Surgical innovation is common and plays a crucial role in advancing surgical practice. It is characterized by a developmental process whereby novel procedures and devices evolve from early ideas and first-in-human studies to longer-term evaluations. Surgeons acquire important learning from incremental cases [1-3], leading to a steep learning curve in the early phases of technique development. Transparent sharing of case-by-case learning is therefore critical to promote efficient and safe innovation and timely evaluation of surgical innovation when a new technique has stabilized [2,4].

Currently, surgeons tend to innovate independently. Early incremental learning, including modifications to the technique and its outcomes, is rarely shared beyond the local team, if at all. Traditionally, dissemination of information about innovative procedures occurs at a relatively late stage through surgeon innovators presenting their technique at meetings and conferences, followed by taught courses and peer-reviewed publications [1,5]. Key incremental case-by-case learning is often not recorded. Evidence in other areas has shown that the outcomes of using an innovation are positively affected when knowledge is shared between external stakeholders [6], which can even be a source of innovation itself [7]. Increased shared learning in surgical innovation may provide similar benefits.

Disseminating new knowledge in health care is known to be challenging [8]. The process of exchanging information can be influenced by a multiplicity of factors, including organizational, cultural, social, and psychological influences [9,10] and facilitated by technology [11]. More recently, surgeon innovators are increasingly utilizing digital platforms and social media to disseminate ideas and practice [12]. While this has notable benefits, the acquired knowledge is shared inconsistently [13,14]. It may promote optimism bias by preferentially favoring positive developments and outcomes and is unsuited to building a robust evidence base [15]. Furthermore, approaches to disseminating information provide little or no scope for feedback, hindering efficient innovation that can address learning curve effects. This may also mean the benefits and harm outcomes of innovation are underreported or not shared transparently, and opportunities to promote patient safety by avoiding repetition of potentially harmful mistakes are therefore lost. Methods for effectively and transparently sharing information in real time are needed to accommodate incremental, case-by-case learning when developing a new surgical technique. Such methods must also include mechanisms for the confidential provision of feedback to avoid patient harm while simultaneously ensuring a safe space for surgical innovation.

A number of digital methods to provide feedback exist, including image analysis [16], artificial intelligence [17], or virtual and telementoring platforms [18]. These have been demonstrated to improve outcomes relevant to patients (eg, reduced operative time) and surgeons (eg, improved surgical skills) for established procedures [19,20]. It is therefore possible that similar methods could be used or adapted to capture the incremental learning associated with an innovative surgical technique in near real time. Digital methods may also enable prompt sharing to facilitate efficient, transparent, and safe innovation of novel surgical procedures.

There is no standard definition of surgical innovation, and descriptions of novelty vary considerably across the literature [21,22]. Identification of relevant literature on surgical innovation is also hindered by poor reporting [4,23,24]. Innovations are “frequently reported as information communications which may not be well organized and are sometimes anecdotal“ (pg 1) [25]. Standard systematic review methodology would therefore be unlikely to identify relevant studies consistently and reliably. A literature synthesis that adopts a broad approach to include a wide variety of publication types is required to capture a range of digital methods for shared learning in surgery in general. Initial scoping searches (using Google) showed no such review has been conducted. A scoping review is considered a suitable approach for mapping a complex topic area where no prior investigation exists [26-28]. Methods allow inclusion of a range of study designs without requiring a formal quality assessment of the included articles. A scoping review of currently available methods used to share learning in surgery can identify potentially relevant digital methods, which in turn can inform the development of novel methods to optimize shared learning in surgical innovation.

The aim of this scoping review is to explore the application of digital methods for training/education and feedback for shared learning in the context of invasive surgical procedures. We aimed to:

Results will inform strategies for embedding suitable methods within an electronic platform for real-time reporting and sharing of outcomes of surgical innovation.

