How to Analyse a Hospital Stock (ARPOB, Occupancy)

In the December 2024 quarter, Narayana Hrudayalaya's flagship Bangalore Health City earned about ₹2.31 lakh per inpatient, against roughly ₹1.20 lakh per patient in its other Indian regions — nearly double, for the same company, in the same country (NH Q1 FY26 concall, "Bangalore's Health City… drives higher inpatient average revenue (INR 231K vs. ~INR 120K in other regions)"). Same surgeons, similar buildings, wildly different economics. The gap is not a quality difference; it is a case-mix and pricing difference — more cardiac and oncology work, more private payors, a wealthier catchment. (Illustration of how to read the numbers, not a view on the stock.)

That single contrast is the whole sector in miniature. A hospital does not sell beds; it sells what happens in the bed. Two hospital chains can each run 70% occupancy and look identical on a screener, and one earns twice as much per bed as the other. The headline that retail investors fixate on — revenue growth, or even occupancy — tells you almost nothing until you decompose it into the few operating metrics that actually decide the margin. Across the concalls Inve tracks, hospital managements guide confidently on bed additions and ARPOB, then quietly let completion dates slip and margins compress while new units fill — and the income statement hides the drag for a year or two.

This is how to read a hospital the way an operator does: the handful of numbers that decide the outcome, where they hide (mostly not on the income statement), and the one red flag that has cost hospital investors more than any bad debt ever could — the new-hospital drag they didn't price.

A boundary first. You will not audit clinical quality or case mix from the outside. What you can do is read whether ARPOB, occupancy and the maturity ladder are moving the way management's account says they are — and whether the guidance survives the next four calls.

What actually drives a hospital's economics?

Strip a hospital chain down and it is a real-estate-plus-talent business with brutal operating leverage. The big costs — the building, the equipment, the senior consultants — are largely fixed once the hospital is open. So profit is overwhelmingly a function of two things: how full the beds are (occupancy) and how much each occupied bed earns (ARPOB). Push either up against a fixed cost base and EBITDA compounds; let either sag and the same base eats you alive.

That framing fixes two beginner errors at once. First, a new hospital is a loss-making machine for its first two to four years — it carries the full fixed cost from day one but fills slowly, so it drags group margin exactly when revenue growth looks most exciting. Apollo, India's largest chain, told analysts it expects its bundle of new hospitals to run an EBITDA loss of about ₹150 crore through the ramp, and that a new hospital sits around 40% occupancy in its first year (Apollo Hospitals Q2 FY26 and Q3 FY26 concalls). Second, revenue growth without ARPOB growth is just more beds, not a better business — the quality of growth lives in price and case mix, not the top line.

The homely version: a hospital is a hotel that also does heart surgery. The hotel half (occupancy) keeps the lights on; the surgery half (ARPOB) is where the money is. A chain growing rooms while the surgery mix stays thin is adding hotel, not adding margin.

The metrics that matter — and where they hide

Here is the spine. For each, what it is, why it matters here, where to find it, what good looks like, and a real number.

ARPOB — average revenue per occupied bed

ARPOB is daily (or annual) revenue divided by occupied beds. It is the single best proxy for case mix and pricing power — a hospital doing more complex surgery on better-paying patients earns more per bed without adding a single bed.

Where it hides: almost never on the income statement. ARPOB lives in the investor presentation and the concall, quarter by quarter, and chains define it differently (per-day vs annualised, network vs cluster), so never compare two companies' ARPOB blindly — compare each against its own trend. Max Healthcare runs a network ARPOB of about ₹77,900 per bed per day (MAXHEALTH Q3 FY26, "ARPOB Q3 FY26 ₹77,900"); Apollo's was ₹60,839, up 8% year-on-year (Apollo Hospitals Q3 FY25 concall, "ARPOB increased 8% year-on-year to INR 60,839"). Fortis reports it annualised — about ₹2.56 crore per bed per year in Q3 FY26 (FORTIS Q3 FY26, Inve data). Same metric, three units; the comparison that matters is each one's own slope.

