Incidence of ill-health related job loss and related social and occupational factors. The “unfit for the job” study: a one-year follow-up study of 51,132 workers

Introduction

The assessment of occupational diseases and their impact is a difficult issue. However, it is important for setting priority actions concerning health and safety at work. Most data concerning these occupational diseases indirectly assesses their incidence or impact. The findings of The Health and Occupation Reporting network (THOR) highlighted the gap between the epidemiological assessment by a network of specialists removed from the workplace (i.e., psychiatrists) and a direct assessment by an occupational physician linked to the workplace (Carder et al., 2009).

The difficulties with assessing the prevalence or the incidence of occupational diseases are (i) the comprehensive identification of the cases, (ii) the reliable notification of work causality concerning multifactorial diseases, and (iii) the knowledge about the population in whom these cases are identified. Over recent years, several occupational disease networks have been developed and data have been collected concerning occupational diseases. Several nationwide networks of clinical specialists have been developed in Europe (Stocks et al., 2015), such as THOR in the UK (McDonald et al., 2006), or the National Occupational Disease Surveillance and Prevention Network (RNV3P) in France (Bonneterre et al., 2010; Faisandier et al., 2011). Another way to assess occupational diseases is to match different data sources. Mustard et al. (2015) matched different and independent data sources: (i) emergency department encounter records, (ii) lost-time workers compensation claims, and (iii) representative samples of workers in a national health interview survey (Mustard et al., 2015).

Numerous countries may also have a compensation system for occupational diseases. Data concerning the recognition of occupational diseases are fairly easy to collect by the organizations which pay the compensation. These systems, however, may provide compensation to workers for benign illnesses, such as occupational contact eczema, but compensation may never be claimed for some frequent and fairly serious diseases, such as work-related depression or burnout syndrome. Consequently, this kind of indicator is a poor representation of the main occupational health problems in a country or a region. Likewise, the analyses of the causes of disability pensions or disability retirements are probably not very suitable for assessing the incidence of occupational diseases. As a result, two of the main difficulties are the collection and use of relevant and reliable indicators.

Unfitness (“inaptitude” in French), is the recognition by the occupational physician that a worker’s health status is no longer fit for their current job and that they will require a job change. This does not mean that the worker cannot work anymore, but that they can no longer work in their current position. In France, the assessment of the fitness or the unfitness to perform the job is exclusively carried out by the occupational physician (OP) during a medical examination. All paid workers are systematically followed by the OP, either yearly or every two years. Additional specific medical examinations are carried out when a medical problem occurs. Moreover, when a paid worker is on sick leave for one month or more, fitness is systematically assessed by the OP when the worker returns to work (during a medical examination called “the reinstatement visit”). An “unfit for the job” outcome may be pronounced for several reasons, including but not limited to: (i) a change in an employee’s health status (e.g., returning to work after recovery from a serious illness or injury), (ii) a medical condition that may limit, reduce or prevent the person from effectively performing a new or current job (e.g., musculoskeletal conditions that limit mobility), (iii) a medical condition that is likely to make it unsafe both for the employee, their co-workers or the public (e.g., driving is essential to the job but the employee is subject to unpredictable and sudden unconsciousness). Each OP follows roughly 3,000 workers (Dutheil et al., 2017). There are currently 5,600 OPs in France who report one of three conditions back to the employer: fit, fit subject to work modification, or unfit for the job. Unfit for the job may lead to being assigned to a new suitable job in the same company, or to the termination of the employment contract due to medical issues. According to a recent study performed in one French region by the General Labor Department of the Ministry of Work, termination of the employment contract is the main conclusion of an unfitness notification (more than 97% of cases) (Fernand, 2012).

A previous study explored the analysis of the “unfit for the job” determination among 55,026 workers by collecting data directly from the medical records of a French occupational health service (Dutheil et al., 2016). The main limitation of this study was that only a univariate analysis was performed. Indeed, the data concerning the followed up workers were non-individual statistics and demographic information in aggregated form (e.g., sex ratio, occupations, average age), not a data frame extracted from the information system. Consequently, the socio-demographics could not be merged with the data frame of unfitness. However, to our knowledge, this is the only study analyzing ill-health related job loss.

The current study aimed to estimate the incidence of unfitness for the job in an occupational health service over a period of one year, and to describe their aetiology and the characteristics of the unfit employees.

Methods

Results

Overall, 51,132 employees were followed up by 17 OPs in 2014 and were included in this study. Among them, 398 cases of “unfit for the job” were reported. The overall incidence of “unfit for the job” was 7.8‰.

Demographics and company characteristics of employees both fit and unfit for the job are presented in Table 1. The mean age of followed up workers was 38.7 years (SD = 11.8). The unfit for the job workers were older (mean = 44.4 years; SD = 11.8, p < 0.001).

When all causes were considered, (women ORa 1.52, 95% CI [1.21–1.91]) and workers over 50 years old (ORa 2.63, 95% CI [2.13–3.25]) were more frequently unfit for the job (Table 2).

Two groups of pathologies caused 85.7% of the cases of unfitness: musculoskeletal disorders (MSD) (47.2%) and mental ill-health (38.4%) (Table 2).

