Original Research

Discrepancy Between Invasive and Noninvasive Blood Pressure

Measurements in Patients with Sepsis by Vasopressor Status
Quincy K. Tran, MD, PhD* *University of Maryland School of Medicine, The R Adams Cowley Shock Trauma
Dominique Gelmann, BS† Center, Baltimore, Maryland
Zain Alam, BS† †
University of Maryland School of Medicine, Department of Emergency Medicine,
Richa Beher, BS† The Research Associate Program in Emergency Medicine and Critical Care,
Emily Engelbrecht-Wiggans, MS† Baltimore, Maryland
Matthew Fairchild†
Emily Hart, CRNP*
Grace Hollis†
Allison Karwoski, BS†
Jamie Palmer, BS†
Alison Raffman, BS†
Daniel J. Haase, MD*

Section Editor: Christopher Tainter, MD

Submission history: Submitted May 14, 2021; Revision received December 11, 2021; Accepted January 7, 2022
Electronically published May 5, 2022
Full text available through open access at http://escholarship.org/uc/uciem_westjem
DOI: 10.5811/westjem.2022.1.53211

Introduction: Blood pressure (BP) monitoring is an essential component of sepsis management.

The Surviving Sepsis Guidelines recommend invasive arterial BP (IABP) monitoring, although
the benefits over non-invasive BP (NIBP) monitoring are unclear. This study investigated
discrepancies between IABP and NIBP measurement and their clinical significance. We
hypothesized that IABP monitoring would be associated with changes in management among
patients with sepsis requiring vasopressors.

Methods: We performed a retrospective study of adult patients admitted to the critical care
resuscitation unit at a quaternary medical center between January 1–December 31, 2017. We
included patients with sepsis conditions AND IABP monitoring. We defined a clinically significant
BP discrepancy (BPD) between NIBP and IABP measurement as a difference of > 10 millimeters of
mercury (mm Hg) AND change of BP management to maintain mean arterial pressure ≥ 65 mm Hg.

Results: We analyzed 127 patients. Among 57 (45%) requiring vasopressors, 9 (16%) patients had
a clinically significant BPD vs 2 patients (3% odds ratio [OR] 6.4; 95% CI: 1.2-30; P = 0.01) without
vasopressors. In multivariable logistic regression, higher Sequential Organ Failure Assessment
(SOFA) score (OR 1.33; 95% CI: 1.02-1.73; P = 0.03) and serum lactate (OR 1.27; 95% CI: 1.003-
1.60, P = 0.04) were associated with increased likelihood of clinically significant BPD. There were no
complications (95% CI: 0-0.02) from arterial catheter insertions.

Conclusion: Among our population of septic patients, the use of vasopressors was associated
with increased odds of a clinically significant blood pressure discrepancy between IABP and NIBP
measurement. Additionally, higher SOFA score and serum lactate were associated with higher
likelihood of clinically significant blood pressure discrepancy. Further studies are needed to confirm
our observations and investigate the benefits vs the risk of harm of IABP monitoring in patients
with sepsis. [West J Emerg Med. 2022;22(3)358–367.]

Western Journal of Emergency Medicine 358 Volume 23, no. 3: May 2022
Tran et al. Discrepancy Between Invasive and Noninvasive BP in Sepsis Patients

