DTIC ADA619106: Computerized Decision Support System Improves Fluid Resuscitation Outcomes Following Severe Burns: An Original Study

0.3 MB·English

by  Defense Technical Information Center

Computerized decision support system improves fluid resuscitation
following severe burns: An original study*
Jose´ Salinas, PhD; Kevin K. Chung, MD; Elizabeth A. Mann, MS; Leopoldo C. Cancio, MD; George C. Kramer, MD;
Maria L. Serio-Melvin, RN; Evan M. Renz, MD; Charles E. Wade, PhD; Steven E. Wolf, MD
Objective: Several formulas have been developed to guide Measurements and Main Results: We found no significant dif-
resuscitationinseverelyburnedpatientsduringtheinitial48hrs ference between the Model and computer decision support system
after injury. These approaches require manual titration of fluid groups in age, total body surface area, or injury mechanism. Total
that may result in human error during this process and lead to crystalloidvolumeduringthefirst48hrspostburn,totalcrystalloid
suboptimal outcomes. The goal of this study was to analyze the intensivecareunitvolume,andinitial24-hrcrystalloidintensivecare
efficacyofacomputerizedopen-loopdecisionsupportsystemfor unitvolumewerealllowerinthecomputerdecisionsupportsystem
burn resuscitation compared to historical controls. group. Infused volume per kilogram body weight (mL/kg) and per
Design: Fluid infusion rates and urinary output from 39 se- percentageburn(mL/kg/totalbodysurfacearea)werealsolowerfor
verelyburnedpatientswith>20%totalbodysurfaceareaburns thecomputerdecisionsupportsystemgroup.Thenumberofpatients
were recorded upon admission (Model group). A fluid-response who met hourly urinary output goals was higher in the computer
modelbasedonthesedatawasdevelopedandincorporatedinto decisionsupportsystemgroup.
acomputerizedopen-loopalgorithmandcomputerdecisionsupport Conclusions: Implementation of a computer decision support
system.Thecomputerdecisionsupportsystemwasusedtoresus- system for burn resuscitation in the intensive care unit resulted in
citate32subsequentpatientswithsevereburns(computerdecision improved fluid management of severely burned patients. All mea-
supportsystemgroup)andcomparedwiththeModelgroup. suresofcrystalloidfluidvolumewerereducedwhilepatientswere
Setting: Burn intensive care unit of a metropolitan Level 1 maintainedwithinurinaryoutputtargetsahigherpercentageofthe
Trauma center. time.Theadditionofcomputerdecisionsupportsystemtechnology
Patients: Acute burn patients with >20% total body surface improvedpatientcare.(CritCareMed2011;39:2031–2038)
arearequiringactivefluidresuscitationduringtheinitial24to48 KEY WORDS: automated systems; burn care; burn resuscitation;
hours after burn. computerdecisionsupport;crystalloidinfusion;informationtechnology
T
he promulgation of computer nology. The critical care arena has been Each year, approximately 40,000 adult
decision support systems typicallyrecalcitranttoacceptcomputers patientswithsevereburnsrequirehospi-
(CDSSs)hasnotbeenaswide- withinthepatientcareenvironment.Au- talization, approximately 4,000 of whom
spreadinmedicineasinother tomating complex clinical decision- dieoftheirinjuries(1–3).Burnmanage-
areas that depend on information tech- making paradigms, coupled with regula- ment requires specialized expertise and
toryandtechnicallimitations,isjustone treatmentoptionsthatmaynotnormally
of the hurdles that has kept advanced be available at nonburn centers. In addi-
*Seealsop.2178. informationsystemsfrombeingdeployed tion,personneltrainedinburncaremust
FromtheU.S.ArmyInstituteofSurgicalResearch beyond simple documentation roles. provide prompt initialization of fluid
(JS, KKC, EAM, LCC, MLS, EMR, SEW), Fort Sam
However, as patient care continues to therapy coupled with continuous atten-
Houston,TX;DepartmentofAnesthesiologyandSur-
evolveintousingadditionaldevices,sen- tive care at admission to the intensive
gery (GCK), University of Texas Medical Branch,
Galveston,TX;andUniversityofTexasHealthScience sors, and information sources related to care unit (ICU) (4). Appropriate fluid
Center(CEW),Houston,TX. thepatientcondition,theneedforautoma- titration during the initial resuscitation
Supported, in part, by the National Institutes of
tion and decision support in this environ- periodofacuteburnisvitalandhasbeen
HealthandtheU.S.ArmyCombatCasualtyCareRe-
mentbecomescritical.Similarly,aspatient thecornerstoneofeffectiveburncare(4).
searchProgram.
