Tight glycaemic control: clinical implementation of protocols

12 Pages·2009·2.27 MB·English

BestPractice&ResearchClinicalAnaesthesiology23(2009)461–472
ContentslistsavailableatScienceDirect
Best Practice & Research Clinical
Anaesthesiology
journal homepage: www.elsevier.com/locate/bean
9
Tight glycaemic control: clinical implementation
of protocols
Frank Nobels, MD, PhDa,*, Patrick Lecomte, MDb, Natascha Deprez, MSc,
APRNa, Inge Van Pottelbergh, MD, PhDa, Paul Van Crombrugge, MDa,
Luc Foubert, MD, PhDb
aDepartmentofEndocrinology,Onze-Lieve-VrouwHospital,Moorselbaan164,9300Aalst,Belgium
bDepartmentofAnesthesiologyandCriticalCareMedicine,Onze-Lieve-VrouwHospital,Moorselbaan,Aalst,Belgium
Implementationoftightglycaemiccontrolinhospitalisedpatients
Keywords:
presents a huge challenge. It concerns many patients, there are
blood-glucosecontrol
algorithms many interfering factors and many health-care professionals are
protocols involved.Thecurrentliteratureprovideslittlepracticalguidance.
tightglycaemiccontrol This article offers the clinical anesthesiologist direction for the
intensive-insulintherapy organisationofinpatientbloodglucosecontrolinacutesituations,
criticallyillpatients in the perioperative setting and in the intensive care unit. An
effective,safeanduser-friendlyalgorithmforintravenousinsulin
administrationispresentedthatcanbeexecutedbyregularnurses
andusedinmanysituations.Practicaladviceisofferedfortheuse
of subcutaneous basal–bolus insulin, for fasting orders and for
transition to discharge care. The main safety considerations are
discussed.
(cid:2)2009PublishedbyElsevierLtd.
Many hospitals seek to institute intensive glucose control, inspired by the results of the Leuven
trials and by observational data showing a relation between elevated blood glucose levels during
hospitalisationandadverseclinicaloutcomes.1–3Theyareconfrontedwithgreatdifficulties,however,
duetoahighnumberofpatients,manyinterferingfactorsandscarcityofpracticalguidanceinthe
availableliterature.Inacutesettings,hyperglycaemiaoccursveryfrequently,mainlyduetoelevated
concentrationsofstresshormones.Withaconservativebloodglucosetargetof(cid:2)180mgdl(cid:3)1nearly
three-fourths, and with an intensive target of (cid:2)110mgdl(cid:3)1 nearly all critically ill patients require
exogenousinsulin.1,2,4Manypublishedintravenous(IV)insulinprotocolsusecomplexalgorithmsthat
* Correspondingauthor.Tel.:þ3253724488;Fax:þ3253724187.
E-mailaddress:[email protected](F.Nobels).
1521-6896/$–seefrontmatter(cid:2)2009PublishedbyElsevierLtd.
doi:10.1016/j.bpa.2009.09.001
462 F.Nobelsetal./BestPractice&ResearchClinicalAnaesthesiology23(2009)461–472
are neither safe nor simple enough for routine use. The blood glucose targets are frequently not
reached because of poor algorithm performance and the incidence of severe hypoglycaemia
((cid:2)40mgdl(cid:3)1)ishigh.1,2,4–6Thequalityofpointofcare(POC)bloodglucosemeasurements,neededto
feedthealgorithms,isoftenweak.7,8Finally,thetransitiontosubcutaneous(sc)insulinisoftendiffi-
cult,duetolackofexpertiseinusingbasal–bolusinsulinregimens.9
Weareconvincedthateffectiveandsafeglycaemiccontrolisfeasible,however,bothinacuteand
perioperativesituationsandintheintensivecareunit(ICU),withsmoothtransitiontopost-acutecare.
Thisarticleoffersguidancefortheclinicalanaesthesiologistinchargeofthesepatients.First,essential
‘buildingblocks’forglycaemiccontrolarepresented.Thensafetyconsiderationsarediscussed.Finally,
perioperativecareisusedasapracticalexampleofintegrationofthedifferentelementsinaworkable
system.
