Objective: To investigate whether consensus differs when reached by the Nominal or the Delphi method.
Study design and setting: Seventeen general practices from North Staffordshire, England were randomly allocated to Delphi (postal feedback only) or Nominal group (also had group discussion). General practitioners classified 56 morbidities according to four scales of severity (chronicity, time course, health care use, patient impact) in two consensus rounds. Consensus outcomes were assessed by between-group comparison of severity scores at baseline and follow-up rounds, and consensus process by within-group change in the variance of severity scores between the two rounds.
Results: Consensus rounds were completed by 21 out of 35 Nominal GPs and 23 out of 43 Delphi GPs. Baseline scores for three of the four severity scales were significantly higher for Nominal compared to Delphi GPs, but there were no differences at follow-up. Between the two rounds, variance reduced within the Nominal and Delphi group, respectively, by 61% and 35% (chronicity), 40% and 62% (time course), 42% and 36% (health care use), and 19% and 38% (patient impact).
Conclusion: The Nominal and Delphi methods did not result in different outcomes and we conclude that either method can be used in health services research.