Prehospital Emergency Care

ISSN: 1090-3127 (Print) 1545-0066 (Online) Journal homepage: https://www.tandfonline.com/loi/ipec20

Impact of Ambulance Crew Configuration on

Simulated Cardiac Arrest Resuscitation

Ryan Bayley, Matthew Weinger, Stephen Meador & Corey Slovis

To cite this article: Ryan Bayley, Matthew Weinger, Stephen Meador & Corey Slovis (2008)
Impact of Ambulance Crew Configuration on Simulated Cardiac Arrest Resuscitation, Prehospital
Emergency Care, 12:1, 62-68, DOI: 10.1080/10903120701708011

To link to this article: https://doi.org/10.1080/10903120701708011

Published online: 26 Aug 2009.

Submit your article to this journal

Article views: 271

View related articles

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=ipec20
 IMPACT OF AMBULANCE CREW CONFIGURATION ON SIMULATED CARDIAC
ARREST RESUSCITATION
Ryan Bayley, EMT-P, BA, Matthew Weinger, MD, Stephen Meador, EMT-P, Corey Slovis, MD

ABSTRACT highest level of care to the patients they serve while at

the same time minimizing costs and maximizing effi-
Background. Despite the widespread use of both two
ciency. As these systems face increasing economic con-
paramedic and single paramedic ambulance crews, there
is little evidence regarding differences between these two
straints and paramedic staffing shortages, many have
staffing configurations in the delivery of patient care. reduced the number of paramedics per advanced life
Objectives. To determine potential differences in care pro- support (ALS) ambulance from two to one in an effort
vided by each of these ambulance configurations in the resus- to maintain or increase the number of ALS ambulances
citation of a cardiac arrest victim in ventricular fibrillation. within their systems. Recent examples include the Fire
Methods. Fifteen paramedic-paramedic and 15 paramedic- Department of New York (FDNY) and the District of
EMT crews were recruited to perform resuscitation on a Columbia Fire Department (DCFD).1,2 It is not surpris-
high-fidelity human simulator (Laerdal SimManTM ). Errors ing that these measures have drawn scrutiny from both
and their nature, time to critical interventions, and compli- the lay public and EMS providers themselves regarding
ance with continuous cardiopulmonary resuscitation (CPR) possible effects on the quality of patient care.1−3
were captured by the simulator and videotape. Results. Two
Currently, approximately 40% of ALS ambulances
paramedic crews averaged 0.7 ± 0.5 more errors of com-
mision, 0.5 ± 0.4 more errors of sequence, and 0.8 ± 0.8
in the United States are staffed by two paramedics.
more total errors per resuscitation (±95% CI; p = 0.008, 0.017, The remainder have one paramedic and a lesser-
and 0.036, respectively). For all interventions analyzed, only trained emergency medical technician (EMT).4 Unlike
time required to achieve intubation differed between the two their paramedic counterparts, EMTs are not trained in
configurations, with two paramedic crews intubating 63.9 ± Advanced Cardiovascular Life Support (ACLS) and
