Panorama over 34 years of night work by profession and branch of activity in France based on census data and a job-exposure matrix by gender | BMC Public Health

This study describes the prevalence and proportion of night work among workers in France for more than 30 years, regardless of their status (employee or self-employed), using job-exposure matrices and census data.

Our study clearly shows an increase in habitual night work in France from the 2000s (4.5% in 1999 to 7.0% in 2015). Conversely, occasional night work was less frequent (12.1% in 1999 to 9.4% in 2015), with overall night work being relatively stable over this period. The most significant change concerns night work for women, who work more and more at night (7.7% in 1982 to 9.9% in 2015). This is explained by the French legislation concerning night work, which until 2001 was different according to gender. Before 2001, women were not allowed to work at night except in specific activities. According to Eurostat, almost 2% of active French women were regular night workers in 1992 compared to almost 5% in 2012 with an increase of 2.4 to 3.8% if we consider the years framing the evolution of the legislation. [26].

The very strong increase in the number of women working at night can also be explained by the growth in female activity over this period (+37%) and in particular in jobs where women usually work at night (+150%). In comparison, the number of workers among men was relatively stable over the period (+0.2%) and regular night work increased moderately than among women (+79% among men). In 2015, regular night workers were mainly found in the service sector (1.4 million men and women) and in the manufacturing and extractive industries (410,000 workers), the same finding applies to workers in occasional night workers (1.9 million and 300,000 workers respectively). Night work was particularly common in public health occupations, e.g. nurses, public administration, e.g. army officers, road transport activities, e.g. drivers, or among blue-collar Agrifoods industries. It should be noted that the decrease in the number of night workers in the manufacturing and mining industries could be explained by the sharp reduction in the workforce in this sector.

The Sumer surveys document the exposure of employees in France to a wide range of occupational risks. These cross-sectional national surveys were conducted in 1994, 2003, 2010 and 2017 by the Department of Research, Studies and Statistics (DARES) and the Ministry of Labor to assess occupational risks among 25,000 to 50,000 French employees from a completed questionnaire. during occupational medicine visits. The 2010 and 2017 surveys show that 14% of employees worked at night between midnight and 5 a.m. even occasionally (20% among men and 8% among women, 3,521,100 employees working at night in 2017) [27,28,29]. Our own estimates for the close years (2007 and 2015) were similar with 16% 95%SI [11–20] night workers (22% [17–28] in men and 9% [6–13] among women, 3,307,100 employees working at night in 2015), despite the difference in the exposure assessment method between the two studies. The occupations and industries with the highest number of night workers were also similar in the two studies.

At European level, Eurostat compiles data on the active population of the Member States [26]. France is comparable in percentage of night workers (usual and occasional) to the Netherlands, Finland and Greece (respectively 14.9%, 15.0 and 15.6% vs 16% in France), but different from Portugal which has the lowest proportion (10%) or Slovakia with the highest rate (23%). Our results are also similar to those of the United States with 9.1% [8.3–10.0] men, and 5.6% [5.0–6.2] of women who usually work at night compared to 9.3 and 4.5% respectively in our study [30]. The proportion of night work in sectors such as health and industry is also comparable at 11.8%.[9.6–14.6] and 10.8%[8.9–13.1] (10.3 and 12.2% respectively in our study). In 2011, in Canada, the proportion of regular night workers was higher than in France (12% vs 7% in 2015) but it includes alternating hours [12]. However, international comparisons of data should be made with caution due to variations in data collection methodology and definition of night work.

The PCS and NAF classifications used both in employment surveys and in population censuses have evolved over time. The matching between the JEM and the population data can therefore be carried out using the versions of the classifications defined by period. Thus, we chose to develop several JEM corresponding to periods with the same versions of classifications, rather than developing a single JEM integrating a single version of classifications and several exposure periods. Only the jobs from the 1982 and 1990 censuses coded in the previous versions of the job classifications had to be cross-referenced to be matched with the JEM, using the tables provided by INSEE. This methodology reduced the matching errors of the JEMs with the census data, but has limitations for the study of temporal trends. Due to changes in the coding rules for certain occupations, we were unable to study the evolution of the prevalence of exposure to night work over the 30-year period in, for example, “Manufacturing of leather and similar products” (Supplementary file 2).

The Employment Survey covers a very large sample of the active population; however, certain jobs (PCSxNAF defined at the finest level) present in the general census may not be represented in the survey and therefore not be evaluated in the JEM PCSxNAF. To limit this problem, a matrix developed on the PCS independently of the NAF (PCS JEM) was retained for non-assessed jobs rather than considering them as not exposed to night work (20% of the overall population was assessed by the PCS JEM in 1999). Finally, few jobs have not been assessed (1016 individuals for the 1999 census out of more than 23 million individuals) but they are little affected by night work (in 2007 and 2015 the unassessed PCS only concern women for field jobs in the construction industry). The change in the frequency of night work raises questions about the increase observed in our results after 2000. For the population as a whole, we observe an increase in the percentage of regular night workers after 2000 (4.5% in 1999 vs 7% in 2015) and conversely a decrease in the percentage of occasional night workers (12.1% vs 9.4%), but this trend is also visible between 1990 and 1999 Moreover, the analysis carried out by professional groups usually working at night before the 2003 law (nurses, military police officers) shows rather the opposite trend, with an increase in usual night work over time. The change in the definition of this frequency modality does not therefore seem to have had an impact on the results after 2003. The surveys after 2012 were not included in this analysis due to a new modification of the question where the Night work exposure was assessed only in the last four working weeks before the interview and with significant changes in frequency modalities depending on percentage of working time (Supplementary File 1).

The JEMs for night work presented in this article were developed from data collected in France from large samples of workers during cross-sectional surveys repeated over several decades. These data provided a solid basis for the development of our job-exposure matrices using an a posteriori method. [31, 32]. The large amount of data retrieved from the census with detailed occupational data allows analysis of exposure to night work at a detailed level. JEMs are simple tools that help assess exposure, especially when information is not available, such as night work. JEM has some limitations such as the use of occupational and industry classifications that may group jobs with different exposures. Therefore, the JEM exposure indices are averaged by job code and take into account the variation in exposure between different jobs or different seasons or different activity characteristics. When exposure to night work is studied as a risk factor for an outcome, it should be considered as a proxy because it does not take into account all the complex combinations leading to circadian disruption [33]. Although this JEM night work is specific to the French work organization, our method is reproducible to obtain JEM specific to each work organization because similar data (censuses and employment surveys) are available in many countries. This study is also easily reproducible on future census data and assesses the exposure of all workers in France regardless of their gender and status (employees and self-employed).

Although only the results on night work are presented in this article, “evening” and “shift work” matrices were developed using the same methodology and are available. It is also planned to develop a matrix combining night work and shift work in order to take into account any type of job rotation. This JEM could be used to estimate the health impact in epidemiological studies (e.g. estimation of fractions attributable to night shift work population for several cancers such as breast and prostate cancer), if additional data is available on exposures to other factors involving circadian disturbances, such as light at night, sleep disturbances, poor diet, lack of physical activity, lack of vitamin D [33,34,35].

Michael A. Bynum