| |
There are
several comprehensive books on statistical analysis of clinical
trials which cover specific methods of analysis and how to interpret
and present the data. Our purpose here is to briefly cover challenging
aspects of analysis, interpretation, and presentation that are
specific to a study evaluating musocitis-related pain.
The main questions
surrounding data analysis are determining who should be analyzed
and how to evaluate the data. Missing data and/or poor quality
data can cause participants to be withdrawn from analysis. Withdrawal
policies should be clearly stated in a protocol before enrolling
participants. Withdrawing participants can complicate data analyses.
If the loss is related to the intervention, a bias can be introduced.
For instance, in this study, if patients on the placebo are generally
doing worse than the study drug, they may skip diary entries,
thereby causing a bias.
Patient cooperation
can make or break a study, especially one involving pain. In this
study, the endpoints surround the patient's daily self-assessment
of pain. If a few entries are missing, it is possible to estimate
the data if there is an observed rate of change seen in the days
prior to and/or after the missing dates. If an observed rate of
change is not seen, then the scores prior to the missing data
should be used. However, this can skew the data and should be
avoided if possible. For instance, if the patient was in severe
pain with a WHO pain ladder of 3 for 5 days, skipped 5 days of
entries and started again with no pain, filling in the missing
5 days would be difficult. The pain was severe without a rate
of change and then a sudden drop was noted. In this scenario,
the missing data can only be filled with the last entry of severe
pain, regardless of whether pain had decreased during those days.
Presenting
the data can be challenging. As with any study, over-analyzing
the data should be avoided. Whether or not the study revealed
positive results, the data should be presented in a straightforward
manner. The goal in presentation is to state what resulted, even
if the treatment--in this case the mouth rinse--did not make a
significant difference. Leave the potential sources of error and
questions for future studies in the discussion section.
|