PRECISE PREDICTION OF BLOOD PRESSURE LEVELS AFTER CONTROLLING FOR SEASONAL TEMPERATURE: A THREE-YEAR PROSPECTIVE COHORT OF ELDERLY POPULATION IN HONG KONG. (April 2021)

• Record Type: Journal Article
• Authors: Tsoi, Kelvin Kf; Li, Jinhui; Xue, Jiashu; Yiu, Karen Kl; Lee, Eric Kp; Wong, Samuel Ys
• Is Part Of: Journal of hypertension. Volume 39(2021)e-Supplement 1
• Publication Date: 2021-04
• Subjects: Hypertension -- Periodicals; Hypertension -- Periodicals; 616.132005
• DOI: 10.1097/01.hjh.0000746908.96693.0e
• Languages: English
• ISSNs: 1473-5598

Abstract

Abstract : Objective: Cardiovascular disease risk has been linked to the environmental temperature changes, but few studies have evaluated the long-term temperature impacts on the blood pressure (BP) projection. Design and method: Regular blood pressure records were uploaded to the data servers from elderly community centres in Hong Kong since 2016. Meteorological data were obtained from the Hong Kong observatory. The Vector Autoregressions, the most common deep learning method was adopted for systolic BP (SBP) projection. Temperature and diastolic BP (DBP) were controlled in the model. The predictions of SBP were compared with the real BP records in the last 6-month and 12-month. The mean absolute percentage error (MAPE) and the mean absolute deviation in terms of mmHg were used to describe the accuracy of prediction. Subgroup analyses were also conducted based on age, gender, social-economic status (SES), drug intake level, and medical diseases (diabetes & hypertension). Results: A total of 733 participants regularly measured BP over three years since 2016. Most of them (67.7%) were aged over 75 years and the majority are female (73.8%). The MAPE of the projection models were 1.8% for 6-month and 1.9% for 12-month BP projections. The mean absolute SBP value in the 6-month projection model was 1.66mmHg with the 95%CI range of 2.3319mmHg. between the projection model and real SBP records. Moreover, the temperature increment of one unit in degree Celsius was associated with a 0.597 mmHg increased in SBP. The temperature effect on BP was more prominent for the participants with hypertension. Subpopulation analysis showed that the impact of gender (p = 0.005) and hypertension status (p = 0.0416) had more significant impacts on the blood pressure projection among the elderly participants. Conclusions: Seasonal temperature showed as a critical factor for the evaluation of hypertension. Understanding the blood pressure readings after removal of seasonal temperatures will help for a better diagnosis.