The Use of Statistics in Health Sciences: Situation Analysis and Perspective
The Use of Statistics in Health Sciences: Situation Analysis and Perspective
IN HEALTH SCIENCES:
SITUATION ANALYSIS
AND PERSPECTIVE
Ricardo Ocaña-Riola
The Andalusian School of Public Health
Article in Statistics in Biosciences ·
October 2016
1. Background
Statistics is a branch of mathematics that studies random events.
This formal science is under continuous development, and provides a knowledge
which is essential in factual sciences for decision-making in uncertain settings.
As a result, over the last decades biomedical journals have published a host of
papers that use some novel statistical method or other
The aim of this paper is to present a review on how the statistical methods are
being applied in health research, as well as, to propose some improvement actions.
2. The beginnings
of a scandal
The foundations of
statistical inference and
modern statistics, as we
know it today, were laid
down in the early 20th
century by great
contemporary
mathematicians
3. The eternal problem
Considering that these first warnings came over fifty years ago and that we have
now entered a new century, we might think that there have been major
improvements in the use of statistics and the quality of research in the health
sciences.
The frequency distribution by year of these 229 documents shows that, since 1950,
the annual number of papers reporting statistical errors in health sciences research
has increased progressively (Figure 1). Half papers were published after 2000, i.e.,
during the last 15 years of a period composed by 65 years (1950-2015).
Table 2 summarizes the most frequently statistical errors found in the published
papers. These errors are related to the planning of the analysis, the analysis of the
data and the interpretation of results, and most of them have been recognized as
common problems by the editors of the medical journals with the highest impact factor
in the Journal Citation Report
4. The causes of the problem
•Health researchers lacking appropriate training in
statistics
•Unqualified reviewers to evaluate statistical
methods in health research
•The seductive nature of complex statistical
methods
5. Social repercussions of
the problem
Most health research requires statistical
methods to reach conclusions
The ethical guidelines for statistics
drawn up by the American Statistical
Association give a clear warning: “The
use of statistics in medical diagnoses
and biomedical research may affect
whether individuals live or die […]
Because society depends on sound
statistical practice, all practitioners of
statistics, whatever their training and
occupation, have social obligations to
perform their work in a professional,
competent, and ethical manner
6. Proposal of solutions
Basic training in statistics for health researchers
Professionalising statistics
1. Only statisticians can provide quality training in statistics
2. Only statisticians can ensure quality in statistical data analysis into research teams
3. Only statisticians can assess quality of statistical methods used in scientific articles
Setting up Statistics Units within Health Schools and Health Research Centres
7. Conclusions
Despite efforts made by editors, errors in
applying and interpreting statistical methods
remain common in scientific papers
published in both high impact factor and less
prestigious journals.
Unfortunately, there is no legislation or
regulation for professionalizing of statistics
in most countries, unlike other professions
such as medicine, pharmacy, psychology,
law, engineering or architecture.
In the 21st century, the challenge for science
and governments all over the world is now
achieving professionalization of the
statistical practice and getting statistically
advanced societies to improve decision-
making based on adequate evidences.
THANKYOU!!!!