Cyrille Delpierre

Background: Social factors like gender and socio-economic position are insufficiently considered in medical practice despite their influence on various diseases. Many studies have focused on the development of respiratory diseases and their symptoms across SEP but have attributed these differences by poorer lung function, yet we don’t know whether the expression of dyspnea differs between people across gender and SEP for the same values of lung function, which could reveal social differences in semiology. We aim to assess whether there is a difference in the expression of dyspnea between educational attainment groups with comparable lung function among women and men. Methods: In this cross-sectional study we used data from the national French CONSTANCES population-based cohort at enrolment from 2012 to 2021. Participant with at least one measure of forced vital capacity and dyspnea, with non-missing values for covariates were selected. Main outcome measure was dyspnea assessed by “Do you get short of breath when hurrying or walking up a slight hill”. Social variables were sex and educational attainment. Potential confusion and intermediate factors were age, weight, height, physical activity, smoking, occupational exposures, constraints and physical demands at work, depressive symptoms, self-rated general health, number of medical antecedents, anaemia and a biological health score. Findings: The sample included 130,916 individuals aged 18 to 75 years (53% Women, 47% Men, 34% High, 44% Middle, 22% Low educational attainment). Dyspnea prevalence was 21%. Dyspnea decreased as FVC increased, from 35% in the lowest sextile to 8% in the highest. Women and men with lower educational attainment reported more dyspnea than those with higher educational attainment rate (40% compared with 28% and 20% in women, 21% compared with 12% and 8% in men) and at every FVC level. After adjustments for all covariables these, differences decreased but persisted for all FVCs. Interpretation: These results highlight the importance of considering the patient's social position in clinical practice, as the expression of symptoms differ for the same pulmonary function. Funding This project has received funding from the European Research Council (ERC) (grant agreement No. [856478]).

Evidence on the dynamics of social inequalities throughout the COVID-19 healthcare pathway in France is sparse. This study examined the relationship between area-level social deprivation and hospitalizations, intensive care unit (ICU) admissions, and deaths in metropolitan France during five pandemic waves (wave 2: 1 July-16 December 2020, wave 3: 23 December 2020-16 June 2021, wave 4: 23 June-20 October, wave 5: 27 October-2 March 2022, wave 6: 9 March-31 August 2022). Using comprehensive data from the French national surveillance hospitalization database, we built spatiotemporal Bayesian Poisson regression models to estimate, for each pandemic wave, Incidence Rate Ratios (IRR) and 95% credible interval (CrI) for the association between outcomes and quintiles of the European Deprivation Index, among the French general population and infected persons. We show a growing social gradient for all outcomes among infected persons. For people living in the most deprived areas the IRR increases from 1.43 95%CrI (1.39-1.47) (wave 2) to 1.60(1.55-1.66) (wave 6) for hospitalizations, from 1.57(1.49-1.66) to 1.72(1.58-1.87) for ICU admissions, and from 1.35(1.28-1.42) to 1.70(1.55-1.87) for in-hospital deaths. Among the general population we also observe a social gradient in all outcomes which declines after wave 5. We find a social gradient related to severe forms of COVID-19 over the five pandemic waves studied, and this gradient increases among persons infected with the disease. Our study highlights the importance of considering social position when managing pandemics and underscores the need for targeted interventions for specific populations to minimize health inequalities.