INTRODUCTION
Sepsis and septic shock are major healthcare problems Population Health Research Capsule
that affect millions of people around the world annually.1
The Surviving Sepsis Campaign (SSC) guidelines suggest What do we already know about this issue?
maintaining a mean arterial pressure (MAP) of at least 65 There are discrepancies in invasive arterial
millimeters of mercury (mm Hg) among these patients.1 blood pressure (IABP) and non-invasive BP
Furthermore, a recent meta-analysis suggested that earlier (NIBP) measurements in patients with sepsis.
administration of vasopressors is associated with improved
short-term outcomes in patients with sepsis.2 Invasive arterial What was the research question?
blood pressure (IABP) monitoring is considered to be the gold Does the difference between IABP and NIBP
standard compared to non-invasive blood pressure (NIBP) lead to change in management among patients
monitoring.3 Despite the SSC recommendation that patients with septic shock.
requiring vasopressors should undergo arterial catheter
placement for IABP monitoring as soon as possible,1 only What was the major finding of the study?
52% of patients on vasopressors from 168 intensive care units Vasopressor use, Sequential Organ Failure
(ICU) across the United States had IABP monitoring.4 Assessment (SOFA) score, and lactate levels
Data regarding the efficacy of IABP monitoring has are associated with change in management
been inconclusive. A previous retrospective study of 30 between IABP and NIBP monitoring. NIBP
patients with septic shock5 suggested there was only a small was typically higher than IABP.
difference in MAP measurements between IABP and NIBP in
its small patient population, and only 10% of those patients How does this improve population health?
had a difference of ≥ 10 mm Hg. However, the study was Invasive arterial blood pressure monitoring
significantly limited by its small patient sample size and lack is associated with detection of occult
of control group. More importantly, the study did not assess hypotension, compared to NIBP, in septic
whether having IABP monitoring would have changed patient patients with shock, high SOFA score, or high
management compared to NIBP monitoring. lactate level.
In our study we investigated the discrepancy between
NIBP and IABP measurement in a large patient population
with septic shock defined by the use of vasopressors,
compared to a control group of patients with sepsis but BP measurements. Patients who do not need further ICU level
without vasopressors. We hypothesized that the use of of care can have BP recorded every 2-4 hours while waiting
vasopressors would be associated with an increased in the CCRU for a bed in an intermediate care (IMC) unit or
discrepancy between NIBP and IABP measurement, medical ward. Most of the arterial catheter cannulations are
which would translate into potential differences in clinical performed by CCRU clinicians upon patients’ arrival as part
management for patients with septic shock. of the resuscitation efforts. The cannulations are performed
under sterile conditions with sterile gloves, sterile fields, and
METHODS hair covers in compliance with our institutional requirements.
Study Setting Additionally, the cannulation process can be aided by point-
We conducted the study in the critical care resuscitation of-care ultrasound at the clinicians’ preference. Our study was
unit (CCRU) at a quaternary academic center. The goal of the approved by our institutional review board.
CCRU (created in July 2013) is to expedite the interhospital
transfer of patients with time-sensitive disease or critical illnesses Patient Selection
when these conditions exceed the capability of the referring This study is a secondary analysis of a previously
hospitals and when our medical center’s adult ICUs do not have collected clinical dataset.7 All adults who were admitted to the
an available bed.6 These patients, depending on their disease CCRU between January 1–December 31, 2017 with arterial
severity, are transferred urgently to the CCRU to undergo catheter cannulation at the CCRU were eligible. We included
diagnostic or therapeutic interventions. Once these patients patients with diagnoses suggesting sepsis conditions and NIBP
receive the necessary interventions and are stabilized, they are and IABP measurement within 60 minutes of each other.
moved to an available in-patient bed at our medical center. We excluded patients who had diagnoses of hypertensive
To resuscitate these patients in the acute phase, the CCRU emergencies (acute aortic diseases, spontaneous intracranial
clinical policy requires that patients have arterial blood hemorrhage, ischemic stroke, etc.) because these patients are
pressure monitoring if they need frequent blood gas analyses managed according to goals of systolic BP,8,9 while patients
or hemodynamic monitoring, whether receiving a vasoactive with sepsis are managed according to goals of MAP.10 We also
infusion or not. The CCRU nursing staff also document hourly excluded patients who did not have three BP measurements

Volume 23, no. 3: May 2022 359 Western Journal of Emergency Medicine
Discrepancy Between Invasive and Noninvasive BP in Sepsis Patients Tran et al.