45.8 seconds more quickly (p = 0.009). CPR compliance was thus do not perform interventions such as cardiac
highly variable, and a meaningful statistical difference could medication administration or intubation.5 It has been
not be determined, although performance overall was poor, hypothesized that during critical scenarios where mul-
with both configurations averaging less than 50% compli- tiple interventions must be carried out in a time-
ance. Conclusion. Two paramedic crews were more error- sensitive manner, paramedic-EMT crews may perform
prone and did not perform most interventions more rapidly
more slowly than paramedic-paramedic crews because
with the exception of intubation. These data do not support
the proposition that two paramedic crews provide higher
of the EMTs reduced skill set.3 It has also been argued
quality cardiac care than paramedic-EMT crews in a simu- that when both providers are ACLS-trained and certi-
lated ventricular fibrillation arrest. Key words: ambulance fied paramedics, there exists a redundancy in critical
crews; paramedic; technician; out-of-hospital; cardiac arrest; care decision making that may reduce errors.2
mannequin. The few relevant studies to date have only indirectly
PREHOSPITAL EMERGENCY CARE 2008;12:62–68 analyzed crew configuration by using “on-scene time”
as a proxy for team efficiency. One study found that re-
ducing the number of paramedics from three to two
per ambulance greatly increased both on-scene time
INTRODUCTION and the time required to complete interventions.6 On
Emergency Medical Services (EMS) systems within the the basis of these findings, it has been hypothesized
United States and abroad are challenged to provide the that further reduction from two paramedics to a single
paramedic paired with an EMT might further increase
on-scene time and time per intervention. In contrast,
Received February 7, 2007, from the Vanderbilt University Medical an Australian study found that for a similar number
Center, Department of Emergency Medicine, Nashville, Tennessee of interventions, paramedic-EMT crews actually spent
(RB, CS); Vanderbilt University Medical Center, Nashville, Tennessee less time on-scene versus paramedic-paramedic crews.3
(MW); Emergency Medical Services, Nashville Fire Department, However, the study concluded that the difference was
Nashville, Tennessee (SM). Revision received August 9, 2007; ac-
cepted for publication August 20, 2007.
so small as to be of clinical irrelevance. Thus, there are
insufficient data to assess the potential impact of dif-
Address correspondence and reprint requests to: Ryan Bayley, EMT-
P, BA, Vanderbilt University Medical Center, Department of Emer-
ferent crew configurations on response efficiency. Fur-
gency Medicine, 703 Oxford House, 1313 21st Avenue S, Nashville, thermore, extant studies do not address differences in
TN 37232-4700. e-mail: Ryan.Bayley@ vanderbilt.edu the quality of interventions or types of errors performed
doi: 10.1080/10903120701708011 by different crew configurations.

