Australian Journal Of Physiotherapy

164 Holland: Physiotherapy management of interstitial lung disease 18. Wallaert B, Monge E, Le Rouzic O, Wemeau-Stervinou L, Salleron J, Grosbois JM. Thoracic Society Clinical Practice Guideline. Am J Respir Crit Care Med. Physical activity in daily life of patients with fibrotic idiopathic interstitial pneu- 2020;202:e121–e141. monia. Chest. 2013;144:1652–1658. 37. Visca D, Montgomery A, de Lauretis A, Sestini P, Soteriou H, Maher TM, et al. Ambulatory oxygen in interstitial lung disease. Eur Respir J. 2011;38:987–990. 19. Bahmer T, Kirsten AM, Waschki B, Rabe KF, Magnussen H, Kirsten D, et al. Clinical 38. Khor YH, Hazard A, Symons K, Westall G, Glaspole I, Goh N, et al. Portable Oxygen Correlates of Reduced Physical Activity in Idiopathic Pulmonary Fibrosis. Respira- Concentrators versus Oxygen Cylinder during Walking in Interstitial Lung Disease: tion. 2016;91:497–502. A Randomized Crossover Trial. Respirology. 2016;22:1598–1603. 39. Khor YH, Goh NSL, McDonald CF, Holland AE. Oxygen Therapy for Interstitial Lung 20. Raghu G, Collard HR, Egan JJ, Martinez FJ, Behr J, Brown KK, et al. An Official ATS/ Disease: A Mismatch Between Patient Expectations and Experiences. Ann Am ERS/JRS/ALAT Statement: Idiopathic Pulmonary Fibrosis: Evidence-based Guide- Thorac Soc. 2017;14:888–895. lines for Diagnosis and Management. Am J Respir Crit Care Med. 2011;183:788–824. 40. Lee JYT, Tikellis G, Corte TJ, Goh NS, Keir GJ, Spencer L, et al. The supportive care needs of people living with pulmonary fibrosis and their caregivers: a systematic 21. Miyashita K, Kono M, Saito G, Koyanagi Y, Tsutsumi A, Kobayashi T, et al. Prognosis review. Eur Respir Rev. 2020;29. after acute exacerbation in patients with interstitial lung disease other than idio- 41. Lee JYT, Tikellis G, Glaspole I, Khor YH, Symons K, Holland AE. Self-management for pathic pulmonary fibrosis. Clin Respir J. 2021;15:336–344. pulmonary fibrosis: Insights from people living with the disease and healthcare professionals. Patient Educ Couns. 2022;105:956–964. 22. King TE, Jr., Bradford WZ, Castro-Bernardini S, Fagan EA, Glaspole I, Glassberg MK, 42. Tikellis G, Lee JYT, Corte TJ, Maloney J, Bartlett M, Crawford T, et al. Peer Connect et al. A phase 3 trial of pirfenidone in patients with idiopathic pulmonary fibrosis. Service for people with pulmonary fibrosis in Australia: Participants’ experiences N Engl J Med. 2014;370:2083–2092. and process evaluation. Respirology. 2020;25:1053–1059. 43. Kreuter M, Bendstrup E, Russell AM, Bajwah S, Lindell K, Adir Y, et al. Palliative care 23. Richeldi L, du Bois RM, Raghu G, Azuma A, Brown KK, Costabel U, et al. Efficacy and in interstitial lung disease: living well. Lancet Respir Med. 2017;5:968–980. safety of nintedanib in idiopathic pulmonary fibrosis. N Engl J Med. 2014;370:2071– 44. Higginson IJ, Bausewein C, Reilly CC, Gao W, Gysels M, Dzingina M, et al. An in- 2082. tegrated palliative and respiratory care service for patients with advanced disease and refractory breathlessness: a randomised controlled trial. Lancet Respir Med. 24. Dowman L, Hill CJ, May A, Holland AE. Pulmonary rehabilitation for interstitial lung 2014;2:979–987. disease. Cochrane Database Syst Rev. 2021;2:CD006322. 45. Bajwah S, Ross JR, Wells AU, Mohammed K, Oyebode C, Birring SS, et al. Palliative care for patients with advanced fibrotic lung disease: a randomised controlled 25. Guler SA, Hur SA, Stickland MK, Brun P, Bovet L, Holland AE, et al. Survival after phase II and feasibility trial of a community case conference intervention. Thorax. inpatient or outpatient pulmonary rehabilitation in patients with fibrotic inter- 2015;70:830–839. stitial lung disease: a multicentre retrospective cohort study. Thorax. Published 46. Lindell KO, Klein SJ, Veatch MS, Gibson KF, Kass DJ, Nouraie M, et al. Nurse-Led online ahead of print, August 30 2021; https://doi.org/10.1136/thoraxjnl-2 Palliative Care Clinical Trial Improves Knowledge and Preparedness in Caregivers 021-217361. of Patients with Idiopathic Pulmonary Fibrosis. Ann Am Thorac Soc. 2021;18:1811– 1821. 26. Dowman LM, McDonald CF, Hill CJ, Lee AL, Barker K, Boote C, et al. The evidence of 47. Polke M, Kondoh Y, Wijsenbeek M, Cottin V, Walsh SLF, Collard HR, et al. Man- benefits of exercise training in interstitial lung disease: a randomised controlled agement of Acute Exacerbation of Idiopathic Pulmonary Fibrosis in Specialised and trial. Thorax. 2017;72:610–619. Non-specialised ILD Centres Around the World. Front Med (Lausanne). 2021;8: 699644. 27. Nakazawa A, Dowman LM, Cox NS, McDonald CF, Hill CJ, Lee AL, et al. Factors 48. Chambers DC, Perch M, Zuckermann A, Cherikh WS, Harhay MO, Hayes D, Jr., et al. predicting progression of exercise training loads in people with interstitial lung The International Thoracic Organ Transplant Registry of the International Society disease. ERJ Open Res. 2019;5. for Heart and Lung Transplantation: Thirty-eighth adult lung transplantation report - 2021; Focus on recipient characteristics. J Heart Lung Transplant. 28. Dowman LM, May AK, Cox NS, Morris NR, Nakazawa A, Parker L, et al. Attenuation 2021;40:1060–1072. of exertional desaturation and preference for interval exercise compared to 49. Wickerson L, Rozenberg D, Janaudis-Ferreira T, Deliva R, Lo V, Beauchamp G, et al. continuous exercise in people with interstitial lung disease. Respirology. Physical rehabilitation for lung transplant candidates and recipients: An evidence- 2021;26:1076–1079. informed clinical approach. World J Transplant. 2016;6:517–531. 50. Florian J, Rubin A, Mattiello R, Fontoura FF, Camargo Jde J, Teixeira PJ. Impact of 29. Harada J, Nagata K, Morimoto T, Iwata K, Matsunashi A, Sato Y, et al. Effect of pulmonary rehabilitation on quality of life and functional capacity in patients on high-flow nasal cannula oxygen therapy on exercise tolerance in patients with waiting lists for lung transplantation. J Bras Pneumol. 2013;39:349–356. idiopathic pulmonary fibrosis: A randomized crossover trial. Respirology. 51. Rochester CL, Fairburn C, Crouch RH. Pulmonary rehabilitation for disorders 2022;27:144–151. other than chronic obstructive pulmonary disease. Clin Chest Med. 2014;35:369– 389. 30. Holland AE, Watson A, Glaspole I. Comprehensive pulmonary rehabilitation for 52. Tarrant BJ, Holland A, Le Maitre C, Robinson R, Corbett M, Bondarenko J, et al. The interstitial lung disease: a consensus approach to identify core education topics. timing and extent of acute physiotherapy involvement following lung trans- Patient Educ Couns. 2019;102:1135–1130. plantation: An observational study. Physiother Res Int. 2018;23:e1710. 53. Mackintosh JA, Glenn L, Barnes H, Dunn E, Bancroft S, Reddy T, et al. Benefits of a 31. Holland AE, Fiore JF, Jr., Goh N, Symons K, Dowman L, Westall G, et al. Be honest and virtual interstitial lung disease multidisciplinary meeting in the face of COVID-19. help me prepare for the future: What people with interstitial lung disease want Respirology. 2021;26:612–615. from education in pulmonary rehabilitation. Chronic Respir Dis. 2015;12:93–101. 54. Tikellis G, Corte TJ, Teoh AKY, Glaspole IN, Macansh S, Holland AE. Barriers and facilitators to best care for idiopathic pulmonary fibrosis in Australia. Respirology. 32. de Andrade JA, Kulkarni T, Neely ML, Hellkamp AS, Case AH, Guntupalli K, et al. 2022;27:76–84. Implementation of guideline recommendations and outcomes in patients with idiopathic pulmonary fibrosis: Data from the IPF-PRO registry. Respir Med. 2021;189:106637. 33. Hoffman M, Mellerick C, Symons K, Glaspole I, Holland AE. Pulmonary rehabili- tation for interstitial lung disease: Referral and patient experiences. Chronic Respir Dis. 2021;18:14799731211046022. 34. Cox NS, McDonald CF, Mahal A, Alison JA, Wootton R, Hill CJ, et al. Tele- rehabilitation for chronic respiratory disease: a randomised controlled equivalence trial. Thorax. Published online ahead of print, Oct 14 2021; https://doi.org/10.1136/ thoraxjnl-2021-216934. 35. Kadura S, Raghu G. Rheumatoid arthritis-interstitial lung disease: manifestations and current concepts in pathogenesis and management. Eur Respir Rev. 2021;30. 36. Jacobs SS, Krishnan JA, Lederer DJ, Ghazipura M, Hossain T, Tan AM, et al. Home Oxygen Therapy for Adults with Chronic Lung Disease. An Official American

Journal of Physiotherapy 68 (2022) 156 j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j p hy s Readers’ Choice Award The Editorial Board is pleased to announce the annual Readers’ Choice Award, which recognises the paper published in Journal of Physio- therapy that generates the most interest by readers. The winning paper from a given year is the paper that is downloaded the greatest number of times in the 6 months after its day of publication. The winning paper from among those published in 2021 is ‘Some types of exercise are more effective than others in people with chronic low back pain: a network meta-analysis’.1 The lead author is Associate Professor Jill Hayden, head of the Back Pain Evidence Synthesis and Translation Program in the Department of Community Health and Epidemiology at Dalhousie University, Canada. Her co-authors are Jenna Ellis, Rachel Ogilvie, Samuel Stewart, Matthew Bagg, Sanja Stanojevic, Tiê Yamato and Bruno Saragiotto. The winning paper expertly summarises the results of an enormous amount of research: 217 randomised trials with over 20,000 participants. The review found evidence that Pilates, McKenzie therapy and functional restoration were more effective than other types of exercise for reducing pain intensity and functional limitations in people with chronic low back pain. Most exercise types were found to be more effective than minimal treatment. In addition to garnering the most downloads, the winning paper also generated the highest activity on social media among the papers published in 2021. The Editorial Board of Journal of Physiotherapy congratulates Associate Professor Hayden and her co-authors on their success. Reference 1. Hayden JA, Ellis J, Ogilvie R, Stewart SA, Bagg MK, Stanojevic S, Yamato TP, Saragiotto BT. Some types of exercise are more effective than others in people with chronic low back pain: a network meta-analysis. J Physiother. 2021;67:252–262. https://doi.org/10.1016/j.jphys.2022.05.008 1836-9553/

Journal of Physiotherapy 68 (2022) 210–212 j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j p hy s Appraisal Research Note: Designing implementation trials in physiotherapy Introduction trials are termed hybrid implementation-effectiveness trials5 (see Figure 1). Despite significant investment in health and medical research and the discovery of therapeutic interventions that can improve patient When should you conduct an implementation trial? health, most effective interventions are never integrated into clinical care. Implementation science is the study of methods to promote An implementation trial is warranted when effective clinical in- the systematic uptake of evidence-based interventions or policies terventions are not routinely implemented in practice; that is: there into routine practice to improve the quality and effectiveness of is an evidence-practice gap. Evidence to support effectiveness of an healthcare.1 While it is increasingly being acknowledged that intervention to be implemented should ideally come from a sys- evidence-based medicine includes evidence-based methods of tematic review of randomised controlled trials or from rigorously implementation,2 conducting implementation research remains un- developed clinical practice guidelines, particularly if implementation common in the field of physiotherapy; for example, a search of the is sought at a system-wide level. However, if evidence is less certain Physiotherapy Evidence Database (PEDro) for trials with the term or indirect, a hybrid type 2 trial (Figure 1) can be used to test ‘implementation’ in the title or abstract in November 2021 identified implementation strategies on a smaller scale together with inter- 658 records among the 40,814 total trials listed in the database. To vention effectiveness, as a means to accelerate research translation.5 support the conduct of implementation research, and in particular implementation trials in the field, this Research Note aimed to define Designing an implementation trial implementation trials and provide guidance for designing these trials in physiotherapy. The following section provides guidance for designing an imple- mentation trial. Table 1 provides definitions of key implementation What are implementation trials? terms used within this section. Implementation trials (often randomised) evaluate the success of Use of theories, models and frameworks strategies to help people or organisations put evidence-based in- terventions or policies into practice. Implementation strategies are Changing the professional behaviour of clinicians to integrate defined as the methods or techniques used to enhance the adoption, evidence-based interventions within health systems is complex and implementation and sustainability of an evidence-based interven- influenced by a range of factors operating at different levels. To help tion.3 Some common examples of implementation strategies include navigate this complexity, the use of theories, models and/or frame- conducting educational meetings, creating a learning collaborative, works is recommended within implementation research.9 Although identifying/preparing champions, and audit and feedback.4 these three terms have different definitions (see Table 1), they are often used interchangeably in the literature. Frameworks can be used: It is important that implementation trials distinguish between to describe or guide the process of implementation (ie, process frameworks); to understand and or explain what influences imple- i) the evidence-based intervention that is the subject of imple- mentation outcomes (ie, determinant frameworks); and to evaluate mentation – interventions are typically directed at the patient, the success of implementation (ie, evaluation frameworks). such as a therapy or drug – and Depending on the theory, model or framework selected, more than one may be needed in the planning and conduct of an implementa- ii) the implementation strategy(s) that is being tested in the trial to tion trial. An excellent web tool, developed and managed by re- facilitate effective adoption, implementation and/or sustainment searchers in the United States, provides a step-by-step guide to of the intervention – these are typically directed at the clinician or selecting the best framework(s) to suit the implementation trial the circumstances surrounding the clinician that will influence (https://dissemination-implementation.org/index.aspx). Examples of their practice, such as performance audit/feedback, training or some commonly used frameworks are provided in the sections below. incentives. Process of implementation The primary outcome of an implementation trial should be to measure the extent and quality of the implementation of the inter- The process of implementation is often divided into several steps vention (eg, implementation fidelity such as provision of care to or phases: pre-implementation, implementation and sustainment.11 patients concordant with guideline recommendations). However, in The preparatory work for an implementation trial is conducted instances where the effectiveness of the intervention on patient within the pre-implementation phase, deciding on an evidence-based health outcomes has been insufficiently established, implementation intervention to implement, exploring the local context where the trials may have a dual focus assessing both the effectiveness of the trial will be conducted, identifying implementation barriers and implementation strategy (on measures of implementation) and intervention effectiveness (on measures of patient outcomes); these https://doi.org/10.1016/j.jphys.2022.05.009 1836-9553/© 2022 Australian Physiotherapy Association. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).