for each modality (IABP and NIBP) because we suspected the MAP from NIBP was 68 mm Hg, this was considered a
that a lower number of BP measurements would not produce clinically significant BPD. In this case, crystalloids or even
reliable average values of the measurements. Patients who vasopressors would have been added to increase the patient’s
arrived at the CCRU with arterial catheters were also excluded MAP of 58 mm Hg, while the MAP of 68 mm Hg, according
because they would not have documentations of IABP to NIBP monitoring, would have suggested no further
measurements at the time of arterial catheter insertions. We interventions. Conversely, a patient with a MAP of 50 mm Hg
defined patients with shock as those requiring any vasopressor per IABP and MAP of 60 mm Hg per NIBP would not have
(eg, norepinephrine, epinephrine, vasopressin) as reported a clinically significant BPD, because both modalities would
previously.5 For our study, we included only patients who have suggested interventions to increase the MAP to reach a
received vasopressors within one hour of arterial cannulation. goal of 65 mm Hg.
Our secondary outcome was the percentage of patients
Data Collection and Management who had MAP differences between IABP and NIBP of at least
We collected data from patients’ electronic health 10 mm Hg. Other outcomes included factors associated with
records at our institution. Relevant data occurring within one either primary or secondary outcomes.
hour of arterial cannulation was collected retrospectively.
Demographic data included age, gender, past medical history, Sample Size Calculation
and body mass index. Clinical data included components We based our sample size calculation on previous results
of the Sequential Organ Failure Assessment (SOFA) score, by Riley et al.5 We planned to detect a difference of 10 mm Hg
white blood cell (WBC) counts, serum lactate levels, and four with a standard deviation of 15 between NIBP-IABP among
consecutive pairs of both IABP and NIBP measurements. patients with vasopressors and those without vasopressors. As
For components of the SOFA score, we imputed missing a result, we calculated that we would need 37 patients for each
components as normal. Three patients in our population did group to have power of 80% with an α value of 0.05.
not have laboratory values for total bilirubin at the time of
arterial cannulation. Because their values were normal at Data Analysis
subsequent laboratory checks we imputed their component for We used descriptive analyses (mean ± standard deviation
the liver SOFA score as normal (score of 0). We also extracted [SD]), median [interquartile range [IQR]), or percentages
data regarding complications from arterial catheter insertions to present continuous variables or categorical variables as
throughout a patient’s hospital stay. We defined complications appropriate. We used unpaired Student’s t-test to compare the
as any necrosis of hand, wrist or extremity, source of blood mean between two groups (without vs with vasopressors). We
stream infection or local infection, bleeding, or aneurysm. performed forward stepwise, multivariable logistic regressions
We performed our retrospective data analyses in to estimate the associations between demographic, clinical
compliance with methodologic standards for health record independent variables with our outcomes (clinically significant
review.11 The research team members, who were not blinded BPD, MAP difference ≥ 10). Our independent variables were
to the study hypothesis, were first trained by the principal determined a priori and are listed in Appendix 1. Additionally,
investigator to extract data into a standardized Excel we assessed the goodness-of-fit, multicollinearity, and
spreadsheet (Microsoft Corp, Redmond, WA). Training discriminatory capability of our multivariable logistic
was performed with sets of 10 patients until results from regression models. For goodness-of-fit tests, a model with
all research team members reached 90% agreement with a Hosmer-Lemeshow test’s P-value > 0.05 is considered to have
senior investigator. Up to 10% of each investigator’s data was a good fit of independent variables.
subsequently double-checked for accuracy. To reduce further We used variance inflation factors (VIF) to assess
bias, investigators independently collected data in separate independent variables’ multicollinearity. Any factor with
sections. For example, investigators who collected data for VIF ≥ 5 were removed from the logistic regression for
SOFA scores did not collect BP measurements, and demonstrating collinearity. We used the area under the
vice versa. receiver operating characteristic (AUROC) curve to assess our
logistic regression models’ discriminatory capability. A model
Outcome Measures with AUROC of 1.0 would be considered to have perfect
Our primary outcome was the percentage of patients who discriminatory capability because this model can perfectly
had a clinically significant BP discrepancy (BPD) in MAP distinguish the difference between dichotomous outcomes
measurement via IABP and NIBP between those receiving (eg, clinically significant BPD vs none), while a model with
vasopressors and those not receiving vasopressors. We defined AUROC of 0.5 would have poor discriminatory capability.
a clinically significant BPD as a difference of at least 10 mm
Hg AND a potential change of clinical management, according Additional Analyses
to patient’s goal MAP ≥ 65 mm Hg. For example, when the Once our multivariable logistic regression identified
MAP from a patient’s arterial catheter was 58 mm Hg but continuous independent variables that were significantly

Western Journal of Emergency Medicine 360 Volume 23, no. 3: May 2022
Tran et al. Discrepancy Between Invasive and Noninvasive BP in Sepsis Patients

