You are on page 1of 4

articles nature publishing group

See COMMENTARY page 246

Comparison Between an Automated and Manual
Sphygmomanometer in a Population Survey
Martin G. Myers1, Natalie H. McInnis2, George J. Fodor2 and Frans H.H. Leenen2

Background Results
An automated sphygmomanometer, the BpTRU, was used in a blood The mean s.d. BP with the automated device was 115 ± 16/71 ±
pressure (BP) survey of 2,551 residents in the province of Ontario. 10 mm Hg compared to 118 ± 16/74 ± 10 mm Hg for the manual BP
Automated BP readings were compared with measurements taken (P < 0.001). A systolic BP ≥140 mm Hg was present for 16 automated
by a mercury sphygmomanometer under standardized conditions and 19 manual readings. Similarly, the diastolic BP was ≥90 mm Hg
in a random 10% sample. for 9 automated and 14 manual readings. Linear regression analysis
showed that automated BP was a significant (P < 0.001) predictor of
Methods both manual systolic and diastolic BP.
BP was recorded in 238 individuals in random order using
both a standard mercury device and an automated BP recorder, Conclusion
the BpTRU. All subjects rested for 5 min prior to the first Conventional manual BP readings can be replaced by readings taken
BP reading, which was then discarded. The mean of the using a validated, automated BP recorder in population surveys.
next three readings was obtained using the mercury device The slightly lower readings obtained with the BpTRU device (in the
whereas the BpTRU was set to record a mean of five readings context of reduced observer–subject interaction) may be a more
taken at 1 min intervals with subjects resting alone in a accurate estimate of BP status.
quiet room. Am J Hypertens 2008; 21:280-283 © 2008 American Journal of Hypertension, Ltd.

Blood pressure (BP) surveys in the community have gener- recording device, which is designed specifically for the office
ally been performed using mercury sphygmomanometry. This or clinic setting. The most widely studied device to date is the
approach requires extensive training for health professionals BpTRU (BpTRU Medical Devices, Coquitlam, BC, Canada)
in order to standardize procedures for BP measurement, so which is able to perform five readings with subjects rest-
that accurate readings can be taken for thousands of subjects ing quietly and alone in an examining room, thus decreasing
by different study personnel. Rigorous programs for monitor- subject–observer interaction, reducing anxiety and decreasing
ing quality control also need to be initiated in order to ensure observer measurement error, all of which would tend to give
continued compliance with the study protocol for BP mea- a better estimate of an individual’s BP status. In hypertensive
surement until the survey is completed. The net result is an populations, the BpTRU has been shown to reduce the white-
estimate of the BP status of a population based upon conven- coat effect with readings exhibiting a significantly higher cor-
tional BP readings, but obtained at great expense with exten- relation with the mean waking ambulatory BP, the current
sive human resource efforts in training and execution. gold standard for assessing BP status, compared to BP readings
The Ontario Survey on the Prevalence of High Blood obtained in routine clinical practice.1,2 Mean BpTRU readings
Pressure (ON-BP) decided to consider alternatives to the con- have exhibited a close approximation to the manual BP in both
ventional measurement of BP using the mercury sphygmoma- validation studies3,4 and in clinical practice.5
nometer, in order to simplify the process of evaluating the BP The ON-BP was undertaken to record the BP of about 3,000
status of a large number of individuals in the community. An randomly selected subjects residing in the community in the
attractive alternative was the use of a validated automated BP province of Ontario. In a random sample of 10% of participants,
BP readings were taken with both the BpTRU and a mercury
1Division of Cardiology, Schulich Heart Centre, Sunnybrook Health Sciences sphygmomanometer under similar standardized conditions in
Centre, Department of Medicine, University of Toronto, Toronto, Ontario, order to compare the automated BP readings with the conven-
Canada; 2Hypertension Unit, Division of Cardiology, University of Ottawa Heart
Institute, Ottawa, Ontario, Canada. Correspondence: Martin G. Myers tional standard for BP surveys, the mercury device.
Received 10 September 2007; first decision 7 October 2007; accepted 17 November Methods
2007; advance online publication 24 January 2008. doi:10.1038/ajh.2007.54 Patient population. Adult residents of the province of Ontario,
© 2008 American Journal of Hypertension, Ltd. aged 20–79 years, were randomly selected for enrollment in


