A Modified Delphi for ECPR Post Resuscitation Care

The study will be performed using a modified Delphi process with an international expert panel to generate recommendation statements related to the management of patients treated with ECPR and to determine the level of agreement for each statement from the panel as a whole. The following topics will be covered:

Sedation Mechanical ventilation management Oxygen management Carbon dioxide management Haemodynamic management ECMO flow management Mean arterial pressure targets Aortic valve not opening management Left ventricular unloading/venting Complication management Infection prevention Anticoagulation Imaging Prognostication and neuro-prognostication Practical recommendations Organ donation Patient followup

Cannulation will not be covered as this has been covered in other Delphi processes.

Management Consensus is achieved through iterative rounds of survey and revision to the statements until a pre-determined level of agreement is met (>70% agreement).

Part 1: planning and evidence review Series of planning sessions with co-chairs Literature is reviewed/summarized Initial statements will be developed based on evidence and study team expertise

Expert panel selection:

The study team identified experts on ECPR for OHCA based on contributions to the field, clinical experiences, range of specialties and peer-reviewed publications with prioritization of diversity in geography and expertise. Using snowball sampling, investigators will ask the recruited experts to identify additional content experts to participate in the expert panel. It is aimed to recruit a total of 50 experts, in accordance with the recommended panel number of >30 as outlined in the original Delphi method paper. To avoid any bias, the study team will not participate in the voting process. The survey and voting process will be online and anonymous to avoid any reciprocal influence.

Part 2: Survey to achieve consensus Survey, based on the initial statements developed by study team, will be developed. This will be reviewed by independent contributors (not part of the study team or the expert panel), to ensure validity, appropriate topic coverage and length.

Some questions will enable a Yes/No or quantitative response. Others will be utilising a Likert scale. The voting process will be anonymous to avoid any reciprocal influence or dredging effects and done with Research and Development/University of California, Los Angeles (RAND/UCLA) Appropriateness Method Scale.

Experts will be asked to rate the importance of each item for ECPR for OHCA using the Grading of Recommendations Assessment, Development and evaluation scale from 1 (strongly disagree) to 9 (strongly agree). Statements will be classified as strong agreement (median score 7-9), neither agree or disagree (median score 4-6), or strongly disagree (median score 1-3). Panel members will also provided an "Unable to Score" response to use if they did not feel comfortable rating any specific domain. Recipients will be asked to recommend any wording/phrasing changes and to suggest any additional items for inclusion.

Based on a previously established Delphi method approach, consensus will be achieved when an item reached a high level of agreement for inclusion. High level of agreement will require greater than 70% of the survey responses for a particular item meeting a score of strong agreement (i.e. a score ≥ 7 on the Likert scale). A medium level of agreement will be defined between 60% and 70%, and low agreement below 60%. Each expert will receive anonymized aggregate statistics (group mean, group standard deviation, percentage agreement). Statements achieving a high level of agreement in the first survey (i.e. a score of 70% of respondents with a score >7) will not be reviewed in the consensus meeting nor in the second survey. Those not meeting a high level of agreement will be rephased during the online meeting or subsequent surveys. If 70% of respondents score a statement at 3 (on the 9-point scale) or less for a statement in the first-round survey, that statement will be deemed not important and removed and not discussed at the online discussion.

An online consensus meeting will be arranged after the first round where the expert panel will discuss the items in detail, including comments on the first survey. Modifications to questions will be possible at this meeting.

Following the online consensus meeting, additional survey rounds will be prepared and sent to each expert as determined by the study group. This survey will include each item with comments indicating changes made during the consensus meeting as well as the mean score and level of agreement (low, medium, high) for each item from the first survey (excluding statements of low importance/agreement as defined above). Expert participants will again rate each item on the same 9-point Likert scale. Descriptive statistics including median, IQR, and level of agreement will be obtained for each item from the final survey. Items that reach a high level of agreement will be included in the final list of consensuses. Items that reached medium or low agreement will be discarded after the final round but will be provided as supplementary data for review.

It is anticipated that 3 rounds of the Delphi process will be required. Additional rounds may be considered if the study group and expert group view additional consensus may be made on key items with restructuring of statements.

The percentage agreement will be quantified as the number of individual scores of more than 7 divided by the number of voting experts. Given the poor- and low-quality data available on many aspects of topic, the study team will not be using the Grading of Recommendation Assessment Development and Evaluation (GRADE) to prepare statements but will referrer to the Appraisal of Guidelines for Research and evaluation (AGREE) statement. This approach is consistent with other recently published guidelines.