| Verified against primary record | |
| Why averaging fails | Medical expenses tend to vary tremendously across diseases and treatment techniques[1] |
|---|---|
| The clustering problem | That variation hinders compilers from forming clusters when calculating average expenses for patients[1] |
| The missing denominator | One statistical agency removed health and education travellers from its traveller counts[2] |
| Records read | An IMF guidance note, two statistical agency publications, a national statistics portal and a survey results page, 9 October 2026 |
| Independently reported | |
| Companion spending | Recorded under other personal travel, so it sits outside the health-related travel figure[1] |
| Bands apply only to the rows beneath them. No average expenditure per medical traveller is published here, because no official figure was located. | |
Average expenditure per medical traveller is the figure on which most medical tourism market sizing rests, and the international body responsible for the relevant statistical standard has written down why the calculation does not work. This entry publishes no such average, because no official one was located and the standard-setter advises against constructing one this way.
The standard-setter’s own objection
A guidance note prepared for the revision of the balance of payments manual addresses this directly. It records that medical expenses tend to vary tremendously across diseases and treatment techniques, and that this hinders compilers from forming clusters when calculating average expenses for patients.[1] Clustering is the standard tourism technique of grouping visitors with similar spending and applying a group mean, and the note’s point is that patients do not cluster.
The same passage states that, unlike the case of estimating expenses of normal tourists, applying an arithmetic or geometric average of medical expenses would not yield reasonably good estimates of health-related expenses.[1] The middle of that sentence could not be retrieved in full when the document was read for this entry, so the two verified fragments are reproduced without the words between them, and this entry does not assert what the omitted clause says.
Even so, the direction is unambiguous: the body that sets the rules for the national accounts in which health-related travel appears has stated that multiplying an average medical expense by a patient count does not produce a good estimate. That is the method most commercial market sizing uses, and it is examined further in medical tourism market size.
Why the inputs are hard to get
The same note sets out the practical obstacles to each possible data source. Immigration records cannot help, because it would not be feasible for the immigration authority to distinguish patients from other types of visitors. Exit surveys face a response problem, since the willingness of patients to provide information to a voluntary exit survey could be a major concern. And ministry of health data is identified as a potential source for the number of patients and their expenses, though such data generally covers patients rather than the people travelling with them.[1]
Companions are excluded by design rather than by accident. Under the treatment adopted in the revision, companions’ expenses are recorded under other personal travel.[1] A published health-related travel figure therefore omits the spending of the people who travelled with the patient, which is real spending attributable to the medical trip. Any comparison with a commercial estimate that includes companions is comparing two different quantities.
A numerator without a denominator
The clearest illustration of why no official per-traveller average exists comes from one country’s accounts. In documenting a methodology change, the statistical agency records two decisions in the same summary table. For health-related travel it adopted an estimation methodology based in part on data from a survey of international air travellers about expenditures of travellers travelling for health purposes. And it removed estimates of travellers travelling for health and education purposes from traveller counts.[2]
So the spending is estimated and the people are not counted. There is a numerator in the accounts and no matching denominator, and a per-traveller average cannot be formed from them.
The spending figures themselves are published. In one statistical table, health-related personal travel exports were 3,925 million dollars in 2017 and 4,097 million in 2018, with imports of 2,316 million and 2,606 million for the same years.[3] Those are aggregate service-account figures, not a market size, and no accompanying traveller count is published for the row.
The one per-unit average that is published, and what it measures
Latvia’s medical tourism collection publishes an indicator labelled average amount spent on one medical service, in euro, alongside the number of manipulations and the total amount spent on medical services, crossed with service form and country of usual residence.[4]
That is an average per service, not per patient and not per trip. One patient can receive many services, so the figure cannot be relabelled as spend per medical traveller, and doing so would be exactly the error the guidance note warns against. No figures from the Latvian tables are reproduced here, because the portal serves them through an interactive query that could not be submitted.
For contrast only, an American survey reports that the average overseas visitor spent 1,829 dollars and stayed 16.9 nights.[5] That is all overseas visitors, not medical travellers, and is given here solely to show the order of magnitude of a genuine per-visitor average where one is actually collected.
How to read a per-patient figure
Any circulating claim that the average medical tourist spends a given amount should be treated as provider-supplied unless a primary record is named. Four questions settle it: what the denominator is, whether companions are included, whether travel and accommodation are included, and which procedures the average spans. Because expenses vary tremendously across treatments, an average across procedures carries almost no information about any individual case.
See also
- Medical tourism and balance of payments statistics, the framework these figures sit in
- Medical tourism market size, the method this entry’s main source warns against
- Medical tourism revenue, what the money figures include and exclude
- Average length of stay in medical tourism, the same averaging problem applied to time
References
- International Monetary Fund, Committee on Balance of Payments Statistics, Current Account Task Team. Guidance note C.7, treatment of travel packages, health-related travel, and taxes and fees on passengers’ tickets. Approved March 2022. Verified against primary record: the measurement annex and the recommendations on companions were read; one sentence could not be retrieved in full and is reproduced only as its two verified fragments. Retrieved 9 October 2026.
- Telles R Jr, Martinez N, Peck T. Annual update of the U.S. international transactions accounts. Survey of Current Business, volume 100 number 7, 21 July 2020, United States Bureau of Economic Analysis. Verified against primary record: the summary of major changes table was read for both the estimation method and the removal from traveller counts. Retrieved 9 October 2026.
- United States Bureau of Economic Analysis. U.S. trade in services, by type of service, table 2.1. Release date 15 October 2019. Verified against primary record: the health related rows were read directly from the published table. Retrieved 9 October 2026.
- Centre for Disease Prevention and Control of Latvia. Medical tourism statistics tables. Official Statistics Portal of Latvia, tables last updated 31 July 2026. Verified against primary record: indicator labels and dimensions read; cell values not retrieved. Retrieved 9 October 2026.
- United States National Travel and Tourism Office. 2025 Survey of International Air Travelers results. Undated on the page. Verified against primary record: average spend and nights read from the published results. Retrieved 9 October 2026.
Sourcing note: the guidance note, the two statistical agency publications, the Latvian tables and the survey results page were opened and read on 9 October 2026. The guidance note’s sentence about averaging is reproduced as two verified fragments because the words between them could not be retrieved; no wording is supplied to bridge them. No average expenditure per medical traveller is published in this entry, because no official figure was located in any source consulted. The all-visitor spending figure is given as contrast and is explicitly not a medical travel figure.