The minimum number of blood pressure measurements needed and thresholds for visit-to-visit blood pressure variability to predict cardiovascular disease in primary care patients
Lukitasari, M.; Argha, R.; Liaw, S.-T.; Jalaludin, B.; Rhee, J.; Jonnagaddala, J.
Show abstract
ObjectivesVisit-to-visit blood pressure variability (VVV BPV) is an underutilised risk factor for cardiovascular disease (CVD). This study aims to determine the minimum number of BP measurements needed and to identify cut-off values for the standard deviation (SD), coefficient of variation (CV), and average real variability (ARV) of systolic and diastolic VVV BPV to predict CVD risk in primary care. MethodsWe analysed data from the electronic practice-based research network (ePBRN) in Southwestern Sydney, including patients aged 18-55 with at least eight BP readings. Patients with incomplete data or no follow-up beyond age 55 were excluded. The agreement between SD calculated from 3-5 measurements and 8 measurements (reference) was evaluated using Pearsons correlation coefficient and the intraclass correlation coefficient. Then, after identifying that a minimum of five BP measurements is needed, another cohort with at least five BP measurements was developed. Percentile-based cut-offs (10th - 90 th, 5-percentile increments) were derived for systolic and diastolic BPV (SD, CV, ARV). Predictive accuracy was assessed using the C-statistic. The outcome was the first CVD occurrence. ResultsA total of 1,549 patients were included in the first study. Five BP measurements showed good agreement with eight measurements (ICC: 0.79; correlation: 0.80). A total of 3,022 patients were included (55.2% women). Higher VVV BPV (SD, CV. ARV) was associated with increased CVD risk. Optimal cut-off values for systolic BP were 19 mmHg (SD), 14% (CV), and 15 mmHg (ARV), and for diastolic BP were 11 mmHg (SD), 12% (CV), and 11 mmHg (ARV). Predictive performance was consistent across time frames. ConclusionsThese BPV cut-offs provide clinically relevant thresholds for CVD risk prediction. At least five BP measurements are sufficient to estimate BPV for this purpose.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Trends in Awareness, Antihypertensive Medication Use and Blood Pressure Control by Race and Ethnicity Among Adults with Hypertension in the United States: A National Health and Nutrition Examination Analysis from 2011 to 2018 96%
- Elevated blood pressure in the emergency department - a risk factor for incident cardiovascular disease: An EHR-based cohort study 96%
- Derivation of an Outcome-Driven Threshold for Aortic Pulse Wave Velocity: An Individual-Participant Meta-Analysis 96%
Similar papers in this journal
- Social Networks and Cardiovascular Disease Events in the Jackson Heart Study 95%
- Joint associations of pregnancy complications and postpartum maternal renal biomarkers with severe cardiovascular morbidities: A US racially diverse prospective birth cohort study 95%
- Prevalence of and risk factors for peripheral artery disease in rural South Africa: A cross-sectional analysis of the HAALSI cohort 94%
Similar papers in this journal
- Differences between brachial and aortic blood pressure in adolescence and their impact on classification of hypertension 96%
- Within-visit Blood Pressure Variability in Children and Adolescents in the National Health and Nutrition Examination Survey (2013-2020) 96%
- Low blood pressure phenotype underpins the tendency to reflex syncope 95%
Similar papers in this journal
Similar papers in this journal
- Prediction of Cardiovascular Markers and Diseases Using Retinal Fundus Images and Deep Learning: A Systematic Scoping Review 93%
- Using ECG Machine Learning for Detection of Cardiovascular Disease in African American Men and Women: the Jackson Heart Study 93%
- Blood pressure management in 1,000 patients with CureApp HT digital therapeutics for hypertension 93%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.