Systematic PubMed search
Data gathering was performed using MEDLINE® database (http://www.ncbi.nlm.nih.gov/pubmed). To identify relevant literature, the search terms “video laryngoscopy” and “video laryngoscope” were used.
A total of 3351 studies were identified by the applied search strategy in PubMed (Fig. 1). First, a filter restricting the time period of the search (10 years range; going from 22/June/2007 to 22/June/2017) was applied. The PubMed® article categories selected were “Clinical Study”, “Clinical Trial”, “Comparative Study”, “Controlled Clinical Trial”, “Evaluation Studies”, “Multicenter Study”, “Observational Study” and “Randomized Controlled Trial”. The final raw dataset consisted of 582 papers (the number of results is referred to a search made on 11/July/2017).
The complete list of items, including whole article names, authors and PubMed® URLs as well as the table of the results sorted by year was downloaded directly from PubMed® in a CSV format. Manuscripts presenting scientific data on video laryngoscopy as well as outcome parameters were included for analysis. If outcome parameters were not presented, the specific manuscript was excluded for analysis.
Data analysis
Papers were screened manually by two anaesthesiologists to identify study endpoints. For each study, “primary outcome”/“primary endpoint” or “secondary outcome”/“secondary endpoint” were collected if clearly stated in the abstract and/or in the full text (when available in the University of Naples Federico II or University of Cologne digital libraries).
If a study did not contain parameters described as “primary outcome/endpoint” or “secondary outcome/endpoint”, alternative measurements were included in the analysis. Alternatively, the “aim/goal/objective/target/purpose/null hypothesis” information was used from the abstract. Also, all studies evaluating directly a video laryngoscopy system or evaluating how the effectiveness of these laryngoscopes could be improved by other ancillary devices (e.g., different endotracheal tubes, stylets, gum elastic boogie, etc.) were included.
A chess-like table was hence built with the several outcomes as columns and 372 papers included as rows (Fig. 2). Outcomes from different studies, having different names but concerning the same variable were fused in a single column (e.g., “dental compression” and “number of audible dental click sounds”). Outcomes concerning a group of variables, slightly different among the studies but mostly overlapping were fused, as well.
Data processing
The number of row-column matches for each column/category was reported in a different table and the ratio between it and the total number of included articles was calculated. Derived from the percentages of each parameter, a suggestion was provided of which parameters should be reported in future video laryngoscopy studies to facilitate study comparisons.
Generation of recommendations
The chosen outcomes outcomes need to be relevant to both health care providers and healthcare users on one hand, and also to those involved in making decisions and choices about health care, on the other hand [12]. However, a lack of attention for using clinical outcomes in studies has led to avoidable losses in both the production and reporting of research. Moreover, the outcomes which have been included in studies have not always been those being most important or relevant for patients [15]. To develop a consensus between the authors concerning use of different parameters, the Delphi system was used [16, 17].
To develop relevant recommendations for video laryngoscopy studies, the four-step Core Outcome Set (COS) process was used. First, the scope was defined, followed by checking if a set exists as a second step. Third, a procedure for the development of the COS was defined and last, it was defined what specific parameters should be measured in future studies [12].
In a COS framework, the method is used for achieving convergence of opinion from experts on the importance of different outcomes in sequential questionnaires (or rounds) sent either by post or electronically [12]. In the present study, all authors (n = 4) participated in the process. Three rounds were planned. The answers for each of the outcomes were summarised and fed back anonymously until a consensus was reached with at least 75%. After considering the views of others before re-rating each item, participants were able to change their initial responses based on the feedback from the previous rounds. Direct communication concerning the specific parameters was not possible. Therefore, the feedback provides a mechanism for reconciling different opinions of participants and is essential to achieving a consensus [12]. In terms of the overall validity for the final consensus, this approach has significant advantages as compared to round-table discussions [18].