associated with a clinically significant BP discrepancy in IABP P< 0.001) (Table 2). In other words, IABP monitoring was
and NIBP measurement between patients with and without associated with a 5.3-times higher likelihood of detecting
vasopressors, we applied those continuous independent MAP level less than the recommended level of 65 mm Hg in
variables in probit analyses. The probit analyses would enable sepsis patients requiring vasopressors.
us to predict the probability of clinically significant BPD The median IQR of catheter days was 3 (1-5). The total
at certain values of the continuous independent variables. number of catheter days for our patient population was 639,
We used the Bland-Altman plot to graphically present the with no complications (95% CI: 0-0.02) from arterial catheter
discrepancy between NIBP and IABP. We performed our insertion (Table 2).
statistical analyses with Minitab version 19 (Minitab Corp,
State College, PA). We considered all tests with two-tailed Primary Outcome: Clinically Significant Discrepancy
P-value < 0.05 as statistically significant. Between NIBP And IABP
Among 57 patients requiring vasopressors, nine patients
RESULTS (16%) had a clinically significant BP discrepancy, compared to
Patient Characteristics two patients (3%) without vasopressor requirement (OR 6.4;
We electronically identified 570 patients who underwent 95% CI: 1.2-30; P = 0.01) (Table 2).
arterial catheter placement at the CCRU during the study The Bland-Altman plot of patients with sepsis but not
period (Figure 1). Among 271 patients with non-hypertensive requiring vasopressors (Figure 2A) showed that the [NIBP-
conditions, we included 127 patients with sepsis conditions IABP] discrepancy was distributed evenly throughout the
(list of diagnoses is included in Appendix 2) based on X-axis, which suggested that the difference between the
their admission diagnoses. Among the included patients, 57 two modalities was distributed evenly when patients were
hypotensive or normotensive. Additionally, in this patient
population, the discrepancy between NIBP and IABP (denoted
as [NIBP-IABP] on the Y-axis) was mostly concentrated
between the level of -10 mm Hg (IABP measurements > NIBP
measurements) and level of +10 mm Hg (NIBP measurements
> IABP measurements) (Figure 2A). This distribution
suggested that there was similar likelihood for IABP to be
higher than NIBP, and vice versa, among patients with sepsis
not requiring vasopressors.
Among patients with sepsis requiring vasopressors, the
NIBP and IABP difference was also distributed evenly along
the X-axis (Figure 2B). However, most values for the NIBP
and IABP difference for this group were above the level of +10,
suggesting that NIBP measurements were in general greater
than IABP in patients with sepsis requiring vasopressors.
Table 3 shows the results of the multivariable logistic
regressions measuring the association between clinical factors
and the primary outcome of clinically significant BP discrepancy
Figure 1. Patient selection diagram. We included 127 patients between NIBP and IABP measurement. Four factors were
with sepsis conditions in our analysis. associated with a clinically significant BPD between NIBP and
IABP. Each unit increase in SOFA score was associated with
increased odds of having a clinically significant difference in
(45%) required vasopressors and 70 (55%) did not require management when comparing NIBP and IABP (OR 1.33; 95%
vasopressors (Table 1). CI: 1.02-1.73; P = 0.034). Similarly, each increase in millimoles
The average (SD) age for the population was 55 (16) per liter (mmol/L) of serum lactate was associated with increased
years (Table 1), and there was no age difference between odds of having a clinically significant BP discrepancy when
patients without vasopressors or those with vasopressors. an arterial catheter was inserted (OR 1.27; CI: 1.003-1.60; P =
Compared to those without vasopressor use, patients who 0.047). The model showed good fit of data (Homes-Lemeshow
required vasopressors had significantly higher WBC counts, test’s P = 0.81), low multicollinearity (all factors had VIF < 5),
serum lactate levels, and SOFA scores (Table 1). Other clinical and very good discriminatory capability (AUROC = 0.92).
factors were similar between both groups. Of the patients Probit logit analyses demonstrated that for patients with
requiring vasopressors, 19 (33%) had MAP of less than or a mean SOFA score of 8 (approximately 5% of all patients
equal to 64 mm Hg by IABP monitoring, compared to 6 (9%) with sepsis, regardless of vasopressor status) had a clinically
of those without vasopressors (OR 5.3; 95% CI:1.9-14.5; significant BP discrepancy causing change in management when

Volume 23, no. 3: May 2022 361 Western Journal of Emergency Medicine
Discrepancy Between Invasive and Noninvasive BP in Sepsis Patients Tran et al.

Table 1. Characteristics of patients with sepsis conditions and arterial pressure monitoring in the critical care resuscitation unit who
were included in the study. Patients who required vasopressors were more likely to have higher SOFA* scores, serum lactate levels.
Variables All patients (N = 127) Without vasopressor (N = 70) With vasopressor (N = 57) P
Age, years (mean, SD) 55 (16) 54 (16) 56 (16) 0.3
Gender, N (%)
Male 78 (61) 42 (60) 36 (63) 0.7
Female 49 (39) 28 (40) 21 (37) 0.7
BMI, mean (SD) 32.4 (11.9) 32.1 (10.7) 32.8 (13.3) 0.6
Past medical history, N (%)
Diabetes 42 (33) 23 (33) 19 (33) 0.9
HTN 57 (45) 31 (44) 26 (46) 0.9
CAD 20 (16) 13 (19) 7 (12) 0.3
PAD 9 (7) 4 (6) 5 (9) 0.5
Any kidney disease 63 (50) 28 (40) 35 (61) 0.02
Mechanical ventilation, N (%) 47 (37) 18 (26) 29 (51) 0.049
Location of arterial catheter, N (%)
Radial 113 (89) 67 (96) 46 (81) 0.007
Femoral 14 (11) 3 (4) 11 (19) 0.007
Left 54 (43) 28 (40) 26 (46) 0.5
Right 73 (57) 42 (60) 31 (54) 0.5
SOFA score, median (IQR) 8 (4-11) 5 (2-8) 11 (8.5-14.5) < 0.001
Diagnoses, N (%)
Bowel obstruction 5 (4) 3 (4) 2 (4) 0.8
Endocarditis 4 (3) 2 (3) 2 (4) 0.8
Incarcerated organs 4 (3) 4(6) 0 (0) 0.3
Ischemic organs 2 (2) 0 (0) 2 (4) N/A
Liver failure 6 (5) 3 (4) 3 (5) 0.8
Pancreatitis 6 (5) 4(6) 2 (4) 0.6
Perforated viscus 12 (9) 4 (6) 8 (14) 0.1
Postoperative infection 11 (9) 6 (9) 5 (9) 0.9
Respiratory failure 9 (7) 7 (10) 2 (4) 0.2
Sepsis, unspecified 21 (17) 7 (10) 14 (57) 0.028
Soft tissue infection 46 (36) 29 (41) 17 (30) 0.2
Other 1 (1) 1 (1) 0 (0) N/A
Time intervals between NIBP and 10 (0-15) 12 (0-16) 8 (0-11) 0.018
IABP (minutes), median (IQR)
White blood cell counts (per 16.0 (10.8) 14.2 (9.7) 18.3 (11.7) 0.001
microliter), mean (SD)
Serum lactate (mmol/L), mean (SD) 3.1 (3.1) 2.1 (1.8) 4.3 (3.9) < 0.001
Hospital disposition, N (%)
Discharge home 40 (32) 26 (37) 14 (25) 0.1
Acute rehabilitation facility 36 (28) 17(24) 19 (33) 0.3
Skilled nursing home 22 (17) 16 (23) 6 (11) 0.7
Dead/hospice 29 (23) 11 (16) 18 (32) 0.03
BMI, body mass index; HTN, hypertension; CAD, coronary artery disease; PAD, peripheral arterial disease; IABP, invasive arterial blood
pressure; NIBP, non-invasive blood pressure; mm Hg, millimeters mercury; PAD, peripheral artery disease; IQR, interquartile range;
SOFA, Sequential Organ Failure Assessment; SD, standard deviation; mmol/L, millimoles per liter.

