This protocol has been designed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols (PRISMA-P) Statement (15, 16). The PRISMA-P checklist is reported (see Additional File 1). The study protocol has been registered within the PROSPERO database (CRD42020191467).
Eligibility criteria
The population, intervention, comparators, outcomes, timing, and setting (PICOTS) framework will be used to define the scope of this review and assist in study screening.
Population: We will include all studies including adults (≥ 18 years of age) with chronic pain conditions of any etiology or experiencing chronic cancer or noncancer pain in both treatment and control settings. We will exclude studies that focus on pain in the perioperative period as well as studies of acute pain. Pain can be in any physical location (e.g., migraine or low-back pain) if it is both chronic and documented via a patient-reported, numerical pain score.
Intervention: We will include studies that implement an intervention incorporating teamwork or teaming. For the purposes of this study, evidence of teamwork or teaming is defined as at least two clinicians interacting with a patient in an ongoing coordination process, evidenced by clinicians interacting among themselves and/or with the patient over at least two distinct timepoints. A single referral will not provide sufficient evidence of teamwork or teaming, nor will layered care modalities with no communication between interdisciplinary clinicians. If a study has multiple intervention arms that incorporate teamwork or teaming, each arm will be included as a distinct intervention in the analysis of our review.
Comparator: The main comparator is treatment as usual; however, studies that compare an intervention incorporating teaming to another care process may also be included (i.e., usual care plus educational pamphlet).
Outcomes: The primary outcome of interest is overall pain, inclusive of psychical, mental, emotional, or spiritual suffering. However, the pain outcome must be presented discretely and overtly as a patient-reported numerical value. Studies that only include a measurement of pain as a component of a composite metric of a greater construct will not be included. Pain outcomes will include measures of bodily pain, pain intensity, severity of pain, pain-related disability, self-reported pain, etc., on validated pain scales (e.g., the Brief Pain Inventory (BPI) (17), Graded Chronic Pain Scale (GCPS) (18), or other validated scales).
Timing: Interventions with any follow-up period will be included.
Setting: Only studies that are primarily set in the context of primary care will be included.
Information sources
The review will include published, peer-reviewed, randomized studies and systematic reviews that meet the inclusion criteria. The following databases will be searched: EMBASE, PubMed, CINAHL, and Cochrane Library. For feasibility, we will search for articles published in English and articles published after 2009. Searching for articles from the last ten years will capture many chronic pain management interventions incorporating teamwork or teaming in a timeframe where pain management guidelines have emphasized the importance of interdisciplinary teams (9). Limiting our review to only systematic reviews and randomized controlled trials will allow us to focus on the highest level of evidence of studies with demonstrated improved pain outcomes.
Review Team
Our core review team is composed of a PhD expert in organizational theory (KG), two physicians (PP, KL) with expertise in pain management, and two research assistants (NC, SZ). This team will meet weekly via video conference or telephone throughout the review.
Search Strategy
Electronic search strategies were developed and tested in an iterative process by a library scientist (HW) in consultation with the review team (KG, KL, SZ, NC, and PP). Our pilot search strategy is included in Appendix I. We will conduct our search twice: once at the commencement of this systematic review, and once to update the search in advance of journal submission to ensure any newly published studies that may meet our inclusion criteria are identified.
Data Collection and Analysis
Data Management
All stages of our review (title and abstract screen, full text screen, data extraction, quality assessment) will be conducted using the Covidence online systematic review tool to streamline screening and extracting processes (19).
Selection Process
The studies will be assessed according to the eligibility criteria and the selection will be divided into two phases. The screening of titles, abstracts, and full texts will be dually screened by two teams of paired reviewers (NC or SZ and KG or PP) to ensure each study is reviewed by a team member with doctoral-level expertise. Any disagreement will be adjudicated by a third “gold standard” reviewer (KL). If uncertainties persist as to the eligibility of an article, the study authors may be contacted for clarification (20). A PRISMA flow diagram showing details of studies included and excluded at each stage will be provided.
