A Bed Built Is a Bed Filled

Good evidence rarely loses an argument on its merits. More often it loses to the decision architecture around it: the number of people with a stake in the choice, the reputations and budgets riding on it, and the headline nobody wants to own.

I was reminded of this over the weekend. Examiner feedback on my doctoral thesis arrived while I was out at dinner on Saturday night, and one of the questions put to me was what Roemer's Law might mean in the Australian healthcare context. It took up a good part of my Sunday, and it has followed me around the farm this Monday morning too.

The freeway that filled itself

Before we talk about Roemer's Law, some background to larger point that I would like to make. Thirty-odd years ago I worked in the centre of Sydney and lived in a small town about 100 kilometres out. In those early years the freeway was a lonely place for those committed to the commute. So was the train.

Slowly but surely, the capacity of those roads and rail services grew. And people came to fill it, until infrastructure that once felt empty was bursting at the seams.

Transport planners call this induced demand. Build the road and the traffic arrives to use it. The new capacity doesn't relieve the pressure; it changes the choices people make, and the pressure returns at a higher volume.

Roemer’s Law - A bed built is a bed filled

Healthcare has its own version, and it has been sitting in plain sight since 1959. Milton Roemer, a specialist in health systems and their capacity, observed that a bed built is a bed filled. If a hospital bed is available, someone, somewhere, will make a choice that fills it.

That was 67 years ago. Look it up and you'll find it explained in a paragraph. It isn't hard to understand.

The wisdom that follows is just as simple. Adding fixed capacity is not your best answer to a system capacity problem. What you need is flexibility: the ability to surge up when demand rises and surge down when it falls.

A problem of flow, not of front doors

Here in Tasmania, and I suspect right across Australia, a version of this argument is still running. Ambulances ramped outside emergency departments are its most visible symptom. The debate keeps circling the same question: do we need more beds in emergency, or do we need to move people through the system more quickly?

There are data and patient flow specialists in the health service who know the answer. One very clever fellow who studies patient flow here in Tasmania framed it for me the way you'd think about a postal system: inputs, processes and outputs. The queue at the front door, he'd tell you, is rarely caused by a shortage of capacity at the front door. It's caused by blockages further along. Ramping is a problem of flow through the system, not of overall system capacity.

The risk register that missed the risk

The point to note here is that the true wisdom to solve the problems that the system is experiencing is available and present if we can make the hard choice to look closely. A perfect example was when I was doing safety and risk work with the NSW Police Force.  I was working with the superintendent of a local area command to review his risk register which was a simple Excel list; and most of it concerned old buildings and ageing plant.

Partway through, the commander voiced his frustration. All this time spent on compliance items, he said, when there were far more serious matters to deal with.

So we turned towards those matters directly. We put three open prompts to the wider team in the room:

• Tell us about your work and experiences.

• Tell us how the law, regulations, policies and procedures applied to those experiences, and where they did or didn't fit.

• Tell us about the outcomes. What happened as a result?

The folks in the room immediately started telling work stories that fit within this simple structure and serious risks surfaced very quickly. The command sat on the edge of the outback in NSW, and officers routinely drove eight to ten hours on patrol to remote areas, including some areas where communications regularly failed. On one occasion, an officer's vehicle had broken down in one of those areas, leaving them stranded for two hours with no way to call for help. It was luck, nothing more, that someone drove past and raised the alarm.

None of this was on the register. All of it was known, held in the hard-won experience of the people doing the work. The fix was straightforward: a protocol requiring a two-hourly safety check-in with the home station, backed by upgraded communications equipment.

The knowledge hadn't been missing. The architecture of the decision had simply never asked for it.

Why the evidence doesn't land

So if the law is simple, and the people closest to the work understand it, why are we still having the same argument nearly seven decades on?

Because a decision like this is never one person weighing one piece of evidence. Gathered around it are ministers, executives, clinicians, budget holders and boards, and every one of them has something riding on the outcome. Reputations are at stake. Careers are at stake. Budgets are at stake. And a queue of patients outside your local hospital guarantees a news story no politician wants to wear.

Under that pressure, building beds is visible, announceable and easy to defend. Redesigning flow is slow, technical and hard to put in a press release. The evidence hasn't lost on its merits. It has lost to the architecture of the decision.

This is the knowing-doing gap in its purest form. Integrating good knowledge into professional practice is difficult. Making choices well in complex environments, with this many hands on the wheel, is harder still.

The question worth asking

If a finding this simple, this old and this well understood still can't settle the argument, the problem isn't a shortage of evidence. Producing more of it won't help.

So the question for anyone leading change in a complex system isn't 'What does the evidence say?' It's 'what would have to change about how this decision gets made for the evidence to win?'

In that outback command, the answer was three open questions put to the people who already knew.