62
Bayley et al. AMBULANCE CONFIGURATION IMPACT ON RESUSCITATION 63

The present study sought to directly compare Experimental Protocol

paramedic-paramedic and paramedic-EMT crews in
their ability to execute one standardized critical On the day of simulation, crews were provided
scenario–specifically, the resuscitation of a patient in a scripted 10-minute orientation to the SimManTM
ventricular fibrillation. Using high-fidelity simulation, Patient Simulator (Laerdal, Norway, software v3.1). The
crews were compared for critical errors, time to com- script reviewed the airway, breathing, and circulatory
plete interventions, and continuous CPR compliance capabilities of the SimManTM as well as how to perform
when evaluated against the then current American all ACLS interventions on the mannequin. Crews were
Heart Association 2000 ACLS guidelines. then given their standard issue defibrillator/monitor
(Zoll M series) and jump bags with sufficient equip-
ment and medications to perform all ACLS algorithms.
The defibrillator was modified with Laerdal hands-
free defibrillation snaps coupled with a Zoll-compatible
METHODS AND MATERIALS adaptor to allow for full-energy defibrillation to be per-
formed directly on the simulator. Crews were given
Study Design time to ask questions and configure the equipment to
Thirty full-time ALS ambulance crews were solicited their own personal preferences before the simulation
from the Nashville-Davidson County Fire and EMS sys- commenced.
tem (NFD-EMS) for participation in this study. This Crews were then asked to wait outside a room in
large, urban EMS system is a single-provider fire-based which the SimManTM was readied and placed supine
service covering 500 square miles. It employs 200 EMS on the floor 15 feet from the door. When ready, crews
paramedics and EMTs and responds to approximately were told to enter the room, assess the patient, and per-
60,000 EMS calls per year. In this system, roughly half form interventions as dictated by their current standing
of ambulances are staffed by paramedic-EMT crews, protocols.
where the EMTs have training and experience that The SimManTM was programmed to generate a
closely approximates that of the nationally recognized rhythm of refractory ventricular fibrillation until the
EMT-Intermediate classification.4 The other half of am- administration of a ventricular antiarrhythmic medica-
bulances are staffed by two paramedics. tion followed by appropriate defibrillation as detailed
A power analysis, assuming a type I error of 5%, cal- under Outcome Measures. This was the standing pro-
culated that 15 crews of each configuration would pro- tocol for NFD-EMS during the study period and con-
vide a 97% chance to detect a 30-second difference in formed to the then current 2000 AHA ACLS guidelines
the time required to execute the complete resuscitation with which all crews had many years of experience.10
assuming a standard deviation of 30 seconds. These The SimManTM logged the occurrence and time of
assumptions were based on data of simulated two- all pulse checks, defibrillations, and intervals during
rescuer ACLS studies using a similar methodology.7,8 which chest compressions were performed. A video
This same sample size would also provide a 97% chance camera recorded the resuscitation on digital videotape
to detect a 15% difference in CPR compliance (assum- (Sony DVCam DSR-1500 ) with time-coding provided
ing a 15% standard deviation) and 97% chance to detect by a Horita RM 50 II unit. The video data were imported
a difference of 0.5 errors per scenario (assuming a stan- to digital files in video-editing software (Macintosh Fi-
dard deviation of 0.5 errors). Sample size calculations nal Cut Pro 4). Frame-by-frame video analysis allowed
were performed by using the PS Power and Sample Size for validation of the SimManTM data and crews’ per-
Program.9 formance elements including the number of chest com-
The first 15 crews of each configuration to volun- pressions and times to intubation, IV access, and each
teer were chosen with no exclusion criteria. All em- medication administration.
ployees in the NFD-EMS system work full-time, and After the simulation, participants completed a
all paramedics are required to maintain current ACLS brief survey and information sheet, intended to as-
certification. At the time of this study, ACLS certifica- sess additional variables that might impact perfor-
tion was based on the AHA 2000 ECC guidelines. Crews mance such as experience level, instructor expe-
were blinded to the nature of the simulated emergency rience, frequency working with other crew mem-
(ventricular fibrillation arrest) and to the variables be- ber, and date of most recent ACLS refresher
ing evaluated, including staffing configuration. training.
Each crew member signed a written informed consent
acknowledging that he or she would be video-recorded
and was compensated $50.00 for participation in the
Outcome Measures
study, which lasted approximately 1 hour. This study Three categories of outcomes were measured: errors,
design was approved by Vanderbilt’s Institutional Re- time required to complete interventions, and compli-
view Board. ance with continuous CPR.
64 PREHOSPITAL EMERGENCY CARE JANUARY / MARCH 2008 VOLUME 12 / NUMBER 1

