Centre for Advanced Study

at the Norwegian Academy of Science and Letters

Statistical Analysis of Complex Event History Data

Information

Former 2005/2006 Natural Sciences - Medicine - Mathematics

Abstract

Survival and Event history analysis denote a set of statistical methods used to analyse and describe life times, durations and more complex event histories. It is the traditional tool for analysing incidence and prognosis of cancer and other chronic diseases, but is also of importance in fields like demography, social science, economics, and technical reliability. Although a large body of theory and methods exists, the methodology for analysing complex event histories, with many events and concomitant information (covariates) measured at the entry to the study and during the follow up, is still not well developed. One example would be a clinical study where success of a treatment cannot only be measured by survival, but also by how the course of the disease develops, e.g., with respect to periods of remission and quality of life. The purpose of the project at the Norwegian Centre for Advanced studies is to contribute to the development of statistical methodology for complex event history data. In particular we aim to • study connections with modern methods of causality • study methods that account in a realistic way for unobserved heterogeneity • study methods for proper inclusion of high dimensional gene expression data in the analysis of survival and event history data • develop alternatives to the commonly used Cox regression that are better suited to handle dynamic covariates and time-varying covariate effects • develop useful summary measures of complex event histories

End Report

Event history analysis is a set of statistical concepts, models and methods for studying the occurrences of events over time for a number of subjects. The subjects may be humans, animals, or technical equipment, while the events may be deaths, the onset of disease, relapses of cancer patients, or failures of technical equipment. The aim of an investigation may be to establish risk factors for a disease, to study the effect of a medical treatment or to predict future occurrences of an event. Medicine is the main area of application for event history analysis, which is also the main focus of the CAS project. Other important applications include demography, econometrics, sociology and technical reliability.

Although great advances have been made in event history analysis in recent decades, the field still remains dominated by the classical methods for single-event times, and existing methodologies are not always easily adapted to the more ambitious research questions and richer data structures of contemporary research.

The aim of the project has been to make substantial contributions to the methodology for analyzing complex event histories, as indicated by the above-mentioned and related problems.

 

The CAS year has been an important epoch in a research process that started prior to the project at CAS and will continue after the project year. The group leaders (Aalen, Borgan) and others (Gjessing, Hjort, Samuelsen) have made important contributions to international development in the field. In 2000, Aalen and Borgan took the initiative to establish a working group in event history analysis in the Oslo area. The aim of the working group (called NOREVENT) was to create a common environment for researchers in event history analysis across institutional boundaries for the purpose of (i) strengthening collaboration among statisticians in academia and collaboration between statisticians in academia and biostatisticians at the Norwegian Cancer Registry and the Norwegian Institute of Public Health, (ii) further strengthening international collaboration, (iii) improving the recruitment of young researchers to the field, and (iv) increasing the awareness and knowledge among medical researchers of upto- date methodology for analysing event history data. NOREVENT was soon designated a thematic research area at the Faculty of Medicine, University of Oslo, and its status that was renewed in 2005. At about the same time, NOREVENT changed its name to Centre for Biostatistical Modelling in the Medical Sciences (BMMS).

The project year at CAS made it possible for key BMMS researchers to focus on a number of methodological problems related to the analysis of complex event history data and to benefit from the close contact and collaboration with colleagues from abroad. Having all researchers working away from their home departments, at first-rate facilities and with the support of the professional and friendly staff at CAS provided the best possible conditions for research progress in the form of joint projects and professional discussions.

An application to the Research Council of Norway that will turn BMMS into a Centre of Excellence is through to the second round, and the decision will made in January 2007. If the application goes through, it will create a leading international centre for research in event history analysis and related fields in the Oslo area. Otherwise, BMMS will continue as a thematic research area at the Faculty of Medicine, University of Oslo. In other words, the research from the project year at CAS will be followed up within the framework of BMMS in any case. To this end, the close personal and professional contacts among key international researchers established during the project year will be invaluable, and arrangements have already been made for further research visits to Oslo.

Fellows

  • Andersen, Per Kragh
    Professor University of Copenhagen 2005/2006
  • Böhm, Christine
    - University of Freiburg 2005/2006
  • Bøvelstad, Hege Marie
    Ph. D. Candidate University of Oslo (UiO) 2005/2006
  • Didelez, Vanessa
    Dr. University College London 2005/2006
  • Eide, Geir Egil
    Associate Professor Haukeland University Hospital 2005/2006
  • Farewell, Vernon Todd
    Senior Researcher Medical Research Council Biostatistics Unit 2005/2006
  • Farewell, Daniel Mark
    Research Fellow Cardiff University 2005/2006
  • Fosen, Johan
    Senior Advisor Statistics Norway 2005/2006
  • Frigessi, Arnoldo
    Professor University of Oslo (UiO) 2005/2006
  • Gandy, Axel
    Lecturer Imperial College London 2005/2006
  • Gjessing, Håkon K.
    Professor Norwegian Institute of Public Health 2005/2006
  • Gran, Jon Michael
    Research Fellow University of Oslo (UiO) 2005/2006
  • Gunnes, Nina
    Research Fellow University of Oslo (UiO) 2005/2006
  • Henderson, Robin
    Professor Newcastle University 2005/2006
  • Hjort, Nils Lid
    Professor University of Oslo (UiO) 2005/2006
  • Keiding, Niels
    Professor University of Copenhagen 2005/2006
  • Kvaløy, Jan Terje
    Associate Professor University of Stavanger (UiS) 2005/2006
  • Lindqvist, Bo Henry
    Professor Norwegian University of Science and Technology (NTNU) 2005/2006
  • Martinussen, Torben
    Associate Professor The Royal Veterinary and Agricultural University 2005/2006
  • Nygård, Ståle
    Research Fellow University of Oslo (UiO) 2005/2006
  • Samuelsen, Sven Ove
    Associate Professor University of Oslo (UiO) 2005/2006
  • Scheike, Thomas Harder
    Lecturer University of Copenhagen 2005/2006
  • van Houwelingen, Hans C.
    Professor Leiden University Medical Center 2005/2006

Group leader

  • Odd Olai Aalen

    Title Professor Institution University of Oslo (UiO) Year at CAS 2005/2006
  • Ørnulf Borgan

    Title Professor Institution University of Oslo (UiO) Year at CAS 2005/2006
LOGO