Appraisal Research Note 211 Hybrid Type 1a Hybrid Type 2 Hybrid Type 3 • Primary aim: Determine • Co-primary aim: Determine • Primary aim: Determine effectiveness of a clinical effectiveness of a clinical effectiveness of an intervention intervention implementation strategy(s) on an implementation outcome • Secondary aim: Understand • Co-primary aim: Determine potential barriers and effectiveness of an • Secondary aim: Describe facilitators for future implementation strategy(s) on health outcomes implementation through an implementation outcome process evaluation • Participants: Clinicians, policy • Participants: Clinicians, policy makers and service providers • Participants: Patients or a makers, service providers, specific population and patients or a specific • Focus of evaluation: population Implementation strategy(s) to • Focus of evaluation: Clinical support implementation of an intervention or policy • Focus of evaluation: evidence-based clinical Implementation strategy(s) intervention or policy • Exampleb: Hassett 2019⁶ alongside and in support of a clinical intervention or policy • Exampleb: Kerlin 2021⁸ • Exampleb: Abbott 2018⁷ Figure 1. Key features of hybrid implementation-effectiveness trials. Figure is informed by Curran et al 2012.5 a Hybrid Type 1 is not considered an implementation trial. b Protocol paper for each hybrid design describing the implementation of physiotherapy -relevant clinical interventions.6–8 Table 1 providers, funders), and is likely to be undertaken both formally and Common terminology used in implementation science. informally using quantitative and/or qualitative research methods. Terminology Definition Determinant frameworks such as the Consolidated Framework for Implementation Research13 can be useful for conducting formative Implementation A discrete method or technique used to enhance the adoption, pre-implementation research. The Consolidated Framework for strategy Implementation Research includes five domains of potential in- Theory implementation and sustainability of an evidence-based fluences on implementation: intervention characteristics (eg, intervention.4 complexity, adaptability), outer setting (eg, patient needs and re- Model sources), inner setting (eg, culture, learning environment), charac- Framework An idea or set of ideas to help inform observations and teristics of the clinicians (eg, knowledge and beliefs about the Organisational intervention), and the planned process of implementation (eg, readiness understanding of a particular topic. They are typically engaging opinion leaders). Identifying barriers and facilitators to predictive with specified relationships between concepts.9 implementation across the five domains can help to guide the se- A simplified representation of a more complex phenomenon. lection of implementation strategies to address barriers (eg, educa- A theory may be operationalised within a model.9 tion to address lack of knowledge about the intervention) and leverage facilitators (eg, use of opinion leaders as clinical champions). A structure, overview or outline consisting of various A single implementation strategy may be selected; however, it is descriptive categories. Causal relationships are not specified.9 more common that multiple implementation strategies are used A construct encompassing both the willingness and perceived together to address multiple barriers to implementation. capacity of stakeholders within an organisation to engage in The choice of implementation strategies can also be informed adopting a new intervention.10 from the literature of known barriers and facilitators to imple- mentation and/or known effective implementation strategies for facilitators, developing a logic model,12 and preparing the local similar settings, clinician groups or clinical interventions. The context for implementation. An implementation trial is usually un- Cochrane Effective Practice and Organisation of Care group publishes dertaken in the implementation phase and may extend into the systematic reviews on effectiveness of different implementation sustainment phase if the trial’s aims include evaluating sustainment strategies (eg, conducting educational meetings).14 It is recom- of the intervention. Alternatively, the sustainment phase may include mended that implementation strategies are named and defined using additional research where further strategies to sustain implementa- a standardised taxonomy to enable comparison between studies and tion may be employed when there is evidence of implementation build the evidence-base for particular strategies;3 for example, the decay. A commonly used process framework is the Exploration Expert Recommendations for Implementing Change taxonomy pro- Preparation Implementation Sustainment framework.11 This frame- vides a list and definitions of 73 discrete implementation strategies.4 work considers the inner and outer contexts (ie, within and external It is also recommended that sufficient detail about the implementa- to the implementation setting), bridging factors that span across the tion strategy is included to enable the strategy to be replicated in inner and outer context, as well as the suitability of the evidence- different settings (eg, who is delivering it; who, how and what is it based intervention to be implemented throughout the different targeting; what is the mode of delivery and dose; when is it deliv- phases of implementation. ered; and what is the justification for this strategy).3 Understanding the implementation context and building Research design and measurement implementation strategies Selecting the type of design to use in implementation trials de- Formative research undertaken in the pre-implementation phase pends on many factors, including: potential sample size and funds is critical for informing the development of an implementation trial. available, risk of contamination, and the outcomes of interest and This work is important to help understand current usage of the how easily they can be collected.15 While randomised designs are evidence-based intervention in a specific setting, the barriers and frequently used in implementation trial designs, random allocation facilitators that hinder or assist local clinicians being able to put the may not be appropriate (or possible) in many circumstances, and so evidence-based intervention into practice, and to assess organisa- tional readiness for implementation (eg, leadership, culture).10 As stakeholders have a key role to play in the success of an imple- mentation effort, this work should include engagement with a range of stakeholders (eg, clinicians, managers, patients, community

212 Appraisal Research Note non-randomised designs are often also employed.16 For randomised stakeholders, a good understanding of the local context and the use of designs, an individually randomised controlled trial where clinicians frameworks to guide implementation and evaluation. Health systems are individually randomised to one of two or more groups is un- should be ‘learning health systems’ where research is embedded into common for implementation trials because clinicians typically work practice and research is valued as a part of a physiotherapist’s in teams and therefore there is risk of contamination between groups. workload, with appropriate time carved out for them to participate in In addition, influences on clinicians’ ability to implement an implementation research, including education and training.20 evidence-based intervention are likely to be multilevel (ie, at the individual, organisation, community and/or system level)17 with Conclusion implementation strategies often targeting organisation or system level determinants of implementation (eg, change record systems).4 Adoption and implementation of effective physiotherapy in- For these reasons, cluster randomised trials are more common for terventions into clinical practice does not happen automatically, and there implementation trials where teams of clinicians (eg, clinics) are is now greater understanding of the complexity of changing clinician randomised to one of two or more groups. A stepped wedged design behaviour and organisation practice. Implementation science seeks to where groups (eg, whole departments) can be randomised to receive generate new knowledge to guide how best such practice change can be the implementation strategy(ies) in a sequential order can also achieved. This article introduces key terminology in implementation reduce the risk of contamination, but require fewer clusters. For a science, includes a reference list of useful methodology papers in the field, more detailed description of different study design options see and provides guidance for conducting implementation trials in physio- Wolfenden 2021.15 therapy. It is hoped that this will provide further interest of physiothera- pists in conducting implementation trials to enable more evidence-based Implementation trials have an implementation outcome as their interventions to be embedded into healthcare systems. primary outcome. McKay et al 201918 provide a framework for rec- ommended implementation outcomes and determinants for use Provenance: Invited. Peer reviewed. in implementation trials. The framework describes five key implementation outcomes – reach, adoption, fidelity, dose and sus- Leanne Hassetta,b,c and Luke Wolfendend tainment – and suggests that these measures should be collected aInstitute for Musculoskeletal Health, The University of Sydney/Sydney both for the delivery of the evidence-based intervention and imple- mentation strategy(ies); for example, fidelity of the evidence-based Local Health District, Sydney, Australia intervention delivered by the clinicians and fidelity of the imple- bSydney School of Health Sciences, Faculty of Medicine and Health, mentation strategies used to facilitate implementation of the inter- vention should both be measured. It is worth noting that some of University of Sydney, Sydney, Australia these measures may only be collected from the group(s) randomised cLiverpool Hospital, South Western Sydney Local Health District, to receive the implementation strategy(ies). Measures of accept- ability, feasibility and cost have also been suggested as useful deter- Sydney, Australia minant measures for inclusion in implementation trials and can be dSchool of Medicine and Public Health, University of Newcastle, assessed using both quantitative and qualitative methods.18 Newcastle, Australia Reporting implementation trials References To report implementation trials, the Standards for Reporting Implementation Studies (STaRI) guideline19 should be used in 1. Eccles MP, Mittman BS. Implement Sci. 2006;1. conjunction with the relevant Consolidated Standards for Reporting 2. Nilsen P, Bernhardsson S. J Physiother. 2013;59:143–144. Trials (CONSORT) guideline for the type of randomised controlled 3. Proctor EK, et al. Implement Sci. 2013;8:139. trial chosen. 4. Powell BJ, et al. Implement Sci. 2015;10:21. 5. Curran GM, et al. Med Care. 2012;50:217–226. Future directions for implementation research in physiotherapy 6. Hassett L, et al. BMJ Open. 2020;10:e034696. 7. Abbott A, et al. BMJ Open. 2018;8:e019906. Strengthening implementation research in physiotherapy can be 8. Kerlin MP, et al. Implement Sci. 2021;16:78. facilitated in a few ways. When designing interventions, future 9. Nilsen P. Implement Sci. 2015;10:53. implementation settings and key stakeholders should be borne in 10. Vax S, et al. Implement Sci. 2021;16:61. mind and effectiveness trials be conducted while collecting data to 11. Aarons GA, et al. Adm Policy Ment Health. 2011;38:4–23. inform future implementation (hybrid type 1 study design).5 For 12. Smith JD, et al. Implement Sci. 2020;15:84. existing effective evidence-based interventions in physiotherapy, 13. Damschroder LJ, et al. Implement Sci. 2009;4:50. consider randomised implementation trials to assess how these 14. Forsetlund L, et al. Cochrane Database Syst Rev. 2021;9:CD003030. interventions can be implemented in practice. This requires 15. Wolfenden L, et al. BMJ. 2021;372:m3721. pre-implementation work, including good engagement with local 16. Wolfenden L, et al. Prev Med Rep. 2016;4:441–443. 17. Tabak RG, et al. Frontiers in PHSSR. 2013;2. 18. McKay H, et al. Int J Behav Nutr Phys Act. 2019;16:102. 19. Pinnock H, et al. BMJ Open. 2017;7:e013318. 20. Moullin JC, et al. Implement Sci Commun. 2020;1:42.

Journal of Physiotherapy 68 (2022) 182–190 j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j p hy s Research Stratified exercise therapy does not improve outcomes compared with usual exercise therapy in people with knee osteoarthritis (OCTOPuS study): a cluster randomised trial Jesper Knoop a, Joost Dekker b, Johanna M van Dongen a, Marike van der Leeden b,c, Mariette de Rooij c, Wilfred FH Peter c,d, Willemijn de Joode a, Leti van Bodegom-Vos e, Nique Lopuhaä f, Kim L Bennell g, Willem F Lems c,h, Martin van der Esch c,i, Thea PM Vliet Vlieland d, Raymond WJG Ostelo a,j,k a Department of Health Sciences, Vrije Universiteit Amsterdam, Amsterdam, Netherlands; b Department of Rehabilitation Medicine, Amsterdam UMC, Location VUmc, Amsterdam, Netherlands; c Amsterdam Rehabilitation Research Centre, Reade, Amsterdam, Netherlands; d Department of Orthopaedics, Leiden University Medical Center, Leiden, Netherlands; e Department of Biomedical Data Sciences, Leiden University Medical Center, Leiden, Netherlands; f Dutch Arthritis Society, Amsterdam, Netherlands; g Department of Physiotherapy, School of Health Sciences, University of Melbourne, Melbourne, Australia; h Amsterdam UMC, location VUmc, Department of Rheumatology, Amsterdam, Netherlands; i Center of Expertise Urban Vitality, Health Faculty, Amsterdam University of Applied Sciences, Amsterdam, Netherlands; j Amsterdam Movement Sciences, Vrije Universiteit Amsterdam, Amsterdam, Netherlands; k Department of Epidemiology and Data Science, Amsterdam UMC, Location VUmc, Amsterdam, Netherlands KEY WORDS ABSTRACT Knee osteoarthritis Question: In people with knee osteoarthritis, how much more effective is stratified exercise therapy that Exercise therapy distinguishes three subgroups (high muscle strength subgroup, low muscle strength subgroup, obesity Dietary intervention subgroup) in reducing knee pain and improving physical function than usual exercise therapy? Cluster randomized controlled trial Design: Pragmatic cluster randomised controlled trial in a primary care setting. Participants: A total of Stratified care 335 people with knee osteoarthritis: 153 in an experimental arm and 182 in a control arm. Intervention: Physiotherapy practices were randomised into an experimental arm providing stratified ex- ercise therapy (supplemented by a dietary intervention from a dietician for the obesity subgroup) or a control arm providing usual, non-stratified exercise therapy. Outcome measures: Primary outcomes were knee pain severity (numerical rating scale for pain, 0 to 10) and physical function (Knee Injury and Osteoarthritis Outcome Score subscale activities of daily living, 0 to 100). Measurements were performed at baseline, 3 months (primary endpoint) and 6 and 12 months (follow-up). Intention-to-treat, multilevel, regression analysis was performed. Results: Negligible differences were found between the experimental and control groups in knee pain (mean adjusted difference 0.2, 95% CI –0.4 to 0.7) and physical function (–0.8, 95% CI –4.3 to 2.6) at 3 months. Similar effects between groups were also found for each subgroup separately, as well as at other time points and for nearly all secondary outcome measures. Conclusion: This pragmatic trial demonstrated no added value regarding clinical outcomes of the model of stratified exercise therapy compared with usual exercise therapy. This could be attributed to the experimental arm therapists facing difficulty in effectively applying the model (especially in the obesity subgroup) and to elements of stratified exercise therapy possibly being applied in the control arm. Registration: Netherlands National Trial Register NL7463. [Knoop J, Dekker J, van Dongen JM, van der Leeden M, de Rooij M, Peter WFH, de Joode W, van Bodegom-Vos L, Lopuhaä N, Bennell KL, Lems WF, van der Esch M, Vliet Vlieland TPM, Ostelo RWJG (2022) Stratified exercise therapy does not improve outcomes compared with usual exercise therapy in people with knee osteoarthritis (OCTOPuS study): a cluster randomised trial. Journal of Physiotherapy 68:182–190] © 2022 Australian Physiotherapy Association. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Introduction strong, high-quality evidence for the effectiveness of exercise therapy Osteoarthritis (OA) is a common chronic health condition and a on knee pain and physical function compared with no exercise leading cause of pain and disability among adults.1 Knee OA is a therapy.5 These effects have been found in mild OA and also in severe highly heterogeneous disease but is likely to consist of homogeneous OA.6,7 However, although effective, the average standardised effect phenotypes or subgroups.2,3 Exercise therapy is recommended as a first-step treatment, next to pain medication and diet.4 There is size of exercise therapy compared with no exercise therapy is only moderate (approximately 0.5).5 This may be attributed to the current ‘one-size-fits-all’ exercise approach. A stratification into subgroups https://doi.org/10.1016/j.jphys.2022.06.005 1836-9553/© 2022 Australian Physiotherapy Association. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).

