This protocol is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analysis Protocols (PRISMA-P) 2015 checklist (see Additional file 1) [26].
Approach
To answer our research questions, we will perform a realist synthesis with coincidence analysis (CNA). Realist synthesis is a systematic, iterative, theory-driven approach that draws on a heterogeneous evidence base to establish what works, how, in what context and for whom [13]. We will apply CIMO logic, which aims to determine the combinations showing that in context C, an intervention I invokes generative mechanisms M that produces outcome O.
Given this complexity, to arrive at the CIMO combinations we will apply a new mathematical, cross-case method called Coincidence Analysis (CNA), which belongs to a broader class of Configurational Comparative Methods (CCMs). These methods have been designed explicitly to support causal inferences, answer research questions about combinations of conditions that are minimally necessary or sufficient for an outcome, and to identify the possible presence of multiple causal paths to an outcome [27]. CNA can be applied to large-n as well as small-n sets. It has recently been used in multiple projects which aimed at discovering minimally sufficient and necessary factors affecting successful implementation, as well as their combinations and multiple paths to outcomes [28–30]. Moreover, CCMs were applied in a recent Cochrane Review which aimed at identifying conditions associated with successful implementation of school-based interventions for asthma self-management [31]. Finally, configurational methods have been included as a relevant method for implementation research in the Handbook on Implementation Science [32,33].
CNA operates based on the Boolean properties of causation, which encompass three dimensions of complexity. The first is conjunctivity: to bring about an outcome, several conditions must be jointly present. For example, the analysis may demonstrate that a job crafting intervention is successful when individuals are white-collar employees AND are provided with tasks to perform (homework) between workshop sessions AND when they are reminded to engage in these tasks. The second dimension of complexity is disjunctivity (equifinality), which means that different paths can lead to the same outcome. For instance, a success in job crafting interventions can be achieved when there is alignment of the intervention with organizational aims and when reminders of JC activities are sent OR when there is no alignment present, but managers have been trained alongside employees. The third dimension of complexity is sequentiality, which points to a possibility that outcomes tend to produce further outcomes, propagating causal influence along causal chains. For instance, an increase in job crafting as a result of the intervention can, in turn, lead to an increase in job performance. Thus, CNA analysis will be instrumental in answering our questions about necessary and sufficient factors, as well as multiple combinations that lead to the intervention success.
Eligibility criteria
To systematize study selection, we will use the PICO (Population; Intervention; Comparison; Outcome) approach. Specifically, to be included in our review, the study:
-
has to involve active employees (i.e., not a student sample, not persons on leave) as participants (P),
-
has to be a job crafting intervention (I),
-
contains a comparator (intragroup, between group, or control group) (C);
-
contains measures of job crafting (general or specific types) as a proximal intervention outcome and/or contains health and wellbeing or performance as distal outcomes (O);
Search terms and strategy
The search will be executed by a professional team from the Karolinska Institutet (KI) library that specializes in literature reviews. It will be conducted in the following databases: PsycINFO, Web of Science, Academic Search Complete, MEDLINE, and CINAHL. The search will be conducted for research published until May 2021 (without a lower time limit). The search terms will be based on the topic of this review, i.e., job crafting (e.g., “job crafting” as well as its specific types, such as “seeking challenges” or “increasing structural job resources”) and the intervention (e.g., “intervention”, “experiment”, or “workshop”)1. Additionally, we will inspect the references of two published literature reviews on job crafting interventions: a systematic review [20] and a meta-analysis with utility analysis [21].
Review and extraction
All retrieved references will be uploaded to open source Rayyan software [34]. Abstracts of the retrieved studies will be screened independently by 2 team members to identify studies that meet the inclusion criteria. Then, full texts of these studies will be independently assessed by the two authors for eligibility. Discrepancies in the screening at these two stages will be resolved by discussion, and when needed, a third person will be consulted.
A standardized, pre-piloted form will be used to extract data from the included studies. We conducted a feasibility test for 5 published articles that contained a job crafting intervention to derive factors of relevance within the CIMO logic. Four groups of data will be extracted: Context (e.g., alignment, participants, co-occurring changes), Intervention (e.g., workshop duration, action plans, feedback), Mechanism (e.g., theory of planned behavior, job demands-resources model, experiential learning), and Outcomes (e.g., effects for job crafting, health and well-being, performance). Each record will be extracted by two independent extractors in duplicate to ensure a high quality. Additional factors will be inductively added during data retrieval if they appear relevant for the CIMO logic. Authors will be contacted via e-mail to clarify or provide more information on the conducted intervention.
