Research

JAMA Internal Medicine | Original Investigation

Arm Position and Blood Pressure Readings

The ARMS Crossover Randomized Clinical Trial
Hairong Liu, MHS; Di Zhao, PhD; Ahmed Sabit, MS; Chathurangi H. Pathiravasan, MS, PhD;
Junichi Ishigami, MD, MPH; Jeanne Charleston, RN, BSN; Edgar R. Miller III, MD, PhD;
Kunihiro Matsushita, MD, PhD; Lawrence J. Appel, MD, MPH; Tammy M. Brady, MD, PhD

Supplemental content
IMPORTANCE Guidelines for blood pressure (BP) measurement recommend arm support on a
desk with the midcuff positioned at heart level. Still, nonstandard positions are used in clinical
practice (eg, with arm resting on the lap or unsupported on the side).

OBJECTIVE To determine the effect of different arm positions on BP readings.

DESIGN, SETTING, AND PARTICIPANTS This crossover randomized clinical trial recruited adults
between the ages of 18 and 80 years in Baltimore, Maryland, from August 9, 2022, to June 1,
2023.

INTERVENTION Participants were randomly assigned to sets of triplicate BP measurements

with the arm positioned in 3 ways: (1) supported on a desk (desk 1; reference), (2) hand
supported on lap (lap), and (3) arm unsupported at the side (side). To account for intrinsic BP
variability, all participants underwent a fourth set of BP measurements with the arm
supported on a desk (desk 2).

MAIN OUTCOMES AND MEASURES The primary outcomes were the difference in differences in
mean systolic BP (SBP) and diastolic BP (DBP) between the reference BP (desk 1) and the 2
arm support positions (lap and side): (lap or side − desk 1) − (desk 2 − desk 1). Results were
also stratified by hypertensive status, age, obesity status, and access to health care within the
past year.

RESULTS The trial enrolled 133 participants (mean [SD] age, 57 [17] years; 70 [53%] female);
48 participants (36%) had SBP of 130 mm Hg or higher, and 55 participants (41%) had a body
mass index (calculated as weight in kilograms divided by height in meters squared) of 30 or
higher. Lap and side positions resulted in statistically significant higher BP readings than desk
positions, with the difference in differences as follows: lap, SBP Δ 3.9 (95% CI, 2.5-5.2) mm
Hg and DBP Δ 4.0 (95% CI, 3.1-5.0) mm Hg; and side, SBP Δ 6.5 (95% CI, 5.1-7.9) mm Hg and
DBP Δ 4.4 (95% CI, 3.4-5.4) mm Hg. The patterns were generally consistent across
subgroups.

CONCLUSION AND RELEVANCE This crossover randomized clinical trial showed that commonly
used arm positions (lap or side) resulted in substantial overestimation of BP readings and may
lead to misdiagnosis and overestimation of hypertension.

TRIAL REGISTRATION ClinicalTrials.gov Identifier: NCT05372328

Author Affiliations: Author

affiliations are listed at the end of this
article.
Corresponding Author: Tammy M.
Brady, MD, PhD, Department of
Pediatrics, Johns Hopkins University
School of Medicine, 200 N Wolfe St,
The David M. Rubenstein Child Health
JAMA Intern Med. doi:10.1001/jamainternmed.2024.5213 Building, Room 3062, Baltimore, MD
Published online October 7, 2024. 21287 ([email protected]).

(Reprinted) E1
© 2024 American Medical Association. All rights reserved.
Downloaded from jamanetwork.com by UFBA, Bruna Barreto on 11/03/2024
 Research Original Investigation Arm Position and Blood Pressure Readings