Research 183 that receive specifically tailored interventions may yield superior within 50 km of one of the two study centres, they had exercise clinical and economic outcomes.8 Randomised trials are needed to therapy facilities and they were treating on average at least one new determine the added value of such a stratified approach of exercise patient with knee OA per month. therapy. A total of 55 practices with 126 physiotherapists were randomly Five phenotypes or subgroups of knee OA patients were recently allocated (1:2 ratio) to the experimental arm (19 practices, 46 phys- identified in one cohort9 and this finding was replicated in a second iotherapists) or control-arm (36 practices, 80 physiotherapists), by cohort.10 Subsequently, a model of stratified exercise therapy was using a web-based randomisation program, with random sequence developed based on three of the identified subgroups that are aligned generation and concealment of randomisation guaranteed. The 1:2 with well-accepted OA phenotypes: a ‘low muscle strength subgroup’ ratio was chosen to reduce the number of participants needed for (LMS) (‘age-induced phenotype3), a ‘high muscle strength subgroup’ each control-arm physiotherapist, thereby minimising their burden, (HMS) (‘post-traumatic phenotype3), and an ‘obesity subgroup’ (OS) as physiotherapists expressed interest in this study primarily because (‘metabolic phenotype3). For each subgroup, a subgroup-specific ex- of the experimental intervention. Because of low participant inclusion ercise therapy intervention was developed, based on existing exercise rates in the experimental arm, a second wave was added to recruit therapy interventions,11,12 supplemented by a dietary intervention for physiotherapists with a 2:1 instead of 1:2 randomisation (in order to the OS. These interventions, together with a simple stratification al- include more experimental-arm than control-arm physiotherapists) gorithm to allocate patients into one of the subgroups, were pilot and without the criterion of at least two participating physiothera- tested for feasibility and further optimisation.13 Moreover, it was pists per practice. This second wave resulted in four practices with found that the construct validity of the final stratification algorithm eight physiotherapists being allocated to the experimental arm and was adequate, as it consistently aligned these subgroups with their two practices with three physiotherapists to the control arm. Each of proposed phenotype.14 the experimental-arm physiotherapy practices were instructed to nominate a dietician with whom they already collaborated for the It was then hypothesised that this model of stratified exercise diet intervention in the OS. therapy – in which the HMS, LMS and OS subgroups receive subgroup-specific exercise therapy, supplemented by a dietary Participants intervention for the OS – would be more effective in reducing pain and improving physical function in patients with knee OA compared Participants were recruited by participating physiotherapists in with usual, ‘non-stratified’ exercise therapy. primary care, and tested for eligibility with the following eligibility criteria. The inclusion criteria were: the presence of knee pain with Therefore, the study question for this cluster randomised trial a duration  3 months and severity during walking  2/10 on the was: pain numerical rating scale; and a clinical knee OA diagnosis.19 The exclusion criteria were: age , 40 or . 85 years; pain severity during In patients with knee osteoarthritis, how much more effective is walking  9/10 on the pain numerical rating scale; physical or stratified exercise therapy that distinguishes three subgroups mental comorbidity severely affecting daily life and contra- (‘high muscle strength subgroup’, ‘low muscle strength subgroup’ indicating exercise therapy; suspicion of chronic widespread pain; and ‘obesity subgroup’) in reducing knee pain and improving (planned) total knee arthroplasty; other reasons for knee pain (eg, physical function than usual exercise therapy? rheumatoid arthritis, gout); physiotherapy or intra-articular in- jections for knee pain in past 6 months; and insufficient Dutch Method language comprehension. Study design Intervention A pragmatic, parallel, two-group, cluster randomised controlled Experimental arm trial in primary care was performed – the Optimisation of exerCise Physiotherapists were trained to provide treatment according to Therapy in patients with knee Osteoarthritis in a Primary care Setting (OCTOPuS) study – which was accompanied by a qualitative process the model of stratified exercise therapy, which consisted of: subgroup evaluation15 and economic evaluation (under review). Detailed allocation by physiotherapists through a simple, stratification algo- description of the trial methods has been published.16 This trial is rithm (Figure 2); and subgroup-specific, protocolised exercise therapy reported according to the CONSORT 2010 checklist for cluster rand- interventions (see Table 1 for a summary and Appendix 2 on the omised controlled trials (Appendix 1 on the eAddenda).17 eAddenda for detailed information). In line with current recommendations for pragmatic cluster Dieticians were instructed to deliver a dietary intervention ac- randomised controlled trials,18 patients were blinded for treatment cording to the current Dutch guideline20 to participants in the OS (see allocation (ie, only informed about their own intervention). Blinding Table 1 for summary and Appendix 3 on the eAddenda for detailed experimental arm physiotherapists or dieticians was not possible, but information). control arm physiotherapists were blinded for the content of the experimental intervention. Blinding of researchers responsible for the Control arm study logistics was not possible. However, an independent researcher Physiotherapists were instructed to provide their usual care who was blinded for treatment allocation performed the random- isation of physiotherapy practices and primary analyses. (ie, standard, ‘non-stratified’ exercise therapy, according to the guideline21). Physiotherapists screened patients for eligibility at their first consultation. After providing informed consent, eligible patients were Outcome measures included and asked to complete questionnaires at baseline (T0), 3 months follow-up (T3), 6 months follow-up (T6), and 12 months Patient-reported outcome measures were administered at T0, T3 follow-up (T12). All physiotherapists and dieticians registered treat- (primary end-point), T6 and T12, while the three physical tests were ment fidelity parameters for each session. As this was a pragmatic applied at T0 and T3 (by physiotherapist) and T12 (by physiotherapist trial, patients from both groups were allowed to receive any addi- or researcher team member). tional usual care and this was monitored in the follow-up question- naires. The design of the study is presented in Figure 1. Physiotherapists Primary outcome measures Physiotherapists were considered eligible if there were at least Average knee pain severity during walking in the past week: assessed two physiotherapists working in one practice, they were located by a numerical rating scale (score 0 = no pain; 10 = worst pain imaginable).22

184 Knoop et al: Stratified exercise therapy for knee osteoarthritis Cluster enrolment Recruitment of PTs (n = 155) Excluded (n = 18) ♦ Not meeting inclusion criteria (n = 4) ♦ > 50 km from research center (n = 2) ♦ < 1 knee OA patient/month (n = 2) ♦ Declined to participate (n = 14) Randomisation of PTs on level PTs withdrawn before start of PT practice (n = 137) trial (n = 5) PTs withdrawn before PTs in intervention Cluster allocation PTs in control ♦ Declined to participate (n start trial (n = 2) group (n = 54) + group (n = 83) = 3) ♦ Stopped working at dieticians (n = 21) for ‘obesity subgroup’ Patients tested for ♦ Stopped working at PT PT practice (n = 2) eligibility (n = 298) practice (n = 2) Patients tested for Patient enrolment PTs discontinued eligibility (n = 290) PTs discontinued during during trial (n = 2) trial (n = 7) ♦ Stopped working at ♦ Stopped working at PT PT practice (n = 2) practice (n = 5) ♦ Unwilling to participate anymore (n = 2) Excluded (n = 137) Excluded (n = 116) ♦ Not meeting inclusion ♦ Not meeting inclusion criteria criteria (n = 79) (n = 105) ♦ Declined to participate (n ♦ Declined to participate = 37) (n = 32) Included, provided informed consent and Included and provided informed allocated (n = 153): consent (n = 182) ‘high muscle ‘low muscle ‘obesity strength strength subgroup’ (n = 35) subgroup’ (n = subgroup’ (n = 64) 54) Lost to FU before T3 (n = 4) Patients Follow-up Patients Lost to FU before T3 (n = 11) assessed at: assessed at: ♦ Withdrawn before T0 (n = 5) ♦ Withdrawn before T0 (n = T0 (n = 151) T0 (n = 177) 2) T3 (n = 141) T3 (n = 165) ♦ Withdrawn between T0 to T6 (n = 130) T6 (n = 156) T3 (n = 6) ♦ Withdrawn between T0 to T9 (n = 123) T9 (n = 148) T3 (n = 2) T12 (n = 131) T12 (n = 155) Lost to FU after T3 (n = 11) ♦ Withdrawn between T3 to Lost to FU after T3 (n = 15) T12 (n = 11) ♦ Withdrawn between T3 to T12 (n = 15) Analysed in ITT T0 to T3 (n = 151) Analysis Analysed in ITT T0 to T3 (n = 177) Analysed in per-protocol T0 to T3 (n = 120/107): Analysed in per-protocol T0 to T3 (n = 165/165): ♦ Excluding major violators (n = 31) ♦ Excluding major violators (n = 12) ♦ Excluding major and minor violators (n = 44) ♦ Excluding major and minor violators (n = 12) Analysed in ITT T0 to T12 (n = 151) Analysed in ITT T0 to T12 (n = 177) Analysed in per-protocol T0 to T12 (n = 111/69): Analysed in per-protocol T0 to =T1125)1(n = 162): ♦ Excluding major violators (n = 40) ♦ Excluding major violators (n ♦ Excluding major and minor violators (n = 82) ♦ Excluding major and minor violators (n = 15) Figure 1. Flow diagram. FU = follow-up, ITT = intention to treat, OA = osteoarthritis, PT = physiotherapist, T = time point. Physical functioning: assessed by subscale function in daily living Secondary outcome measures Global perceived effect: measuring the participant’s subjective (activities of daily living) of the Dutch translation of the Knee Injury global change using a 7-point scale ranging from ‘worse than ever’ to and Osteoarthritis Outcome Score (KOOS) questionnaire (score 0 = ‘completely recovered’. This was dichotomised as follows: recovered maximal problems; 100 = no problem).23,24

