Five Years After COVID, Japan Is Still Dying at Pandemic Rates, and Confirmed COVID Explains Only a Tenth

Epidemiologists trust one number more than any death certificate, precisely because it does not care what the certificate says. Excess mortality asks a blunt question: how many more people died this year than a normal year would have predicted? It is the tool that pinned the true scale of the 1918 influenza long after the official counts were filed and forgotten, and the tool that later exposed how badly some governments undercounted their COVID dead. It does not argue. It counts the coffins and subtracts what you expected.

Which is why what that number is doing in Japan should stop you cold.

In 2020, the first year of the pandemic, fewer Japanese died than the models predicted. The country ran a negative excess mortality of -1.67 percent, according to a two-stage interrupted time-series analysis led by Shuhei Nomura’s group at the University of Tokyo and published in BMJ Public Health in April 2025. That is not a typo, and it is not even surprising: masking, distancing, and a collapse in ordinary respiratory infection meant that in the year the virus arrived, Japan buried fewer people than usual. This was the model pandemic performer, the country that seemed to have outrun the thing.

Then the line turned, and it turned the wrong way. Excess mortality rose to 2.19 percent in 2021 and spiked to 7.55 percent in 2022, the worst year of the entire period. Here is what refuses to resolve. After the government downgraded COVID to a routine infectious disease in May 2023, when the emergency was supposedly over and the curve should have been relaxing back toward zero, excess mortality for the rest of that year still ran at 5.76 percent. Across 2020 through 2023, the same analysis counts roughly 219,500 excess deaths. A separate 2025 review by Hiroshi Kusunoki of Osaka Dental University, drawing on national and municipal vital statistics, found the crude mortality rate still climbing through 2024. The instrument that should have gone quiet is, five years later, still ringing.

The official explanation is not stupid, and it deserves to be stated at its strongest. Japan is the oldest society on earth, and an aging population dies at rising rates almost mechanically. Lockdowns delayed cancer screenings and cardiac care, and delayed care kills on a lag, a pattern the Journal of Epidemiology documented in shifting cancer-death locations through early 2023. There is undercounted COVID, people who died of the virus without a positive test or in the crush of overwhelmed wards. Pile the aging curve, the deferred care, and the missed diagnoses together and you can build a story that never once mentions a syringe. Parts of that story are certainly true.

But an explanation has to fit the shape of the data, and this one has a hole in it you could drive a hearse through. If undercounted COVID were doing the heavy lifting, the excess should track the infection waves. It doesn’t. By the reckoning of Hideki Kakeya, a University of Tsukuba engineer, and his co-authors in the JMA Journal, Japan’s excess deaths per million exceeded 1,400 in 2023, roughly three times the American rate, and confirmed COVID deaths accounted for only about 10 percent of them. Confirmed COVID, in other words, cannot explain the other ninety percent of the dying. Undercounted COVID remains a live candidate for some of that gap, but it is a hypothesis, not a measured cause, and it has to compete with the timing. The surge did not begin with the first wave. It began in 2021 and detonated in 2022, after the country had already vaccinated harder than almost any nation on the planet.

That timing is the uncomfortable part. Japan reached one of the world’s highest per-capita rates of mRNA vaccination, with more than 80 percent of the population fully vaccinated and boosters layered on top, and it was after that campaign that both infections and excess deaths climbed. Kakeya’s group went further, reporting that in prefecture-level analysis the number of booster doses tracked excess mortality more tightly than infection rates or population density did. We have watched ecological correlations mislead careful people before, and I will come back to that. But it is not a small thing when the curve of the dying lines up better with the shots than with the disease the shots were meant to stop.

Then there is the ledger the government keeps itself, which is harder to wave away than any regression. Under Japan’s vaccine-injury relief system, the state had by November 18, 2024 paid out on 8,432 injury claims, including 903 deaths, following COVID-19 vaccination. Sit with the comparison Kakeya’s paper draws. Those figures already exceed the total injuries and deaths compensated after all other vaccines combined over the previous 47 years, since the program began in 1977. The system is deliberately no-fault, which means a payout is not an admission of cause and proves nothing on its own. But a state does not certify 903 death claims against a single vaccine in three years, against a few thousand for everything else across half a century, without a signal that at minimum warrants an audit.

Now the honesty this beat demands, and the sharper point buried inside it. Kakeya’s paper is an opinion piece, not a controlled trial, and it leans on ecological data, the population-level correlation that has fooled good scientists before. Kusunoki’s review openly concedes he could not even compute age-standardized mortality rates, because the granular data was withheld from him. Those are signals, not verdicts, and anyone selling them as a closed case is overselling. But watch what the establishment does with that uncertainty. It treats the softness of the skeptics’ data as a reason to dismiss the question, when the same softness is a reason to run the one study that would harden the answer. Uncertainty is the argument for looking, not the excuse for looking away.

And Japan can look. It sits on individual-level records: who was vaccinated, when, with how many doses, and who died. Linking those two datasets would convert a decade of ecological guesswork into a clean answer, and Japan is one of the few countries with the registries to do it. That linkage has not been run. Instead the institutional response has been to publish rebuttals, one in the JMA Journal accusing the skeptics of “methodological flaws” and reliance on “retracted evidence”, to reach for the word misinformation, and to offer every explanation except the one an audit of the campaign might confirm or kill. When the people holding the data answer a mortality signal by grading the critics’ methods instead of releasing the records that would settle the methods, the choice of what to withhold is itself the tell.

So the mystery deepens for a reason that is not mysterious at all. It persists because the definitive study has not been run, and the records that would run it sit unlinked in the hands of the institutions with the most to lose from the result. The excess-mortality number, that stubborn old lie-detector, keeps counting coffins Tokyo has not accounted for.

The next reading is scheduled. Japan’s finalized 2024 vital statistics came out in September 2025, and the finalized 2025 figures are due this coming September; they will tell us whether the curve is finally bending back to baseline or hardening into a grim new normal. The card that matters is not a statistic, though. It is a decision, and it belongs to Japan’s Ministry of Health, Labour and Welfare: release the record-level linkage between vaccination and death, or keep explaining a five-year surge with everything except the intervention that reached more Japanese arms than any in the country’s history.

Sources

1. BMJ Public Health – Nomura et al., excess mortality in Japan during and after the COVID-19 emergency, two-stage interrupted time-series (April 2025)
2. JMA Journal – Kakeya et al., “Significant Increase in Excess Deaths after Repeated COVID-19 Vaccination in Japan” (2025, PMC full text)
3. JMA Journal – Kakeya et al., journal landing page (10.31662/jmaj.2024-0298)
4. Healthcare / Osaka Dental University – Kusunoki, review of rising mortality in Japan with international comparisons (2025)
5. JMA Journal – “Concerns about Methodological Flaws and Use of Retracted Evidence in the Opinion on COVID-19 Vaccination and Excess Mortality in Japan” (2025)
6. Journal of Epidemiology – impact of the COVID-19 pandemic on cancer death locations in Japan, excess mortality through February 2023
7. Ministry of Health, Labour and Welfare – Summary of Vital Statistics (release schedule for finalized annual figures)
8. TrialSite News – Japan’s excess death mystery deepens: persistent mortality surge five years after COVID (July 2026)