Japan’s Flu Season Started in August—Why an Early Wave Matters Even as Weekly Cases Ease

Japan’s influenza season did not wait for winter. According to the Japan Institute for Health Security, the national number of reported influenza patients per sentinel medical institution reached 1.11 in week 34, covering August 17–23. That crossed the benchmark historically used to mark the start of an epidemic and made the 2026/27 season the second-earliest start since the current surveillance era began in 1999, behind only the 2009 pandemic year.^[1]

By week 39, September 21–27, the national sentinel rate stood at 7.17, representing 26,678 reported patients. That was down from 8.48 the week before. Yet influenza hospitalizations reported by core sentinel hospitals increased from 752 to 859. Among those hospitalized were 273 people aged 80 or older and 150 in their 70s.^[2]

What “nationwide outbreak” actually means

For years, a national average above 1.00 reported influenza cases per sentinel per week has been used as the practical marker of the beginning of seasonal epidemic activity. The 2025/26 season crossed that line in week 39, already earlier than usual. This season crossed it five weeks earlier, in week 34. JIHS explicitly calls it the second-earliest start since 1999.^[1]

The 2009 comparison needs care. That year involved the emergence of a novel A(H1N1)pdm09 virus and a global pandemic. Japan’s health ministry later estimated that roughly 20 million people were infected over about a year, with about 18,000 hospitalizations and 203 deaths. Since 2011, descendants of that virus have been treated as seasonal influenza rather than a novel pandemic strain.^[3]

Okinawa at 42.50, with high levels across Kanto and Tokai

Week 39 showed sharp geographic differences. Okinawa recorded 42.50 cases per sentinel, followed by Saitama at 14.17, Ibaraki at 12.50, Aichi at 12.33, Gifu at 10.18, Chiba at 9.76, Tokyo at 9.51 and Kanagawa at 8.59. Nineteen prefectures increased from the prior week while 28 declined. A falling national average therefore masks local epidemics moving on different timetables.^[2]

Among influenza viruses detected in Japan over the most recent five weeks, JIHS reported 110 AH1pdm09 detections, or 98%, and two AH3 detections. The circulating H1N1 lineage descends from the virus that caused the 2009 pandemic, but it now circulates as seasonal influenza.^[2]

Schools are already operating as if winter had arrived

Japan’s school surveillance system recorded 157 facilities taking closure measures in week 39. Three schools closed entirely, 34 implemented grade closures and 120 closed individual classes. From August 31 through September 27, the cumulative number of affected facilities reached 2,119. The week-39 measures involved 2,493 identified patients immediately before closure decisions.^[4]

Tokyo reported measures at 23 facilities that week, Ibaraki 13, Chiba 11 and Hyogo nine. Nagoya City alone reported 11. Those figures matter beyond education. A class closure can remove a parent from work, send siblings into different care arrangements and increase pressure on local clinics.

Why did it start so early? The evidence does not support one answer

JIHS has not assigned the early start to a single cause. Its week-35 review noted increasing influenza positivity in parts of the Northern Hemisphere and a reported rise in cases and positivity in Singapore. International movement, population immunity, weather, school schedules and contact patterns may all matter, but their relative contributions in Japan have not been established.^[1]

The episode is a useful warning against the assumption that influenza disappears during warm weather. Seasonality remains powerful, but circulation does not fall to zero, and a globally connected population can experience substantial transmission well before conventional winter timing.

This is also the first full season under a new surveillance architecture

Japan introduced acute respiratory infection surveillance in April 2025. Influenza reporting moved from roughly 5,000 influenza/COVID sentinels to about 3,000 ARI sentinel institutions, including about 2,000 pediatric and 1,000 internal-medicine sites.^[1]

Because the institutions and geographic continuity changed, JIHS temporarily stopped publishing its former influenza warning and advisory map and its estimate of total infections. That means comparisons such as “2024 week 39 was 0.63 and 2026 week 39 is 7.17” should not be presented as a clean eleven-fold increase. The surveillance denominator is different.^[5]

What can be said more confidently is that the new system has recorded a steep rise since August and that independent indicators—hospital admissions and school closures—show real social impact.