Dr. Salinas, Dr. Chung, Ms. Mann, Dr. Cancio, Dr. care becomes more specialized while the Standard pathophysiologic response to a
Kramer,andDr.WolfhaveaU.S.patentpendingforthe numberofprimarycarepersonnelisbeing thermal injury results in a burn-induced
invention described in the manuscript. The remaining
reduced, having automated systems that intravascularfluiddeficitfeaturingasub-
authorshavenotdisclosedanypotentialconflictsofinterest.
have the skills and knowledge of expert stantialplasmavolumedeficitduringthe
The opinions or assertions contained herein are
the private views of the authors and are not to be providersbecomesincreasinglyimportant. initial48hrspostburn,whichengenders
construedasofficialorasreflectingtheviewsofthe An example of the need for decision hypovolemic shock and generalized
DepartmentoftheArmyortheDepartmentofDefense.
support technology is the burn critical edema formation (4). These include a
For information regarding this article, E-mail:
careenvironment.Inthisarena,effective principalfluidshiftintothesurrounding
[email protected]
Copyright©2011bytheSocietyofCriticalCare initial management of severe burns is interstitialspace(i.e.,third-spacing)that
MedicineandLippincottWilliams&Wilkins critical for minimizing both resuscita- has to be treated to avoid burn shock
DOI:10.1097/CCM.0b013e31821cb790 tion-related morbidity and mortality. conditions (5–6).
CritCareMed2011Vol.39,No.9 2031
Report Documentation Page Form Approved
OMB No. 0704-0188
Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington
VA 22202-4302 Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it

does not display a currently valid OMB control number
1. REPORT DATE 2. REPORT TYPE 3. DATES COVERED
01 SEP 2011 N/A -
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
Computerized decision support system improves fluid resuscitation
5b. GRANT NUMBER
outcomes following severe burns: an original study
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) 5d. PROJECT NUMBER
Salinas J., Chung K. K., Mann E. A., Cancio L. C., Kramer G. C.,
5e. TASK NUMBER
Serio-Melvin M. L., Renz E. M., Wade C. E., Wolf S. E.,
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION
United States Army Institute of Surgical Research, JBSA Fort Sam REPORT NUMBER
Houston, TX
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)
11. SPONSOR/MONITOR’S REPORT
NUMBER(S)
12. DISTRIBUTION/AVAILABILITY STATEMENT
Approved for public release, distribution unlimited
13. SUPPLEMENTARY NOTES
14. ABSTRACT
15. SUBJECT TERMS
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF
ABSTRACT OF PAGES RESPONSIBLE PERSON

a REPORT b ABSTRACT c THIS PAGE UU 8
unclassified unclassified unclassified
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std Z39-18
Effective early fluid management in MATERIALS AND METHODS dations because of clinical complications,
these patients has been shown to reduce death, or other factors were excluded. The
poor outcomes and decrease complica- CDSS Development controlcohortincludedonlypatientswhosur-
tion rates (7–15). Similarly, delayed or vivedfortheinitial24hrspostadmissionand
inadequate fluid resuscitation is associ- We obtained local approval from the had undergone at least 24 hrs of fluid resus-
BrookeArmyMedicalCenterInstitutionalRe- citation.Student’sttestsandchi-squaretests
ated with increased morbidity and mor-
view Board for a study to collect and analyze were used to compare variables using the
tality rates (8). Most providers strive to
data from 40 consecutive adult patients with SPSSVersion16(SPSS,Chicago,IL)statisti-
standardizeresuscitationtomaintainad-
(cid:1)20%TBSAburnsadmittedtoourburnICU calanalysissystem.Left-skewedvariablesthat
equate urinary output (UOP) volumes as
from November 2004 to February 2007. A werenotnormallydistributedunderwentlog-
a proxy to cardiac output by continued computerized digital acquisition system was arithmic(log)transformationtoachievenor-
hourly titration of crystalloid fluids. Cal- used to capture UOP volumes from a digital mality. Other nonparametric variables were
culation of initial infusion rates is typi- urimeter (Bard Criticore, Murray Hill, NJ) comparedbyusingMann-WhitneyUtests.An-
callybasedontheBrookeortheParkland during the initial 48-hr resuscitation. One alyzed variables included mortality, 24-hr
formulas (16–18) that define total vol- subjectwasdroppedfromthestudybecauseof crystalloid volume, 48-hr crystalloid volume,
umerequirementsovertheinitial24hrs missingdata,resultinginatotalof39patients prehospital volume, 48-hr mL/kg volume,
afterburn,withaprescribedgoalranging used for model development. Research per- 48-hrmL/kg/TBSAvolume,andpercentageof
between 2 mL/kg/total body surface area sonnel recorded crystalloid infusion rates timewithintargetUOP.Analysisofoutcomes
manuallyatthetopofeachhour.Fluidintake includedICU-freedays(IFDs),andventilator-
(TBSA) and 4 mL/kg/TBSA and target
andoutputrateswereanalyzedforallpatients freedays(VFDs).IFDsandVFDsweredefined
UOPs between 30 and 50 mL/hr. This
todetermineaveragefluidratesateachhour, as the number of each individual’s total days
formulaincludesgivinghalftherequired
rate of change from previous hour, and per- subtractedfromthecohortmean.
