Background: Hospitals increasingly make decisions about early development of and investment in innovative medical technologies (IMTs), but at present often without an early assessment of their potential to ensure selection of the most promising candidates for further development. This paper explores how early assessment is carried out in different health organisations and then discusses relevant learning points for hospitals.
Methods: A qualitative study design with a structured interview guide covering four themes was used. Content analyses of interview notes were performed covering four predetermined themes: context, basis for decision-making, process and structure, and perceptions. A fifth theme, handling cognitive bias, was identified during data analysis.
Results: A total of 11 organisations participated; eight from the private health industry and three public hospitals. The interviews identified four areas in which early assessment is performed in similar manner across the studied organisations and four areas where differences exist between public hospitals and private organisations. Public hospitals indicate a lower degree of formalised early assessment and less satisfaction with how early assessment is performed, compared to private organisations. Based on the above findings, two learning points may carry promise for hospitals. First, having dedicated prioritising committees for IMTs making stop/go decisions. This committee is separate from the IMT development processes and involved staff. Secondly, the committee should base decisions on a transparent early assessment decision-support tool, which include a broad set of domains, is iterative, describes uncertainty, and minimise cognitive biases.
Conclusions: Similarities and differences in the way early assessment is done in different health organisations were identified. These findings suggest promising learning points for the development of an early assessment model for hospitals.
Keywords: Early assessment; Health technology assessment; Innovation; Interview.