H
ypertension is the leading cause of cardiovascular dis-
ease and preventable mortality worldwide.1 Accu- Key Points
rate blood pressure (BP) measurement is a corner-
Question What is the effect of commonly used arm positions on
stone of hypertension diagnosis and management. The latest blood pressure (BP) measurements compared to the standard,
clinical practice guidelines emphasize several key steps for ac- recommended position?
curate measurement, including appropriate cuff size selec-
Findings This crossover randomized clinical trial of 133 adults
tion, back support, feet flat on the floor with legs uncrossed,
showed that supporting the arm on the lap overestimated systolic
and appropriate arm position (ie, midcuff positioned at heart BP by 3.9 mm Hg and diastolic BP by 4.0 mm Hg. An unsupported
level with arm supported on a desk or table). Despite these rec- arm at the side overestimated systolic BP by 6.5 mm Hg and
ommendations, proper arm position is commonly over- diastolic BP by 4.4 mm Hg, with consistent results across
looked in daily practice.2,3 For example, in the US, BP is often subgroups.
measured with patients seated on an examination table with- Meaning Commonly used, nonstandard arm positions during BP
out any arm support or with inadequate support (eg, resting measurements substantially overestimate BP, highlighting the
on their lap or supported by health care professionals holding need for standardized positioning.
the patient’s arm). In resource-limited settings, a desk or table
for arm support is often unavailable.
Few studies have rigorously evaluated the effects of arm po- (hereafter, lap), and (3) arm unsupported on the side (hereaf-
sition on BP. Previous studies documenting statistically signifi- ter, side). To account for intrinsic BP variability, all partici-
cant BP overestimation when the arm is unsupported or is po- pants underwent a fourth set of triplicate BP measurements
sitioned with the BP cuff lower than heart level were limited by with the arm supported on a desk with midcuff at midheart
suboptimal design (eg, nonrandomized comparisons with the level (hereafter, desk 2), which is the same condition as desk
reference condition, small sample size, or evaluations in which 1. Thus, each participant underwent a total of 12 BP measure-
patients were supine or standing).4-7 In this context, we per- ments (3 sets of triplicate measurements in randomized or-
formed a crossover randomized clinical trial comparing 3 seated der plus 1 set of triplicate measurements with the arm on the
arm positions: (1) the standard reference position (arm sup- desk [desk 2]) (Figure 1).
ported on a desk with midcuff at heart level), (2) arm resting on An institutional review board at the Johns Hopkins
the participant’s lap, and (3) arm unsupported on the partici- University School of Medicine reviewed and approved the pro-
pant’s side while adhering to all other recommended BP mea- tocol (Supplement 1).8 All participants provided written in-
surement steps in each condition. We also investigated whether formed consent, and the study adhered to the Consolidated
hypertensive level of systolic BP (SBP), older age, obesity sta- Standards of Reporting Trials (CONSORT) reporting guidelines.
tus, and no access to health care within the past year affected
the effect of arm position on BP readings. Study Population
Eligible participants were adults between the ages of 18 and
80 years. We excluded individuals presenting with 1 or more
of the following conditions: rashes, gauze dressings, casts,
Methods edema, paralysis, tubes, open sores or wounds, or arteriove-
Study Design nous shunts on both arms; mental impairment; pregnancy; or
This was a randomized crossover trial conducted among adults a mid-upper arm circumference of more than 55 cm.
in Baltimore, Maryland. The 3 measurement conditions that
were conducted in random order were (1) arm supported on Recruitment
a desk with midcuff at approximately midheart level (hereaf- From August 9, 2022, to June 1, 2023, we recruited partici-
ter, desk 1; reference), (2) hand supported on the lap pants using multiple approaches: (1) BP screenings at a public

Figure 1. Randomization Diagram

133 Participants

133 Randomized

12 Desk 1, lap, 25 Desk 1, side, 22 Lap, desk 1, 23 Lap, side, 20 Side, lap, 31 Side, desk 1,
side, desk 2 lap, desk 2 side, desk 2 desk 1, desk 2 desk 1, desk 2 lap, desk 2

Participants were randomly assigned to 1 of 6 groups to ensure that each (lap), and (3) arm unsupported at the side (side). To account for intrinsic blood
participant experienced all blood pressure measurement conditions in a pressure variability, all participants underwent a fourth set of triplicate
sequence designed to minimize any potential order effects and biases. measurements with the arm supported on a desk with midcuff at midheart level
Measurement conditions included (1) arm supported on a desk with midcuff at (desk 2).
approximately midheart level (desk 1; reference), (2) hand supported on the lap

E2 JAMA Internal Medicine Published online October 7, 2024 (Reprinted) jamainternalmedicine.com

© 2024 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by UFBA, Bruna Barreto on 11/03/2024
 Arm Position and Blood Pressure Readings Original Investigation Research

