Decision rights redesign in a hospital network

A regional hospital network of four hospitals and one merged administration had arrived at governance by accretion. It ran 31 standing committees, most of them created to solve a problem nobody could still name. Clinical quality was strong. Administrative decisions were not. A capital request took a median 61 days to decide, and operational questions routinely visited three committees before finding one that felt authorised enough to answer, or before whoever raised them gave up.

The culture was collegiate and consensus-minded, which everyone valued and nobody wanted destroyed. The brief was precise. Keep the consensus where it earns its cost, take it out of the decisions where it is only delay, and do not pretend those two categories are obvious from an org chart.

Over six weeks we traced 74 actual decisions through minutes and approval trails. The network did not have a consultation problem. It had an ownership problem. For 60% of recurring decision classes no single accountable owner existed, so committees consulted each other because none of them believed it could decide alone, which is a rational response to an unclear mandate rather than a cultural failing.

The latency was priced at roughly $3.9M a year in delayed works, locum cover extensions and equipment idle time. The board had never seen that number, because it lived in no budget line. It is an estimate assembled from operational records rather than an accounting entry, and we said so when we put it in front of them, because a number like that is easier to dismiss than to defend.

The decision inventory became a ledger. Every recurring decision class got one accountable owner, a deliberately short consulted list, and a named forum where the decision actually happens, with the consulted list kept short on purpose because every name added to it adds days. Clinical-safety decisions kept their consensus machinery intact, because that is consensus earning its cost. Administrative and capital decisions moved to named owners on a fixed cadence.

Before adoption we tested the draft ledger against the last ten significant decisions the network had actually made. The test reshaped a third of it. That is the reason the ledger was adopted as a description of how the network already works rather than an aspiration it would file and forget, and it is the step we would give up last if a client asked us to compress the work.

Latency read from calendars and approval trails, not interviews; month-11 readings verified under MKM-F-003.