Buy the biggest rectangle, not the biggest area

Every other decision in a hospital programme is reversible at some price. The equipment can be replaced. The layout can be refitted, expensively. The staffing model, the service mix, the brand and the operating policy can all be changed by people who arrive later and find them wanting.

The parcel cannot. There is no version of being wrong about the land that is recoverable, which makes it the only genuinely irreversible purchase in the programme.

And it is bought first. Before the massing exists, before the service mix is settled, before anybody has run the hospital against anything, and usually under time pressure from an option that expires or a vendor with another interested party. The information available at the moment of maximum consequence is at its minimum, and that ordering is not a failing of this project. It is how development works everywhere.

Which produces the whole strategy of this page, and it is one move rather than many. If the decision cannot be made later, and cannot be made well now, then what you buy first is not land. It is time.

Start with the part where a group whose only hospital is in London will have the strongest and most wrong instincts.

Nigeria has no freehold in the sense a British developer means it. Under the Land Use Act of 1978 all land in a state is vested in the Governor, held in trust, and what is acquired is a right of occupancy, normally for ninety nine years, evidenced by a Certificate of Occupancy. That is not a technicality about wording. It changes what is being bought, how long it lasts against a sixty year asset, and who has to agree before it can be transferred at all.

Because transfer is a separate permission. A holder cannot assign, mortgage or transfer without the Governor’s consent obtained first, and without that perfection the transaction can be void, which means money paid and no legal title held, and no lender willing to take the asset as security.

Then the layer underneath, which is where most of the losses actually happen. Much Lagos land was compulsorily acquired by the state decades ago. Where the state has released parts of it back to the original communities, that release is an excision, formally published in the state gazette. Land sold as having an excision in process, or with a file number, has nothing. A file number records that somebody applied. The application can be refused.

The table sets out all eight strands separately, because a parcel can be excellent on seven and worthless on the eighth, and the last column is the part to act on rather than to read.

Before shape, quantity, because the quantity is routinely underestimated and the underestimate is discovered at massing when the option has already been exercised.

A hospital is not a building on a plot. It is a building, plus a plant compound with fuel separation distances, plus parking for a population who mostly cannot arrive any other way, plus ambulance turning and a service yard, and then whatever is being held back so the place can still be right in year twenty.

Run those together and the answer is several hectares for a modest hospital, most of which is not the hospital.

One thing is deliberately absent from that figure. The staffing strategy rests on a training academy, and the academy is not being planned onto this parcel. That is a decision rather than an omission, and it is defensible: a partnership with an accredited institution supplies the campus, the accreditation route and the placement capacity without the group holding land for a decade against a programme that has not started. Clinical placement happens in the hospital regardless. Classrooms, simulation and a library do not have to.

It carries one condition, and the rest of this page is why. The academy has to be an off site model, stated as such and contracted as such. What it cannot become is an on site academy that is simply deferred, because everything below about parcels, setbacks and buildable rectangles applies exactly as much to the campus as to the hospital. Room found later is room that does not exist.

Which leaves the expansion reserve as the one line in the calculation still most likely to be deleted, by a board under pressure, because it has no revenue attached to it. Deleting it is a decision to build a smaller hospital than the numbers say, taken by somebody who thinks they are economising on land.

Here is the part that almost never appears in a land appraisal, and it is the reason for the title of this page.

Four parcels of identical area, differing only in proportion. Apply the same boundary setback to each and the buildable rectangle shrinks as the shape gets thinner, which is mildly interesting. The fourth column is the one that matters, and it is not mild at all.

A hospital has one part that cannot be spread out: the hot floor. Theatres, intensive care, imaging and the emergency department want to sit on a single large horizontal plate, next to each other, on one level, because the adjacency between them is the most time critical in the building. That plate needs depth. A site that is long and thin cannot provide it at any area, because depth is bounded by the short dimension and no amount of length compensates.

When the plate will not fit, the hot floor splits across two storeys, and theatre to intensive care becomes a lift journey. Which is one of the permanent constants on this site, arriving here as a consequence of a parcel shape chosen years earlier by somebody comparing price per square metre.

So the usable measure of a parcel is not its area. It is the largest rectangle you can inscribe in it after setbacks, and on an irregular or acutely angled site that can be dramatically less than the title area, because clinical rooms are rectangular, corridors are straight, and an acute corner is landscape.

