AbstractBackgroundIn the face of pressure to contain costs and make best use of scarce nurses, flexible staff deployment (floating staff between units and temporary hires) guided by a patient classification system may appear an efficient approach to meeting variable demand for care in hospitals.ObjectivesWe modelled the cost-effectiveness of different approaches to planning baseline numbers of nurses to roster on general medical/surgical units while using flexible staff to respond to fluctuating demand.Design and SettingWe developed an agent-based simulation model, where units move between being understaffed, adequately staffed or overstaffed as staff supply and demand, measured by the Safer Nursing Care Tool, varies. Staffing shortfalls are addressed firstly by floating staff from overstaffed units, secondly by hiring temporary staff. We compared a standard staffing plan (baseline rosters set to match average demand) with a ‘resilient’ plan set to match higher demand, and a ‘flexible’ plan, set at a lower level. We varied assumptions about temporary staff availability. We estimated the effect of unresolved low staffing on length of stay and death, calculating cost per life saved.ResultsStaffing plans with higher baseline rosters led to higher costs but improved outcomes. Cost savings from low baseline staff largely arose because shifts were left under staffed. Cost effectiveness for higher baseline staff was improved with high temporary staff availability. With limited temporary staff available, the resilient staffing plan (higher baseline staff) cost £9,506 per life saved compared to the standard plan. The standard plan cost £13,967 per life saved compared to the flexible (low baseline) plan. With unlimited temporary staff, the resilient staffing plan cost £5,524 per life saved compared to the standard plan and the standard plan cost £946 per life saved compared with the flexible plan. Cost-effectiveness of higher baseline staffing was more favourable when negative effects of high temporary staffing were modelled.ConclusionFlexible staffing can be guided by shift-by-shift measurement of patient demand, but proper attention must be given to ensure that the baseline number of staff rostered is sufficient.In the face of staff shortages, low baseline staff rosters with high use of flexible staff on hospital wards is not an efficient or effective use of nurses whereas high baseline rosters may be cost-effective. Flexible staffing plans that minimise the number of nurses routinely rostered are likely to harm patients because temporary staff may not be available at short notice.Study registration: ISRCTN 12307968Tweetable abstractEconomic model of hospital wards shows low baseline staff levels with high use of flexible staff are not cost-effective and don’t solve nursing shortages].What is already known?Because nursing is the largest staff group, accounting for a significant proportion of hospital’s variable costs, ward nurse staffing is frequently the target of cost containment measuresStaffing decisions need to address both the baseline staff establishment to roster, and how best to respond to fluctuating demand as patient census and care needs varyFlexible deployment of staff, including floating staff and using temporary hires, has the potential to minimise expenditure while meeting varying patient need, but high use of temporary staff may be associated with adverse outcomes.What this paper addsOur simulation shows that low baseline staff rosters that rely heavily on flexible staff increase the risk of patient death and provide cost savings largely because wards are often left short staffed under real world availability of temporary staff.A staffing plan set to meet average demand appears to be cost effective compared to a plan with a lower baseline but is still associated with frequent short staffing despite the use of flexible deployments.A staffing plan with a higher baseline, set to meet demand 90% of the time, is more resilient in the face of variation and may be highly cost effective