Data Collection
Data will be extracted independently by two reviewers (NC, SZ, KG, or PP) using a customized data extraction form, piloted before use. The extraction form will be informed by organizational theories and will be used to extract relevant team structures and processes, as well as pain outcomes. In the design of our abstraction guide, we relied upon Donabedian’s structure-process-outcome model (21) to inform the extraction of relevant information from all included studies on the care setting, team structures (who is involved), teamwork and team processes (how they communicate, collaborate, coordinate, cooperate), follow-up duration, patient demographics, and pain outcomes. The questions on our abstraction guide about teamwork and teaming are based upon Hackman’s Five Factor Model of Teaming (14) and Edmondson’s dynamic teaming (1); Table 1 highlights how these models will be used to inform our data extraction and analysis. Primary data will be collected and stored in Covidence (19). Discrepancies between the two evaluators in extracted data will be resolved by team consensus. When data are not available in the manuscripts or in case of uncertainty, the authors may be contacted for clarification or study protocols referenced (20).
Table 1
Theoretical models used to inform data extraction
Donabedian Model (21)
|
Organizational Theory
|
Abstraction Guide Questions
|
Structure
|
Hackman’s Five Factor Model (14)
|
Real team
|
Is the team bounded?
Who is on the team?
|
Compelling direction
|
Is the team working towards a common goal?
|
Enabling structure
|
How many clinicians are on the team?
|
Process
|
Hackman’s Five Factor Model (14)
|
Supportive context
|
Were the clinicians responsible for attending any training or educational sessions?
Important facilitators like incentives, additional resources, dedicated provider time?
|
Expert coaching
|
Important facilitators like mentorship?
|
Edmondson’s Dynamic Teaming (1)
|
Communication, coordination, collaboration processes
|
Does the team of providers have meetings? Phone contacts/other forms of communication?
|
Does more than one team member work together on any particular task/role?
|
Did the team members contact the patient throughout the intervention?
Were patient workshops/sessions part of the intervention?
|
Outcome
|
Patient reported outcome
|
Pain
|
Did patient-reported pain outcomes improve (e.g., improved BPI, GCPS, etc.)?
|
Data items
Information extracted during the data abstraction will include author information, date of publication, journal of publication, number of study participants, type of population (chronic cancer or non-cancer pain), chronic pain outcomes reported as numeric, patient-reported pain outcomes (e.g., BPI, GCPS, etc.), study findings, overall statistical findings, and author’s conclusions. Additionally, we will capture and describe the evidence of teams and teaming, including information about team structure (e.g., team member roles, number of team members, role of clinicians involved in delivering the intervention), team training (e.g., educational session for either clinicians or patients), and teaming processes (e.g., huddles, checklists, referral templates, coordination between multiple providers, the use of technology like telehealth, teleconferences, or notes in the electronic medical record).
Risk of bias/quality assessment
Risk of bias in randomized control studies will be assessed using the Cochrane Risk of Bias tool (22). This tool considers several domains of bias: randomization, allocation concealment, blinding, accounting of patients and outcome events, and selective outcome reporting bias. Risk of bias in systematic reviews will be assessed using the AMSTAR tool (23). The risk of bias assessments will inform our assessment of study limitations across the body of evidence.
Data synthesis
We anticipate a high degree of heterogeneity within the included studies, and thus we will perform a narrative synthesis of the included studies. We will follow Cochrane methods for conducting a narrative synthesis (22). We will identify the key team structures and teaming processes of the included interventions.
All studies will undergo a dual review with gold standard adjudication. We will evaluate included studies by focusing on differences in team structures and processes between effective interventions (those with at least one improved patient-reported pain outcome in the intervention group compared to control) and ineffective interventions (lack of a significant improvement in pain outcomes in intervention groups over control). We will seek feedback from study advisors with expertise in pain management and organizational theory on the team structures and processes identified through our analysis.