Every major medical-trend survey tells the same story: healthcare costs are rising faster than general inflation, and in international private medical insurance the pressure is acute. When the cost of care climbs year on year and claims outpace premium growth, the loss ratio does the rest of the talking.
The instinct is to reach for the two familiar levers: put premiums up, and squeeze the network for better rates. Both matter. Neither is enough on its own.
Why medical inflation is not ordinary inflation
General inflation is largely a price story. Medical inflation is a price story and a volume story, and the volume half is the part that compounds.
Prices rise for the familiar reasons: staff costs, facility costs, the price of new drugs and devices. But utilisation rises alongside them. New diagnostics create new findings, which create new pathways. Care that was inpatient becomes day-case, which lowers unit cost and raises volume. Populations covered by international policies are getting older and carrying more chronic disease. Provider markets that serve insured foreign patients have every commercial incentive to admit, investigate and extend, and very little counterweight.
That is why a book can be on excellent rates and still deteriorate. You are being charged less per unit for steadily more units.
Premium increases have a ceiling
Raise premiums too far and you lose the account. Regulators, brokers and corporate buyers are all watching affordability, and in a competitive IPMI market the member simply moves. Premium is a blunt instrument, and it treats the symptom rather than the cause.
Network discounts are the floor, not the ceiling
A negotiated rate lowers the price of a given treatment. It does nothing about whether that treatment was necessary, whether the length of stay was justified, or whether the bill was itemised fairly. Once a book is on good rates, the marginal saving from pushing them further is small. The market has largely accepted that discounts alone are no longer sufficient.
The three levers, honestly ranked
There are really only three things an insurer can do about a deteriorating loss ratio, and it is worth being clear-eyed about what each is capable of.
Price. Fastest to implement, immediate effect, and self-limiting. Push it and the good risks leave first, which worsens the book you are left with. It also does nothing to the underlying trend, so you are back next year.
Network rates. Genuine value, and every book should do it. But it is a one-off gain on each renegotiation, it works only on price, and once you are on good terms the marginal return on pushing harder is small. It is a floor, not a strategy.
Clinical challenge on volume. Slower to stand up, requires clinicians rather than administrators, and is the only lever that touches the half of medical inflation that compounds. It also improves as it runs, because the data on what your book actually spends money on gets better.
Most books over-index on the first two because they are easier to buy. The third is the one that changes the trend line.
The underused lever is clinical
The larger, more durable savings sit upstream of the invoice, in the question of whether the care was appropriate in the first place. That is a clinical question, and answering it means having clinicians who can engage the treating team on medical necessity, level of care and treatment pathway, not administrators applying a fee schedule after the fact.
Framed properly, this is not about withholding care. It is about the right care, in the right place, at the right time, which is usually both the most clinically defensible option and the most cost-effective one.
Where the trend concentrates
Medical inflation is not evenly distributed, and treating it as a single number hides where the money actually goes. On an international book the pressure concentrates in a few places:
- High-tariff private markets. The United States above all, but also the Gulf, where a single extended admission can move a small book’s numbers on its own.
- Elective and medical tourism complications, which arrive as acute surgical problems with no records and an unclear coverage position.
- Long-stay overseas admissions, where every additional day carries pharmacy, nursing and investigation costs that never appear in a rate negotiation.
- Repatriation decisions made by reflex, where a dedicated air ambulance is chartered for a patient a stretcher and a nurse would have carried safely.
Each of those is a clinical decision with a price attached, which is the whole argument. A blended inflation figure tells you the temperature. It does not tell you which room is on fire.
Make it defensible
None of this works if it cannot be evidenced. When an insurer or regulator asks why a decision was made, the answer has to be clinical, documented and auditable. That is the standard we hold on every case: the reasoning is recorded as the case progresses through Atlas, so the loss-ratio benefit and the clinical rationale are one and the same record.
What to measure instead of a percentage
If you are trying to work out whether any of this is working, the blended saving figure is close to useless. A percentage off billed charges means very little when the billed charges were never a credible starting point.
More honest measures:
- Average length of stay on comparable admissions, tracked over time. It is the single cleanest proxy for whether anyone is driving the pathway.
- Proportion of cases seen clinically before authorisation rather than after discharge. This is a leading indicator; the savings follow it by a quarter or two.
- Case-level saving with an attached clinical rationale, so the number is traceable to a decision a named clinician made and can defend.
- Escalation-to-review time, because on the cases that matter the window in which cost is still influenceable is measured in hours.
Medical inflation is not going away. The insurers who weather it best will be the ones who pair their network rates with genuine clinical governance over what care happens at all, and who move the clinical conversation before the bill rather than after it.
