Article Text
Abstract
Objective: In Spain, sick pay benefits for work-related sick leave episodes are higher than for non-work-related episodes. Our aim is to assess whether time to return to work is longer for higher paid sick leave episodes than for lower paid episodes.
Methods: We used data from 62 376 work-related and 76 932 non-work-related sick leave episodes occurring among 338 226 workers from 56 099 companies in Spain in 2002. All episodes were followed for up to 18 months. Episodes were classified by a physician as being work- or non-work-related according to medico-legal judgments. The median episode duration and the 25th and 75th percentiles were calculated. The probability of remaining absent from work was estimated by a non-parametric estimator of the marginal survival function. The time ratio between both types of sick leave was estimated by a log-logistic regression model, using non-work-related episodes as the reference.
Results: Median episode duration (25th–75th percentiles) was 11 (6–21) days for work-related episodes and 9 (4–29) days for non-work-related episodes. Time to return to work was longer for work-related episodes than for non-work-related episodes of less than 16 days (time ratio: 1.19 in men and 1.08 in women), while the opposite was observed for episodes of more than 15 days (0.58 in men and 0.40 in women).
Conclusions: Sick pay benefits have a limited effect on time to return to work after a sick leave episode.
Statistics from Altmetric.com
Studies on time to return to work after a sick leave episode have mainly focused on the nature of health problems1–3 or on exposure to poor working conditions,4–7 while the work or non-work origin of the health problem resulting in the sick leave has not been addressed. Earlier studies have suggested that higher sick pay benefits could be associated with more cumulative compensated work absence days.8 9
Spanish social security law defines work-related sickness absences as those caused by occupational injuries and diseases. An occupational disease is legally defined as any disease listed in the official adaptation of the European schedule of occupational diseases (2003/670/CE), which includes a list of diseases and occupational activities carrying a risk of such diseases (eg, carpal tunnel syndrome in workers exposed to repetitive motion or back pain due to overexertion). Thus, by definition any sick leave episode due to an injury or disease that has not been legally considered to be work related is a non-work-related sickness absence. For instance, mental health disorders such as depression are not recognised as occupational diseases but as non-work-related disorders.
In Spain, workers who are unable to work as a consequence of a health problem have a maximum of 18 months of temporary sick leave to recover while receiving medical care, independent of the origin of the health problem resulting in the sick leave. Sick pay benefits (as a temporary disability pension) and medical care, however, do differ according to whether the sick leave is due to work- or non-work-related health problems. When the sick leave is due to work-related health problems, certification and medical care for occupational injuries and diseases is provided by insurance companies (called mutuas) collaborating with the social security system, which are also responsible for the follow-up of sick workers. During the sick leave episode, the worker receives a sick pay benefit paid by the mutuas equivalent to approximately 75% of the worker’s monthly base salary from the first day of sick leave. When the sick leave is due to non-work-related health problems, certification and medical care is provided by general practitioners from the national health service. During the episode, the sick pay benefit is equivalent to approximately 60% of the monthly base salary from the 4th to the 20th day of the sick leave and increases to 75% from the 21st day onwards. It is paid by the employer from the 4th to the 15th day of sick leave, and by the social security system from the 16th day onwards. The first 3 days are unremunerated. Since 1995, at the employer’s request, mutuas are allowed to manage the worker’s sick pay benefit and to offer some care complementary to the national health service medical care (eg, diagnostic tests, surgery treatments or medical monitoring) for non-work-related sick leave episodes from the 16th day onwards. However, mutuas are not allowed to perform either the initial or the final medical certification.10
Thus, the Spanish social security system provides us with an opportunity to assess the influence of benefit protection schemes on time to return to work after a sick leave episode. We hypothesised that work-related sick leave episodes would have longer duration due to higher sick pay benefits than non-work-related episodes. We carried out a retrospective cohort study to examine differences in time to return to work between incident episodes of work- and non-work-related sick leaves.
METHODS
In Spain, work- and non-work-related sick leave episodes are defined according to a medico-legal judgment made by a physician. Work-related sick leaves are those due to work-related health problems, injuries occurring at workplaces or during commuting, and officially listed diseases.11 Non-work-related sick leaves are those due to any other injuries and diseases that have not legally been considered as being caused by working conditions.
The study base included 338 226 workers from 56 099 companies associated with one mutua (the fifth-largest mutua in Spain), which managed both the work- and non-work-related sick leave episodes of those workers. Sick leave policies and procedures for the notification of sick leave are similar in all the companies covered by the mutua. Data for all sick leave episodes starting in 2002 were retrieved from the administrative file of the mutua. Maternity leaves were not included in the analyses as they are not considered sickness spells in Spain. A total of 62 376 work-related episodes and 76 932 non-work-related episodes were registered by the mutua; 21 726 (15.6% of the total) sick leave episodes occurred in workers who had had a previous episode in 2002.