Risk of bias
Following the meta-analysis of JC interventions [21], we will use the Cochrane Collaboration tool [35] to evaluate the risk of bias. We will consider selection bias (i.e., sequence generation, allocation concealment), performance bias (blinding of participants/personnel), detection bias (blinding of outcome assessment), attrition bias (i.e., incomplete outcome data), reporting bias (selective outcome reporting), and other potential sources of bias. Each domain will be evaluated for each intervention as either a low, high, or unclear risk of bias. Consequently, the more domains are assigned low risk in a particular study, the higher its quality.
Synthesis/Data analysis
As reflected in the types of hypotheses scrutinized by CNA, regularity theoretic causation is a relation that holds between variables/factors taking on specific values [27]. In our synthesis, factors will represent categorical properties that will partition sets of units of observation (cases) into two sets (binary properties). These binary properties will take the form of a crisp set that can take on 0 and 1 as possible values. Thus, for the purpose of CNA, the extracted information for each factor will be coded as 1 (reflecting a presence of a factor) and 0 (reflecting a factor absence) based on clearly defined criteria. The definitions and criteria will be developed by the research team, pilot-tested based on a sample of relevant articles and refined if needed. Again, coding will be done by two separate coders in duplicate and codes will be compared. Table 1 presents examples of factors for each CIM category as well as coding criteria.
Table 1 Examples of factors in the categories of Context, Intervention, and Mechanism with coding criteria
Category and factor
|
Coding criteria: Assign 1 when
|
Coding criteria: Assign 0 when
|
Context (C)
|
Fit
|
The intervention is described as implemented in response to a specific organizational need.
|
The intervention is implemented without consideration of a specific need.
|
Presence of co-occurring changes
|
Changes co-occurring with the intervention are mentioned
|
No co-occurring changes are described.
|
Homogeneity of the profession
|
Intervention participants all have the same/similar job (e.g., all are nurses, all are teachers, etc.).
|
Participants in the interventions have mixed jobs/professions.
|
Intervention (I)
|
Job analysis
|
The intervention included an analysis of resources (mental, physical, social and organizational factors, enabling professional goals to be achieved and reduction of costs), demands (mental, physical, social and organizational, requiring effort or skills from the employee), organizational barriers and/or constraints.
|
The intervention activities did not include any job analysis pertaining to job demands, resources, organizational barriers or constraints.
|
Action plans
|
The intervention involved planning future JC activities by participants.
|
No plans for JC activities were created by participants as a result of an intervention.
|
Reminders
|
The participants received reminders about fulfilling actions plans or post-workshop homework.
|
No reminders were provided for participants about fulfilling actions plans or post-workshop homework.
|
Mechanism (M)
|
Job demands-resources model
|
The authors indicate that their intervention study and hypotheses are based on the JD-R model.
|
No or theories other than the JD-R model are mentioned.
|
Self-determination theory
|
The authors clearly indicate that their intervention study and hypotheses are based on self-determination theory.
|
No or theories other than self-determination theory are mentioned.
|
Theory of planned behavior
|
The authors clearly indicate that their intervention study and hypotheses are based on the theory of planned behavior.
|
No, or other, theories than the theory of planned behavior are mentioned.
|
Data analysis will be performed in a devoted CNA package in R. The output will be interpreted in terms of conditions sufficient (RQ4), conditions necessary (RQ5), and possible alternative paths to the outcome (RQ6). We will also investigate whether a success in a JC intervention is sufficient and/or necessary with an increase in well-being and performance after the intervention (RQ7). The interpretation of the output from configurational methods will be done with specific attention to consistency (degree to which the solution always yields the expected outcome; [27]) and coverage (degree to which the solution covers all cases; [27]) of the results, non-redundancy of the factors, and consistency with logic, theory, and prior knowledge.
The strength of the body of evidence will be assessed using relevant domains from the Guide to Community Preventive Services [36], which serves as a framework to evaluate “confidence that changes in outcomes are attributable to the interventions” (p. 38). This framework is suitable for narrative synthesis. Based on this framework, evidence will be rated as strong, sufficient, or insufficient.
1The final refined list of search terms will be included in the systematic review upon publishing the results.