What good looks like: mid-to-high single-digit ARPOB growth, driven by case mix (more cardiac/oncology/transplant), not just a list-price hike. KIMS is the cautionary version — consolidated ARPOB jumped over 20% year-on-year, but that was mostly mix shift as it folded in higher-priced units, not the same hospital pricing up. Its mature Andhra cluster runs around ₹23,500 while mature Telangana runs ₹70,000 (KIMS Q4 FY25 and Q3 FY26, Inve data) — a 3x spread inside one company, which tells you "consolidated ARPOB" can move for reasons that have nothing to do with the business getting better.

Occupancy — how full the beds are

Occupancy is occupied bed-days over available bed-days. It is the operating-leverage dial: against a fixed cost base, the marginal patient is enormously profitable.

Where it hides: investor PPT and concall, occasionally the press release. Mature Indian metro hospitals cluster around 65–75% — Max ran network occupancy of about 75% (MAXHEALTH Q4 FY25), Apollo group-wide around 67% (Apollo Hospitals Q3 FY26), Fortis around 67–71% (FORTIS, FY26 quarters, Inve data).

The trap: occupancy looks low precisely when a chain is adding capacity — KIMS reported census occupancy around 50–52% (KIMS Q4 FY25 to Q3 FY26, Inve data) not because it is failing but because it just opened large new units running near-empty. So read occupancy together with the maturity ladder, never alone. A falling group occupancy can mean trouble or a wave of new beds; only the split tells you which.

Bed additions and capacity — the growth engine and the drag

Bed count is the chain's installed base; gross additions are its growth pipeline. This is the one metric that is easy to find — management trumpets it. The discipline is treating every announced bed as a future cost that arrives before its revenue.

Where the real signal hides: in the completion dates, which slip constantly. This is exactly what Inve's Promise Tracker is built to catch. Max Healthcare guided at its Q3 FY25 call that three large projects — Nanavati Phase 1 (268 beds), Mohali (155 beds) and Max Smart at Saket (400 beds) — would complete by Q1 FY26 (MAXHEALTH Q3 FY25 concall). In Inve's tracker, all three are marked delayed (Inve data) — bed guidance that quietly went past its date. A bed announced is a cost you can bank; a bed delivered on time is the rarer thing.

Mature vs new hospitals — the metric that explains the margin

This is the one that separates investors who understand hospitals from those who don't. A chain's reported margin is a blend of mature hospitals (often 25%+ EBITDA margins) and new ones (loss-making). When a chain is expanding hard, the blend falls even as every individual hospital improves.

Where it hides: the segment split in the PPT, and direct concall questions. Apollo splits it explicitly: established hospitals targeting a further 500 bps of margin while new hospitals run that ~₹150 crore loss (Apollo Hospitals Q2 FY26 concall, "the internal target is to take it higher by 500 basis points" and "EBITDA losses from these hospitals should be around the INR 150 crore number"). KIMS quantified its drag precisely — new units contributed an EBITDA loss of ₹18.31 crore in FY25 (KIMS Q4 FY25 concall, "contributed an EBITDA loss of ₹18.31 crores in FY25"). Without that split you cannot tell a chain whose mature business is weakening from one whose mature business is fine and is simply paying the entry cost of growth.

EBITDA per bed — the productivity number

EBITDA per occupied bed is ARPOB and occupancy and cost control rolled into one — the truest single measure of how hard a bed works. Max Healthcare reported annualised EBITDA per bed of about ₹82.6 lakh for its existing units, rising to ₹84 lakh the next quarter (MAXHEALTH Q3 FY25 and Q4 FY25 concalls). That is the metric to track over years: a mature chain should grind it upward; a stalling number under rising ARPOB means costs are leaking.

