Comparing the occurrence of chronic physical disorders in self-employed individuals with that of employees: A systematic review

3.4.1 Mortality and incidence of physical disorders in longitudinal studies

We identified six longitudinal studies of good quality, namely, four from Sweden, and two from Denmark (Tables 2 and 4).

The largest study with over four million subjects and five years of follow-up was a nationwide Swedish register-based study. In this study, Toivanen et al. showed that the proportion of self-employed sole proprietors hospitalized for stroke or myocardial infarction was the same as that for employees. However, self-employed individuals with own personnel had a significantly lower incidence rate of hospital admissions for myocardial infarction and the same incidence rate for stroke compared with those of employees [21].

The two studies with a 23- and a 12-year follow-up of the Swedish Total Population Registry and the Swedish Cause of Death Registry, respectively, found higher incidence and mortality rates of CVD in self-employed individuals compared with those found in nonmanual workers. When compared with manual workers, self-employed individuals had considerably lower incidence rates of cardiovascular diseases and mortality [32, 33].

Based on a four-year follow-up of national Danish registry data, Tüchsen et al. (1992) found a considerable excess risk of being admitted to hospital due to ischemic heart disease for self-employed female hotel and restaurant workers aged 20–59 years. The corresponding risks for employed women and self-employed men, as well as for employed hotel and restaurant workers, were also increased but to a much smaller extent (not significant) than those for self-employed female hotel and restaurant workers (significant). Furthermore, self-employed individual males in the textile industry and self-employed male hairdressers, as well as self-employed women with agricultural occupations and self-employed women from sectors that include hardware, radios, and bicycles, had increased rates of ischemic heart disease risk. For these occupational groups, the authors did not publish comparable data for employees. Among bakers, their slight increased risk of ischemic heart disease was not considerably different between self-employed and employed individuals. Other occupational groups, e.g., drivers and fishermen, also showed a notably increased risk of ischemic heart disease among both self-employed and employed workers. The authors reported only the results that separately regarded significant differences between self-employed and employed individuals, but further occupational groups were not examined. Therefore, only the selected significant results are presented in Table 4 [31].

A nationwide Swedish register and census-based study from 1989 reported some inconsistent results for various groups of workers and diagnoses. In summary, the majority of the outcomes showed higher standardized mortality ratios for CVD parameters in self-employed individuals versus nonmanual workers, whereas in comparison with manual workers, the majority of the results showed no differences between the groups [29].

A more recent Danish register study investigated incidences of hospital admissions for arthrosis of the hip (coxarthrosis). Over a 5-year follow-up, self-employed individuals had a considerably (not significant) higher standardized hospitalization ratio compared with that of employees who were in leading positions or those who were not in leading positions [30] (Tables 2 and 4).

Most of the cross-sectional studies were from Europe (7 studies) [20, 34–38], and only three were from Asia [39–41]. The majority of these studies (8 studies) showed significantly higher prevalence of physical illness, especially musculoskeletal diseases, among self-employed individuals compared with that among employees [20, 41] by investigating different occupation types and the use of varying assessment tools (Tables 3 and 5).

3.4.2.1. Nonmusculoskeletal disease. Only one cross-sectional study used the current International Statistical Classification of Diseases and Related Health Problems (ICD). Stephan et al. investigated a variety of aggregated physical disorders and showed lower risks of hypertension and any (i.e., at least one) of the following somatic diseases for self-employed individuals compared with those for employees: hypertension, gastrointestinal ulcers, diabetes, rheumatoid arthritis, lower back pain, or shoulder/neck pain. The risks were similar for diabetes, back pain, and upper and lower extremity pain. Higher risks for gastrointestinal ulcers were found in self-employed individuals than in occupation type-matched employed persons [20] (Tables 3 and 5).

Kim et al. reported no differences in the prevalence of headache and eye strain between self-employed individuals with personnel and employees [41], whereas Lee et al. found significantly more eye strain among all self-employed individuals combined than in employees [40] (Tables 3 and 5).

Two studies from France investigated respiratory outcomes: one measured lung function using spirometry, and the other assessed self-reported chronic bronchitis symptoms [36, 38]. Ferré et al. compared self-employed individuals with nonmanual workers and reported a higher prevalence of chronic bronchitis symptoms among self-employed workers, whereas the bronchitis prevalence of the latter showed no difference in comparison with that of manual workers [36]. Lesage et al. reported reduced lung function among self-employed joiners compared with that among their employed counterparts [38] (Tables 3 and 5).

Two studies applied questionnaires to assess fatigue [34, 41]. Benavides et al. used a nonvalidated questionnaire with a permanent employed reference group and showed a significantly higher prevalence of fatigue among self-employed individuals [34]. Kim et al. found no difference between self-employed individuals with their own personnel and employees of various occupations using a validated questionnaire [41] (Tables 3 and 5).