fluid in the initial 8 hrs of resuscitation
centageoftimewithinacceptableUOPtargets
(18–21). However, burn studies reported
of 30 to 50 mL/hr. Mean fluid rate and UOP RESULTS
intheliteratureconsistentlyreportover- werecomputedforeachhourafterburnforup
resuscitationofburnpatientswithvalues to48hrstodeterminetheexpectedmeanfluid
System Development
that are consistently higher than the rateforthecohort.Asetofcurve-fitmethods
maximum 4 mL/kg/% TBSA directed by was used on both the infusion and the UOP Results of overall total crystalloid in-
the Parkland rate. A meta-analysis of 31 data sets to derive their best-fit function. A fusionperhourshowedacontinuousde-
burn studies demonstrated that 86% re- fluid-response equation was derived for esti- cay pattern over the initial 48 hrs of re-
portedmeanfluidvolumesexceedingthe matingaveragecrystalloidfluidrequirements suscitation (Fig. 1A). An exponential
Parkland estimates (22). for each milliliter of UOP generated at each decay function given by f(hpb) (cid:3)
One solution is the use of informa- hour. Colloid use during resuscitation in the Xe((cid:4)Y(cid:5)hbp)hadthebestfitforthecrys-
first24hrsisnotstandardpracticeinourunit
tionsystemscoupledwithdecisionsup- talloid infusion rates during this time,
and was not considered for model develop-
port technology to provide recommen- withX,Yrepresentingthefunctioncoef-
ment in this instance (the model is based on
dations for fluid volumes based on ficients of decay and hpb representing
titrationofhourlylactatedRinger’ssolution).
measuredbiologicalresponsesinasim- A target UOP rate of 40 mL/hr was used for hours post burn. The UOP values had a
ilar population of burned patients. model development (average of upper and linearfunctionfittothedataoftheform
These recommendations consider the lower bounds of 30 mL/hr and 50 mL/hr). f(hpb) (cid:3) M (cid:5) hpb (cid:6) N, with M, N rep-
patient’s own responses to assist care EquationmodifiersforTBSAandweightwere resenting the linear coefficients of the
personnel during the resuscitation developed to adjust recommendations based function(Fig.1B).Theexponentialdecay
phase. CDSSs have been successfully on these additional parameters. Final CDSS function model suggested the need for
used in the clinical setting for several model equations were programmed into a larger infusion rates at the beginning of
computer algorithm to generate recommen- the resuscitation period to achieve the
years (23–24). Using decision support
dationsforcrystalloidtitrationforeachhour. same UOP than in later hours. The ratio
technology,wedevelopedanovelCDSS
TheCDSSwasdevelopedusingtheJava(Sun offluidinfusiontotheUOPforeachhour
for resuscitation of patients with acute
Microsystems, Palo Alto, CA) programming postburnisthereforerepresentedbythe
burns (TBSA (cid:1)20%) during the initial
languageanddeployedonacomputersystem nonlinearratiooff(fluid_ratio)(cid:3)fluid_rate/
48 hrs after burn. The system imple- withineachroominourICU. UOP(cid:3)()Xe((cid:4)Y(cid:5)hbp)/(M(cid:5)hpb(cid:6)N)for
ments an open-loop concept that pro-
each postinjury hour given by hbp. This
vides recommendations to assist users
Analysis equationshowstheexpectedfluidrateper
in making decisions during patient
milliliter of UOP of the model cohort. For
care. These recommendations are then Historical control data from the 39 pa-
algorithm and software implementation,
usedbylicensedproviderstohelpthem tientsenrolledduringmodeldevelopmentwas
thefunctionwasdividedintothreephases
make better treatment decisions when compared with data from patients on CDSS
(I,II,III)bydividingthemaximumamount
admitted from November 2007 to January
determining fluid rates over the next offluidpredictedintothreeequalfluidrate
2009andanalyzedforimprovementsincrys-
hour. The CDSS was deployed in our sections.Thethreephasescorrespondedto
talloidfluidmanagementandoutcomes.Dur-
burn ICU in November 2007 and has periods when patient fluid needs changed
ing this period, 66 consecutive patients were
becomeourcurrentpracticeforallnew resuscitatedwiththeCDSSduringtheinitial fromahighvolumeduringthefirstperiod
patients admitted to our burn ICU. The 48 hrs post-ICU admission. We included pa- totheminimumrateattheendofthe48
goal of this study was to analyze the tientsontheCDSSwhohadatleast24hrsof hrs. They allowed the system to reduce
efficacy of the CDSS in burn resuscita- recommendations from the software for this fluid change recommendations as the pa-
tion compared to historical controls. analysis.Patientswith(cid:2)24hrsofrecommen- tientbecamemorestableandpreventedthe
2032 CritCareMed2011Vol.39,No.9
Table 1. Rule of 10 for determining initial
crystalloidresuscitationvolume
Weight
(kg) InitialRate(mL/hr)
(cid:2)80 %TotalBodySurfaceArea(cid:5)10
(cid:1)80 %TotalBodySurfaceArea(cid:5)10(cid:6)
100mLforevery10kgover80
the 40-mL/hr UOP target. Additionally,
because changes in UOP at each hour
may be driven by several factors in addi-
tiontorenalfunctionandfluidvolume,a
UOP projection formula was imple-
mentedtocalculatetheexpectedUOPfor
thenexthour.Thisformulaisbasedona
projectionofthelastthreenonzeroUOP
values using a linear estimator function
to estimate the next hour’s UOP. The
linear estimator fits a line to the last
three UOP measures to reduce the
amountofnoiseintheUOPresultsdueto
otherfactorsnotrelatedtoresuscitation.
