The influence of therapeutic alliance on adult obesity interventions in primary care: A systematic review protocol

Trial registration: CRD42018091338 in PROSPERO (International prospective register of systematic reviews).

Obesity is one of the world’s most burdensome and escalating population health challenges.¹ More effective strategies for managing patients with obesity are needed within primary care  to curb the health burden for communities,  the cost implications for health systems and the reduced quality of life for individuals.^2–5 Obesity is defined by a body mass index (BMI) ≥30 kg m², with associated health impairments such as impaired glucose metabolism, hypertension, musculoskeletal pain or mental health distress.⁶ Current intensive strategies for managing obesity include bariatric surgery, medications and very low energy diets.⁷ There is a strong international consensus from the World Health Organization (WHO)⁸ and multiple national government and obesity coalitions^7,9–12 that all approaches to obesity management should be underpinned by support for behavioural and lifestyle change, as this increases the effectiveness of all interventions.^13,14

This review will focus on the behavioural and lifestyle supports that are essential for all approaches to obesity management.

Obesity is a chronic, relapsing, progressive disease that requires costly healthcare over the long term. Primary care is well placed to assist patients with obesity.^{2,3,5,12,15,16} Primary care is the first point of contact in a healthcare system. It offers whole-person care (ie care involving any body system); assists patients in a person-centred manner (ie care individualised based on a person’s value and beliefs and within the context of their family and community life); and provides continuing care to people over the long term.¹⁷

As obesity affects many body systems and is a chronic disease, it is essential that primary care services and professionals are supported to manage the condition effectively.³ Primary care provides the most efficient and cost-effective avenue for chronic disease care within health systems.^17,18

The relationship between healthcare providers and patients is increasingly being recognised as a key contributor to effectiveness.^19,20 The practitioner–client relationship has been most fully investigated in psychology, with a meta-analysis confirming the strength of the relationship to be moderately associated with outcomes (Spearman’s rho = 0.22), which is more than is attributable to any particular form of psychotherapy.^21,22

The concept of alliance is now well established in the theoretical framework of Bordin²³ from the late 1970s, which extended the concept of the therapeutic relationship from being ‘just’ about warmth and empathy, to an understanding of the practical and collaborative nature of a helping relationship that he termed the ‘Working Alliance’. This three-part model frames the relationship as:

1. ‘Bond’ – respect, empathy, trust and warmth

2. ‘Goals’ – collaborative goal setting

3. ‘Tasks’ – agreed steps to be undertaken to reach the goals.

This model has since been used to develop the Working Alliance Inventory for measuring the practitioner–client alliance in psychology; a high score is associated with better client outcomes.^22,24 Further applications of the tool to other areas of medicine and primary care have also shown the strength of the alliance to be associated with better patient outcomes.^25–27

With an understanding of the chronic nature of obesity, the importance of supporting primary care and the strength of therapeutic alliances, this systematic review aims to appraise trials of interventions in primary care for obesity management in adults with a focus on the influence of the therapeutic alliance.

The review will answer the following questions:

1. What effect does therapeutic alliance have on the effectiveness of interventions for obesity in adults in primary care?

2. What effect does therapeutic alliance have on the withdrawal rates in obesity management trials in primary care?

3. What effect does therapeutic alliance have on the rates of loss to follow-up in obesity management trials in primary care?

Methods and analysis

A systematic review incorporating a meta‑analysis will be used. The review will be prospectively registered in PROSPERO using the PRISMA-P format. We will search MEDLINE, EMBASE, PsycINFO, CINAHL, Cochrane Central Register of Controlled Trials (CENTRAL) and Scopus for original literature and citations. Authors of unpublished trials that are registered with the WHO International Clinical Trials Registry Platform will be contacted up to three times by email for information on their trials. Theses published on Open Access Theses and Dissertations and Networked Digital Library of Theses and Dissertations will be included.

Search strategy

The search strategy is based around the terms ‘obesity’, ‘primary care’ and ‘intervention’. The Ovid string is detailed in Box 1. This search strategy will be modified where necessary for the other databases. The search strategy has been assessed by a specialist primary care librarian. 
 

Types of study to be included

Controlled trials of interventions (randomised controlled trials [RCTs], cluster RCTs).
Participants and population
To be included in the review, studies must:

• include an adult sample (aged ≥18 years) with confirmed obesity, where obesity is defined as either a BMI of >30 kg/m², or by the diagnosis of obesity as defined by the intervention if different. Studies that focus primarily on other chronic conditions or with a purpose of general improvement in nutrition or physical activity will be excluded.

• be published as full papers or theses in English between 1 January 1998 and March 2018.

Intervention

Interventions will include those delivered in primary care and aimed at optimising the health, and/or reducing the body weight, of adults living with obesity. ‘Primary care’ is defined as the first point of contact in a community-based system. Interventions may be delivered by a single health professional or by a multidisciplinary team of health professionals. Any intervention that occurs for screening or diagnostic purposes will be excluded.

