What the building has already decided

During design, everything on this page is a matter of opinion, and the opinions are held by people with different interests and unequal influence. Somebody thinks the ward is too long. Somebody else has a budget. The argument is real, it is winnable, and for a period of about two years it is the most consequential conversation anybody involved will ever have.

Then the building opens, and the same sentences stop being arguments and become measurements. The ward is not too long any more. The ward is one hundred and four metres, and that is not a view, it is a number, and it will be that number in 2086.

This distinction matters more than it sounds, because organisations behave badly at the boundary. A complaint about a constant is indistinguishable, in a meeting, from a complaint about something fixable. It is raised with the same words, at the same meeting, by the same person, and it consumes the same time. The difference is that one of them can be acted on.

So the list below is not a lament. It is a filter. Everything on it is closed, which means the effort of everybody now working in the building belongs somewhere else, and knowing exactly where the line falls is worth a great deal to whoever is running the place.

Eleven of them. Read the third field of each, because that is the one that decides whether an item belongs on this list at all: not what it costs, but what undoing it would actually take.

Almost none of them are undone. They are absorbed, which is a different thing and is the reason they stay invisible. An absorbed constant does not appear in any account. It appears as a slightly tired workforce, a discharge that happens after lunch, a list that overruns on Thursdays, and a set of explanations that are all about people.

Take the first item and price it, because the scale of a distance error is genuinely difficult to believe until it is written out.

Ten metres. Not a wing in the wrong place, not a department on the wrong floor. One store cupboard positioned ten metres further from the bed bays than it needed to be, which is the kind of thing that gets settled in a fifteen minute conversation about a drawing, if it gets discussed at all.

The third figure is the one to sit with, and it is one ward. A hospital has many. Nothing in that calculation is exotic: it is a stated trip count, an ordinary walking speed and a sixty year life, and the trip count is the only assumption in it worth arguing with.

That is what makes this the cheapest thing in the entire study to get right and one of the most expensive to get wrong. The correction costs nothing at design stage, because at design stage it is a line on a drawing. Afterwards it is not available at any price, and it is paid every day by people who were not consulted, in a currency nobody records.

The list is only useful if it changes what somebody does on Monday, so here it is as five tiers ordered by how reversible they are.

The pattern worth noticing runs the wrong way. Effort and attention in most organisations concentrate in the middle tier, equipment, because that tier has a budget line, a supplier, a business case format and a visible outcome. The bottom two tiers, which are where nearly all of the available improvement actually sits, have none of those things and are therefore chronically under resourced relative to what they return.

And the top tier, which determines more of the outcome than the rest put together, is closed before most of the people who will live with it have been hired. Which is the argument for doing this work at the gates rather than afterwards, and the reason a hospital that has already opened should stop spending meetings on tier one entirely.

Everything above is written from the position of somebody about to build. Turn it round and it becomes something more immediately useful, because the eleven constants are all measurable in a building that is already standing.

That matters because there is no good way to compare two hospitals. Clinical outcomes are confounded by case mix to the point where a fair comparison takes a research team. Reputation is a lagging indicator of something else. Inspection ratings measure compliance, which is a floor rather than a ceiling. And management quality is real but transient, since it can change with one appointment and frequently does.

The constants have none of those problems. They are physical, they are stable, they do not vary with who is on shift, and they can be measured by two people with a tape measure, a stopwatch, a floor plan and two days on site. They are, by some distance, the most objectively comparable thing about a hospital, and almost nobody looks at them.

So the same eleven items become a survey. Each row below says what to measure, how, what a good answer looks like, and, in the last column, what a bad one permanently caps. That last column is the point of the exercise: a benchmark that does not tell you what a poor score costs is a scorecard rather than a diagnosis.

There is one way to misread this survey and it is the obvious one, so it is worth closing off before anybody uses it in anger.

The score is a ceiling. It says what the building permits, not what the organisation achieves. A hospital can measure well on all eleven and still be mediocre, because constants are permissive rather than causal. And a hospital can measure badly and be genuinely excellent, run by people who have spent years building processes specifically to work around a layout nobody should have had to work around.

Which is why the survey is only useful next to the operating data. The constants give you the ceiling. Length of stay, discharge times, cancellations, turnover and infection rates give you the actual. The gap between them is the only honest measure of management there is, because it is the only one that has already subtracted the part the team did not choose.

Read that way it separates two findings that are confused almost universally. A large gap is a management finding, and management findings are cheap to act on and quick to move. A low ceiling is a capital finding, and capital findings are slow, expensive and sometimes not available at any price. Telling an excellent team to try harder against a low ceiling is the most common waste in hospital management, and the most demoralising.

