Our first hypothesis was that there are factors related to LBP, to the worker, to the job and to the psychosocial environment that influence duration of an episode of sick leave. The results presented in Table
2 show that factors within the clinical, psychosocial and workplace categories are associated with RTW. Understanding these factors can help practitioners dealing with patients during the RTW process. There does not seem to be consensus between researchers on a core set of prognostic factors that should be included in prognostic studies in LBP and work disability in particular. While some may argue for the use of meta-analysis, like we did in our first review [
6], we deemed meta-analysis inappropriate for this review because of the lack of consensus on adjustment of confounders. Moreover, studies measured factors in different ways, there was inconsistency in reporting methods, and a large variability in quality of the studies.
Our second hypothesis was that in the sub acute and even more so in the chronic phase, psychological and social issues would likely become more prominent compared to the acute phase. This hypothesis cannot be confirmed, mainly due to the lack of high quality studies and a lack of consensus among researchers on what to measure, how to measure, and how to analyze the associations. In our previous reviews [
6,
17] we found strong evidence for no association of ‘pain catastrophising scale’ with RTW. We hypothesized that pain catastrophising might play a role at a later stage in the work disability process. However, there are not enough high quality studies to go beyond moderate evidence for any of the pain catastrophising and fear avoidance factors in later phases.
Limitations of the Literature
The psychosocial work environment is clearly understudied in later phases. There has been lack of consensus among researchers on how to measure psychosocial constructs and how to analyze the data based on the available theoretical models [
41]. Considering the theoretical underpinnings of the fear avoidance model (FAB) [
41], straightforward predictive analysis might not be appropriate and techniques that take the complexity of concepts and their interrelationships in the FAB model into account might be preferred.
In the acute phase, we found strong evidence for an association between radiating pain—distinctly different from ‘non-specific’ low-back pain—and RTW [
17]. Surprisingly, this factor was only examined in one study [
25] from the sub acute phase and in none from the chronic phase. More research seems warranted based on the importance of this factor in the acute phase.
Unlike in our previous review [
17], workplace factors were often not considered in the later phases. The related factors: SES, physical demands and modified duties were examined in a number of studies and, despite the crudeness of measures the results show some association with RTW.
For treatment related factors and for the effect found for modified duties, it should be noted that a prognostic study is not the most appropriate study design to examine effectiveness of interventions. Results on effectiveness of interventions can be biased in many ways when an appropriate control group is missing. The association of a
delay of referral could very well be caused by immortal time bias [
42] or time dependent bias [
43,
44] since none of the studies applied time dependent analysis to examine this bias. Those that receive intervention are likely to differ from those who do not receive intervention either at baseline or over time.
Strengths and Limitations of this Review
The strength of this systematic review is that we restricted the analysis to studies with a defined inception point. In an inception cohort, patients are included in the study at the same point in the course of their disease. In many studies on RTW the study population consists of a mixture of workers on sick leave and workers still at work at the time of inception. The number of patients at work during follow-up depends on both this mixture and on the presence of prognostic factors. Making inferences about the prognosis of RTW from such mixed studies may be misleading. It might be, however, that some researchers do not agree with the phases of disability [
45] we used, as a framework for analysis in this review. The cut-offs of 6 and 12 weeks from the Frank et al. publication [
45] are somewhat data driven: based on the median and 75
th percentile. Populations in different settings and jurisdictions have shown to have different medians and 75
th percentiles [
46,
47] which could have important consequences for the effectiveness of interventions [
46]. This classification of time on work related benefits has been extrapolated to outcomes of return to work and functional disability which might not always be appropriate [
48].
The seminal paper by Frank et al. [
45] seems to assume the outcome “end of benefits” to have a clear inception and a firm endpoint, more recent studies on recurrences [
49] and trajectories [
50] of low back pain have argued successfully that this is not always the case.”
For this review we used the quality assessment that we used in our previous reviews on the acute phase [
6,
17] to have a consistency in methods. Based on our experience, we recognise that further research is needed in the development of a tool to assess the quality of prognostic research [
51].
A prognostic study is not the most appropriate study design to examine the effectiveness of interventions. Especially because immortal time bias [
42] or time dependent bias [
43,
44] are not considered by the selected studies, and because those that receive intervention are likely to differ from those who do not receive intervention either at baseline or over time. Our findings on interventions should be interpreted with that limitation in mind.
Due to the time passed because of the magnitude of the review and the complicated analysis an update of the literature would be worthwhile, however we had to postpone publication because of knowledge transfer workshops and the development of a handbook for our funder. A quick screen of an updated search revealed few new high quality studies that could impact our findings in a substantial way. Some new findings on the importance opioid use in workers compensation settings are in our previous study [
19], it should be noted that most of these studies were in the acute phase.
Comparison of Factors in Different Phases
Workers’ recovery expectations seem important in later phases of work disability, despite a lack of high quality studies. It makes sense to ask an injured worker about their expectation for RTW. Unfortunately, there is no consensus among researchers on how to do so, nor have any of the questions used in the studies undergone psychometric testing. However, predictive validity was confirmed in all studies.
The impact of pain, functional status and radiating pain changes with duration when compared to the results from our review on the acute phase [
17]. This is somewhat puzzling, although it could be that after some time, when the worst pain has subsided, other factors become more prominent. Workplace physical factors remain important over the entire course of work disability. Therefore, an injured worker should always be asked about the work he/she did when he/she hurt his/her back and/or what kind of job he/she will return to.