food market located near the Johns Hopkins University School BP readings was taken, with measurements spaced 30
of Medicine; (2) direct, personalized mailings to previous study seconds apart, using an upper-arm cuff selected based on the
participants; (3) informational brochures about the trial placed participant’s measured mid-upper arm circumference. On com-
at Johns Hopkins University hypertension clinics; and (4) rec- pleting the initial set of triplicate BP measurements, the cuff
ommendations from physicians who specialize in treating in- was removed, and the participant walked for another 2 min-
dividuals with hypertension. utes. After resting again for 5 minutes, another set of 3 BP read-
The research team collected self-reported data from par- ings was obtained in the same manner. This cycle was re-
ticipants, including age, sex, racial and ethnic background, peated until 4 sets of triplicate BP measurements (totaling 12
weight (body mass index [BMI; calculated as weight in kilo- readings) were completed. All of the measurements were con-
grams divided by height in meters squared]), and medical his- ducted in a quiet and private space, and participants were asked
tory of hypertension, diabetes, chronic kidney disease, and not to talk to researchers or use their smartphones during BP
myocardial infarction, along with the use of antihypertensive measurements. BP used in the analysis was the average of the
medication. Participants were asked to report the approxi- triplicate BP measurements in each set.
mate date of their last health care professional visit (for either
acute or chronic care); health care utilization in the past year Outcomes
was dichotomized as any health care visit in the past 365 days The primary outcomes were the difference in differences in
(yes/no). mean SBP and diastolic BP (DBP), between the reference BP
(desk 1) and the 2 arm positions (lap and side). Specifically, the
Sample Size difference between (1) lap and desk 1 (lap − desk 1) and (2)
Assuming 80% power and a type I error probability of a 2-sided side and desk 1 (side − desk 1) was compared to the difference
α of .05, the target sample size was at least 100 participants between desk 2 and desk 1 (desk 2 − desk 1). The secondary out-
to allow us to detect a clinically meaningful difference of 2.5 comes were the difference in differences in mean SBP and DBP
mm Hg, based on observed standard deviation of BP differ- among subgroups. In the subgroup analysis, we examined dif-
ences in our previous studies of 8 to 10 mm Hg.9,10 To allow ferences in the primary outcomes by hypertensive BP status
for prespecified subgroup analyses, we tried to enrich the num- (SBP of ≥130 mm Hg vs <130 mm Hg), age (≥60 years vs
ber of individuals with SBP of 130 mm Hg or higher; thus, the <60 years), obesity status (BMI of ≥30 vs <30), and health care
final sample size exceeded 100 participants. utilization (no health care visit vs ≥1 health care visit in the past
365 days).
Randomization
Using the RANDBETWEEN function in Excel (Microsoft), we Statistical Analysis
created randomization allocations; the allocation table was up- Using paired t tests, we assessed the difference in differences
loaded to REDCap (Vanderbilt University).11 After consent, re- of the mean BPs obtained when the arm was in a nonstan-
search staff accessed REDCap to determine the participant’s dard position (lap or side) and when the arm was positioned
randomization assignment; there was no way for the re- properly (desk). Specifically, for SBP and DBP, we determined
search staff to know the order of measurement conditions be- (lap − desk 1) − (desk 2 − desk 1) and (side − desk 1) − (desk
fore this step. Participants were randomly assigned to 1 of 6 2 − desk 1). This method of calculating the difference in dif-
groups based on the orders mentioned previously, ensuring ferences, incorporating (desk 2 − desk 1), allowed us to take into
that each participant experienced all measurement condi- account intrinsic, within-person variability of BP. Bland-
tions in a sequence designed to minimize any potential order Altman plots were used to show BP variability among differ-
effects and biases. ent arm positions. We conducted these analyses for the study
population overall and then by the a priori defined subgroups
BP Measurement Procedure noted previously.
All BP measurements were conducted by 2 research staff mem- During the analysis of the study, we found that there were
bers who received standardized training and completed a cer- unequal numbers of participants allocated to each condition.
tification test in BP measurement, administered by an author Exploration of the Excel function RANDBETWEEN11 revealed
(J.C.). Measurements took place from 9 AM to 6 PM using a vali- that this program did not have the capability of assigning equal
dated oscillometric BP device (ProBP 2000 Digital Blood Pres- numbers of participants to each assignment; thus, the alloca-
sure Device [Welch Allyn]).12 Unless a specific condition was tion tables that were developed and used during randomiza-
present, such as the presence of an open sore, the right arm tion were unequal. This finding prompted us to conduct sen-
was used for all measurements. Other than the arm position, sitivity analyses to explore the effect of this unequal
all other patient preparatory and positioning recommenda- distribution on the primary outcomes using linear mixed ef-
tions were consistently applied per guidelines for the 3 mea- fect models. For these sensitivity analyses, we defined the main
surement conditions in this study. exposure through a dummy variable representing the treat-
After obtaining consent and asking participants to empty ment groups: 0 for the reference group (desk 2 − desk 1), 1 for
their bladders, participants walked for 2 minutes to replicate the (lap − desk 1) group, and 2 for the (side − desk 1) group. The
a typical clinical scenario before arriving at a BP measure- outcome was defined as the difference in mean BP readings
ment station. They then underwent a 5-minute seated rest pe- between each condition (lap, side, and desk 2) and desk 1. Par-
riod with their back and feet supported, after which 1 set of 3 ticipant identification number was included as a random in-

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine Published online October 7, 2024 E3