The resulting projected UOP value con-
stituted the UOP input into the recom-
mendation calculations. Initial recom-
mendations (in milliliters per hour) of
crystalloids within the model were based
Figure 1. A, Average crystalloid infusion rates captured from patient data (bar) compared with
on the Rule of 10 approach (Table 1),
exponential decay model (line). B, Average urine output (UOP) volume captured from patient data
which provides a simple and rapid deri-
(bar)comparedwithlinearmodel(line).
vation of the initial fluid rate by multi-
plying the TBSA by a factor of 10 with
additional fluid volume for overweight
system from dramatic changes in recom- thepatientmodelcohortsandreflectde-
patients over 80 kg (25).
mendationratesasthepatientmovedfrom viations of TBSA and weight from the
CDSSdeploymentconsistedofadual-
one phase to another. Phase I included model (i.e., patients with TBSA (cid:1) model
screencomputersystemrunningtheJava
hours0to13afterburn;PhaseII,hours14 averagewillrequireadditionalfluid).The
run-time environment. Normal configu-
to33afterburn;andPhaseIII,hour33and resulting equations are independent of
rationforaresuscitationpatientusedthe
greater.Foreachphase,theaveragerateof hourspostburnandrepresentaconstant
top computer screen to run the main
infusion change at each hour was calcu- used to further modify the results of the
CDSS application and the bottom screen
latedwithacorrespondingsetofconstants. model equation based on the patient’s
to run the Essentris electronic charting
Using these constants and the previous TBSA and weight. A final burn modifier
system (Clinicomp, San Diego, CA). UOP
hour’sinfusionrate,wederivedanewfor- coefficient was included to reduce the
dataweremanuallyinputintothesystem
mulaforcomputingtheamountofcrystal- effect of large variations in fluid recom-
fromthedigitalreadoutoftheelectronic
loid infusion to change from the previous mendations given by the function BM (cid:3)
urimeter(BardCriticore,MurrayHill,NJ).
hourtobringthepatientdowntoatarget I /(TBSA (cid:5) 10), where BM is the burn
t-1 FinallactatedRinger’srateswereadjusted
levelgiventheexpectedresponseandhour modifier coefficient and I is the crystal-
t onthefluidinfusionpumps(HospiraPlum
postburn. loid infusion at hour postburn t. Results
A(cid:6),Hospira,LakeForest,IL).
To further compensate for intra- ofthesemodifiersweremultipliedbythe
The system included several charac-
patientvariationsinTBSAandweight,we previoushour’srateandusedasthefinal
teristics that allow for ease of use and
implementedasetofgeneralizedlogistic crystalloid recommendation for the next
integration into the nursing and critical
curves as the modifiers to the base fluid hour or half-hour rate. To guarantee
care workflow environment, including
rate model equation. These provided a minimalchangetotherecommendedin-
the following:
patient-specificmodifierbasedonthepa- fusion as the patient’s UOP approaches
tient’s weight and TBSA given by the the target of 40 mL/hr, we used an addi- ● Network deployability. CDSS design
equation Y (cid:3) A (cid:6) C/(1 (cid:6) Te(cid:4)B(X(cid:4)M))1/T, tional weighting factor based on an in- and implementation were based on a
whereA,B,X,M,andTrepresentcoeffi- vertedGaussianfunctionasanadditional centralized application system that re-
cients defining the logistic curve for filter. This equation, minimized at the sides on a common file server within
valueY(TBSAorweight).Coefficientsfor target fluid rate, is given by G (cid:3) our institute. The CDSS application is
these two equations were chosen based 1-e(cid:4)(UOP(cid:4)40)2ˆ/25 to further limit changes automatically downloaded and exe-
onthedistributionofTBSAandweightof to fluid rates as the patient approaches cuted on the bedside computer in the
CritCareMed2011Vol.39,No.9 2033
ICUwhenrequiredthroughastandard
shared network drive available to all
ICU computers.
● Java implementation. The CDSS was
written in the Java programming lan-
guage to allow the application to exe-
cute independently of the underlying
hardware or computer system.
● Link into electronic charting. The
CDSS provided a link into the elec- Figure2.Consortdiagramfortestcohorts.CDSS,computerdecisionsupportsystems.
tronic charting system for patient up-
dates and data verification. These data Table2. Demographicsandinjurycomparisonbetweencontrolandcomputerdecisionsupportsystem
included additional fluids and physio- cohorts
logic parameters of the patient.
ComputerDecisionSupport
Workflow of the CDSS was incorpo- Parameter Control(n(cid:3)38) System(n(cid:3)32) p
rated into the critical care procedures
required during patient admission and Age(years) 50(cid:7)21 44(cid:7)16 .18
subsequent resuscitation. The main Weight(kg) 88(cid:7)24 87(cid:7)23 .83
Gender Male:n(cid:3)28,74% Male:n(cid:3)25,78% .67
CDSSapplicationscreeniscomposedofa %TotalBodySurfaceAreaa 40(cid:7)19 39(cid:7)16 .94
two-panel window application. Warning %FullThicknessa 11.5(QI:0,11.50,40.75) 9(QI:0,9,16.75) .07
ruleswereimplementedbothgraphically InhalationInjury(%)b Positive:n(cid:3)11,29% Positive:n(cid:3)10,31% .83
and verbally when the patient reached
200 mL/kg and 250 mL/kg in the last 24 QI,quartiles(25%,50%,75%).
aVariablesarenotnormal.LogtransformusedforStudent’sttestcomparison.bmeasuredviaa
hrs. Additional markers for 2 mL/kg/
fiberopticbronchoscopytest.
