Article Text

Download PDFPDF

Louis-René Villermé (1782–1863), a pioneer in social epidemiology: re-analysis of his data on comparative mortality in Paris in the early 19th century
  1. C Julia1,
  2. A-J Valleron1,2,3
  1. 1AP-HP, Hôpital Saint-Antoine, Unité de Santé Publique, Paris, France
  2. 2INSERM, U707, Paris, France
  3. 3UMPC Univ Paris 06, Faculté de Médecine Pierre et Marie Curie, UMR S 707, Paris, France
  1. Correspondence to Chantal Julia, Unité de Santé Publique, Hôpital Saint Antoine, 27, rue de Chaligny, F-75012, Paris, France; julia{at}


Background During the early 19th century, contagionists' and anti-contagionists' explanations of disease causes opposed one another, and the Hippocratic miasma theory still predominated. According to that theory, geographic health disparities could be explained by topographical factors: differences in altitude, population density or proximity to a river. This article summarizes the life of Louis-René Villermé (1782–1863) and his major contributions to social epidemiology that proved the association between poverty and mortality.

Methods In this study, data reported by Villermé to study the mortality-rate variations across the 12 districts (arrondissements) of Paris—that is, 1817–1826 Parisian death rates by district, population density and income indicators—are presented and reanalyzed.

Results Results obtained with today's statistical techniques (correlation analysis) support Villermé's claims of a direct poverty–high death rate link: the three income indicators that he chose were significantly correlated with at-home mortality: taxation index (r=–0.83, p<0.002), average rent (r=–0.83, p<0.002), trade taxation index (r=–0.67, p<0.05), while population density variables were not associated with mortality.

Conclusion Villermé was not only a forerunner of social epidemiology, he was also a scientific pioneer by relying on data, not opinions, to challenge or support medical hypotheses.

  • Biography
  • history of medicine
  • social medicine
  • epidemiology
  • public health Europe
  • social epidemiology

Statistics from

At the dawn of the 19th century, medicine was about to be torn between the rise of quantification in all scientific fields and persistent unproved medical theories dating back several centuries. The miasma theory, derived from Hippocrates' theory of medicine, was still widely prevalent in 1852, at the time when Snow (1813–1858) and Farr (1807–1883) analysed the London cholera epidemics.1 Pierre-Charles Louis (1787–1872), the founding father of evidence-based medicine with his numerical method (1835), attacked the then generalized use of leaches to treat many diseases, including pneumonia.2 Less known is Louis-René Villermé (1782–1863), whose contributions during the early 19th century pioneered the field of social epidemiology.3 4 Herein, we briefly recall his contributions and focus on his analysis of the 1817–1826 death rate disparities in Paris that, because of his perspicacious choices of social indicators, enabled him to demonstrate that miasma variables (using population density as a surrogate marker) were not linked to mortality, while poverty was.

Louis-René Villermé (1782–1863)

Biographical sketch

Louis-René Villermé, the son of a magistrate, was born in 1782, near Paris. Between 1801 and 1804, he studied surgery in Paris, where Guillaume Dupuytren (1777–1835) was his anatomy professor. He then served in the Napoleonic armies for 10 years. He wrote and defended his medical thesis after the wars, practiced for 4 years and, in 1818, abandoned his clinical activities to do research. His activities as Secretary-General of the Société Médicale d'Emulation, starting in 1818, served as a springboard for his career, propelling his election to the Académie de Médecine, in 1823, less than 1 year after its re-opening. There, he was rapporteur for its Statistic Commission, chaired by the mathematician Fourier (1768–1830), the inventor of the Fourier's transform, which is still used daily in signal analysis. From 1829, Villermé's academic success was echoed in his professional activities: he participated in the foundation of the Annales d’Hygiène Publique et de Médecine Légale (1829), entered the Conseil de Salubrité de Paris (1831) and was elected, in 1832, to the newly re-established Académie des Sciences Morales et Politiques. From 1834, he was commissioned by this Academy to conduct fieldwork to evaluate the impact of industrialization on society. His report, published in 1840, led to the first child labour laws in France.5 He died 16 November 1863.

Research subjects addressed by Louis-René Villermé

Villermé made three major contributions to social epidemiology. The first issue he addressed was prisoners' health.6 7 He argued that prisoners' abnormally high death rates could be explained by the type of prison and that the decline of their mortality rate over the years reflected improvement of detention conditions. He based his arguments on the analysis of each prison's mortality charts, which were obtained from penitentiary directors. Studies in his second field of interest, the relationship between poverty and mortality at different geographic levels, from the department to the street, and more specifically in Paris arrondissements, were published between 1822 and 1830,8–14 and are described in detail below. Villermé's third notable contribution concerned working-class sanitary conditions, which was a still quite neglected issue that was emerging in industrial societies. In 1840, Villermé published a report, the Tableaux, for the Academy in which he described the physical and moral conditions of the workers in the silk, wool and cotton industries.5 He described trades, working conditions, lifestyle and developed life-expectancy tables according to age and trade. Villermé's Tableaux was the first publication on working-class sanitary conditions, and was later followed by those of Chadwick (1800–1890) in Great Britain (1842), Griscom (1809–1874) in New York (1845) and Shattuck (1793–1859) in Massachusetts (1850).15–17