Background: Late HIV diagnosis (LD) remains a major challenge in controlling the HIV epidemic, contributing to higher morbidity, mortality, and ongoing transmission. We aimed to estimate its prevalence and to identify associated socio-territorial inequalities in Greater Paris. Methods: This observational study included PLWH diagnosed during 2014-2021 and treated in all HIV hospital centers across Greater Paris. Seven transmission groups were defined according to birth region, gender, and transmission mode. Neighborhood deprivation was assessed using the French Deprivation Index (FDEP) based on residence place. Factors associated with LD were identified using multilevel modeling. Findings: Among 8,346 PLHW, 58% were born abroad, 43% were MSM, and 16% lived in precarious housing (3% among French-born MSM (FMSM) vs. 26% and 30% among foreign-born heterosexual men (fHM) and women (fW)) Overall, 47% lived in the most deprived neighborhoods, reaching 64% among fHM and 62% among fW. Nearly half had LD (47%), ranging from 30% among FMSM, to 65%/57% among fHM/fW. In multivariate analysis, LD was associated with living in deprived neighborhoods (aOR=1.39, 95%CI [1.17-1.65]). After adjustment for transmission group, all groups showed higher LD risk than FMSM, and previous associations disappeared. Conclusion: Living in deprived areas was associated with higher LD risk, driven by the concentration of high-risk populations (fHM and fW) and reflecting the socio-spatial segregation of PLWH. These findings highlight the need for innovative, geography-based, screening strategies that go beyond traditional risk-group approaches. Moreover, the seven-category “transmission group” classification represents a valuable social indicator for studying health inequalities among PLWH.

Epidemiologists often handle large datasets with numerous variables and are currently seeing a growing wealth of techniques for data analysis, such as machine learning. Critical aspects involve addressing causality, often based on observational data, and dealing with the complex relationships between variables to uncover the overall structure of variable interactions, causal or not. Structure learning (SL) methods aim to automatically or semi-automatically reveal the structure of variables’ relationships. The objective of this study is to delineate some of the potential contributions and limitations of structure learning methods when applied to social epidemiology topics and the search for determinants of healthcare system access. We applied SL techniques to a real-world dataset, namely the 2010 wave of the SIRS cohort, which included a sample of 3006 adults from the Paris region, France. Healthcare utilization, encompassing both direct and indirect access to care, was the primary outcome. Candidate determinants included health status, demographic characteristics, and socio-cultural and economic positions. We present two approaches: a non-automated epidemiological method (an initial expert knowledge network and stepwise logistic regression models) and three SL techniques using various algorithms, with and without knowledge constraints. We compared the results based on the presence, direction, and strength of specific links within the produced network. Although the interdependencies and relative strengths identified by both approaches were similar, the SL algorithms detect fewer associations with the outcome than the non-automated method. Relationships between variables were sometimes incorrectly oriented when using a purely data-driven approach. SL algorithms can be valuable in exploratory stages, helping to generate new hypotheses or mining novel databases. However, results should be validated against prior knowledge and supplemented with additional confirmatory analyses.

Research has shown that our socially structured experiences elicit a biological response, leading to the observation that numerous biomarkers (objective biological measures that are representative of various biological processes) are socially patterned. This ‘social-to-biological’ research is of interest to researchers across multiple disciplines and topics and especially to those with an interest in understanding the biological embodiment of the ‘social environment’. Combining social and biomarker data is also of relevance to those examining the biological determinants of social behaviours (for example, the relationship between genetics and certain behaviours like smoking). However, as much of the research involving biomarkers and social data are multidisciplinary, researchers need to understand why and how to optimally use and combine such data. This article provides a resource for researchers by introducing a range of commonly available biomarkers across studies and countries. Because of the breadth of possible analyses, we do not aim to provide an exhaustive and detailed review of each. Instead, we have structured the glossary to include: an easy-to-understand definition; a description of how it is measured; key considerations when using; and an example of its use in a relevant social-to-biological study. We have limited this glossary to biomarkers that are available in large health and social surveys or population-based cohort studies and focused on biomarkers in adults. We have structured the glossary around the main physiological systems studied in research on social to biological transition and those that go across systems and highlight some basic terms and key theoretical concepts.

