Sources of Error in Office Blood Pressure Measurement

Roy N. Morcos, MD, FAAFP; Kimbroe J. Carter, MD; Frank Castro, MS; Sumira Koirala, MD; Deepti Sharma, MD; Haroon Syed, MD


J Am Board Fam Med. 2019;32(5):732-738. 

In This Article


Terminal Digit Bias

Table 1 presents the distribution of terminal digits for 1 of the nurses using a manual device. In the absence of TDB, the predicted prevalence of each terminal digit is 150 and expected to be equally distributed. In this example, 350 measurements ended with a zero, indicating highly significant bias for that digit (P < .01). A preference for the terminal digit zero was found for each of the 3 nurses when measuring BP with a manual device, Table 2. The degree of TDB varied between the nurses but was highly significant for each (P < .01). No such TDB was found for any of the nurses when BP was measured with the automated device.

Effect of Patient Positioning on BP Measurements

The outcomes of the randomized controlled trial showed that the sequence of patient positioning (table to chair position followed by chair to table position vs chair to table position followed by table to chair position) did not affect the differences in BP. The results of the χ 2 test for independence with 1 degree of freedom were 0.37 for systolic BP and 1.00 for diastolic BP. These findings were not significant for the positional effect at a critical value of 3.84, representing 95% probability of no difference. Thus, the sequence of table position first followed by chair position was adhered to in the subsequent part of the study.

The results of BP measurements comparing table and chair positions are summarized in Table 3. The BP was significantly lower in the chair position compared with the table position in 128 individuals (43.5%). Compared with BP measurements in the table position, 46 patients (15.6%) would have been misclassified with prehypertension, and 48 patients (16.3%) would have been misclassified with hypertension, based on the JNC-7 definition.

Based on the ACC/AHA guideline, 4.8% of patients would have been misclassified as having elevated BP, and 20.1% of patients would have been misclassified as having hypertension (Table 3). Logistic regression analysis of age, sex, the presence of diabetes, cardiovascular disease, hypertension, hyperlipidemia, and smoking showed that these independent factors did not predict significant diastolic BP lowering with change in patient positioning (Table 4). Interestingly, cardiovascular disease was associated with more systolic BP lowering while hyperlipidemia was associated with less systolic BP lowering.