Before Buying a LINAC: Why the Purchase Order Comes Eighth

What actually destroys the money

Three things destroy the return on a radiotherapy investment, and the price of the machine is not one of them.

Idle capacity — a machine planned for a patient load that never arrives, or that arrives and then leaves halfway through treatment. A stalled clock — capital committed and interest running while permission to operate waits on drawings that were started too late. An irreversible siting decision — a bunker placed where the department can never grow and the next machine can never be delivered.

All three are settled before the purchase order is signed. In a properly sequenced radiotherapy project the purchase order is the eighth document, and seven decisions precede it. This article covers the four that most often go wrong; the full sequence is set out at the end.

All three also tend to arrive by inheritance rather than by choice, because the plan is modelled on a cancer centre the promoter has seen and admired — built somewhere else, for conditions that are not present here.

A metropolitan cancer centre is a highly optimised building. It is worth being precise about what it is optimised for.

Reverse the conditions and the building changes — not in size, in kind. A plan carried across that line solves problems the site does not have and ignores the ones it does.

One: what actually fills a machine

Capital consequence — revenue stability.

Most business plans for a cancer centre begin with catchment population. Radiotherapy does not work that way. Two districts with identical populations, identical cancer incidence and identical referral networks can produce completely different machine utilisation, because a linear accelerator is filled by completed courses, not by diagnosed patients.

Start with the ceiling. International planning practice sizes a machine at approximately 450 patients per machine per year. A curative course averages around 22 fractions and a palliative course around 4, giving roughly 18 fractions for an average first course; a machine treats four to five patients an hour, so capacity depends on the hours it runs. That is about 8,100 fractions a year — roughly 32 fractions a day across a normal working year.

The figure holds from the other direction. Departments tracking throughput over two decades report a decline from 40 to 32 patients per eight-hour shift as image guidance and advanced delivery techniques increased complexity. Two methods, one answer. Which makes it a test: any projection materially above this describes a department that does not exist.

Three corrections before it enters a business plan.

Case-mix. Head and neck and cervical cancers — which dominate the burden across much of northern and central India — carry long fractionation schedules. A programme weighted toward them consumes machine hours faster than the benchmark assumes.

Headroom. Utilisation should sit 5–15% below maximum capacity, or random fluctuation in referrals produces unacceptable delay. A department booked to capacity is a department with a waiting list and referral leakage.

Completion. This is the correction nobody models, and in a regional market it decides the number.

Capacity planning assumes a patient who starts a course finishes it. Treatment is delivered in daily fractions over several weeks, which strains populations living far from a facility, who may receive suboptimal treatment or none at all. In India specifically, patients from smaller towns often travel hundreds of kilometres for each session, leading to dropout from treatment cycles. The geography is measurable: about 76% of the world’s population lives within two hours’ travel of a radiotherapy facility, falling to roughly 17% in low-income countries. And ICMR’s national assessment treats inequitable distribution of radiotherapy services as being as critical as the raw equipment shortfall.

A machine treating sixty per cent of its planned fractions is usually read as a referral problem. In a referral geography of this kind it is more often an architectural one: the facility was designed around six weeks of machine operation rather than six weeks of a patient’s life.

What that costs, in relationships

No projections below — only the arithmetic of the figures already stated.

Two hospitals buy the same machine, hire the same oncologist, obtain the same approvals. Five years on, one runs near capacity with a second bunker under construction; the other sits at half utilisation with nowhere to expand. The machine did not decide that. The plan did — through who finished treatment, how fast patients moved, whether staff stayed, whether downtime could be absorbed, and whether growth had anywhere to go.

Two: the two approvals that precede permission to buy

Capital consequence — time to revenue.

Under the eLORA framework, AERB requires RSO approval in order to grant procurement permission for radiotherapy equipment at a new facility. Separately, site and layout approval is issued on submitted drawings — and a separate application is required for each individual installation: teletherapy, brachytherapy and simulator each on their own. Construction must then be carried out in accordance with the AERB-approved plan, and any deviation must be reported.

Three consequences.

The physicist and the drawing are pre-purchase conditions, not commissioning tasks. Without both, the hospital does not have permission to buy.

There is no type approval for a centre format. For a group building several sites this is the constraint that sets rollout speed. Approvals do not amortise across a network — they scale linearly with expansion. Ten centres with a simulator each is twenty separate approvals.

