This post relates to a longer article, ‘Recommendations for managing missing data, attrition and response shift in palliative and end of life care research: Part of the MORECare research methods guidance on statistical issues’ that has been selected as ‘Editor’s choice’ in the December issue of Palliative Medicine. Dr Nancy J Preston, Lecturer, International Observatory on End of Life Care, Lancaster University, UK, explains the background to the study
Anyone who has tried to conduct research with palliative care populations will know that we face particular problems. As our patients become unwell they struggle to continue on research studies even when research has been planned to be of a short duration and the burden to patients and their carers is as little as possible.
When patients are too ill to complete research missing data can result. Missing data might refer to missing responses to some questions on a quality of life questionnaire, or a whole questionnaire could be missing. This is not uncommon in palliative care research as the burden of progressive illness increases. Alternatively, patients may die during the study or become too unwell to continue, which results in high rates of ‘attrition’. So how can researchers cope with this when analysing data?
The MORECare project aimed to develop guidelines on how to conduct research with palliative and end of life care patients. Recommendations were based upon systematic reviews and consensus methods with experts. A specific workshop was held to address these issues and experienced palliative care researchers and medical statisticians attended. In the morning, presentations were made about missing data and attrition and in the afternoon attendees made recommendations about how to overcome these problems, which they later ranked using an online survey tool. In addition we also addressed the issue of response shift. We have all heard patients express something like, ‘I thought I knew what pain was until this week’. This change in the perception of an experience also changes how we rate it, for example on a questionnaire.
High levels of missing data and attrition are traditionally seen as indicators of poor study design; however, a recommendation from the workshop was that such high levels should be anticipated as they are indicative of a palliative population. Also, researchers need to report why they think they have missing data and how they managed it. However, the key recommendation was to develop a classification of attrition that would be used as a way of easily reporting why participants dropped out of a study. This could include attrition due to death (ADD), attrition due to illness (ADI) and attrition at random (AAR). This could be applied to research beyond palliative care.
The overarching recommendation, which also applied to response shift, was that more research is needed but hopefully these recommendations give us an opportunity to explore and report these issues.
Download a free copy of this article…
‘Recommendations for managing missing data, attrition and response shift in palliative and end of life care research: Part of the MORECare research methods guidance on statistical issues’ by Nancy J Preston, Peter Fayers, Stephen J Walters, Mark Pilling, Gunn E Grande, Vicky Short, Eleanor Owen-Jones, Catherine J Evans, Hamid Benalia, Irene J Higginson, Chris J Todd, and on behalf of MORECare. Palliat Med December 2013 vol. 27 (10): 899-907. First published on 7 May 2013 doi: 10.1177/0269216313486952.
EAPC members and registered users of the EAPC website can download a free copy of this article and other ‘Editor’s choice’ papers from the EAPC website. (Just follow the instructions in the top right-hand corner of EAPC home page to register or login, and scroll down to download the article). Click here to view other EAPC-originated papers on the EAPC website.