Errors were quantified by using an 11-item check- time to complete each intervention variables, and CPR
list of ordered clinical actions derived directly from compliance were consistent with a normal distribution.
the AHA 2000 ECC/ACLS guidelines. This checklist All measures of error were found not to be consistent
was similar to standardized tools used in ACLS prac- with a normal distribution. Data for variables consis-
tical skills assessment for the management of ventric- tent with a normal distribution were compared by us-
ular fibrillation.10,11 Per the checklist, each team was ing a one-way ANOVA. Data for all error variables
expected to (in order): check pulses, administer three were analyzed by using the Kruskal-Wallis rank sum
defibrillations, intubate, initiate IV access, administer 1 test, which does not assume a normal distribution. For
mg epinephrine, defibrillate, administer 300 mg amio- non-normal data, 95% confidence intervals and their
darone, defibrillate a fifth time, and perform a final means were generated via an Efron bootstrap calcula-
pulse check after rhythm change. Each team’s actions, tion which does not assume a normal distribution.
as recorded by the SimManTM log, were compared to Multivariate linear regressions were also run by
this checklist. For each of these 11 actions absent from using time to complete each intervention and total
the SimManTM log, an error of omission was recorded. scenario time as dependant variables. Independent
Actions recorded by the log that were superfluous to variables included team configuration, CPR compli-
these 11 actions were recorded as errors of commission. ance, individual and combined years of experience of
Once corrected for commission and omission errors, the EMS providers, instructor status, and frequency
each log was reviewed for the correct order of inter- with which providers worked with each other as self-
ventions, with each out-of-order intervention counting reported on a five-point Likert scale. Given the size of
as one sequence error. Total errors was the sum of errors the data set, each of the above independent variables
of commission, omission, and sequence for each team. was individually investigated by regression, along
For this categorical analysis, all errors were given equal with team configuration, against time required per
weight. intervention.
Time required to complete each intervention was de- With the exception of bootstrap sampling, analysis
fined as the time elapsed from completion of one inter- was carried out by using the CoStat software package
vention to the completion of the next intervention in the (CoHort Software, Monterey, CA, PC version 6.311).
sequence. This was calculated by using the time stamps Bootstrap sampling was performed with Resampling
of the SimManTM log. Time to complete the whole re- Procedures Software (University of Vermont, Burling-
suscitation was calculated similarly. If a crew failed to ton, VT, PC version 1.3) at 5,000 resamplings per cal-
complete the scenario to the point of ROSC, a cutoff culation. For all calculations, a p ≤ 0.05 was consid-
time of 12 minutes was used as the time to complete ered significant. Unless otherwise stated, normally dis-
the scenario. tributed data are presented as means ± standard de-
The third outcome measure, compliance with contin- viation, and non-normal data are presented as median
uous CPR, was calculated as the total aggregate time and 95% confidence interval.
during which chest compressions were performed, di-
vided by the total time from scenario start to ROSC. RESULTS
The SimManTM mannequin registered each chest com-
Fifteen paramedic-paramedic crews and 14 paramedic-
pression meeting a threshold depth of approximately
EMT crews were included in the final analysis. One
one inch. Intervals of continuous compressions were
calculated as time periods during which there was a
no greater than 5-second pause during compressions TABLE 1. Descriptive Statistics for Each Crew Configuration
(intended to allow for ventilation). No time corrections Paramedic Paramedic
were made for actions during which chest compres- –Paramedic –EMT
sions were correctly suspended (e.g., defibrillation). Total Years of Experience 16 ± 9.1 18 ± 8.4
This method is consistent with other prehospital studies ACLS Instructors per team 0.4 ± 0.5 0.3 ± 0.5
of CPR performance during resuscitation.12 Compres- Total errors per resuscitation ‡∗ 1 (0.8–2.0) 0 (0.1–1.1)
sion rate per minute was also calculated, by dividing the Errors of commission ‡∗ 1 (0.4–1.2) 0 (−0.1–0.3)
total number of effective compressions for the scenario Errors of omission ‡ 0 (−0.1–0.3) 0 (0.1–0.8)
by the total aggregate time in minutes during which Errors of sequence ‡∗ 0 (0.1–0.8) 0 (0–0) ∗∗
chest compressions were performed. Continuous CPR compliance 48% ± 20% 44% ± 20%
Compressions (rate/min) 82 ± 22 90 ± 18
Completion of scenario (sec) 519 ± 101 516 ± 86
Data Analysis Statistics regarding crew experience, instructor status, and basic performance
in regard to errors, CPR compliance, and speed are presented. Statistical
Descriptive statistics for the two groups and the re- differences in the number of total errors, commission errors, and sequence
sults of each outcome measure as described above were errors are noted.
‡Reported as median value (95% confidence interval), otherwise mean ± 1 SD.
tested for normal distribution using the D’Agostino- ∗
Statistically significant difference; see Figure 1.
Pearson test for non-normality. Descriptive statistics, ∗∗
Paramedic-EMT crews performed no sequence errors during the study.
Bayley et al. AMBULANCE CONFIGURATION IMPACT ON RESUSCITATION 65

