GENERAL MEDICINE/ORIGINAL RESEARCH

The Vexatious Vital: Neither Clinical Measurements by Nurses

Nor an Electronic Monitor Provides Accurate Measurements of
Respiratory Rate in Triage
Paris B. Lovett, MD From the Columbia University Medical Center, New York, NY (Lovett); Memorial West
Jason M. Buchwald, MD Hospital, Pembroke Pines, FL (Buchwald); Beth Israel Medical Center, New York, NY
(Stürmann); and the Albert Einstein College of Medicine, New York, NY (Bijur).
Kai Stürmann, MD
Polly Bijur, PhD

Study objective: Of all the vital signs, only respiratory rate is still measured clinically in most US
triage systems. Previous studies have demonstrated the inaccuracy, poor interobserver agreement,
and low variability of routine measurements of respiratory rate. We assess the variability and
accuracy of triage nurses’ measurements of respiratory rate against a criterion standard. Also, we
assess electronic measurement of respiratory rate against the same criterion standard.
Methods: Consecutive patients presenting to an urban teaching emergency department (ED) were
enrolled in this prospective study. Electronic measurement of respiratory rate was recorded
throughout the triage encounter when nurses were recording measurements of respiratory rate.
Electronic respiratory rate was measured using transthoracic impedance plethysmography.
Immediately after each triage evaluation, criterion standard measurements of respiratory rate were
made by research assistants using the World Health Organization recommendation of auscultation
or observation for 60 seconds.
Results: We enrolled 159 patients. Variability was low for triage nurses’ measurements of
respiratory rate (SD 3.3) and electronic measurement of respiratory rate (SD 4.1) compared with
criterion standard measurements of respiratory rate (SD 4.8; P\.05). Triage nurses’ measurements
of respiratory rate and electronic measurement of respiratory rate showed low sensitivity in
detecting bradypnea and tachypnea. In a Bland-Altman analysis, triage nurses’ measurements of
respiratory rate and electronic measurement of respiratory rate showed poor agreement with
criterion standard measurements of respiratory rate. Subgroup analysis of patients presenting with
cardiac and respiratory symptoms yielded similar results.
Conclusion: Neither triage nurses nor an electronic monitor provides accurate measurements of
respiratory rate in the ED. Emergency physicians should search for new electronic modalities for
measuring respiratory rate to bring respiratory rate into line with other vital signs. Emergency
physicians should also consider new clinical strategies for measuring respiratory rate. [Ann Emerg
Med. 2005;45:68-76.]

0196-0644/$-see front matter

Copyright ª 2005 by the American College of Emergency Physicians.
doi:10.1016/j.annemergmed.2004.06.016

INTRODUCTION Respiratory rate has an important role in the early recognition

Emergency physicians depend heavily on vital signs. In of such diverse illnesses as pulmonary embolus, pneumonia,
recent years, pulse oximetry, glucometry, the Glasgow Coma congestive cardiac failure, and toxicologic and metabolic
Scale, and various pain scales have all been honored with the emergencies. Respiratory rate is incorporated into guidelines
title of ‘‘fifth vital sign.’’ The utility of the 4 classic vital signs is and decision rules for diagnosing and managing these
rarely questioned. conditions.1-5 Finally, because of its role in determining
Of the 4 classic vital signs, respiratory rate is the only one diagnosis and severity, respiratory rate plays an important role
that is still measured clinically, not electronically, in most in prioritizing care after triage.
emergency department (ED) settings. The reliability of triage We set out, therefore, to evaluate the measurement of
respiratory rate measurement matters for several reasons. respiratory rate in the triage of our urban teaching hospital ED.

