There is no doubt that the outbreak of COVID-19 has caused enormous economic loss and health burden around the world. Under such circumstances, independent and robust scientific evidence will undoubtedly provide a powerful weapon to deal with this crisis. We believe that it is significant to clarify the relationship between ambient temperature and mortality of the COVID-19, not only in Wuhan but also in other epidemic areas in the world. In this study, we used a rigorous and scientific mathematical model to reveal the unique relationship between temperature and death caused by COVID-19, even if the association between death and temperature in non-communicable diseases has been established [39]. We hope that the research results can provide some methodological guidance for the response to this crisis.
DLNM model was verified to be a useful tool in this study to assess the non-linear relationship between ambient temperature and COVID-19 mortality on a daily basis, including properly evaluating the non-linear associations and cumulative death relative risks related to temperatures for lag days. The model figures out the non-linear and negative correlation between ambient temperature and COVID-19 mortality [26, 39, 40]. The increase in temperature could reduce the death risk of patients, and the relationship between temperature and death effect was U-shaped.
Our study found that the relationship between death risk of COVID-19 and low temperature was different from the high one. The low temperature effect on the death risk of COVID-19 is first enhanced and then weakened. With increasing of outdoor temperature, the death risk of COVID-19 is decreasing. The increase in temperature may reduce the lethal intensity of COVID-19, which is related to the increase of virus inactivation caused by high temperature [35, 41]. When the ambient temperature rose to around 10.0 °C and continued to rise, the temperature and the death risk of COVID-19 gradually decreased and then increased, which is consistent with the findings in non-communicable diseases [38]. This may be due to that COVID-19 is most likely to occur in the elderly with weakened immune systems and combined diseases [15], such as diabetes and hypertension. When the temperature is getting higher and higher beyond the inflection, the death risk of these combined diseases may also increase [23], which potentially increases the death risk of patients with COVID-19. Besides, the low temperature effects are acute and short-term [42], with the most considerable risk occurring in 5–7 days of exposure. High temperature mainly reflects the acute effect, and the maximum effect occurs on the day of temperature exposure, which is similar to some studies [43, 44].
The results of the study show that low temperature has a more significant impact on the death risk of COVID-19 than high temperature is consistent with a meta-analysis [45]. At low temperatures, deaths from respiratory illnesses are greatly affected. Exposure to low temperature in humans can cause cardiovascular stress, which is affected by factors such as peripheral blood vessel constriction, plasma cholesterol, plasma fibrinogen, red blood cell count, blood viscosity, and inflammatory response [46, 47]. These factors together lead to respiratory distress, thus contribute to the deterioration of COVID-19 patients. At high temperatures, the number of patients dying from chronic non-communicable diseases increased, which forms a potential competitive relationship, leading to a gentle change in the number of COVID-19 deaths directly attributable to temperature [39, 48].
Overall, the temperature was negatively correlated with the cumulative effect of COVID-19 deaths [24]. At low temperatures, the cumulative death risk of COVID-19 was higher. With the increase of daily average temperature, the delayed effects of temperature exposure in patients with COVID-19 decrease rapidly and show protective effects. This data indicates that the risk of death of COVID-19 patients gradually decreases due to the increase in ambient temperature. With the advent of summer, the COVID-19 patient population may benefit from the high temperature effect.
Sensitivity analysis showed that the results of this study were robust. Firstly, the distributed lag non-linear method can flexibly dig out the possible relationship between temperature changes and daily mortality and cumulative delayed effects. Although the model is involved with many parameters, our sensitivity analysis shows that the results are robust [48]. Secondly, during the analysis, we adjusted a group of potential confounding factors, including daily average temperature, relative humidity and AQI, and compared the model results after excluding relative humidity or AQI. Generally, our results were relatively robust.
Some limitations should be considered in interpreting our findings: Firstly, this is an ecologically designed study, and the use of environmental monitoring data may not accurately reflect actual personal exposure. Secondly, COVID-19 patients basically receive isolation treatment in the designated hospital, and the patients live in the closed space, so the relationship between the ambient temperature and death may be different from that of indoor temperature. Third, this study did not adjust the social and demographic factors such as population mobility and economy, which may affect the population structure and mortality [9]. Finally, in the process of treating and curing COVID-19 patients, clinical diagnosis and treatment guideline is continuously updated, and the impact from this inconstancy was not included in this study.