Introduction The National University Hospital is a tertiary hospital with around 6400 employees. A ‘Total Workplace Safety & Health’ program was started in 2014 by assessing organisational and individual health risks using a health survey, workplace visits, accident reports, results of staff health screening and an internet-administered survey regarding health, stress, behaviour, and program engagement. Interventions included evidence-based health screening, a ‘healthy-eating’ campaign introducing a traffic-light system identifying healthier food at worksite canteens, needle-stick injuries reduction programs and motivational campaigns to increase physical activity. In spring 2017 the impact evaluation started. We report on lessons learned on program implementation, methodology and results of this impact evaluation.

Methods Barriers to program success and evaluation were assessed through quantitative and qualitative methods focus-group discussions, employee survey, vendor interviews and direct observations for Healthy-Eating and Physical-Activity Interventions, including environment assessment, plate-counts of meals, sedentary and eating behaviours, activity habits, acceptance of messages.

Result Feedback for lifestyle/behavioural interventions suggests modest participation and unknown impact. Participation in medical programs increased (vaccination +30%), previously undiagnosed chronic diseases were identified (unclear/pathological results 5% high blood-pressure, 5% high blood-glucose, 11% cancer screenings), a follow-up process was implemented. Organisational and environmental barriers for long-term program success were identified. Great enthusiasm got the programs started but lack of planning for evaluation hinders impact assessment. Survey, focus-groups and observations are completed in July 2017.

Discussion Lessons learned include: Clarify responsibilities and reporting lines, plan for monitoring/evaluation at the beginning, get a baseline, define program goals. Clearly defined and medical-style programs showed more positive results and participation. Lifestyle interventions need to set realistic targets. ‘Asking’ people on behaviour is notoriously biassed. Unplanned evaluations ‘ex-post’ are unsatisfactory or impossible. Only some programs are open for cost-effective evaluation on hindsight and coordination from the top is crucial for impactful interventions and their evaluation.