Its possible the reason is unknowable at this time. As an example one theoretical reason you didn't mention is changing bandages twice as often reduces post-op

Its possible the reason is unknowable at this time. As an example one theoretical reason you didn't mention is changing bandages twice as often reduces post-op infections by 75%, although no one has discovered or proven that yet, its sounds possible, and post-op infections are expensive and painful compared to cheap bandage replacement. Minimization of cost of any random individual component almost always doesn't result in minimization of total system cost. Its important to build robust fault tolerant fail safe systems. By definition, half of all management will be below median. A system designed to give the bad ones maximally overpowered weapons to wreck havoc is not a good system design. Numerology as a weapon can be destructive. It seems reasonable that as a system it might enable the top half of managers to take three steps forward, while enabling the bottom half of managers to take five steps back, not forgetting the fixed cost due to documentation and processing labor (bean counting) and micromanagement somewhere around a fixed two to four steps back just to gather and process the data, regardless if its used or not. That's a badly designed system. This is before design criticism along the lines of never implement a system more complicated than its minimal need. Simplicate and add lightness. The substantial money spent counting, processing, and graphing every band aid could have been spent more simply, probably more productively, by spending $X extra on surgeon hiring salary to get a slightly measurably higher class of surgeon... or manager... Or the mental effort spinning around inside a surgeon's brain while cutting could be spent on improving patient outcomes by perfecting incisions or sutures or during surgery judgement calls ... or those brain cycles can be spent on minimization of band aids to meet band aid quota.