Noninferiority Margin Size and Acceptance of Trial Results: Contingent Valuation Survey of Clinician Preferences for Noninferior Mortality. (August 2022)

• Record Type: Journal Article
• Authors: Pong, Sandra; Fowler, Robert A.; Mitsakakis, Nicholas; Murthy, Srinivas; Pernica, Jeffrey M.; Gilfoyle, Elaine; Bowen, Asha; Fontela, Patricia; Seto, Winnie; Science, Michelle; Hutchison, James S.; Jouvet, Philippe; Rishu, Asgar; Daneman, Nick
• Is Part Of: Medical decision making. Volume 42:Number 6(2022)
• Page Start: 832
• Page End: 836
• Publication Date: 2022-08
• Subjects: antimicrobials -- contingent valuation -- duration of therapy -- evidence uptake -- mortality -- noninferiority margin; Medical policy -- Periodicals; Clinical medicine -- Decision making -- Periodicals; Medicine -- Periodicals; Médecine clinique -- Prise de décision -- Périodiques; 362.1
• DOI: 10.1177/0272989X221099493
• Languages: English
• ISSNs: 0272-989X

Abstract

Objectives: We used modified contingent valuation methodology to determine how noninferiority margin sizes influence clinicians' willingness to accept clinical trial results that compare mortality in critically ill children. Methods: We surveyed pediatric infectious diseases and critical care clinicians in Canada, Australia, and New Zealand and randomized respondents to review 1 of 9 mock abstracts describing a noninferiority trial of bacteremic critically ill children assigned to 7 or 14 d of antibiotics. Each scenario showed higher mortality in the 7-d group but met noninferiority criterion. We explored how noninferiority margins and baseline mortality rates influenced respondent acceptance of results. Results: There were 106 survey respondents: 65 (61%) critical care clinicians, 28 (26%) infectious diseases physicians, and 13 (12%) pharmacists. When noninferiority margins were 5% and 10%, 73% (24/33) and 79% (27/33) respondents would accept shorter treatment, compared with 44% (17/39) when the margin was 20% ( P = 0.003). Logistic regression adjusted for baseline mortality showed 5% and 10% noninferiority margins were more likely to be associated with acceptance of shorter treatment compared with 20% margins (odds ratio [OR] 3.5, 95% confidence interval [CI]: 1.3–9.6, P = 0.013; OR 5.1, 95% CI: 1.8–14.6, P = 0.002). Baseline mortality was not a significant predictor of acceptance of shorter treatment. Conclusions: Clinicians are more likely to accept shorter treatment when noninferiority margins are ≤10%. However, nearly half of respondents who reviewed abstracts with 20% margins were still willing to accept shorter treatment. This is a novel application of contingent valuation methodology to elicit acceptance of research results among end users of the medical literature. Highlights: Clinicians are more likely to accept shorter treatment durations based on noninferior mortality results when the noninferiority margin is 5% or 10% than if the margin is 20%. However, nearly half of clinicians would still accept shorter-duration treatment as noninferior with margins of 20%. Baseline mortality does not independently predict acceptance of shorter-duration treatment. Contingent valuation is a novel approach to elicit the acceptance of research design parameters from the perspective of endusers of the medical literature.