Work

El Niño, La Niña and recorded heat deaths in India: index choice and early-warning value in a warming climate

Year: 2026
Type: preprint
Study design: Observational Study
Abstract: We asked whether El Niño–Southern Oscillation (ENSO)–health estimates depend on the index, now adjusted operationally for tropics-wide warming, and whether ENSO, forecast before South Asia’s April–June hot season, adds lead time to heat early warning. We analysed 23,821 heat/sun-stroke deaths recorded in 34 Indian states and union territories, 2001–2024, in negative-binomial mixed models; because a year random intercept was estimated at zero, year-level uncertainty was assessed with a national 24-year model, leave-one-year-out refits and a year-block bootstrap. Per 1°C of the preceding December–February traditional Oceanic Niño Index (ONI), the rate ratio was 1.28, with a year-block 95% confidence interval (CI) of 1.06–1.50; it was 1.27 with the Relative Oceanic Niño Index (RONI) and 1.27 nationally. Fewer deaths followed La Niña winters in every specification reported (0.74, 95% CI 0.60–0.91, per 1°C cooler), also without 2020–2023. The El Niño-phase estimate was fragile: 1.21 with the ONI, 1.13 with RONI and 1.13 without 2024. Withholding whole El Niño episodes, adding ENSO to a baseline without the previous year’s count reduced forecast error by 0.6% (skill − 1.1% without 2024), but not once observed local heat was known. Illustrative + 2.0°C scenarios implied 359–510 additional recorded deaths in 2027 with a separate El Niño term (including zero with RONI) and 586–725 with a single symmetric term, whose slope partly reflects La Niña winters; the indices’ +2.0°C values describe different events. ENSO–health studies should report both indices; whether ENSO outlooks improve heat-health preparedness beyond local temperature forecasts and surveillance remains untested.
Language: English
Open Access status: green
OpenAlex ID: W7226000246
Cites: 25
Cited by:
Related to: 10