The factors ‘self report of disability’ [
47,
52‐
60] and ‘pain intensity’ [
36,
52‐
65] were supported by strong evidence in the acute phase, but the evidence is less clear in the sub acute phase [
24,
25,
36,
37]. In the chronic phase, there is moderate evidence for a negative association of functional status [
20,
23,
34] and of pain Intensity [
21,
23,
29‐
33]. This might indicate a somewhat puzzling U-shape relationship between these factors and RTW over time. It could also be explained by the fact that studies adjust for different confounders.
One factor that was supported by strong evidence in the acute phase is the treatment-related factor: content of care [
47,
52,
53]. In other words, it matters with which health-care provider the worker is in contact. We found moderate evidence for an association between treatment and RTW in the sub acute phase [
24,
26,
27]. A delay in referral to intervention was associated with a delay in RTW [
21‐
23,
28,
38]. Overall, experience with and content of treatment matters [
22,
31] across all phases.
One prognostic factor that was not considered in the acute phase was the impact of functional capacity evaluations on RTW. In the sub acute phase, moderate evidence was found for an association with RTW [
20,
23,
32,
35]. In the chronic phase we found moderate evidence for a positive association of a higher score on a functional capacity evaluation (FCE) on RTW [
37]. It is not clear whether a full assessment of functional capacity is needed and whether it also predicts sustained RTW [
66‐
68]. It should be noted that not only functional capacity evaluation systems were used in work disability assessments, but more traditional “objective measures” of functional capacity like a sub maximal bike ergometer test [
35] and trunk flexibility [
20] were also included.
In the acute phase, we found moderate evidence for no association of depression on RTW [
36,
54,
58]. In the sub acute phase, a negative association between distress and RTW was reported [
25]. In the chronic phase, one high quality [
21] and one lower quality study [
23] found no statistically significant association between depressive symptoms and RTW, resulting in moderate evidence for no association of depressive symptoms on RTW. These findings are consistent with the findings from the acute phase, although there are only a limited number of studies available. Some injured workers might suffer from mental health issues, but scores on different questionnaires do not seem to predict RTW.
Earlier [
6,
17], we found that the offer of modified duties, or workplace accommodation improved RTW outcomes [
52,
58,
69]. The evidence is not as strong in later phases, mainly because the factor does not seem to be considered by many researchers [
31]. Also, when considering the evidence from the intervention literature [
70], modified duties should be considered for RTW of injured workers. Timing of the intervention seems best in the acute phase [
6,
17], although it might also be effective in the late phases [
71].
Physical demands are often measured by occupation in the acute phase [
52,
54]. Those classified as having more physical jobs are slower to return to work where self-reported physical demands were not associated with RTW [
17]. In the later phases, very few studies examined the factor
physical demands resulting in insufficient evidence in the sub acute phase [
24] and moderate evidence in the chronic phase [
22,
30,
31]. We did find strong evidence for SES on RTW [
22,
23,
28,
29]. If SES is considered a proxy for physical demands at work, the association between physical demands and RTW seems consistent across phases and should be taken into consideration in the RTW process. Future research on RTW in the later phases of work disability should examine physical demands by using more objective measures.
Job satisfaction was supported by strong evidence in our previous review [
17]. It was not examined in the sub acute phase and only one publication [
29] reported on it in the chronic phase. The impact of job satisfaction might diminish after a longer time away from the job; however evidence for that hypothesis is lacking.
We found insufficient evidence for an association between age and sex and RTW in acute LBP [
17]. There is moderate evidence for a negative association of older age on RTW [
24,
26] in the sub acute phase. In the chronic phase, most studies also reported a negative association. Across all phases, the evidence is conflicting and calendar age might not be the most appropriate measure to capture the concept.
In the sub acute phase we found strong evidence for no association of sex on RTW [
24‐
27]. However, one medium quality study found a longer time until RTW in men [
26]. There is moderate evidence for an association between male sex and RTW [
20,
21] in the chronic phase. Although two studies [
22,
23] did not find an association between sex and RTW, this is not contradictory and could be due to small sample sizes. Overall, the association between sex and RTW is inconsistent across phases and might be the result of gender specific workplace based exposures [
72].
Future Research
Prognostic research in work disability prevention would benefit from consensus among research and practitioners on what factors are deemed important and how they should be measured and analysed. Claim-related factors are supported by strong evidence in the chronic phase, and in all cases, are related with delays and experiences in the claims process. This factor was not considered in earlier phases [
17]. Some of the claim related factors might be time dependent: they start to play a role at later stages of work disability due to negative side effects of being in the administrative and adjudicative process that happens alongside the RTW process. Further study into claim-factors seems justified.
When presenting the findings from our review to practitioners, it was clear that there is little consensus on what “psychosocial” means in research but great consensus on the importance of the construct in practise. There seems to be a clear disconnect between research and practice that should be resolved.
Understanding of the importance of different prognostic factors at various times in the RTW process can inform stakeholders about the most appropriate actions that can be taken to improve RTW outcomes. To transfer the messages from this review we have presented the findings in a number of workshops. Based on the feedback from stakeholders we are currently developing a Handbook on Prognosis of RTW in LBP for use in practise. The handbook emphasizes the role of recovery expectations and the importance of the workplace and physical demands on the job, and provides suggestions to uncover these constructs when dealing with injured workers trying to RTW. The impact of providing such information to work disability practitioners should be studied.