All episodes were followed by the mutua until the worker’s return to work or for a maximum of 18 months. Each episode was anonymised. The following variables were available for each worker: gender, age (<30, 30–40 and >40 years old), economic activity of the company (11 categories according to the European Classification of Economic Activities) and region of the workplace (17 regions). Analyses by specific medical diagnosis were not attempted since this information was only available for 25.2% of the episodes.
Because duration (in days) was not normally distributed (Kolmogorov–Smirnov = 120.49; p<0.001), the median and the 25th and 75th percentiles (25th–75th percentile) were estimated. Data structure included recurrent events that could not be considered independent of each other and therefore could not be treated with standard survival methods.12 Thus, the probability of staying absent from work for work- and non-work-related episodes was estimated using the estimator proposed by Wang and Chang.13 This function is a non-parametric estimator of the marginal survival function and could be considered as analogous to the Kaplan–Meier estimator for the case of having correlated events. This was calculated based on complete sample and using R software (survrec function).14 Data are presented for men and women separately.
The duration (in days) of work- and non-work-related sick leave episodes was compared using a log-logistic regression model with a frailty gamma distribution term.15 The frailty term was added given that recurrent events were present in the data, and it was assumed that they followed a gamma distribution. This model allowed the estimation of time ratios (TR) and their 95% confidence intervals (95% CI), using non-work-related sick leave as the reference category, and adjusting for age, economic activity and region of the workplace. The time ratio arises naturally from this model (instead of the hazard ratio used in more common survival methods). This parametric regression model was used because hazards were not proportional,16 and presented an acceptable Akaikes’s criterion AIC value.17 We also estimated the time ratio for short-term (less than 16 days) and long-term (more than 15 days) sick leave episodes, because the first 15 days of non-work-related sick leave benefits are paid by the employer and by the social security system from the 16th day onward. This analysis was carried out for men and women separately. Only episodes with complete data for all variables were included in this second analysis, that is, 62 011 (99.4%) work-related episodes and 48 932 (63.6%) non-work-related episodes. Episodes with incomplete data were shorter than episodes with complete data (median of 7 vs 11 days, respectively).
RESULTS
There were a total of 3 806 578 lost days. Of those, 36% were work-related and 64% were non-work-related sick leave episodes. As shown in table 1, both work-related and non-work-related sick leave episodes were more common in men, in workers younger than 30 years of age, and those working in the region of Cataluña. By economic activity, work-related episodes were more frequent for those working in construction, while non-work-related sick leave episodes were more common in the manufacturing industry.
Work-related sick leave episodes had a median duration of 11 days (25th–75th percentiles: 6–21). The longest median duration was observed in women (12 days; 25th–75th percentiles: 7 to 24 days), workers more than 40 years of age (13 days; 25th–75th percentiles: 7–28 days), education centres (13 days; 25th–75th percentiles: 6–31 days) and the Asturias region (15 days; 25th–75th percentiles: 8–31). For non-work-related episodes, the median was 9 days (25th–75th percentiles: 4–29). The longest median durations were observed in women (11 days; 25th–75th percentiles: 5–35), workers more than 40 years of age (18 days; 25th–75th percentiles: 7–56), hotel services (12 days; 25th–75th percentiles: 5–40) and the region of Galicia (14 days; 25th–75th percentiles: 6–40).
The probability distribution of continuing absence from work was quite similar between men and women (fig 1). Initially, time to return to work was slightly longer for work-related sick leaves compared to non-work-related episodes (10 days vs 8 in men, and 12 vs 11 in women, respectively). After that, the probability distribution of continuing absence from work was lower for work-related sick leaves than non-work-related episodes. Thus, at the 75th percentile non-work-related sick leaves were 5 days longer than work-related sick leave episodes (25 vs 20 days) in men, while differences in women were greater at 11 days (35 vs 24). These differences between both types of sick leave remained during the rest of the follow-up in both men and women (data available upon request).
Table 2 presents the analysis of time to return to work by sex and by duration periods from multivariate models. In men, time to return to work for short-term sick leave was 19% longer for work-related episodes compared to non-work-related episodes (95% CI 1.17 to 1.21), while time to return to work was 48% shorter for long-term sick leave episodes (TR = 0.52; 95% CI 0.49 to 0.53). Adjusting for age, economic activity and region of the workplace only slightly attenuated this difference. A quite similar pattern was found in women: short-term episodes were 8% longer for work-related compared to non-work-related episodes (TR = 1.08; 95% CI 1.05 to 1.11), while time to return to work was 50% shorter for long-term sick leave episodes (TR = 0.40; 95% CI 0.39 to 0.42). Adjusting for age, economic activity and region of the workplace produced similar results to those for men.
DISCUSSION
The present study showed that time to return to work between work- and non-work-related sick leave episodes differed by type of sick leave. In both men and women, time to return to work for short-term sick leave episodes (less than 16 days) was longer for work-related episodes compared to non-work-related episodes, while the opposite was observed in long-term sick leave episodes (more than 15 days).