68 Annals of Emergency Medicine Volume 45, no. 1 : January 2005

Lovett et al Measurement of Respiratory Rates in Triage

Setting
Editor’s Capsule Summary This study was conducted at an urban teaching ED with an
What is already known on this topic annual adult census of 52,000 visits. Triage is performed
Respiratory rate is a traditional emergency department independently of, and usually before, registration.
(ED) vital sign. Studies have indicated that routine
clinical respiratory rate assessment is an unreliable Selection of Participants
measurement. Consecutive patients presenting to the ED triage during
designated study periods were enrolled in the study. Patients
What question this study addressed who arrived in critical condition bypassed the triage area and
This study compared respiratory rate obtained by 2 were not enrolled, and no data were collected on them.
methodsdtriage nurse and electronic transthoracic Pediatric patients (\18 years) were enrolled but were
impedance plethysmographydwith the World Health subsequently excluded from the data analysis. The decision to
Organization criterion standard of auscultation or exclude pediatric patients was made after data collection was
observation for 1 minute in ambulatory adult ED finished.
patients. Study periods occurred on a convenience basis, ranging from
8 AM to 8 PM or from 8 PM to 8 AM. Study periods occurred
What this study adds to our knowledge
between August 4, 2002, and December 7, 2002. A minimum
In 159 subjects, both triage nurse and electronically sample size of 150 was chosen according to comparisons with
obtained respiratory rate had low sensitivity (\40%) for other studies of the reliability of vital sign measurements.6,7
bradypnea (respiratory rate \12 breaths/min) or As patients entered the triage area, research assistants applied
tachypnea (respiratory rate O20 breaths/min) compared standard 3-lead cardiac monitors (Escort Prism, Medical Data
with the study’s criterion standard. Electronics, Orlando, FL). During the triage encounter, nurses
How this might change clinical practice measured respiratory rate as part of standard triage assessment.
Assuming that this study’s criterion standard is valid, this Nurses were aware that their measurements of respiratory rate
study highlights the imprecision of respiratory rate as were being collected for the study. Nurses were not able to see
measured by ED triage nurses or transthoracic the electronic monitors or electronic measurement of respiratory
impedence plethysmography. This study should rate collected from them.
stimulate additional research to determine a standard for At the end of the triage visit, research assistants immediately
the measurement of this important vital sign. performed 2 ‘‘criterion standard’’ respiratory rate measure-
ments: auscultation for 1 minute and observation for 1 minute
(criterion standard measurements of respiratory rate).8
Patients were not moved, and their positions did not alter,
Specifically, we sought to measure the variability and accuracy between triage nurses’ measurements of respiratory rate,
of triage nurses’ measurements of respiratory rate relative to electronic measurement of respiratory rate, and criterion
criterion standard measurements. A finding of low accuracy standard measurements of respiratory rate.
would indicate that triage nurses’ measurements of respiratory The 7 research assistants who collected data for this study
rate are unlikely to detect clinically important bradypnea or were medical students from Albert Einstein College of
tachypnea. Low variability would indicate that triage nurses’ Medicine, New York, NY. They received training and oversight
measurements of respiratory rate are more clustered than are for their electronic measurement of respiratory rate and
criterion standard measurements of respiratory rate, which criterion standard measurements of respiratory rate. No testing
might suggest conscious or unconscious result selection. of or comparisons between research assistants was performed.
We also sought to evaluate the variability and accuracy of Triage nurses’ measurements of respiratory rate were
electronic measurements of respiratory rate recorded using obtained from the medical record.
a cardiac monitor equipped with transthoracic impedance The cardiac monitors were capable of electronic measure-
plethysmography, a feature found in many cardiac monitors in ment of respiratory rate by transthoracic impedance plethys-
common use in EDs. We sought to determine whether mography.
transthoracic impedance plethysmography might offer an The monitors in this study used an algorithm that calculated
alternative to clinical measurement of respiratory rate. electronic measurement of respiratory rate by continuously
averaging the last 3 breath cycles. The monitors also displayed
MATERIALS AND METHODS a list of electronic measurement of respiratory rate values
Study Design captured at 60-second intervals as ‘‘snapshots.’’ Research
This study used a cross-sectional design to assess the accuracy assistants were instructed to use the electronic measurement of
of 2 methods of measuring respiratory rate and was approved by respiratory rate ‘‘snapshot’’ value captured during a ‘‘quiet’’
the institutional review board of Beth Israel Medical Center, period of minimum movement and conversation. Usually, this
New York, NY. Verbal consent was required for participation. ‘‘quiet’’ moment occurred when the patient had a thermometer

Volume 45, no. 1 : January 2005 Annals of Emergency Medicine 69

Measurement of Respiratory Rates in Triage Lovett et al

in his or her mouth, although in some cases it occurred during measurements of respiratory rate and comparing electronic
note taking by the nurse or during breaks in the triage process. measurement of respiratory rate with criterion standard mea-
Electrical impedance across the thorax is less in smaller surements of respiratory rate. The Bland-Altman analysis
individuals, leading to greater amplitudes in the tracing. examines the extent to which 2 methods of measurement of the
Consequently, as the monitors were attached to each new same phenomenon ‘‘agree’’ with each other and also how this
patient, research assistants had to adjust the gain. For larger level of agreement varies across the range of respiratory rate.15
individuals, the gain had to be increased, and for smaller To explain by example, in comparing triage nurses’ measure-
patients, it was decreased. We attempted to standardize this ments of respiratory rate with criterion standard measurements
process by instructing the assistants to adjust gain until the of respiratory rate, the discrepancy between these 2 measure-
peaks came closest to the maximum recordable amplitude but ments for each patient is plotted against the average of the 2
not so high that the peaks were ‘‘cut off.’’ measurements, which yields a graphic representation of
Research assistants were trained in standardized methods for agreement across the range of respiratory rate values. It also
collecting criterion standard measurements of respiratory rate, yields measures of bias (systematic difference between triage
which they obtained by observation and then auscultation for 1 nurses’ measurements of respiratory rate and criterion standard
minute. Observation was performed as unobtrusively as measurements of respiratory rate) and 95% limits of agreement
possible, during a quiet period when the patient was not active (the interval that includes 95% of the differences between
and was not talking. For auscultation, patients were asked to sit measures). We did not perform other measures of agreement
quietly, relax, not talk, and breathe normally. The stethoscope such as the intraclass correlation coefficient and the k statistic.
was applied in a single location. The intraclass correlation coefficient is used to estimate the
If auscultation and observation were obtained successfully, degree of association between 2 measures, but it does not reveal
auscultation was the measurement used for data analysis. information about individual differences between measures. It is
Observation was used for analysis in cases where auscultation expected that 2 measures of the same phenomenon would be
was not performed. Research assistants also collected de- highly correlated, but the differences between the measures for
mographic data, presenting complaint, and time of presenta- any individual could be disparate. Alternative measures of the
tion.
Nurses working in triage were not individually tracked or
Table 1. Characteristics of enrolled patients.
compared in this study.
Characteristic No. %