© 2024 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by UFBA, Bruna Barreto on 11/03/2024
 Research Original Investigation Arm Position and Blood Pressure Readings

pants (82%) participants reported recent (<365 days) health care

Table. Characteristics of Participants (N = 133)
utilization. Detailed participant characteristics by random-
Characteristics No. (%) ized sequence can be found in eTable 1 in Supplement 2.
Demographic information
Age BP Difference Across Arm Positions in the Entire Population
Mean (SD), y 57 (17) Average SBP/DBP was 126/74 mm Hg for each of the desk 1 and
≥60 y 74 (56) desk 2 positions, 130/78 mm Hg for the lap position, and 133/78
Sex mm Hg for the side position (Figure 2). The distribution of SBP
Female 70 (53) and DBP by arm position is shown in eFigures 1 and 2 and
Male 63 (47) eTable 2 in Supplement 2. The mean (SD) difference between
Racea desk 2 and desk 1 was −0.21 (7.26) mm Hg for SBP and 0.09
Asian 3 (2) (3.78) mm Hg for DBP (eTable 3 in Supplement 2). Using the
Black 103 (77) planned difference-in-differences analysis, the lap arm posi-
White 21 (16) tion resulted in statistically significant higher BP readings for
≥1 Race 6 (5) both SBP and DBP: ([lap − desk 1] − [desk 2 − desk 1]: mean Δ
Ethnicitya SBP of 3.9 [95% CI, 2.5-5.2] mm Hg and mean Δ DBP of 4.0 [95%
Hispanic 2 (2)
CI, 3.1-4.9] mm Hg). The side arm position resulted in even
greater BP differences: ([side − desk 1] − [desk 2 − desk 1]: mean
Non-Hispanic 129 (97)
Δ SBP of 6.5 [95% CI, 5.1-7.9] mm Hg and mean Δ DBP of 4.4
Other 2 (2)
[95% CI, 3.4-5.4] mm Hg) (Figure 2). Bland-Altman plots for
Medical history
each contrast are shown in eFigures 3 and 4 in Supplement 2.
Prescribed antihypertensive medications 79 (59)
The variation of BP between 2 arm positions was largely con-
Took antihypertensive medications that day 64 (48)
sistent across the ranges of BP.
History of hypertension or myocardial infarction 44 (33)
Acute care visit in past 365 d 71 (53)
Subgroup Analysis
Chronic care visit in past 365 d 109 (82)
Results were largely consistent across subgroups (Figure 3). No-
Anthropometric measurements tably, there was a statistically significant larger difference in
Weight, mean (SD), lb 189 (52) lap BPs among those who had not received medical care in the
BMI ≥30 55 (41) past year (vs those who had) and in side SBPs among those with
Blood pressure measurements an SBP of 130 mm Hg or higher (vs <130 mm Hg) (Figure 3).
SBP, mean (SD), mm Hg 126 (21)
SBP ≥130 mm Hg 48 (36) Sensitivity Analysis
DBP, mean (SD), mm Hg 74 (10) The sensitivity analysis to explore the limitations of the
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided randomization procedure revealed similar results as the pri-
by height in meters squared); DBP, diastolic blood pressure; SBP, systolic blood mary analysis (eTable 4 in Supplement 2). Therefore, the
pressure. planned analysis by randomized group is presented as the main
a
Race and ethnicity were self-selected by the patient. analysis.

tercept to account for individual variability. In these models,

we adjusted for the following covariates: age (years), body
weight (kg), antihypertensive medication use (yes/no), upper-
Discussion
arm length (cm), arm circumference (cm), SBP (mm Hg), DBP This crossover randomized clinical trial demonstrates the con-
(mm Hg), and the order of BP measurement sets. siderable effect of arm position on BP readings. Specifically,
A 2-sided P < .05 was considered statistically significant. when BP measurements are obtained with arm positions fre-
All analyses were performed using R, version 4.2.3 (R Project quently used in clinical practice (ie, on the lap or at the side),
for Statistical Computing). the readings obtained are markedly higher than those ob-
tained with the arm positioned according to published guide-
lines. Although the error in SBP with the arm supported in the
lap was less striking in magnitude than when the arm was un-
Results supported at the side (approximately 4 mm Hg vs approxi-
Participant Characteristics mately 7 mm Hg), BP readings in either position were
A total of 133 participants were randomized into 1 of 6 pos- sufficiently high to raise concerns for overdiagnosis and over-
sible groups that differed by order of the 3 arm positions used treatment. These findings were consistent, and also more ex-
for sets of triplicate measurements, with the number of par- treme, among higher-risk groups: SBP was overestimated by
ticipants in each group ranging from 12 to 31 (Figure 1). The approximately 9 mm Hg among individuals with hyperten-
mean (SD) age of the study participants was 57 (17) years, and sive BP when their arm was positioned at the side.
70 participants (53%) were female (Table). There were 55 par- This statistically significant BP measurement error when
ticipants (41%) with a BMI of 30 or higher, and 109 partici- guideline-recommended arm positions are not implemented

E4 JAMA Internal Medicine Published online October 7, 2024 (Reprinted) jamainternalmedicine.com

© 2024 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by UFBA, Bruna Barreto on 11/03/2024
 Arm Position and Blood Pressure Readings Original Investigation Research