Our findings are only partially in agreement with earlier research,8 9 which found that higher sick pay benefits were associated positively with the duration of sick leave episodes. Given that in Spain sick pay benefits for short-term work-related episodes are greater than for non-work-related episodes, we expected the higher paid workers to take longer to return to work. This was true for sick leave episodes of less than 16 days. Another explanation of our findings might be that during the first 15 days many of the non-work-related sick leaves were due to minor health problems. Minor respiratory infection diseases, such as flu and cold, are some of the most frequent medical diagnoses of non-work-related sick leave.18 Another reason could be that companies may manage non-work-related episodes better because they pay the sick leave benefits. In contrast, for sick leave episodes longer than 15 days the duration of work-related episodes was shorter (48% in men and 60% in women) compared to non-work-related episodes. This reversed trend might indicate that the health problems causing the sick leave were more severe. Unfortunately, the lack of information on diagnoses in our data precluded further examination of these factors.
Episodes in women were longer than in men for both work- and non-work-related sick leaves. This difference was greater for long-term episodes, especially for non-work-related sick leaves. These differences could be explained in terms of the social role of women in present Spanish society, with growing family and job demands, together with still poorly developed social services and gender differences in the distribution of family responsibilities.19 20 The so-called “dual role”, that is, the combination of job and family demands,21 22 may be related to certain types of health problems (eg, especially musculoskeletal disorders of the neck and upper extremities23 and poor mental health24) which are more easily recognised as non-work-related rather than work-related sick leave.25 In fact, women had more non-work-related sick leaves (n = 31 835 episodes) than work-related sick leaves (n = 14 101), and their duration was also longer than for men, both for work-related (12 days in women vs 10 days in men) and non-work-related (11 vs 8) episodes. Confirmation of these results would imply that time to return to work in women is more dependent on the characteristics of the initial health problem and their severity rather than the features of the sick paid benefit, which would have a minor effect on the duration of sick leave episodes in women.
Main messages
-
Earlier studies suggest that higher sick pay benefit is associated with more cumulative compensated work absence days.
-
In short-term episodes, the level of sick pay benefits has a limited effect on the time to return to work after a sick leave episode.
-
Differences in the duration of long-term sick leave may be due to differences in the severity of medically diagnosed disorders underlying sick leave episodes.
Policy implications
Sick pay benefits do not seem to have a significant influence on the duration of episodes.
Furthermore, as reported in prior studies,26 the duration of sick leave episodes differed by age, economic activity and region of the workplace. These differences were more evident for non-work-related sick leave than for work-related episodes. However, adjusting for these variables in our analyses did not substantially alter our findings. This minor effect is probably due to a similar association of those variables with time to return to work for both types of sick leave for men and for women. Specific analyses of the differences by age, economic activity and region of the workplace, although potentially interesting, were beyond the scope of this study.
Others factors such as lack of information reported by workers on their health and working conditions, inadequate training of primary care physicians in occupational health, and especially the economic interests of insurance companies, in addition to difficulties in adherence to the legal definition of work-related health problems, have to be taken into account to explain the difficulties in distinguishing between work-related and non-work-related sick leave episodes. We cannot exclude that this misclassification of types of sick leave could bias our results. More research is needed to clarify the implications of this potential bias.
As mentioned above, an important limitation of this study is that specific medical diagnoses were not available. The severity of the medically diagnosed disorders for both types of sick leaves could be different which, as discussed above, could influence the duration of sick leave episodes. Unfortunately, in our study the analysis of sick leave episodes by diagnostic groups was precluded given that medical diagnoses were not recorded for the majority of episodes. Thus, there is an urgent need to increase the availability of this information, and to study time to return to work by specific diagnostic groups.
The strengths of the present study should also be mentioned. First, our study was based on data from a large sample of incident sickness absence episodes followed for 18 months and recorded systematically. Second, all sick leave data were obtained from a single mutua, thus controlling for potential variability in management between different medical care providers. And, third, the statistical method used for survival data analysis has taken into account that hazards were not proportional and recurrent events were present in the data. Hence, the estimations are more precise.27
In conclusion, when sick pay benefits for work-related sick leave are higher than for non-work-related sick leave, we found small differences in the duration of short-term sick leave episodes between both types of sick leave. That is to say, time to return to work in short-term sick leaves was slightly longer for work-related sick leave episodes than for non-work-related episodes. However, time to return to work in work-related episodes was much shorter for long-term sick leave episodes, probably because underlying health problems are more severe for non-work-related episodes.28
REFERENCES
Footnotes
-
Funding: This study was supported by grants from the Instituto de Salud Carlos III (FIS/04/1062) and the Ministerio de Trabajo y Asuntos Sociales (FIPROS/2006/78).
-
Competing interests: None.