ALOS — average length of stay

ALOS is how many days the average patient stays. Counter-intuitively, shorter is usually better — a hospital that discharges in 4 days instead of 6 turns its beds over faster, lifting effective throughput without adding a single bed. NH runs an ALOS of about 4.3 days (NH Q1 FY26 concall, "ALOS: 4.3 days"), among the leanest in Indian hospitals, which is part of why it earns strong returns on a relatively asset-light footprint. Watch it for deterioration: a creeping ALOS quietly steals the capacity a chain is spending crores to build.

Payor mix — the quality of the rupee

Who pays — cash, private insurance (TPA), or government scheme (CGHS, Ayushman, state schemes) — decides both the price and the cash conversion. Cash and private-insurance patients pay more and pay sooner; government schemes pay less, slower, and sometimes get re-rated downward. NH gets roughly 19–20% of revenue from government payors (NH Q4 FY25 concall, "19 to 20% of our revenue comes from government payors") and is candid that it is a lower-margin, working-capital-heavy slice. The risk shows up as receivable days: Fortis flagged government receivable days of about 180 (FORTIS Q2 FY26, Inve data) — half a year of waiting for money already booked as revenue. A chain quietly leaning harder on government volume to fill beds is buying occupancy with margin and cash flow.

How do you value a hospital — and which multiple is right?

Forget P/E first. A hospital chain mid-expansion has a depressed blended margin and a depressed near-term profit, because the new-hospital losses and the fresh depreciation hit the P&L before the revenue ramps. P/E on that trough earnings looks absurdly high and tells you nothing. The two lenses that work:

EV/EBITDA, because it is capital-structure-neutral and sees through the new-hospital drag better than net profit does — but only if you ask whose EBITDA. The honest exercise is to value the mature book at a mature multiple and treat the new hospitals as an option that is currently costing you money. A chain trading at 25x blended EV/EBITDA might be far cheaper on its mature estate alone, with the new beds thrown in near-free — or far dearer, if the "ramp" never ramps.

EV per bed, the hospital equivalent of EV per subscriber — it strips out the maturity blend entirely and asks what the market is paying for installed capacity. A premium metro chain commanding high ARPOB deserves a higher EV/bed than a tier-2 chain; the question is whether the gap the market is pricing matches the ARPOB and occupancy gap the operations actually show. Max Healthcare's roughly ₹82–84 lakh EBITDA per bed versus a tier-2 chain's far lower figure is precisely why their EV/bed should not be the same — and why a flat EV/bed comparison across a metro chain and a regional one is a category error.

The owner's framing: you are buying a portfolio of mature hospitals throwing off cash, plus a pipeline of new ones consuming it. Price the two separately, and the "expensive" hospital often turns out to be the one whose mature engine you got at a fair price.

A worked case: Max Healthcare, said vs did

Take Max Healthcare through FY25 into FY26 — a genuinely well-run chain, which is why it teaches better than a weak one. Its consolidated operating margin compressed from the low-30s in late FY25 toward the high-20s through FY26 (MAXHEALTH financialreport: ~33–34% operating margin Q2/Q3 FY25 falling to 27–29% across FY26, Inve data). A reader watching only that line would conclude the business was deteriorating.

The operating metrics say the opposite. Existing-unit ARPOB kept growing ~7% (MAXHEALTH Q3/Q4 FY25, Inve data), network ARPOB rose from ₹75,900 to ₹77,900 (MAXHEALTH Q3 FY25 to Q3 FY26), occupancy held near 75% (MAXHEALTH Q4 FY25), and existing-unit EBITDA per bed rose to ₹84 lakh (MAXHEALTH Q4 FY25 concall). The mature book got better. What pulled the blended margin down was new capacity: Max-Dwarka alone reported ₹59 crore of revenue against a ₹5 crore EBITDA loss in one quarter (MAXHEALTH Q3 FY25 concall, "reporting Rs. 59 crores in revenue and an EBITDA loss of Rs. 5 crores for the quarter"), and management said a new hospital runs 5–6% below mature margin until it matures (MAXHEALTH Q3 FY25 concall, "it should be lesser by 5-6%").