Figure 2. Difference-in-Differences Analysis for Blood Pressure Obtained

With the Arm in Different Positions Compared to the Reference Standard

A Net SBP difference B Net DBP difference

10.0 10.0

6.5 (95% CI, 5.1−7.9)

7.5 7.5

4.4 (95% CI, 3.4−5.4)

DBP, mm Hg
SBP, mm Hg

3.9 (95% CI, 2.5−5.2) Results compare the difference in

4.0 (95% CI, 3.1−4.9)
5.0 5.0 differences in blood pressure when
the arm is positioned on the lap or
hanging at the side compared to
when blood pressure is measured
2.5 2.5 with the arm resting on a desk with
the cuff at midheart level (reference).
DBP indicates diastolic blood
0 0 pressure; SBP, systolic blood
Lap Side Lap Side pressure. Error bars represent 95%
CIs.

Figure 3. Difference-in-Differences (DiD) Analysis for Systolic Blood Pressure (SBP)

Obtained With the Arm in Different Positions Compared to the Reference Standard Among Subgroups

No. of Favors Favors Favors Favors

Subgroup participants DiD (95% CI) lap reference DiD (95% CI) side reference
All 133 3.9 (2.5-5.2) 6.5 (5.1-7.9)
SBP
<130 mm Hg 85 4.0 (2.5-5.3) 5.3 (3.8-6.9)
≥130 mm Hg 48 3.5 (0.9-6.2) 8.5 (5.7-11.4)
Age
<60 y 59 4.4 (2.9-5.8) 5.2 (3.5-7.0)
≥60 y 74 3.3 (1.3-5.6) 7.5 (5.3-9.6)
BMIb
<30 78 4.9 (3.2-6.5) 7.5 (5.7-9.3)
≥30 55 2.4 (0.2-4.7) 5.1 (2.8-7.4)
Last health care utilization
<365 d 101 3.6 (2.0-5.2) 6.3 (4.5-8.0)
≥365 d 8 9.6 (6.4-12.7) 9.9 (6.9-12.9)

0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
DiD (95% CI) DiD (95% CI)

Results compare the difference in differences in BP when the arm is positioned body mass index (calculated as weight in kilograms divided by height in meters
on the lap or hanging at the side compared to when BP is measured with the squared).
arm resting on a desk with the cuff at midheart level (reference). BMI indicates

aligns with prior research.5-7 Earlier studies have shown that ous data regarding the effect of arm position on BP measure-
unsupported or arm positioning below heart level can over- ment when using an automated device instead of a manual
estimate SBP by 4 to 23 mm Hg and DBP by 3 to 12 mm Hg.4 sphygmomanometer,6,14,16 reflecting contemporary clinical
However, the present study has notable strengths over these practice.17,18
earlier studies. The randomized crossover design we imple- In addition to overcoming the limitations of other stud-
mented is ideal for the study of differences in BP, which is in ies, this study also offers several unique observations not pre-
contrast to the majority of published studies where the order viously reported. We demonstrate that when the arm is com-
of arm positions before seated BP measurement was not ran- pletely unsupported and hanging at the side, as is often the case
domized or not clearly described.5,7 The sample size in the pre- when arm support on a desk or chair is not possible or when a
sent study was also substantially greater than that of prior trials patient is seated on the examination table in a clinic room, BP
(<50 participants).6,7 Another differentiating characteristic of is greatly overestimated. Furthermore, positioning the arm in
this trial is that we focused the investigation on arm positions the lap, a typical compromise for the above scenarios, also re-
commonly used in BP screening environments and clinical set- sults in considerable BP overestimation. Thus, these arm po-
tings. Most published studies compare BP measurements sitions should not be used, even in the setting of limited time
obtained with the arm positioned while standing or supine, or resources.19-23 Proper arm position may be even more im-
which are not the recommended postures for diagnosing portant for individuals at higher risk for cardiovascular dis-
and managing hypertension.4,13-15 Finally, we provide rigor- ease, particularly those with hypertensive SBP.

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine Published online October 7, 2024 E5

© 2024 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by UFBA, Bruna Barreto on 11/03/2024
 Research Original Investigation Arm Position and Blood Pressure Readings