Research 185 Presence of obesity yes Obesity subgroup (BMI ≥ 30 kg/m²) High muscle strength subgroup no yes Presence of ‘high’ muscle strength (30-second CST score ≥ 12) no Low muscle strength subgroup Figure 2. Stratification algorithm. BMI = body mass index, CST = chair stand test. (‘completely recovered’ and ‘much recovered’) versus not recovered Healthcare utilisation (all other responses). At baseline and every follow-up time point, participants registered Physical functioning: assessed by the short version of the Dutch in the questionnaires the number of received consults of any other translation of the KOOS questionnaire (0 = maximal problems; 100 = healthcare in the past 3 months. Healthcare utilisation was sub- no problem).23,24 divided into general practitioner, other primary care (eg, occupational therapist), secondary care (eg, orthopaedic surgeon) and alternative Pain interference: assessed by the short version of the Patient Re- care (eg, acupuncturist). ported Outcomes Measurement Information System (PROMIS) (4 = best score; 20 = worst score).25 Additional baseline measures General participant characteristics (age, gender, duration of knee Fatigue: assessed by the short version of PROMIS (4 = best score; 20 = worst score).25 symptoms, left/right knee, history of knee surgery, comorbidity) and physiotherapist characteristics (age, gender, years employed, years of Patient-reported knee instability: assessed by frequency of episodes experience treating knee OA patients, number of knee OA patients of knee instability in the past 3 months, and severity of impact of treated monthly, education level and additional OA-related educa- knee instability on daily living.26 tion) were collected through questionnaires. Upper leg muscle strength: assessed by the 30-second chair stand Data analysis test, as a measure of lower body strength and physical functioning.27 A between-group difference of 0.5 on the 0 to 10 pain numerical Body mass index (BMI): body weight (in kilograms) divided by rating scale was a priori expected, based on the effectiveness of usual squared body height (in metres). exercise therapy5, the pilot study13 and a previous study29 comparing stratified exercise therapy versus usual care in low back pain. With an Waist circumference: distance around the abdomen in the hori- estimated standard deviation (SD) of 1.4, a = 0.05 (two-sided testing), zontal plane midway between the lowest rib and iliac crest. power = 90% and design effect of 1.05, 346 participants were needed. With a 15% drop-out rate, the sample size was 408 participants (204 Treatment fidelity measures per group). The treatment fidelity measures were: number of sessions All data were analysed according to the intention-to-treat prin- (physiotherapy and dietary intervention); intervention modalities ciple. In a deviation to the statistical analysis plan, multiple (physiotherapy and dietary intervention); adverse events (physio- therapy and dietary intervention); training intensity on 6-to-20 Borg scale28 as perceived by the patient (physiotherapy intervention); and patient-reported adherence to home exercises and (moderate and vigorous) physical activities (physiotherapy intervention). Table 1 Summarya of description of subgroup-specific, protocolised exercise therapy interventions. High muscle strength subgroup Low muscle strength subgroup Obesity subgroup Exercise therapy from physiotherapist Exercise therapy from physiotherapist Exercise therapy from physiotherapist Number of sessions: Number of sessions: Number of sessions: - 3 to 5 individual sessions in a - 8 to 12 individual sessions in a - 12 to 18 individual sessions in a 12-wk treatment period - 2 to 3 ‘booster’ sessions in the post-treatment period 12-wk treatment period 12-wk treatment period Content: - 1 ‘booster’ session in the post- - 1 to 2 ‘booster’ sessions in the a) subgroup-specific education/advice b) supervised exercise therapy adapted to obesity, targeting treatment period post-treatment period Content: Content: upper leg muscle strength, aerobic capacity and weight loss a) subgroup-specific education/advice a) subgroup-specific education/advice c) home exercises b) home exercises b) supervised exercise therapy, Dietary intervention from dietician Number of sessions: primarily targeting upper leg - 5 to 8 individual sessions, of 150 minutes in total muscle strength Content: c) home exercises - advising and monitoring healthy diet and active lifestyle, aiming at  10% weight loss Interprofessional consultation between physiotherapist and dietician - at least one consultation after 3 to 4 wks of treatment to agree on an approach to achieve sustainable lifestyle change a Detailed information provided at Appendices 1 (physiotherapy intervention) and 2 (diet intervention) on the eAddenda.

186 Knoop et al: Stratified exercise therapy for knee osteoarthritis imputation was used instead of maximum likelihood estimation for Table 2 missing values, due to the number of missing data. Baseline characteristics. A multilevel, regression analysis – with levels of physiotherapy Exp Con practice, patient and time point – was performed for the primary (n = 151) (n = 177) outcomes. The primary end-point was T3, with the other time points and the overall effect for the total follow-up period as secondary time General patient characteristics 66 (9) 64 (9) points. Analyses were adjusted for the baseline value of the outcome Age (years), mean (SD) 95 (63) 114 (64) variable, and of the presumed effect modifiers (ie, pain severity, up- Gender (female), number (%) 9 (10) per leg muscle strength and BMI). Similar analyses for secondary Duration of knee symptoms (yr), mean (SD) 50 (33) 7 (8) outcome measures were performed with continuous scales, while History of knee surgery (yes), n (%) 63 (42) 55 (31) logistic multilevel analysis was performed for the dichotomous sec- Using pain medication (yes), n (%) 86 (49) ondary outcome measures. Subgroup analyses were also performed Comorbidity, n (%) 90 (60) comparing the two trial arms separately for each of the three sub- 32 (21) 103 (58) groups. Sensitivity analyses were performed: excluding all cases with present 31 (18) a major protocol violation; and excluding all cases with any protocol present and affecting daily life 54 (36) violation (per-protocol analyses), with all a priori formulated protocol Work status, n (%) 16 (11) 75 (42) violations described in Appendix 4 on the eAddenda. The intended paid work 79 (52) 16 (9) sensitivity analysis accounting for the number of participating phys- no work, but not retired 2 (1) 85 (48) iotherapists within each physiotherapy practice (as a fourth level retired 10 (7) 1 (1) alongside physiotherapy practice, patient and time point) could not unknown 13 (7) be performed due to small numbers. Finally, we calculated relative Sick leave in past 3 mths, n (%) 63 (42) improvements (%), number of responders (based on currently Allocation to subgroup, n (%) 54 (35) 65 (37)a accepted cut-off points for minimally clinically important change) High muscle strength subgroup 34 (23) 53 (30)a and within-group and between-group Cohen’s effect sizes. Low muscle strength subgroup 53 (30)a Obesity subgroup 5.1 (2.1) StataSE softwarea was used for multilevel regression analyses. IBM Unknown 6 (3)a softwareb was used for description of baseline, treatment fidelity and Clinical variables 68 (19) healthcare utilisation characteristics. Knee pain severity (NRS, 0 to 10), mean (SD) 55 (22) 5.3 (2.1) Physical function (0 to 100), mean (SD) 10.1 (3.5) Results KOOS ADL 8.0 (3.7) 65 (18) KOOS Short form 51 (20) Flow of participants through the study Pain interference (PROMIS, 4 to 20), mean (SD) 111 (74) 10.5 (3.4) Fatigue (PROMIS, 4 to 20), mean (SD) 65 (43) 8.1 (3.7) Figure 1 shows the flow of physiotherapists and participants Knee instability, n (%) 11.7 (3.1) through the trial. A total of 137 physiotherapists were randomised to in past 3 mths 27.1 (4.0) 137 (77) the experimental arm (n = 54) and control arm (n = 83). In addition, affecting daily life 97.9 (11.4) 84 (47) 21 dieticians were recruited for the diet intervention of the OS. Pa- 30-s chair stand test (reps), mean (SD) 11.5 (3.7) tients were enrolled between January 2019 and May 2020, and Body mass index (kg/m2), mean (SD) 28.6 (4.9) enrolment was stopped at 335 participants (ie, 82% of the intended Waist circumference (cm), mean (SD) 100.7 (13.5) 408) due to a COVID-19 lock-down that obstructed any further in- clusion for a substantial period. The intention-to-treat analysis General physiotherapist characteristics Exp Con included 328 participants, which was 95% of the 346 that were (n = 54) (n = 80) needed based on the sample size calculation. Age (yr), mean (SD) 34 (11) 39 (11) Baseline characteristics of the study participants Gender (female), n (%) 26 (48) 37 (45) Years working as physiotherapist, mean (SD) 10 (9) 14 (11) The baseline characteristics of participants from the experimental arm (n = 151) and control arm (n = 177), and of the physiotherapists ADL = activities of daily living, Con = control group, Exp = experimental group, KOOS = are described in Table 2. In the experimental arm, 42% were allocated Knee Injury and Osteoarthritis Outcome Score, NRS = numerical rating scale, PROMIS = to HMS, 35% to LMS and 23% to OS. No stratification algorithm was the Patient Reported Outcomes Measurement Information System. used in the control arm, but based on baseline characteristics, 38% would have been allocated to HMS, 31% to LMS and 31% to OS. a Control arm was unaware of and did not use stratification algorithm for subgroup allocation. not shown). Both arms showed comparable average relative im- provements at T3 and T12 (ie, 20 to 30% for knee pain and 10 to 20% for physical function), which only slightly varied across subgroups (see Table 4). Finally, the two arms showed comparable responder rates (ie, 65% responders for knee pain; 42% and 45% responders in experimental and control arm, respectively, for physical function) and comparable within-group effect sizes (ie, 0.5 to 0.7 for knee pain; 0.3 to 0.6 for physical function) (see Appendix 7 on the eAddenda). In- dividual participant data are presented in Table 5 on the eAddenda. Clinical outcomes Treatment fidelity The adjusted mean differences between the experimental and The mean number of physiotherapy sessions was 8.4 (SD 4.7) in control arms on the primary outcomes knee pain (numerical rating the experimental arm and 9.6 (SD 4.8) in the control arm (Table 6). scale) and physical function (KOOS ADL) at T3 were 0.2 (95% CI –0.4 to The mean number of sessions in the experimental arm differed across 0.7) and –0.8 (95% CI –4.3 to 2.6), respectively (see Table 3, and subgroups, as instructed (5.3 for HMS, 9.6 for LMS, 12.2 for OS), while Appendix 5a on the eAddenda). These differences can be considered they were similar across subgroups in the control arm (9.9 for HMS, negligible. Subgroup analyses in which the experimental and control 9.4 for LMS, 9.9 for OS). The OS participants from the experimental groups were compared in each of the three subgroups separately arm additionally received on average 3.2 sessions (SD 1.8) from a resulted in similar results of negligible differences between arms (see dietician. The mean training intensity was marginally higher in the Appendix 5b to 5d and Appendix 6 on the eAddenda). experimental arm – Borg score (scale 6 to 20) of 13.1 (SD 2.0) versus 12.3 (SD 2.0) – based on 2,021 registered sessions. Home exercises Negligible differences were also found between arms for the were performed for on average 3 days/week throughout the 12- secondary outcome measures. Small effects were found favouring the month follow-up period in both groups. In 61% and 58% of the par- experimental arm for pain interference (PROMIS; at T3, T12 and for ticipants, the intended level of physical activity (ie, 150 minutes/week overall effect) and upper leg muscle strength (30-second chair stand of at least moderately intense activities) was reached at 12-months test; for overall effect only), but these were all too small to be of follow-up, in the experimental and control groups, respectively. clinical relevance. Per-protocol analyses yielded similar findings (data

Research 187 Table 3 Primary and secondary outcomes. Scoresa Estimated mean difference (MD) between arms Exp (n = 151) Con (n = 177) Crude analyses Adjusted analysesb MD (95% CI) MD (95% CI) Primary outcomes Knee pain severity (NRS, 0 to 10) 5.1 (0.17) 5.3 (0.16) n/a n/a Baseline 4.1 (0.19) 4.1 (0.20) 0.1 (–0.5 to 0.6) 0.2 (–0.4 to 0.7) 3 mths follow-up 4.1 (0.20) 4.0 (0.20) 0.1 (–0.5 to 0.7) 0.3 (–0.3 to 0.8) 6 mths follow-up 3.6 (0.19) 3.9 (0.18) –0.2 (–0.8 to 0.4) –0.2 (–0.7 to 0.4) 12 mths follow-up 0.0 (–0.5 to 0.5) 0.1 (–0.3 to 0.5) Overall n/a n/a n/a n/a Physical function (KOOS ADL, 0 to 100) c 68 (1.6) 65 (1.4) 1.1 (–2.9 to 5.1) –0.8 (–4.3 to 2.6) Baseline 74 (1.4) 73 (1.5) 1.4 (–2.7 to 5.6) –0.5 (–4.1 to 3.1) 3 mths follow-up 76 (1.6) 74 (1.5) 0.8 (–3.2 to 4.8) –0.6 (–4.0 to 2.8) 6 mths follow-up 77 (1.4) 76 (1.4) 1.1 (–2.4 to 4.6) –0.7 (–3.5 to 2.2) 12 mths follow-up Overall n/a n/a Secondary outcomes Global perceived effect (at least ‘much improved’) 48 (32%) 60 (34%) 0.9 (0.4 to 1.9)d 0.9 (0.4 to 1.8)d 3 mths follow-up 54 (36%) 67 (38%) 0.9 (0.4 to 1.9)d 0.8 (0.4 to 1.8)d 6 mths follow-up 65 (43%) 73 (41%) 1.1 (0.5 to 2.6)d 1.1 (0.5 to 2.5)d 12 mths follow-up 1.0 (0.5 to 1.7)d 0.9 (0.5 to 1.7)d Overall n/a n/a n/a n/a Physical function (KOOS short form, 0 to 100) c 55 (1.8) 51 (1.5) 1.3 (–3.2 to 5.8) –1.08 (–4.9 to 3.0) Baseline 60 (1.6) 58 (1.6) 2.0 (–2.8 to 6.7) –0.3 (–4.6 to 4.0) 3 mths follow-up 63 (1.8) 61 (1.7) 0.2 (–5.4 to 5.7) –2.1 (–7.2 to 3.0) 6 mths follow-up 63 (1.8) 63 (1.9) 1.2 (–2.9 to 5.3) –1.1 (–4.6 to 2.3) 12 mths follow-up Overall n/a n/a Pain interference (PROMIS, 4 to 20) 10.1 (0.28) 10.5 (0.26) n/a n/a Baseline 8.0 (0.24) 9.0 (0.30) –0.9 (–1.8 to –0.1) –0.7 (–1.4 to –0.1) 3 mths follow-up 7.9 (0.30) 8.4 (0.26) –0.5 (–1.3 to 0.4) –0.3 (–0.9 to 0.4) 6 mths follow-up 7.3 (0.27) 8.3 (0.28) –1.0 (–1.8 to –0.1) –0.8 (–1.5 to 0.0) 12 mths follow-up –0.8 (–1.5 to –0.1) –0.6 (–1.1 to –0.1) Overall n/a n/a Fatigue (PROMIS, 4 to 20) 8.1 (0.30) 8.1 (0.28) n/a n/a Baseline 7.7 (0.27) 7.4 (0.26) 0.2 (–0.8 to 1.1) 0.3 (–0.3 to 1.0) 3 mths follow-up 7.4 (0.32) 7.9 (0.30) –0.6 (–1.6 to 0.5) –0.4 (–1.2 to 0.4) 6 mths follow-up 7.2 (0.27) 7.5 (0.28) –0.3 (–1.3 to 0.6) –0.2 (–0.8 to 0.5) 12 mths follow-up –0.2 (–1.4 to 0.7) –0.1 (–0.6 to 0.5) Overall n/a n/a Knee instability in past 3 mths 112 (74%) 136 (77%) n/a n/a Baseline 122 (81%) 149 (84%) 0.8 (0.3 to 2.2)d 0.9 (0.4 to 2.3)d 3 mths follow-up 110 (73%) 140 (79%) 0.7 (0.3 to 1.6)d 0.7 (0.3 to 1.7)d 6 mths follow-up 104 (69%) 120 (68%) 1.1 (0.5 to 2.6)d 1.3 (0.6 to 2.7)d 12 mths follow-up 0.9 (0.5 to 1.7)d 1.0 (0.6 to 1.7)d Overall n/a n/a Knee instability affecting daily life 65 (43%) 85 (48%) n/a n/a Baseline 59 (39%) 76 (43%) 0.7 (0.3 to 1.8)d 0.9 (0.4 to 1.9)d 3 mths follow-up 56 (37%) 64 (36%) 1.0 (0.4 to 2.5)d 1.2 (0.5 to 2.8)d 6 mths follow-up 45 (30%) 62 (35%) 0.7 (0.3 to 1.8)d 0.8 (0.3 to 2.0)d 12 mths follow-up 0.81 (0.40 to 1.63)d 0.96 (0.53 to 1.76)d Overall n/a n/a Upper leg muscle strength (30-s CST, repetitions) 11.7 (0.26) 11.4 (0.28) n/a n/a Baseline 14.9 (0.28) 14.0 (0.36) 0.9 (–0.3 to 2.0) 0.8 (–0.1 to 1.7) 3 mths follow-up 15.3 (0.68) 14.3 (0.38) 0.9 (–0.4 to 2.2) 0.8 (–0.4 to 2.0) 12 mths follow-up 0.9 (–0.1 to 1.9) 0.8 (0.1 to 1.5) Overall n/a n/a n/a n/a Body mass index (kg/m2) 27.1 (0.33) 28.6 (0.37) –1.5 (–2.6 to –0.5) –0.2 (–0.7 to 0.2) Baseline 26.9 (0.32) 28.4 (0.35) –1.4 (–2.5 to –0.3) –0.1 (–0.6 to 0.4) 3 mths follow-up 27.1 (0.32) 28.5 (0.40) –1.5 (–2.5 to –0.4) –0.2 (–0.6 to 0.2) 12 mths follow-up Overall n/a n/a Waist circumference (cm) 97.7 (0.93) 100.9 (1.03) n/a n/a Baseline 96.6 (0.90) 99.5 (0.98) –3.0 (–6.2 to 0.2) –0.2 (–1.6 to 1.3) 3 mths follow-up 96.3 (1.04) 99.5 (1.05) –3.3 (–6.9 to 0.2) –0.5 (–3.0 to 1.9) 12 mths follow-up –3.2 (–6.3 to 0.0) –0.4 (–1.8 to 1.1) Overall n/a n/a ADL = activities of daily living, BMI = body mass index, Con = control group, CST = chair stand test, Exp = experimental group, KOOS = Knee Injury and Osteoarthritis Outcome Score, NRS = numerical rating scale, PROMIS = the Patient Reported Outcomes Measurement Information System. a Mean (standard error) for continuous outcomes; % for dichotomous outcomes. b Adjusted for baseline value of outcome measure, in addition to baseline values of NRS pain, 30-s chair stand test and BMI. c Level of physiotherapist practice could not be added. d Odds ratio.

