China's childhood lower respiratory infection mortality collapsed from 330,194 deaths in 1990 to 12,098 in 2021—a 96.3% reduction—while mean national temperature rose 0.364°C per decade and diurnal temperature range narrowed. A parallel 94.9% decline in asthma deaths occurred over the same period. Using GBD 2021 mortality data linked to high-resolution C-LSAT gridded temperature records across five climate zones, the authors tested four temperature indicators against log mortality rates with Newey-West and bootstrapped quadratic models.
Despite initially large regression coefficients suggesting temperature associations, every relationship evaporated upon detrending and a future-exposure falsification test—a critical methodological discipline rarely applied in climate-health literature. The apparent minimum mortality temperature of 35.84°C proved a mathematical artifact far outside observed data (6.66–8.13°C range), underscoring how quadratic extrapolation can mislead. The honest conclusion: parallel secular trends in warming and mortality reduction are confounded by unmeasured drivers—almost certainly improved antibiotics, vaccination coverage, neonatal care, air quality interventions, and rising incomes.
This finding matters broadly because climate-health research frequently reports temperature-mortality associations without adequately ruling out shared long-run trends. The falsification framework demonstrated here should become standard practice. Limitations include ecological design, national aggregation masking regional heterogeneity, and PM2.5 adjustment being partial rather than comprehensive. As a preprint not yet peer-reviewed, these methodological conclusions warrant independent verification, but the detrending discipline itself represents a genuinely valuable contribution to epidemiological rigor—confirmatory of skepticism toward unadjusted time-series climate inferences.