1 INTRODUCTION

Obesity is an increasingly common public health issue; the worldwide prevalence of obesity nearly tripled between 1975 and 2016.1 In 2016, more than 1.9 billion adults aged 18 years and older were overweight worldwide. Of these, over 650 million adults were obese. Overall, about 13% adult population (11% of men and 15% of women) were obese in 2016.1 Obesity is a major risk for non-communicable diseases such as cardiovascular diseases; diabetes; musculoskeletal disorders; and some cancers (including endometrial, breast, ovarian, prostate, liver, gallbladder, kidney and colon).2 Obesity is associated with many adverse outcomes, including morbidity, disability, premature death,1 poor mental health,3 stigma and discrimination.4

Self-management (SM) is a dynamic, interactive, and daily activity in which individuals engage to manage their chronic illness.5 For the purpose of this study, we defined SM as ‘actions that individuals, families, and communities engage in to promote, maintain, or restore health and cope with illness and disability, with or without the support of health professionals, and including but not limited to self-prevention, self-diagnosis, self-medication, and coping with illness and disability’.6 Optimal SM can help patients to manage symptoms, treatments, lifestyle changes and address emotional strategies to maintain a satisfactory quality of life.7 Self-management interventions (SMIs) are supportive interventions systematically delivered or led by healthcare staff or other patients with the aim of building patients' confidence and equipping them with the necessary skills. Their purpose is to actively engage patients (and informal caregivers where appropriate) in the management of their disease.8 Research evidence suggests that SMI can improve clinical outcomes in chronic conditions, such as body weight in patients with obesity,9 and patient-reported outcomes such as quality of life, self-efficacy and adherence.11

However, outcomes that measure effectiveness of an intervention often do not reflect what matters most to patients.12 This is probably because most used measurement tools were not co-developed by involving patients.13 Therefore, the appropriate selection of outcomes is an essential stage of research aimed to guide decision-making and inform policy.14 Research results can only be optimally compared if similar outcomes are used across studies. To overcome this, researchers are beginning to develop core outcome sets (COS); defined as ‘agreed, standardised set of the outcomes that should be measured and reported at a minimum in clinical trials and studies of specific conditions and interventions, and are also suitable for research other than randomised trials’.15

Therefore, the aim of this study was to develop the first COS for SMIs for patients living with obesity, from the perspective of patients and other stakeholders dealing with this condition. This work was conducted as part of the COMPAR-EU project; an EU-funded project that aims to bridge the gap between current knowledge and practice on SMIs for patients living with four high-priority chronic conditions: type 2 diabetes, obesity, chronic obstructive pulmonary disease and heart failure.6 Three other COS specific to type 2 diabetes mellitus, chronic obstructive pulmonary disease and heart failure have also been developed in the context of this project, in which manuscripts are under preparation.

2 METHODS

We developed and published our protocol in BMJ Open.6 The COS-STAR guidelines were used to ensure the reporting of this study.16 The development of this COS for SMIs for patients living with obesity included four steps: (1) Development of the initial catalogue of outcomes; (2) Scoping review on preferences and experiences of patients and caregivers; (3) Delphi survey (two rounds) with patients and patients' representatives; and (4) Consensus workshop with patients, patient representatives, healthcare professionals and researchers (Figure 1).

F Core outcomes set development process

2.1 Development of the initial catalogue of outcomes 2.1.1 Data sources and searches

Following methods used in previous COS studies including literature reviews,17, 18 we reviewed selected references of two EU funded projects: Promoting Self-Management for Chronic Diseases in the EU (PRO-STEP) and Empowering patients in the management of chronic diseases (EMPATHiE) that included the development of several overviews of systematic reviews focusing on SMIs for chronic diseases.19, 20 We additionally searched for COS in relevant organization databases such as Core Outcome Measures in Effectiveness Trials (COMET) (http://www.comet-initiative.org/Resources/Database) (Table S1), and also in Medline (via PubMed) by using specific search terms such as ‘core outcome set’; ‘Obesity’ (MeSH terms); ‘patient preference’ (MeSH terms); and ‘Outcome Assessment, Health Care’ (Mesh). We restricted the search to studies published in English, published from 2012 onward.