The area calculation earlier on this page answers a narrow question: how much ground does a hospital of a given size stand on. It takes the bed number as given. The real question runs the other way, because the land requirement and the bed number are not independent.

Sizing a site is the same problem as sizing everything else in this study. Expected demand, plus a deliberate spare. What makes land different from beds, people, fuel and medicines is that the spare is not stored somewhere convenient. It is a larger parcel, larger parcels are further out, and the obvious conclusion is that further out is thinner.

That conclusion is half wrong, and the half it gets wrong is the half that pays. The three streams do not share a geography.

Unplanned demand is an isochrone and nothing else. Minutes are clinical, an ambulance cannot be asked to drive further, and a site that is twenty minutes further from where people collapse has a smaller emergency catchment in the most literal sense. Returning demand is tighter still, because a journey made three times a week for years is not measured the way a journey made once is, and a dialysis patient will change hospital over twenty minutes.

Planned demand is not an isochrone at all. This is the segment that flies to London, Delhi and Dubai. A person who boards a six hour flight with a parent is not going to be deterred by an extra twenty five minutes on a Lagos road. For that segment distance has already been demonstrated not to bind, and what does bind is trust.

So the table splits the demand accordingly. Only the local term decays with distance. The elective pool, which is the outbound money this whole proposition exists to capture, stays where it is wherever the hospital is put.

Read down the bed column and the fall is gentle rather than catastrophic. Read the proportion of that demand which is elective and it climbs steadily. Going further out does not destroy the hospital. It changes what kind of hospital it is.

Demand is one arm and cost is the other, and the second is where the correction above really shows.

The land bill barely moves across the whole range, because price per acre falls faster than the requirement rises. Land cost is not what is being traded here, which is worth knowing before anybody spends a board meeting on it.

What does move sharply is what a tight site does to the building. A parcel too small for the hospital its catchment justifies does not refuse to accommodate it; it accommodates it expensively, through a basement that was going to be a choice, piling, tanking, decked parking and more storeys than the journey optimum wants. That premium is large at the tight end and disappears once the site can hold what it needs to hold.

Put both arms together and the last column is close to flat from the middle of the range outward. It falls steeply as the site stops strangling the building, and then it stops falling and stays there, because the demand that would have been lost to distance is mostly demand that does not measure distance.

That is the real finding on this page and it is not the one the exercise set out to produce. A large peripheral parcel is not the expensive mistake the naive model made it look like. The capital per bed is much the same, the land bill is much the same, and what changes is the composition of the hospital rather than its economics.

Which moves the decision somewhere else entirely. If cost does not decide it and distance only decides part of it, then the land question is not really a land question. It is a question about what kind of hospital this is intended to be, and that has to be answered before an acreage means anything.

The reason shape is decisive is that a hospital is five buildings with incompatible requirements, sharing one parcel and pretending to be a single thing.

Read the first two together, because they contain the whole problem. The hot floor wants depth and does not want perimeter. The ward block wants perimeter and cannot use depth. A form that serves one starves the other, and every hospital typology that has ever worked is an answer to that single tension.

Every hospital plan ever drawn is a position on a single axis, and once that is visible the choice between them stops being a matter of taste. Compactness buys short journeys. Perimeter buys daylight, natural ventilation and somewhere to expand into. You cannot have both, and each form is a different compromise between them.

Five forms at roughly the same floor plate, with the two numbers that matter computed from the geometry rather than asserted. The square gives the shortest journeys and starves the middle of the building of any facade at all. The slab gives every room a window and produces a corridor a third of a kilometre long, which is the walking distance argument on this site arriving as a plan decision.

The ring is the interesting one and the arithmetic explains why this form keeps reappearing in hospitals for a century. It has almost exactly the facade of the slab, because it has an outside and a courtyard, and it halves the worst journey, because a loop can be walked in either direction and a dead end cannot. That is not a stylistic preference. It is two ways round.

The cruciform does well on both measures and pays for it elsewhere: everything passes through one core, which becomes the bottleneck, the wings are short so departments are small, and lengthening an arm to expand makes every journey worse. The comb has the most perimeter of any form and the most walking, which is the trade it exists to make.

Two things the table cannot show. The ring is a closed figure, so it is the hardest of the five to extend, and a ring hospital that needs a bigger theatre suite in year fifteen has nowhere to put it. And the comb is the easiest, because you add another tooth, which is why it became the standard form for hospitals that expected to grow.

