The Role of Decision Impact Studies in Genomic Medicine in Cancer Care: A Scoping Review and Bibliometric Analysis Protocol

Decision impact studies have become increasingly prevalent in oncology in recent years, particularly in breast cancer prognostic research. Such studies, which aim to evaluate the impact of a test on clinical decision-making, appear to be a new form of knowledge with the potential to impact clinical practice and regulatory decision-making in genomic medicine. Yet their origins, intended purpose and usage have not yet been explored. The objectives of this review are to identify and characterize decision impact studies in genomic medicine in cancer care. This review is comprised of two parts. First, we will conduct a scoping review to catalogue the characteristics of decision impact studies. The scoping review will be followed by a bibliometric analysis to understand the role of actors and institutions in the production and dissemination of this new knowledge, by identifying inuential articles, authors, global research trends and collaboration networks. We will conduct a scoping review and a bibliometric analysis of the scoping review results. The search will include four databases, Medline, Embase, Scopus and Web of Science, using a comprehensive search strategy developed through a preliminary review of the literature. Arksey & O’Malley’s scoping review methodology, with updates by Levac et al. will be used, and the review will be reported following the PRISMA-ScR checklist. The FT Model will be used to collect and analyze data on clinical utility of decision impact studies. Our bibliometric analysis, using Bibliometrix software, will elucidate the evolution of these studies and provide data on the trends, inuences and networks emerging in the eld.

rst step to exploring their integration into regulatory decision-making, including market access and coverage.
We conducted a preliminary search of the literature to gain insight into the research, reviews and critical appraisals of this type of study. An initial search of Medline and Embase for the term "decision impact study" found studies that used the term as a methodology or referenced other "decision impact studies".
The search located a systematic review and meta-analysis of the decision impact of a 21-gene assay used for breast cancer prognostics. These results would indicate that this term is related to a distinct style of research, but no methodology or explanatory literature was found. A more extensive search returned a signi cant number of studies that reported on the impact of a genomic test on decision-making but did not use the exact term. These results illuminate an area of research that is currently understudied and con rmed the need for a more comprehensive and rigorous examination of the literature.
There is a growing literature on the production and authorization of knowledge in regulatory processes [5][6][7]. Scienti c knowledge is critical in regulatory processes, which are inherently knowledge intensive.
New technologies have uncertain properties and effects, which regulatory agencies aim to adjudicate.
Previous research has interrogated the relationships between researchers and industry and the production of evidence for regulatory and commercial success in areas of genomic medicine [5,[8][9][10][11][12], but the purpose and role of decision impact studies has yet to be explored. As new types of evidence are developed and disseminated, it is the responsibility of researchers to interrogate these sources of knowledge, to understand their place within regulatory processes and evidence-based medicine.
Health technology assessments (HTA) are important tools to assess the validity of a medical test, device or drug. HTAs establish standards and expectations for knowledge production and legitimize evidence and regulatory decisions [13][14]. Diagnostics do not directly act on health outcomes. Instead, they inform decision making about risk pro les or the use of therapeutic interventions. Efforts to measure the clinical and economic value of a test must therefore consider a "chain of evidence" linking intermediate to ultimate outcomes [15][16]. The links in this chain typically assess the analytic validity, clinical validity and clinical utility of the test or device. Clinical utility has been de ned as something that improves patient outcomes and adds value to the clinical decision-making process [16]. Clinical utility is viewed as a key standard for reimbursement decision-making, but a lack of evidence and ambiguity regarding standards for the clinical utility of genomic tests has been identi ed in the literature [7,14,[17][18][19]. As decision impact studies report on the impact of a test on decision making, these studies appear to position themselves as a form of evidence to assess clinical utility.
HTAs typically use an evaluative framework to support the assessment of the medical device, test or drug. For our review, we will leverage the FT Model [21], with Walcott et al.'s expanded categories [16] to collect and analyze data on how clinical utility is reported in DISs. This model offers the most comprehensive categorization of the components of clinical utility. The largely hierarchical and nested nature of the framework is well-suited to the context of genetics because the components of effectiveness are speci c, well de ned, and linked as a chain of evidence [16]. The Model consists of six domains of e cacy, with domains 2-6 pertaining to clinical utility: 1. Technical e cacy (i.e., laboratory performance), 2. Diagnostic accuracy e cacy (i.e., clinical sensitivity and speci city), 3. Diagnostic thinking e cacy (i.e., impact on clinician's diagnostic process), 4. Therapeutic e cacy (i.e., impact on clinical management), 5. Patient outcome e cacy (i.e., patient bene t) and . Societal outcome e cacy (i.e., cost-bene t, cost-effectiveness, societal acceptability) Walcott et al. (2021) found that the diagnostic thinking e cacy domain was not prevalent in the literature. The authors acknowledge the importance of measuring "the extent to which a test result helps a clinician come to a diagnosis and/or how the test results compare to a clinician's pretest estimate of the probability of disease"

Methods
We will conduct a scoping review and a bibliometric analysis of the scoping review results.