Buildingblocks
For effective and safe in-hospital blood glucose regulation, a systematic approach is needed. A
clinicalpathshouldbedevelopedwithprotocolarguidanceforeverysituationthatcanbeencountered
duringhospitalisation.Keystakeholdersshouldbeidentified,workinggroupsappointed,protocolsand
algorithms created and educational programmes developed. The protocols should be as simple as
possible, taking into account staffing requirements and safety. It is recommended to use as few
‘buildingblocks’aspossiblethatcanbeusedindifferentsituations.Thisfacilitateseducationofthe
nursing and medical staff and allows experience gained in a certain situation to be used in other
protocols.Ourhospitalmanagementconsistsofthefollowingprincipalbuildingblocks:
1. AdynamicIVinsulininfusionalgorithmthatcanbeusedfordifferentbloodglucosetargets,
2. Ascbasal–bolusinsulinschemeforpatientsrecoveringfromanacutesituation,
3. Atransitionschemetomakethelinkfromabasal–bolusinsulinschemetoatreatmentplanfor
dischargeand
4. Aprotocolforbloodglucoseregulationwhenfastingforaninvestigationortreatment.
IVinsulininfusionalgorithm
An essential condition for obtaining good glycaemic control in acute situations, during major
surgeryandincriticalillness,istheavailabilityandcreativeuseofagoodIVinsulin-infusionprotocol.It
should be effective, safe and simple enough to be used throughout the hospital by regular nurses,
keepingtheneedforexpertsupervisiontoastrictminimum.Mostofthepublishedalgorithmsdonot
meettheserequirements.Hypoglycaemiaisaparticularconcern.IntheLeuvenstudies,theincidence
of severe hypoglycaemia (defined as a blood glucose level (cid:2)40mgdl(cid:3)1) in the intensively treated
patientswas5.2%inthesurgicaland18.7%inthemedicalICU.1,2TheGlucontrolandVISEPstudieswere
stoppedearlyduetotheincidenceofhypoglycaemia,at9.8%and17.0%,respectively,intheirtightly
controlled groups.5,6 In the Normoglycemia in intensive care evaluation and survival using glucose
algorithm regulation (NICE-SUGAR) study, theincidencewaslower, butstill toohighat 6.8% inthe
intensivegroup.4
Duringthepastyears,betterprotocolsforIVinsulinadministrationwerepublished,basedonthe
principle,initiallypublishedbyMarkovitzetal.,thattheinsulininfusionrateisgraduallyadaptedtothe
individualinsulinsensitivityofthepatient.10–12Davidsonetal.createdacomputeralgorithminspired
on this principle, using the formula: insulin dose/h¼(blood glucose in mgdl(cid:3)1–60)(cid:4)multiplier.12
Whenthebloodglucoselevelisnotdecreasingfastenough,theinsulinadministrationcanbemade
moreaggressivebyincreasingthemultiplierandviceversa.Thiscanbeexpressedvisuallyinagrid
(Fig.1),withrowsrepresentingbloodglucoserangesandcolumnsinsulindoses,movingfromleftto
righttomoreaggressiveadministration(highermultiplier).Theinsulininfusionisusuallystartedin
column 2 (multiplier 0.02) at a rate corresponding to the current blood glucose. When the blood
glucosehasdecreasedsignificantly((cid:5)1range)atthenextmeasurement,thesamecolumnisusedfor
theinsulinadministration,andthedoseisdecreasedaccordingtothenewbloodglucoselevel.When
F.Nobelsetal./BestPractice&ResearchClinicalAnaesthesiology23(2009)461–472 463
Fig.1. ProtocolforcontinuousivinsulinadministrationforBGtarget80–110mg/dl(O.L.V.hospitalAalst,Belgium,2009).
thebloodglucosefailstodecrease,theinsulindosageisincreasedbymovingonecolumntotheright
(increasing the multiplier with 0.01). By repeating this, a column corresponding to the individual
insulin sensitivityof the patientis graduallyreached. Oncetheblood glucose target (greenzone in
Fig.1)hasbeenreached,theinsulinadministrationstaysinthesamecolumn,but,unlikeinmostother
protocols,smalladaptationscanstillbemade.Ifthebloodglucosefallsbelowthetarget(orangezonein
Fig.1), theaggressivenessof the insulinalgorithm is decreasedtoavoidhypoglycaemia,bymoving
acolumntotheleft.
