CHICAGO — Terminal digit bias analysis can be used to gauge the accuracy of blood pressure measurements at primary care clinics, a new study suggests.
Terminal digit bias is the tendency of an observer to round off a measurement to a digit of his or her own choosing. It has been shown in several studies to cause significant discrepancies in BP measurements in both the clinical and research settings.
No clear guidelines have been established for what is an acceptable range of end-digit preference, especially for zero—the most common terminal digit recorded. One study proposed a “0” end-digit preference by BP observers of less than 25% as being “optimal,” less than 30% as “acceptable,” and less than 35% as “minimal standard” (J. Clin. Epidemiol. 1996;49:869-77).
In the current study, Dr. Ahmed Dalmar and associates measured the terminal digit preference of 2,202 blood pressure readings at an inner-city hospital before caregiver training and 2,154 readings 3 months after training, which included a video on proper techniques.
Terminal digit bias for zero significantly decreased among systolic and diastolic BP measurements from 57% (a very minimal standard) before training to 16% (optimal) after training, Dr. Dalmar, a research scientist with Aurora Health Care in Milwaukee, reported in a poster at a meeting sponsored by the International Society on Hypertension in Blacks.
The use of 2, 4, 6, and 8 as the last digit significantly increased after training, although the terminal-digit bias remained in the optimal range post intervention for all of the digits except 2, which fell into the minimal standard range.
The goal was to decrease or eliminate the zero-digit bias, with the hope that subsequently the bias for other digits would be in the acceptable or optimal range, Dr. Dalmar said in an interview.
“Unfortunately this did not happen in our study and, yes, these digits are also prone to bias,” he said. “We think this happened because after training many providers tended to avoid zero.”
The researchers disclosed no conflicts of interest and received no funding for the study.