Several physiological mechanisms likely explain why BP measurement technique, including nonadherence to sup-
measurements are higher when the arm is not optimally posi- ported arm positioning, likely contributes to these differ-
tioned or supported. First, the vertical distance between the heart ences. Since the number of studies using electronic health rec-
and the cuff increases when the arm is positioned in the lap or ord data is increasing, researchers should cautiously interpret
at the side (vs when it is supported on a desk with midcuff at heart BP data in this context and understand their limitations. Si-
level). This increase in distance when the arm is positioned at multaneously, health care systems should continue efforts to
levels below the heart leads to an increase in hydrostatic pres- improve and maintain the quality of BP measurements ob-
sure (the force exerted due to gravitational pull) in the brachial tained for patient management and research, and even con-
artery.24 Additionally, with these lower arm positions, there is sider a regulatory approach to promote standardized mea-
decreased venous return and compensatory vasoconstriction surements of BP. 31 Out-of-office BP measurement, often
leading to an increase in vascular resistance and an increase in conducted in the home environment, is important in the di-
BP.13,25 Moreover, an unsupported arm can lead to muscle con- agnosis and management of hypertension.32,33 Appropriate pa-
traction, which may cause transient increases in BP.26,27 It should tient preparation and positioning prior to BP measurements,
be noted that the present trial was not designed to distinguish including using the appropriate arm position, is as important
between the effects of arm position and support (eg, the side arm for home measurements as it is for measurements obtained in
position included both lower arm position and lack of support). the clinic. Therefore, education and training of both clinical
Although clinical guidelines emphasize positioning the arm staff and patients regarding BP measurement is essential for
at midheart level with support during BP measurements, this hypertension control and cardiovascular disease prevention.
practice is often overlooked in clinical settings.17,18 Several fac-
tors contribute to this discrepancy: lack of health care profes- Limitations
sional awareness about the effect of arm position on measure- This study has limitations. First, as noted previously, the use
ment accuracy, as well as limitations in training, resources, and of the RANDBETWEEN function resulted in the unequal ran-
equipment, particularly in resource-limited environments.28 dom distribution of participants to each group. However, an
The error in BP measurement resulting from nonadherence to extensive set of sensitivity analyses adjusting for participant
this recommendation has the potential to lead to substantial characteristics and order demonstrated consistent results with
hypertension overdiagnosis, unnecessary patient follow- the a priori analysis. Second, some subgroups included rela-
ups, and overtreatment. Inaccurate arm positioning can over- tively small sample sizes; thus, the results of subgroup analy-
estimate BP by up to 5 mm Hg generally and close to 10 mm ses need to be interpreted carefully. Finally, it is uncertain to
Hg in individuals with high levels of SBP. Based on our calcu- what extent the present results can be generalizable to other
lations using data from the National Health and Nutrition Ex- settings (eg, different BP devices).
amination Survey, improper arm position would result in 16%
of US adults, equating to 40 million individuals, being mis-
classified as hypertensive when using a SBP cutoff of 140 mm
Hg and higher, and 22% (54 million individuals) would be mis-
Conclusions
classified when using a SBP cutoff of 130 mm Hg and higher.29 This crossover randomized clinical trial shows that not
Considering the varied health care practices, equipment stan- adhering to the guideline-recommended arm position and sup-
dards, and training levels across different countries, the like- port during BP measurement can result in overestimation of
lihood of misdiagnosis could be even greater. BP by 4 to 10 mm Hg. This degree of BP error could lead to a
Discrepancies in BP data between electronic health rec- substantial number of people being overdiagnosed with
ords and research settings have been reported30; suboptimal hypertension.

ARTICLE INFORMATION Author Contributions: Mr Sabit and Dr Brady had submitted work. Dr Matsushita reported grants
Accepted for Publication: August 7, 2024. full access to all of the data in the study and take from the National Institutes of Health and personal
responsibility for the integrity of the data and the fees from Kowa Company, RhythmX AI, and Fukuda
Published Online: October 7, 2024. accuracy of the data analysis. Denshi outside the submitted work. No other
doi:10.1001/jamainternmed.2024.5213 Concept and design: Sabit, Ishigami, Charleston, disclosures were reported.
Author Affiliations: Department of Epidemiology, Miller, Matsushita, Appel, Brady. Funding/Support: This study was supported by
Johns Hopkins Bloomberg School of Public Health, Acquisition, analysis, or interpretation of data: Liu, Resolve to Save Lives. Resolve to Save Lives is
Baltimore, Maryland (Liu, Zhao, Ishigami, Zhao, Sabit, Pathiravasan, Miller, Matsushita, funded by Bloomberg Philanthropies, the Bill and
Charleston, Matsushita, Appel, Brady); Welch Appel, Brady. Melinda Gates Foundation, and Gates Philanthropy
Center for Prevention, Epidemiology, and Clinical Drafting of the manuscript: Liu, Matsushita, Brady. Partners, which is funded with support from the
Research, Johns Hopkins University, Baltimore, Critical review of the manuscript for important Chan Zuckerberg Foundation.
Maryland (Liu, Zhao, Ishigami, Charleston, intellectual content: All authors.
Matsushita, Appel, Brady); Department of Statistical analysis: Liu, Zhao, Sabit, Pathiravasan. Role of the Funder/Sponsor: Resolve to Save Lives
Biostatistics, Johns Hopkins Bloomberg School of Obtained funding: Matsushita, Appel. had no role in the design and conduct of the study;
Public Health, Baltimore, Maryland (Sabit, Administrative, technical, or material support: collection, management, analysis, and
Pathiravasan); Department of Medicine, Johns Appel, Brady. interpretation of the data; preparation, review, or
Hopkins University School of Medicine, Baltimore, Supervision: Pathiravasan, Charleston, Miller, approval of the manuscript; and decision to submit
Maryland (Miller, Matsushita, Appel); Department Matsushita, Brady. the manuscript for publication.
of Pediatrics, Johns Hopkins University School of Conflict of Interest Disclosures: Ms Liu reported Data Sharing Statement: See Supplement 3.
Medicine, Baltimore, Maryland (Brady). grants from Resolve to Save Lives outside the