Western Journal of Emergency Medicine 362 Volume 23, no. 3: May 2022
Tran et al. Discrepancy Between Invasive and Noninvasive BP in Sepsis Patients

Table 2. Comparison between blood pressure from IABP and NIBP monitoring modalities for septic patients. Patients requiring vasopressors
had a greater likelihood of clinically significant discrepancy between IABP and NIBP compared to patients without vasopressor requirement.
Arterial blood pressure monitoring was more likely to detect MAP ≤ 64 mm Hg among sepsis patients with vasopressors.
Variables All patients (N = 127) Without vasopressor (N = 70) With vasopressor (N = 57) P
Catheter-days (days), median [IQR] 3 [1-5] 2 [1-4] 4 [2-8.5] <0.001
Type of vasopressor, N (%) 1

Norepinephrine 54 (43) 0 (0) 54 (95) N/A

Epinephrine 11 (9) 0 (0) 11 (19) N/A
Vasopressin 16 (13) 0 (0) 16 (28) N/A
Mean arterial pressure of NIBP (mm 82 (19) 87 (20) 76 (16) <0.001
Hg), mean (SD)
Mean arterial pressure of IABP (mm 79 (19) 84 (19) 73 (16) <0.001
Hg), mean (SD)
Difference in Mean Arterial Pressure 11 (12) 10 (10) 12 (15) 0.08
Between IABP and NIBP (mm Hg),
mean (SD)
Number of patients MAP of NIBP ≤ 64 12 (9) 5 (7) 7 (12) 0.3
mm Hg, N (%)
Number of patients with a clinically 11 (9) 2 (3) 9 (16)3 0.01
significant discrepancy in MAP2
Number of patients with MAP of 25 (20) 6 (9) 19 (33)4 <0.001
IABP ≤ 64 mm Hg, N (%)3
Number of any complications, N (%)5 0 (0) 0 (0) 0 (0) N/A
1
Patients were eligible to receive more than one vasopressor.
2
Clinically significant discrepancy was defined as Mean Arterial Pressure Difference ≥ 10 mm Hg and either NIBP’s or IABP’s reading
was ≤ 64 mm Hg.
3
OR 6.4, 95% CI 1.2-30, P = 0.01.
4
OR 5.3, 95% CI 1.9-14.5, P < 0.001.
5
Complications from arterial catheters were defined as necrosis, source for blood stream infection, local infection, infiltration, bleeding,
aneurysm. The 95% confidence interval (CI) for complications was 0 (95% CI 0-0.02).
NIBP, non-invasive blood pressure; IABP, invasive arterial blood pressure; IQR, interquartile range; MAP, mean arterial pressure; mm
Hg, millimeters of mercury; OR, odds ratio; SD, standard deviation.

an arterial catheter was inserted (Figure 2C). Similarly, when a between IABP and NIBP in our population of patients with
patient’s serum lactate level was 2 mmol/L (approximately 6% of sepsis. Among all patients with sepsis (both those requiring
all patients with sepsis, regardless of vasopressor status), IABP vasopressors and those not on vasopressors), a few clinical
monitoring resulted in a change in clinical management (Figure factors were associated with increased odds of change in BP
2D). Approximately 9% of patients had a change in clinical management when arterial catheters were inserted. Higher
management when their serum lactate was 4 mmol/L. SOFA score and higher serum lactate levels were both
associated with higher likelihood of clinically significant BP
Secondary Outcome: MAP Difference ≥ 10 mmHg Between discrepancy between NIBP and IABP monitoring modalities.
NIBP And IABP Higher SOFA score, history of peripheral artery disease, and
Three factors were significantly associated with high the diagnosis of incarcerated organs were associated with
likelihood of patients having a MAP difference ≥ 10 mm Hg higher likelihood of larger MAP difference between NIBP
between the two modalities (Table 3). These three factors were and IABP.
higher SOFA score (OR 1.27; 95% CI: 1.03-1.3; P = 0.012), Our study provides support for the use of IABP
having peripheral artery disease (OR 6.7; 95% CI: 1.3-22.5; P monitoring in patients requiring vasopressors, as clinically
= 0.021), and the diagnosis of incarcerated organs (OR 16.4; significant BP changes may be missed with NIBP
95% CI: 1.4 to +100; P = 0.027). monitoring. Findings from our study population showed
that IABP monitoring in patients with sepsis requiring
DISCUSSION vasopressors was significantly associated with higher
The use of vasopressors was associated with an likelihood of detecting MAP ≤ 64 mm Hg (Table 2).
increased incidence of clinically significant BP discrepancy Similarly, IABP values were more frequently lower than