Scoping review
A scoping review is a useful methodology to determine the coverage of a body of literature on a given topic and identify and analyze knowledge gaps [22]. To our knowledge, there is nothing in the current literature summarizing these studies, therefore a scoping review will examine the extent, range and nature of research activity [22]. We will use Arksey and O'Malley's rationale for scoping reviews and incorporate enhancements made to this methodology by Levac et al., [23]. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement (PRISMA-ScR) will also be used to guide the reporting process of this review [24] and has been completed for this protocol (see additional le 1).

Bibliometric analysis
To give a more comprehensive and complete overview of the extent and complexity of the topic, we will conduct a bibliometric analysis of the scoping review results as a second step in our research study [25].
A bibliometric analysis is a type of social network analysis that can provide an overview of regional and time-based publishing trends, including research elds and keyword co-occurrence networks; in uential journals, authors and collaboration networks between in uential authors, countries, and institutions [25].

Identifying the research questions
Through an iterative process and based on the results of the preliminary literature review, the following research questions have been selected for this review: Identifying relevant studies We will conduct a comprehensive search of 4 databases, Medline, Embase, Scopus, and Web of Science including publications from inception of the databases. Due to resource constrains, only English language studies will be included. All publication types in these databases, e.g., reviews, conference abstracts, research studies will be included, and no study designs will be excluded. The rationale to include conference abstracts is to track origins and growth of these studies. As we are exploring the production and validation process of this new form of knowledge, this review will only include publications listed in the selected databases and not include gray literature.

Search strategy
We have developed two search strategies to identify decision impact studies in the literature. For the primary search we will use a broad search strategy to identify decision impact studies in genomic medicine in cancer care. Currently there are numerous different names and phrases used to describe "genomic testing" in cancer and this heterogeneity and lack of overarching taxonomy makes this type of search challenging. Therefore, we conducted a review of the literature and collected 47 terms, and from this list we developed a search strategy containing 28 search terms to capture genomic testing. In addition, we will include the names of 17 commercial assays. These terms, along with terms identi ed in our preliminary search to denote characteristics of a decision impact study, will be used for the primary search. Table 1 illustrates the search terms for the primary database search. Test/tool/ assay (("21-gene" or "70-gene" or "80-gene" or "recurrence score" or "50-gene" or molecular* or "genetic* pro l*" or genomic* or ("tumour* sequenc*" or "tumor* sequenc*") or ("tumour* pro l*" or "tumor* pro l*") or "gene-expression" or (multi-gene* or multigene*) or multianalyte* or "next generation" or "whole-genom*" or whole-exom* or transcriptomic* or biomarker*) adj2 (Diagnostic* or assay* or test* or pro l* or technolog* or sequenc* or panel* or signature* or classi er* or prognostic*)) or (Mammaprint or endopredict or "endo predict" or prosigna or MammoStrat or Oncotype or "Breast cancer index" or BCI or IHC4* or "genomic grade test" or "Radiotype Dx" or "AIR-CIS" or "UroVysion FISH" or Cologuard or "Lung RS" or "MyPRS Plus" or ResponseDX or "FoundationOne CDx" or Percepta) Action (test* or assess* or recommend*) and (reduc* or unnecessary or decreas* or chang* or impact* or avoid* or minimiz* or "less likely" or "bene t") and (decide* or decision* or recommend*) and (adjuvant or "treatment recommendation*" or "treatment decision*" or "therapy change*" or "therapy recommendation*" or "treatment selection" or referral* or evaluation*) "Clinical utility" and ((individual* or direct* or guide* or tailor* or target*) and (therap* or treatment*)) or ((treatment* or therap*) and (selection* or management*)) or (actionable or prognostic* or decision* or reduc* or unnecessar*) Area cancer* or neoplasm* As we are interested in the publication of decision impact studies and the origin and evolution of these type of studies broadly, we will search for items that use the exact phrase ("decision impact stud*" or "decision impact analys*" or "decision-making impact" or "decision making impact" or "decision impact") without limitation of the other search terms. This secondary search will identify articles that use this term in any area of research, topic or publication and will potentially illustrate the breadth and scope of these studies and provide data within which we will be able to contextualize decision impact studies and their role in evidence production. The results of the two searches will follow the same process, but the screening and data collection will be conducted separately. As we do not know at this time what type of study will be identi ed in the second search, we anticipate we will need to modify the data collection sheet for these results.

