From Epidemium
Jump to: navigation, search

Scientific question


Cancer is a complex disease and numerous factors can impact the health state of diseased and non diseased patients.

The aim of the AROUND project is to describe the multiple dimensions of the environment of cancer and investigate their impact on cancer incidence and evolution (progression, death, …).

Dimensions taken into account will focus on medical environment such as existence of randomized clinical trials, launch of new molecules, development of prescribing guidelines, existence of diagnostic tools and digital applications, number of physicians, health care pathway, funding for research, … (non exhaustive list).

The AROUND project aims at providing a relevant picture of factors impacting health state of populations. The final aim is to provide insight on the most efficient measures to reduce cancer impact on people.


In order to provide early results, a first simple question has been dealt with: what is the impact of clinical trials on cancer diagnosis and mortality? This subproject is nammed AROUND-trial and its aim is to see how clinical research translates into clinical practice.

In a first time, the results to this question will have limited interpretation due to the fact that not all dimensions of medical environment will be taken into account. Nevertheless, this first question will provide a first insight in the overall AROUND project.

Status and timelines

Timelines and objectives

The project has been launched on 7th September 2017.

Presentation on midterm meeting in la Paillasse (7th October 2017) is available:

State of the art: November 2017

Data selection and method: December 2017

Research and analysis: April 2018

Finalization and writing of a draft for a scientific paper: September 2018

Present status

In January 2018, after presentation to the Epidemium final for challenge 2, the project is still ongoing with new developments to define.

Team members

Data analysis: Aurelie Bardet, Elie Toledano, Thomas Berdugo

Literature review and state of the art: Anne Gillard, Nicolas Barbé

Scientific input, brainstorming and environment: Romain Fillaudeau, Valentin Legros, Josiane Versini

Ethics and concept: Antoine Lassauge

Help in results interpretation from other Epidemium participants: Nasséra Chalabi and David Magrez

Data selection and method

Data on clinical trials for AROUND-trial project

Open data for cancer trials have been looked for in and (INCa register). After assessment of the comparative advantages of the databases regarding exhaustiveness, covered time period and easyness for use (from unstructured data), e-cancer data were chosen. A G-app script was used to recover structured data and the number of clinical trials per department (summing trials (all phases) in all centers) was derived.

Data on cancer mortality

For AROUND-trial, open data on cancer have been provided by national registers (INSEE/INSERM Cepi-DC data, provided on Score Santé website). Standardized mortality rates (on Europeen population) for all tumours were used, per department.

Scope of analysis

In order to be based on data considered as of sufficient quality, AROUND-trial has focused on mainland France, from 2002 to 2012.

Research and analysis


All the analyses have been performed using R ( and R studio.

Visual representations of geographical distribution of number of clinical trials and cancer mortality are provided (package maptools), as well as time evolution.

In order to dig the data, a specific focus has been performed on five departments with a high number of clinical trials: Bouches-du-Rhône, Gironde, Nord, Paris, Rhône and on one department with an intermediate number of clinical trials: Bas-Rhin.

Furthermore, all tumours mortality have been studied, as well as mortality associated with 4 different organs: breast, pancreas, lung and colorectal cancer.


Geographical representations show a visual evidence for regional heterogeneity of death rates and number of clinical trials but the differences do not present a similar pattern.

Time evolution on all cancers mortality show raw similar trends in all 6 specifically-focused departments.

Looking at data focused on each cancer type, some signals of a link between an increase in the slope of growth of number of clinical trials and a decrease in the slope of mortality can be found, but not for all cancers and not in all departments.

All codes and databases are stored in the AROUND GitLab [1]


Some limits of the AROUND-trial subproject have been identified:

- in the data choices

- in the data preparation

- in the dataset choices

- in the data analysis

Thus, AROUND-trial provides first insights on the considered problematics and requires further exploration through appropriate analytical methods, and in particular taking into account the expected time lags (specific to each tumour type) between a clinical trial and its impact on disease management.


Presentation on AROUND-trial subproject at final meeting in la Paillasse (20th January 2018) is available:

Poster on AROUND-trial subproject at final meeting in la Paillasse (20th January 2018) is available:

Abstract of AROUND-trial:


Defining the medical expertise of a geographical zone in a given pathology based only on number of clinical trials by centers can be subject to refinement. Similarly, impact of clinical trials on cancer mortality depends on each cancer type evolution. Thus, the conclusions of the AROUND-trial project must be seen as an hypothesis generation and have to be strengthened by sensitivity analyses, taking into account number of centers by department, overall survival time, …

Defining a fine problematic in such a complex pathology as cancer, dealing with lots of open datasets, requires time and clinical  as well as advanced analytical skills. AROUND-trial provides first insights on the impact of medical environment on cancer and in-depth studies with consideration of all dimensions are required to draw robust conclusions.

The whole AROUND project is going to be continued after deadline for Epidemium challenge 2 and aims at providing elements for a publication.