Biostatistics is a key component of
biomedical research and is essential for the interpretation of
study results. Statistical techniques inform, among others, about
the experimental design, treatment plan, evaluation of a new versus
a standard treatment, and identify prognostic factors. The
Biostatistics Center provides statistical expertise to researchers
and doctors on diverse topics from all areas of observational and
experimental biomedical cancer research. This involves developing
and implementing statistical approaches to cover a wide range of
topics including the design and analysis of epidemiologic studies
and clinical trials, the identification of prognostic and
predictive biomarkers, sample size calculations, risk prediction,
as well as animal and in vitro experiments. In the recent past, the
Biostatistics Center has been involved in, among others, prediction
of hearing loss after cisplatin-based chemoradiation in patients
with locally advanced head-and-neck cancer, biomarker-based
treatment of breast cancer with adjuvant tamoxifen, comparison of
platforms for array comparative genomic hybridization (aCGH) to
identify breast cancers with a BRCA-like copy number profile, and
dysphagia and trismus after concomitant chemo-Intensity-Modulated
Radiation Therapy (chemo-IMRT) in advanced head and neck cancer.
Moreover, the Biostatistics Center is closely involved with several
international epidemiological studies on cancer incidence and
exposure to chemicals and radiation, and it offers the annual Basic
Medical Statistics Course.
Any researcher or doctor can
contact us for advice on statistical and general research
methodology and data analysis. Support may range from one-time
advice to extensive scientific collaboration. You can contact us by
e-mail (Mr. Zavrakidis: email@example.com) or phone (Mr.
Zavrakidis: 7990) to make an appointment. In general, after a brief
intake, your project will be assigned to one of us based on
availability and subject-specific knowledge. In order to make
collaboration as smooth as possible, we have developed the
following list of guidelines based on our past experience:
- For all but very minor
projects/issues, we prefer collaboration over consultation.
- Please contact us early in the
planning stage of a new study/experiment. We rather avoid bias or
other issues by design than trying to fix it by analysis.
- For all other than small projects,
we recommend a meeting with all participating collaborators
(including the statistician) during the planning stage and possibly
at regular intervals throughout the course of the project.
- We will decide together with you
who will perform the statistical analyses.
- We expect you to provide us with
the final product of collaboration or consultancy (usually a
manuscript or grant proposal), so that we can make sure the
description of the applied methods and the interpretation of the
results of the analyses is appropriate. If the analysis (or a
non-trivial part of it) is performed by one of us, authorship on
the resulting paper is strongly encouraged according to the
guidelines of most scientific journals.
- We ask you to not contact more
than one statistician for the same project. If we are not familiar
with the specific analyses required for your data, we will consult
with other statisticians or bioinformaticians within and outside
NKI to make sure that state-of-the-art advice is being
- If you previously worked with one
of us on a project (e.g., for a project proposal), please contact
that person directly for any further advice on the same or closely
related topics (e.g., study design after funding has been obtained,
or analysis after data have been collected). If you contact
somebody else, please mention previous assistance provided.
- Make sure we have seen and
approved of any document on which one or more of us is mentioned
before it is submitted outside the NKI (manuscripts, project
proposals, abstracts of presentations, etc.). This applies also to
manuscripts being re-submitted to another journal after they have
previously been approved no matter whether any changes have been
- We expect that any editorial
decision on manuscripts we collaborated on and corresponding
reviewer comments will be forwarded to us without delay.
- Do not send any data unless we ask
you to. In this case, please send the data as described here.
Basic Medical Statistics Course
This full week course explains
statistical techniques for the evaluation of biomedical data. We
provide an introduction into design aspects, methods of summarizing
and presenting data, estimation, confidence intervals and
hypothesis testing, including multivariable regression methods for
the assessment of association. For more information click here.
We offer a series of workshops on state-of-the-art analytic
techniques commonly used in biomedical research. For this first
series, we selected the following topics: sample size and power
calculation, multiple testing, interaction analysis, and imputation
of missing data. For more information click here.
Online statistical tools and statistics
Click here for a list of links to
online statistical tools and statistics references.
Statistical web-tool for pre-clinical experiments
We have built a web-tool tailored to mice experiments, providing
fundamental statistical theory, as well as sample size calculations
and guidelines for performing basic statistical tests for the most
common designs. Researchers conducting such experiments are highly
encouraged to visit and use this web-tool. Temporarily, it is
available only internally on Antonet here.
Staff of the Biostatistics Center
John Zavrakidis, Junior
Roberti, PhD student