Exploring physician approaches to conflict resolution in end-of-life decisions in the adult intensive care unit: protocol for a systematic review of qualitative research

Introduction Conflict is unfortunately well-documented in the adult intensive care unit (AICU). In the context of end-of-life (EOL) decision-making (ie, the withdrawal or withholding of life-sustaining treatment), conflict commonly occurs when a consensus cannot be reached between the healthcare team and the patient’s family on the ‘best interests’ of the critically ill, incapacitated patient. While existing literature has identified potential methods for conflict resolution, it is less clear how these approaches are perceived and used by stakeholders in the EOL decision-making process. We aim to explore this by systematically reviewing and synthesising the published evidence, which addresses the following research question: what does existing qualitative research reveal about physician approaches to addressing conflict arising in EOL decisions in the AICU? Methods and analysis Peer-reviewed qualitative studies (retrieved from MEDLINE, Project Muse, Scopus, EMBASE, Web of Science, PsycINFO, CINAHL, and LILACS) examining conflict and dispute resolution in the context of EOL decisions in the AICU setting will be included. Two reviewers will independently screen either all or a randomly selected sample of studies, with a third reviewer independently screening studies of uncertain eligibility. The ‘thematic synthesis’ approach will be employed to analyse the resulting data. The quality of included papers will be assessed using the 2018 Mixed-Methods Assessment Tool. The ‘Grading of Recommendations, Assessment, Development, and Evaluations-Confidence in the Evidence from Reviews of Qualitative research’ approach will be used to assess our confidence in the findings. Ethics and dissemination Ethical approval is not required for this review, as only published data will be included. We anticipate that the findings will be of interest to healthcare professionals working in AICUs and individuals working in bioethics, given the ethically contentious nature of EOL decisions. The findings will be disseminated at academic conferences and through open-access publication in a peer-reviewed journal. PROSPERO registration number CRD42021193769.

The primary issue is a potential concept issue within the search strategy. Definition of the EOL decisions and relating them to "the withdrawal or withholding of life-sustaining treatments" conflict with the search term "Refusal to Treat" and focus on the aim of this protocol. Therefore, the authors should clarify the following section to avoid confusion. Please bear in mind that the "withdraw*" and "withhold*" keywords can connect to the "Withholding Treatment" MeSH term (https://www.ncbi.nlm.nih.gov/mesh/68028761), which is opposite of the "Refusal to Treat" MeSH term (https://www.ncbi.nlm.nih.gov/mesh/68016079) as is defined as: "Refusal of the health professional to initiate or continue treatment of a patient or group of patients".
Due to the wide range of years in the search, the search strategy needs to be more flexible to gather possible terms. For example, not all papers are indexed under the MeSH term "Terminal Care" from 1980until today. Therefore, the recommendation is that the authors need to include all possible MeSH and all possible keywords to the related term(s). In this specific case, it could also be helpful to use the MeSH term "Terminal Ill", "Palliative Care", etc. To be sure that all (possible) papers are retrieved.
One suggestion, for the keyword, search the .mp search field is excellent in the Medline database for solo search, but because there is a multiple database search, I advise authors to be consistent in the search and search fields across the other databases. Please consider to use .ab (abstract) and .ti (title) fields instead of the .mp. Also, please consider using the abbreviations of the EOL, (A)ICU in the search strategy because we never know how the title, abstract, or keywords are constructed. For example, authors can use (Ovid MEDLINE) for the "End of life": (End adj2 life).ab,ti (End adj2 life care).ab,ti EOL.ab,ti. EOLC.ab,ti. EOL care.ab,ti.
Suggestions are on the same page as the definition and application of the methodological filters that the authors mentioned in the discussion. The description of Boolean operators is redundant in describing the search strategy. (Information source) Because of the diversity of the selected databases, I would suggest that authors divide databases by types: bibliographic, complete text databases, etc. Because of the inclusion of medicine, social sciences, and bioethics, I would suggest including a few more databases covering all fields, such as Scopus and PubMed and related bibliographic databases. The authors should consider explaining or defining why did they take the 1980 year as a star year. The commencement dates of databases it is not relevant for the search strategy. For the protocol search, it would advise having the same start year. Regarding the grey literature, I would like to reassure the authors that they do not need to be afraid of the grey literature and consider using this protocol for application in the grey literature. To conclude, there are some crucial methodological changes, but if the suggestions are not applicable from the authors, I would like to receive more information to clarify the choice of the problems mentioned above.