E6 JAMA Internal Medicine Published online October 7, 2024 (Reprinted) jamainternalmedicine.com

© 2024 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by UFBA, Bruna Barreto on 11/03/2024
 Arm Position and Blood Pressure Readings Original Investigation Research

Additional Contributions: We thank the 12. Alpert BS. Validation of the Welch Allyn Pro BP 22. Sekisui Diagnostics. YouTube. Accessed
participants who volunteered their time and the 2000, a professional-grade inflation-based January 22, 2024. https://www.youtube.com/watch
staff who conducted the study. automated sphygmomanometer with arrhythmia ?v=AdddVV94xHg&t=100s
detection in a combined pediatric and adult 23. Our physicians. The Heart Center of Northern
REFERENCES population by ANSI/AAMI/ISO standard testing. Anne Arundel County. Accessed January 22, 2024.
1. Hypertension. World Health Organization. March Blood Press Monit. 2018;23(6):315-317. doi:10.1097/ https://www.heartcenterdocs.com/our-physicians/
16, 2023. Accessed November 13, 2023. https:// MBP.0000000000000350
24. Simulation of blood pressure measurement
www.who.int/news-room/fact-sheets/detail/ 13. Olufsen MS, Ottesen JT, Tran HT, Ellwein LM, hydrostatic artifact. University of Florida Health.
hypertension Lipsitz LA, Novak V. Blood pressure and blood flow October 13, 2022. Accessed January 12, 2024.
2. Beevers G, Lip GY, O’Brien E. Blood pressure variation during postural change from sitting to https://simulation.health.ufl.edu/education-
measurement: part I—sphygmomanometry: factors standing: model development and validation. J Appl training/online-simulations/blood-pressure-
common to all techniques. BMJ. 2001;322(7292): Physiol (1985). 2005;99(4):1523-1537. doi:10.1152/ measurement-hydrostatic-artifact/
981-985. doi:10.1136/bmj.322.7292.981 japplphysiol.00177.2005
25. Klabunde RE. Effects of gravity on venous
3. Merendino J, Finnerty FA Jr. Importance of the 14. Terént A, Breig-Asberg E. Epidemiological return. Cardiovascular Physiology Concepts
position of the arm on the level of arterial blood perspective of body position and arm level in blood website. Accessed January 17, 2024. https://
pressure. JAMA. 1961;175:51-53. doi:10.1001/jama. pressure measurement. Blood Press. 1994;3(3):156- cvphysiology.com/cardiac-function/cf017
1961.63040010010015c 163. doi:10.3109/08037059409102246
26. Smith ML, Hudson DL, Raven PB. Effect of
4. Kallioinen N, Hill A, Horswill MS, Ward HE, 15. Netea RT, Lenders JW, Smits P, Thien T. Both muscle tension on the cardiovascular responses to
Watson MO. Sources of inaccuracy in the body and arm position significantly influence blood lower body negative pressure in man. Med Sci
measurement of adult patients’ resting blood pressure measurement. J Hum Hypertens. 2003;17 Sports Exerc. 1987;19(5):436-442. doi:10.1249/
pressure in clinical settings: a systematic review. (7):459-462. doi:10.1038/sj.jhh.1001573 00005768-198710000-00003
J Hypertens. 2017;35(3):421-441. doi:10.1097/HJH. 16. Reeves RA. The rational clinical examination: 27. France CR, France JL, Patterson SM. Blood
0000000000001197 does this patient have hypertension? how to pressure and cerebral oxygenation responses to
5. Familoni OB, Olunuga TO. Comparison of the measure blood pressure. JAMA. 1995;273(15):1211- skeletal muscle tension: a comparison of two
effects of arm position and support on blood 1218. doi:10.1001/jama.1995.03520390071036 physical maneuvers to prevent vasovagal reactions.
pressure in hypertensive and normotensive 17. Whelton PK, Carey RM, Aronow WS, et al. 2017 Clin Physiol Funct Imaging. 2006;26(1):21-25. doi:
subjects. Cardiovasc J S Afr. 2005;16(2):85-88. ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/ 10.1111/j.1475-097X.2005.00642.x
6. Silverberg DS, Shemesh E, Iaina A. The NMA/PCNA Guideline for the prevention, 28. Mourad A, Carney S. Arm position and blood
unsupported arm: a cause of falsely raised blood detection, evaluation, and management of high pressure: an audit. Intern Med J. 2004;34(5):290-