AMERICAN JOURNAL OF HYPERTENSION | VOLUME 21 NUMBER 3 | MARCH 2008 281 . Linear BpTRU (model BPM 300) after patients rested alone in a quiet regression analysis showed that the automated systolic BP is a room for 5 min. To assess the differences 40 40 between males and females an independent t-test was used to analyze the delta systolic and delta diastolic values.70) readings (Figure 3). A total the two methods were tabulated and displayed as frequency of 2. histograms (Figures 1 and 2). and diastolic BP between the automated and manual measure- ments. There was a 2-min break between automatic and manual BP readings. to see if they were interested in participating in the survey and Individual readings for systolic and diastolic BP for each of attending a BP clinic (see Leenen et al.551 individuals attended the special clinic for BP measure. Figure 1 | Histogram representing number of participants (n = 238) for each stolic BP—automated diastolic BP. each 1 min The comparison of the manual versus the automated readings apart and the mean value and heart rate were recorded for each showed a high coefficient of correlation (r2) for systolic BP (r2 = individual. respectively. The differences between systolic and left arm at heart level. the auto- BP measurement. those Mean (±s. ance was used to analyze the differences between delta systolic and delta diastolic by age and cuff size.93 × automated BP reading and heart rate were noted for each sub- ject. Mean manual and automated systolic and dia. A one way analysis of vari. of these. For the automated/manual order of readings. The mean with a regression equation (Adjusted systolic BP = 11.d. The device was set to take six readings at 1 min significant predictor (β = 0. 40 60 80 100 40 60 80 100 Automated diastolic BP (mm Hg) Manual diastolic BP (mm Hg) ticipated in this substudy. mean (±s.A. and cuff size did not predict the differences in systolic appropriate cuff size was used for each device. cury sphygmomanometer. Fifty-nine subjects remembered 5 mm Hg intervals of manual and automated diastolic blood pressure (BP). The device was positioned in such a way as to prevent the 40 40 subject from seeing the readings as they were taken. They were allowed to rest for 5 min manual readings and systolic automated readings were signifi- in a quiet room before the first reading was taken. NY) according participants having manual BP followed by automated BP read- to the procedures of the Canadian Hypertension Education ings was 113 ± 17/70 ± 11 for the automated versus 117 ± 17/ Program. Parameters such as device first.d. 20 20 mated systolic and diastolic BPs as the independent variables. 10 10 Results 0 0 A 10% sample (n = 238) of the entire survey population par. A paired 0 0 t-test was used to assess the differences between the manual 80 100 120 140 160 180 80 100 120 140 160 180 Automated systolic BP (mm Hg) Manual systolic BP (mm Hg) BP and automated BP readings. Systolic BP was BP recorded using both the BpTRU device and a standard mer. respectively. The corresponding results for sphygmomanometer (W. used was determined by the individual’s study number. being told by a health professional that they had hyperten- Both men and women with and without high BP were eligible sion and.) BP taken with the automated BpTRU device was subjects with even last digits were measured with BpTRU first 115 ± 16/71 ± 10 mm Hg compared to 118 ± 16/74 ± 10 mm Hg and those with odd last digits were measured with the mercury for the manual BP readings (P < 0. 5 mm Hg intervals of automated and manual systolic blood pressure (BP). 42 were receiving antihypertensive drug for the survey.01) smaller when automated was used first. Automated BP readings were obtained using the 0. All 30 30 measurements were obtained under similar conditions except Frequency Frequency for the two different BP recording techniques used.001). Delta values were calculated as manual systolic BP—automated systolic BP and manual dia.4 + 0. Potential subjects were contacted by home visits therapy. A linear 30 30 regression analysis was performed to examine the relationship Frequency Frequency between the automated and manual BP readings with the auto. for the manual BP. then discarded. sex. for more details6). Baum. Diastolic BP was ≥90 mm Hg ment. Manual BP was recorded by a registered mated and manual values (mm Hg) were 116 ± 14/72 ± 9 and nurse or registered nurse practitioner using a Baumanometer 118 ± 15/75 ± 10. The subjects in this substudy com- prised 96 males and 142 females. this was cantly (P < 0. The left arm circumference was measured and the age.0.) age 50 ± 15 and Figure 2 | Histogram representing number of participants (n = 238) for each 47 ± 15 years. One of every 10 participants was selected to have their for 14 manual and 9 automated BP readings. Three more readings were taken. ≥140 mm Hg for 19 manual and 16 automated measurements. P < 0.Automated Recording Devices for BP Surveys articles the ON-BP survey using census data from Statistics Canada.93. 10 10 stolic BP values were tabulated as histograms showing the num- ber of participants for each 5 mm Hg interval of BP.001) of manual systolic BP intervals with the first reading being discarded. The order in which the devices were Each data set was normally distributed. All data was analyzed using SPSS version 15. Copiague. 20 20 Data analysis.7 Subjects remained seated with the back supported 74 ± 10.84) and diastolic BP (r2 = 0.