2.1.2 Study selection

We included systematic reviews and individual studies that reported outcomes on SMIs for patients living with obesity. We excluded studies if the included outcomes were not related to obesity or when the outcomes were not developed considering the perspective of patients. Pairs of authors (RS, CO, MB and EC) conducted independently title and abstract screening and full-text assessment. Disagreements were solved by consensus or with the help of a third author. We also checked references from included studies.

2.1.3 Data extraction

For each included study, pairs of authors (RS, CO, MB and EC) independently extracted the following data: (i) name of the database source, (ii) the type of publication (i.e., published COS, literature reviews or systematic reviews), (iii) age groups, and (iv) list of outcomes and definitions.

2.1.4 Data synthesis

We synthesized and classified outcomes into seven subdomains using a predefined taxonomy developed in the COMPAR-EU project21: (i) basic empowerment components, (ii) adherence to expected SM behaviours, (ii) clinical-related outcomes, (iv) quality of life of patients and caregivers, (v) perceptions and satisfaction with care, (vi) healthcare use, and (vii) costs. Outcomes of the initial list and their definitions were iteratively reviewed and discussed by the research team and when outcomes were similar to others were merged. An external review process included a review by an external clinician and researcher (VP), followed by a discussion with multidisciplinary experts of the COMPAR-EU consortium. The list was edited in plain language by experts in health literacy and patient representatives. This list of outcomes was used for the first round of the Delphi process (Step 3).

2.2 Scoping review on preferences and experiences of patients and caregivers

We conducted a scoping review of reviews using the methodology proposed by Arksey and O'Malley's22 to identify and describe key concepts related to outcomes exploring patients' and caregivers' preferences and experiences with SM for the evaluation of SMIs for people living with obesity.

2.2.1 Data sources and searches

We applied a content search strategy for values and preferences23 in combination with terms specific to obesity and searched in MEDLINE, CINAHL, and PsycINFO (from inception to February 2018).

2.2.2 Study selection

We included quantitative, qualitative and mixed-methods reviews that explored preferences and experiences of patients or caregivers through a systematic search.

2.2.3 Data extraction

Using a previously pilot-tested data extraction form, pair of authors collected the general characteristics and main findings of each review.

2.2.4 Data synthesis

Finally, we conducted a descriptive thematic synthesis including identification of codes, descriptive themes, and main themes relevant to SMI outcomes.24, 25 In addition, we developed an infographic to illustrate the themes by outcome. The infographic included the outcomes of the initial catalogue of outcomes informed by the scoping review, classified according to the preliminary version of the outcome taxonomy. This material was sent to participants 1 week in advance to inform and facilitate the selection of the final COS during the consensus workshop (Appendix S2, Infographic).

2.3 Delphi survey

Following standard methodologies used in previous studies,26, 27 we conducted a two-round modified Delphi online survey with a convenience sample to investigate and obtain consensus on what outcomes were important to participants living with obesity to measure the success of SMIs.

2.3.1 Study population and recruitment strategy

Patients with obesity, caregivers and patients' representatives, who were able to understand and speak English, provided informed consent to participate. Participants were identified by the European Patients' Forum network of more than 70 patient organizations (https://www.eu-patient.eu/about-epf/about-us/) and other patient groups (e.g., those involved in the International Consortium for Health Outcomes Measurement) (https://www.ichom.org/). Recruitment started in February 2018 and ended in May 2018.

2.3.2 Two-round Delphi survey

We administered a Delphi online survey in two separate rounds. In both rounds, participants were asked to rate the importance of outcomes using a 9-points Likert scale (1 being the least and 9 being the most important). In addition, in the second round, participants were able to see the ratings of participants in the first round. All participants received weekly reminders to complete the questionnaire.

All outcomes were then categorized into three groups based on the level of agreement participants reached. If an outcome reached a 70% level of agreement on importance with a Likert-scale mean score larger or equal to 7, meant the outcome was considered important to measure the success of an SMI, and it was categorized in the ‘high consensus and high support outcomes’ group. If less than 70% of participants agreed on the importance of one outcome, it was categorized in the group ‘low consensus and mixed support’; finally, if an outcome reached a 70% level of agreement on non-importance with a Likert-scale mean score smaller or equal to 6, the outcome was categorized in the ‘high consensus of moderate and low support’.