Volume 23, no. 3: May 2022 363 Western Journal of Emergency Medicine
Discrepancy Between Invasive and Noninvasive BP in Sepsis Patients Tran et al.

A. B.

C. D.

Figure 2. (A)Bland-Altman plot displaying blood pressure differences among septic patients without vasopressors. The noninvasive blood
pressure (NIBP) and invasive arterial (IA) BP discrepancy was distributed evenly throughout the X-axis, demonstrating that the difference
between the two modalities occurred when patients were hypotensive or normotensive. Additionally, the difference between NIBP and
IABP on the Y-axis was mostly concentrated between the level of -10 mm Hg and +10 mm Hg, demonstrating that the NIBP modality has
equal likelihood to be higher or lower than IABP. (B) Bland-Altman plot displaying blood pressure differences among septic patients with
vasopressors. There were even distributions of NIBP-IABP* discrepancies along the X-axis, demonstrating that the difference between the
two modalities occurred when patients were hypotensive or normotensive. However, most values for [NIBP-IABP] difference were above
the level of +10 mm Hg, demonstrating that NIBP measurements were usually greater than IABP in our patient population with sepsis
requiring vasopressors. (C) Probit logit analysis showing probability of having clinically significant discrepancy between noninvasive and
intra-arterial blood pressure (Y-axis) and its association with SOFA score (X-axis). Patients who had a SOFA* score of 20 (X-axis) would have
50% probability (Y-axis) of requiring change in clinical management when arterial catheters were inserted. (D) Probit logit analysis showing
probability of having a clinically significant discrepancy between noninvasive and intra-arterial blood pressure (Y-axis) and its association with
serum lactate level. Patients who had serum lactate of 4 mmol/L (X-axis) would be associated with approximately 9% probability (Y-axis) of
having change of clinical management when arterial catheters were present.
IABP, invasive arterial blood pressure; LOA, limit of agreement; mm Hg, millimeter of mercury; NIBP, non-invasive blood pressure; SOFA,
Sequential Organ Failure Assessment; mmol/L, millimoles per liter.

NIBP among patients with sepsis on vasopressors (Figure real-time continuous monitoring. For both reasons, IABP
2B). Although the mechanism for the difference is still monitoring would enable clinicians to detect MAP < 65
unknown, this observation may have important clinical mm Hg sooner and intervene early. This has important
implications. Patients who require vasopressors may have implications for patient outcomes, as previous studies
unrecognized hypotension when they arrive at a resuscitation suggested that hypotension may lead to significant morbidity
or ICU due to the use of NIBP monitoring. Additionally, for patients.12,13
NIBP measurements are typically taken intermittently, In a patient population that is similar to ours, inserting
while IABP measurement provides the additional benefit of an arterial catheter would be associated with a change in

Western Journal of Emergency Medicine 364 Volume 23, no. 3: May 2022
Tran et al. Discrepancy Between Invasive and Noninvasive BP in Sepsis Patients

Table 3. Results from forward stepwise multivariable logistic regression measuring association between clinical factors and the
likelihood of clinically significant discrepancy between NIBP and IABP*. All predetermined factors were entered into the models and
only factors with significant association were reported. The models for each outcome measure showed both good fit of the independent
variables and good discriminatory capability (higher AUROC**).
Variables OR 95% CI P VIF
Outcome: Clinically Significant Blood
Pressure Discrepancy1
SOFA – each unit 1.33 1.02-1.73 0.034 2.0
Serum lactate – each mmol/L 1.27 1.003-1.60 0.047 2.1
Any kidney disease 0.03 0.002-0.51 0.015 2.6
Bowel obstruction 34 1.2-100+ 0.035 1.4
Secondary outcome: MAP difference ≥ 10
mm Hg2
SOFA – each unit 1.17 1.03-1.3 0.012 1.9
Peripheral artery disease 6.7 1.3-33.5 0.021 1.1
Incarcerated organs 16.4 1.4-100+ 0.027 1.1
1
Homes-Lemeshow test chi-square 4.5, D(f) = 8; P = 0.81; AUROC: 0.92.
2
Homes-Lemeshow test chi-square 6.5, D(f) = 8, P = 0.59; AUROC: 0.72.
AUROC, area under the receiver operating characteristic curve; OR, odds ratio; CI, confidence interval; D(f), degree of freedom;
mmol/L, millimoles per liter; IABP, invasive arterial blood pressure; NIBP, non-invasive blood pressure; SOFA, Sequential Organ Failure
Assessment score; VIF, variance inflation factor.