paramedic crews averaging 149 seconds versus 209 sec-

onds for paramedic-EMT crews (p = 0.018). More im-
portant perhaps was the substantial variability within
each group (see standard deviations of time required
per intervention in Table 3).
To determine if factors other than team configura-
tion might be predominantly driving the time required
to compete certain interventions, multivariable regres-
sions were performed by using independent variables
including CPR compliance, individual and combined
years of experience of the EMS providers, instructor sta-
tus, and frequency with which providers worked with
each other. Controlling statistically for each of these
independent variables did not significantly affect the
results.
FIGURE 1. Comparison of error rate per resuscitation. The 95% confi- There was appreciable intragroup variation in
dence intervals for the mean difference in errors performed by each CPR compliance. Paramedic-paramedic crews aver-
crew type are presented and contrasted, showing that paramedic-
paramedic crews perform significantly more total errors, errors of
aged 48 ± 20% compliance versus 44 ± 20% compliance
addition, and errors of sequence. for paramedic-EMT crews (mean ± 1 SD), with a non-
significant 95% confidence interval difference of 3.8 ±
14.2%. Adjustment for individual and combined years
paramedic-EMT team was excluded because of a sim- of experience of the EMS providers, instructor status,
ulator malfunction where a loose ECG connector re- and frequency with which providers worked with each
sulted in no rhythm generation on the crew’s moni- other did not significantly impact the CPR adherence
tor/defibrillator, which they interpreted as asystole. results.
Descriptive statistics for the two different crew con-
figurations are summarized in Table 1. The two config-
DISCUSSION
urations did not differ significantly in total years of ex- This controlled simulation study of prehospital cardiac
perience or instructor status. Intragroup performance resuscitation provides additional insight into factors
was highly variable. Crews ranged from zero to four that affect the ACLS performance of two-person ambu-
errors per resuscitation, required anywhere from 323 lance crews. The results do not support an assertion for
to 702 seconds to complete the resuscitation, and had superiority of paramedic-paramedic crews. Moreover,
continuous CPR compliance ranging from 1.6% to 84%. the most notable finding was the substantial range of
A wide range of types of errors were observed performance of operational crews.
(Table 2). Paramedic-paramedic crews had signifi-
cantly more commission errors, sequence errors, and Errors
total errors than paramedic-EMT crews (Figure 1). Paramedic-paramedic crews averaged almost one
The two configurations were next compared re- whole error more per resuscitation. These crews had
garding the elapsed time to complete each interven-
tion (Table 3). The only time point for which a sta-
tistically significant difference was detected was the TABLE 3. Elapsed Time to Complete Intervention as a
time required to achieve intubation, with paramedic- Function of Crew Configuration
Paramedic Paramedic P
–Paramedic –EMT value
TABLE 2. Examples of Error
Scenario start to initial pulse check 15 ± 6.5 32 ± 29 0.101
Errors of omission: Initial pulse check to defib #1 55 ± 26 51 ± 32 0.466
• Failure to administer medication such as epinephrine (4 crews) Defib #1 to defib #2 17 ± 6.0 19 ± 6.6 0.252
• Failure to follow-up drug administration with defibrillation (1 Defib #2 to defib #3 32 ± 6.1 36 ± 7.8 0.257
team) Defib #3 to intubation 149 ± 23 209 ± 71 .0179 ∗
Errors of addition: Defib #3 to IV access 294 ± 79 275 ± 71 0.489
• Performing an extra defibrillation in addition to the initial 3 Defib #3 to epinephrine 193 ± 57 234 ± 65 0.105
stacked defibrillations and “drug-shock” combinations (10 crews) Epinephrine to defib #4 38 ± 16 46 ± 32 0.644
Errors of sequence: Defib #4 to amiodarone 167 ± 46 173 ± 138 0.215
• Intubation before initial defibrillation (2 crews) Amiodarone to defib #5 102 ± 51 119 ± 87 0.870
• Performing defibrillation followed by drug administration Defib #5 to ROSC pulse check 10 ± 9.7 16 ± 28 0.570
instead of drug administration followed by 30–60 sec of CPR and Total time to complete scenario 510 ± 79 524 ± 106 0.678
then defibrillation (3 crews)
Regression outcomes using crew configuration as the independent variable
Examples of the more common errors for each category of error analyzed are and elapsed time to complete each intervention as the dependant variable. All
provided. values are means ± 1 SD (sec).
66 PREHOSPITAL EMERGENCY CARE JANUARY / MARCH 2008 VOLUME 12 / NUMBER 1