the manual readings in this sample from the ON-BP survey. the “white-coat reaction” tends to Manual diastolic BP (mm Hg) Manual systolic BP (mm Hg) 160 100 provoke a greater pressor response. BP readings of 80 and 90 mm Hg are equivalent to manual ulations. readings of 82 and 90 mm Hg.0/5. auto- aspects of bias associated with conventional BP measurement mated readings taken with the BpTRU device closely approxi- using mercury sphygmomanometry.4 mean BpTRU values for systolic diminishing as BP values increased. 8. respectively (P < 0. This conversion makes it possible to compare data ence was observed in hypertensive patients when manual read. Figure 3 | A linear regression analysis was performed to examine the Kaczorowski and colleagues11 found systolic BP readings (taken relationship between the automated and manual blood pressure (BP) in a community pharmacy using the BpTRU device). the mean values are quite similar. accurate.. In a study in clinical practice. manual Linear regression analysis of the automated and manual BP readings exceeded automated readings by 3/3 mm Hg. when the manual readings were taken by ­physicians. Estimates of the cut-point for normal BP for or reading up or down to influence the patient’s BP status. sphygmomanometer when taken under similar conditions. which is what measurement. sive patients. BP at 140/90 mm Hg is equivalent to an automated reading of which is a factor known to increase the BP. with the difference reported by Wright et al. The r2 values for systolic and diastolic BP are 0. In the above studies.83 × reaction.d. the mean manual office BP was 20/5 mm Hg higher 120 60 than the mean of five automated BpTRU readings. Thus.10 Many individuals 135/85 mm Hg taken in the office with subjects resting alone exhibit a fall in BP within a minute or two after being left alone in a quiet room. dated. too rapid deflation of the cuff. automated systolic BP were 3/2 mm Hg higher with the mercury sphygmomanometer. the BpTRU device with subjects resting alone in a quiet room cant predictor (β = 0.5 Similarly. points for a diagnosis of hypertension with the two methods of ventional manual readings obtained by a nurse. In a formal validation study ings obtained using the automated device.70. readings of 120 and 140 mm Hg are equivalent to manual read- These differences between automated and manual ­readings ings of 123 and 142 mm Hg.7/8. respectively.2 mm Hg.9% would be in a quiet room especially in the context of a treatment setting designated as being hypertensive according to a cut-point of such as a doctor’s office or clinic. However.8 This approach removes several in the presence of the observer (162 ± 27/85 ± 12).6 + 0. At lower BP values.0% with a The benefits of taking readings using an automated device normal automated BP set at <135/85 mm Hg. one would anticipate lower reliable comparative estimates of normal versus abnormal cut- mean BP values for the automated device compared to con. The routine manual versus automated (BpTRU) values are still absence of the observer from the room during readings also preliminary but the available data suggest conventional office precludes conversation between the subject and the observer. readings taken with and manual BP measurements under standardized conditions. the mean BP was sub- at specific time intervals. The number of such as the BpTRU are generally applicable to population subjects in this subset of the ON-BP was too small to provide surveys such as the ON-BP. readings. was observed in the Ontario survey.9 The role of the observer mate conventional measurements recorded with a mercury in recording the BP is eliminated and replaced with a vali.001) of manual diastolic BP with in order to minimize factors that tend to provoke a white-coat a regression equation (adjusted diastolic BP = 15. The automated BpTRU device was designed to minimize the of two readings taken using a mercury sphygmomanometer impact of observer–subject interaction on the measurement of (163 ± 23/86 ± 12) was similar to the first BpTRU reading taken BP in the office/clinic setting.2 ± 4. if one performs the automated readings automated diastolic). articles Automated Recording Devices for BP Surveys 180 Under these circumstances. In a series of 50 hypertensive 140 patients referred to a specialty hypertension centre for their man- 80 agement. data provided a “correction factor” to convert the automated This difference represents the white-coat effect for a random readings obtained in the survey into comparable manual BP sample of adult residents residing in Ontario. P < 0.1 In this instance. derived from BP surveys performed using an automated BP ings recorded by a research technician outside of the treatment recorder with previous surveys that have employed manual BP setting were compared with the mean automated readings taken measurement techniques. According to the linear regression using the BpTRU device. Discussion respectively.4 ± 4. In a population survey of persons 65 years of age and older with hypertension.84 and 0. 282 MARCH 2008 | VOLUME 21 NUMBER 3 | AMERICAN JOURNAL OF HYPERTENSION . Automated diastolic BP was also a signifi. Automated diastolic are substantially less than reported for other hypertensive pop.5 <140/90 mm Hg for normal manual BP and 9. respectively. the manual BP recorder were higher than corresponding read.83.3/−1. the automated readings were taken with automated systolic).2 In the present subset. and diastolic BP differed from reference readings taken with a standard mercury device by only –0. In this instance.5 the mean ± s. Thus. digital device programmed to take readings In this subset of the ON-BP survey. produced readings with the automated systolic and diastolic BP as the independent a mean value 9 mm Hg lower than the last routine office BP variables. 100 Beckett and Godwin2 noted a difference of 11/3 mm Hg between 80 40 the last routine office BP taken by a patient’s own family phy- 80 100 120 140 160 Automated systolic BP (mm Hg) 180 40 60 80 100 Automated diastolic BP (mm Hg) sician and the mean automated BpTRU value.001). A similar differ. thus eliminating imprecision due to stantially lower than in previous studies involving hyperten- factors such as digit preference. taken by the subject’s own family doctor.