The authors of the proposed protocol are off to a good start; however, this protocol requires additional clarification and possible changes in the search strategy and information sources. Although it is a methodological issue, it should not disturb the flow of the proposed protocol.
Thank you to the reviewer for their comments.
In the search strategy, the authors are referring to the search strategy from the paper (ref. 18 -Schram A, Hougham G, Meltzer D, Ruhnke G. Palliative care in critical care settings: A systematic review of communication-based…) which is not so applicable for this protocol because it is off-topic. Therefore, the proposition is that the authors use mentioned article as an inspiration or starting point for key search terms in this protocol and explain why they used their selected MeSH terms and keywords.
While the systematic review by Schram and colleagues does not address the same research question as this systematic review, there is some overlap in the issues (end of life care) and setting (critical care). Hence, their search strategy was used as a starting point, but it was further defined through consultation with a librarian and review of the MeSH database. This has been explained in the manuscript.
The primary issue is a potential concept issue within the search strategy. Definition of the EOL decisions and relating them to "the withdrawal or withholding of life-sustaining treatments" conflict with the search term "Refusal to Treat" and focus on the aim of this protocol. Therefore, the authors should clarify the following section to avoid confusion. Please bear in mind that the "withdraw*" and "withhold*" keywords can connect to the "Withholding Treatment" MeSH term (https://www.ncbi.nlm.nih.gov/mesh/68028761), which is opposite of the "Refusal to Treat" MeSH term (https://www.ncbi.nlm.nih.gov/mesh/68016079) as is defined as: "Refusal of the health professional to initiate or continue treatment of a patient or group of patients".
EOL decisions are here defined as the withdrawal or withholding of life-sustaining treatments, and in the context of ICU decision-making, the decision to withdraw/withhold could be the decision of the patient, the patient's representatives, or the ICU team. The "Withholding Treatment" MeSH term relates to withholding treatment from a patient or research subject. This is differentiated from, but not necessarily the opposite of, "Refusal to Treat" -where the emphasis is on the health professional's refusal to treat a patient, when the patient or patient's representatives request treatment. Both of these scenarios can give rise to conflicts in EOL decision-making (although conflict is more implicit in the "Refusal to Treat" MeSH term), hence both of these MeSH terms have been included in the search strategy. Other terms, such as "Resuscitation Orders" have also been included by using the 'Explode' term. A footnote has been added to explain this -thank you to the reviewer for pointing this out.
Due to the wide range of years in the search, the search strategy needs to be more flexible to gather possible terms. For example, not all papers are indexed under the MeSH term "Terminal Care" from 1980until today. Therefore, the recommendation is that the authors need to include all possible MeSH and all possible keywords to the related term(s). In this specific case, it could also be helpful to use the MeSH term "Terminal Ill", "Palliative Care", etc. To be sure that all (possible) papers are retrieved.
Further MeSH terms and keywords to the related terms have been added to the search strategy, to ensure that all (possible) papers are retrieved. These have been evidenced in the supplementary file.
One suggestion, for the keyword, search the .mp search field is excellent in the Medline database for solo search, but because there is a multiple database search, I advise authors to be consistent in the search and search fields across the other databases. Please consider to use .ab (abstract) and .ti (title) fields instead of the .mp.
Thank you to the reviewer for this suggestion. We have amended the search strategy to include "titles, abstracts, and keywords" as these are consistent fields across the included databases. Subject headings across the databases have still been utilised, where possible, and modified for the databases as needed. The full search strategy is documented in the supplementary file.
Also, please consider using the abbreviations of the EOL, (A)ICU in the search strategy because we never know how the title, abstract, or keywords are constructed. For example, authors can use (Ovid MEDLINE) for the "End of life": • (End adj2 life) Suggestions are on the same page as the definition and application of the methodological filters that the authors mentioned in the discussion.
be feasible to assess the full-texts of PhD thesis, for example, for eligibility. Similarly, conference abstracts or blogs may not offer enough detail. Therefore, grey literature has not been included.
To conclude, there are some crucial methodological changes, but if the suggestions are not applicable from the authors, I would like to receive more information to clarify the choice of the problems mentioned above. Thank you to the reviewer for their comments.