Three situations make this worth doing rather than merely interesting.

Buying. An acquisition is usually assessed on financial performance, case mix and the state of the equipment, all of which are consequences. The constants are causes, and they set the bound on everything the buyer intends to improve. A building with no expansion land and a shallow ceiling void will resist every plan in the investment case, and the two days it takes to find that out are the cheapest two days in the transaction.

Running what you already have. The survey tells an operator which of their recurring problems are constants and which are not, which is exactly the filter the first half of this page argued for. It converts an argument that has run for years into a fact, and it redirects the effort to the tiers where effort still works.

And selling, which is our own case. Meridian sells its hospitals at maturity, to a buyer with a technical adviser who is paid to find the weaknesses. A seller who has already run this survey on its own building, published the result and acted on what was actionable arrives in that room with the awkward questions already answered and the unanswerable ones already priced. That is a considerably stronger position than being taken through it by somebody else, and it costs two days.

This case study has one organising law, set out on the entity page. The ability to see what something costs the patient falls as you move up the provider scale, while control over the decision that caused it rises. The two cross somewhere in the middle of any organisation of this kind, and below the crossing people can see and cannot act, while above it they can act and cannot see.

This page is that same law rotated ninety degrees. It holds across time exactly as it holds across hierarchy, and once you look at it that way a great deal of what is strange about hospital buildings becomes ordinary.

The designer has total control and no sight. They decide where the store goes and they will never once watch a nurse walk to it at four in the morning in the eleventh hour of a shift. The operator has total sight and no control. They watch it several hundred times a day for a career and cannot move it by a single metre. Neither of them is behaving badly. They are the two ends of the same axis, separated by five years instead of by five levels of hierarchy, and the separation is structural rather than personal.

What makes the time version worse than the hierarchy version is that there is no meeting. A ward sister can at least be in a room with a director. Nobody can be in a room with the person who drew this in 2027, because by the time the consequence is visible the project is closed, the team is dispersed and the drawings are in an archive. The feedback that would improve the next building has nowhere to go, which is why hospital design improves so slowly despite being practised constantly.

Follow that through to the consequence, because it is the part that shapes behaviour rather than just explaining it.

An operator is appraised on length of stay, on discharge times, on infection rates, on staff turnover and on cost per case. Every one of those is substantially determined by the eleven constants above, all of which were fixed before they were appointed. They are being measured on a performance whose ceiling somebody else set.

Three things follow, and all three are visible in real hospitals. Operators become defensive about numbers they cannot move, which reads as excuse making and is usually accurate. Nobody has any incentive to attribute a bad number to the building, because that conversation goes nowhere and makes the speaker sound like they are dodging. And so the building is never named as a cause, which means the evidence that would improve the next one is never collected, which guarantees the next one repeats it.

That is a closed loop, and it is closed in the wrong direction. Breaking it does not need a new theory. It needs somebody to record, from the first day of operation, which of the problems are constants and which are not, and to send that record somewhere it can be used.

Almost nobody can act on the previous chapter, and the reason is structural rather than a failure of will. A trust that builds a hospital once every thirty years has no next building to improve. A contractor who builds many has no operating experience of any of them. An architect gets told what went well. The two halves of the knowledge sit in different organisations and there is no commercial mechanism that joins them.

Meridian is unusual in exactly the way that matters here. It designs, it operates, and it builds again. The person who will run the building is inside the same company as the person drawing it, and the record of what the constants cost in hospital one is an asset that hospital two can be designed against. The feedback loop that is broken across the whole industry is closable inside a single firm, purely because the firm happens to contain both ends of it.

That is worth stating as the commercial point rather than the virtuous one. This group’s product is demonstrated, transferable operating performance. A register of what the last building got wrong, measured rather than remembered, is the most direct evidence of that product there could be, and it is something a competitor who builds once cannot assemble at any price.

It also gives the sixth gate its content. That gate never closes, and this is what runs through it: the constants list from every building in the group, updated, priced, and handed forward. The first site earns twice, and this page is the second earning.

The first is the one that cannot be engineered away, and it is worth reading before anybody concludes that enough modelling would have prevented all of this.

A building can be correct when it opens and wrong thirty years later, through no error by anybody, because medicine moved. Day surgery, imaging and length of stay have already stranded buildings designed by careful people who were right at the time. The conclusion is not to predict harder. Prediction at that horizon is not available.

The conclusion is to buy optionality instead: structure with spare capacity in it, risers larger than today needs, land held back, and rooms whose function is not welded to one clinical model. Every one of those costs money at the start and shows no return for years, which puts it with the beds, the people, the fuel and the medicines, and makes it the fifth appearance of the sentence this study keeps arriving at.