The research behind the method.

The argument

Patient experience and staff engagement are one problem. Hospitals measure them separately, fund them separately, and assign them to different departments. The data from the study unit showed them moving as a single variable.

The lever is local. A multiple linear regression across 243 shift-level sentiment readings found that the biggest announced reorganization in the health authority's history, community events, overtime, and sick time had no significant effect on whether staff felt valued. What did: negatively perceived local decisions, the individual frontline leader on shift, unit outbreaks, and the better-resourced night shift. The model explained about 40% of the variance.

The mechanism is subtractive. The intervention did not add a program. It identified the visible, ordinary leadership behaviours that staff and patients were responding to, and removed the meetings, reporting, and process that kept leaders from doing them.

Negativity erases gains. Positively perceived leadership actions did not register as significant. Negatively perceived ones did, and the closer the leader to the frontline, the larger the effect. Turning toward dissenting voices in the final weeks, through three town halls and an anonymous channel, is what flipped the remaining negative sentiment.

The results. Staff feeling negatively valued by leadership fell from 40% to 2% over 12 weeks. Patients reporting they always felt listened to rose 32 points to 86%, against a goal of 74%. Patients reporting they were always involved in decisions about their care rose 28 points to 93%, against a goal of 85%. At the 24-month follow-up the unit's engagement gain held at two standard deviations.