2.4 Consensus workshop and final COS development

The final stage of the obesity COS development process was a 2-day consensus workshop. In the workshop patients and patients' representatives, healthcare professionals, researchers and project members participated. We aimed to invite a group of participants balanced in terms of gender, socioeconomic status, nationality and professional experience. Participants were provided with the results of the two-round Delphi survey (Step 3), and with the synthesis of results from the scoping review together with the infographic (Step 2).

The list of the outcomes was presented into the same seven subdomains as for the Delphi survey, and sorted by level of agreement as described above. Participants were first presented with the outcomes of the ‘low consensus mixed support’ category, which were individually discussed and voted for inclusion (yes or no answers by raising hands) using the same criteria of importance and non-importance described above (i.e., at least of 70% level of agreement on importance). In the same way, participants reviewed the outcomes from the ‘high consensus of moderate and low support’, and from the ‘high consensus and high support’ groups to determine which outcomes to include in the final COS. During this process, participants had the opportunity to discuss results, propose new outcomes and when outcomes were similar to others were merged. These data were entered by one author into a Microsoft Excel computer file and checked by a second author at the time of the voting. Those outcomes with at least 70% of agreement on high importance were included as part of the draft COS. After all outcomes had been voted on, and the draft COS was reviewed, participants were asked to select a maximum of 15 outcomes for the final COS following the same inclusion criteria previously described. Outcomes that reached a vote of ‘high consensus and high support’ but were not included in the final COS, became ‘supplementary outcomes’ (maximum 10 outcomes). The COMPAR-EU research team led step-by-step the discussion and addressed discrepancies across stakeholders (Figure 2).

Consensus workshop process

3 RESULTS 3.1 Development of the Initial catalogue of outcomes 3.1.1 Study selection

The review of previous EU funded projects identified 257 records focusing on SMIs in chronic diseases in general. We included 44 systematic reviews specific to obesity (see references in Appendix S1, Systematic Reviews References). The search in MEDLINE (via PubMed) and COMET yielded 21 articles of which six were considered eligible. After full-text screening, we included five studies.28-32 Figure S1 presents the study flowchart with the search results and study selection process.

3.1.2 Study characteristics

Five studies included a COS,28 a comprehensive International Classification of Functioning, Disability and Health (ICF) Core Set,29 two systematic reviews30, 31 and a protocol of a clinical trial design.32 Three of the studies were on outcomes related to surgery, and one was about a specific drug. The 44 SR included 948 individual studies with a total of 323 303 patients. The individual studies were conducted between the years 1950 and 2006, and were published between 2010 and 2016. Thirty-two of 44 SR (72.73%) included at least one European study.

3.1.3 Catalogue of outcomes and outcomes classification

We identified 82 outcomes, which were categorized into the seven predefined subdomains and 24 elements (Table 1). The majority of outcomes were either ‘Clinical related outcomes’ (n = 24, 29.3%), or were included in the ‘Quality of life of patients and caregivers’ subdomain (n = 22, 26.8%). The remaining were distributed in: ‘Adherence to expected self-management behaviours’ (n = 13, 15.9%), ‘Health care use’ (n = 9, 11.0%), ‘Perceptions and satisfactions with quality of care’ (n = 6, 7.3%); ‘Basic empowerment components’ (n = 4, 4.9%), and ‘Costs’ (n = 4, 4.9%).

