急诊科就诊与体温：来自墨西哥的证据

Emergency Department Visitsand Temperature: Evidencefrom Mexico Luis Sarmiento, Francesco Pietro Colelli, and Filippo Pavanello Emergency department visits and temperature:Evidence from Mexico∗ Luis Sarmiento†Francesco Pietro Colelli‡Filippo Pavanello March 21, 2025 We estimate the impact of temperatures on emergency department visits using dailydata from the universe of public hospitals in Mexico from 2008 to 2022. We find thatcold temperatures decrease visits by up to 8.9 percent on the same day, and warmtemperatures increase visits by as much as 3.6 percent. Using distributed lag models,we then show that cold temperatures can reduce visits for the next 30 days by up to16.3 percent.For warm temperatures, contemporaneous and cumulative effects aresimilar (limited harvesting). These findings suggest that, unlike mortality, tempera-tures affect the demand for emergency services linearly. Leveraging the granularity ofour dataset, we also document significant heterogeneities (e.g., higher sensitivity forchildren and teenagers) and relevant mechanisms, such as ecosystem dynamics andbehavioral changes.Finally, we project that temperature-driven annual emergencydepartment visits will increase by 0.24 percent by midcentury, resulting in an esti-mated increase of 92 million USD in annual medical expenditures in Mexico. Keywords: Temperature, Morbidity, Mexico, Climate ChangeJEL: I12, O13, Q54 1. Introduction In the context of global warming, the impact of adverse temperature conditions on healthis a significant public health concern worldwide. Although extensive research has doc-umented temperature’s effect on mortality, its influence on subfatal health conditions re-mains largely overlooked. A primary reason for this gap in the literature is the lack ofcomprehensive morbidity data, hindering large-scale assessments and leading to an in-complete estimate of the health costs associated with temperature changes (White, 2017;Gould et al., 2024). This issue is especially pronounced in developing countries, whereclimate change is likely to have more severe effects due to warmer climates and loweradaptive capacities (Davis et al., 2021). We estimate the relationship between temperature and morbidity in Mexico using dailycase-level data on emergency department (ED) visits from all public hospitals between2008 and 2022. To assess the causal effect of temperature on ED services, we exploit ex-ogenous daily variations on average temperatures within the same municipality, month,and year, while controlling for correlated weather variables, such as precipitation andrelative humidity, and day-of-the-week seasonal factors.Using distributed lag models(Deschenes and Moretti, 2009), we provide evidence of the effects of temperature changeson same-day and cumulative demand for ED services over the following 30 days. Our results show that an additional day with an average temperature above 30°C in-creases contemporaneous ED visits by 3.6 percent.In contrast, an additional day be-low 10°C reduces visits by 8.9 percent. For the cumulative effect over the next 30 days,the increase from heat slightly decreases to 2.5 percent (harvesting), and the reductionsfrom cold almost double, to 16.3 percent. These findings align with research in Califor-nia, which indicates a linear relationship between the temperature gradient and ED visits(White, 2017; Gould et al., 2024).Observing the same effect in Mexico and Californiaenhances the external validity of our results, suggesting that this relationship holds indifferent contexts. Finding a linear relationship between ED visits and temperature changes is crucial be- cause it contrasts with the U-shaped relationship observed for mortality (Cohen and Deche-zleprˆetre, 2022; Deschenes and Moretti, 2009; Deschˆenes and Greenstone, 2011).Usingdata from death certificates during the same period, we confirm this U-shaped relation-ship in Mexico. This difference in functional forms implies significant heterogeneities inthe potential effects of climate change. Nonlinear and concave relationships make the con-sequences context dependent, influenced by climatic and social conditions. In contrast,when the relationship is linear, climate change’s shift toward warmer temperatures willconsistently increase demand for ED services across the entire temperature distribution. Our analysis by disease category shows that cumulative cold temperatures decrease over-all visits, particularly those related to infectious diseases, despite raising admissions forrespiratory diseases over time.In contrast, hot temperatures raise admissions for en-docrine, genitourinary, and infectious-parasitic diseases and external causes.The coef-ficients across disease categories align in sign and significance with estimates from Cali-fornia (White, 2017). However, variations in the prevalence of certain conditions lead todifferences in the average effect. Specifically, the cumulative increase in cold-related visitsin California p