Article Text

Download PDFPDF
Extended report
A discrete event modelling framework for simulation of long-term outcomes of sequential treatment strategies for ankylosing spondylitis
  1. An Tran-Duy1,
  2. Annelies Boonen2,
  3. Mart A F J van de Laar3,
  4. Angelinus C Franke4,
  5. Johan L Severens5
  1. 1Department of Health Services Research, Maastricht University and Caphri Research Institute, Maastricht, The Netherlands
  2. 2Department of Internal Medicine, Division of Rheumatology, Maastricht University Medical Center and Caphri Research Institute, Maastricht, The Netherlands
  3. 3Department of Rheumatology and Clinical Immunology, Twente University Medical Center, Enschede, The Netherlands
  4. 4Department of Plant Production Systems, Wageningen University, Wageningen, The Netherlands
  5. 5Institute of Health Policy and Management, Erasmus University Rotterdam, Rotterdam, The Netherlands
  1. Correspondence to Dr An Tran-Duy, Department of Health Services Research, Maastricht University and Caphri Research Institute, PO Box 616, 6200 MD Maastricht, The Netherlands; a.tranduy{at}


Objective To develop a modelling framework which can simulate long-term quality of life, societal costs and cost-effectiveness as affected by sequential drug treatment strategies for ankylosing spondylitis (AS).

Methods Discrete event simulation paradigm was selected for model development. Drug efficacy was modelled as changes in disease activity (Bath Ankylosing Spondylitis Disease Activity Index (BASDAI)) and functional status (Bath Ankylosing Spondylitis Functional Index (BASFI)), which were linked to costs and health utility using statistical models fitted based on an observational AS cohort. Published clinical data were used to estimate drug efficacy and time to events. Two strategies were compared: (1) five available non-steroidal anti-inflammatory drugs (strategy 1) and (2) same as strategy 1 plus two tumour necrosis factor α inhibitors (strategy 2). 13 000 patients were followed up individually until death. For probability sensitivity analysis, Monte Carlo simulations were performed with 1000 sets of parameters sampled from the appropriate probability distributions.

Results The models successfully generated valid data on treatments, BASDAI, BASFI, utility, quality-adjusted life years (QALYs) and costs at time points with intervals of 1–3 months during the simulation length of 70 years. Incremental cost per QALY gained in strategy 2 compared with strategy 1 was €35 186. At a willingness-to-pay threshold of €80 000, it was 99.9% certain that strategy 2 was cost-effective.

Conclusions The modelling framework provides great flexibility to implement complex algorithms representing treatment selection, disease progression and changes in costs and utilities over time of patients with AS. Results obtained from the simulation are plausible.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Funding This study was financially supported by the Pfizer BV, The Netherlands.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.