It is the season when American workers find out what their health plan will cost next year, and the number behind this year’s envelopes is unusually blunt. Mercer’s National Survey of Employer-Sponsored Health Plans projects that health benefit costs per employee will rise 8.2% in 2027 — the largest annual increase since 2003, a figure confirmed in preliminary survey results and driven, per Mercer’s own analysis, by provider consolidation, rising prices and high-cost drugs such as the GLP-1 class. The figure is still a projection from preliminary results rather than the final release — but nothing in the 2026 numbers suggests the direction will surprise. For readers arriving at this site from the US side of the ocean, this is the machinery behind the receipts our library collects.
The two-step escalation
This is not a one-year shock. Employers already absorbed a 6.7% increase in 2026 — at the time, the highest jump in 15 years, pushing the average cost per employee above $18,500. The 2027 projection layers 8.2% on top of that base. The drivers are structural, not incidental: provider consolidation and rising prices concentrate bargaining power on the hospital side of the table, while high-cost drugs — the GLP-1 class above all — add a utilisation shock on top of the price shock. Mercer’s June survey documented how employers respond: by shifting more of the cost to employees through plan design — higher deductibles, richer cost-sharing, tighter networks — even as they add support programmes. When the average grows 8% and the share shifts too, the employee’s line moves faster than the headline.
The per-episode view of the same machine
Averages per employee are abstract. The price machine shows up concretely in the figures our cost pages track: a ~US$2,453 average total payment per US emergency-room visit against a first-hand $80 public-hospital visit in China; a US$861 national average for a lower-back MRI against a ¥486 (~US$67) self-pay scan at a private hospital in Kunming. The spread exists inside the United States too — metro averages for that same MRI ranged from $404 to $1,106 — which is worth remembering before any cross-border comparison: the “US price” is itself a range, not a number. These are not the same measurements — one side is claims averages, the other documented anecdotes — but they are measurements of the same divergence: what an episode of care costs to deliver and bill in one system versus another. When the cost of covering American care compounds at 8%-a-year, the gap those receipts illustrate does not stay still either.
The honest caveats, as always
Three, unchanged from every comparison we publish. An employer-cost average is not a traveller’s price — self-pay patients abroad negotiate separately, and foreign insurers almost never fund planned treatment in China. An anecdote is not an average — our method for reading these videos exists precisely because a best-case receipt and a blended average are different things. And an episode is not the full cost — follow-up care, complications and revisions are priced wherever you end up receiving them, at local rates. The vloggers who film cash visits without any insurance are documenting the transaction, not the decade.
None of this is an argument that anyone should book a flight. It is the observation that the American cost curve and the genre this site catalogues are the same story told at two zoom levels — one compounding at 8.2% a year, the other filmed on a phone in a hospital lobby. For UK readers the compounding number is different — the queue rather than the bill — and that version of the story is the NHS breakdown. Nothing here is treatment or insurance advice.