more commission, sequence, and total errors, but did clude intraossesous infusion and needle cricothyroido-
not differ significantly in omission errors. tomy. Because intubation and these other complex clin-
On the basis of a qualitative review of the data, the ical skills are performed during only a minority of EMS
authors speculate that this counterintuitive result may responses, it is understandable why other studies that
be due in part to differences in how the members of each used total scene time as a proxy for efficiency may not
crew configuration interact. When two paramedics are have detected any differences between crew configura-
present, both providers may act as equals contributing tions.
to the resuscitation. Without a clear leadership hierar-
chy, each provider may be more likely to contribute to
the resuscitation as each sees fit. This may create an
CPR Compliance
environment permissive to redundancy and erroneous Total mean compliance for all crews was poor at 46%.
sequencing. There was also large intragroup variation in both the
In contrast, there is a clear leader of the paramedic- CPR compliance rate and compressions per minute.
EMT configuration, leading to better organization. Thus, this study found no significant difference be-
However, the demands on a single paramedic may be tween CPR compliance rates for the two configurations.
so onerous at times that tasks can be delayed or inad- These results are similar to those of a study of CPR
vertently omitted. This supposition is supported by the performance during actual prehospital cardiac arrests.
slower time to intubation in the paramedic-EMT group. Using a device to measure CPR compliance during
Further human factor studies are warranted in this area. prehospital resuscitations, Wik and colleagues (2005)
Little data exist quantifying the individual impact found that European EMS providers performed CPR
of many of these interventions on patient outcome. only 52% of the time during actual cardiac resuscita-
In fact, only defibrillation and CPR have been clearly tions using the same 2000 ACLS guidelines as this study.
shown to improve patient outcome.13 It is thus difficult Research emphasizes the importance of continu-
to ascertain the impact per error or assert unequivo- ous CPR using a high-compression rate.15 There is a
cally that one type of error is more clinically signif- large body of evidence showing that slow compres-
icant than another. Eighty-three percent (10 of 12) of sions, frequent interruptions, and significant “hands-
the additional interventions performed by paramedic- off” time during CPR precludes adequate cardiac and
paramedic crews were defibrillations, which one might cerebral perfusion pressures, thus adversely affecting
argue may be less detrimental to a patient than would outcome.12,16,17 Thus, it is notable that neither crew
be the omission, for example, of an antiarrhythmic configuration was able to accomplish CPR that would
drug, or a sequence error, such as intubation before likely have been of clinical benefit to an actual patient.
initial defibrillation. Regardless, this study shows a Why did some crews dramatically outperform others
substantial incidence of care process deviations, which regardless of crew configuration? In the early minutes
many clinical and patient safety experts believe are a of a resuscitation, CPR compliance may be poor due
meaningful proxy for lower quality care. to providers dividing their time between CPR perfor-
mance and completion of all of the other ACLS inter-
ventions. However, CPR compliance does not improve
Speed of Interventions dramatically even after all of these other interventions
There were no significant differences between the two are completed.18 While the large variability in CPR com-
crew configurations in terms of the efficiency with pliance in this study could be an artifact of it being a
which most interventions were performed, with the ex- simulated study, our experience with in-hospital resus-
ception of time required to complete intubation. There citations suggests otherwise and further study seems
are limited data to suggest that time to intubation may warranted, particularly with the increased focus on
independently affect patient outcome. When intubation CPR in the 2005 ECC guidelines.
times are controlled for other variables and analyzed by
quartiles, one study found that patients whose intuba-
tion time was in the fastest quartile were twice as likely
Study Limitations
to survive.14 In our study, six of the seven crews in the This study used the 2000 ACLS guidelines, because
fastest quartile were paramedic-paramedic crews. it was initiated just prior to publication of the 2005
In comparison with cardiac medication administra- ACLS guidelines and the crews were still operating
tion or defibrillation, intubation requires significant under the 2000 guidelines. Some states did not revise
time to not only perform but also to prepare for it. their EMS protocols with the new guidelines until early
Having two providers who perform intubation regu- 2007,19 and some providers will not undergo formal
larly and thus are familiar with the setup and execution ACLS recertification under the new guidelines until
well may facilitate the speed with which it is accom- early 2008. The use of guidelines with which crews had
plished. Other interventions that may similarly benefit years of experience in actual clinical encounters is ad-
from the involvement of advanced providers might in- vantageous in that results are less likely to be driven by
Bayley et al. AMBULANCE CONFIGURATION IMPACT ON RESUSCITATION 67