3.4.2.2. Musculoskeletal disease. The majority (5 studies) of the eight studies that investigated musculoskeletal disorders reported higher prevalences among self-employed individuals compared with those among employees. The same applies to male and female workers [9, 39], except in the study by Park et al., who differentiated manual workers, service and sales workers, and nonmanual workers with cognitive demands. The authors found fewer musculoskeletal diseases among self-employed women than among employed female service and sales workers [39]. Two out of three studies that used occupation-matched references described similar prevalences of back pain among self-employed individuals compared with those among employees [20, 38] (Tables 3 and 5).

Almost all of the studies that investigated mixed occupation groups reported a significantly higher risk of musculoskeletal disease among self-employed individuals compared with that among employees [34, 41]. In contrast, studies comparing self-employed and employed people in one specific occupational group showed inconsistent findings [20, 38]. Holmberg et al., who excluded farmers from two other groups reported even less lower back pain among self-employed nonfarmers than among nonfarming employees [37] (Tables 3 and 5).

4.2.1 Longitudinal studies

The results of the six longitudinal studies included in this review were partly inconsistent. Although all the studies were based on register data evaluating ICD diagnoses, there were some considerable differences in the study designs that may explain the varying outcomes.

The register study of Tiikkaja et al. in Sweden showed higher mortality rates for stroke and myocardial infarction among self-employed individuals compared with those among nonmanual workers and lower mortality rates compared with those among manual workers [32]. These results are in line with Rosén et al. who reported a 50–60% higher risk of death from myocardial infarction for manual workers than that for nonmanual workers in higher socioeconomic positions [42]. However, Vagerö et al. reported predominantly higher mortality rates regarding cardiovascular diseases in self-employed individuals compared with those for nonmanual workers, but hardly any differences in the mortality rates compared with those for manual workers [29]. Reasons for the different outcomes might be changes in working conditions, as well as lifestyle habits, e.g., smoking, over the long period between the studies (1961–68 and 1990–2002). Vagerö et al. mentioned high smoking rates in self-employed nonmanual workers at the time of their study [29], whereas manual workers were at higher risk for smoking at a later time [43, 44]. In contrast to nonmanual workers, who drastically decreased their overall smoking rate from 1978 to 1994, manual workers' smoking rate diminished in the beginning and remained stable after a few years [44]. These time-dependent behaviors may have influenced the results of the two studies that investigated cardiovascular mortality [45, 46]. Additionally, there might be a lack of comparability with respect to the data that were collected such a long time apart because of the changes made to work environments since then. Sweden enjoyed a booming economy in the second half of the 20th century [47], which slowed down in 1970 and ended in a crisis in the 1990 s [48]. These economic circumstances might have influenced the working conditions for self-employed individuals as well as for employees, and, consequently, may have affected the health of workers. Neither of the studies that investigated mortality examined nonmanual and manual self-employed workers separately; rather they compared all self-employed individuals combined with nonmanual and manual employees [29, 32]. The distribution of nonmanual and manual workers within self-employed individuals may cause disparities between the studies because a twofold risk of CVD mortality among manual workers has been previously reported [42].

Two recent nationwide register studies from Sweden investigated the incidences of stroke and myocardial infarction [21, 33]. The 5-year follow-up study of Toivanen et al. reported no differences in the stroke incidence rates between the groups in the fully adjusted model (adjusted for sex, age, country of origin, previous health status, education, family, enterprise size). One self-employed subgroup, namely, the limited liability company owners, showed even lower incidence rates than those of employees. Therefore, various other factors may influence the physical health status of workers. Toivanen et al. also reported large differences between industrial sectors regardless of occupational group. They stated that “the results highlight the importance of the industrial sector for hospitalization rates due to acute CVD among sole proprietors and limited liability company owners” [21]. A limitation of the study by Toivanen was the lack of data for the occupational group and industrial sector together. This may explain the similar outcomes found for sole proprietors and self-employed workers.

Similar to Tikkaja et al., the 23-year follow-up study of Malki et al. indicated more protective health effects regarding stroke and myocardial infarction for self-employed individuals when compared with those for manual workers, whereas self-employed individuals showed higher incidence rates in comparison with those of nonmanual workers. Additionally, the authors reported similar outcomes for low-level nonmanual, self-employed and high-level manual workers, whereas high-level nonmanual and low-level manual workers showed clear differences in the manner of lower and higher incidence rates than those of self-employed individuals [33]. Although the included register studies did not integrate questionnaires to explicitly assess lifestyle factors, it is well known that risk factors such as smoking, physical inactivity and socioeconomic inequalities are associated with cardiovascular morbidity [49]. Therefore, lower socioeconomic positions indicate worse health outcomes compared with those of higher statuses [21, 42]. According to the Eurofound and International Labour Organization (2019), the tasks of upper-level nonmanual workers, including higher responsibility or having to solve more unforeseen problems, may be responsible for their better health compared with the tasks of manual workers [50]. Manual workers often have a lower decision latitude and heavier physical work. As Karasek‘s demand-decision-model described, these reasons could lead to problems of strain and feelings of dissatisfaction and consequently result in poor general health [51]. Self-employed individuals enjoy entrepreneurial scope but may perceive more financial pressure and less long-term security [13].

