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


An accurate and reliable BP measurement is essential for diagnosing and managing hypertension. The family physician is ideally positioned to identify an early BP rise in asymptomatic individuals and can have a major influence on reducing hypertension-related morbidities. To achieve this goal, meticulous attention to the BP measurement technique and instruments must be followed. Sources of error may be due to the equipment used or to the individual measuring the BP. In this study, we have examined 2 such sources of error that family physicians are likely to encounter. The importance of minimizing errors is confirmed in a large meta-analysis showing that a decrease in systolic BP by 10 mm Hg results in significant reduction in the risk of coronary artery disease, stroke, and heart failure.[5] Another large study by Greiver[6] showed that TDB decreased from 26.6% to 15.4% since the acquisition of automated devices, and patients in sites with a high level of TDB had a higher frequency of strokes, acute myocardial infarction, and angina. This highlights the relevance of TDB and the clinical importance of minimizing or eliminating it. This study confirms and expands on previously reported findings regarding the behavior and limitations of BP measurements.[12] First, we identified a highly significant TDB for all 3 nurses tested using a manual device. There was a bias for the number zero as a terminal digit. No such bias was identified with any of the same 3 nurses using an automated device. This TDB was previously described in numerous other studies,[19,21–29] most of which show that TDB is reduced but not completely eliminated by the introduction of automated devices in measuring BP. Myers and Campbell[11] found evidence of TDB of 14% of readings when using the BpTRU automated device, when the expected proportion of zero terminal digits is 10%. In another study, no TDB was identified when BP were measured with a BpTRU device although the actual data are not shown.[31] It is important to note that not all automated devices are necessarily similar since the BP is not directly measured but calculated based on a proprietary algorithm that differs according to each manufacturer. One study by Mengden[30] showed that use of automated devices minimized TDB but there was another bias in data recording because BPs were clustered around therapeutic cutoff levels. In our study, there was no evidence of TDB when BP was measured with the OMRON automated device.

We previously reported that BP measurements obtained with the patient sitting on the examination table rather than in a chair often results in elevated levels, which can lead to misclassification of hypertension.[9] A concern was raised about the chosen sequence, table first then chair position, and whether the opposite sequence would have had the same effect. A need for randomization of the order of BP measurements was also discussed in a recent review of the implications of BP measurement techniques.[10] Herein, we found that the sequence of BP measurements did not affect the difference in BP between the 2 positions. No previous studies have evaluated the difference in BP between table versus sitting positions and the impact of such improper positioning on misclassification of prehypertension and hypertension when BP is measured with an automated device. Lacruz et al[17] found a significant increase in BP in the sitting versus lying-down position.

We previously found that the chair position resulted in a significant decrease in BP compared with the table position in 30.4% of patients when using a manual device.[9] In this study, we also found that the chair position resulted in a significant and even greater decrease in BP compared with the table position in 42.7% of patients when using an automated device. Further, we found more misclassification of prehypertension and hypertension using either the JNC-7 or ACC/AHA guidelines when the BP is predominantly measured with the automated compared with the manual method. The reasons for these differences between devices are not known but may result from another type of observer bias. When using a manual device, an observer's knowledge of the BP initially measured in the table position may affect the BP reading in subsequent measurements. This is an example of anchoring bias,[16] which is not expected to occur with an automated device. Further studies exclusively using automated devices are needed to confirm the lack of anchoring bias. Furthermore, highly significant TDB was identified with the use of a manual but not with an automated device.


A weakness of our study is that BP was measured only twice in each position. Other studies have obtained 3 or more measurements to ensure a stable and reliable BP level.[13,17] However, in a recent study of BP measurements and mortality, only 2 measurements were obtained, and an average was calculated.[18] We have similarly chosen to obtain only 2 BP readings to better simulate real-life conditions that are likely to be encountered in a busy primary care practice.