crew unfamiliarity or lack of clinical experience with variability in all aspects of performance regardless of
recently changed guidelines. However, given the 2005 crew configuration.
ACLS guidelines’ emphasis on CPR and attempts to The results of this study suggest other possibilities
streamline other interventions, the results of this study for future investigation. Larger multicenter simulator-
cannot be assumed to carry over to the new guidelines based studies could be undertaken to further eluci-
and must be reconfirmed. date possible differences between crew configurations
A second limitation is that this was a simulation. in terms of efficiency. The differences in error rate noted
Simulator-based studies allow the direct observation of in this study could be further elucidated by using field
participants under highly controlled and reproducible data to try to quantify their actual impact on patient
circumstances. However, participants may act differ- care, if any. Perhaps the most important avenue for fu-
ently than they might during actual patient care. The ture research is delineation of the causes of the wide
nature of this confound is unknown: participants may variation in performance and interventions to decrease
regard a simulation less seriously because it is not it. In both configurations, a minority of crews were able
“real,” or they may perform with more diligence know- to achieve error-free resuscitation with high CPR com-
ing that they are being observed and reimbursed. Fur- pliance. These crews could not be reliably identified
thermore, participants who volunteer in any study may by any of the independent variables collected includ-
differ from the actual population. Nevertheless, the re- ing crew configuration or experience. Understanding
sults of this study do corroborate those of prior field the factors driving this variability and developing in-
studies of prehospital cardiac resuscitation. terventions to ensure maximal performance and de-
A third limitation of this pilot hypothesis-generating creased variability could provide significant benefit to
study was that it had a relatively small sample size. EMS systems regardless of the crew configuration em-
Furthermore, the intragroup variability observed was ployed.
greater than assumed when making initial power cal-
culations. Regardless, the study was sufficiently pow- We thank Ray Booker, simulation engineer for Vanderbilt University
ered to detect statistical differences in the error rates Medical Center, for his contributions in executing the simulations
between the two crew configurations. For the resuscita- for this study; David Sewell, assistant Chief, Nashville Fire Depart-
tion time overall and the times to complete many major ment, for his contributions in participant recruitment and his role as
fire department liaison; and Vanderbilt University Medical Center
interventions, differences between the two configura- Department of Biostatistics, for contributions to the analysis of this
tions as small as 20% would likely have been detected data.
if present. The study was not sufficiently powered to
detect meaningful statistical differences in CPR com-
pliance. However, the high variability demonstrated is References
itself noteworthy and serves to highlight future direc-
tions for research. 1. Wilbur DQ. D.C. paramedic shortage causes concern. Washington
Post, May 7, 2005, B03.
2. Medina J. Plan to place one paramedic per ambulance draws
CONCLUSION anger. New York Times, January 10, 2005;B01;3.
3. Kelly AM, Currell A. Do ambulance crews with one advanced
This study does not support the assertion that paramedic skills officer have longer scene times than crews with
two? Emerg Med J. 2002;19(2):152–4.
paramedic-paramedic crew configurations provide bet-
4. Williams, DM. 2005 JEMS 200 City Survey. J Emerg Med Serv.
ter resuscitation care than paramedic-EMT crews. In 2006;31(2):44–61, 100–1.
contrast, paramedic-paramedic crews in this study ex- 5. Lilja GP, Swor RA. In Prehospital Care Emergency Medicine, 5th
hibited more total errors, more errors of commission, edition. Tintinalli, ed. New York: McGraw–Hill, 2000.
and more errors of sequence per resuscitation. More- 6. Brown LH, Owens CF Jr, March JA, Archino EA. Does ambu-
over, the two configurations did not differ signifi- lance crew size affect on–scene time or number of prehospital
interventions? Prehosp Disas Med. 1996;11:214–7.
cantly in terms of speed to perform most interventions. 7. Wik L, Steen PA. The ventilation/compression ratio influences
Paramedic-paramedic crews did, however, outperform the effectiveness of two rescuer advanced cardiac life support on
their paramedic-EMT counterparts in the efficiency a manikin. Resuscitation. 1996;31(2):113–9.
with which intubation was performed. Whether these 8. Kill C, Giesel M, Eberhart L, Geldner G, Wulf H. Differences in
findings continue to hold true during actual resusci- time to defibrillation and intubation between two different ven-
tilation/compression ratios in simulated cardiac arrest. Resusci-
tations under the new 2005 ACLS guidelines or are tation. 2005;65(1):45–8.
sufficient to affect actual patient outcomes requires fur- 9. Dupont WD, Plummer WD. PS Power and Sample Size Program.
ther investigation. In regards to CPR compliance, the Controll Clin Trials. 1997;18:274.
wide intragroup variations reduced the power of this 10. Part 6: Advanced Cardiovascular Life Support. Section 7: al-
study to detect meaningful statistical differences. How- gorithm approach to ACLS. 7C: a guide to the international
ACLS algorithms. European Resuscitation Council. Resuscita-
ever, the data show that crews of both configurations tion. 2000;46 (1–3):169–84.
fail to achieve high compliance with CPR guidelines. 11. Part 6: Advanced Cardiovascular Life Support. Section 7: al-
Notably, the crews in this study demonstrated marked gorithm approach to ACLS. 7C: a guide to the international
68 PREHOSPITAL EMERGENCY CARE JANUARY / MARCH 2008 VOLUME 12 / NUMBER 1