Age, y
Primary Data Analysis 18–29 54 34.0
Variability of triage nurses’ measurements of respiratory rate, 30–39 36 22.6
electronic measurement of respiratory rate, and criterion 40–49 23 14.5
50–59 14 8.8
standard measurements of respiratory rate was estimated by 60–69 12 7.5
calculating the SD of each of the measures. The F test was used 70–79 9 5.7
to test whether these variabilities were statistically different from 80–89 2 1.3
each other. The F test compares values for variance (the square NR 9 5.7
of SD). Measurements were treated as unpaired. Sex
Women 81 50.9
We calculated sensitivity and specificity for triage nurses’ Men 74 46.5
measurements of respiratory rate and electronic measurement of NR 4 2.5
respiratory rate as follows. Triage nurses’ measurements of Race
respiratory rate and electronic measurement of respiratory rate White 74 46.5
values were cross-tabulated against criterion standard measure- Black 31 19.5
Asian 9 5.7
ments of respiratory rate values. All of these values were Other 29 18.2
categorized as low (\12 breaths/min), normal (12 to 20 NR 16 10.1
breaths/min), or high (O20 breaths/min). We chose these cut Ethnicity
points according to commonly reported reference ranges.9-14 Non-Hispanic 80 50.3
Sensitivity and specificity for detecting bradypnea (criterion Hispanic 66 41.5
NR 13 8.2
standard measurements of respiratory rate \12 breaths/min) Data collection
and for detecting tachypnea (criterion standard measurements CSRR data (total) 151 95.0
of respiratory rate O20 breaths/min) were calculated for triage CSRR and RNRR data 135 84.9
nurses’ measurements of respiratory rate and electronic CSRR and ERR data 149 93.7
measurement of respiratory rate, with 95% confidence intervals No CSRR data 8 5.0
(CIs) around these estimates. NR, Not reported; CSRR, criterion standard measurements of respiratory rate;
RNRR, triage nurses’ measurements of respiratory rate; ERR, electronic
We also performed Bland-Altman analyses comparing triage
measurement of respiratory rate.
nurses’ measurements of respiratory rate with criterion standard

70 Annals of Emergency Medicine Volume 45, no. 1 : January 2005

Lovett et al Measurement of Respiratory Rates in Triage

Table 2. Variability in respiratory rate measurements: SD and variance.

Minimum, Maximum, Mean, SD, Variance,
Respiratory Rate Measurement Method No. Breaths/Min Breaths/Min Breaths/Min Breaths/Min Breaths2/Min2

Results for overall analysis

RNRR 139 12 28 19.5 3.3 10.9
ERR 152 5 31 17.8 4.1 17.2
CSRR 151 8 32 18.9 4.8 23.1
Results for subgroup analysis of patients with complaints
suggestive of cardiac or respiratory illness
RNRR 28 12 28 20.1 4.0 15.8
ERR 30 10 31 19.2 4.3 18.4
CSRR 30 8 29 17.9 5.3 28.0

