The Development and Validation of a Generic Instrument, QoDoS, for Assessing the Quality of Decision Making.
Salek, Mir-Saeed Shayegan
Introduction: The impact of decision-making during the development and the regulatory review of medicines greatly influences the delivery of new medicinal products. Currently, there is no generic instrument that can be used to assess the quality of decision-making. This study describes the development of the Quality of Decision-Making Orientation Scheme QoDoS© instrument for appraising the quality of decision-making. Methods: Semi-structured interviews about decision-making were carried out with 29 senior decision makers from the pharmaceutical industry (10), regulatory authorities (9) and contract research organizations (10). The interviews offered a qualified understanding of the subjective decision-making approach, influences, behaviors and other factors that impact such processes for individuals and organizations involved in the delivery of new medicines. Thematic analysis of the transcribed interviews was carried out using NVivo8® software. Content validity was carried out using qualitative and quantitative data by an expert panel, which led to the developmental version of the QoDoS. Further psychometric evaluations were performed, including factor analysis, item reduction, reliability testing and construct validation. Results: The thematic analysis of the interviews yielded a 94-item initial version of the QoDoS© with a 5-point Likert scale. The instrument was tested for content validity using a panel of experts for language clarity, completeness, relevance and scaling, resulting in a favorable agreement by panel members with an intra-class correlation coefficient value of 0.89 (95% confidence interval = 0.56, 0.99). A 76-item QoDoS© (version 2) emerged from content validation. Factor analysis produced a 47-item measure with four domains. The 47-item QoDoS© (version 3) showed high internal consistency (n=120, Cronbach’s alpha = 0.89), high reproducibility (n=20, intra-class correlation =0.77) and a mean completion time of 10 min. Reliability testing and construct validation was successfully performed.