Now the said-vs-did, the part you only get by reading a sequence of calls. At Q3 FY25 Max guided three major projects to complete by Q1 FY26; Inve's Promise Tracker marks all three delayed (Inve data). An institutional price increase guided "within a month or two" is marked missed (Inve data). None of this makes Max a bad operator — but it is the precise mechanism by which the drag lasts longer than the headline guidance implied: beds slip, losses persist an extra quarter or two, and the blended margin stays soft while the screen-watcher gives up. (Illustration, not a view on the stock; a read on how management communicated through one expansion cycle, not a lifetime verdict.)

Red flags specific to hospitals

• A "stable" group margin during heavy expansion. New hospitals must drag the blend. If a rapidly-expanding chain reports a flat or rising blended margin, either the new beds aren't really new (recycled capacity) or something in the mature book is being flattered. Ask for the mature-vs-new split; vagueness is the answer.
• Occupancy bought with payor mix. Occupancy climbing while ARPOB stalls and government/scheme share rises is a chain filling beds with low-yield, slow-paying patients. Check receivable days alongside — 180 government receivable days is half a year of revenue you've booked but not collected.
• Bed guidance that keeps slipping. One delayed project is operations; a pattern of completion dates moving right (and the EBITDA-breakeven date moving with them) is a management that systematically under-prices its own ramp — and the new-hospital drag you modelled for two years runs for four.
• ARPOB growth that's all price, no mix. A chain leaning on list-price hikes rather than richer case mix is borrowing growth from patient affordability and insurer tolerance, both of which have limits.
• ALOS creeping up unremarked. It silently consumes the capacity the chain is spending crores to add — the cheapest bed expansion is a shorter stay, and the most expensive miss is a longer one nobody flagged.

Frequently asked questions

The discipline comes down to refusing to be impressed by the blended number. A hospital's reported margin is a weighted average of a cash machine and a cash furnace, and the whole craft is separating the two before you decide what you're paying for. So invert the question you bring to the results. Don't ask "is the margin good?" Ask: if this management were quietly funding tomorrow's empty new beds with today's mature-hospital profit and calling the blend "steady," what would the numbers look like — and does the mature-vs-new split rule that out? A flat group margin during heavy expansion does not rule it out; it is the thing to interrogate.

Where this lens can be wrong. The strongest case against everything above is that a great new hospital in a great location can ramp faster than the two-to-four-year rule, in which case the cautious investor who waited for the drag to clear paid up for proof they could have underwritten earlier. NH's Bangalore unit and Max's mature Saket complex earn what they earn because someone built ahead of demand and absorbed the early losses — read every new hospital as a guaranteed multi-year loss and you will systematically underpay the best operators and miss them entirely. The honest claim is narrower: the maturity split tells you why a margin is soft and whether the softness is the price of growth or the symptom of decay. It does not tell you which new hospitals will ramp — that is a judgement about location, clinical talent and catchment that no transcript settles. And we have not modelled clinical quality, regulatory price caps on scheme patients, or doctor-retention risk, any one of which can break a hospital that screens beautifully.

And the owner's question to sit with before buying a single share of any hospital chain: five years out, when today's loss-making new beds are full, what ARPOB and occupancy must they reach for this to have been worth the wait — and is that ramp something the mature hospitals already prove the chain can do, or something you're taking on faith? If the answer leans on management's confidence rather than the maturity ladder's own record, you've read the headline, not the business.

For the lender's version of this same discipline — reading a blended number that hides a slower-forming problem — see how to analyse an NBFC.

Related reading

• How to Analyse a Bank Stock (NIM, CASA, GNPA, ROA)
• How to Analyse a Building Products Company (Pipes, Tiles)
• How to Analyse a Cables and Wires Stock