same phenomenon are clinically useful only if the difference triage nurses’ measurements of respiratory rate was possible for
between them is small enough, in the majority of patients, so 135 patients, and comparison between criterion standard
that treatment would not be altered according to which measure measurements of respiratory rate and electronic measurement of
was used. The k statistic could be used if the interest were respiratory rate was possible for 149 patients.
simply in determining whether there was agreement above and Variability for triage nurses’ measurements of respiratory rate
below a specific cut point but not at agreement over the range of (SD 3.3) was significantly lower than for electronic measure-
values of respiratory rate. ment of respiratory rate (SD 4.1) or for criterion standard
Data from pediatric patients were omitted from analysis measurements of respiratory rate (SD 4.8; P\.01). Variability
because of the difference in reference ranges between adults and for electronic measurement of respiratory rate was significantly
children. This decision was made after completion of data lower than for criterion standard measurements of respiratory
collection. rate (P\.05; Table 2).
Subgroup analysis was performed on patients presenting For triage nurses’ measurements of respiratory rate and
with complaints suggestive of respiratory or cardiac illness (eg, electronic measurement of respiratory rate, we determined
shortness of breath, cough, chest pain, palpitations, dizziness, sensitivity and specificity for detecting bradypnea and for
syncope). detecting tachypnea relative to criterion standard measurements
Microsoft Excel (Microsoft Corporation, Redmond, WA) was of respiratory rate. Neither triage nurses’ measurements of
used to produce the Bland-Altman analyses. CIs were calculated respiratory rate nor electronic measurement of respiratory rate was
using the VassarStats online clinical research calculators (http:// accurate in detecting abnormal criterion standard measurements
faculty.vassar.edu/lowry/VassarStats.html), and the F test was of respiratory rate. Cross-tabulation of raw numbers is shown in
performed using the HyperStat online statistical calculator Table 3, and accuracy calculations are shown in Table 4.
(http://davidmlane.com/hyperstat/F_table.html). Bland-Altman analysis showed poor agreement between
triage nurses’ measurements of respiratory rate and criterion
standard measurements of respiratory rate (95% limits of
RESULTS agreement ÿ8.6 to 9.5; Figure 1*).
One hundred fifty-nine adult patients were enrolled in the There was also poor agreement between electronic mea-
study (Table 1). Thirty-one adult patients qualified for the surement of respiratory rate and criterion standard measure-
subgroup with respiratory or cardiac presenting complaints. ments of respiratory rate (95% limits of agreement ÿ9.9 to 7.5;
Data were collected from 28 pediatric patients but were Figure 2). These limits of agreement are wider than the range of
excluded from analysis. normal respiratory rate values. Systematic bias was small for
Measurements of respiratory rate were missed in some triage nurses’ measurements of respiratory rate (C0.5) and for
patients (see bottom of Table 1). Missed criterion standard electronic measurement of respiratory rate (ÿ1.2).
measurements of respiratory rate occurred because of patient Subgroup analysis yielded results similar to those for the
refusal, severity of illness (patients immediately moved into the overall study but lacking in statistical significance. Variability
ED, or unusually brisk triage times [triage nurse refusal]). was low for triage nurses’ measurements of respiratory rate (SD
Missed triage nurses’ measurements of respiratory rate occurred
only when the triage nurse left blank the box for recording the
respiratory rate on the triage form. Missed electronic measure-
ment of respiratory rate occurred for reasons similar to those for *In all figures in this article, a small amount of random error has been
criterion standard measurements of respiratory rate misses and added to the x and y coordinates of each point so that superimposed
points are revealed. As an example, 12 patients each have triage
also because of technical difficulties, particularly when elec- nurses’ measurements of respiratory rate of 18 and criterion standard
trodes would not adhere to patients’ chests. Comparison measurements of respiratory rate of 20; adding random error teases
between criterion standard measurements of respiratory rate and these points apart visually.

Volume 45, no. 1 : January 2005 Annals of Emergency Medicine 71

Measurement of Respiratory Rates in Triage Lovett et al

Table 3. Cross-tabulation of criterion standard measurements of respiratory rate results against triage nurses’ measurements of
respiratory rate (top) and electronic measurement of respiratory rate (bottom).
Measurement, CSRR \12 CSRR 12–20 CSRR O20
Breaths/Min Breaths/Min Breaths/Min Breaths/Min Total

RNRR \12 0 0 0 0
RNRR 12–20 7 71 26 104
RNRR O20 0 15 16 31
Total 7 86 42 135
ERR \12 2 1 2 5
ERR 12–20 5 80 26 111
ERR O20 1 13 19 33
Total 8 94 47 149

