Causal discovery of heatwave-mortality pathways from syndromic surveillance under unmeasured confounding: a climate-analytics evaluation of attribution error in municipal heat-action triggers

Municipal heat-action plans typically trigger interventions when a forecast temperature exceeds a fixed threshold, treating the trigger as a proxy for excess mortality risk. Using a formal causal-discovery framework, the paper defines attribution error as the divergence between the trigger-implied mortality effect and the true causal effect, and derives bounds on that error under plausible confounding structures. The framework is illustrated with observed monthly air-temperature reanalysis for Harare, Zimbabwe (2000–2023; 312 monthly observations; mean 18.914 °C; warming trend 0.2001 °C per decade), which establishes the exposure context within which trigger calibration must operate. Attribution error is not a fixed property of a trigger but a function of the confounding structure, the surveillance signal's specificity, and the exposure contrast the trigger creates.