Davidsonetal.publisheddataof5080IVinsulinrunswiththisprotocol.12Theyachievedamean
glucoselevel<150mgdl(cid:3)1in3h,thatremainedstableforaslongastheruncontinued(mean24h).
Theprevalenceofseverehypoglycaemia(<40mgdl(cid:3))was2.6%.
Wehaveadaptedthisprotocolinourinstitutiontomakeitevenmoreefficientandsafe.13Thegoal
of our adaptations was to proactively react to rapidly changing insulin needs in cardiac surgical
patients,bothduringtheoperationaslaterintheICU.Duringcardiacsurgerywithcardiopulmonary
bypass, we proactively increase the insulin dosage by moving three columns to the right during
re-warming,becausethisinducesasudden,transientincreaseininsulinresistance.Wereturnthree
columnstotheleftwhentheoesophagealtemperaturereaches(cid:5)36(cid:6)C.Wemoveafurthercolumnto
theleftattheendoftheoperationtoanticipateadecreaseininsulinrequirementswhensurgicalstress
fades.Similarly,weanticipatetheeffectofstressinducedbystoppingthesedationforextubationby
movinginsulinonecolumntotheright.Intheoriginalprotocol,onecanonlyshifttothelefttoaless
aggressiveinsulin therapy,whenthebloodglucoselevelisalready undertarget(intheorangerisk
zone).Wehaveincreasedtheupperlimitoftheorangezoneto85mgdl(cid:3)1toanticipatehypoglycaemia.
464 F.Nobelsetal./BestPractice&ResearchClinicalAnaesthesiology23(2009)461–472
In745patientsundergoingcardiacsurgerywithcardiopulmonarybypass,bloodglucoseremained
(cid:2)130mgdl(cid:3)1 during surgery and in the first 24h in the ICU in, respectively, 92% and 95% of the
measurements in non-diabetic patients and 84% and 91% in diabetic patients.14 All blood glucoses
remainedabove40mgdl(cid:3)1.
Recently,wecreatedacomputerisedexpertsystemincludingtheseadaptations,allowingnursesto
safelyusethisIVinsulinprotocol.Weincludedanticipationtosuddenchangesincarbohydratesupply,
andalsousedthecomputersystemtocalculatetheanticipatedscinsulindosagewhenthepatientis
transferredfromICUtotheward.Thecomputerexpertsystemalsoallowsustostudychangesinthe
protocol without increasing the workload for the nurses. We are currently investigating whether
movingtotheleftwhenthebloodglucosedecreasestoofastbetweentwosuccessivemeasurements,
evenwhenitisstillabovetarget,betteravoidshypoglycaemiawithoutjeopardisingtheeffectivenessof
theprotocol.
Thealgorithmisdesignedforglucosemeasurementsevery60min,butmeasurementsevery30min
maybenecessarywhentheinsulinsensitivityischangingrapidly.Wheninastablepatienttheblood
glucoseremainsintargetatfoursuccessivemeasurements,checkingcanbedecreasedtoevery120min.
This protocol is not only effective and safe, but also very user-friendly. Since it automatically
searchesthemosteffectiveinsulindosage,itcanoperatewithallinfusionfluidsandfluidadminis-
trationrates.Thesameprotocolcanbeusedfordifferentbloodglucosetargetsbychangingtheupper
and/orlowertargetlevels(greenzoneinFig.1).Itcaneasilybetaughttonursesofallunits.Allthese
advantagesallowitsuseindifferentsituationsand/orhospitalunits.
Scbasal–bolusinsulinscheme
Whenthepatientstartstoeat,theIVinsulinprotocolwillinduceup-titratingoftheinsulindose,
increasingtheriskofhypoglycaemiaafterafewhours.Thiscouldbeavoidedbygivinga2-hsquare
bolusofIVinsulin,butthiswouldmakethetreatmentmorecomplex.Itiseasiertoconverttoabasal–
bolus scheme with sc administration of short-acting insulin before the meals and intermediate- or
long-actinginsulin,usuallyatbedtime.3,15Inpatientswhosemedicalstressstillinduceshighinsulin
requirements,theIVinsulininfusioncanbecontinuedtoprovidebasalinsulin,withscshort-acting
insulincoveringthemeals.
