Skip to main content

Evaluation of bias and precision in methods of analysis for pragmatic trials with missing outcome data: a simulation study

Randomised controlled trials (RCTs) in arthritis and musculoskeletal conditions generally necessitate long-term follow up of largely self-reported outcomes; thus, such RCTs are prone to missing outcome data, mainly because of participant dropout/non-response. Recent years have seen a rise in the application of methods for dealing with missing outcome data (e.g. mixed models for repeated measures or multiple imputation). However, the implications of the missing data and their handling in pragmatic RCTs (as in arthritis and musculoskeletal conditions) have not received widespread attention to date. In a review of 91 published RCTs in arthritis and musculoskeletal conditions in 2010-11, we found that complete case analysis and single imputation – such as last observation carried forward – are still the most commonly used approaches to analysis of the primary endpoint. None of the RCTs reported a primary analysis or sensitivity analysis based on an assumed ‘missing not at random’ mechanism. The findings indicate a possible belief among researchers that if the dropout rate is low and/or equal between treatment arms, bias is not a concern and advanced methods to handle dropouts are unnecessary. In this study we perform a detailed simulation aimed at understanding the nature and degree of bias in estimates of treatment effect in terms of the level of dropout, the pattern of dropout, the analysis used, and the type of missing data mechanism.

Author information

Authors and Affiliations

Authors

Rights and permissions

Open Access This article is published under license to BioMed Central Ltd. This is an Open Access article is distributed under the terms of the Creative Commons Attribution License ( https://creativecommons.org/licenses/by/2.0 ), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver ( https://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated.

Reprints and Permissions

About this article

Cite this article

Joseph, R., Sim, J., Ogollah, R. et al. Evaluation of bias and precision in methods of analysis for pragmatic trials with missing outcome data: a simulation study. Trials 14, P110 (2013). https://doi.org/10.1186/1745-6215-14-S1-P110

Download citation

  • Published:

  • DOI: https://doi.org/10.1186/1745-6215-14-S1-P110

Keywords

  • Arthritis
  • Primary Endpoint
  • Dropout Rate
  • Primary Analysis
  • Multiple Imputation