4.0) versus criterion standard measurements of respiratory rate blinding would produce more careful, not less careful,
(SD 5.3; P=.07; Table 2). Assessing sensitivity, triage nurses measurements of respiratory rate.
detected tachypnea in 4 of 7 subgroup patients with criterion Criterion standard measurement of respiratory rate is not
standard measurements of respiratory rate greater than 20 perfect. Even the most careful clinical measurement of
breaths/min. Bland-Altman analysis of triage nurses’ measure- respiratory rate may suffer from inaccuracy and poor in-
ments of respiratory rate against criterion standard measure- terobserver agreement, as discussed below. Nor could electronic
ments of respiratory rate in the subgroup yielded wide 95% measurement of respiratory rate offer an alternative criterion
limits of agreement (ÿ7.2 to 10.7; Figure 3). standard.
Patients triaged at the highest level of severity bypassed
enrollment in the study. The findings of this study should not
LIMITATIONS
be applied to such patients. Our results should also not be
Nurses were aware of the purpose of the study, potentially
applied to pediatric patients, who were excluded from the
influencing their practice in measuring respiratory rate.
reported analysis.
However, common sense would lead us to believe that a lack of
Patients were not followed up beyond triage. Clinical
correlation is provided by subgroup analysis, but no in-
Table 4. Accuracy data for triage nurses’ measurements of formation on diagnosis or clinical status at disposition or
respiratory rate and electronic measurement of respiratory discharge can be provided by this study.
rate, using criterion standard measurements of respiratory A further limitation is that the study was performed at 1
rate as the criterion standard (95% CIs in parentheses). institution.
A. RNRR accuracy in detecting bradypnea. In retrospect, we could have improved the study by tracking
Bradypnea: CSRR and analyzing individual triage nurses’ measurements of
\12 Breaths/Min respiratory rate, rather than offering only an analysis of the
whole group. In the present study, if some nurses produced
RNRR sensitivity 0/7=0 0.00 (0.00–0.35)
RNRR specificity 128/128=1 1.00 (0.97–1.00)
more accurate measurements than others, these results may have
been obscured statistically.
B. ERR accuracy in detecting bradypnea.
Bradypnea: CSRR DISCUSSION
\12 Breaths/Min Pulse oximetry gained rapid acceptance in EDs during the
ERR sensitivity 2/8=0.25 0.25 (0.07–0.59) 1980s. Not only was oximetry an early claimant to the title of
ERR specificity 138/141=0.98 0.98 (0.94–0.99) ‘‘fifth vital sign’’16,17 but many authors began to ask whether
C. RNRR accuracy in detecting tachypnea.
oximetry had rendered routine measurement of respiratory rate
unnecessary.
Tachypnea: CSRR Any debate about oximetry and respiratory rate harkens to
O20 Breaths/Min
a much older discussion about ventilation versus oxygenation.
RNRR sensitivity 16/42=0.38 0.38 (0.25–0.53) There are many clinical situations in which ventilation and
RNRR specificity 78/93=0.84 0.84 (0.75–0.90) oxygenation are altered in tandem, and in such situations,
D. ERR accuracy in detecting tachypnea. oximetry and respiratory rate may provide redundant diagnostic
Tachypnea: CSRR
information. However, there are also situations in which there is
O20 Breaths/Min little relationship between hypoxia and tachypnea.17 Figure 4
lists examples of conditions in which respiratory rate may help
ERR sensitivity 19/47=0.40 0.40 (0.28–0.55)
determine diagnosis, urgency of triage, or management,
ERR specificity 88/102=0.86 0.86 (0.78–0.92)
independent of oximetry.1-3,14,18-20

72 Annals of Emergency Medicine Volume 45, no. 1 : January 2005

Lovett et al Measurement of Respiratory Rates in Triage

Figure 2. Electronic measurement of respiratory rate plotted

Figure 1. Triage nurses’ measurement of respiratory rate
against criterion standard measurement of respiratory rate. A,
plotted against criterion standard measurement of respiratory
Scatter plot; B, Bland-Altman plot of differences between
rate. A, Scatter plot; B, Bland-Altman plot of differences
measurements against mean of measurements.
between measurements against mean of measurements.

If respiratory rate is to be clinically useful, we must have Finally, even careful clinical measurement of respiratory rate,
meaningful reference ranges. Reports from as early as 184921 accepted as the criterion standard, has often been shown to have
have studied respiratory rate ranges, yet such attempts have been poor interobserver agreement. In practice, triage respiratory rate
beset by certain intrinsic problems. First, respiratory rate is measurement rarely meets the exacting requirements for
probably more subject to voluntary control than any other vital accurate measurement.6,7,25
sign. When subjects are aware that respiratory rate is being Hutchinson’s21 early report gave a range for respiratory rate
measured, the respiratory rate may change. Many methods for of 6 to 40 breaths/min in men (n=1,714), with a mean of 20.2
observing respiratory rate are sufficiently invasive or intrusive breaths/min; the mean in patients with pneumonia (n=255) was
that they further influence respiratory rate.14,17,22,23 28 breaths/min. More recently Hooker et al6 examined ranges
Second, it is to be expected that respiratory rate is context obtained by observation for 1 minute in ED patients without
specific. Even when accounting for organic illness, sedated cardiac or pulmonary complaints (n=110). Mean respiratory
patients in an ICU setting would be expected to have different rate was 20.1 breaths/min, with an SD of 4.0. The mean was
ranges than clinic patients, people living their day-to-day lives 20.9 breaths/min for women and 19.4 breaths/min for men.
outside of hospital, people who are sleeping or crying, or Moving from the issue of reference ranges, the next issue is
patients attending an ED triage.9,16,24 accuracy. There is no lack of articles critical of the accuracy of

Volume 45, no. 1 : January 2005 Annals of Emergency Medicine 73

Measurement of Respiratory Rates in Triage Lovett et al

Pulmonary
Asthma and chronic obstructive pulmonary disease
Pneumonia
Pulmonary embolism
Cardiac
Congestive heart failure
Metabolic Acidoses
Diabetic ketoacidosis
Other
Toxicologic
Cyanide; carbon monoxide
Opioids
Salicylates
Shock
Pain
Sepsis
Thoracic trauma
Other
Allergic reactions
Anxiety; hyperventilation
Dehydration
Neuromuscular disease

Figure 4. Conditions in which respiratory rate has diagnostic

value independent of oximetry.

