OBJECTIVES: Delphi methodology offers a robust approach to gather expert opinions. We conducted an online modified Delphi panel to gain insights into early PCa epidemiology and treatment pathways in Europe; here, we reflect on our experience of this methodology.

METHODS: We utilized a pre-specified Delphi protocol and anonymous online market research panel to seek consensus on areas of early PCa epidemiology and treatment pathways. Questions were generated with an expert Steering Committee. Panelists were urologists and radio-oncologists from France, Germany, Spain, Italy and Belgium. Three rounds were planned; two were completed (Round 1, N=113; Round 2, N=72), followed by qualitative interviews of Delphi respondents (n=10), to investigate results.

RESULTS: Whilst consensus was reached for several questions, some results from both Delphi rounds showed variation by and within geography, specialty, and hospital setting. Qualitative interviews indicated that some variation in responses (e.g., in Round 2, metastatic cases at first diagnosis: mean 37.2%, range 0–75%) stemmed from genuine differences in respondents’ clinical experience and caseload. However, despite efforts to clarify wording, add definitions, and include schematics in Round 2, it was suggested that question concepts/definitions were not always interpreted as intended and potentially led to additional variation. This was partly attributed to using an online survey and market research panel, which limited opportunities for respondents to raise uncertainties and discuss with the research team and peers.

CONCLUSIONS: A combination of inherent variation in caseload, practice, and between-country differences combined with methodological limitations with an online modified Delphi panel using market research techniques limited the ability to reach consensus on some questions. We hypothesis that conducting Delphi panels using an online survey and market research panel, even with written guidance, yields limitations in achieving consensus. Allowing flexibility for dialogue to facilitate understanding of questions, whilst maintaining independence in voting to minimize social bias, should be considered.