Why hospitalizations can rise while outpatient reports fall

Week 39 produced an apparently contradictory picture: the outpatient sentinel rate fell from 8.48 to 7.17, while influenza hospitalizations rose from 752 to 859. The measures track different systems and different stages of illness. Severe disease often follows infection with a delay, and prefectures can be at different points in their local epidemic curves.

The age distribution is also important. Hospital reports included 273 people aged 80 or older, 150 in their 70s and 82 in their 60s. At the other end of the age range, there were 102 admissions among children aged 1–4 and another 102 among those aged 5–9.^[2]

Vaccination meets an unusually early season

Japan’s 2026/27 influenza vaccines are trivalent, covering two influenza A subtypes and the B/Victoria lineage. Authorities expect roughly 51.9 million doses to be available over the season. Routine vaccination under Japan’s Immunization Act covers people 65 and older and certain people aged 60–64 with serious heart, kidney, respiratory or immune impairment.^[6]

This season also lost a planned new option. A high-dose inactivated influenza vaccine for older adults had been expected to enter routine vaccination from October 2026, but supply difficulties led the government to abandon its use in the routine program this season. The Japanese Association for Infectious Diseases says the domestic launch timing remains uncertain.^[7]

That does not mean Japan lacks standard influenza vaccine. It means one additional higher-dose option intended for older adults will not arrive on the timetable originally expected—a notable issue in a season when older age groups already account for a substantial share of reported hospitalizations.

What workplaces should plan for

An early flu season is a continuity problem for employers. Absence comes not only from infected workers but from parents whose children are sent home by school closures. Businesses with single-person dependencies, inflexible onsite requirements or thin staffing are more vulnerable.

The realistic goal is not zero infections. It is resilience when infections occur: clear stay-home rules for symptomatic workers, remote-work options where possible, cross-training for critical tasks, ventilation, hand hygiene and appropriate mask use in crowded settings. JIHS recommends cough etiquette, hand hygiene, ventilation, appropriate mask use, preventing introduction of respiratory viruses into medical and care facilities, and considering vaccination.^[1]

Symptoms alone cannot tell flu from COVID-19

Fever, headache, fatigue, muscle and joint pain, cough and nasal symptoms are common with influenza, but JIHS cautions that symptoms alone cannot reliably distinguish influenza from COVID-19. In its week-35 analysis, COVID-19 sentinel reporting was also rising.^[1]

People at higher risk of severe illness—including older adults, very young children and people with certain underlying conditions—may need earlier medical assessment. Decisions about diagnosis or antiviral treatment belong with clinicians rather than with national surveillance numbers.

What 2009 teaches—and what it does not

Calling 2026 the earliest start since the 2009 pandemic can sound more alarming than the evidence supports. The two events are not equivalent. In 2009 a novel H1N1 virus spread through a population with little pre-existing immunity. In 2026, AH1pdm09 is a seasonal lineage descended from that pandemic virus.^[3]

The useful comparison is calendar timing. Systems that normally use autumn to prepare for winter—schools, employers, municipalities, clinics and hospitals—were already responding in August and September. The pressure arrived before the usual preparation window was over.

Winter has not started yet

The decline in the national week-39 sentinel rate may prove encouraging, but seasonal influenza does not always produce one clean peak. The 2025/26 season was explicitly bimodal, reaching 51.12 per sentinel in week 47 and then 43.34 in week 6 of 2026.^[1]

This year, school closures and rising hospitalizations have arrived before the coldest months. The most important question is therefore not whether one week’s national number has fallen. It is how much capacity schools, workplaces, hospitals and long-term-care facilities will still have if another wave arrives after winter actually begins.