Scheduling and float-pool patterns that stabilize clinical coverage.
Night coverage gaps are usually solved by asking the same reliable people to absorb more. That works for a quarter and then produces the resignations that widen the gap.
A dedicated night cohort outperforms rotation for most units. People who choose nights and are staffed permanently to nights report better sleep stability and leave less often than staff rotated in and out.
Pair that cohort with a float pool sized to actual variance, not to the average. Coverage models built on the mean will fail on every above-average night, which is where burnout accumulates.
Differentials matter, but predictability matters more. Publishing schedules further ahead consistently reduces call-outs more than incremental pay increases do.
Finally, measure consecutive-shift counts per person, not just total hours. The count, not the total, is what predicts who resigns next.