What has actually killed Americans, and what is circulating right now. Life expectancy and the major causes of death run from 1900 — long enough that the 1918 influenza pandemic shows up as a hole in the record rather than a statistic. Alongside them: flu surveillance across all ten HHS regions, hospital load, and the notifiable diseases CDC counts every week.
Nearly thirty flu seasons on one screen.
Where the current respiratory season stands against decades of the same measure.
Six groups, from a century of mortality down to what is circulating this week.
Life expectancy was 47.3 years in 1900, and it fell to 39.1 in 1918. That notch is the influenza pandemic, and it is the most violent single movement anywhere in this data — a better sense of what a severe pandemic does to a country than any modern comparison.
Influenza and pneumonia killed 612 per 100,000 in 1918, against 267 the year before and 14.9 by 2018. A forty-one-fold decline, from the leading cause of death to a footnote, achieved with antibiotics, vaccines and oxygen.
Worth its own paragraph. US measles peaked at 894,134 cases in 1941. The vaccine was licensed in 1963 (385,156 cases that year), and five years later the count was 22,231 — down 94%. By 2000, when measles was declared eliminated in the US, there were 85. 2025 ran to 2,289, the worst year since 1992.
Read it against kindergarten MMR coverage (94.4% in 2009-10, 92.4% now) and exemptions (1.6% to 4.2%). Measles needs roughly 95% coverage to stop circulating, which is why a couple of percentage points matters far more than it sounds. Almost all of the exemption rise is non-medical; medical exemptions have stayed near 0.2%.
Sources: OWID’s compilation of Public Health Reports, Census Bureau and CDC for the case history; CDC’s SchoolVaxView for coverage.
The cause-of-death rates are age-adjusted: they correct for the fact that the population got much older over the century. Without that correction almost every disease would appear to be “rising” for demographic reasons alone. The all-cause death rate is crude, so some of its recent flattening genuinely is just an aging population — the two are not directly comparable.
Beyond flu surveillance:
Death records take weeks to register, so the most recent weeks of any provisional mortality series are always under-reported and slope downward. Plotted raw that looks like deaths falling. The trailing weeks that sit below 90% of the preceding half-year’s median are dropped here for that reason — three weeks were removed from the end of the all-cause series. CDC froze that dataset after 2023.
The notifiable-disease counts come from the cumulative year-to-date column, differenced back into weeks, rather than from the weekly column directly. The weekly figure is the count as first filed and never picks up late reports — for 2024 it totals 11,145 pertussis cases against the 35,493 CDC actually published. Differencing the cumulative keeps weekly resolution while still reconciling to the real annual total.
This is surveillance data with a short reporting lag, tracking illness that looks like flu (not lab-confirmed cases of any one bug). It’s for seasonal context, not diagnosis or forecasting.