A scoping review was chosen to investigate this topic due to the breadth and type of the data of interest for the research question. This scoping review will be conducted adhering to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) extension for scoping reviews (PRISMA-ScR) and established frameworks for conducting scoping reviews [26,28-30]. An initially completed checklist can be found in Multimedia Appendix 1 [28], and an updated checklist will be provided upon publication of results. Two trained reviewers will conduct the review, with input from a multidisciplinary team consisting of surgeons, methodologists, health services researchers, and social scientists. Any necessary deviations from the current protocol will be reported in the completed manuscript.

Ethical approval for this scoping review was not required, because it does not involve human participants, their tissue, data or samples or has ethical implications, as outlined in institutional policies including Section 2 of the University of Bristol's Ethics of Research Policy and Procedure.

A total of 6 approaches will be followed to identify relevant data sources detailed above. Collaboration with a subject librarian will aid optimization of searches and inclusivity of search terms throughout.

Study eligibility criteria are defined according to the Population, Concept, and Context framework [38] and are presented in Table 2. Publications will only be considered if they are dated within 5 years of their original publication date to ensure data is contemporary. Articles will also be excluded if they are not published in the English language, due to resource restrictions that prevent the translation of non-English articles.

For records with common publication formats (ie, published a title, structured abstract, executive summary, or synopsis), a 2-stage screening process will be undertaken to assess records for inclusion against the study eligibility criteria.

Search results will be downloaded from their respective online databases, deduplicated, and uploaded to an online review manager (Rayyan) [39]. In a first step, 3 review authors (CH, MK, and JR) will independently (each review author will be blind to the screening choice of the others) screen the titles and abstracts, executive summaries, or synopses of the retrieved records, with 10% of records double-screened. Full texts of articles will be obtained from records meeting the inclusion criteria and from those where inclusion remains uncertain (eg, because of a lack of information from the abstract). In a second step, 2 reviewers will each screen half of the retrieved full texts independently to assess full eligibility. Duplicate assessment of eligibility will be performed on 10% of all full texts with further duplicate reviews in case of poor agreement (<80%).

It is anticipated that some potentially relevant records do not follow conventional publication formats (eg, tweets, opinion pieces, news articles). In this case, 2 reviewers will independently review the content of the record in full.

Discrepancies at any stage of the screening process will initially be discussed between the 2 review authors. A third independent reviewer (SP) will arbitrate where agreement on inclusion could not be reached, and input from the wider team will be sought where necessary.

Data extraction will be performed directly into a purposely designed electronic data extraction form (eg, Microsoft Access; Microsoft Corp). Details about (1) study and publication characteristics (eg, author, study design, funding, and sponsorship statements), (2) the method for shared learning (eg, purpose, type, operationalization, and modality), and (3) impact of methods for shared learning (eg, methodology of evaluating methods for shared learning and their results, limitations, and author recommendations) will be extracted. Additional items of interest for social media posts will be explored to capture further information on their content (eg, presence of a link to an external website). An initial data extraction form will be piloted with a small number of relevant articles (see Multimedia Appendix 3). The form will be iteratively refined to comprehensively capture all relevant detail emerging during the pilot.

One review author will extract data from all included studies, and a second reviewer will independently perform double data extraction for at least 10% of articles. Consistency in the approach to data extraction will be ensured through constant dialogue between the 2 reviewers.

Findings will be summarized in tables using descriptive statistics and in narrative form. Verbatim extracted data will be analyzed by 2 reviewers adhering to principles of thematic analysis [40]. Identified themes will be displayed in schematics. Any verbatim extracted data will be reviewed to identify barriers and benefits to the implementation of shared learning methods. Barriers will be considered factors that impede the implementation of methods for shared learning in clinical practice. Benefits will be considered those that enable implementation [41]. Two reviewers will code data as a barrier or benefit, whereby regular meetings will be held to discuss coding results, and senior authors will be involved where consensus is required. In case an automated dashboard vendor is used to identify social media posts, the content will undergo additional analysis using the software’s existing classification algorithms and analyses (eg, sentiment analysis).