Home > Research > Publications & Outputs > MORECare research methods guidance development


Text available via DOI:

View graph of relations

MORECare research methods guidance development: recommendations for statistical methods in palliative and end of life care research

Research output: Contribution to Journal/MagazineMeeting abstractpeer-review

  • Nancy Preston
  • Gunn Grande
  • P. Fayers
  • M. Pilling
  • I. J. Higginson
  • V. Short
  • E. Anscombe
  • C. Evans
  • H. Benalia
Article numberOA42
<mark>Journal publication date</mark>06/2012
<mark>Journal</mark>Palliative Medicine
Issue number4
Number of pages2
Pages (from-to)416-417
Publication StatusPublished
<mark>Original language</mark>English


Aims: To identify agreed best practice for statistical methods
in palliative and end of life (P&EoLC) research.

Background: Carrying out statistical analysis in P&EoLC
research can be problematic due to high levels of missing
data and attrition as patients’ disease progresses.

Methods: We used the MORECare Transparent Expert
Consultation approach to conduct consultation workshops
with experts in statistical methods in P&EoLC research.
Prior to workshops participants were sent overviews of pertinent
issues in statistical methods in P&EoLC. Following
workshop presentations and discussion, nominal group
techniques were used to produce candidate recommendations.
These were subsequently rated online by participating
experts. Descriptive statistics were employed to permit
analysis of consensus and rated importance. Narrative comments
were collated.

Results: The statistical methods expert workshop comprised
20 participants making 19 recommendations. There
was broad agreement across most recommendations, the
top 5 recommendations were:
1. Types and amount of missing data should be reported
along with imputation methods.
2. Pattern of missing data should be investigated to inform
imputation method.
3. A statistical analysis plan should be in place.
4. A taxonomy should be devised to define types of
5. Use of transition questions and minimally important difference
approach provides insight into response shift.

Conclusions: When designing studies in P&EoLC it is recommended
that high rates of attrition should not be seen as
indicative of poor design and that a clear statistical analysis
plan is in place to account for missing data and attrition. More
research is required in statistical methods to assess these
areas but also develop the emerging area of response shift.