How we compute Trust Score, savings, and verification — line by line.

Three things on every clinic card and profile page deserve a plain-English explanation: the Trust Score we put next to the star rating, the "X% less than US" savings figure in the sidebar, and the "JetPatient verified" claim itself. Here is exactly what each one means.

1. Trust Score

The Trust Score is a composite quality signal computed from completed journeys in the JetPatient network. It is a number between 0 and 100 — or, when we do not yet have outcomes for a clinic, no number at all. The formula is fixed in code at assets/js/jp-trust-score.js, which is the single implementation used by both the server and the marketplace, and only changes through engineering review.

The four components

Each component is scored 0–100 and contributes at the weight below. Weights are renormalised across whichever components actually have data, so a missing component is excluded rather than filled with a guess.

• Accreditation — 25%. Driven by the rigour of the strongest body a clinic holds, not by how many certificates it has collected: JCI 85, ISQua 80, NABH 70, TEMOS/ICONTEC 65, ISO 9001 40, other recognised 30, plus 5 points per additional accreditation. A site-inspected award should not be outranked by a stack of paper ones.
• Complications — 35%. 100 − adjusted_rate% × 3. Where JetPatient has attested a complication rate against completed journeys, the attested figure is used, not the clinic's own — the commitment set out in §3.5, applied in the calculation itself.
• Readmission — 20%. A first-class component, not a footnote. See below.
• Patient-reported outcomes — 20%. Mean PRO score, 0–100, reported by patients during post-operative follow-up.

Why readmission carries its own weight

In domestic care a readmission is a quality signal. In cross-border care it is the signal. A complication managed at the treating hospital, while the patient is still in the destination and under the surgeon's care, is a serious but contained event. A readmission that happens after the patient has flown home is the outcome patients are actually afraid of — a different hospital, a different country's records, a surgeon eight time zones away.

So readmission is scored separately and more steeply than complications, and it is decomposed by where it happened:

readmission_score = 100 − (adjusted_30day_rate% × 4) − (adjusted_30day_rate% × share_at_home × 2)

• 30-day readmission is what is scored, because it is the measure that remains comparable to domestic benchmarks.
• 90-day readmission is recorded and shown for context but deliberately not scored — it has no comparable domestic benchmark, and scoring on it would make our numbers look worse or better than anyone else's for reasons of definition rather than quality.
• Share readmitted at home attracts the additional penalty above. A clinic whose readmissions mostly happen while the patient is still with them scores better than one whose patients get home and then deteriorate — which is exactly the distinction a cross-border patient needs and no domestic quality measure captures.

Small samples are shrunk, not flattered

Observed rates are pulled toward the network prior in proportion to how little data supports them: adjusted = (n × observed + 5 × prior) / (n + 5). A clinic with two uneventful journeys therefore cannot outrank one with two hundred. Below five completed journeys a score is labelled provisional; at or above five it is established. Sample size and a confidence figure are published alongside every score, on the face of it rather than in a tooltip.

What the Trust Score is not

2. Savings percentage

The "Up to X% less than US average" figure on each clinic sidebar is computed per procedure as:

savings_pct = round(100 × (us_reference − clinic_price) / us_reference)

Reference

• US reference. The Healthcare Bluebook / FAIR Health median for the procedure in the patient's home metro, USD, cash-pay. We use median rather than the high end of the range to keep the savings figure conservative.
• Clinic price. The lower bound of the clinic's quoted range for that procedure, USD, all-inclusive of facility fee, surgeon fee, anaesthesia, and standard implants. International transfer fees and travel/lodging are not included in the savings figure — they're called out separately on the Trip Plan tab on each profile.

For procedures where the clinic has not published a price range, the savings figure is suppressed (no number shown) rather than estimated.

2.5 Price X-Ray estimates

The Price X-Ray tool answers a different question from the clinic-sidebar figure above: before any clinic has quoted your case, roughly what does this procedure cost abroad? Because no clinic quote exists yet, its numbers are estimates by design, computed and labelled differently:

• Your anchor. The patient-responsibility or billed total read from the document you provide, on your device. If no figure can be read, the tool anchors to the U.S. national median for the procedure and says so on the result — it never invents a personal figure.
• International estimates. Indicative price levels at six accredited flagship hospitals across JetPatient corridors, derived from U.S. median benchmarks for each procedure. These are corridor-level estimates, not clinic-published prices, and every result carries that label.
• The number that binds. When you request quotes, clinics bid on your actual case. A binding quote always replaces the estimate, and only clinic-published or quoted prices ever feed the sidebar savings figure defined in §2 above.

The X-Ray's estimates are deliberately held apart from §2's clinic-sidebar figures: the sidebar suppresses rather than estimates because a clinic profile implies clinic-level precision; the X-Ray estimates rather than suppresses because a patient holding a $54,000 bill deserves an order-of-magnitude answer immediately — clearly labelled as indicative — while the binding quotes are on their way.

3. "JetPatient verified"

Before a clinic can appear in the marketplace as a verified partner it must satisfy five evidence requirements. These are what we require; the staged pipeline that collects and checks them — application, document review against issuing bodies and sanctions lists, on-site audit, and ongoing monitoring — is set out in full in Trust & Methodology.

1. Facility licensure in the destination country, evidenced by current government documentation.
2. Independent accreditation by a body recognised in the destination jurisdiction (JCI, NABH, TEMOS, ISO 9001, ICONTEC, or equivalent).
3. Board certification of every operating surgeon listed on the profile.
4. Malpractice coverage in force, with policy summary on file.
5. Outcome reporting agreement — the clinic commits to submitting de-identified outcome data after each JetPatient-routed case, which is what populates Trust Score.

Each clinic's verification status (applied, under review, verified) is tracked in the Credentialing console and is the same state machine that drives the "Newly verified" panel inside Guardian (the employer-facing product). Status changes are time-stamped, attributable to the reviewing administrator, and audit-logged.

3.5 Claimed versus attested

Clinics report their own complication rates. We do not take those numbers at face value, and we do not quietly discard them either.

For every clinic we hold two figures. The claimed rate is what the clinic reports. The attested rate is the rate JetPatient computes from completed journeys in its own network, using the same definition for every clinic. Both live on the clinic's digital twin as claimedComplication and attestedComplication, alongside attestedJourneys — the number of completed journeys the attested figure is based on, and networkMedian for context against the rest of the network.

Where the two agree within tolerance, the clinic profile reads Verified — attested outcomes. Where they diverge, it reads Adjusted, states the size of the gap, and your ranking uses the attested figure, not the claimed one. Both numbers stay visible, along with the journey count behind the attested one.

Where a clinic has too few completed journeys for the figure to be meaningful, no attested rate is shown and the profile says so — the same suppression rule used for savings percentages in §2. We would rather show you a gap than fill it.

A clinic is not removed for divergence alone. Reporting differences are frequently definitional rather than dishonest; what counts as a complication varies more than most people expect. What matters is that the number affecting your ranking is one we can stand behind, and that you can see both.

4. When this methodology changes

Any change to the Trust Score formula, the savings calculation, or the verification process is tracked in this document with a dated entry. The current canonical version is recorded in _CANONICAL_MASTER_DEPLOY.md in the source repository.

If you'd like more detail on any of the above, or if you spot something on a clinic profile that doesn't match what's described here, write to support@jetpatient.com and we'll respond inside two business days.