Now apply the power argument to the plan, because that is where the answer for this market stops matching the answer a temperate country would give.

A deep square block has no usable facade in the middle. Everything inside it is mechanically ventilated, mechanically cooled and artificially lit, continuously, forever. That is a perfectly sound arrangement in a place with a reliable grid. In a place without one, it means the building is uninhabitable within hours of losing power, and the hospital has converted its plan form into a permanent dependency on the standby plant.

A shallow form with openable windows fails differently. When the power goes it becomes hot, dim and uncomfortable, and it keeps working. That is the whole difference, and it is the same distinction the automation page draws between a frequent small failure and a rare total one. The deep plan is more efficient on every ordinary day and it fails completely; the shallow plan is worse on every ordinary day and degrades.

Which does not resolve to a single form, because a hospital is not one building. The hot floor has no choice: theatres and intensive care need controlled air, sealed envelopes and heavy servicing, so that part is deep, mechanical and on the essential supply whatever anybody would prefer. It is a small proportion of the floor area and it is where the generator argument is genuinely unavoidable.

The wards are the opposite, and they are most of the building. A ward block that can be naturally ventilated when it has to be is a resilience decision taken at plan stage, for nothing, and it is quietly given away every time a ward is drawn deeper than about fifteen metres because deep is cheaper per square metre of envelope.

So the form for this market is a deep, heavily serviced hot floor with shallow ward accommodation over or beside it, arranged as rings or combs rather than as deep towers, oriented with the long facades north and south, and shaded. That is a specific enough answer to be wrong, which is the point of writing it down before the parcel is chosen rather than after.

Which produces a small number of moves, and they are worth knowing before the parcel is chosen rather than after, because most of them impose a requirement on the land.

The first is the one the previous chapters argue for: a deep podium with shallow blocks above or beside it. Put the hot floor in a low, deep plate that uses the middle of the site where daylight is worthless and where mechanical servicing is unavoidable anyway, and put the wards in a shallow ring or comb where every bed reaches a window and the building survives a power cut. It resolves the tension rather than splitting the difference, and it needs a parcel deep enough for the plate, which is the constraint the table above measures.

The second is courtyards, which is what you do when the site is too small to spread out. A courtyard pushes daylight and air into the middle of a deep plan and creates perimeter where there was none, and it is the traditional answer in hot climates for good reason. It costs footprint, so it is a trade rather than a free move.

The third is orientation, and in this climate it is worth real money. Run the long axis roughly east to west and the big facades face north and south, where the sun is high and a modest overhang shades them. Run it north to south and the long facades face east and west, where the sun is low, shading barely works and the cooling load is permanent. The parcel proportions largely decide which of those is available, which makes site orientation an energy decision taken by a land agent.

The fourth is the growth axis. A hospital does not expand evenly. It expands where the pressure is, usually theatres, critical care and imaging, all of which sit in the hot floor. So the massing has to leave one side of that plate clear and keep the expansion land on that side. Reserving land is not sufficient; it has to be reserved in the right direction, and putting the ward block across the growth axis is one of the more common and less recoverable planning errors.

And the fifth is to read a slope as a gift rather than a cost. A site that falls lets you have two different ground levels, so ambulances arrive at one and goods at another, both at grade, without ramps or a lift. On a flat site that separation costs a floor. It is the one respect in which the awkward parcel beats the easy one.

Large contiguous parcels do not exist in the dense core of this city at a price any hospital can pay, so specifying an area specifies, implicitly, how far out the site will be. That much was always true. What changes is what the implication costs.

At the tight, central end the hospital is a general one. It has a substantial emergency department, a meaningful local population walking and being driven to it, cyclical treatment that people can actually get to three times a week, and an elective service alongside all of that. It is more building per acre, more storeys, a compulsory basement and a parking structure, and it is expensive to construct for exactly those reasons.

At the large, peripheral end the hospital is a destination. Most of its work is planned, most of its patients travel to it deliberately, and its emergency and cyclical roles are thin because the local population around it is thin. It is cheaper and simpler to build, it can be low, it has room for everything that is not clinical, and it can expand for decades.