BP management in 9% of all patients with sepsis regardless Although our study was not designed to investigate
of vasopressor status, and in 16% of patients with sepsis the economics of IABP monitoring, we calculated the
requiring vasopressors. In other words, for every 11 septic cost required to detect change in clinical management
patients with IABP monitoring regardless of vasopressor via arterial line BP measurement for patients with
use, one patient would be identified as requiring change septic shock. The one-time supply cost to set up IABP
in clinical management. For patients with sepsis requiring monitoring at our institution is approximately $55 US
vasopressors, IABP monitoring would detect one need for dollars (USD) per patient. The prevalence of clinically
change in management for every seven patients. Within our significant BP discrepancy was 16% among patients with
population of patients with sepsis, the probability of change sepsis on vasopressors, which equates to one change
in management when arterial catheter was inserted was in management for approximately every seven patients
approximately 5% for patients whose SOFA score was 8, for with IABP monitoring. Therefore, the total cost of IABP
an estimated change in one of every 20 patients. For those monitoring would be approximately $385 USD to detect
with a SOFA score of 16, IABP monitoring was associated clinical change in management for every seven patients
with management change in one of every two patients with sepsis requiring vasopressors. Further analysis is
(Figure 2C). Our probit logit analysis suggested that IABP necessary to investigate whether the cost for IABP will
monitoring would detect one change in management for offset the cost of patients’ hospitalization if they develop
approximately every 17 patients with a serum lactate level acute kidney injuries or other comorbidities.
of 2 mmol/L, and one change in management for every 11 Other authors have questioned the need for arterial
patients with a serum lactate level of 4 mmol/L. catheters among critically ill patients because of the
The baseline differences in patients with sepsis associated risks and unclear benefit of ABP.14 Results
requiring vasopressors and those not requiring vasopressors from our study suggest that IABP monitoring offers the
may have influenced our findings of an increased rate of benefit of potential change in clinical management due
BP discrepancy in the former group. Patients with higher to early detection of hypotension, and with low cost and
SOFA scores indicating increased illness severity may low complication rates. We found no complications from
undergo more vasodilatory changes that may contribute arterial catheter cannulation in our patient population. Our
to a higher prevalence of BP discrepancy between the two findings agreed with the previously reported low arterial
measurement modalities. While our patient population cannulation risk of less than 1%.15 This suggests that early
was not large enough for propensity score matching of insertion of an arterial catheter is a low-risk procedure
these groups, further studies should explore such factors that enables clinicians to detect and remedy hypotension
potentially influencing outcomes. effectively, thus reducing the likelihood of hypotension-

Volume 23, no. 3: May 2022 365 Western Journal of Emergency Medicine
Discrepancy Between Invasive and Noninvasive BP in Sepsis Patients Tran et al.

related complications. As a result, we presented information Address for Correspondence: Quincy K. Tran, MD, University
suggesting that arterial catheters are associated with a high of Maryland School of Medicine, The R. Adams Cowley Shock
benefit-to-harm ratio in patients with sepsis, especially those Trauma Center, 22 South Greene Street, Suite T3N45, Baltimore,
MD 21201. Email: [email protected].
requiring vasopressors. Further studies are necessary to
confirm our observations. Conflicts of Interest: By the WestJEM article submission
agreement, all authors are required to disclose all affiliations,
LIMITATIONS funding sources and financial or management relationships that
Our exploratory study has several limitations. First, could be perceived as potential sources of bias. No author has
patients with sepsis requiring vasopressors are not similar professional or financial relationships with any companies that are
relevant to this study. There are no conflicts of interest or sources
to those without vasopressors, but the small sample of funding to declare.
size prevented us from performing propensity score
matching. However, by including the group of patients Copyright: © 2022 Tran et al. This is an open access article
without vasopressors, we provided a glimpse of the distributed in accordance with the terms of the Creative Commons
potential discrepancy between NIBP and IABP in both Attribution (CC BY 4.0) License. See: http://creativecommons.org/
groups of patients with sepsis. Additionally, we could licenses/by/4.0/
not retrospectively identify whether the BP cuff was on
the same or opposite arm as the arterial catheters, as
our nursing staff usually does not document the location REFERENCES
of the BP cuffs. Further, we based our outcome on the
1. Rhodes A, Evans LE, Alhazzani W, et al. Surviving Sepsis Campaign:
potential change of clinical management, according to
International Guidelines for Management of Sepsis and Septic Shock:
MAP ≥ 65 mm Hg, but we could not ascertain what types
2016. Intensive Care Med. 2017;43(3):304-77.
of interventions were given to the patients. Finally, our
2. Li Y, Li H, Zhang D. Timing of norepinephrine initiation in patients
multivariable logistic regressions showed wide 95%
with septic shock: a systematic review and meta-analysis. Crit Care.
CIs for a few clinical factors (past medical history of
2020;24(1):488.
kidney disease, diagnosis of bowel obstructions, and
incarcerated organs) due to the very small sample sizes 3. Lakhal K, Macq C, Ehrmann S, et al. Noninvasive monitoring of blood