ACLS algorithms. European Resuscitation Council. Resuscita- Emergency Cardiovascular Care. Circulation. 2005;112:IV-19– IV-
tion. 2000;46 (1–3):169–84. 34.
12. Wik L, Kramer–Johansen J, Myklebust H, Sorebo H, Svensson L, 16. Sato Y, Weil MH, Sun S, Tang W, Xie J, Noc M, Bisera J. Adverse
Fellows B, Steen PA. Quality of cardiopulmonary resuscitation effects of interrupting precordial compression during cardiopul-
during out–of–hospital cardiac arrest. JAMA. 2005;293(3):299– monary resuscitation. Crit Care Med. 1997;25:733–6.
304. 17. Yu T, Weil MH, Tang W, Sun S, Klouche K, Provoas H, Bisera J.
13. Part 7.2: Management of Cardiac Arrest. 2005 American Heart Adverse outcomes of interrupted precordial compression during
Association Guidelines for Cardiopulmonary Resuscitation and automated defibrillation. Circulation. 2002; 106;368–72.
Emergency Cardiovascular Care. Circulation 2005;112(suppl I): 18. Kramer–Johansen J, Wik L, Steen PA. Advanced cardiac life sup-
IV–58–IV–66. port before and after tracheal intubation–direct measurements of
14. Shy BD, Rea TD, Becker LJ, Eisenberg MS. Time to intubation quality. Resuscitation. 2006;68(1):61–9.
and survival in prehospital cardiac arrest. Prehosp Emerg Care. 19. McVicar D. Letter to New Hampshire EMS providers regarding
2004;8(4):394–9. 2005 ACLS Guideline Updates. State of New Hampshire Depart-
15. Part 4: Adult Basic Life Support. 2005 American Heart As- ment of Safety Division of Fire Standards and Training and Emer-
sociation Guidelines for Cardiopulmonary Resuscitation and gency Medical Services, March 11, 2006.