Data SourcesData Sources This study utilized data from four French Health Barometer surveys conducted in 2017, 2019, 2020, and 2021. These surveys have been used since 1992 to monitor the health behaviors, attitudes, and perceptions of the French population. Health Indicators The surveys included two health indicators: * Self-Rated Health (SRH): Assessed on a five-point scale from “very good” to “very poor.” * General Activity Limitation Index (GALI): Measured on a three-point scale from “severely limited” to “not limited at all.” Analysis The study focused on the population aged 18 to 75 years in all four surveys. The analyses examined: * Socio-demographic and geographical variables associated with SRH and activity restriction. * Prevalence of “less than good SRH” and “some limitation of activity” by Barometer year and socio-demographic variables. * Relative risk of “less than good SRH” and “some limitation of activity” between 2017 and 2021, adjusted for covariates. * Interactions between Barometer year and age, education level, socio-professional category, income, and geographic region on health outcomes. Statistical Methods Poisson regression models were used to estimate prevalence ratios (PRs) and 95% confidence intervals (CIs) for the health indicators. Interactions were assessed using interaction terms within fully adjusted models. Analyses were weighted to account for the sample size and demographic structure of the French population.
Data sources
This study uses data from four French Health Barometer surveys conducted in 2017, 2019, 2020, and 2021. Since 1992, these surveys have been used as an epidemiological surveillance tool to monitor the main behaviors, attitudes, and perceptions of the French general population regarding their health. Health barometers, regularly conducted since 1992 by Santé Publique France, are among the most widely used data sources to assess the health status of the French population. These barometers are cross-sectional telephone surveys using a random sampling method to ensure representativeness of adults living in the community at the regional level, with the random generation of landline and mobile phone numbers in an overlapping dual-frame approach (15). To be included in the survey, individuals must live in metropolitan France and speak French.
In 2017, the survey included people aged 18 to 75; in 2019, 2020 and 2021, the survey was expanded to include people aged 18 to 85. Residents of institutions, collective housing and hospitals were excluded. As for the 2020 Barometer, the survey field took place during the COVID-19 pandemic and was disrupted by the implementation of the first lockdown measures. Therefore, the survey was terminated prematurely before reaching the desired sample size, although this did not affect its representativeness. The pre-lockdown sample size was nevertheless used to conduct analyses at the national level. Subjects were included from 8 January to 16 March 2020.
In 2020, the sample included 9,178 subjects, with a participation rate of 40.0% (16). In 2017, the survey included 25,319 subjects, with a participation rate of 48.5% (17). In 2019, 10,352 subjects were included, with a participation rate of 50.8% (18). In 2021, 24,514 subjects were included, with a participation rate of 44.3% (19).
In order to be as representative as possible of the French population, the data from the four surveys were weighted by calibration in terms of age, gender, region, size of the city, education level and number of persons per household. For the 2017 Barometer, the population structure was provided by the 2016 Labor Force Survey of the French population (20). Data from the 2019 and 2020 Barometers were calibrated on the structure of the 2018 Labor Force Survey (21). For the 2021 Barometer, the calibration of the sample was based on the 2020 Labor Force Survey (22). The Labor Force Surveys are conducted by INSEE (Institut National de la Statistique et des Etudes Economiques, Paris).
The detailed methodology and questionnaires of the four barometers are available online (16,17,18,19, 23,24,25,26).
Health indicators
In the four surveys, the same formulation was used for the SRH and GALI questions (3) respectively:
-
“How is your health in general? Is it…” with five possible answers: “Very good, good, fair, poor, very poor”;
-
“To what extent have you been limited in the activities that people usually do because of a health problem for at least 6 months?” with three possible answers: “Limited severely, limited but not severely, not limited at all.”
According to WHO recommendations (27) and in line with EUROSTAT usage (28), SRH responses were categorized into two groups: “very good, good” versus “fair, poor, very poor” (i.e. “less than good”) health. GALI responses were also categorized into two groups: “severely limited, limited but not severely” (i.e. “some limitation”) versus “not limited at all” in line with EUROSTAT usage (29).
static analysis
Because the age limit in the 2017 Barometer was 75 years, the analyses in this study focus on the population aged 18 to 75 years in all four surveys; subjects aged 76 to 85 years in the 2019, 2020 and 2021 Barometers were therefore excluded. For the 2020 and 2021 Barometers, 8,473 and 22,625 subjects were included, respectively. In the 2019 Barometer, 9,460 persons were aged 18 to 75 years, although the MEHM questions were only asked to a subgroup (approximately half of the sample, 4,909 persons) that was representative of the population.
For each barometer, the population was described with respect to socio-demographic and geographical variables: gender; age (six classes: 18–24, 25–34, 35–44, 45–54, 55–64, 65–75 years); socio-economic position (SEP) measured by socio-professional category (divided into three categories: farmer/manual worker, employee/middle manager, manager/intellectual profession), education level (< middelbareschooldiploma, middelbareschooldiploma, > high school diploma) and monthly income (four categories: < 1.500€, 1.500€-3.000€, > €3,000, not reported); and geographical region (five main regions: North-East, North-West, South-East, South-West and the Greater Paris region, based on groupings of the 13 administrative regions created in 2018). SRH (in five and two categories) and activity restrictions (in three and two categories) were also described.
The values and absolute changes of the indicators ‘less than good SRH’ and ‘some limitation of activity’ were described for each Barometer year and for each of the socio-demographic and geographical variables considered here.
Poisson regression models were constructed to assess the relative risk of “less than good SRH” and “any limitation of activity” between 2017 and 2021 (dependent variables), while Barometer years were coded as dummy variables and 2017 was used as the reference year to obtain prevalence ratios (PRs) and 95% confidence intervals (CI), adjusted for all other covariates or independent variables: age, sex, socio-professional category, education level, monthly income, and geographic region as described above (30). Effect modifications of age, education level, socio-professional category, monthly income, and geographic region on the relationship between Barometer years and the two health outcomes were examined by examining interaction terms between Barometer years and all modifiers within the fully adjusted model (31). In case of significant interaction terms (P< 0.05), stratified Poisson regression analyses of the risk of “less than good SRH” and “some limitation of activity” according to Barometer year were then performed, adjusting for all covariates except the modifier under consideration. Analyses regarding education level, socio-professional category and income were performed while excluding the category of 18-24 year olds due to the high percentage of students in this age group for whom these variables are not relevant.
All analyses were performed using appropriate weights with the SAS 9.4 software. The weights took into account the selection probability of the individual and were calibrated to take into account the demographic structure of the French population of the year in question in terms of sex by age (10-year intervals), region, degree of urbanization of the place of residence, household size, and education level (15).
