThe probability of acquiring a shoulder related diagnosis or receiving shoulder surgery after 5 years from inclusion.OutcomesTreatment groupsSupervision (n = 137)Home-based exercise (n = 71)Primary outcome Record of shoulder diagnose, n (%)45 (33)22 (31) OR10.91 95% CI0.52-1.61Secondary outcome Record of shoulder surgery, n (%)24 (18)14 (20) OR11.16 95% CI0.42 - 3.22OR, odds ratio; CI, confidence interval.Results from the logistic regression analyses are presented.Regression analyses adjusted for site and pretrial management. Table IIThe probability of acquiring a shoulder related diagnosis or receiving shoulder surgery after 5 years from inclusion.OutcomesTreatment groupsSupervision (n = 137)Home-based exercise (n = 71)Primary outcome Record of shoulder diagnose, n (%)45 (33)22 (31) OR10.91 95% CI0.52-1.61Secondary outcome Record of shoulder surgery, n (%)24 (18)14 (20) OR11.16 95% CI0.42 - 3.22OR, odds ratio; CI, confidence interval.Results from the logistic regression analyses are presented.Regression analyses adjusted for site and pretrial management. The probability of acquiring a shoulder related diagnosis or receiving shoulder surgery after 5 years from inclusion. OR, odds ratio; CI, confidence interval. Results from the logistic regression analyses are presented. Regression analyses adjusted for site and pretrial management. Discussion In this 5-year register-based follow-up of a randomized controlled trial, we found no evidence of a difference in shoulder-related secondary health-care contacts or shoulder surgery between participants receiving home-based and supervised exercise rehabilitation. However, the estimates were imprecise, and the findings should therefore be interpreted with caution. To our knowledge, this is the first study to investigate the long-term association between the mode of exercise delivery and subsequent secondary health-care utilization in people with subacromial pain. The findings suggest that the level of supervision during exercise rehabilitation was not associated with subsequent secondary health-care utilization within the limitations of the present study. However, the relatively small number of outcome events resulted in wide CIs, limiting the precision of the estimates. The study's strengths include its prospective design, complete follow-up through register data, and random allocation to treatment groups, which reduces the risk of confounding by balancing baseline characteristics across groups. Minor differences were, however, observed in pain duration, and analgesic use. Although analyses were adjusted for site and pretrial management, the relatively small number of outcome events limited the ability to adjust for additional covariates, such as symptom duration, age and diagnoses. This represents a limitation of the study. We used data from the Danish National Patient Register, which is generally considered a valid and reliable source of information on contacts to secondary health care. However, the long follow-up period increases the likelihood that some health-care contacts may relate to new or unrelated shoulder conditions that cannot be distinguished using registry data, potentially introducing a degree of non-differential misclassification. Other diagnoses, such as DM19.0∗, might also have been relevant to include, but as this diagnosis is also relevant for other non–shoulder-related conditions, we decided not to include those. We, however, do not believe this potential information problem would differ between the groups and therefore could only bias our results toward non-differential misclassification and find it unlikely that it would change the overall conclusion of the study. Furthermore, the registry-based outcomes captured only contacts within secondary health care and therefore did not include health-care utilization in primary care, private or municipal physiotherapy, medication use, or persistent symptoms not resulting in hospital contact. Consequently, the present findings should be interpreted as reflecting long-term secondary health-care utilization rather than long-term overall health-care utilization or clinical outcomes. Similarly, the surgical outcome was based on registry-recorded shoulder procedure codes (KNB) irrespective of the accompanying diagnosis to maximize sensitivity for identifying subsequent shoulder surgery. However, registry data did not permit distinction between primary and revision procedures or confirmation of the operated side, which may have introduced limited non-differential misclassification. The absence of a sample size calculation for this secondary analysis introduces a risk of limited statistical power, which should be considered when interpreting the findings. The wide CIs indicate substantial statistical uncertainty and limit the ability to draw firm conclusions regarding differences between groups.6 Finally, the inclusion of both conservatively and surgically treated patients increases external validity but may introduce clinical heterogeneity; however, this was accounted for analytically and reflects real-world patient populations. In the primary study, home-based exercises were found to be equally effective in reducing pain and improving function. Moreover, no difference was found with respect to adherence to the exercise program and patient satisfaction. There are several studies with results that point in the same direction, and they show that conservative treatment is as effective as surgery in patients experiencing shoulder pain.1,5,7 Home-based physiotherapy is as effective as other treatments choices regarding the patient's adherence and reducing pain and disability.8 Often studies have a limited follow-up time but being aware of the high frequency of shoulder problems and the potential burden on individuals and society. The present 5-year follow-up therefore provides important information on long-term secondary health-care utilization following different exercise rehabilitation strategies. Conclusion In this 5-year register-based follow-up of a randomized controlled trial, we found no evidence that home-based exercise rehabilitation was associated with increased shoulder-related secondary health-care utilization or surgery compared with supervised exercise rehabilitation. However, the estimates were imprecise and should be interpreted with caution. These findings support home-based exercise rehabilitation as a potentially scalable treatment option, although adequately powered studies are needed to confirm long-term health-care utilization outcomes. Declaration of Generative AI and AI-Assisted Technologies in the Writing Process During the preparation of this work, the authors used ChatGPT (OpenAI, GPT-5.5) to assist with language editing, improving clarity, grammar, style, and the overall presentation of the manuscript. All AI-generated suggestions were critically reviewed, revised where appropriate, and verified by the authors. The authors take full responsibility for the content of the published article. Disclaimers: Funding: This was a researcher-initiated study, supported with funding from the Practice Research Foundation of Danish Physiotherapists and the Danish Rheumatism Association. Conflicts of interest: The author, their immediate family, and any research foundation with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.