Villermé's analysis of the impact of geographical inequalities on mortality in Paris


As of the 16th century, physicians began harbouring opposing opinions on the causes of death: contagionists were convinced that diseases would pass from a sick individual to a healthy one through a living, but invisible, process; anti-contagionists, on the other hand, saw the primary cause of diseases in the environment. During the early 19th century, the anti-contagionist view prevailed.18 Among the environmental influences on health valued by the anti-contagionists, miasmas were the most frequently cited. Miasmas—exhalations derived from decaying bodies that could be detected by their smell—were transported in the air and concentrated at certain geographical points.19 This approach was the mere continuation of the Hippocratic miasma theory.20 Under the supervision of the Royal Medical Society, from the late 18th century, medical topographies were written that linked specific geographical sites and their physical characteristics to observations on diseases.21 22 Winds, rivers, compass exposure and population density were described as factors corrupting air and contributing to death. In 1822, Lachaise (1797–1881) wrote in his medical topography of Paris: “mortality is in direct link to street narrowness, house elevation and piling up of households”.23 It is that topographical theory that Villermé contradicted.22 23 His primary concern was the social environment, and as Ackerknecht18 stated, his sociological approach to disease represented a third division of the epidemiological theories of the early 19th century, after the geographic (including the “miasmatists”) and the biological. Because Villermé was familiar with the miasmatic variables that were usually linked to mortality, he studied them along with poverty before concluding his social thesis: “compass exposure, Seine proximity, winds […], population density, all circumstances doctors unanimously cite as playing a role in our health, have no […] evident action (I am not saying no real action), on mortality; their effect is masked by the role of economic privilege or misery”.14


Data Villermé used were drawn from the Recherches statistiques sur la ville de Paris et le département de la Seine,24 hereafter referred as the Recherches, which were the first rigorous statistics compiled for Paris, based on the 1817 nominative census, from the land registry, from the 1820 fiscal statistics (which were later published in a subsequent volume of the same Recherches) and from the handwritten charts of Villot, the chief statistician of the Seine Département (County).24

He used three categories of variables: (1) at-home mortality rate (column 2, table 1), which he computed from the yearly numbers of deaths from 1817 to 1826 (transforming values given as 1/66, to 151.5/10 000, a more common denominator today). The 1817 death rates were known precisely because a census had been conducted that year, but the rates for the other years were only approximated. (2) Population density (columns 5–7, table 1): Villermé used three variables: the ratio of constructed area to total area of the arrondissement, and the arrondissement area per inhabitant for which he made two calculations, including or excluding private gardens, courtyards and streets. (3) Income (columns 8–11, table 1): Villermé used fiscal statistics to create four variables: average rent per arrondissement, and ratios of untaxed, taxed individual (for annuities) and trade taxed rents (for professional activity) to total amount of rents paid per arrondissement. For clarity purposes, we modified this non-taxation index into a taxation index—that is, (1−(non-taxation index)) (see figure 1 for a spatial representation of this variable). For trade taxed rents, Villermé excluded from his calculations rents taxed <30 French francs, considering “modest trade in a great poverty”.14

Table 1

The 1817 data on mortality, poverty, taxes and population density in Paris, France

Figure 1

Paris in 2009 (external border in bold) and in 1820. The numbers identify the different arrondissements of Paris, which were the statistical units of Villermé's studies. The map of the total taxation index identifies the eastern part of Paris as being poorer.

We added to our reanalysis the percentage of poor (column 4, table 1) that we found in the Recherches,24 but that Villermé chose not to use: this information concerned the people subsidized at home by charitable institutions.

Villermé analysed the “at-home” mortality. However, at that time, roughly 30% of deaths occurred in hospitals. For the purpose of our reanalysis, we therefore estimated the total mortality rates for each arrondissement for the year 1817, using original data from the Recherches.24 For the year 1817 (the census year), the arrondissements of origin of all public hospital patients were known because this information was recorded by the hospital administrators. Unfortunately, hospices only recorded the number of patients, and hospitals and hospices only recorded the number of deaths. Therefore, we hypothesized that hospice patients and deaths were distributed as for public hospitals, and estimated total death rates for the year 1817 (column 3, table 1).

We calculated Spearman's rank (table 2) and Pearson's correlation coefficients between mortality and the other variables. We performed a bootstrap analysis (1000 replications, using R software) to obtain the distribution and confidence intervals of the correlation coefficients.

