Terminal Digit Preference and Threshold Avoidance in Digital Blood Pressure Measurements During Pregnancy: Secondary Analysis of Data From the CLIP and PRECISE Cohorts

Background: Screening for, detecting, and managing pregnancy hypertension is a core function of antenatal care. To reduce both training requirements and the risks of measurement error in blood pressure (BP) values, automated and semiautomated BP devices have been validated in pregnant women with normal BP and pregnant women with hypertension and introduced for serial antenatal measurement of BP. Objectives: The study aimed to (1) determine whether or not repeated BP measurements reduced the presence of terminal digit preference and (2) discern whether or not there was evidence of threshold avoidance in the Community-Level Interventions for Preeclampsia (CLIP) trials compared with the purely observational Pregnancy Care Integrating Translational Science, Everywhere (PRECISE) cohorts. Methods: The BP 3AS1-2 and CRADLE Vital Signs Alert low-cost Microlife BP devices were used by trained research staff in the CLIP trials conducted in India, Mozambique, Nigeria (pilot trial only), and Pakistan and the PRECISE cohorts of unselected pregnant women and nonpregnant women of reproductive age recruited in the Gambia, Kenya, and Mozambique. Both devices algorithmically calculate systolic blood pressure and diastolic blood pressure values displayed on digital read-outs. All BP readings were entered manually into a digital platform, which averaged them as the BP for that visit; the first and second readings were averaged unless they were more than 10 mm Hg different, which triggered a third reading, and the second and third readings were averaged. Results: A total of 51,875 participants had their BP measured 438,404 times. Using raw BP values, there was terminal digit preference (129,539/911,500, 14.21% vs 10%; P<.001 values ended in zero). A total of 28,929 out of 437,446 (6.61%) dBP values were 62 mm Hg, compared with 9310 of 195,349 (4.77%) from the averaged values (P<.001); errors were obviated by averaging BP values. There was evidence of both threshold preference and avoidance in the CLIP trials and the PRECISE cohort. Conclusions: Given the excess of 62 mm Hg values, there is a shared inherent algorithmic error in the calculation of dBP in the BP 3AS1-2 and CRADLE Vital Signs Alert devices. Averaged BP measurements are important to reduce the impact of user errors in manually recording BP values. We recommend that automated and semiautomated BP devices should be connected wirelessly to automatically transfer readings to digital health records to further optimize care.