188 Knoop et al: Stratified exercise therapy for knee osteoarthritis Table 4 Relative improvements in primary outcome measures for each arm and subgroup. Exp (n = 151) Con (n = 177)a Total HMS LMS OS Total HMS LMS OS (n = 151) (n = 63) (n = 54) (n = 34) (n = 171) (n = 65)b (n = 53)b (n = 53)b Knee pain severity (NRS) –20% –19% –23% –14% –23% –30% –13% –26% 3 mths follow-up –29% –26% –34% –30% –26% –30% –29% –21% 12 mths follow-up 119% 16% 19% 117% 111% 111% 18% 113% Physical function (KOOS ADL) 114% 18% 117% 122% 117% 114% 114% 120% 3 mths follow-up 12 mths follow-up ADL = activities of daily living, Con = control group, Exp = experimental group, HMS = high muscle strength subgroup, KOOS = Knee Injury and Osteoarthritis Outcome Score, LMS = low muscle strength subgroup, NRS = numerical rating scale, OS = obesity subgroup. a subgroup allocation was unknown for six participants in control arm. b control arm was unaware and did not use stratification algorithm for subgroup allocation. Major protocol violations were identified in 34 (23%) and 11 (6%) intervention), 87 (49%) (median 0, IQR 0 to 2) for secondary care and in the experimental and control arm, respectively (see Appendix 8 on 14 (8%) (median 0, IQR 0 to 0) for alternative care. the eAddenda). The most frequently reported major protocol viola- tion in the experimental arm was ‘too few physiotherapy sessions’ Discussion (with minimum number of sessions differing across subgroups) (n = 30) and ‘less than two physiotherapy sessions’ (n = 5) in the control In this large, pragmatic cluster randomised controlled trial in a arm. Minor protocol violations were identified in 52 (34%) partici- primary care setting, the clinical effectiveness of a new model of pants in the experimental arm, mostly due to ‘no booster session stratified exercise therapy for an HMS, LMS and OS (supplemented provided’ (n = 47). No minor protocol deviations were defined in the by a dietary intervention for the OS) was evaluated in people with control arm. No adverse events were reported. knee OA. This model of stratified exercise therapy was based on effective exercise programs,11,12 well-accepted3 and empirically Other healthcare utilisation observed phenotypes,9,10 and a valid stratification algorithm.14 In contrast to the hypothesis, the promising results in a pilot study13 During the 12-month follow-up period, 101 participants (67%) in and the model being perceived by patients and therapists as the experimental arm received general practitioner care (median highly applicable,15 it showed no added value regarding clinical number of consults 1, IQR 0 to 3), 30 (20%) received primary care outcomes compared with usual exercise therapy, neither in the total (other than experimental intervention) (median 0, IQR 0 to 0), 71 sample nor in any of the three subgroups. Although the study (47%) received secondary care (median 0, IQR 0 to 2) and 15 (10%) estimated effects on two secondary outcome measures in favour of received alternative care (median 0, IQR 0 to 0). In the control arm, the experimental arm, these were clearly too small to be clinically this was 122 (69%) (median 2, IQR 0 to 3) for general practitioner, 33 important. (19%) (median 0, IQR 0 to 0) for primary care (other than the control Table 6 Treatment fidelity. Exp (n = 151) Con (n = 171)a Total HMS LMS OS Total HMS LMS OS (n = 151) (n = 63) (n = 54) (n = 34) (n = 171) (n = 65)b (n = 53)b (n = 53)b Physiotherapy treatment Physiotherapy sessions Number of sessions (12-mth period), mean (SD) 8.4 (4.7) 5.3 (3.1) 9.6 (3.6) 12.2 (5.1) 9.6 (4.8) 9.9 (5.0) 9.4 (4.4) 9.9 (4.7) Fewer sessions than recommended, n (%) 30 (20) 3 (5) 12 (22) 15 (44) n/a n/a n/a n/a More sessions than recommended, n (%) 15 (10) 9 (14) 4 (12) n/a n/a n/a n/a 2 (4) Type of physiotherapy interventions, n (%) Patient education 141 (95) 56 (92) 51 (94) 34 (100) 141 (84) 50 (81) 47 (89) 42 (84) Home exercises 137 (91) 56 (92) 48 (89) 33 (97) 135 (80) 50 (81) 42 (79) 41 (82) Exercise therapy (any) 132 (89) 46 (75) 52 (96) 34 (100) 168 (96) 59 (95) 51 (96) 49 (98) 129 (87) 43 (71) 52 (96) 34 (100) 155 (92) 57 (92) 49 (93) 47 (94) Strength training 78 (52) 14 (23) 36 (67) 28 (82) 105 (63) 36 (58) 62 (33) 35 (70) Aerobic training 90 (60) 27 (44) 38 (70) 25 (74) 100 (60) 40 (65) 31 (59) 29 (58) Functional training 92 (62) 27 (44) 37 (69) 28 (82) 135 (80) 50 (81) 43 (81) 41 (82) Balance/ stabilisation training Other interventions (any) 33 (22) 13 (21) 11 (20) 9 (27) 63 (38) 22 (36) 20 (38) 20 (40) Active/passive mobilisation 27 (18) 10 (16) 11 (20) 6 (18) 44 (26) 13 (21) 13 (25) 17 (34) Massage 9 (6) 3 (5) 2 (4) 4 (12) 18 (11) 9 (15) 5 (9) 3 (6) Taping 2 (1) 0 (0) 1 (2) 1 (3) 15 (9) 6 (10) 7 (13) 2 (4) TENS 0 (0) 0 (0) 0 (0) 0 (0) 3 (2) 0 (0) 1 (2) 2 (4) Dry needling 0 (0) 0 (0) 0 (0) 0 (0) 1 (1) 0 (0) 0 (0) 1 (2) Dietary intervention Number of sessions (12-mth period), mean (SD) n/a n/a n/a 3.2 (1.8) n/a n/a n/a n/a Total minutes, mean (SD) n/a n/a n/a 105 (5) n/a n/a n/a n/a Fewer sessions than recommended, n (%) n/a n/a n/a 10 (29) n/a n/a n/a n/a More sessions than recommended, n (%) n/a n/a n/a 0 (0) n/a n/a n/a n/a Con = control group, Exp = experimental group, HMS = high muscle strength subgroup, LMS = low muscle strength subgroup, OS = obesity subgroup, TKA = total knee arthroplasty. a subgroup allocation was unknown for six participants in the control arm. b control arm was unaware and did not use stratification algorithm for subgroup allocation.

Research 189 Stratified care is considered to have major potential to optimise therapy is not as high as proposed, with usual exercise therapy current treatment effects in knee OA and is a research priority in the already being the best available treatment option. Therefore, further field.8,30 It is believed that this is the first study to perform a research on stratified approaches of exercise therapy seems to be less randomised controlled trial evaluating a stratified approach to ex- relevant. Instead, it is recommended that future research focuses on ercise therapy in knee OA. The results align with trials in other optimising clinical reasoning process and applying a personalised musculoskeletal patient groups where stratified care appeared to approach by physiotherapists, such as tailoring exercise therapy to have no31–34 to minimal29 added value over usual care. Results from the major comorbidity.42 ongoing trials35–37 on stratified exercise therapy are expected soon. A slightly different approach, namely a stepped care approach, has This trial demonstrated no added value regarding clinical out- recently been tested in knee OA, demonstrating an unclear effect.38 comes of the model of stratified exercise therapy compared with This means that the evidence underlying stratified exercise therapy usual exercise therapy. This could be attributed to the experimental- in musculoskeletal patient groups is currently very limited. arm therapists facing difficulty in effectively applying the model (especially in the OS) and to elements of the model possibly being The lack of difference in effects in this study can be attributed to applied in the control arm. the contrast between both interventions being smaller than intended for two reasons. First, the experimental intervention seemed to not What was already known on this topic: There is strong, have been provided fully as planned, especially in the OS. In this high-quality evidence for the effectiveness of exercise therapy on subgroup, a substantial proportion (ie, 44% regarding physiotherapy knee pain and physical function. The population of people with sessions, 29% regarding dietary sessions) received too few sessions. knee osteoarthritis is heterogeneous but it contains homoge- This may have impeded the potential of the intervention, thereby neous phenotypes or subgroups. reducing the contrast with the control arm. In addition, the qualitative What this study adds: Stratification of exercise therapy so evaluation15 revealed several barriers in applying the stratified care that it is tailored to subgroups with well-preserved muscle model, which is in line with qualitative evaluations regarding the strength, low muscle strength or obesity does not substantially most-applied other stratified care model (ie, STarT BACK approach in improve clinical outcomes in people with knee osteoarthritis. low back pain).39,40 Again, especially in the OS, multiple barriers were observed, namely physiotherapists being hesitant to address obesity, Footnotes: a StataSE 16, StataCorp, College Station, TX, USA. and both physiotherapists and dieticians reporting barriers regarding b IBM SPSS statistics 27, IBM Corp., Armonk, USA. their interprofessional collaboration. These barriers, in addition to the eAddenda: Table 5 and Appendices 1 to 8 can be found online at lower number of provided sessions, might have been responsible for https://doi.org/10.1016/j.jphys.2022.06.005 the fact that, in contrast with other studies combining exercise with Ethics approval: This study was approved by the Medical Ethical diet (eg,41), no participant in the OS reached the intended 10% weight Committee of the VU University Medical Centre (2018.563). This loss and clinical effects in this subgroup were lower than expected. It study was conducted in agreement with the Declaration of Helsinki should be noted that participants in the current trial initially con- (2013), in accordance with the Dutch Medical Research Involving sulted their physiotherapist for their knee symptoms, with no inten- Human Subjects Act (WMO), and the General Data Protection Regu- tion of additionally following a diet intervention from a dietician. It lation (in Dutch: Algemene Verordening Gegevensbescherming, was also found that across all subgroups, a large number of partici- AVG). Written informed consent was obtained from each subject, pants (31%) did not receive the recommended booster session(s), due after an information letter had been provided. The researchers made to a lack of perceived necessity by the physiotherapists.15 The concise sure that the participants were given complete, adequate, written and training of physiotherapists (ie, e-learning course on the physio- oral information regarding the nature, aims, possible risks and ben- therapy guideline and a 4-hour training course on the experimental efits of the study. The participants were informed that they were free intervention, supplemented by online videos and site visits by the to interrupt their participation in the study at any moment without researcher) may have been insufficient. By addressing the barriers any consequence, and that they were able to receive a digital copy of more extensively in this training of physiotherapists, and by providing their personal data. Participants received a copy of the information tools and resources that support physiotherapists and dieticians, the sheet and informed consent form. An independent expert was effectiveness of the experimental intervention might improve. appointed to provide participants with the opportunity to ask ques- tions about the study to someone not related to the study. A second reason for the smaller contrast is that it is possible that Competing interests: RWJG Ostelo is member of the editorial elements of the stratified model were already integrated in usual board of Journal of Physiotherapy. He was also member of the Scien- care. In contrast to our hypothesis, physiotherapists no longer seem to tific Board Physical Therapy of the Royal Dutch Society for Physical apply a ‘one-size-fits-all’ but a personalised approach, meaning that Therapy until March 2021, which is the funder of this project. All the treatment is tailored to the person-specific rather than subgroup- other authors declare that they have no competing interests. specific patient factors, needs and preferences. Such a personalised Sources of support: This study was funded by the Scientific Board approach could make a stratified, subgroup-specific approach Physical Therapy of the Royal Dutch Society for Physical Therapy redundant. Moreover, especially in physiotherapy, considering the (grant number: WCF.7201.01.2017.01). The funder played no role in wide range of factors that should be taken into account, a stratified the design, conduct or reporting of this study. approach with only two factors seems to be too simplistic and could Acknowledgements: We thank all participating patients, physio- even hinder the process of personalisation and clinical reasoning by therapists and dieticians for their participation and effort, and Dr limiting physiotherapists in their treatment options. Therefore, this Lothar Kuiper for conducting the randomisation and supervising the stratified approach may only be of added value in physiotherapy for blinded analyses. only a few specific purposes, such as for deciding for which patient Provenance: Not invited. Peer reviewed. minimal supervision is sufficient and for which patient other Correspondence: Jesper Knoop, Department of Health Sciences, healthcare professionals (eg, dietician) need to be consulted. It is also Vrije Universiteit Amsterdam, Amsterdam, Netherlands. Email: possible that in this process of personalisation, physiotherapists in [email protected] the control arm tailored their treatment to the two stratification factors of the model (ie, upper leg muscle strength and BMI), which References further minimised the contrast between arms. It should be noted that the control-arm physiotherapists, who were very keen to participate 1. Allen KD, Golightly YM. State of the evidence. Curr Opin Rheumatol. 2015;27:276– in this trial, may have had more expertise in providing knee OA 283. https://doi.org/10.1097/BOR.0000000000000161 treatments and were therefore not fully generalisable compared with physiotherapists outside the study. Based on these results and recent other trials,29,31–34 it was concluded that the potential of stratified, subgroup-specific exercise

