Although several studies have shown the relation between temperature/atmospheric pressure and pulmonary embolism (PE), their results are inconsistent. Furthermore, diurnal temperature range (DTR) and diurnal pressure range (DPR) were not fully evaluated for their associations with hospital admissions for PE. Study subjects comprised cases of 1,148 PE treated at institutions belonging to the Tokyo Cardiovascular Care Unit Network from January 2005 to December 2012. Patient data were combined with a variety of daily local climate parameters obtained from the Japan Meteorological Agency. Every 1°C increase in the DTR at lag0 corresponded to an increased relative risk of hospital admission for PE (odds ratio [OR] 1.036, 95% confidence interval [CI] 1.003 to 1.070). In the cooler season (November to April), an increase of 1 hPa (barometric pressure) in the DPR at lag4 and lag5 was associated with an increased relative risk of hospital admission for PE (OR 1.042, 95% CI 1.007 to 1.077 and OR 0.952, 95% CI 0.914 to 0.992, respectively). An increase in the PE hospitalization rate was seen only in the cool season. Using a metropolitan database, we showed that DTR and DPR have different impacts on hospital admissions for PE. In conclusion, we found that an increase in the DTR increases the PE hospitalization rate, especially during the cooler season. The impact of DTR and DPR on PE incidence and related hospitalizations needs to be further evaluated.
Weather changes including temperature and atmospheric pressure, possibly influenced the pulmonary embolism (PE) hospitalization rate. These seasonal variations of PE occurrence were considered to be mediated through hematologic alterations such as changes in blood viscosity and coagulation. Of several weather variables, the difference between the daily maximum and minimum temperature is recently considered as one of the meteorologic indicators related to a variety of cardiovascular diseases such as acute coronary syndrome. However, the relation between these temperature/pressure changes within the day and PE admissions has not been fully evaluated. We hypothesized that a large diurnal temperature range (DTR) and diurnal pressure range (DPR) might be the sources of additional environmental stress and risk factors for PE and related hospitalizations.
Methods
This study was performed using data from the Tokyo Cardiovascular Care Unit (CCU) Network. This network operates in 71 hospitals with the help of ambulance units that are dispatched through the control room of the Tokyo Fire Department. Institutions belonging to the Tokyo CCU Network routinely submit details of all the patients treated in their CCUs; these details are recorded using survey forms. The diagnosis of PE was accepted without independent review if confirmed by pulmonary angiography, enhanced computed tomography, high-probability lung scan, or autopsy. In this study, subjects were 1,148 patients with consecutive PE treated at institutions belonging to the Tokyo CCU Network from January 2005 to December 2012. Baseline demographics of the patients are presented in Table 1 .
| Age, (years) | 65.0±16.3 |
| Male | 477(41.6 %) |
| Pulmonary embolism severity index | 96.9±41.8 |
| SpO2, (%) | 91.4±8.6 |
| Systolic blood pressure, (mmHg) | 124±60.6 |
| Mortality | 95(8.3 %) |
| Dyspnea on exersion | 697(60.7 %) |
| Chest pain | 157(13. 7 %) |
| Altered mental stasus | 130(11.3 %) |
We obtained data for a variety of daily local climate parameters from the Japan Meteorological Agency, including daily station average/maximum/minimum barometric pressure (hPa) and daily average/maximum/minimum temperature (°C). DTR was calculated by subtracting the minimum temperature from the maximum temperature for the same day. Similarly, DPR was calculated by subtracting the minimum barometric pressure from the maximum pressure for the same day.
To identify influenza epidemics, we retrieved surveillance data from the Infectious Agents Surveillance Report published by the National Institute of Infectious Diseases of Japan. We also collected the suspended particulate matter (SPM) data provided by the National Institute for Environmental Studies of Japan. We treated weekly influenza infection occurrence and mean daily SPM concentrations as a potential confounder in data analysis.
Generalized additive Poisson regression models were used to fit the relation of DTR and DPR with the number of PE hospitalizations. We used the smoothing spline, s (.) to filter out seasonal patterns and long-term trends in daily hospitalizations as well as the daily mean temperature, daily mean barometric pressure, and relative humidity. We also adjusted our data for the day of the week and dichotomous variables, such as public holidays (Holiday), influenza outbreaks, and air pollutions. We followed previous studies’ methods for selecting a priori, the model specification, and the df for the time trend, and other meteorologic variables. We used a df of 8/ y for the time trend, a df of 6 for the mean temperature of the current day (Temp 0 ) and the previous 3 days’ moving average (Temp 1–3 ), a df of 3 for the current day’s relative humidity (Humid 0 ) and a df of 3 for the mean barometric pressure of the current day (Atm 0 ). We included the day of the week; Holiday, the weekly occurrence of influenza; and daily SPM concentrations during the study period. This was done to adjust for the confounding effect of an influenza epidemic on hospital admissions.
Briefly, we set up a core model to remove long-term trends and seasonal variations and to adjust for time-varying confounders, as follows:
log ( E ( Y ) ) = α + s ( t , d f = 8 y × no . of years ) + s ( Temp 0 , d f = 6 ) + s ( Temp 1 − 3 , d f = 6 ) + s ( Humid 0 , d f = 3 ) + s ( Atm 0 , d f = 3 ) + β 1 DOW + β 2 Holiday + β 3 influenza + β 4 SPM ,
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