TBSA and 4 mL/kg/TBSA were depicted
graphically on this window. The bottom
applicationpanelwasusedfortheintake/
patient may be restarted on the CDSS against 38 historical control patients
output table and fluid balance display.
system after a prolonged break at the (modeldevelopmentcohort)thatmetcri-
Thelowerwindowpanelwasusedfordata
request of the attending physician (i.e., teria for inclusion collected between No-
input and hourly graphing displays.
afterreturningfromsurgery)ifneededif vember2004andFebruary2007(Fig.2).
Hourly input was done through a table
thepatientisstillwithinthe48-hrresus- Both the CDSS and the control cohorts
interface where users were able to input
citation window. hadvaluesforTBSA,age,weight,gender,
all fluid values given to the patient for
eachhouroftheresuscitation.Anhourly The system incorporated a set of rules and rate of inhalation injury that were
graphingsystemprovidedafluidbalance andlimitsbasedonasetofconsensusdef- not statistically different (Table 2). A log
viewshowinginfusionfluidrates,recom- initionsagreedtobyaclinicalpanelofburn transform was used on the TBSA and
mendations, and UOP values for each careproviders.Recommendationsfromthe full-thickness continuous variables to
hour. Fluids were color coded according CDSS outside normal institute guidelines normalize them before analysis. The
tothefluidtypeandfurtherbrokendown requiredapprovalfromalicensedcarepro- CDSS patients had statistically lower re-
byprehospitalandICUrates.UOPvalues vider. A real-time system clock prompted suscitationvolumeovertheinitial48hrs
the user at the recommended time points
werecolorcodedaccordingtowherethey (including prehospital, emergency de-
during the resuscitation (every 60 or 30
fell within the specified range of 30 partment, and ICU volumes of crystal-
mins) for data from the current infusion
mL/hr to 50 mL/hr. Values higher than loids) (Fig. 3A). Total fluid volume over
pumpsandelectronicurimeter.
50mL/hrwerecodedred,whereasvalues 48hrs(prehospitalandresuscitation)was
below the range were coded yellow and TheCDSSgeneratedrecommendations reduced from 26,309 (cid:7) 14,454 mL to
green when on target. for new lactated Ringer’s rates automati- 15,605 (cid:7) 5707 mL (p (cid:2) .05) (Fig. 4A).
cally once the user input and verified the
Termination of the CDSS is deter- Prehospital crystalloid infusion volume
required data. If the user chose not to ac-
mined by a combination of clinical prac- was significantly different between the
cept the new fluid rate recommendations,
tice guidelines and built-in system rules two groups; controls received an average
theCDSSpromptedtheusertodocument
including the following: of4993(cid:7)4081mLvs.3222(cid:7)2290mL
the reason for deviating from the recom-
for the CDSS group (p (cid:2) .05). When
● Terminationhasbeenrequestedbythe mended system rate. All user interactions
prehospital volume was eliminated from
attending physician. withthesystemweresavedtothecentralized
theanalysis,thetotalofcrystalloidspost-
● Patient has been on maintenance rate databaseforreviewanddocumentation.
ICUadmissionovertheinitial24hrswas
(125mL/hr)for6hrs(butnot(cid:2)24hrs).
stillreducedfrom14,973(cid:7)10,681mLto
● Patient has been stable at 48 hrs post Analysis Results
9679(cid:7)4776mL(p(cid:2).05).Additionally,
burn.
FrominitialdeploymentthroughJan- totalcrystalloidvolumesafterICUadmis-
● Patienthasbeentransferredtosurgery.
uary 2009, 32 patients were resuscitated sion over the entire resuscitation were
System exit rules are treated as rec- with the CDSS with at least 24 hrs of reducedfrom21,316(cid:7)12,974mLinthe
ommendations and are followed at the recommendations during the initial 48 control to 13,088 (cid:7) 5644 mL in the
discretionoftheprovider.Additionally,a hrs postburn. These were analyzed CDSS group (p (cid:2) .05).
2034 CritCareMed2011Vol.39,No.9
Ratio comparisons resulted in CDSS reduced mL/kg/TBSA from 6.5 (cid:7) 4.1 mL/ to 4.6 (cid:7) 2.5 mL/kg/TBSA over the entire
patientsalsohavinglowermL/kgandmL/ kg/TBSAto4.2(cid:7)1.8mL/kg/TBSAoverthe resuscitationperiod(p(cid:2).05).UOPvalues
kg/TBSA values during the resuscitation initial 24 hrs after burn (p (cid:2) .05). Ratios were compared at each hour during the
period(Fig.4,BandC).CDSSpatientshad werereducedfrom7.3(cid:7)5.6mL/kg/TBSA initial48hrstodeterminewhetherpatients
mettargetrangesof30to50mL/hrrates.