190 Knoop et al: Stratified exercise therapy for knee osteoarthritis 2. Felson DT. Identifying different osteoarthritis phenotypes through epidemiology. 24. de Groot IB, Favejee MM, Reijman M, Verhaar JA, Terwee CB. The Dutch version of Osteoarthritis Cartilage. 2010;18:601–460. https://doi.org/10.1016/j.joca.2010.01.007 the Knee Injury and Osteoarthritis Outcome Score: a validation study. Health Qual Life Outcomes. 2008;26:16. 3. Bijlsma JW, Berenbaum F, Lafeber FP. Osteoarthritis: an update with relevance for clinical practice. Lancet. 2011;377:2115–2126. https://doi.org/10.1016/S0140- 25. Khanna D, Krishnan E, Dewitt EM, Khanna PP, Spiegel B, Hays RD. The future of 6736(11)60243-2 measuring patient-reported outcomes in rheumatology: Patient-Reported Out- comes Measurement Information System (PROMIS). Arthritis Care Res. 4. Bannuru RR, Osani MC, Vaysbrot EE, Arden NK, Bennell K, Bierma-Zeinstra SMA, 2011;63(Suppl 11):S486–S490. et al. OARSI guidelines for the non-surgical management of knee, hip, and poly- articular osteoarthritis. Osteoarthritis Cartilage. 2019;27:1578–1589. https://doi. 26. Knoop J, van der Leeden M, van der Esch M, Thorstensson CA, Gerritsen M, org/10.1016/j.joca.2019.06.011 Voorneman RE, et al. Association of lower muscle strength with self-reported knee instability in osteoarthritis of the knee: results from the Amsterdam 5. Fransen M, McConnell S, Harmer AR, Van der Esch M, Simic M, Bennell KL. Exercise Osteoarthritis cohort. Arthritis Care Res. 2012;64:38–45. https://doi.org/10.1002/ for osteoarthritis of the knee. Cochrane Database Syst Rev. 2015;1:CD004376. acr.20597 https://doi.org/10.1002/14651858.CD004376.pub3 27. Dobson F, Hinman RS, Hall M, Marshall CJ, Sayer T, Anderson C, et al. Reliability and 6. Knoop J, Dekker J, van der Leeden M, van der Esch M, Klein JP, Hunter DJ, et al. Is measurement error of the Osteoarthritis Research Society International (OARSI) the severity of knee osteoarthritis on magnetic resonance imaging associated with recommended performance-based tests of physical function in people with hip outcome of exercise therapy? Arthritis Care Res. 2014;66:63–68. https://doi.org/10. and knee osteoarthritis. Osteoarthritis Cartilage. 2017;25:1792–1796. https://doi. 1002/acr.22128 org/10.1016/j.joca.2017.06.006 7. Skou ST, Roos EM, Laursen MB, Rathleff MS, Arendt-Nielsen L, Rasmussen S, et al. 28. Borg G. Perceived exertion as an indicator of somatic stress. Scand J Rehabil Med. Total knee replacement and non-surgical treatment of knee osteoarthritis: 2-year 1970;2:92–98. outcome from two parallel randomized controlled trials. Osteoarthritis Cartilage. 2018;26:1170–1180. https://doi.org/10.1016/j.joca.2018.04.014 29. Hill JC, Whitehurst DG, Lewis M, Bryan S, Dunn KM, Foster NE, et al. Comparison of stratified primary care management for low back pain with current best practice 8. Foster NE, Hill JC, O’Sullivan P, Hancock M. Stratified models of care. Best Pract Res (STarT Back): a randomised controlled trial. Lancet. 2011;378(9802):1560–1571. Clin Rheumatol. 2013;27:649–661. https://doi.org/10.1016/j.berh.2013.10.005 https://doi.org/10.1016/S0140-6736(11)60937-9 9. Knoop J, van der Leeden M, Thorstensson CA, Roorda LD, Lems WF, Knol DL, et al. 30. Hay E, Dziedzic K, Foster N, Peat G, van der Windt D, Bartlam B, et al. Optimal Identification of phenotypes with different clinical outcomes in knee osteoar- primary care management of clinical osteoarthritis and joint pain in older peo- thritis: data from the Osteoarthritis Initiative. Arthritis Care Res. 2011;63:1535– ple: a mixed-methods programme of systematic reviews, observational and 1542. https://doi.org/10.1002/acr.20571 qualitative studies, and randomised controlled trials. Prog Grants Appl Res. 2018;6:1–260. 10. van der Esch M, Knoop J, van der Leeden M, Roorda LD, Lems WF, Knol DL, et al. Clinical phenotypes in patients with knee osteoarthritis: a study in the Amsterdam 31. Foster NE, Mullis R, Hill JC, Lewis M, Whitehurst DG, Doyle C, et al. IMPaCT Back osteoarthritis cohort. Osteoarthritis Cartilage. 2015;23:544–549. https://doi.org/10. Study team. Effect of stratified care for low back pain in family practice (IMPaCT 1016/j.joca.2015.01.006 Back): a prospective population-based sequential comparison. Ann Fam Med. 2014;12:102–111. https://doi.org/10.1370/afm.1625 11. Knoop J, Dekker J, van der Leeden M, van der Esch M, Thorstensson CA, Gerritsen M, et al. Knee joint stabilization therapy in patients with osteoarthritis of 32. Cherkin D, Balderson B, Wellman R, Hsu C, Sherman KJ, Evers SC, et al. Effect of Low the knee: a randomized, controlled trial. Osteoarthritis Cartilage. 2013;21:1025– Back Pain Risk-Stratification Strategy on Patient Outcomes and Care Processes: the 1034. https://doi.org/10.1016/j.joca.2013.05.012 MATCH Randomized Trial in Primary Care. J Gen Intern Med. 2018;33:1324–1336. https://doi.org/10.1007/s11606-018-4468-9 12. de Rooij M, van der Leeden M, Cheung J, van der Esch M, Häkkinen A, Haverkamp D, et al. Efficacy of tailored exercise therapy on physical functioning in 33. Foster NE, Konstantinou K, Lewis M, Ogollah R, Saunders B, Kigozi J, et al. Stratified patients with knee osteoarthritis and comorbidity: a randomized controlled trial. versus usual care for the management of primary care patients with sciatica: the Arthritis Care Res. 2017;69:807–816. https://doi.org/10.1002/acr.23013 SCOPiC RCT. Health Technol Assess. 2020;24:1–130. https://doi.org/10.3310/ hta24490 13. Knoop J, van der Leeden M, van der Esch M, de Rooij M, Peter WF, Bennell KL, et al. Is a model of stratified exercise therapy by physical therapists in primary care 34. Delitto A, Patterson CG, Stevans JM, Freburger JK, Khoja SS, Schneider MJ, et al. feasible in patients with knee osteoarthritis? A mixed methods study. Physio- Stratified care to prevent chronic low back pain in high-risk patients: The TARGET therapy. 2019. https://doi.org/10.1016/j.physio.2019.01.013 trial. A multi-site pragmatic cluster randomized trial. E Clin Med. 2021;34:100795. https://doi.org/10.1016/j.eclinm.2021.100795 14. Knoop J, Ostelo RWJG, van der Esch M, de Zwart A, Bennell KL, van der Leeden M, et al. Construct validity of the OCTOPuS stratification algorithm for allocating pa- 35. Hill J, Garvin S, Chen Y, Cooper V, Wathall S, Bartlam B, et al. Computer-Based tients with knee osteoarthritis into subgroups. BMC Musculoskelet Disord. Stratified Primary Care for Musculoskeletal Consultations Compared With Usual 2021;22:633. https://doi.org/10.1186/s12891-021-04485-1 Care: Study Protocol for the STarT MSK Cluster Randomized Controlled Trial. JMIR Res Protoc. 2020;9:e17939. https://doi.org/10.2196/17939 15. Knoop J, de Joode JW, Brandt H, Dekker J, Ostelo RWJG. Patients’ and clinicians’ experiences with stratified exercise therapy in knee osteoarthritis: a qualitative 36. van Tilburg ML, Kloek CJJ, Pisters MF, Staal JB, van Dongen JM, de Weerd M, et al. study. BMC Musculoskelet Disord. 2022; in press. https://doi.org/10.1186/s12891- Stratified care integrated with eHealth versus usual primary care physiotherapy in 022-05496-2 patients with neck and/or shoulder complaints: protocol for a cluster randomized controlled trial. BMC Musculoskelet Disord. 2021;22:143. https://doi.org/10.1186/ 16. Knoop J, Dekker J, van der Leeden M, de Rooij M, Peter WFH, van Bodegom- s12891-021-03989-0 Vos L, et al. Stratified exercise therapy compared with usual care by physical therapists in patients with knee osteoarthritis: A randomized controlled trial 37. Koppenaal T, Arensman RM, van Dongen JM, Ostelo RWJG, Veenhof C, Kloek CJJ, protocol (OCTOPuS study). Physiother Res Int. 2019:e1819. https://doi.org/10. et al. Effectiveness and cost-effectiveness of stratified blended physiotherapy in 1002/pri.1819 patients with non-specific low back pain: study protocol of a cluster randomized controlled trial. BMC Musculoskelet Disord. 2020;21:265. https://doi.org/10.1186/ 17. Hopewell S, Clarke M, Moher D, Wager E, Middleton P, Altman DG, at al. CONSORT s12891-020-3174-z for reporting randomized controlled trials in journal and conference abstracts: explanation and elaboration. PLoS Med. 2008;5:e20. https://doi.org/10.1371/ 38. Allen KD, Woolson S, Hoenig HM, Bongiorni D, Byrd J, Caves K, et al. Stepped Ex- journal.pmed.0050020 ercise Program for Patients With Knee Osteoarthritis: A Randomized Controlled Trial. Ann Intern Med. 2021;174:298–307. https://doi.org/10.7326/M20-4447 18. Weijer C, Grimshaw JM, Taljaard M, Binik A, Boruch R, Brehaut JC, et al. Ethical issues posed by cluster randomized trials in health research. Trials. 2011;12:100. 39. Karstens S, Kuithan P, Joos S, Hill JC, Wensing M, Steinhäuser J, et al. Physiother- https://doi.org/10.1186/1745-6215-12-100 apists’ views of implementing a stratified treatment approach for patients with low back pain in Germany: a qualitative study. BMC Health Serv Res. 2018;18:214. 19. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al. Development https://doi.org/10.1186/s12913-018-2991-3 of criteria for the classification and reporting of osteoarthritis. Classification of osteoarthritis of the knee. Diagnostic and Therapeutic Criteria Committee 40. Middleton A, Fitzgerald GK, Delitto A, Saper RB, Gergen Barnett K, Stevans J. of the American Rheumatism Association. Arthritis Rheum. 1986;29:1039– Implementing stratified care for acute low back pain in primary care using the 1049. STarT Back instrument: a process evaluation within the context of a large prag- matic cluster randomized trial. BMC Musculoskelet Disord. 2020;21:776. https://doi. 20. van Binsbergen JJ, Langens FNM, Dapper ALM, Van Halteren MM, Glijsteen R, org/10.1186/s12891-020-03800-6 Cleyndert GA, et al. NHG-standaard Obesitas. Huisarts Wet. 2010;53:609–625. 41. Messier SP, Beavers DP, Mihalko SL, Miller GD, Lyles MF, Hunter DJ, et al. The effects 21. Peter WF, Jansen MJ, Hurkmans EJ, Bloo H, Dekker J, Dilling RG, et al. Guideline of intensive dietary weight loss and exercise on gait in overweight and obese Steering Committee - Hip and Knee Osteoarthritis. Physiotherapy in hip and knee adults with knee osteoarthritis. The Intensive Diet and Exercise for Arthritis (IDEA) osteoarthritis: development of a practice guideline concerning initial assessment, trial. J Biomech. 2020;98:109477. https://doi.org/10.1016/j.jbiomech.2019.109477. treatment and evaluation. Acta Reumatol Port. 2011;36:268–281. Epub 2019 Oct 31. 22. McCaffery M, Pasero C. Using the 0-to-10 pain rating scale. Am J Nurs. 42. van der Leeden M, Stuiver MM, Huijsmans R, Geleijn E, de Rooij M, Dekker J. 2001;101:81–82. Structured clinical reasoning for exercise prescription in patients with comorbidity. Disabil Rehabil. 2020;42:1474–1479. https://doi.org/10.1080/09638288.2018.1527953 23. Roos EM, Roos HP, Lohmander LS, Ekdahl C, Beynnon BD. Knee Injury and Osteo- arthritis Outcome Score (KOOS)-development of a self-administered outcome measure. J Orthop Sports Phys Ther. 1998;28:88–96.