Table 2

Spearman's rank correlation coefficients (r) between mortality and 1817 poverty, taxes and population density in Paris, France

Results of the reanalysis

In Villermé's time, statistical techniques to test correlations were not available. However, had they been, his conclusions concerning the link between taxation index and at-home death would not have changed, as demonstrated by the strong correlation, shown graphically in figure 2A. Spearman's rank correlation coefficients, obtained for the 1817 data (table 2), between at home and total mortality rate, on the one hand, and total taxation index and average rents, on the other, differed highly significantly from zero. In contrast, population density variables are not significantly correlated to mortality. The same results were obtained with Pearson's correlation coefficients, which were very close to Spearman's rank correlation coefficients. The bootstrap analysis showed that, despite the small number of arrondissements, the distribution of the estimated correlation coefficients between total taxation index and at-home mortality is narrow (figure 2B). The CIs of the other correlation coefficients are given in table 2.

Figure 2

(A) Scattergram for at-home mortality and total taxation index (Spearman's rank correlation coefficient=−0.83; Pearson correlation coefficient=−0.88). (B) Bootstrap analysis of Spearman's rank correlation coefficient.

Villermé did not incorporate the percentage of poor in his analysis (column 4, table 1). We found this variable to also be significantly correlated to at-home and total mortality rates. Therefore, had he included it, his theory would have been further strengthened.


The compilation of statistics expanded in France in the early 19th century. A statistics prize (the Prix Montyon) was created in 1817 by the Académie des Sciences, and a Statistics Commission, chaired by Fourier, was established by the Académie de Médecine in 1824. At that time, emphasis focused on the collection of data, and little attention was accorded to secondary analyses of existing data sets. It is indeed one of the reasons why Villermé was repeatedly denied the Prix Montyon, because he preferred to analyze existing data sets rather than acquire new ones.26 Retrospectively, his originality lies in his approach to challenge the dominant—and wrong—hypothesis. His insight and rigor are particularly apparent when compared to other authors who addressed the subject: Benoiston de Châteauneuf (1776–1856), a hygienist,20 calculated the longevity of rich and poor people in the Annales d'Hygiène Publique et de Médecine Légale in 1830, through a direct comparison of two lists.27 He compiled a list of 1600 names of men and women considered to be rich (members of the royal family, for example), and computed death rates by age of 10-year increments. Then, he presented death rates by age of a group of 2000 poor individuals but he failed to explain how he chose them and how he calculated their mortality rates. Retrospectively, it appears that he lacked the data required for those calculations, as neither his contemporaries nor more modern investigators have been able to understand how he obtained the numbers he reported.22 28 29 Over several years, Penot (1801–1886) compared the death rates by age in the industrial city of Mulhouse to the average death rates by age in France, and concluded that the higher Mulhouse mortality resulted from the higher immorality—including drinking and gambling—of the working classes.30

During his epoch, Villermé's methods and conclusions were criticized: to Bayard (1812–1852), a doctor of forensic medicine and contributor to the Annales d'Hygiène Publique et de Médecine Légale who wrote medical topographies of the Paris arrondissements,31 the arrondissements represented too wide a scale to allow “any positive conclusion”, and only thorough research examining the death rates for a distinct district, street or house, for several years could yield a definite answer to the question.32 However, by 1847, Marc d'Espine (1806–1878), a member of the Société d'Observation Médicale created by Pierre-Charles Louis,33 summarized the general feeling in a synthesis of the works on the subject of differential mortality between rich and poor: Villermé “seems to have solved the question [of the influence of poverty on mortality] in its most general way”.29

More worrying for the 21st century epidemiologist is that his conclusions are subject to the ecological bias fallacy. Today's social epidemiology would certainly apply more sophisticated techniques, such as multilevel or contextual analysis. However, Villermé's papers must be viewed with a clear understanding of his era and his foresightedness: he subjected the available data and current methodology to undertake novel analyses and to forge a new understanding of medical and environmental factors affecting life expectancy.


During the early 19th century, Villermé in social epidemiology, like Pierre-Charles Louis in clinical research, was a precursor for whom the quantitative analysis of systematically collected real data was the only way to defend or refute medical hypotheses. He pioneered the field by showing quantitatively that poverty was a predictor of high mortality and, more generally, by laying the foundation of social epidemiology as a science.

What is already known on this subject?

  • During the early 19th century, geographic health disparities were poorly understood and were generally explained by geographical factors, in the line of the still vivid Hippocratic miasma theory. Social epidemiology was absent.

What does this paper add

  • This paper reminds the life and scientific achievements of Louis-René Villermé (17821863) who can be considered as one of the founding fathers of social epidemiology. The paper confirms, by reanalysing his data with today's statistical methods, one of his major contributions where he claimed that poverty was a major determinant of the observed mortality variations.

Implications for policy and practice

  • Evidence-based assessment of social determinants of health should be taught as part of evidence-based medicine.

  • Historical contents should be included in the curriculum of medical students.


The authors thank Dominique Julia and Stephane Baciocchi for helpful discussions to understand Villermé in his historical context, Janet Jacobson for editorial assistance and Sophie Valtat for the map of Paris.



  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.