Abstract
The presence of systematic errors caused by terminal digit preference was examined in the blood pressure measurement of 9696 female subjects (40 to 69 years old) who participated in general health examinations for the elderly conducted by 195 municipalities extracted from all prefectures in 2000. Terminal digit preference equality was tested for even numbers with normal mercury sphygmomanometer and for 0 to 9 with automated blood pressure measurement devices. The results were as follows;
1) In 6560 subjects, both systolic and diastolic blood pressure values by mercury sphygmomanometer had a significant high frequency of terminal digit preference for zero (χ2=772, p<0.001, χ2=1855, p<0.001, respectively).
2) In 2919 both systolic and diastolic blood pressure terminal digit preference equality by automated blood pressure measurement devices was denied for 0 to 9 (χ2=53.4, p<0.001, χ2=72.9, p<0.001, respectively), and even terminal digit numbers occurred more frequently than odd numbers significantly (χ2=36.6, p<0.001, χ2=38.4, p<0.001, respectively).
3) When terminal digit zero readings of this 2000 survey were compared with the 1985 survey for both systolic and diastolic blood pressure readings, the rate of zero readings in even numbers were decreased from 41.1% to 33.7% for systolic and from 45.5% to 41.5% for diastolic.
Because of the classification of blood pressure levels which have usually a boundary value of terminal digit of zero for example 140-159 mmHg or 90-99 mmHg, the difference of the rate of zero in measured values have an influence on the prevalence of hypertension