CDSS patients had higher rates of UOP
values within target than control. CDSS
patients achieved a percentage target in
UOPrangeanaverageof31%(cid:7)16%over
48 hrs compared to 23% (cid:7) 13% for the
controlgroup(p(cid:2).05)(Fig.3B).
Results from the hour-by-hour algo-
rithm performance resulted in higher
UOP rates in target (30 to 50 mL/hr)
when providers did not deviate (cid:1)100
mL/hr from the CDSS recommendations
compared with deviations (cid:1)100 mL/hr
during the initial 24 hrs post burn (p (cid:2)
.05)(Fig.5).AsdeviationsfromtheCDSS
recommendation increased, the fre-
quencytargetpercentageinUOPtarget
rangewasfurtherdecreased.UOPhour-
by-hour analysis of variance using Lev-
ene’s test for equality of variances also
showedasignificantdifferencebetween
the two groups, with the control group
showing more variability across the re-
Figure 3. A, Comparison of infusion rates between control and computer decision support systems suscitation time vs. the CDSS group
(CDSS)patients.B,Volumecomparisonofurineoutput(UOP)betweencontrolandCDSSpatients. (Fig. 6).
Figure 4. A, Volume comparison between the control and the computerized decision support system (CDSS) groups. B, Total resuscitation (mL/kg)
comparisonbetweencontrolandtheCDSS.C,Totalresuscitation(mL/kg/%totalbodysurfacearea[TBSA])comparisonbetweencontrolandCDSS.ICU,
intensivecareunit.
CritCareMed2011Vol.39,No.9 2035
example, showed in 1968 that successful
resuscitationcouldbeaccomplishedwith
aformulaconsistingof4mL/kgperTBSA
oflactatedRinger’ssolutionintheinitial
24 hrs postburn. Subsequently, Pruitt et
al (12) provided a modification to the
Brooke formula of 2 mL/kg per TBSA in
whichalbuminwaseliminatedduringthe
first24hrs.However,recentstudieshave
suggested that these formulas will not
accuratelydeterminetheamountoffluid
needed for resuscitation, and do not ac-
curatelyreflectthetotalvolumerequired
bythepatient(28).Similarly,theissueof
fluid creep has been documented in sev-
eral studies, showing that fluid volumes
havebeenincreasingconstantlyoverthe
lastseveralyears(29–31).However,over-
resuscitationhasthepotentialtoincrease
morbidity and life-threatening complica-
tions, such as abdominal compartment
syndrome. On the other hand, not pro-
Figure5.Urineoutput(UOP)targetsasafunctionofthedifferencebetweenrecommendationsofthe
viding the patient with sufficient fluid
computerizeddecisionsupportsystem(DSS)andactualinfusedfluidvolumes.
may also increase rates of serious prob-
lems by not addressing the inherent in-
travascular fluid deficit associated with
the burn injury. Therefore, effective re-
suscitationbecomesachallengewhenpa-
tients have to be kept to an appropriate
fluid balance regimen that will mitigate
possiblecomplicationsfromtoomuchor
too little fluid. For the clinician or care
provider, one of the main challenges for
appropriate fluid management may be
duetothecomplexityoftheresuscitation
guidelines themselves. Depending on
careproviderstocalculateandderivethe
Figure6.Comparisonofhour-by-hoururinaryoutput(UOP)variancebetweencontrolandcomput-
necessary fluid delivery rates for each
erizeddecisionsupportsystem(CDSS)patients.
hour while providing appropriate care to
major burn injuries may be problematic
in many instances. This issue becomes
increasinglycriticalifthepatientisbeing
treatedataregionalhospitalornonburn
facility.Furthermore,whencoupledwith
masscasualtysituations,theworkloadand
requirements for care of multiple large
burnsmayoverwhelmmanycenters.
One of the advantages of using our
CDSS is the ability to better model the
expected fluid response from burns such
that fluid titration is more accurate and
Figure7.A,Mortalitycomparisonbetweencontrolandcomputerizeddecisionsupportsystem(CDSS) effective throughout the resuscitation
patients.B,OutcomescomparisonbetweencontrolandCDSSpatients.ICU,intensivecareunit. phase compared to standard manual ap-
proaches that divide the injury response
TheCDSSgrouphadalowermortality .05) (Fig. 7B). IFDs were not different be- into two broad phases. Phase 1 is the
than the historical cohort (29% vs. 44%) tweengroups. initial8hrs,whenitisassumedthatthe
(p(cid:2).05)(Fig.7A).MeanventilatorandICU patient will require the most fluid, and
daysforthetwogroupswere13daysand30 DISCUSSION phase II is the next 16 hrs of the resus-
days, respectively. These values were used citation, where much less fluid is given.
to derive the VFDs and IFDs for compari- The need for fluid volume therapy for However, these approaches assume a
son. Mean VFDs were also higher in the burnshockhasbeenrecognizedformany constant physiologic response that is
CDSSgroup(6.5(cid:7)5.5vs.3.8(cid:7)5.2)(p(cid:2) years(26–27).BaxterandShires(18),for fixed on an 8-hr and 16-hr response pe-
2036 CritCareMed2011Vol.39,No.9
riod.Physiologically,thisresponseisnot acuteburnsintermsoffluidadministra- CDSS, patients had a significantly lower
likelytooccur.Ourexperienceusingour tion and hospitalization outcomes (e.g., mortality and increased VFDs and IFDs.
