We have recently published a systematic review of the long term outcomes of both surgical and conservative treatment for radial dysplasia. This was prospectively registered with the PROSPERO database (CRD42016036665) and conducted using the Cochrane highly sensitive search strategy. We searched Medline and Embase via OvidSP, PubMed, Cochrane, ClinicalTrials.gov and the WHO International Clinical Trials portal for published and unpublished studies. Searches were not restricted by date or language. From all studies identified which reported outcomes for RD treatment, we have extracted a list of outcomes measured for radial dysplasia following the process in the COMET handbook (section 18.104.22.168), which will form the starting point for our Delphi process.
Identification of outcomes of importance to patients, parents and clinicians
To achieve consensus on a COS for RD within and between groups, we propose to use an online Delphi process, adapted and simplified from the protocol laid out by Harman et al . This will include four groups of participants;
· RD patients aged over 16
· Parents of RD patients aged under 16
· Hand therapists who treat RD patients
· Hand surgeons who treat RD patients
These groups were chosen to reflect patient and family perspectives equally with surgical and therapist perspectives. We aim to make the groups of similar size. Patient and parent groups will be drawn from across the UK, and clinician groups drawn from specialist centres internationally. The study will be managed from Great Ormond St Hospital in London, and participants will be recruited via email by the central research team, following identification by participating specialist centres worldwide. The Delphi process will be administered using the secure DelphiManager software at the University of Liverpool. The study process is summarised in figure 1.
Identification of potential outcomes
Our initial outcome list has been generated from the outcomes identified during our systematic review. We included all outcomes that we identified in the global literature on RD treatment. Composite outcome scores have been split into their component parts, where possible. The outcome list will be presented thematically, by outcome domain.. Outcomes will be described using lay language, then medical language beneath, with a diagram or picture as required. The draft outcome list will be piloted with members of all stakeholder groups before use to ensure it is easily understood and clear.
Participants and stakeholder selection
We aim to recruit the patient and parent groups via specialist centres across the UK, and the therapist and surgeon groups from global specialist centres. All participants will be required to be proficient with spoken and written English, and to have access to a computer and internet connection.
Round 1 – initial ranking and finalising outcomes considered
Participant identification centres have been identified by the review authors. Potential participants will be identified by each centre locally, then invited to participate in the Delphi process by the central research team at Great Ormond St. Those who agree will be invited to register with the DelphiManager platform, then sent an email linking to an online survey, listing each identified outcome thematically, by outcome domain, and providing an open question at the end to nominate important missing outcomes. They will be asked to rank each outcome 1-9, where 1-3 is ‘not important’, 4-6 is ‘important but not critical’ and 7-9 is ‘critical’. The online survey will allow a review author (GM), who will not himself participate, to identify who has completed the survey. Participants will be given three weeks to complete the survey, with a reminder email being sent after one and two weeks.
Analysis of round 1
Newly suggested outcomes will be reviewed by two review authors (GM and BS) to ensure they are genuinely novel, then combined and added to the round two questionnaire accordingly. Results will be analysed by participant group and in overall summary, noting the number participating and the distribution of scores per outcome. Outcomes meeting the ‘consensus out’ definition will be excluded from the round two questionnaire. Individual participation in round two will be contingent upon completing the survey in round one.
Participants will again be contacted by email with a link to the online survey. For each previously scored outcome, they will be presented with a histogram summary of the responses for each group and for all groups combined, plus a reminder of their previous score. They will then be asked to re-score each outcome, again from 1-9, and then to score any newly suggested outcomes identified in round one. Participants will be given three weeks to complete the round two survey, with a reminder email being sent after one and two weeks.
Analysis of round 2
Results will be analysed both by participant group and in overall summary, noting the number participating and the distribution of scores per outcome. Outcomes meeting the ‘consensus out’ definition will be excluded from the round three questionnaire. Individual participation in round three will again be contingent upon completing the survey in round two.
Participants will again be contacted by email with a link to the online survey. For each outcome they will be presented with a histogram summary of the responses for each group and for all groups combined, plus a reminder of their round 2 score. They will then be asked to re-score each outcome, again from 1-9. Participants will be given three weeks to complete the round three survey, with a reminder email being sent after one and two weeks.
Analysis of round 3
Results will be analysed both by participant group and in overall summary, noting the number participating and the distribution of scores per outcome. Outcomes will be classified as consensus in, consensus out or no consensus using the criteria in table 1. The distribution of scores and consensus result for each outcome will be displayed by group and overall and used to structure the final consensus meeting.
A final consensus will be reached during a consensus meeting, which may involve a mixture of face to face and teleconference participation. All participants in the Delphi survey will be invited. All participants will receive the results of round three in advance, presented by group and overall. We will then follow the COSMIN guidelines to select outcome measurement instruments. To avoid duplication of effort by patients, clinicians and researchers, we will measure outcomes which overlap with the ICHOM standard set for congenital upper limb anomalies in the same way. The final COS will be published in a peer-reviewed journal.
Definition of consensus
We will use the definition of consensus from Harman et al , summarised in table 1. This will be applied to the combined group scores from round 3.
Statistical analysis and Sample size
Scores for each item will be presented as a histogram of responses, and the percentage of responses in each group (1-3, 4-6, 7-9) calculated. This will be done by individual stakeholder group, and for all groups combined.
As there is no standard model for the sample size required for a Delphi process, we will aim for between 10 - 15 participants per group, with the patient and parent groups covering patients with a variety of disease severity and age.