So the only compressible variable is design repeatability. What can be standardised is the chassis: a validated department module with settled structural, services and access parameters, adapted per site rather than redrawn per site. What cannot be standardised is precisely what AERB reviews — the site-specific geometry and the occupancies surrounding the treatment room. Knowing which is which is the difference between a rollout and a queue.

For a single hospital the same gates apply once. The sequence does not change; only the multiplier does.

Bunker geometry, shielding sequence and the approval timeline as lived on a built project are covered in LINAC bunker design in India and appointing a LINAC architect.

Three: the scarce resource is no longer capital

Capital consequence — operational continuity.

A hospital can order a machine in a week. It cannot produce a medical physicist in a year.

Eligibility to hold the RSO role in a radiotherapy department runs through a narrow national pipeline. BARC’s post-M.Sc. Diploma in Radiological Physics admits around 30 candidates a year; recognised university M.Sc. Medical Physics programmes admit roughly 15–25 each. A 12-month internship in a recognised radiation therapy department is required on top of the degree. Pass rates in the RSO certification examination run below 30% at most institutions outside BARC’s programme. Approval is renewable, not permanent. And every new centre requires its own approved RSO before it may purchase equipment.

The conventional planning order is money, then building, then staff. Indian radiotherapy has inverted it: people, then building, then money. No amount of capital increases the number of people the national system produces each year, which means recruitment belongs in the feasibility study rather than the commissioning plan. It is not the vendor’s constraint, so it does not appear in the vendor’s proposal.

Architecture cannot solve a national shortage. It can make a department operable by a thinner on-site team — short distances between simulation, planning and treatment; physics space that is usable rather than notional; remote planning designed in rather than retrofitted. A physicist weighing an offer in a smaller city is also weighing whether the department will be workable.

Four: what the drawing decides that you think the budget decides

Capital consequence — capital efficiency, and everything that cannot be undone.

AERB’s guidance on preparing radiotherapy layout drawings carries an instruction most promoters never see. The applicant must indicate the occupancy around the treatment room — and must name it specifically rather than describing it as full or partial: Ward, Toilet, Corridor, Accounts Office.

Those room names are an input to the shielding assessment. The shielding assessment drives the largest single civil cost in the project. How shielding then propagates through the rest of the building — penetrations, adjacencies, workflow — is a subject of its own, covered elsewhere in this publication.

The same guidance directs that room and site layout drawings be prepared in consultation with expert medical physicists, radiation oncologists, architects and the supplier of the unit. The regulator has already written the architect into the process. But the stronger argument is one of sequence:

Architecture is the only discipline in the project whose decisions become physically irreversible before the first rupee of clinical revenue is earned. The vendor can be changed. The planning software can be upgraded. The physicist can be recruited. The machine itself will be replaced inside a working lifetime. The position of the bunker, the expansion axis and the service approach are fixed before the first patient is treated, and permanent after.

A board should spend its attention in proportion to irreversibility, not in proportion to invoice value.

A board believes it is approving a civil budget. It is approving a drawing decision whose cost was fixed months earlier.

Who decides what

The supplier optimises the machine. The architect optimises the building the machine will sit inside for the next forty years.

Beyond India

Secondary cities across East Africa and the Gulf present the same conditions this article describes — patients spread across long distances, thin specialist supply, single-machine dependency, and national capacity concentrated in the primary metropolitan centres. The access gap is starkest in low-income settings, where the share of population within two hours of a radiotherapy facility falls to roughly one in six. IAEA guidance on establishing radiotherapy services addresses this condition directly, and its logic — facility, staffing and regulatory readiness planned together rather than in sequence — is the logic AERB enforces domestically.

The name of the regulator changes. The order does not.

The corrected sequence

1. Establish clinical intent and case-mix
2. Test capacity against completion, not catchment
3. Test the site — position, access, expansion, replacement route
4. Appoint architect and medical physicist
5. Obtain RSO approval
6. Obtain site and layout approval, per installation
7. Obtain procurement permission
8. Issue the purchase order
9. Construct in accordance with the approved plan
10. Commission, survey, licence, treat

Board checklist: before the purchase order

Do not issue a purchase order until the board can answer yes to each of the following.

If the answer is no

If the honest answer to any question on that list is no, the correct response is not to delay procurement. It is to delay the project.

A purchase order signed before the sequence is resolved does not advance anything. It converts capital into commitment, and buys nothing that could not have been bought later on better terms.

Studio Athenos designed the AERB-licensed LINAC facility at Jeevan Raksha Complete Cancer Care Centre, Bikaner, alongside its PET-CT and gamma camera departments — planned as one comprehensive cancer centre from the first site sketch.