Repository logo
 

Impact of Blood Pressure Measurement Errors on CVD Risk Prediction in New Zealand

Files

Thesis(3.54 MB)

Date

2025

Authors

Supervisor

Miranda, Victor
Lowe, Andrew
Lee, Tet Chuan

Item type

Thesis

Degree name

Doctor of Philosophy

Journal Title

Journal ISSN

Volume Title

Publisher

Auckland University of Technology

Abstract

Cardiovascular disease (CVD) includes conditions affecting the heart and blood vessels, such as coronary artery disease, stroke, and heart failure. It is a leading global cause of death, increasing healthcare costs and reducing quality of life. CVD develops due to genetic, lifestyle, and environmental factors, making early identification and prevention essential. Accurate CVD risk prediction helps identify high-risk individuals, enabling timely interventions. Risk models estimate cardiovascular event likelihood based on factors like age, sex, smoking, diabetes, blood pressure (BP), and cholesterol, guiding clinical decisions and treatment strategies. BP is a key CVD risk factor and is widely used in risk prediction equations due to its strong association with cardiovascular events. Accurate BP measurement is essential, as errors can misclassify individuals into incorrect risk categories. These errors stem from factors such as improper cuff size, white-coat hypertension, observer bias, and device inaccuracies. Previous studies found that rounding BP to zero end-digits mostly led to overtreatment. New Zealand introduced the PREDICT-1 equation in 2018, but the impact of rounding on this model remains unknown. Automated BP measurement devices, though widely used, have inherent inaccuracies. International standards such as ISO and ANSI/AAMI SP10 acknowledge these errors and set acceptable thresholds for reliability. However, some degree of error remains unavoidable, and its impact on CVD risk prediction is not fully understood. Additionally, many clinical studies on device accuracy do not adhere to international standards, either exceeding allowable error margins or using insufficient sample sizes (N<85). The implications of using smaller samples remain unexplored, leading to inconsistencies in adherence to these criteria. Objective: The main objective of this research is to evaluate the impact of BP measurement errors, specifically rounding to the nearest zero end-digit and device inaccuracies, on CVD risk prediction in the NZ population. This study also aims at examining how BP measurement errors affect risk estimation for Māori and Pacific individuals compared to Europeans. Furthermore, this research aims at providing a new statistical framework to assess the impact of not adhering to the international standards and necessary modifications to be adapted to use smaller sample sizes. Findings: The results show that even though there were slight variations in the overall Cox PH model predictability, there were notable changes in the risk classification. With just rounding approximately 4.24% of high-risk men and 3.21% of high-risk women are misclassified into lower-risk categories. Additionally, 1.19% of men and 0.62% of women are overclassified into the moderate-risk group, while 0.47% of men and 0.20% of women are overclassified into the high-risk group. Over 5 years, these misclassifications could cost the healthcare system approximately NZD $1.57 million, with potential expenses reaching up to NZD $8.2 million.. After adjusting the BP readings for the device inaccuracies within the acceptable range, the maximum observed misclassification rates showed that up to 7.50% of men were overclassified into higher-risk categories, which is around 5.65% for women. The findings also suggest disparities among ethnic groups, with Māori exhibiting the highest rates of misclassification, hence increasing the risk of both undertreatment and overtreatment. Additionally, the results also highlight the changes in the acceptance region for different sample size to ensure adherence to the existing standards. The use of proposed methodology resulted in identifying various clinical studies not adhering to the criteria. Conclusion: While risk equations like PREDICT-1 play a crucial role in guiding clinical decisions and improving health outcomes, the continued use of measurement practices with known errors, without fully understanding their consequences, remains a concern. The results confirm that even small inaccuracies in BP measurements can lead to notable misclassification, resulting in both overtreatment and undertreatment, with direct implications for healthcare costs. These results emphasise the importance of improving BP measurement practices to enhance the accuracy of CVD risk prediction and ensure equitable healthcare outcomes.

Description

Keywords

Source

DOI

Publisher's version

Rights statement

Permanent link

http://hdl.handle.net/10292/19936

Collections

Doctoral Theses