• Niels Egholm Pedersen, Lars Simon Rasmussen, John Asger Petersen, Thomas Alexander Gerds, Doris Østergaard, and Anne Lippert.
• Herlev Hospital, Copenhagen Academy for Medical Education and Simulation, Capital Region of Denmark, University of Copenhagen, 25th floor, Herlev Ringvej 75, 2730, Herlev, Denmark. niels.egholm.pedersen@regionh.dk.
• J Clin Monit Comput. 2018 Feb 1; 32 (1): 109-116.

AbstractThe national early warning score (NEWS) is recommended to detect deterioration in hospitalised patients. In 2013, a NEWS-based system was introduced in a hospital service with over 250,000 annual admissions, generating large amounts of NEWS data. The quality of such data has not been described. We critically assessed NEWS data recorded over 12 months. This observational study included NEWS records from adult inpatients hospitalized in the Capital Region of Denmark during 2014. Physiological variables and the use of supplementary oxygen (NEWS variables) were recorded. We identified implausible records and assessed the distributions of NEWS variable values. Of 2,835,331 NEWS records, 271,103 (10%) were incomplete with one or more variable missing and 0.2% of records containing implausible values. Digit preferences were identified for respiratory rate, supplementation oxygen flow, pulse rate, and systolic blood pressure. There was an accumulation of pulse rate records below 91 beats per minute. Among complete NEWS records, 64% had NEWS ≥ 1; 29% had NEWS ≥ 3; and 8% had NEWS ≥ 6. In a large set of NEWS data, 10% of the records were incomplete. In a system where data were manually entered into an electronic medical record, digit preferences and the accumulation of pulse rate records below 91 beats per minute, which is the limit for NEWS point generation, showed that staff practice influenced the recorded values. This indicates a potential limitation of transferability of research results obtained in such systems to fully automated systems.

41 keywords...

hide…