Modeling the Present and Future Incidence of Pediatric Hand, Foot, and Mouth Disease Associated with Ambient Temperature in Mainland China

Abstract

Background: There is limited evidence about the association between ambient temperature and the incidence of pediatric hand, foot, and mouth disease (HFMD) nationwide in China.

Objectives: We examined the childhood temperature-HFMD associations across mainland China, and we projected the change in HFMD cases due to projected temperature change by the 2090s.

Methods: Data on daily HFMD (children 0-14 y old) counts and weather were collected from 362 sites during 2009-2014. Daily temperature by the 2090s was downscaled under the Representative Concentration Pathway (RCP) 4.5 and 8.5 scenarios. Temperature-HFMD associations were quantified using a two-stage Poisson regression with a distributed lag nonlinear model. The impact of changes in temperature on the incidence of HFMD was estimated by combining the fitted temperature-HFMD associations with projected temperatures under each scenario, assuming a constant population structure. Sensitivity analyses were performed to assess the influence of primary model assumptions.

Results: During 2009-2014, >11 million HFMD cases were reported. In most regions, the temperature-HFMD association had an inverted U shape with a peak at approximately 20°C, but the association leveled off or continued to increase in the Inner Mongolia and Northeast regions. When estimates were pooled across all regions and the population size was held constant, the projected incidence of HFMD increased by 3.2% [95% empirical confidence interval (eCI): −13.5%, 20.0%] and 5.3% (95% eCI: −33.3%, 44.0%) by the 2090s under the RCP 4.5 and 8.5 scenarios, respectively. However, regional projections suggest that HFMD may decrease with climate change in temperate areas of central and eastern China.

Conclusion: Our estimates suggest that the association between temperature and HFMD varies across China and that the future impact of climate change on HFMD incidence will vary as well. Other factors, including changes in the size of the population at risk (children 0-14 y old) will also influence future HFMD trends. https://doi.org/10.1289/EHP3062.

Figure 1.

(A) Annual HFMD morbidity standardized by age (per 1,000 children 0–14 y old) and (B) average daily mean temperature ($°\mathrm{C}$) in four municipalities, 332 prefectures, and 26 counties under the jurisdiction of province (CJPs) in mainland China, 2009–2014. Data on daily hand, foot, and mouth disease (HFMD) and weather for each city were provided by Chinese Center for Disease Control and Prevention through the China Information System for Disease Control and Prevention and the China Meteorological Data Sharing Service System. Population data for calculating HFMD morbidity were extracted from National Bureau of Statistics of China. Maps in this publication were generated using ArcMap (version 10.5; ESRI, Inc.) with topographical basemap content from GADM.

Figure 2.

Pooled cumulative associations between daily mean temperature and hand, foot, and mouth disease (HFMD) lagged over 0–14 d for 2009–2014. Black solid lines indicate region-specific associations, and shaded areas indicate 95% confidence interval bands. $I^2$ values and the number of provinces or municipalities included in each pooled estimate are provided. Temperature-HFMD associations were estimated by pooling the site-specific estimates using random-effect meta-analyses. Natural cubic splines (with three degrees of freedom) were used to model temperature and lag days. Note: No.Prov/Muni, the number of provinces or municipalities in the region; RR, relative risk.

Figure 3.

Projected percent change in hand, foot, and mouth disease (HFMD) incidence among children 0–14 y old due to climate change [Representative Concentration Pathway (RCP) 4.5 and 8.5 scenarios] relative to baseline estimates for 2009–2014, holding population sizes and temperature-HFMD associations constant over time. Data shown in this figure are the point estimates across 362 sites with the 95% empirical confidence intervals provided in Table S4. Maps in this publication were generated using ArcMap (version 10.5; ESRI, Inc.) with topographical basemap content from GADM.