N E Pedersen1,2, D Oestergaard1,2, A Lippert1. 1. Copenhagen Academy for Medical Education and Simulation, Herlev Hospital, Herlev, Denmark. 2. Department of Clinical Medicine, Copenhagen University, Copenhagen N, Denmark.
Abstract
INTRODUCTION: When investigating early warning scores and similar physiology-based risk stratification tools, death, cardiac arrest and intensive care unit admission are traditionally used as end points. A large proportion of the patients identified by these end points cannot be saved, even with optimal treatment. This could pose a limitation to studies using these end points. We studied current expert opinion on end points for validating tools for the identification of patients in hospital wards at risk of imminent critical illness. METHODS: The Delphi consensus methodology was used. We identified 22 experts based on objective criteria; 17 participated in the study. Each expert panel member's suggestions for end points were collected and distributed to the entire expert panel in anonymised form. The experts reviewed, rated and commented the suggested end points through the rounds in the Delphi process, and the experts' combined rating of the usefulness of each suggestion was established. RESULTS: A gross list of 86 suggestions for end points, relating to 13 themes, was produced. No items were uniformly recognised as ideal. The themes cardiac arrest, death, and level of care contained the items receiving highest ratings. CONCLUSIONS: End points relating to death, cardiac arrest and intensive care unit admission currently comprise the most obvious compromises for investigating early warning scores and similar risk stratification tools. Additional end points from the gross list of suggested end points could become feasible with the increased availability of large data sets with a multitude of recorded parameters.
INTRODUCTION: When investigating early warning scores and similar physiology-based risk stratification tools, death, cardiac arrest and intensive care unit admission are traditionally used as end points. A large proportion of the patients identified by these end points cannot be saved, even with optimal treatment. This could pose a limitation to studies using these end points. We studied current expert opinion on end points for validating tools for the identification of patients in hospital wards at risk of imminent critical illness. METHODS: The Delphi consensus methodology was used. We identified 22 experts based on objective criteria; 17 participated in the study. Each expert panel member's suggestions for end points were collected and distributed to the entire expert panel in anonymised form. The experts reviewed, rated and commented the suggested end points through the rounds in the Delphi process, and the experts' combined rating of the usefulness of each suggestion was established. RESULTS: A gross list of 86 suggestions for end points, relating to 13 themes, was produced. No items were uniformly recognised as ideal. The themes cardiac arrest, death, and level of care contained the items receiving highest ratings. CONCLUSIONS: End points relating to death, cardiac arrest and intensive care unit admission currently comprise the most obvious compromises for investigating early warning scores and similar risk stratification tools. Additional end points from the gross list of suggested end points could become feasible with the increased availability of large data sets with a multitude of recorded parameters.
Authors: Niels Egholm Pedersen; Lars Simon Rasmussen; John Asger Petersen; Thomas Alexander Gerds; Doris Østergaard; Anne Lippert Journal: J Clin Monit Comput Date: 2017-02-25 Impact factor: 2.502
Authors: Patricia Kipnis; Benjamin J Turk; David A Wulf; Juan Carlos LaGuardia; Vincent Liu; Matthew M Churpek; Santiago Romero-Brufau; Gabriel J Escobar Journal: J Biomed Inform Date: 2016-09-20 Impact factor: 6.317
Authors: Maj Juhl Skov; Jacob Dynesen; Marie K Jessen; Janet Yde Liesanth; Julie Mackenhauer; Hans Kirkegaard Journal: Scand J Trauma Resusc Emerg Med Date: 2020-04-10 Impact factor: 2.953
Authors: Idar Johan Brekke; Lars Håland Puntervoll; Peter Bank Pedersen; John Kellett; Mikkel Brabrand Journal: PLoS One Date: 2019-01-15 Impact factor: 3.240