A Study On The Meteorological Incentives Of Upper Respiratory Tract Infection And Forecasting Methods

Objective (1) To make a quantitative analysis of the meteorological factors influencing the daily Emergency Room (ER) visits of the Upper Respiratory Tract Infection (URTI) in Beijing; (2) To make a prediction of the daily URTI-related ER visits, in order to prevent the diseases induced by the climate changes.Methods The ER records of URTI were collected in three general hospitals in Haidian, Beijing from January 2009 to December 2012, and the data of the air pollutants in that period were also collected from the local authorities. Generalized additive models (GAM) were used to explore the exposure-response relationship between the meteorological factors and the ER visits on URTI. Four methods, including GAM, multiple stepwise regression, BP neural network and decision tree were used to evaluate the potential patients of ER caused by URTI. The levels of the prediction were divided into proportion, equal-width (distance) partitioning and equal-depth (frequency) partitioning methods while analyzing the whole population, the vulnerable and the young individuals respectively.Results (1) The number of daily ER visits on URTI was periodically changed as a "U" shaped curve annually from 2009 to 2012. Moreover, it had a negative correlation with the temperature, humidity, precipitation and the average wind speed; while it had a positive correlation with the air pressure, SO2, NO2 (P<0.01). (2) The exposure-respond relationship between the mean daily temperature and the daily ER visits of URTI showed an approximate U-shaped distribution. The number of the daily URTI-related ER visits among the whole population, the vulnerable and the young individual groups reached to the minimum when the mean temperature daily was 18.1℃,7.1℃ and 20.0℃. Furthermore, the mean daily temperature which lagged of 4 days (tlag4) had the greatest impact on the daily number of the URTI-related ER visits when the relative risk (RR) was 1.0053 (95%CI:1.0043～1.0063); while the air pressure which lagged of 4 days (plag4) had the greatest impact on the number of these patients when the RR was 1.0037 (95%CI:1.0029～1.0045). With the changing of the relative humidity and the average wind speed, the relative risk (RR) of the daily ER URTI patients were 1.0007 (95% CI:1.0003～1.0011) and 1.0003 (95% CI: 1.0050～1.0050). The threshold of the average air pressure, relative humidity, average wind speed affecting the daily number of ER visits of the URTI were 1019.5hPa,41% and 3.9m/s. (3) The equal-width partitioning method analyzed on three groups of URTI:the whole population, the vulnerable and the young group worked best among the different data processing methods. The prediction accuracy rate on three group people of GAM were 84.50%,83.40% and 89.42%; the multiple stepwise regression were 71.16%,78.65% and 69.66%; the BP neural network were 74.66%,63.70% and 80.14%; the prediction accuracy rate of the decision tree were 78.58%,66.05% and 85.22%.Conclusions (1) The number of the daily ER visits of URTI correlates with the meteorological and environmental changes, and it shows a certain lagged effect. Low temperature, dry and gale are proved to be the potential risks of URTI. The meteorological elements can also regarded as dynamic preventable factors of the URTI. (2) GAM is the best model in predicting the diseases, which can eliminate other confounding effects in this research. However, the effectiveness of these models needs to be testified more by further studies.