Journal of Physiotherapy 68 (2022) 197–202 j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j p hy s Research Trunk stabilising exercises promote sternal stability in patients after median sternotomy for heart valve surgery: a randomised trial El-Sayed Essam El-Sayed Felaya a, Eman Hassan Abd Al-Salam b, Alshaymaa Shaaban Abd El-Azeim c a Department of Physical Therapy for Internal Medicine, Faculty of Physical Therapy, Cairo University, Giza, Egypt; b Department of Diagnostic Imaging, National Heart Institute, Giza, Egypt; c Department of Basic Science, Faculty of Physical Therapy, Cairo University, Giza, Egypt KEY WORDS ABSTRACT Trunk stabilising exercises Question: What is the effect of trunk stabilising exercises on sternal stability in women who have undergone Sternal instability heart valve surgery via median sternotomy? Design: Randomised controlled trial with concealed allocation, Median sternotomy assessor blinding and intention-to-treat analysis. Participants: Thirty-six women aged 40 to 50 years who Heart valve surgery had undergone heart valve surgery via median sternotomy 7 days before enrolment. Intervention: All Randomised trial participants in both groups received cardiac rehabilitation during hospitalisation and three times per week for 4 weeks after discharge. In addition, participants in the experimental group were prescribed a regimen of trunk stabilising exercises to be performed three times per week for 4 weeks. At each exercise session, each of 11 exercises were to be performed with five to ten repetitions. Outcome measures: The primary outcome was sternal separation (the distance between the two halves of the bisected sternum). The secondary outcome was the Sternal Instability Scale from 0 (no instability) to 3 (an unstable sternum with substantial movement or separation). Measures were taken before and after the 4-week intervention period. Results: After the 4-week intervention period, the experimental group had a greater decrease in sternal separation by 0.09 cm (95% CI 0.07 to 0.11). The experimental group was twice as likely to improve by at least one grade on the Sternal Instability Scale by 4 weeks (RR 2.00, 95% CI 1.07 to 3.75). The experimental group was almost three times as likely to have a clinically stable sternum (grade 0 on the Sternal Instability Scale) by 4 weeks (RR 2.75, 95% CI 1.07 to 7.04). Conclusion: Trunk stabilising exercises were an effective and feasible method of promoting sternal stability in women who underwent heart valve surgery via median sternotomy. Trial registration: NCT04632914. [Essam El-Sayed Felaya E, Abd Al-Salam EH, Shaaban Abd El-Azeim A (2022) Trunk stabilising exercises promote sternal stability in patients after median sternotomy for heart valve surgery: a randomised trial. Journal of Physiotherapy 68:197–202] © 2022 Australian Physiotherapy Association. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Introduction healing complications to become major (ranging from 0.3 to 5% of cases), major complications are very disruptive.11 Sternal instability Valvular heart disease is a growing health issue, with an overall often occurs during the first 2 weeks after open heart surgery, after prevalence of 2 to 5% among the general population and a prevalence which healing of the sternum should lead to sternal stability.12 Poor, of 13% in people aged . 75 years.1,2 Heart valve surgery can poten- incomplete or delayed healing of the divided sternum can prolong the tially be a life-saving intervention for people with serious symp- hospitalisation of patients, increase healthcare expenditure and tomatic valve disease.1 restrict the return of patients to work or social activities. Further- more, delayed sternal healing increases the likelihood of infection Up to 27% of patients undergoing valve surgery may need reho- deep in the sternal wound.13 spitalisation within 30 days of discharge from hospital.3–6 Some postoperative patients experience distress and fear regarding hospital A view of the underlying non-physiological motion of the edges of readmission, re-operation, post-surgical complications or decondi- the divided sternum is helpful to guide recommendations for in- tioning,7 which may preclude or hinder their functional recovery and terventions to reduce excessive motion of the sternal edges and affect their activities of daily living.8 thereby usually minimise the symptoms; this situation is especially important in the case of patients whose sternum cannot be treated Complications related to the surgical approach can often be surgically.14 identified early after heart valve surgery and increase the need for unexpected healthcare contacts and hospital readmission.9,10 Healing Trunk stabilising exercises recruit and facilitate the muscles of the complications after median sternotomy include instability in the abdomen and anterior chest wall to help in the stabilisation of the sternum, non-union and wound infection. Although it is rare for divided sternum, with the intention of minimising unwanted motion https://doi.org/10.1016/j.jphys.2022.06.002 1836-9553/© 2022 Australian Physiotherapy Association. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).

198 Essam El-Sayed Felaya et al: Trunk stabilising exercises after median sternotomy in both the sagittal and horizontal planes during trunk movements. provided by a trained physiotherapist. Trunk stabilising exercises Patients with sternal instability after open heart surgery often have commenced on the seventh day after surgery and were adminis- symptoms of pain and restlessness; these can be minimised by tered to every patient from different positions (supine, sitting and reducing the amount of improper displacement between the edges of standing). Participants were prescribed 5 to 10 repetitions of each the bisected sternum, thus providing conservative therapy for post- exercise, one session per day, 3 days per week, for 4 consecutive operative sternal instability.15 weeks. El-Ansary and colleagues (2007), who conducted a study to correlate pain with upper limb movement in patients suffering from Few studies have investigated the effects of therapeutic modalities sternal instability after cardiac surgery, reported that the degree of and nonpharmacological treatment on sternal stability after heart sternal separation was not correlated with the type of upper limb valve surgery. Therefore, this study aimed to implement trunk sta- activity, but both unilateral and unilateral loaded upper extremity bilising exercises in patients who had undergone heart valve surgery movements were associated with more sternal pain than bilateral via median sternotomy, and then evaluate the effect of this thera- movements.18 Accordingly, the trunk stabilising exercises were peutic modality on sternal stability. applied to the patients in line with the recommendations from that study. The trunk stabilising exercises are presented in Appendix 1 Therefore, the research question for this randomised trial was: on the eAddenda. Participants in the experimental group also un- derwent a routine cardiac rehabilitation program, described below. What is the effect of trunk stabilising exercises on sternal stability in women who have undergone heart valve surgery via median Control group sternotomy? Participants in the control group were not shown, taught or pre- Method scribed the trunk stabilising exercises that were prescribed for the experimental group. Participants in the control group underwent a Design routine cardiac rehabilitation program only. Inpatient cardiac reha- bilitation (phase I) was conducted for patients during the hospital- An assessor-blinded, parallel, two-group, randomised controlled isation period; it comprised early progressive mobilisation of the trial enrolled patients who underwent heart valve surgery via median patient to be independent in basic household activities. The program sternotomy. On day 7 after surgery, participants were randomly also provided concise instructions on: the nature of the disease, allocated to one of two groups via a computer-generated, concealed rehabilitation, control of risk factors and follow-up scheduling.19 After allocation schedule. Participants allocated to the experimental group discharge, the patients commenced the outpatient phase (phase II) of were taught and prescribed trunk stabilising exercises in addition to a the cardiac rehabilitation program for 40 to 60 minutes, three times cardiac rehabilitation program. Participants allocated to the control per week for 4 consecutive weeks. The exercise undertaken during group received the cardiac rehabilitation program only. Both groups this phase used an arm ergometer, a leg ergometer and a treadmill. As received the same medical care and physical rehabilitation program described by Bartels,20 a moderate-intensity program was prescribed during their hospitalisation period. Clinical data were measured at by establishing a target heart rate (in beats/minute) of 75 to 80% of baseline and 4 weeks after the baseline measure. Data measurement maximum heart rate (defined as 220 – age in years). In the presence and extraction were each carried out without knowledge of the group to which each participant belonged. of b-blockers, the target HR was 20 beats/minute above the resting Participants, therapists, centres heart rate. Each patient’s heart rate during the cardiac rehabilitation sessions was monitored via a telemetry system.20 This research was conducted at the National Heart Institute in Giza, Egypt, on patients who were undergoing heart valve surgery. All Outcome measures the patients referred to the National Heart Institute between November 2020 and January 2021 for heart valve surgery were The baseline characteristics collected for each participant were screened for possible participation in this study if they met the in- age, weight, height and BMI. clusion criteria. Patients at the National Heart Institute who undergo elective heart valve surgery receive a longitudinal median sternotomy Primary outcome and a period of extracorporeal circulation. The primary outcome was sternal separation. An ultrasound unit The inclusion criteria were: age between 40 and 50 years; female was used to assess the transverse sternal separation distance (ie, gender; haemodynamic stability; body mass index (BMI) between 25 between the two halves of the sternum). The point of greatest sep- and 29.9 kg/m2; and acute sternal instability following cardiac sur- aration of the two halves of the sternum was marked for each patient gery. The presence of sternal instability was based on the patients in the supine lying position with the head of the treatment bed at 30 having had recent median sternotomy and then confirmed as sternal inclination.18,21,22 The separation at this point was then quantified in separation measured by ultrasound at baseline. The exclusion criteria centimetres from the projected images of the ultrasound unit, as were: previous thoracic surgery; elective and urgent coronary artery shown in Figures 1 and 2. bypass surgery; respiratory insufficiency after surgery, manifesting as hypoxaemia (partial pressure of oxygen in arterial blood , 60 Secondary outcome mmHg);13 renal insufficiency after surgery (serum creatinine  1.8 The Sternal Instability Scale is a clinical physical evaluation mg/dl);16 low cardiac output syndrome with ST-segment elevation in multiple electrocardiogram leads, cardiac arrhythmias or hypotension instrument that is designed to evaluate the integrity of the ster- according to the American College of Cardiology Foundation and num and transfer the results of the assessment to an acceptable American Heart Association;17 other medical conditions such as dia- grade. It consists of a 4-point scale bounded by grade 0 (a clinically betes, uncontrolled hypertension and obesity; and a history of con- stable sternum with no noticeable motion or separation of the ditions that may have affected the provision of physiotherapy edges of the sternum) and grade 3 (a separated sternum with interventions such as severe asthma, chronic airflow limitation, substantial movement or separation between the two halves of the bronchiectasis, ankylosing spondylitis or lumbar disc prolapse. sternum).21 Intervention Data analysis Experimental group The number of participants required for this trial was calculated The experimental intervention consisted of trunk stabilising using commercial softwarea, based on identifying a standardised ef- fect size of 1.05, with the anticipated standard deviation calculated exercises in addition to a routine cardiac rehabilitation program from the pilot study on 10 patients.

Research 199 Figure 1. Pre-treatment ultrasound assessment. Figure 2. Post-treatment ultrasound assessment. Analyses were performed using commercial softwarea. The Effect of the intervention continuous data (age, height, weight, BMI and sternal separation) were subjected to a test of normality (Shapiro-Wilk test). All data Primary outcome were normally distributed within each group. Therefore, the At Week 4, sternal separation had improved by 0.10 cm (SD 0.04) continuous variables were summarised using means and standard deviations. The effect of the intervention on sternal separation was in the experimental group and 0.01 cm (SD 0.02) in the control group. estimated as the mean between-group difference with a 95% Therefore, the between-group difference in change provided an es- confidence interval. The Sternal Instability Scale provides ordinal timate of the effect of the exercise regimen on sternal separation: MD data, so this was analysed by comparing the proportion of par- 0.09 cm (95% CI 0.07 to 0.11) less separation in the experimental ticipants in each group who improved by at least one grade on the group (Table 2). scale by Week 4 and who exhibited a Grade 0 (stable sternum) at the Week 4 assessment. These proportions were compared be- Secondary outcomes tween groups and reported as relative risk with a 95% confidence When the Sternal Instability Scale scores were used to assess the interval. proportion of participants in each group who improved by at least Results one grade by Week 4, those in the experimental group were twice as likely to demonstrate this degree of improvement: RR 2.00 (95% CI Flow of participants through the trial 1.07 to 3.75). Figure 3 shows the design of the trial and flow of participants When the Sternal Instability Scale scores were used to assess the through the trial. A total of 43 female patients were screened for this proportion of participants in each group that had a grade 0 (stable study. During the screening for eligibility, seven patients were sternum) at Week 4, those in the experimental group were almost excluded for the reasons shown in Figure 3. A total of 36 patients met three times as likely to demonstrate this degree of improvement: RR the selection criteria and were randomly assigned to one of two 2.75 (95% CI 1.07 to 7.04). These results are presented in Table 3 and groups using the random ordered allocations in the sealed envelopes, individual participant data are presented in Table 4 on the eAddenda. which had a 1:1 ratio of experimental:control allocations. Discussion Characteristics of the participants This randomised clinical trial is the first to analyse the effects of The two groups were similar at baseline regarding age, weight, early postoperative trunk stabilising exercises on sternal instability height, and BMI (Table 1). The two groups were also similar regarding and separation in patients who underwent median sternotomy for their baseline sternal separation (see the first two columns of data in cardiac valve surgery. The analyses of the trial’s data estimate that the Table 2) and their baseline scores on the Sternal Instability Scale average effect of the exercise program in this population is to reduce (Table 1). sternal separation by about 0.1 cm and to double the likelihood of improving by at least one grade on the Sternal Instability Scale. Furthermore, the exercise program leads to an almost threefold in- crease in the likelihood of having a clinically stable sternum 4 weeks after median sternotomy. Therefore, the exercise program could play a major role in increasing sternal stability after median sternotomy

