Background: The clinical impact of risk score use in end-of-life settings is unknown, with reports limited to technical properties.

Methods: We conducted a mixed-methods study to evaluate clinical impact of a validated mortality risk score aimed at informing prognosis and supporting clinicians in decision-making in dementia patients with pneumonia. We performed a trial (n = 69) with physician-reported outcomes referring to the score's aims. Subsequently, physician focus group discussions were planned to better understand barriers to clinical impact, and we surveyed families (n = 50) and nurses practicing in nursing homes (n = 29). We finally consulted with experts and key persons for implementation.

Results: Most (71%) physicians who used the score considered it useful, but mainly for its learning effects. Families were never informed of numerical risk estimates. Two focus group discussions revealed a reluctance to use a numerical approach, and physicians found that outcomes conditional on antibiotic treatment were inadequate to support decision-making. Nurses varied in their perceived role in informing families. Most families (88%) wished to be informed, preferring a numerical (43%), verbalized (35%), or other approach (18%) or had no preference (5%). Revising the score, we added an ethical framework for decision-making to acknowledge its complexity, an explanatory note addressing barriers related to physicians’ attitudes, and a nurses’ form.

Conclusion: The combined quantitative and qualitative studies elicited: substantial barriers to a numerical approach to physicians’ end-of-life decision-making; crucial information for revisions and further score development; and a need for implementation strategies that focus on education.