TABLE 1. Initial catalogue of outcomes Subdomains Elements Outcome Basic empowerment components Patient activation level 1. Patient activation Level of self-efficacy 2. Self-efficacy Level of knowledge 3. Knowledge Level of health literacy 4. Health literacy Level of adherence to expected SM behaviours Clinical self-management behaviours 5. Adherence to programme 6. Taking medication or other treatment as advised (adherence) 7. Self-monitoring (Including self-recording) Lifestyle-related measures 8. Dietary planning 9. Eating management 10. Energy intake 11. Dietary journal 12. Sugar-sweetened beverages consumption 13. Calorie burn 14. Physical Activity Psychological self-management 15. Addictive behaviours 16. Bingeing 17. Purging Clinical-related outcomes Complications 18. Adverse events 19. Complications Disease progression 20. Blood pressure control 21. BMI 22. Bowel motion issues 23. Breathing problems during sleep 24. Breathlessness (dyspnoea) 25. Cardiac rhythm 26. Heat waves 27. Lipid control 28. Nausea 29. Numbness 30. Physical fatigue 31. Sleep quality 32. Snoring 33. Stable weight 34. Swallowing disorders 35. Symptoms (general) 36. Waist size 37. Weight loss 38. Cardiovascular risk 39. Chance of developing other diseases 40. General metabolic functions Mortality 41. Mortality Quality of life of patients and caregivers Caregivers' quality of life and competences 42. Caregiver quality of life 43. Caregiver burden 44. Caregiver knowledge 45. Caregiver self-efficacy 46. Caregiver anxiety and/or depression Physical functioning 47. Usual activities 48. Mobility 49. Work 50. Physical activities 51. Sex life 52. Normality 53. Pain Psychological functioning 54. Depression 55. Anxiety 56. Stress 57. Coping 58. Hostility 59. Happiness 60. Self-esteem Social relations and activities 61. Family relationships 62. Friends 63. Social activities Perceptions and satisfaction with care Overall satisfaction with self-management interventions 64. Care satisfaction 65. Consultation time Perception of being well and sufficiently informed (quality of information provision) 66. The patient feels s/he has enough information Personalized care 67. Participation and decisions making Perception of patient–provider relationship 68. Patient healthcare provider relation 69. Communication with healthcare professionals Healthcare use Number of visits or contacts with healthcare providers 70. (Number of) primary care or outpatient (ambulatory) visits 71. (Number of) nurse visits 72. (Number of) visits to specialist doctors 73. Number of visits with other healthcare professionals 74. (Number of) virtual visits or contacts with healthcare providers Emergency departments visits 75. (Number of) emergency department visits (hospital) Hospital admissions and length of stay 76. (Number of) hospital admissions 77. The length of time spent in hospital 78. (Number of) re-hospitalizations, unexpected return to hospital Costs Cost for the healthcare system 79. Impact on healthcare costs for the healthcare system 80. Cost savings for the healthcare system as a result of the self-management intervention Cost for patient (out of pocket payments) 81. Direct medical costs for patient (out of pocket expenses) Cost-effectiveness 82. Value for money of the self-management intervention 3.2 Scoping review on preferences and experiences of patients and caregivers 3.2.1 Study selection

After removing duplicates, searches yielded 2154 unique references resulting in 67 records selected for full-text screening, and 15 reviews finally included.33-47 Figure S2 presents the study selection flow diagram.

3.2.2 Study characteristics

Of the 15 reviews included, seven were published between 2015 and 2017 (46.6%). The majority were qualitative evidence synthesis (n = 8, 53.3%) and the number of included studies ranged from 642 to 29 primary studies.44 Most of the reviews (n = 14, 93%) addressed patients' perspectives, except for one review45 (7%) that addressed only the perspective of healthcare providers. Most reviews (n = 10, 66.6%) explored the experiences of patients in the process of SM, four explored the experiences with SMI,37, 38, 40, 47 and one review explored the patients' preferences regarding weight change.39

3.2.3 Identified themes

We identified 16 themes according to the COMPAR EU taxonomy subdomains (Table 2).

TABLE 2. Identified themes according to the COMPAR-EU taxonomy subdomains Taxonomy subdomains Themes Basic empowerment components Health knowledge40, 42 Help-seeking behaviour35, 36 Technological literacy38 Level of adherence to expected SM behaviours Adherence to treatment33, 34, 37 The perceived benefit of the intervention37, 38 Weight control34, 42 Clinical-related outcomes Progression of disease34, 39 Quality of life of patients and caregivers Physical functioning40 Psychological distress34-36, 38, 40-42 Social support37, 42 Perceptions and satisfaction with care Patient-provider interaction35, 36, 43 Perceived quality of care36 Healthcare use Access to healthcare36, 45 Visits or contacts with healthcare professionals36 Costs Cost for patients (out of pocket)33 Cost-effectiveness34 3.2.4 Mapping of themes

Of the 82 outcomes of the initial catalogue of outcomes, 31 were covered in the thematic synthesis, meaning nearly one out of four were informed by the scoping review (n = 31, 37.80%). All outcomes of the subdomain ‘Perception and satisfaction with care’ were informed by the scoping review findings (n = 6, 100%), while the ‘Clinical outcomes’ subdomain had the least number of outcomes informed (n = 3, 12.50%). Figure 3 reports the number of outcomes informed by the thematic synthesis of the scoping review.

