Our study makes several novel contributions to the existing literature. Firstly, we found that only about one third of the eligible employees were actually offered OIM services, and secondly, that roughly three out of four employees accepted that offer. Further, our study identified several independent predictors of receiving an OIM offer which relate to employees’ health (i.e., mental impairment, skin disease, cancer, and long-term sickness absence), psychosocial characteristics (i.e., OJ and neuroticism) and, in particular, company size. Acceptance of an OIM offer was weakly associated with only two independent predictors; those were mental impairment and the company size.
4.1. Interpretation of Findings
The observation that only one third received an OIM offer, but that three quarters decided to accept that offer may highlight considerable unmet needs for support services among employees returning to work form long-term sickness absence. Reasons to decline OIM offers may pertain to the fear of losing one’s job when functional limitations are discussed or concerns regarding data protection [11
]. Based on the available literature, the high level of acceptance among employees documented by our study can be contrasted with the acceptance levels among employees that relevant organizational stakeholders assume: in a survey addressing OIM for mental illness [12
], 60% of the participating OIM experts reported that OIM offers are always or often accepted by eligible employees according to their experience. By contrast, in another survey (mostly involving employee representatives, representatives of employees with severe disability and human resources staff), only 37.1% of the stakeholders assumed that the acceptance of OIM among eligible employees would be high or very high [11
]. While those estimates are not readily comparable (e.g., due to larger potential for selection bias in prior surveys [11
] and sample differences), the current evidence possibly suggests that OIM experts underestimate the acceptance of OIM offers among employees.
We found that mental impairment and cancer were associated with an increased probability of receiving an OIM offer. In Germany (just like, assumedly, in most other European countries), employees are not obliged to disclose their illness to their employer. Employees may nevertheless disclose their illness or characteristic functional limitations—despite the risk of stigmatization—because disclosure increases the likelihood of receiving more specific support [28
]. Upon illness disclosure, employees with mental impairment may be more likely to receive an OIM offer because occupational stakeholders involved in OIM perceive the needs of those employees as particularly demanding [12
] and/or because there is increasing awareness, e.g., among supervisors [29
], of the importance of their support in that type of RTW process. Employees with cancer may be more likely to receive an OIM offer, because their illness may be perceived as particularly severe and elicits fear in individuals without cancer [30
]. Conversely, one may speculate that skin conditions are viewed as less severe or less disabling or that solutions for employees with such conditions are primarily sought through other services (e.g., occupational safety services). This may lower the probability of OIM offers. Individuals with the longest sick leave may be more likely to receive an OIM offer, since long-term absence signals a poor RTW prognosis [31
] and therefore employers may be particularly committed to facilitate successful RTW. OJ may positively relate to OIM offers, because OJ is an indicator of employee-oriented workplaces. The positive relationships between neuroticism (i.e., proneness to experience psychological stress [22
]) and the probability of an OIM offer may be due to the fact that individuals with high neuroticism are more likely to report somatic symptoms [32
]. Alternatively, just as employees with mental impairment, those employees’ needs may be perceived as very demanding, thereby eliciting OIM offers.
Increasing company size was the strongest predictor of future OIM offers. This finding may be explained by the fact that larger companies have more resources: more manpower implies that there is more staff to facilitate the OIM processes, for instance, in terms of administrative tasks, the involvement of specialized OIM teams [11
] or in-house occupational health services. Higher financial means imply that larger companies are able to offer a broader set of interventions and more expensive interventions [11
], e.g., concerning the adaptation of workplaces. The fact that OIM offers are less likely to occur in small companies should not interpreted as evidence of less support for workers returning from long-term sick leave in small companies. Support in small companies may take a more informal shape and may in fact be particularly strong [11
] due to more trustful and closer relationships [33
]. Trustful relationships may enable employees to better express their functional limitations, which should result in more suitable interventions in the OIM process.
Regarding acceptance of OIM offers, we found that mental impairment was a weak positive predictor. Possibly, employees with mental impairment are more aware (e.g., due to psychotherapy treatment or counseling) of the importance of support resources to facilitate their coping and are thereby more likely to accept offers. Regarding the company size, we found that the probability of accepting OIM services decreased with company size. This may be because large companies have standardized OIM processes. Therefore, interventions may not be suitably tailored to individual employees who thus decline participation. Furthermore, as mentioned above, trust may be higher in small companies and increases the acceptance of OIM offers (e.g., by reducing the fear of job loss or of misuse of personal data).
