Public Defence: Doris Tove Kristoffersen

Siv.ing. Doris Tove Kristoffersen at Institute of Basic Medical Sciences will be defending the thesis “On 30-day mortality as a quality indicator for hospitals. Evaluation based on a simulation study and Norwegian data.” for the degree of PhD (Philosophiae Doctor).

Trial Lecture – time and place

See Trial Lecture.

Adjudication committee

  • First opponent: Professor Paul Aylin, Imperial College,
  • Second opponent: Professor Anette Hylen Ranhoff, University of Bergen
  • Third member and chair of the evaluation committee: Professor Nanna Lien, University of Oslo

Chair of the Defence

Professor Jarl Åsbjørn Jakobsen, University of Oslo

Principal Supervisor

Researcher Jocelyn Clench-Aas, Norwegian Institute of Public Health

Summary

This project has studied 30-day mortality as a quality indicator for Norwegian hospitals. Hospitals with high mortality are thought to be a result of potentially suboptimal quality of care and patient safety problems, whereas low mortality may identify good practice. Statistical properties of two methods for identifying hospitals with low or high mortality were compared in a simulation study: the ratio of observed to expected number of deaths (OE) and logistic regression (LR). This also included a variance corrected OE (OE-Faris), bias corrected LR, and the use of a trimmed mean. The methods were compared using administrative data from ca. 50 Norwegian hospitals for patients with acute myocardial infarction (AMI), stroke, and hip fracture. All methods performed satisfactorily for the detection of high mortality hospitals. OE and OE-Faris did not satisfactorily detect low mortality hospitals.

For patients admitted to two or more hospitals during their episode of care, the outcome (alive/dead) was assigned to each hospital by fraction of time spent at each hospital (W30D). By using data from all Norwegian hospitals, W30D was compared to mortality for patients treated at only one hospital, or to in-hospital mortality. The results of the three ways of counting were similar for patients with acute myocardial infarction and stroke, but less so for hip fracture.

Survival curves were shown to be successful for all three medical conditions, in identifying areas in the clinical pathway amenable to quality improvements at three high mortality hospitals. Additional information

Additional information

Contact the research support staff.

Published May 9, 2019 2:50 PM - Last modified May 9, 2019 2:50 PM