alternative. Respiratory rate is out of step with the other vital

signs, which are measured electronically. There is a long list of
proposed alternatives: capnography by mask, nasal prongs, or
mouthpiece30-32; pneumotachography (direct flow measure-
ment) by mask or mouthpiece33,34; acoustic monitoring by
nasal microphones30,35; nasal thermistors (temperature)18,34,36;
fiberoptic nasal sensors that gauge condensation during
exhalation30,37,38; inductive plethysmography, which measures
chest wall and abdominal movement by bands encircling the
trunk (extensometry)26,33; single-compartment and multicom-
Figure 3. Subgroup triage nurses’ measurement of respiratory
partment air mattresses that measure and compare chest and
rate plotted against criterion standard measurement of
respiratory rate. A, Scatter plot; B, Bland-Altman plot of abdominal volumes39-41; mask-mounted pyroelectric polymer
differences betweeen measurements against mean of mea- strips, which detect temperature variations32,42; transdermal
surements. fiberoptic photoplethysmography30,43,44; and transthoracic
impedance plethysmography.13,18,26,30,32,37,45-47
routine clinical respiratory rate measurements.6,7,13,25-27 In his All of the above-mentioned modalities have been reported to
1957 article, Kory28 found that in a review of 248 inpatient correlate well with ‘‘criterion standard’’ clinical measurement of
charts, 94% documented respiratory rate between 18 and 22 respiratory rate. We chose to use transthoracic impedance
breaths/min. Our data add support to this literature. plethysmography in this study because it is noninvasive and can
There is an interesting contrast between our results and those be applied quickly in triage, with minimal removal of attire.
of another recent study. Worster et al29 compared triage nurse The equipment (cardiac monitors) is familiar to medical
measurements of respiratory rate with criterion standard personnel and is user friendly.
measurements in 78 ED patients and found no significant Transthoracic impedance plethysmography measures varia-
differences. However, that study used a 30-second measurement tions in electrical impedance across the thorax that occur as solid,
of respiratory rate as the criterion standard, and a 2-tailed paired liquid, and gas elements shift during respiration. Transthoracic
t test was used to test for differences. These represent important impedance plethysmography suffers from artifact, as seen in
distinctions in methodology between the 2 studies. shivering, tremors, talking, crying, and agitation and with other
If clinical measurement of respiratory rate in triage is movement. There is also potential for transthoracic impedance
inaccurate or impractical, an obvious imperative is to seek an plethysmography to produce higher respiratory rate measure-