The objective was to explore the contributions and limitations of structure learning methods within an epidemiological analysis of real-world data. The specific aim was to use these networks to identify determinants of access to healthcare among various social factors. We analyzed data from the 2010 wave of the SIRS cohort, which included a sample of 3,006 adults from the Paris region, France. Healthcare utilization, encompassing both direct and indirect access, was the primary outcome. Candidate determinants included health status, demographic characteristics, and socio-cultural and economic positions. We employed a dual approach: a non-automated epidemiological method (initial expert-knowledge network and logistic regression models) and structure-learning techniques based on several algorithms, with and without knowledge con-straints. We compared the results based on the presence, direction, and strength of specific links within the produced network. Although the interdependencies and relative strengths identified by approaches were similar, the structure-learning algorithms detected fewer associations with the outcome than the non-automated method. Relationships between variables were sometimes incorrectly oriented when using a purely data-driven approach. Structure learning algorithms can be valuable in exploratory stages, helping to generate new hypotheses or mining novel databases. However, results should be validated against prior knowledge and supplemented with additional confirmatory analyses.

Background: Vulnerable social groups have greater difficulty in accessing care and a lower quality of care. Health systems focused on primary care appear to be more effective, efficient and equitable. However, difficulties in accessing primary care are persisting. We focused on primary care screening for cervical cancer through Cervico-Uterine Smear (CUS), which has been shown to be effective in reducing disease incidence and mortality. In this study, we aimed to investigate the characteristics of women who undergo CUS according to the category of health professionals (general practitioners or gynaecologists) performing CUS and to analyse potential differences in access to care in terms of socioeconomic and geographical characteristics. Methods: This was a retrospective observational study based on data from the main health insurance schemes in France, allowing analysis of health care consumption according to socioeconomic levels and proximity to health care services. We included women aged 25 to 64 years in 2012 for whom CUS would be a relevant procedure (695,694). The sociodemographic and territorial indicators were age, geographical area deprivation, and the availability of gynaecological care. The analysis was performed using multinomial logistic regression. Results: A total of 202,271 (29%) patients underwent CUS; of whom 68% underwent CUS administered by gynaecologists and 28% were administered by general practitioners (GPs). However, inequalities in CUS screening rates were observed, with a decrease in the number of CUSs performed with increased age, a rural location, deprivation, and sparse health care provisions. Deprived people seemed less penalised by GPs. Conclusions: Involvement of General Practitioners may improve cervical screening equality in France. The organisation of health systems around primary care may allow a better access to care and to account for the specific needs of deprived populations.

Purpose Survivors after acute respiratory distress syndrome (ARDS) due to coronavirus disease 2019 (COVID-19) are at high risk of developing respiratory sequelae and functional impairment. The healthcare crisis caused by the pandemic hit socially disadvantaged populations. We aimed to evaluate the influence of socio-economic status on respiratory sequelae after COVID-19 ARDS. Methods We carried out a prospective multicenter study in 30 French intensive care units (ICUs), where ARDS survivors were pre-enrolled if they fulfilled the Berlin ARDS criteria. For patients receiving high flow oxygen therapy, a flow ≥ 50 l/min and an FiO₂ ≥ 50% were required for enrollment. Socio-economic deprivation was defined by an EPICES (Evaluation de la Précarité et des Inégalités de santé dans les Centres d’Examens de Santé - Evaluation of Deprivation and Inequalities in Health Examination Centres) score ≥ 30.17 and patients were included if they performed the 6-month evaluation. The primary outcome was respiratory sequelae 6 months after ICU discharge, defined by at least one of the following criteria: forced vital capacity < 80% of theoretical value, diffusing capacity of the lung for carbon monoxide < 80% of theoretical value, oxygen desaturation during a 6-min walk test and fibrotic-like findings on chest computed tomography. Results Among 401 analyzable patients, 160 (40%) were socio-economically deprived and 241 (60%) non-deprived; 319 (80%) patients had respiratory sequelae 6 months after ICU discharge (81% vs 78%, deprived vs non-deprived, respectively). No significant effect of socio-economic status was identified on lung sequelae (odds ratio (OR), 1.19 [95% confidence interval (CI), 0.72–1.97]), even after adjustment for age, sex, most invasive respiratory support, obesity, most severe P/F ratio (adjusted OR, 1.02 [95% CI 0.57–1.83]). Conclusions In COVID-19 ARDS survivors, socio-economic status had no significant influence on respiratory sequelae 6 months after ICU discharge.