BackgroundWork participation among people with rheumatic and musculoskeletal diseases (RMDs) remains reduced when compared to the general population. A EULAR taskforce was established to agree on Points to Consider (PtC) to support people with RMD in healthy and sustainable work participation. Non-pharmacological interventions (NPI) could have an important role in improving work participation in RMDs. However, a comprehensive evidence synthesis of the effectiveness of NPIs in people with RMDs is lacking.ObjectivesTo summarise the literature on effectiveness of NPIs on work participation in people with RMDs.MethodsA search in four databases (MEDLINE, EMBASE, CENTRAL and CINAHL) was performed. Randomised Controlled Trials (RCTs) and Longitudinal Observational Studies (LOS) assessing non-pharmacological/non-surgical interventions until August 2020 were screened. Studies including people with any RMD (except low back pain or work-related RMDs) and assessing a work participation outcome domain (sick leave, work status, presenteeism) were considered eligible. For qualitative evidence synthesis, RCTs and LOS were considered. For quantitative evidence synthesis, only RCTs were considered. For each randomized comparison, standardised mean differences (SMDs) were calculated for the three outcome domains and used as effect size in the meta-analyses; i.e. a negative SMD favouring the NPI over control. Next, Mixed Effects Meta-Regression Analyses were performed, with random effects for randomised comparisons, and a fixed effect factor for the stratified subgroups of clinical interest. Subgroups within diseases (musculoskeletal pain disorders vs. other types of RMDs), risk status for sick leave at baseline (on sick leave or at risk for sick leave; not at risk for sick leave; a combination; or not specified) and single vs. multiple component NPIs were pre-defined. Risk of Bias (RoB) of RCTs was assessed using the Cochrain tool.ResultsOut of 8,864 records, 64 studies (71 treatment comparisons) were included. Studies usually included a mixed population of several RMDs (42%). Most NPIs were conducted in a clinical setting (n=44, 62%) and NPIs usually had multiple components (n=57, 80%), such as vocational support combined with physical training (n=18, 25%). Sick leave was the most frequently reported outcome domain (n=56, 88%). In the qualitative syntheses, 30%/50%/29% of interventions were considered plausible in improving sick leave, work status and presenteeism, respectively.In the quantitative synthesis, NPIs (37 RCTs, 42 comparisons with mostly moderate to high RoB) showed small to moderate effect sizes, favouring NPIs over comparators for sick leave (SMD=-0.23, 95%CI -0.33 to -0.13), work status (SMD=-0.38, 95%CI -0.63 to -0.12) and presenteeism (SMD=-0.25, 95%CI -0.39 to -0.12). The forest plot for sick leave is shown (Figure 1).Subgroup analyses for sick leave revealed that, compared to control, NPIs were not effective in musculoskeletal pain disorders, in contrast to the other types of RMDs. For both other subgroup analyses (baseline risk for sick leave; single vs. multicomponent NPI), NPIs improved sick leave similarly in subgroups compared to the control. Subgroup analyses for work status and presenteeism had generally similar effects in subgroups, but the interpretation requires caution in view of the small number of comparisons. Of note, clinical and methodological heterogeneity between studies was substantial, with some concerns about methodological quality related blinding and completeness of follow up.ConclusionOverall, NPIs seem to have significant, but on the average population level only small to moderate effects on sick leave, work status and presenteeism in RMDs. However, effects on sick leave varied substantially between subgroups. This synthesis suggests that tailoring NPIs to individuals’ needs and context could be clinically valuable.Disclosure of InterestsNone declared.