Both are real businesses and this study is not going to pretend one of them is obviously correct. What it will say is that they are different businesses, they need different staffing, different marketing, different payer mixes and different resilience arguments, and that choosing an acreage without choosing between them means the acreage chooses for you.

It also says which way the group is already leaning without having decided to. The four specialties the demand case is built on are elective. The money it intends to capture is money that currently boards aeroplanes. The comparator it has chosen for scale is a tertiary centre people travel to. Every one of those points at the destination model, and the destination model is the one that tolerates a large peripheral site comfortably.

Follow the destination model through and it arrives somewhere uncomfortable for a property discussion, which is that the land question stops mattering very much.

If the paying patient will travel, then the site is not competing on proximity. It is not competing on price either, since the alternative is an international flight and a hotel. It is competing on one thing, which is whether a family believes this place will not harm somebody they love.

That is the finding of the family testimony on this site, arrived at from a completely different direction: the barrier keeping patients on aeroplanes is evidence and accountability rather than capability. Published volumes and outcomes. A named consultant who is guaranteed to be the one operating. A written escalation path for the bad night. Somebody who owns the case. A price fixed before consent. A route to complain.

None of those are bought with land, and none of them are bought with a better location. They are bought with years of keeping a promise publicly, which is why the helipad matters out of proportion to its transfer volumes: it is the one trust instrument that acts in months rather than years.

So the honest ordering is the reverse of the one most projects use. Decide what kind of hospital this is. If it is a destination hospital, accept that distance is a second order constraint and spend the argument on the trust instruments instead, because those are the binding ones. The parcel then follows, and it can be larger and further out than the arithmetic at the top of this page suggested, because the arithmetic at the top of this page was measuring the wrong constraint for the segment that pays.

Every other building on a street is assessed on how it performs on an ordinary day. A hospital is the one building whose value peaks precisely when everything around it is failing, and a site chosen on average conditions has been chosen against the wrong test.

The flooding question is the sharpest version and in coastal Lagos it is not hypothetical. A hospital that is cut off, or whose basement plant is underwater, is unavailable in exactly the week the city needs it most. Site levels, drainage and where the plant sits are therefore clinical decisions, and the relevant question about a parcel is not whether it floods but what happens to it in the worst event anybody has recorded, plus a margin for the fact that the record is getting worse.

The same test applied to everything else changes what you look for. Access when the main road is blocked, which means a second route rather than a better one. Fuel deliverable when the city is short. Water when the mains is not there. Staff able to reach the site when transport stops, which is an argument about where staff live and therefore about accommodation. Grid failure, which the infrastructure page treats at length and which is a daily event rather than an emergency.

None of this is resilience as a virtue. It is the core function. A hospital that works only on good days is a clinic with a large car park, and the parcel is where most of that is decided.

Everything above describes a decision that has to be made before the information exists. There is one instrument that dissolves that, and it is ordinary development practice rather than anything clever: buy an option rather than a parcel.

An option converts an irreversible decision into a reversible one for a fee. It buys a defined period at a fixed price, during which the searches, the charting, the gazette check, the ground investigation, the utility capacity confirmations, the planning position and the traffic analysis all happen, and, on this project, during which the hospital is run against that specific parcel’s constraints before a penny of purchase money moves.

That is the answer to the ordering problem stated at the top of this page. The decision still has to be taken first. It no longer has to be taken blind, because the option period is where the information arrives, and the fee is the price of moving the decision to after the information instead of before it.

It also changes the negotiation. Approach a vendor as a hospital and the price reflects it, particularly for the adjacent parcel you will need in year fifteen. Options taken quietly, before the use is visible, are cheaper than land bought afterwards by somebody with no alternative.

And it gives a test for whether the process is real. If every option is exercised, the diligence is decorative and the decisions were taken before it started. Abandoning options is what the fee is for, and the abandonment rate belongs on the same list of numbers that should be read the opposite way round to the instinct, alongside the escalation calls on the family page and the interception rate on the medicines page.

Four, and the first is the one that cannot be engineered away in this market.

The last one is the fifth appearance of the sentence this study keeps arriving at, and by now it is less a finding than a structural feature of the whole subject. Beds, people, fuel, medicines and now land. Every one of them is a buffer that looks like waste, is not waste, and survives the cost round only where somebody wrote down what it was for. The land version is the most exposed of the five, because it is the only one that shows up on a balance sheet as an asset doing nothing.