Sullivan SM. Myers MG. Metzger JP. pharmacy setting. automated recorder. supported clinical practice. 273:1211–1216. Zorn A. the automated readings may actually reflect the Blood Press Monit 2006. Schiffrin EL. nometry can be replaced by a validated. Abbott C.F. Wilson T. vascular risk. Nemeth K.J. Validation normal or mild hypertension readings.Automated Recording Devices for BP Surveys articles The results of the ON-BP have demonstrated that conven. The BpTRU automatic blood pressure monitor compared to 24 hour ambulatory blood pressure monitoring in the assessment tional manual BP readings taken using mercury sphygmoma. Reeves RA. MM. 9:267–270. Mann K. Automated measurement of blood pressure in routine the Pfizer Chair in Hypertension Research. by Pfizer Canada. Disclosure: The authors declared no conflict of interest. true hypertension status in the population. Turton P. tionship between automated BP readings and mean waking 7. Hemmelgarn BR. Use of an automated blood pressure recording device. an endowed chair. Can J Cardiol 2007. Can Med Assoc J (in press).H. Wright JM. Grover S. 22:573–581. Valdivieso MA. Mattu GS. Penner B. Le Pailleur C. AMERICAN JOURNAL OF HYPERTENSION | VOLUME 21 NUMBER 3 | MARCH 2008 283 . Milot A. Strange KD. J Clin Hypertens 2007. 11. Gelfer ME. 1. Landais P. Dumais J. McInnis N. Myers MG. Perry TL Jr. 11:59–62. Canadian Hypertension Education Program: The 2006 Canadian Hypertension Education Program Acknowledgment: This survey was supported by a Heart and Stroke recommendations for the management of hypertension: Part 1—Blood pressure measurement. Foundation of Ontario contract awarded to F. Karwalajtys T. University of Ottawa Heart Institute Foundation. 6:161–165. Chambers LW. Myers MG. The effects of talking. and 9. Montgermont P. Helft G. Myers MG. and silence on the “White Coat” phenomenon in hypertensive patients. Myers MG. Wright JM. reading. Mattu GS. 5:18. 11:203–207.H. and G. Feder JM. ambulatory BP. Perry TL Jr. by minimizing the 6. Tobe SW. pressure. Does this patient have hypertension? How to measure blood Canadian Institutes of Health Research. Fodor G: 2006 Ontario survey on the prevalence and control of hypertension (ON-BP).H. Instead of underestimating monitor. McAlister FA. Bolli P. Comparison of the oscillometric blood pressure If the BpTRU minimizes the white-coat effect it would not be monitor (BPM-100β) with the auscultatory mercury sphygmomanometer. Use of the BpTRU. JAMA 1995. diagnosis and assessment of risk. 10. 3. Vacheron A. Blood Press Monit 2001. 16:494–497.L. 6:161–165. possibly indicating a of a new algorithm for the BPM-100 electronic oscillometric office blood pressure lower prevalence of hypertension. F. Beckett L. surprising to see lower BP readings in participants with high 4. Am J Hypertens 1998.H. hypertension. the current gold standard for assessing cardio. Tremblay G. BMC Cardiovasc Disord 2005. Am J Hypertens automated blood pressure measurement to reduce white coat response in a 2003. Chockalingam A. Godwin M.L. Stratychuk L. to reduce the “white coat effect” in routine practice. Can J Cardiol 2006. 2. Leenen FHH. Burgess E. 5. Automated blood pressure measurement in routine clinical practice. Honos G. Touyz RM. 23(Suppl C): 85C. holds 8. Blood Press Monit 2001. Kaczorowski J. Lum-Kwong white-coat effect when one takes into account the close rela. Godwin M. Chen Y. of blood pressure in patients with hypertension. Mckay DW.