Effect of an E-mental health approach to workers' health surveillance versus control group on work functioning of hospital employees: a cluster-RCT.

To evaluate an e-mental health (EMH) approach to workers' health surveillance (WHS) targeting work functioning (WF) and mental health (MH) of healthcare professionals in a randomised controlled trial.Nurses and allied health professionals (N = 1140) were cluster-randomised at ward level to the intervention (IG) or control group (CG). The intervention consisted of two parts: (a) online screening and personalised feedback on impaired WF and MH, followed by (b) a tailored offer of self-help EMH interventions. CG received none of these parts. Primary outcome was impaired WF (Nurses Work Functioning Questionnaire), assessed at baseline and after three and six months. Analyses were performed in the positively screened subgroup (i) and in all participants (ii).Participation rate at baseline was 32% (NIG = 178; NCG = 188). Eighty-two percent screened positive for at least mild impairments in WF and/or MH (NIG = 139; NCG = 161). All IG-participants (N = 178) received part (a) of the intervention, nine participants (all positively screened, 6%) followed an EMH intervention to at least some extent. Regarding the subgroup of positively screened participants (i), both IG and CG improved over time regarding WF (non-significant between-group difference). After six months, 36% of positively screened IG-participants (18/50) had a relevant WF improvement compared to baseline, versus 28% (32/115) of positively screened CG-participants (non-significant difference). In the complete sample (ii), IG and CG improved over time but IG further improved between three and six months while CG did not (significant interaction effect).In our study with a full compliance rate of 6% and substantial drop-out leading to a small and underpowered sample, we could not demonstrate that an EMH-approach to WHS is more effective to improve WF and MH than a control group. The effect found in the complete sample of participants is not easily interpreted. Reported results may be useful for future meta-analytic work.Dutch Trial Register NTR2786