Comparator

The comparator will be guided by the primary studies and classified as either a control (conventional health care) or alternative intervention (variation on the intervention). Pooling will be done according to homogeneity of the interventions.

Primary outcome(s)

The primary outcome wil be a reduction in body weight or BMI.

Timing and effect measures

We will synthesise outcomes of studies using a random-effects model at similar time points for the data that are available (eg at six months and/or two years). Length of follow-up will not be used as an exclusion criterion.

While we acknowledge that long-term weight reduction is much more problematic in obesity than short-term weight loss, this systematic review aims to investigate the effectiveness of the therapeutic alliance in existing obesity studies.

Secondary outcome(s)

Secondary outcomes include:

• loss to follow-up

• withdrawal of participants

• quality-of-life measures

• metabolic risk outcomes (hypertension, blood glucose levels, lipids).

All secondary outcome measures included in trials will be collected.

Timing and effect measures

Outcomes will be measured and/or assessed over a specified period, depending on the data presented in the studies. Length of follow-up will not be used as an exclusion criterion; however, proportion of participants lost to follow-up will be examined.

Data extraction (selection and coding)

Two reviewers will screen all titles and abstracts independently using Covidence software on the basis of the headings detailed above.

Conflicts will be resolved by discussion, with studies retained for full text screening if agreement is not reached. Where there is ambiguity, abstracts will be retained for full text assessment. Where no abstract is available, records will be retained for full text assessment unless they can be judged from the title not to relate to the review topic.

Full text screening

Two reviewers will assess full texts of all potentially eligible studies using a piloted checklist. Disagreement about eligibility will be resolved by discussion, involving a third reviewer where necessary.

Data extraction

Two reviewers will independently extract data using piloted data extraction forms (Box 2). Data extracted will be categorical or continuous. Qualitative extraction will be avoided, if possible, to aid with pooling of the data.

 
The presence or absence of therapeutic alliance within the intervention will be defined on the basis of the Bordin framework for working alliance:

1. the bond between the two parties:

   1. is there a pre-existing relationship with the provider of the intervention? (Yes/No)
      OR
   2. will there be an ongoing relationship with the provider of the intervention after the trial? (Yes/No)

2. collaborative goal setting: did the intervention involve collaborative goal setting? (Yes/No)

3. agreement on the required tasks to reach the goals: did the intervention involve agreement on the tasks to be undertaken (eg not protocol driven, room for individualisation for the patient)? (Yes/No)

Three subgroups for the therapeutic alliance will be defined as:

• Group 1 – none of the above components and constitutes ‘general education’

• Group 2 – one or two components (ie bond plus either tasks or goals) 

• Group 3 – all three components, and/or the intervention specifically measured therapeutic alliance (although must not have ‘bond’ as the only component described).

Where studies of relevant interventions are identified but data relating to therapeutic alliance is not reported separately, authors of included papers will be contacted by email to attempt to obtain this data.

Risk of bias (quality) assessment

The quality of included studies will be assessed independently by two reviewers using the risk of bias criteria detailed in Cochrane Collaboration Risk of Bias Tool.²⁸ Risk of bias relating to the primary outcome will be assessed and presented for each study in a risk of bias table.

Strategy for data synthesis

A meta-analysis will be performed if studies of sufficiently homogenous primary outcome data are found. Homogeneity will be determined statistically via I², as well as descriptively focusing on methodological and clinical variation. Change from baseline data or endpoint date will be used. For variables with the same scale, weighted mean difference with 95% confidence intervals (CI) will be used; for those with a different scale, standardised mean difference with 95% CI will be calculated. All pooling of data will use the DerSimonian and Laird random effects model. This reflects the diverse and complex nature of the interventions included. Statistical heterogeneity will be measured using I² statistics and classified as negligible (I² = 0%), minimal (I² <20%), moderate (20%< I² <50%) or substantial (I² <50%). In instances in which studies are too heterogeneous to combine quantitatively, a narrative synthesis will be done.

Analysis of subgroups

1. Therapeutic alliance presence

   • Group 1 – will have none of the above components and will represent general education
   • Group 2 – will have one to two components (ie bond plus either tasks or goals) 
   • Group 3 – will have all three, and/or the intervention specifically measured therapeutic alliance (although must not have ‘bond’ as the only part they describe)

2. Time-intensiveness – both as per protocol and what actually eventuated in the trial

   • Group 1 – one consultation
   • Group 2 – two to five consultations
   • Groups 3 – more than five consultations

3. Was the intervention delivered in regular practice?

4. Did usual provider refer to the intervention?

Discussion

This systematic review and meta-analysis will provide new and important knowledge of how and why obesity interventions in primary care might work. To improve the effectiveness of interventions in primary care, unpacking the how and why interventions work is important. The information from this systematic review will be used to inform future intervention development, translation of intervention trials into clinical practice and assessment of interventions for effectiveness.

This systematic review may have wider applications outside the field of obesity. The findings may be important for primary care interventions in general.