CDSS has shown that patients not only VFDs, mortality). When looking at all This study provides an example of patient
varywidelybutalsorequireacontinuous measures of fluid use, the CDSS signifi- healthcareimprovementusinginformation
increasing need for fluid that is maxi- cantlyreducesallfluidvolumesandpro- anddecisionsupporttechnology.
mized during the initial 13 hrs postburn videsbetterfluidmanagementduringthe
in addition to requiring fluid at a con- 48-hrresuscitationperiod.Thisispartic-
REFERENCES
stantly decreasing rate thereafter. Plot- ularly apparent during the initial 13 hrs
tingtheinfusedfluidduringtheinitial24 inwhichlarge-volumeshiftsarepossible
1. AhrensM:AnoverviewoftheU.S.fireprob-
hrsshowsacontinuumoffluidneedthat while trying to maintain adequate UOPs.
lem. Quincy, MA, National Fire Protection
can only be modeled as a complex set of Eventhoughthereareotherconfounders Association,2007
mathematical equations. Unfortunately, in this study, these results show that an 2. National Burn Repository 2005 Report.
implementingsuchasystemofcontinuous improvementinmortalitymaybeattrib- AmericanBurnAssociation,2006
fluidtitrationcannoteasilybedonemanu- utedtotheuseoftheCDSS.However,it 3. Karter M: Fire loss in the United States.
allyandrequirestheuseofcomputertech- can be argued that the use of the CDSS Quincy,MA,NationalFireProtectionAssoci-
nology capable of implementing such a may not be completely responsible for ation,2008
4. HerndonDN(Ed):TotalBurnCare.ThirdEdi-
complexresuscitationstrategy. this improvement.
tion.Philadelphia,PA,SaundersElsevier,2007
When comparing the rates of UOP Overall, when all variables are exam-
5. Oliver RI Jr: Burns, Resuscitation and
within target, the CDSS does achieve a ined,itcanbearguedthattheuseofthe
EarlyManagement.Availableat:http://www.
higher UOP on target (31%) than the CDSS provides an improvement in the
emedicine.com/plastic/topic159.htm. Ac-
control group (23%). However, even resuscitationofseverelyburnedpatients. cessedJuly28,2009
though this improvement is statistically It allows care providers to achieve better 6. Kramer GC, Lund T, Herndon DN: Patho-
significant,clinicallyitisstillamarginal resuscitation results and may provide a physiology of burn shock and burn edema.
increase in target percentages. Overall, means for allowing nonburn users to ef- In: Total Burn Care. Second Edition. Hern-
patientshaveaUOP(cid:1)50mL/hrinmany fectively resuscitate severely burned pa- don DN (Ed). Philadelphia, PA, Saunders
cases. One explanation for this low in- tients. When coupled with effective clin- Elsevier,2002,pp78–87
7. Azzopardi EA, McWilliams B, Iyer S, et al:
crease can be seen in the difference be- ical practice guidelines within our burn
Fluid resuscitation in adults with severe
tweentherecommendationsbytheCDSS center, the CDSS provides an effective
burns at risk of secondary abdominal com-
andtheactualinfusionratesgivenbythe adjuncttoburncarethatresultedinim-
partment syndrome–an evidence based sys-
healthcare providers. On average, users proved outcomes. The use of decision
tematicreview.Burns2009;35:911–920
are infusing more crystalloid than the support technology can effectively assist 8. BarrowRE,JeschkeMG,HerndonDN:Early
system recommended. In accordance inthecareofburnswithoutsupplanting fluidresuscitationimprovesoutcomesinse-
withimplementationofthissystem,fluid the expertise available at the bedside. verely burned children. Resuscitation 2000;
rates are strictly provided as a recom- 45:91–96
mendation; it is expected that the pro- CONCLUSIONS 9. WolfSE,RoseJK,DesaiMH,etal:Mortality
vider use clinical judgment when appro- determinantsinmassivepediatricburns.An
analysis of 103 children with (cid:1) or (cid:3) 80%
priate and provide a documented reason Currentresuscitationapproachesbased
TBSAburns((cid:1)or(cid:3)70%full-thickness).Ann
for not following the system recommen- on standard formulas are typically only
Surg1997;225:554–565;discussion565–569
dation. To better understand the actual used as guidelines and lead to significant
10. KirkpatrickAW,BallCG,NickersonD,etal:
performanceofthealgorithm,weneedto inconsistencies and variability between
Intraabdominal hypertension and the ab-
comparetheinstanceswheretheprovid- healthcare providers and patients. Using a dominalcompartmentsyndromeinburnpa-
ersacceptedthesystemrecommendation mathematicalapproachtodeveloparesus- tients.WorldJSurg2009;33:1142–1149
vs. using their own judgment. In this citation-response model, we developed an 11. Ivy ME, Atweh NA, Palmer J, et al: Intra-
case, UOP rates within target were up to algorithmtoprovideuserswithcrystalloid abdominal hypertension and abdominal
20% higher when providers followed the infusion recommendations during the ini- compartment syndrome in burn patients.
