Terminal digit bias in a specialty hypertension faculty practice

Traditional blood pressure (BP) methodology is subject to observer error such as terminal digit preference and single number preference leading to inaccuracies in measurement. A high percentage (60-90%) of terminal BP readings digit being zero has been reported from general medical- and hospital-based clinics. This study examined terminal digit preference in a hypertension specialty practice and assessed clinical factors that may be associated with zero preference in this setting. A retrospective chart review of patients presenting to the hypertension clinic at the University of Connecticut Health Center during the month of September 2001 was performed. Data were extracted on age, gender, height, weight, treatment status, and systolic and diastolic BP measurements taken by nursing staff and attending physicians. Terminal digit preference was apparent in BP readings taken by both nursing staff and physicians. Zero was the terminal systolic BP digit in 40% of readings taken by the nursing staff and 31% of readings taken by physicians. For diastolic BP readings, the percentages were 23 and 36%, respectively. Nurses also recorded 43% of diastolic BP readings with terminal digit 2. Age was significantly higher in those persons in whom the physician diastolic BP terminal digit was zero than in those with nonzero terminal digits (67+/-14 vs 59+/-18 years, P=0.008). Body mass index was lower in the patient group with diastolic terminal digit zero bias compared to those with nonzero terminal digits (28+/-5 vs 32+/-6 kg/m(2), P=0.02). In conclusion, although the frequency of zero digit preference did not reach the 60-80% levels found in previous studies, there was evidence of terminal digit preference in the systolic and diastolic measurements taken by nursing staff and attending physicians in a specialist hypertension clinic. We believe that the lower levels of terminal digit preference observed are an effect of increased training in proper BP measurement and technique. However, the observed bias in measurement even in a hypertension unit argues for regular monitoring and feedback to minimize such errors.