Social deprivation and SARS-CoV-2 testing: a population-based analysis in a highly contrasted southern France region

Abstract

Background: Testing was the cornerstone of the COVID-19 epidemic response in most countries until vaccination became available for the general population. Social inequalities generally affect access to healthcare and health behaviors, and COVID-19 was rapidly shown to impact deprived population more drastically. In support of the regional health agency in Provence-Alpes-Côte d'Azur (PACA) in South-Eastern France, we analyzed the relationship between testing rate and socio-demographic characteristics of the population, to identify gaps in testing coverage and improve targeting of response strategies.

Methods: We conducted an ecological analysis of SARS-CoV-2/COVID-19 testing rate in the PACA region, based on data aggregated at the finest spatial resolution available in France (IRIS) and by periods defined by public health implemented measures and major epidemiological changes. Using general census data, population density, and specific deprivation indices, we used principal component analysis followed by hierarchical clustering to define profiles describing local socio-demographic characteristics. We analyzed the association between these profiles and testing rates in a generalized additive multilevel model, adjusting for access to healthcare, presence of a retirement home, and the age profile of the population.

Results: We identified 6 socio-demographic profiles across the 2,306 analyzed IRIS spatial units: privileged, remote, intermediate, downtown, deprived, and very deprived (ordered by increasing social deprivation index). Profiles also ranged from rural (remote) to high density urban areas (downtown, very deprived). From July 2020 to December 2021, we analyzed SARS-CoV-2/COVID-19 testing rate over 10 periods. Testing rates fluctuated strongly but were highest in privileged and downtown areas, and lowest in very deprived ones. The lowest adjusted testing rate ratios (aTRR) between privileged (reference) and other profiles occurred after implementation of a mandatory healthpass for many leisure activities in July 2021. Periods of contextual testing near Christmas displayed the largest aTRR, especially during the last periods of 2021 after the end of free convenience testing for unvaccinated individuals.

Conclusion: We characterized in-depth local heterogeneity and temporal trends in testing rates and identified areas and circumstances associated with low testing rates, which the regional health agency targeted specifically for the deployment of health mediation activities.

Keywords: COVID-19; France; access to health care; deprivation; geoepidemiology; population-based; testing.

Figure 1

IRIS (spatial units) selection flow chart.

Figure 2

Characteristics of the socio-demographic profiles of the IRIS (spatial units). (A) population density (log scale), (B) European Deprivation Index (EDI), (C) proportion of white collars among active individuals aged ≥15 years, (D) proportion of blue collar among active individuals aged ≥15-years, (E) median income, (F) proportion of population aged ≥65 years, (G) localized potential accessibility (LPA) to healthcare indicator based on the average number of potential medical doctor visits available per inhabitant (defined at municipality level), and (H) number of primary healthcare professionals active in the IRIS. Socio-demographic profiles: (1) “privileged” (red); (2) “remote” (green); (3) “intermediate” (yellow); (4) “downtown” (cyan); (5) “deprived” (light purple); (6) “very deprived” (dark purple). Corresponding histograms are presented in Supplementary Figure S4.

Figure 3

Spatial distribution of the socio-demographic profiles. (A) Provence Alpes Côte d’Azur (PACA) region; (B–E) main cities.

Figure 4

Evolution of daily SARS-CoV-2/COVID-19 (A) testing and (B) incidence rates, by sociodemographic profile in PACA region, from July 2020 to December 2021.

Figure 5

Forest plots of adjusted testing rate ratios (aTRR) across all 10 periods: (A) changing patterns between socio-demographic profiles (reference class, “privileged” profile), (B) decreasing effect of retirement homes over time, and (C) changing patterns according to age profiles (reference class, “family” profile).