Akademik Veri Yönetim Sistemi
Van Medical Journal, cilt.31, sa.4, ss.245-246, 2024 (Hakemli Dergi)
Biostatistical Errors in Medical Journals: A Critical
Evaluation
Dear Editor;
I would like to emphasize that Biostatistical errors in
studies published in medical journals are an important
problem. These errors affect the accuracy and
reliability of studies and can lead to the spread of
misunderstandings and incorrect practices in the field
of health sciences. The quality and accuracy of
scientific literature is highly dependent on the
accuracy and appropriateness of the statistical
analyzes performed. However, in recent years, it has
been observed that serious statistical errors have been
made in many studies. A significant portion of these
errors in health studies arise from statistical analyzes
made by people outside the field who are not
competent in biostatistics. First, errors related to
sample size are common. In most studies, results are
attempted to be obtained with insufficient sample
sizes without adequate power analysis, which
endangers the reliability and validity of the results.
Insufficient sample sizes can lead to false negatives
(Type II error) or false positives (Type I error) in the
study's results (1,2). Secondly, insufficient attention to
the suitability of statistical methods used in data
analysis is a major problem. In particular, the use of
parametric tests, performed without evaluating
whether the data are suitable for normal distribution,
may damage the reliability of the results. The
deficiencies of researchers outside the field in
choosing and applying appropriate statistical methods
cause the analyzes to be inaccurate. Additionally,
failure to make necessary corrections when making
multiple comparisons may lead to misinterpretation
of the results (1,2,3). Thirdly, the lack of transparency
in reporting data draws attention. Not clearly stating
each step and the methods used in the analysis
process makes the reproducibility of the study
difficult and undermines scientific confidence.
Therefore, researchers need to report in detail how
data were collected, how analyzes were conducted,
and how the results were interpreted (3,4). Fourth,
published studies often make the mistake of
confusing statistical results with clinical significance.
The misconception that statistically significant results
are always clinically significant can lead to
misdirection and unnecessary treatments. Therefore,
it is of great importance to consider clinical
significance as well as statistical significance in studies.
In this context, p-value is a statistical tool frequently
used in medical and health research.
Misunderstanding the use of the P-value can lead to
misdirection that can result in unnecessary treatment
and inaccurate results. Therefore, it is important for
medical and healthcare researchers to have the pvalue interpreted and reported accurately by
biostatisticians (5). Fifth, authors have been known to
manipulate results by sometimes inappropriately using
statistical analyzes to establish clinical significance in
line with the literature. For this reason, reliable results
can be obtained by performing statistical analyzes
within the framework of ethical rules and examining
them by an expert (1,2,3). Finally, international
cooperation and setting standards can also play a big
role in this regard. Common standards for statistical
analyzes should be determined worldwide and these
standards should be disseminated. This will improve
the quality of scientific research and contribute to
more reliable results in the field of health (4,5). The
Biostatistical errors mentioned above negatively affect
the reliability and validity of scientific publications,
slowing down advances in medicine and health. To
prevent these errors, researchers need to receive
better training in biostatistics and journals need to
implement stricter control mechanisms in their
publication policies (e.g., applying the decisions of
statistics editors, publishing after the statistical editorSadi Elasan/ Biostatistical errors in medical journals
Van Med J Volume:31, Issue:4, October/2024
246
checks the revisions, etc.). It is especially important
for researchers who do not have sufficient training in
biostatistics and statistics to collaborate with an
expert when performing such analyses.