74 Annals of Emergency Medicine Volume 45, no. 1 : January 2005

Lovett et al Measurement of Respiratory Rates in Triage

ments than other modalities by picking up on small respiratory data. PBL drafted the manuscript, and all authors contributed
efforts. substantially to its revision. PBL takes responsibility for the
Despite these potential drawbacks, many investiga- paper as a whole.
tors13,18,26,30,32,37,45-47 have reported that transthoracic imped- Received for publication December 22, 2003. Revisions
ance plethysmography correlates well with reference standard received April 6, 2004, and June 15, 2004. Accepted for
clinical measurement of respiratory rate. However, these studies publication June 18, 2004. Available online December 7, 2004.
were performed on subjects who were lying down at rest for
Presented orally at the Society for Academic Emergency
prolonged periods. Such studies mirror an ICU environment far Medicine New York regional meeting, New York, NY, April 9,
better than an ED triage environment. 2003; as a poster at the Society for Academic Emergency
Our study indicates that in a triage setting, transthoracic Medicine annual meeting, Boston, MA, May to June, 2003; as
impedance plethysmography is just as lacking in accuracy as is triage a finalist for Best Resident Research Paper at the Society for
nurses’ measurement of respiratory rate. This is demonstrated with Academic Emergency Medicine annual meeting, Boston, MA,
direct evidence (accuracy and agreement data) and indirect May to June, 2003; and as a poster at the Second Mediterra-
evidence (low variability). No techniques for electronically nean Emergency Medicine Congress, Barcelona, Spain, Sep-
measuring respiratory rate have been validated in a triage setting. tember 2003.
In summary, our study contributes to the evidence that Previously published as an abstract in Lovett PB, Buchwald
routine clinical measurement of respiratory rate in triage is JM, Stürman K, et al. The vexatious vital: a comparison of
lacking in accuracy. The authors believe that this and other clinical vs electronic measurement of respiratory rate in
studies should prompt an active search for an accurate way of triage. Acad Emerg Med. 2003;10:552-553.
measuring respiratory rate electronically in triage. Two Escort Prism monitors for transthoracic impedance
In the meantime, consideration might be given to plethysmography measurement were loaned free of charge for
alternative strategies for improving clinical measurement and the study by Medical Data Electronics.
use of respiratory rate in the ED. Physicians and nurses
Reprints not available from the authors.
might benefit from education on technique, particularly
emphasizing a 60-second recording period for respiratory rate. Address for correspondence: Paris B. Lovett, MD, Department
The value of repeated measurements also deserves emphasis. of Emergency Medicine, Columbia University Medical Center,
The accuracy, or lack of accuracy, of respiratory rate 622 West 168th Street, PH1-137, New York, NY 10032; 212-
measurements should be borne in mind by those who 305-2995; E-mail [email protected].
develop or use clinical guidelines and decision rules in which
REFERENCES
respiratory rate plays a role. The adoption of computerized
1. Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify
triage and patient record systems will offer new opportunities low-risk patients with community-acquired pneumonia. N Engl J
for continuous quality improvement efforts targeting Med. 1997;336:243-250.
respiratory rate assessment. 2. Egermayer P, Town GI, Turner JG, et al. Usefulness of D-dimer,
The discipline of emergency medicine places great blood gas, and respiratory rate measurements for excluding
emphasis on the value of the vital signs as tools for pulmonary embolism. Thorax. 1998;53:830-834.
3. Stein PD, Terrin ML, Hales CA, et al. Clinical, laboratory,
prioritizing care and for diagnosing and managing illness. roentgenographic, and electrocardiographic findings in patients
Respiratory rate is no less valuable than the other vital signs, with acute pulmonary embolism and no pre-existing cardiac or
yet the accuracy of its measurement lags behind. This pulmonary disease. Chest. 1991;100:598-603.
problem deserves the attention and the energies of emergency 4. Christenson J, Etherington J, Grafstein E, et al. Early discharge of
nurses and physicians. patients with presumed opioid overdose: development of a clinical
prediction rule. Acad Emerg Med. 2001;7:1110-1118.
5. Galle C, Papazyan JP, Miron MJ, et al. Prediction of pulmonary
We thank the nursing staff of the Department of Emergency embolism extent by clinical findings, D-dimer level and deep vein
Medicine, Beth Israel Medical Center, New York, NY. We thank thrombosis shown by ultrasound. Thromb Haemost. 2001;86:
Haviva Y. Malina, BA, Sherrie Corbin, BS, Vladimir Fridman, 1156-1160.
BS, Ingrid Lin, MD, Patrick Kelly, BA, Rachel Bakst-Sisser, BS, 6. Hooker EA, O’Brien DJ, Danzyl DF, et al. Respiratory rates in
and Nava Bak, MD, all of the Albert Einstein College of Medicine, emergency department patients. J Emerg Med. 1989;7:129-132.
7. Edmonds ZV, Mower WR, Lovato LM, et al. The reliability of vital
New York, NY, who worked as research assistants. We are also sign measurements. Ann Emerg Med. 2002;39:233-237.
grateful to Medical Data Electronics (VIASYS Healthcare) for 8. World Health Organization. Fourth Programme Report,
supplying 2 Escort Prism monitors for transthoracic impedance 1988-1989: ARI Programme for Control of Acute Respiratory
plethysmography measurement. Infections. Geneva, Switzerland: WHO; 1990:31.
9. Tintinalli JE, Kelen GD, Stapczynski JS, eds. Emergency Medicine:
A Comprehensive Study Guide. 6th ed. New York, NY: McGraw-
Author contributions: PBL obtained equipment on loan free of Hill; 2003.
charge for use in the study. PBL, JMB, and KS conceived 10. Gunn VL, Nechyba C, eds. The Harriet Lane Handbook: A Manual
the study, designed the trial, and supervised the conduct of for Pediatric House Officers. 16th ed. St. Louis, MO: Mosby;
the trial. PBL and JMB collected data. PB and PBL analyzed the 2002:514.