4.2. Methodological Considerations
The strength of our study is that we were able to draw on data which allowed examining numerous potential predictors. Moreover, our study was based on a prospective design, which usually introduces a temporal sequence between exposures and outcomes and thereby increases confidence in the causality of the observed associations. It needs to be mentioned though that we were unable to establish a temporal sequence in our study with absolute certainty. Due to the wording of the OIM items, we cannot rule out that OIM offers had been made prior to baseline assessments. Further—though it seems unlikely—we cannot rule out that some participants changed their employers throughout the follow-up period. Moreover, our study relied solely on self-reported data, which may be partially misreported. With regard to the OIM items, for instance, we cannot rule out that, in some instances, employees had received an OIM offer but did not recognize it as such (i.e., if the offer was made in an informal way, e.g., in small companies). Another weakness is that our study assessed only a small range of workplace-related data (i.e., company size and OJ perceptions). Further, our response rates and potential selection bias need discussing. The response rate at the follow-up was decent (67.79%). Overall however, only 22.56% of those who received an invitation to complete the baseline questionnaire provided data at both baseline and follow-up assessments for the current analyses (i.e., 2233/9897). Notably though, the extent of potential selection bias is contingent upon the relationship of participation with the exposures of interest, with the outcomes or with the association of those two in a given study [34
]. For the current study, the data on exposures (except for age and gender) and outcomes were not available for baseline non-participants. At the follow-up, we observed differences between participants and non-participants with regard to sociodemographic characteristics at baseline (see above and Online Resource 1: Table S1
). There was no evidence of a consistent trend though towards better health among follow-up participants as compared to non-participants (including days with sickness absence). With regard to the many psychological variables considered, only social support and effort–reward imbalance seemed to be slightly higher among follow-up participants versus non-participants. Furthermore, at the follow-up, participants were more likely than non-participants to work for a large company. Overall, we thus observed that some of the considered exposures at baseline were associated with follow-up participation. However, potential associations of follow-up participation with the outcomes (i.e., receiving or accepting an OIM offer) remain unknown due to lack of such data. Thus, based on the available data, we are unable to comprehensively examine potential selection bias. It deserves mentioning though that low response rates alone do not necessarily imply selection bias [34
]. This notion is supported by previous research involving original data from health surveys [34
We had the opportunity to utilize data from a unique sample of employees eligible for OIM. Due to this special focus, the generalizability of our findings may be limited though: our sample is not representative of the total workforce in Germany. Among other things, this is due to the sampling within a restricted age range (i.e., 40–54 years), the inclusion criterion of having received sickness benefits (i.e., rendering our sample likely less healthy than the general workforce) and the recruitment through the GPI. Workers enrolled in that pension scheme are characterized by a higher socioeconomic status in terms of their educational levels, vocational qualifications and income [35
]. Furthermore, they are less frequently exposed to high work-related physical demands and feature a higher proportion of employees (e.g., as opposed to self-employed individuals) compared to the general population [35
]. Based on our sampling approach, we can assume good generalizability of our findings specifically to employees who had received sickness benefits. However, while all employees who receive sickness benefits are eligible for OIM, there may be employees who are entitled to OIM, but have not received sickness benefits. These may be employees with several short sick leave episodes that accumulate to six weeks across 12 months and/or who are on a sick leave for varying conditions. Those individuals are not represented in our study.
4.3. Recommendation for Research and Practice
Additional studies are needed to corroborate our findings. Preferably, such studies should be based on prospective designs and utilize administrative data whenever possible and suitable (e.g., to define whether an OIM service offer was sent out). Given that company size was the strongest predictor in our study, it seems useful to further explore explanations for that association (see above) and to examine additional workplace-related or economic factors (e.g., the company’s financial means) [33
]. Further, characteristics of key players in the OIM process may be of interest. It has been found, for instance, that supervisors’ support of OIM may be higher when they have themselves faced impaired workability [33
As mentioned above, low availability of OIM offers accompanied by their frequent acceptance suggests a gap in the supply of OIM services. Awareness of this issue needs to be increased among service providers and employers alike. Providers of health or social services (e.g., in rehabilitation clinics) need to inform employees about the fact that they are entitled to OIM, should explain the aims of OIM, the potential procedures and legal rights and support the employee’s decision-making (e.g., whether and how to claim OIM in case that offer is not made). It may be particularly promising to support employees in initiating OIM services themselves, in particular in small companies which may not have established OIM procedures yet or are unaware of OIM [33
Some practice guidelines for employers on how to carry out OIM already exist (e.g., [10
]) and awareness of those resources needs to be increased in companies, especially in small companies. Many OIM experts seem to feel though that their company will likely not be able to cope with future OIM cases, in particular due to mental health conditions [12
]. Thus, it seems promising to assess the suitability of the available guidelines and how to possibly improve them. For instance, it may be helpful to expand guidelines with illness-specific (e.g., mental illness) or sector-specific (e.g., service sector) information that highlights typical barriers for successful OIM, how to overcome those challenges and context-specific interventions.
While employers may perceive OIM services to be effective, e.g., in terms of reduced absenteeism [11
], experimental evidence is needed to empirically establish such effectiveness. A meta-analysis addressed the potential effectiveness of “RTW coordination programs”, which the authors defined as programs that (i) aim to promote RTW, (ii) build on at least one face-to-face contact between the returnee and a RTW coordinator, (iii) assess the returnee’s needs and devise individualized RTW plans and (iv) whose implementation is managed by a RTW coordinator [3
]. Overall, such programs seemed to offer no benefits to workers when compared to usual care, e.g., in terms of successful RTW or reduced absenteeism. Another meta-analysis [4
] addressed the effectiveness of workplace-related interventions (e.g., modification of work design, working conditions or environments) to improve RTW and reduce absenteeism and delivered varying findings. It is challenging though to contextualize OIM in light of those meta-analyses and other prior work as comparability is restricted. For example, both meta-analyses examined RTW as the outcome based on studies, which recruited individuals who were on sick leave. However, OIM is offered in the early stages of the actual return to work process. Secondly, except for a consensus on general features of OIM (see Introduction), there is no clear definition of the OIM process and the inclusion criteria of one of the meta-analyses [3
] may not cover all cases of OIM as defined in this paper. Overall, high-quality evidence specifically evaluating OIM effectiveness is thus needed.