pressure readings. BMJ. 1977;2(6098):1331. doi:10. blood pressure in adults: executive summary: 291. doi:10.1111/j.1444-0903.2004.00592.x
1136/bmj.2.6098.1331 a report of the American College of
Cardiology/American Heart Association Task Force 29. National Health and Nutrition Examination
7. Beck FM, Weaver JM, Blozis GG, Unverferth DV. on Clinical Practice Guidelines. Hypertension. 2018; Survey: 2017-2018 examination data—continuous
Effect of arm position and arm support on indirect 71(6):1269-1324. doi:10.1161/HYP. NHANES. Centers for Disease Control and
blood pressure measurements made in a dental 0000000000000066 Prevention. Accessed January 23, 2024. https://
chair. J Am Dent Assoc. 1983;106(5):645-647. doi: wwwn.cdc.gov/Nchs/Nhanes/search/datapage.
10.14219/jada.archive.1983.0146 18. Williams B, Mancia G, Spiering W, et al; aspx?Component=Examination&CycleBeginYear=
Authors/Task Force Members. 2018 ESC/ESH 2017
8. The arm rest and support study (ARMS). guidelines for the management of arterial
ClinicalTrials.gov identifier: NCT05372328. hypertension: the task force for the management of 30. Ahmad FS, Chan C, Rosenman MB, et al.
Updated September 11, 2023. Accessed December arterial hypertension of the European Society of Validity of cardiovascular data from electronic
08, 2023. https://clinicaltrials.gov/study/ Cardiology and the European Society of sources: the multi-ethnic study of atherosclerosis
NCT05372328 Hypertension: the task force for the management and HealthLNK. Circulation. 2017;136(13):1207-1216.
9. Ishigami J, Charleston J, Miller ER III, Matsushita of arterial hypertension of the European Society of doi:10.1161/CIRCULATIONAHA.117.027436
K, Appel LJ, Brady TM. Effects of cuff size on the Cardiology and the European Society of 31. Appel LJ, Miller ER III, Charleston J. Improving
accuracy of blood pressure readings: the Cuff(SZ) Hypertension. J Hypertens. 2018;36(10):1953-2041. the measurement of blood pressure: is it time for
randomized crossover trial. JAMA Intern Med. 2023; doi:10.1097/HJH.0000000000001940 regulated standards? Ann Intern Med. 2011;154(12):
183(10):1061-1068. doi:10.1001/jamainternmed. 19. Collins D. Doctors botch blood pressure 838-840. doi:10.7326/0003-4819-154-12-
2023.3264 readings more often than you think. Vox. July 3, 201106210-00014
10. Brady TM, Charleston J, Ishigami J, Miller ER III, 2018. Accessed January 17, 2024. https://www.vox. 32. Krist AH, Davidson KW, Mangione CM, et al; US
Matsushita K, Appel LJ. Effects of different rest com/science-and-health/2018/7/3/17510132/new- Preventive Services Task Force. Screening for
period durations prior to blood pressure blood-pressure-guidelines-ranges-hypertension hypertension in adults: US Preventive Services Task
measurement: the best rest trial. Hypertension. 20. Kaiser Permanente nation’s best at controlling Force reaffirmation recommendation statement.
2021;78(5):1511-1519. doi:10.1161/ high blood pressure. Kaiser Permanente. December JAMA. 2021;325(16):1650-1656. doi:10.1001/jama.
HYPERTENSIONAHA.121.17496 6, 2018. Accessed January 22, 2024. https:// 2021.4987
11. RANDBETWEEN—maybe not so random. Excel permanente.org/kaiser-permanente-nations-best- 33. Green BB, Anderson ML, Cook AJ, et al. Clinic,
Forum. Accessed December 09, 2023. https:// at-controlling-high-blood-pressure/ home, and kiosk blood pressure measurements for
www.excelforum.com/excel-formulas-and- 21. Using telehealth for your care needs. Kaiser diagnosing hypertension: a randomized diagnostic
functions/1222116-randbetween-maybe-not-so- Permanente. Accessed January 22, 2024. https:// study. J Gen Intern Med. 2022;37(12):2948-2956.
random.html healthy.kaiserpermanente.org/learn/what-is- doi:10.1007/s11606-022-07400-z
telehealth-services

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine Published online October 7, 2024 E7

© 2024 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by UFBA, Bruna Barreto on 11/03/2024