Most of the cases of unfitness were caused by a MSD. The average age of these workers was 46.0 years (SD = 11.3). Among the 188 cases of unfitness caused by a MSD, 32.4% were recognized as an occupational injury/disease, whereas 64.0% of these pathologies were estimated to be work related by the OP. The average length of sick leave was significant (13.1 months, SD 12.3). After adjustment, unfitness caused by a MSD was associated with gender (women ORa 1.89, 95% CI [1.35–2.66]) and the odds ratio increased with the age groups, notably that of over 50 year olds (Table 2). With regard to the association with the industry, the activity sector for which the number of cases of unfitness was below 10 (manufacturing, construction, and information & communication) were excluded from the analysis. Moreover, the financial and insurance sectors were aggregated with real estate activities. This aggregated sector was considered as the reference group. All of the other sectors were significantly associated with unfitness, but after adjustment, only one sector (trade, repair, transportation, accommodation and food services) remained statistically significant (ORa 3.65, 95% CI [1.32–10.1]) (Table 2). The group containing the higher grade white collar workers was less associated (ORa 0.12, 95% CI [0.02–0.88]) with unfitness than the reference group (intermediate occupations and lower supervisors). By contrast, the blue collar workers were highly concerned by unfitness resulting from a MSD (ORa 8.13, 95% CI [4.52–14.62]) compared with the same reference group (Table 2).

There were 153 declarations of unfitness caused by mental ill-health. Eight cases (5.2%) were recognized as an occupational injury or an occupational disease by the health insurance. Conversely, the OP estimated that 64.7% of these diseases were work related. Compared to MSD, the average length of sick leave (10.3 months SD 10.5) was lower (p < 0.05) and workers were younger (40.9 years versus 46.0, p < 0.001). Unlike for MSD, age and gender were less associated with mental ill-health related unfitness. Moreover, all occupation groups and activity sectors were concerned (Table 2).

Discussion

Data for this study were directly collected by the OP who determined whether the patient was “unfit for the job” or not. Consequently, the job losses and the medical causes are as reliable as possible. Moreover, they are based on a medical examination, supported by an expert medical opinion, and directly extracted from the medical records. Therefore, the data may be considered as reliable and of high quality.

The study design allowed information concerning the denominator to be completed. One strength of our design is that the population followed by the OP involved in this study is known and the requested information was extracted directly from the information system. Most of the experts in the network have accurate data concerning the cases (i.e., an occupational disease) but cannot collect reliable data on the population observed and, consequently, the indicators (e.g., incidence or associated risk).

Determining whether the patient’s health is insufficient for the related job, known as “unfit for the job” in France, is exclusively performed by the worker’s OP. The number of cases of ill-health related job loss is probably under-evaluated. For example, an employee with anxiety or depression caused by a conflict at work may choose to resign. He could also leave his job with the termination by mutual consent of employment contract. This employee would not be identified as “unfit for the job” although he should be recognized as “unfit for the job.” A termination for a medical reason will maintain the unemployment insurance and termination indemnity. This is in the interest of a worker with a medical problem to be determined unfit for their job rather than to resign from the company. However, an employee do not always choose the most interesting way to leave his job in the specific context of mental ill-health caused by a conflict at work. Particularly severe or life-threatening diseases (e.g., most cancers) are probably under-evaluated in this study design, which is probably not appropriately adapted to reliably assess the impact of such pathologies on jobs.

Ill-health related job losses and socio-demographic data are easy to extract from the information systems using a simple SQL query. Most of the data are comprehensively coded in the databases of the occupational health services. Moreover, such an epidemiological system is very cost effective and is based on the French occupational health system.

One of the major advantages of integrating a denominator (the demographics of all the workers followed up by each OP) is the ability to carry out multivariate analyses. These analyses can provide risk assessments which would allow these risks to be monitored and geographical comparisons to be made in France, or other European countries. This would help to accurately identify ill-health related job losses.

The reader should be aware that this study design does not identify work related diseases, neither can it provide a job loads attributable risk, but rather assesses the impact of diseases, work-related or not, on the capacity to maintain the current job.

In our opinion, this must not be considered as a weakness of this study. The mechanisms of the observed pathologies may be partly, completely or not at all work-related. However, it is a delicate issue to determine if a disease is work-related or not (the work causality) for each individual. Diseases are often multifactorial with personal, professional and extra professional components (e.g., mental ill-health). In fact, all data concerning the issue of work causality should be cautiously interpreted. In our study, the outcome is job loss, which is an objective data, rather than the work causality of cases, which requires a judgement. The analysis of the “unfitness for the job,” even if it does not allow the work related causality to be determined, allows at-risk groups to be identified and preventative actions to be promoted in order to support the continuation of employment.

Conclusions

French occupational health systems have an important and cost effective epidemiological capacity. The supervisory authority of the occupational health services, the General Labor Department of the Ministry of Work, should consider promoting such an epidemiological approach based on occupational health service data, which could provide useful, accurate, and reliable information in the field of occupational health. Moreover, occupational health information systems contain accurate and high quality data on workplace analysis, medical records, or sociodemographic data. Several European countries, such as Belgium, could also collect and analyze similar data (Godderis et al., 2015). For these very reasons, an Occupational Health System Data Analysis network should be promoted.

Supplemental Information