A third national register study from Sweden examined occupations that seemed to be at high risk for ischemic heart disease based on the outcomes of previous studies [31]. The authors argued that some occupational groups have an increased risk of ischemic heart disease, whereas the differences between self-employed and employees in the same industrial sector were mostly not significant. These results might be due to misclassifications of occupations because the study assessed the most important occupation during only one year. Unknown migration from one occupational group to another might have put the results into perspective. Self-employed and employed bakers showed minor differences, whereas female self-employed hotel and restaurant workers showed a higher increased risk of ischemic heart diseases compared with that of employees and males self-employed in hotels and restaurants [31]. A reason for these contradictory outcomes may be the lack of adjustment for potential risk factors such as work-family conflicts or alcohol consumption and smoking. Self-employed workers in food service industries seem to exhibit greater levels of alcohol consumption compared with the employees in this sector [17]. Work-family conflicts, especially for women, are related to physical health problems [52]. In hotel and restaurant working conditions, there could be more work-family conflicts because of shift work, including late work hours. In contrast, bakers can join in family activities in the afternoon. Their different working conditions might explain the different results between bakers and hotel and restaurant workers.

The only longitudinal study assessing coxarthrosis investigated the hospital diagnosis of all economically active men in Denmark. Since only 15% of those with severe arthritis undergo hip replacement, there might have been a high proportion of missing data. The study showed higher hospitalization rates among self-employed individuals versus employees, without an assessment of industrial sector or adjusting for social class. A dramatic fall in coxarthrosis incidence in self-employed individuals over time might be explained by the decrease in the number of self-employed farm workers that accounted for 70% of the cases in this group in the earliest period of the study. Tüchsen et al. also analyzed the differences between industrial sectors, but did so without a breakdown of the occupational groups; the authors concluded that “occupations with low risk can be characterised as nonmanual jobs performed mostly while sitting” [30]. Consequently, the Danish results might have been similar to the results of the Swedish studies by Malki et al. and Tiikkaja et al., if Tüchsen et al. would have subdivided nonmanual and manual workers instead of specific occupational groups [31–33].

The included longitudinal studies only considered ICD codes for their analysis. Further prospective research should aim for longitudinal designs to combine diagnoses from routine data sources or hospitalization rates with questionnaires because the latter may capture conditions at an early stage. These questionnaires should also include information on possible confounders, such as the industrial sector, the extent of partner or family support, dependent children and head-of-household status [53].

4.2.2 Cross-sectional studies

The ten cross-sectional studies of good quality that we included in our systematic evaluation showed a rather heterogeneous picture in terms of the study characteristics and outcomes examined. Studies that found associations of self-employment with negative health outcomes predominantly investigated musculoskeletal, respiratory or malignant diseases [34–36, 41]. The following reasons may explain these findings. First, the most dangerous industries that include hard physical working conditions (e.g. forestry) often hire self-employed workers [39]. Second, some personal factors, such as long working hours or an increased age, apply to many self-employed individuals. Third, in Europe, self-employed individuals are not subject to occupational health legislations [39]. Furthermore, the majority of the included studies used no validated questionnaires. Cross-sectional study designs do not allow us to conclude whether the observed health problems occurred because of or independent of employment status. The employed reference groups of these studies were younger, had a higher educational status and income, and/or were more often nonmanual workers.

According to some studies, low-qualified or low-income workers are at higher risk of injuries and health-related problems [21, 54]. For example, Kim et al. investigated a self-employed group comprising 60% caregivers and an employed group consisting of 50% nonmanual workers [41]. Consequently, the predominant negative effect of self-employment on musculoskeletal conditions could be due to a sample bias. The study by Holmberg et al. was the only one that found a lower prevalence for lower back pain in self-employed individuals compared with that in employees [37]. A reason for these contrary outcomes could be that this study excluded farmers from the group of self-employed individuals and rather examined them separately, whereas the group of employees included all occupations. Previous studies have shown that farmers are in poorer health than other occupational groups [55–57]. Holmberg et al. also found a higher prevalence of lower back pain among self-employed farmers than among employees. Nevertheless, the sick leave rate of farmers due to lower back pain was lower than that of employees [37], which indicated high presenteeism rates for self-employed individuals.

Regarding hypertension, diabetes, headache and eyestrain, the results showed slightly better or similar health outcomes among self-employed individuals than among employees [20, 55]. Different reasons might explain these contradictory results. The study by Stephan et al. was the only one that used ICD codes for assessment [20]. Since only a doctor determines ICD codes, a doctor visit provides the basis for a diagnosis. Previous studies have found that health problems and work hours negatively affect health care utilization. The loss of earning when being absent from work may discourage the solo self-employed from visiting a doctor [58]. Lee et al., who conducted the only study that examined dry eye symptoms, suggested that self-employed individuals may regulate their workplace environment [40]. This flexible regulation may help to relax the eyes, regularly.