Study selection
Database search results will be imported into Covidence, a Cochrane technology platform, (www.covidence.org) for screening. The title and abstract screening process will be conducted by the full research team. Two reviewers (GP and HV) will screen all title/abstracts. As stated above, the title and abstract screen and full-text review for the two searches will be conducted separately. The title and abstract screening will be piloted with 25 articles for each search and results will be discussed and resolved with the research team. Full-text review for both searches will be conducted by two reviewers (GP and HV) with the third reviewer (FM) checking a random 10% sample of articles to ensure reliability. Discrepancies will be discussed and resolved collaboratively. For the results of the primary database search, articles will be excluded if they are not human, not healthcare, not genomic medicine or not cancer focused.

Data collection and extraction
The data collection worksheet will be designed iteratively. It will be piloted with 5 studies that meet the eligibility criteria and revised based on the results. The data collected for the scoping review relate to the nature and characteristics of the studies and the data collected for the bibliometric analysis relate to the nature of publication, author details and geographic and a liations. As described above, data collection for the two searches will be processed separately and we anticipate we will need to modify the data collection worksheet depending on the type of studies identi ed in the second search.
A component of our data analysis will be guided by the FT Model [21], with expanded thematic categories [16]. We will utilize the domains of the FT Model that demonstrate clinical utility to collect data pertaining to decision-making, treatment plans, patient and societal outcomes.  Analysing the data For the scoping review, the data will be collected and entered into an Excel spreadsheet for analysis and reporting. Descriptive statistics will be used to categorise and summarize the data. Categorisation will be completed by two members of the research team (GP and HV), with discrepancies resolved independently by a third reviewer (FM). All members of the research team will review the nal summary of ndings. The FT Model [21] with expanded thematic categories [16] will be used to categorize the clinical utility of the included studies. Aligning with scoping review methodology, the focus of this review is to identify and characterize decision impact studies and the quality of the studies will not be assessed. For the bibliometric analysis we will use Bibliometrix software to analyze the bibliometric data in the included studies. This analysis will be completed by two members of the research team (GP and HV). We will create a time map distribution of the literature and produce descriptive statistics and/or network visualizations for publication dates, publishing countries, impact (Impact Factor, CiteScore, and EigenFactor Score), citation count, keyword co-occurrences, coauthorships, funding, and collaboration networks for authors, institutions and countries. In addition, we plan to create hierarchical clusters based on the relevance of the words in the titles and keywords.

Discussion
The eld of genomic medicine in cancer care is rapidly advancing. Decision impact studies are an emerging form of knowledge with a potential to inform the adoption and use of genomic tests in cancer care. To date, however, little is known about how this new form of knowledge has evolved, is organized, or structured to inform assessments of clinical utility. This review is the rst step in developing this important body of knowledge. Understanding how these studies are motivated, developed and used to guide decision-making in cancer care is critical to understand their role in evidence-based regulatory and funding decision-making.
The scoping review methodology brings rigour and breadth to this task, enabling a comprehensive capture and analysis of the literature. Bibliometrics methods are an important source of complementary insight, supporting an understanding of the evolution and distribution of the actor networks that have developed this body of knowledge. There will, however, be a few limitations to this review. Due to resource limitations, we will only include English language articles and, as previously mentioned, the current heterogeneity of the terminology used to describe genomic tests may mean that the search criteria do not capture all studies on the topic.

Implications for research, policy and practice
Clinical practice, funding and market access are critical areas of healthcare that require complex decisionmaking. Understanding the origins, motivations and intended purposes of this new form of knowledge is critical to situating it in the context of its decision-making and regulatory role. This scoping review and bibliometric analysis will provide an in-depth synthesis of the research eld to-date on decision impact studies. By making the knowledge production process explicit this review can add valuable insights to evidence-informed decision-making processes.
We anticipate the ndings of this review will identify important research data to support future investigations. In our subsequent study we will leverage the knowledge gained from this review to further understanding of how decision impact studies are being used in decision-making at various levels including clinical practice, clinical practice guidelines, health technology assessment, market access regulation and funding decision-making in cancer care.