200 Essam El-Sayed Felaya et al: Trunk stabilising exercises after median sternotomy Patients screened physically for inclusion (n = 43) Excluded (n = 7) • declined to participate (n = 3) • ineligible (n = 2) • other reason (n = 2) Measured sternal separation and the Sternal Instability Scale Week 0 Randomised (n = 36) (n = 18) (n = 18) Experimental group Control group • trunk stabilising • routine cardiac exercises rehabilitation program • routine cardiac • sternal rehabilitation precautions program • sternal precautions Week 4 Measured sternal separation and the Sternal Instability Scale (n = 18) (n = 18) Figure 3. Design and flow of participants through the trial. Table 1 the control group (RR 2.75). Again, the confidence interval spanned Baseline characteristics of the participants. from a very slight improvement to a very large increase in the likelihood of having a clinically stable sternum at Week 4 (1.07 to Characteristic Groups 7.04). Exp Con These enhancements in the trunk stabilising exercise group may (n = 18) (n = 18) be attributed to the bracing effect of the anterior trunk muscles that are oriented in the transverse direction. Therefore, in the lower trunk, Age (y), mean (SD) 46 48 the sacroiliac joint was efficiently locked.23–25 It was also suggested (3) (2) that the contraction of these transversely oriented muscles, such as Height (cm), mean (SD) 166 167 the transverse abdominis, can create forces that produce a ‘corset- (7) (12) like’ motion, which effectively retains sacroiliac joint stability and Weight (kg), mean (SD) 79 78 decrease its laxity.23–25 Additionally, trunk stabilising exercises re- (6) (12) cruit muscles in a horizontal direction, such as the transverse BMI (kg/m2), mean (SD) 28.5 27.8 abdominis muscle, which operates perpendicularly to the sagittal (1.4) (1.2) plane in which the sternotomy incision and the SIJ are similarly SIS score (0 to 3), n (%) directed, thereby preserving the stability of the trunk, particularly the 0 4 6 bisected sternum, as well as reducing sternal motion and 1 14 12 separation.23–25 2 0 0 3 0 0 El-Ansary and colleagues (2007) conducted a prospective rand- omised crossover trial on patients with chronic sternal instability BMI = body mass index, Con = control group, Exp = experimental group, SIS = Sternal (ie, lasting months to years after heart surgery) and found greater Instability Scale. degrees of complicated sternal motion and separation. Their research intervention comprised trunk stabilising exercises. Their for cardiac valve surgery, which should help the patient to regain findings revealed that sternal separation reduced in the supine lying their independence and re-commence activities of daily living. and sitting positions more during trunk stabilising exercise appli- cation than during the control period.15 Helmy and colleagues26 It is important to consider the magnitude and precision of these investigated the effects of low-level laser therapy versus trunk estimates of the treatment effects. The estimate of the effect on stabilising exercises on the post-sternotomy healing of 45 patients sternal separation (0.09 cm) had a very narrow confidence interval who underwent coronary artery bypass grafting surgery. Their re- (0.07 to 0.11), so the average effect in the wider population of people sults showed that both interventions were effective methods for undergoing valve surgery via median sternotomy is fairly certain to healing after surgery, with the laser therapy showing superiority. be about 0.1 cm less sternal separation. While this may seem very Furthermore, their effects were associated with a reduction in pain small, the average amount of separation was only a little over 0.2 cm and recovery of activities of daily living.26 These previous studies in both groups at baseline. The analyses of the Sternal Instability Scale reinforce the current results because they suggest that, in addition strongly favoured the trunk stabilising exercises but there was greater to trunk stabilising exercises after heart valve surgery accelerating uncertainty in the estimates. For example, although the experimental sternotomy healing, it can be anticipated that they would also group was twice as likely to improve by at least one grade on the reduce sternal pain and prevent complications that can arise during Sternal Instability Scale (RR 2.00), the confidence interval (1.07 to prolonged sternal healing. A limitation of the current trial was that 3.75) shows that the true average effect in the wider population may it did not measure pain, longer-term complications and return to lie somewhere between a very slight increase or almost a fourfold increase in the likelihood of improving by at least one grade. Partic- ipants in the experimental group were almost three times as likely to have a clinically stable sternum at Week 4, compared with those in

Research 201 Table 2 Between-group difference Mean (SD) of groups, mean (SD) within-group difference and mean (95% CI) between-group difference for sternal separation. Week 4 minus Week 0 Exp minus Con Outcome Groups Within-group difference –0.09 Week 4 minus Week 0 (–0.11 to –0.07) Week 0 Week 4 Exp Con Exp Con Exp Con (n = 18) (n = 18) (n = 18) (n = 18) Sternal separation (cm) 0.23 0.23 0.13 0.22 –0.10 –0.01 (0.04) (0.11) (0.03) (0.10) (0.04) (0.02) Con = control group, Exp = experimental group. Table 3 References Number (proportion) of participants in each group who improved their Sternal Instability Scale grade to achieve the outcomes shown, and the relative risk (95% CI) 1. Baumgartner H, Falk V, Bax JJ, et al. 2017 ESC/EACTS guidelines for the manage- between the groups. ment of valvular heart disease. Eur Heart J. 2017;38:2739–2791. https://doi.org/10. 1093/eurheartj/ehx391 Outcome Groups Relative risk (95% CI) 2. Nkomo VT, Gardin JM, Skelton TN, et al. Burden of valvular heart diseases: a Exp Con Exp relative to Con population-based study. Lancet. 2006;368:1005–1011. https://doi.org/10.1016/ (n = 18) (n = 18) S0140-6736(06)69208-8 Improvement  1 on Sternal 14 7 2.00 3. Danielsen SO, Moons P, Sandven I, et al. Thirty-day readmissions in surgical and Instability Scale (0.78) (0.39) (1.07 to 3.75) transcatheter aortic valve replacement: a systematic review and meta-analysis. Int Grade 0 on Sternal Instability J Cardiol. 2018;268:85–91. https://doi.org/10.1016/j.ijcard.2018.05.026 Scale at Week 4 11 4 2.75 (0.61) (0.22) (1.07 to 7.04) 4. Vejpongsa P, Bhise V, Charitakis K, et al. Early readmissions after transcatheter and surgical aortic valve replacement. Catheter Cardiovasc Interv. 2017;90:662–670. Con = control group, Exp = experimental group. https://doi.org/10.1002/ccd.26945 independent daily activities. However, the outcome measures that 5. Borregaard B, Dahl JS, Riber LPS, et al. Effect of early, individualised and intensified were included were analysed with complete follow-up of all rand- follow-up after open heart valve surgery on unplanned cardiac hospital read- omised participants. missions and all-cause mortality. Int J Cardiol. 2019;289:30–36. https://doi.org/10. 1016/j.ijcard.2019.02.056 In conclusion, this study showed that trunk stabilising exercises are an effective and feasible method of promoting sternal stability in 6. Borregaard B, Sørensen J, Ekholm O, et al. Sociodemographic, clinical and patient- women who underwent heart valve surgery via median sternotomy. reported outcomes and readmission after heart valve surgery. J Heart Valve Dis. The estimated effect on the sternal separation was small but very 2018;27:78–86. precise. Although the main estimates of the effects on the clinical measures of sternal stability were very favourable, they come with 7. Berg SK, Zwisler AD, Pedersen BD, Haase K, Sibilitz KL. Patient experiences of re- substantial uncertainty. covery after heart valve replacement: suffering weakness, struggling to resume normality. BMC Nurs. 2013;12:23. https://doi.org/10.1186/1472-6955-12-23 What was already known on this topic: Sternal instability after median sternotomy is common. Delayed or incomplete 8. Lapum J, Angus JE, Peter E, Watt-Watson J. Patients’ discharge experiences: healing of the divided sternum can prolong hospitalisation, Returning home after open heart surgery. Heart Lung. 2011;40:226–235. https:// increase healthcare costs and delay return to independent doi.org/10.1016/j.hrtlng.2010.01.001 activities of daily living. What this study adds: Among women who have had a ster- 9. Sibilitz KL, Berg SK, Thygesen LC, et al. High readmission rate after heart valve notomy for heart valve surgery, trunk stabilising exercises surgery: a nationwide cohort study. Int J Cardiol. 2018;189:96–104. https://doi.org/ commenced 7 days after surgery decreased the amount of sternal 10.1016/j.ijcard.2015.04.078 separation and increased clinical measures of sternal stability. 10. Lie I, Danielsen SO, Tonnessen T, et al. Determining the impact of 24/7 phone Footnotes: a G*Power version 3.1.9.2, G*Power software Inc, Kiel, support on hospital readmissions after aortic valve replacement surgery (the Germany. AVRre study): study protocol for a randomised controlled trial. Trials. 2017;18:246. https://doi.org/10.1186/s13063-017-1971-y b SPSS version 23.0, IBM Corp, New York, USA. eAddenda: Table 4 and Appendix 1 can be found online at https:// 11. Losanoff JE, Collier AD, Wagner-Mann CC, et al. Biomechanical comparison of doi.org/10.1016/j.jphys.2022.06.002 median sternotomy closures. Ann Thorac Surg. 2004;77:203–209. https://doi.org/ Ethics approval: This study was approved by the Institutional 10.1016/S0003-4975(03)01468-1 Ethics Committee of Physical Therapy Faculty, Cairo University, Giza, Egypt. Registration approval number: (P.T.REC/012/002895). A writ- 12. Tekümit H, Cenal AR, Tataroglu C, Uzun K, Akıncı E. Comparison of figure-of-eight ten declaration of informed consent was signed by all the patients and simple wire sternal closure techniques in patients with non-microbial sternal after surgery for participation and publication of results. The trial was dehiscence. Anatol J Cardiol/Anadolu Kardiyol Derg. 2009;9:411–416. conducted according to the Declaration of Helsinki. Competing interests: All authors completed a uniform disclosure 13. Iwakura A, Tabata Y, Miyao M, et al. Novel method to enhance sternal healing after form and declare that they have no conflicts of interest. harvesting bilateral internal thoracic arteries with use of basic fibroblast growth Source(s) of support: Nil. factor. Circ. 2000;102:III307–III311. https://doi.org/10.1161/01.CIR.102.suppl_3.III-307 Acknowledgements: The authors thank all the patients and team members in the National Heart Institute for their general and tech- 14. El-Ansary D, Adams R, Waddington G. Sternal instability during arm elevation nical support. observed as dynamic, multiplanar separation. Int J Ther Rehabil. 2009;16:609–614. Provenance: Not invited. Peer reviewed. https://doi.org/10.12968/ijtr.2009.16.11.44942 Correspondence: El-Sayed Essam El-Sayed Felaya, Department of Physical Therapy for Internal Medicine, Faculty of Physical Therapy, 15. El-Ansary D, Waddington G, Adams R. Trunk stabilisation exercises reduce sternal Cairo University, Giza, Egypt. Email: [email protected] separation in chronic sternal instability after cardiac surgery: a randomized cross- over trial. Aust J Physiother. 2007;53:255–260. https://doi.org/10.1016/S0004- 9514(07)70006-5 16. Fernandes GA, Lima ACG, Gonzaga ICA, de Barros Araújo R, de Oliveira RA, Nicolau RA. Low-intensity laser (660 nm) on sternotomy healing in patients who underwent coronary artery bypass graft: a randomized, double-blind study. Lasers Med Sci. 2016;31:1907–1913. https://doi.org/10.1007/s10103-016-2069-6 17. Hillis LD, Smith PK, Anderson JL, et al. 2011 ACCF/AHA guideline for coronary artery bypass graft surgery: executive summary: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guide- lines developed in collaboration with the American Association for Thoracic Sur- gery, Society of Cardiovascular Anesthesiologists, and Society of Thoracic Surgeons. J Am Coll Cardiol. 2011;58:2584–2614. https://doi.org/10.1016/j.jacc.2011.08.009 18. El-Ansary D, Waddington G, Adams R. Relationship between pain and upper limb movement in patients with chronic sternal instability following cardiac surgery. Physiother Theor Pract. 2007;23:273–280. https://doi.org/10.1080/09593980 701209402 19. Mampuya WM. Cardiac rehabilitation past, present and future: an overview. Car- diovasc Diagn Ther. 2012;2:38–49. https://doi.org/10.3978/j.issn.2223-3652.2012.01.02 20. Bartels MN. Essential Physical Medicine and Rehabilitation. Springer Science and Business Media; 2006:119–145. 21. El-Ansary D, Adams R, Toms L, Elkins M. Sternal instability following coronary artery bypass grafting. Physiother Theor Pract. 2000;16:27–33. https://doi.org/10. 1080/095939800307584 22. El-Ansary D, Waddington G, Adams R. Measurement of non-physiological move- ment in sternal instability by ultrasound. Ann Thorac Surg. 2007;83:1513–1516. https://doi.org/10.1016/j.athoracsur.2006.10.058

202 Essam El-Sayed Felaya et al: Trunk stabilising exercises after median sternotomy 23. Snijders CJ, Bakker MP, Vleeming A, Stoeckart R, Stam HJ. Oblique abdominal 25. Richardson CA, Snijders CJ, Hides JA, Damen L, Pas MS, Storm J. The relation be- muscle activity in standing and in sitting on hard and soft seats. Clin Biomech. tween the transversus abdominis muscles, sacroiliac joint mechanics, and low back 1995;10:73–78. https://doi.org/10.1016/0268-0033(95)92042-K pain. Spine. 2002;27:399–405. https://doi.org/10.1097/00007632-200202150-00015 24. Snijders CJ, Ribbers MT, de Bakker HV, Stoeckart R, Stam HJ. EMG recordings of 26. Helmy ZM, Mehani SHM, El-Refaey BH, Abd Al-Salam EH, Felaya EEE. Low-level abdominal and back muscles in various standing postures: Validation of a laser therapy versus trunk stabilisation exercises on sternotomy healing after biomechanical model on sacroiliac joint stability. J Electromyogr Kinesiol. coronary artery bypass grafting: a randomized clinical trial. Lasers Med Sci. 1998;8:205–214. https://doi.org/10.1016/S1050-6411(98)00005-4 2019;34:1115–1124. https://doi.org/10.1007/s10103-018-02701-4