Environment and health: rich and unique data from the Constances cohort

The data on the home environment of the nearly 220,000 volunteers in the Constances cohort (and for a sub-sample of 80,000 of them on that of all their successive homes since birth) combine, thanks to a major collection and geocoding effort, information from questionnaires sent to the volunteers (use of products), future measurements in the homes of some of them, and satellite maps or databases provided by Constances partners over a period of time that may extend to several decades.

This includes:

  • pollutants and atmospheric deposition in the air,
  • moulds and indoor cleaning products,
  • surrounding green areas,
  • nitrates and disinfection by-products in tap water,
  • ionising radiation, ultraviolet radiation and artificial light at night…

These data, with numerous developments in progress or to come, will enable progress to be made in the analysis of the relationships between these environmental factors and the health of individuals, on which Constances already has a wide range of data, enriched by the matching with :

  • SNDS data (detailing health care consumption, hospitalisations, causes of death, sick leave),
  • the socio-professional history of individuals (CNAV data on careers),
  • and a lot of thematic information collected directly from the volunteers in the cohort.

The very wide and varied geographical coverage (cities and countryside in all regions of France) of this database is particularly interesting for this field of research.

The data are available on the CASD. To access the data, a request must be submitted to the cohort leaders and the simplified procedures (MR) must be completed with the Cnil.

Air quality, water quality, green spaces, temperature and climate… Research is increasingly showing links between a multitude of environmental characteristics and human health. But this current state of knowledge remains limited and much remains to be done to better understand how our environment influences our health and to propose appropriate prevention measures.

To this end, the Constances cohort, accessible on the CASD, is an efficient system with nearly 220,000 volunteers, and the regular collection of data on many aspects of health and its determinants, in particular through matching with data from the National Health Data System (SNDS), but also on the environment in which the participants live.

In order to study the relationship between the environment and health, the major difficulty is to assess the exposure to environmental factors for each of the Constances volunteers, and this over the longest possible periods, as most of the effects on health result from exposure extending over years or even decades. To do this, we obviously cannot rely on measurements taken by sensors directly in the environment of each participant, especially as it would be necessary to go back in time! The chosen solution is based on two complementary devices.

Firstly, participants’ residential addresses are collected and geolocated since their entry into the cohort and updated throughout the follow-up of the cohort. In addition, 80,000 participants have agreed to reconstruct the history of all their addresses since birth. An additional advantage of Constances is that the volunteers live in a wide variety of locations, from the city to the countryside and in all regions of France, which makes this database an extremely rich source of information for this type of study, without equivalent.

With these location elements, it is then possible to match the geocodes of the volunteers’ living areas to environmental databases providing spatialized exposure data as well as data on a wide variety of environmental factors. Some of these are already available and others are under development. A range of research on health and the environment has already benefited from the resource provided by the Constances cohort.

The first theme concerns outdoor air quality. Thanks to collaborations with environmental experts, Constances has a set of maps, the result of different and complementary spatial modelling:

  • The ATMO federation has developed a chemistry-dispersion model, and was able to provide Constances with concentration maps for six pollutants (fine particles PM5, and larger particles PM10, nitrogen and sulphur dioxides, ozone, benzene) between 1989 and 20161.
  • The European ELAPSE project was able to provide Constances with “land use” modelling based on the relationships between pollutants and various spatial characteristics (distance to road, altitude, satellite measurements, etc.) for a set of pollutants (PM5, soot carbon, nitrogen dioxide and ozone) between 1990 and 2019.
  • A collaboration with the UMS Patrinat (MNHN, OFB, CNRS), which is developing a biomonitoring programme for metallic atmospheric deposition by mosses in forest environments, has made it possible to estimate the spatial variations, over the whole of rural France2 , of 14 metallic trace elements (including cadmium, mercury, lead, but more originally antimony or zinc) in 1996, 2000, 2006, 2011 and 2016, and to develop maps for Constances, especially for four urban areas3 , for the period between 2018 and 2020.

In terms of indoor air quality, information was collected by questionnaire on moulds and the domestic use of cleaning and disinfecting products, and measurements will be carried out in the homes of 2,000 Constances volunteers, using electrostatic dust collectors to refine knowledge of the chemical quality of indoor air.

Another theme concerns exposure to green spaces, whether natural (e.g. forests) or more anthropogenic (urban parks, agricultural crops). Fine satellite data and European land use inventories, such as the Corine Land Cover, are used to estimate the degree of greenery around the volunteers in the cohort, to define their type (forest, field, meadow, etc.) and to calculate distances to the nearest green spaces of interest. The satellite data allow to go back to 1984 with a monthly scale, and the “Corine” data are available for 1990, 2000, 2006, 2012 and 2018.

Finally, other projects have started involving other types of data: ionising radiation, water quality, light, pesticides (non-professional).

For ionising radiation, residential exposure will be estimated using maps and spatial databases on radon, cosmic and terrestrial radiation. These estimates will be refined by questionnaire data and radon dosimeter measurements in the homes of 1000 participants.

For water quality and in particular nitrate and disinfection by-products, exposure will be estimated by means of the SISE-Eaux database of the Direction Générale de la Santé, currently being formatted for use in research, and supplemented by a dedicated questionnaire in 2020 on water use, and by in situ measurements in Paris and Rennes

The light themes include artificial light at night and ultraviolet radiation. Exposure to artificial light will be estimated using satellite data reworked to cover the needs of health research. Additional measurements using ad hoc sensors will be carried out in the homes of a sample of 200 Constances volunteers. Exposure to ultraviolet radiation will be estimated using dedicated maps derived from satellite data.

Exposure to non-occupational pesticides is currently estimated by means of a questionnaire, and the exploitation of new databases, such as Phytatmo, is being organised.

Data on air pollution and green spaces have already led to the publication of ten papers in leading journals, showing for example for the first time the relationship between air pollution and a range of cognitive functions (paper in Lancet Planetary Health4 ), or to a better understanding of the relationship between soot carbon in fine particles and cancer risk (paper in Environmental Health Perspectives5 ).

1 Bentayeb et al 2014 : https://www.sciencedirect.com/science/article/abs/pii/S135223101400291X http://www.elapseproject.eu/
de Hoogh et al 2018 : https://www.sciencedirect.com/science/article/pii/S0160412018309759?via%3Dihub

2  Lequy et al 2019 : https://www.sciencedirect.com/science/article/pii/S0160412019301321#f0015

3 Lequy et al 2022 : https://www.sciencedirect.com/science/article/pii/S0269749122003116?via%3Dihub

4 https://www.thelancet.com/journals/lanplh/article/PIIS2542-5196(22)00001-8/fulltext
Communiqué de presse :

5 https://ehp.niehs.nih.gov/doi/10.1289/EHP8719
Communiqué de presse :