system recommendations. Furthermore, tial48hrsofburnresuscitation.TheCDSS JTrauma2000;49:387–391
12. Pruitt BA Jr: Fluid and electrolyte replace-
theseresultsrevealthatasthedifferences was implemented as the standard clinical
mentintheburnedpatient.SurgClinNorth
between recommendations and actual practiceinourunitandcomparedagainst
Am1978;58:1291–1312
rates diverge, percentages of UOP in tar- historical controls to assess for any im-
13. BrandstrupB,TønnesenH,Beier-Holgersen
get will tend to also decrease. Based on provements in the resuscitation manage-
R,etal:Effectsofintravenousfluidrestric-
these results, we can conclude that the ment of burn patients. This study showed tion on postoperative complications: Com-
computerizedalgorithmwillincreaselev- that the CDSS was able to maintain pa- parisonoftwoperioperativefluidregimens:
els of compliance in burn injury and re- tientsonUOPtargetalargerpercentageof A randomized assessor-blinded multicenter
sult in better care for patients, and also thetime,whileachievinglowerfluidinfu- trial.AnnSurg2003;238:641–648
may explain the increased fluid adminis- sionscomparedtoourcontrolcohort.Pa- 14. TricklebankS:Modern trendsinfluidther-
tration that led to many of the UOP val- tientsontheCDSSsystemhadstatistically apyforburns.Burns2009;35:757–767
15. ArieffAI:Fatalpostoperativepulmonaryede-
uesmissingtherecommendedtargetrate lowermL/kgandmL/kg/TBSAvaluesthan
ma: Pathogenesis and literature review.
upper limit of 50 mL/hr. controls, as well as reduced overall fluid
Chest1999;115:1371–1377
Furthermore, results from this analy- requirements at both 24 and 48 hrs post-
16. AdvancedBurnLifeSupportCourse(ABLS):
sis show promise that the use of the burn. Additionally, the CDSS achieved the
Instructor’s Manual. Chicago, IL, American
CDSS may provide an improvement in target UOP a greater percentage of the BurnAssociation,2001
overall resuscitation management of time.Finally,resultssuggestthatusingour 17. Warden GD: Fluid resuscitation and early
CritCareMed2011Vol.39,No.9 2037
management. In: Total Burn Care. Second 22. SalinasJ,DrewG,GallagherJ,etal:Closed- 27. Reiss E, Stirman J, Artz C, et al: Fluid and
Edition.HerndonDN(Ed).Philadelphia,PA, loopanddecision-assistresuscitationofburn electrolyte balance in burns. JAMA 1953;
SaundersElsevier,2002,pp88–97 patients.JTrauma2008;64:S321–S332 152:1309–1313
18. Baxter CR, Shires T: Physiological response 23. ClinicalDecisionSupportSystems.Availableat: 28. Cartotto RC, Innes M, Musgrave MA, et al:
to crystalloid resuscitation of severe burns. http://en.wikipedia.org/wiki/Clinical_decision_ HowwelldoestheParklandformulaestimate
AnnNYAcadSci1968;150:874–894 support_system.AccessedJuly28,2009 actual fluid resuscitation volumes? J Burn
19. EngravLH,ColescottPL,KemalyanN,etal: 24. CoieraE:GuidetoHealthInformatics.Sec- CareRehabil2002;23:258–265
AbiopsyoftheuseoftheBaxterformulato ond Edition. London, Oxford University 29. FriedrichJB,SullivanSR,EngravLH,etal:Is
resuscitateburnsordowedoitlikeCharlie Press,2003 supra-Baxter resuscitation in burn patients a
didit?JBurnCareRehabil2000;21:91–95 25. ChungKK,SalinasJ,RenzEM,etal:Simple newphenomenon?Burns2004;30:464–466
20. Kemalyan N, Dimick PL, Heimbach DM, et derivation of the initial fluid rate for the 30. SaffleJI:Thephenomenonof“fluidcreep”in
al:TheBaxterFormula1995.ProcAmBurn resuscitation of severely burned adult com- acute burn resuscitation. J Burn Care Res
Assn1996;28:72 batcasualties:Insilicovalidationoftherule 2007;28:382–395
21. Shah A, Kramer GC, Grady JJ, et al: Meta- of10.JTrauma2010;69:S49–S54 31. WanekSM,WolfSE,SalinasJ,etal:Modern
analysisoffluidrequirementsforburninjury 26. Evans EI, Purnell OJ, Robinett RW, et al: resuscitationpracticeinsevereburns:Over-
1980–2002. Abstr. J Burn Care Rehabil 24: Fluidandelectrolyterequirementsinsevere resuscitation and fluid creep. J Burn Care
S118 burns.AnnSurg1952;135:804–817 Rehabil2006;27:S129
2038 CritCareMed2011Vol.39,No.9