Volume 45, no. 1 : January 2005 Annals of Emergency Medicine 75

Measurement of Respiratory Rates in Triage Lovett et al

11. Luce JM. Respiratory monitoring in critical care. In: Goldman L, 31. Egleston CV, Ben Aslam H, Lambert MA. Capnography for
Bennet JC, eds. Cecil Textbook of Medicine. 21st ed. Philadel- monitoring non-intubated spontaneously breathing patients in an
phia, PA: WB Saunders; 2000:485. emergency room setting. Emerg Med J. 1997;14:222-224.
12. Mathers LH, Frankel LR. Stabilization of the critically ill child. 32. Dodds D, Purdy J, Moulton C. The PEP transducer: a new way of
In: Behrman RE, Kliegman RM, Jenson HB, eds. Nelson measuring respiratory rate in the non-intubated patient. J Accid
Textbook of Pediatrics. 17th ed. Philadelphia, PA: WB Sa- Emerg Med. 1999;16:26-28.
unders; 2004:280. 33. Manczur T, Greenough A, Hooper R, et al. Tidal breathing
13. Haborne D. Measuring respiratory rate. Arch Emerg Med. 1992;9: parameters in young children: comparison of measurement by
377-378. respiratory inductance plethysmography to a facemask pneumo-
14. Gravelyn TR, Weg JG. Respiratory rate as an indicator of acute tachograph system. Pediatr Pulmonol. 1999;28:436-441.
respiratory dysfunction. JAMA. 1980;244:1123-1125. 34. Marks MK, South M, Carter BG. Measurement of respiratory rate
15. Bland JM, Altman DG. Statistical methods for assessing and timing using a nasal thermocouple. J Clin Monit. 1995;11:
agreement between two methods of clinical measurement. 159-164.
Lancet. 1986;1:307-310. 35. Hök B, Wiklund L, Henneberg S. A new respiratory rate monitor:
16. Morley CJ, Thornton AJ, Fowler MA, et al. Respiratory rate and development and initial clinical experience. Int J Clin Monit
severity of illness in babies under 6 months old. Arch Dis Child. Comput. 1993;10:101-107.
1990;65:834-837. 36. Marks MK, South M, Carlin JB. Reference ranges for respiratory
17. Mower WR, Sachs C, Nicklin EL, et al. A comparison of pulse rate measured by thermistry (12-84 months). Arch Dis Child.
oximetry and respiratory rate in patient screening. Respir Med. 1993;69:569-572.
1996;90:593-599. 37. Vegfors M, Lindberg L, Pettersson H, et al. Presentation and
18. Arnson LA, Rau JL, Dixon RJ. Evaluation of two electronic evaluation of a new optical sensor for respiratory rate monitoring.
respiratory rate monitoring systems. Respir Care. 1981;26: Int J Clin Monit Comput. 1994;11:151-156.
221-227. 38. Pettersson H, Stenow END, Cai H, et al. Optical aspects of
19. Taylor JA, Del Beccaro M, Done S, et al. Establishing clinically a fibre-optic sensor for respiratory rate monitoring. Med Biol Eng
relevant standards for tachypnea in febrile children younger than Comput. 1996;34:448-452.
2 years. Arch Pediatr Adolesc Med. 1995;149:283-287. 39. Chow P, Gangadharan N, Abisheganaden J, et al. Respiratory
20. Edwards SM, Murdin L. Respiratory rate: an under-documented monitoring using an air-mattress system. Physiol Meas. 2000;21:
clinical assessment [letter]. Clin Med. 2001;1:85. 345-354.
21. Hutchinson J. Thorax. In: Todd RB, ed. Cyclopaedia of Anatomy 40. Carlson BW, Neelon VJ, Hsiao H. Evaluation of a non-invasive
and Physiology. London, United Kingdom: Longman, Brown, respiratory monitoring system for sleeping subjects. Physiol
Green, Congmans and Roberts; 1849:1079-1087. Meas. 1999;20:53-63.
22. Gilbert R, Auchincloss JH Jr, Brodsky J, et al. Changes in tidal 41. Alihanka J, Vaahtoranta K, Saarikivi I. A new method for long-term
volume, frequency, and ventilation induced by their measurement. monitoring of the ballistocardiogram, heart rate, and respiration.
J Appl Physiol. 1972;33:252-254. Am J Physiol. 1981;240:R384-392.
23. Berman S, Simoes EA, Lanata C. Respiratory rate and pneumonia 42. Cyna AM, Kulkarni V, Tunstall ME, et al. AURA: a new respiratory
in infancy. Arch Dis Child. 1991;66:81-84. monitor and apnoea alarm for spontaneously breathing patients.
24. Simoes EAF, Roark R, Berman S, et al. Respiratory rate: Br J Anaesth. 1991;67:341-345.
measurement of variability over time and accuracy at different 43. Lindberg LG, Ugnell H, Oberg PA. Monitoring of respiratory and
counting periods. Arch Dis Child. 1991;66:1199-1203. heart rates using a fibre-optic sensor. Med Biol Eng Comput.
25. Huang C, Koulouris M, Singer A. Accuracy of automated triage 1992;30:533-537.
vital signs [abstract]. Ann Emerg Med. 2002;40:S87. 44. Nilsson L, Johansson A, Kalman S. Monitoring of respiratory
26. Clancy MJ, Williams MJ. The respiratory rate [letter]. Arch Emerg rate in postoperative care using a new photoplethy-
Med. 1991;8:222. smographic technique. J Clin Monit Comput. 2000;16:
27. Krieger B, Feinerman D, Zaron A, et al. Continuous noninvasive 309-315.
monitoring of respiratory rate in critically ill patients. Chest. 45. Hamilton LH, Beard JD, Carmean RE, et al. An electrical
1986;90:632-634. impedance ventilometer to quantitate tidal volume and ventila-
28. Kory RC. Routine measurement of respiratory rate: an expensive tion. Med Res Eng. 1967;6:11-16.
tribute to tradition. JAMA. 1957;165:448-450. 46. Allison RD, Holmes EL, Nyboer J. Volumetric dynamics of
29. Worster A, Elliott L, Bose TJ, et al. Reliability of vital signs respiration as measured by electrical impedance plethysmogra-
measured at triage. Eur J Emerg Med. 2003;10:108-110. phy. J Appl Physiol. 1964;19:166-173.
30. Vegfors M, Ugnell H, Hök B, et al. Experimental evaluation of two 47. Ashutosh K, Gilbert R, Auchincloss JH, et al. Impedance
new sensors for respiratory rate monitoring. Physiol Meas. 1993; pneumograph and magnetometer methods for monitoring tidal
14:171-181. volume. J Appl Physiol. 1974;37:964-966.

76 Annals of Emergency Medicine Volume 45, no. 1 : January 2005