of these variables from a population with various causes pressure in the critically ill: reliability according to the cuff site (arm, thigh,
of sepsis. Although our results suggest there may be an or ankle). Crit Care Med. 2012;40(4):1207-13.
association between these factors and clinically significant 4. Gershengorn HB, Wunsch H, Scales DC, et al. Association between
BP discrepancy, the sample size was too small to draw arterial catheter use and hospital mortality in intensive care units. JAMA
meaningful conclusions. Intern Med. 2014;174(11):1746-54.
Despite these limitations, our exploratory study had 5. Riley LE, Chen GJ, Latham HE. Comparison of noninvasive blood
strength over the previous study by Riley et al.5 We included pressure monitoring with invasive arterial pressure monitoring in medical
a larger number of patients requiring vasopressors along with ICU patients with septic shock. Blood Press Monit. 2017;22(4):202-7.
a group of patients without vasopressors. We demonstrated 6. Scalea TM, Rubinson L, Tran Q, et al. Critical care resuscitation unit: an
relevant clinical benefits from IABP, not just the existence innovative solution to expedite transfer of patients with time-sensitive
of discrepancy between NIBP and IABP measurements. critical illness. J Am Coll Surg. 2016;222(4):614-21.
Additionally, we identified a few clinical factors that may 7. Keville MP, Gelmann D, Hollis G, et al. Arterial or cuff pressure: clinical
help clinicians practicing in the acute phase, such as in an predictors among patients in shock in a critical care resuscitation unit.
emergency department, resuscitation unit, or ICU, to decide Am J Emerg Med. 2021;46:109-15.
whether IABP is indicated.
8. Hiratzka LF, Bakris GL, Beckman JA, et al. 2010 ACCF/AHA/AATS/
ACR/ASA/SCA/SCAI/SIR/STS/SVM Guidelines for the Diagnosis and
CONCLUSION
Management of Patients with Thoracic Aortic Disease: a report of the
In patients with sepsis conditions requiring
American College of Cardiology Foundation/American Heart Association
vasopressors, there was an increased likelihood of clinical
Task Force on Practice Guidelines, American Association for Thoracic
change in blood pressure management with the use of
Surgery, American College of Radiology, American Stroke Association,
invasive arterial blood pressure monitoring. There were
no complications from arterial catheter insertion observed. Society of Cardiovascular Anesthesiologists, Society for Cardiovascular

Higher Sequential Organ Failure Assesment score and Angiography and Interventions, Society of Interventional Radiology,
higher serum lactate levels were both associated with a Society of Thoracic Surgeons, and Society for Vascular Medicine.
higher likelihood of a blood pressure discrepancy leading Circulation. 2010;121(13):e266-369.
to clinical change in management. Further studies are 9. Hemphill JC, Greenberg SM, Anderson CS, et al. Guidelines for the
necessary to confirm our observation and investigate the Management of Spontaneous Intracerebral Hemorrhage: A Guideline for
risks of arterial catheter cannulations. Healthcare Professionals from the American Heart Association/American

Western Journal of Emergency Medicine 366 Volume 23, no. 3: May 2022
Tran et al. Discrepancy Between Invasive and Noninvasive BP in Sepsis Patients

Stroke Association. Stroke. 2015;46(7):2032-60. Anesthesiology. 2013;119(3):507-15.

10. Dellinger RP, Levy MM, Carlet JM, et al. Surviving Sepsis Campaign: 13. Asfar P, Meziani F, Hamel J-F, et al. High versus low blood-
international guidelines for management of severe sepsis and septic pressure target in patients with septic shock. N Engl J Med.
shock: 2008. Crit Care Med. 2008;36(1):296-327. 2014;370(17):1583-93.
11. Worster A, Bledsoe RD, Cleve P, et al. Reassessing the methods of 14. Garland A. Arterial lines in the ICU: a call for rigorous controlled trials.
medical record review studies in emergency medicine research. Ann Chest. 2014;146(5):1155-8.
Emerg Med. 2005;45(4):448-51. 15. Scheer BV, Perel A, Pfeiffer UJ. Clinical review: complications and
12. Walsh M, Devereaux PJ, Garg AX, et al. Relationship between risk factors of peripheral arterial catheters used for haemodynamic
intraoperative mean arterial pressure and clinical outcomes after monitoring in anaesthesia and intensive care medicine. Crit Care.
noncardiac surgery: toward an empirical definition of hypotension. 2002;6(3):199-204.

Volume 23, no. 3: May 2022 367 Western Journal of Emergency Medicine