The impact of unrecorded readings on the precision and diagnostic performance of home blood pressure monitoring: a statistical study.

Abstract

Hypertension is a major cardiovascular risk factor. To address the disease adequately, most clinicians rely on home blood pressure monitoring (HBPM). However, the impact of unrecorded BP values on the precision and diagnostic performance of BP schedules is unknown. We obtained 103 HBP patients schedules from a previous study. Then, readings were randomly removed from each schedule in order to create new incomplete schedules using a resampling technique. We obtained 10,000 new incomplete schedules. For each number of randomly removed readings, the percentages of incomplete schedules outside a systolic/diastolic blood pressure (SBP/DBP) range of 5/3 mmHg were calculated from the same complete patient's schedule. The sensitivity and specificity of incomplete HBPM schedules regarding BP control were also assessed. One hundred three HBPM schedules were analyzed. Mean patients' age was 67.9 ± 9.9 years. In non-diabetic patients, the mean BP of complete schedules' means was 131.9 ± 12.4/75.5 ± 10.5. In diabetic patients, the mean BP of complete schedules' means was 135.5 ± 14.0/73.4 ± 8.2 mmHg. When schedules were composed of 14 and 21 random measures, differences over 5 mmHg were seen in 2.6% and 0.1% of non-diabetic patients' schedule and 3.7% and 0.1% of diabetic patients' schedule, respectively. At 21 measurements, sensitivity and specificity were approximately 95% and 98% in non-diabetic patients and 90% and 99% in non-diabetic patients, respectively. HBPM precision and diagnostic performance improve rapidly with accumulation of readings. Incomplete schedules composed of 21 readings can provide an almost perfect diagnostic tool compared with the complete schedule reference.