Mapping of themes per COMPAR-EU taxonomy subdomains

3.2.5 Infographics

One infographic on obesity was developed for the final consensus workshop. The infographic included a summary of the main findings and topic-related images (Appendix S2, Infographic).

3.3 Delphi survey

Ten participants participated in the two-round Delphi online survey and only one participant did not complete the second round of the survey. The majority of the participants were women (70%) and were patient advocates or patient representatives. Five of the included participants (50%) were between 45 and 54 years old and the majority had a bachelor's degree or a higher education (masters' or equivalent; doctoral or equivalent) (Table S2).

After the Delphi survey, of the 82 included outcomes, 27 were categorized in the ‘high consensus and high support outcomes’ group, 49 in the ‘low consensus mixed support outcomes’ group and five in the ‘high consensus of moderate and low support’.

3.4 Consensus workshop and final COS development

Ten participants attended the consensus workshop, five were patients living with obesity who participated in the Delphi survey and five were healthcare professionals who were informed on the Delphi results before the workshop took place. Of the 10 participants, there was only one male and the age of all participants ranged from 25 to 65. Four members of the COMPAR-EU team (CV, RS, VS, NA) also participated.

Participants selected and prioritized 15 outcomes for the final COS, five supplementary outcomes were also provided. The COS included five individual outcomes originally discussed and voted during the Delphi, nine composed outcomes resulting from the merging of 22 ‘high consensus and high support outcomes’ with 14 ‘low consensus mixed support outcomes’ and finally, one new outcome ‘Integration at work’ proposed by the participants. None of the outcomes included in the final COS were from the ‘high consensus of moderate and low support’ group. The five supplementary outcomes included: (1) Addictive behaviours; (2) Sleep quality; (3) Medication adverse events; (4) Pain; and (5) Sex life. Table 3 presents the final list of outcomes and Table 4 presents the five supplementary outcomes.

TABLE 3. Outcomes included in the core outcome set Outcome Definition Self-efficacy A person's belief that s/he is capable of doing something, often related to a specific goal s/he wants to achieve; feeling of confidence and of being in control. According to the group health, literacy is a prerequisite and can be included here Participation and decisions making Feeling able to participate actively in her/his own care and treatment decisions (as much as s/he wishes) Patient activation The knowledge, skills and confidence a person has in managing their own health and healthcare, including a feeling of being responsible for taking care of their own health Adherence to programmea (1) Adherence to programme: Adherence to diet, exercise plan, etc. (as a whole) The extent to which a patient follows the prescribed treatment, such as taking medication as advised and following lifestyle advice. (2) Taking medication or other treatment as advised: The extent to which a patient follows the prescribed treatment, such as taking medication as advised and following lifestyle advice. (3) Direct medical costs for patient: Out of pocket expenses Self-monitoring (including self-recording) The extent to which a patient (regularly) monitors herself/himself as agreed with her/his healthcare professionals, for example her/his symptoms or her/his weight Healthy nutrition habits/personalized nutritiona (1) Dietary planning: Having a healthy/balanced eating pattern. (2) Eating management: Healthy eating (personalized nutrition). (3) Sweet tasting beverages consumption: Sugary drinks with sugar or artificial sweetened. (4) Mindful eating: Being able to stop eating when feeling full Quality of life—Physical and psychological functioninga (1) Usual activities: Being able to do usual activities, such as personal hygiene, housework, managing finances. (2) Normality: Feeling able to live a ‘normal’ life. (3) Mobility: Being able to walk, climb stairs, bend, cross legs, get up from chairs). (4) Work: Being able to do work tasks, or to take up work/paid employment. (5) Physical Activities: Being able to participate in and enjoy physical activities. (6) Depression: Feeling depressed. (7) Anxiety: Feeling anxious. (8) Stress: Feeling stressed Coping with the diseasea