Althoughrapid-andlong-actinginsulinanaloguestheoreticallyofferamorephysiologicprofile16,it
iseasiertoworkwithstandardregularandNPHinsulinsintheICUsetting.Criticallyillpatientshave
slower gastric emptying, start out by eating poorly and usually receive between-meal nutritional
supplements.17Thiscanbetterbecoveredwithregularinsulinthanwithrapid-actinganalogues.Their
insulin needs usuallychange rapidlyduring the first days, due tofadingofmedical stress, tapering
medication with hyperglycaemic effect (inotropics and corticosteroids) and fasting for technical
investigations.Inthesecircumstances,theultra-longaction(20–24h)oftheanaloguesglargineand
levemirisdisadvantageousincomparisonwiththeshorteraction(12–18h)ofNPHinsulin.Regular
andNPHarealsomoreconvenientfortransitioningfromIVtosctherapy.Theactionofregularinsulin
islongenoughtobeabletojumptothenextmeal.Thisisnotthecasewithrapid-actinganalogues,
necessitatingsimultaneousinjectionofbasalinsulin,usuallyatatimeofthedaywhenthebasalinsulin
wouldnormallynotyetbeinjected.
Sinceregularinsulincoversapproximatelyone-fourthoftheday,aneasyruleofthumbtodeter-
minethestartingdoseistosumuptheinsulinadministeredduringthelast6handadd20%forthe
prandialrequirements.Forexample,whentheinsulinpumpprovidedameandoseof2Uh(cid:3)1during
thepast6h,(6(cid:4)2)þ20%thatis,approximately14Uofregularinsulincanbegivenbeforethefirst
meal.Thisshouldbeconsideredasatestdose,helpingtodeterminethenextdosesonthebasisofits
effect.When duringthe pasthoursbloodglucose levels werewell controlled with (cid:2)0.5units ofIV
insulinperhour,insulincanusuallybestopped,exceptintype1diabetes.Stoppinginsulinintype1
diabeteswillrapidlycauseketosis.Whenindoubtaboutthetypeofdiabetes,itiswisetocontinue
insulintoavoidproblems.
Regularinsulinhasadelaytoonsetofactionof30min,requiringscinjectionatleast30minbefore
stoppingtheIVinsulinpump.15Itshouldalsobeinjected30minpriortomeals,agoalthatmaybe
F.Nobelsetal./BestPractice&ResearchClinicalAnaesthesiology23(2009)461–472 465
difficulttomeetinabusyunit.SincearegimenwithregularandNPHinsulininduceshighinsulinlevels
intheearlynight,abedtimesnackisneededtopreventnocturnalhypoglycaemia.
Bedsideglucosemonitoringshouldbeperformedbeforemealsandatbedtime.Itcanalsobeuseful
to measure between 2:00 and 3:00 a.m. to assess for nocturnal hypoglycaemia, particularly if the
patientjustconvertedtoscinsulin,orifacorrectiondosewasusedatbedtime.
Alldataconcerningbloodglucoselevelsandinsulinadministrationshouldbenotedonasummary
chartorcomputerfile(exampleinFig.2).Goodcommunicationisveryimportant,becausetheblood
glucose treatment of these patients will be executed by different caregivers. On the basis of the
patient’sresponsetopriorinsulindoses,theresponsiblephysician(ordiabetesnurse)candetermine
the next doses. The nurse who administers the insulin should adapt the scheduled dose with
acorrectionalgorithm.Thisshouldtakeintoaccounttheinsulinsensitivityofthepatientbyproviding
largercorrections at higher insulin doses. Atour institution, we use the algorithm shown in Fig. 2,
basedontheformuladevelopedbyDavidson:(actualbloodglucose–targetbloodglucose)/correction
factor.18,19Thecorrectionfactorcorrespondstohowmuchthebloodglucoselevelisloweredby1unit
ofshort-actinginsulin.Thisisusuallyestimatedwiththeformula1700/totaldailyinsulindose.Weuse
asomewhatmoreconservativeestimationof2200/totaldailyinsulindosetoavoidhypoglycaemia.At
bedtime,weuseevenmoreconservativecorrectiondosestoavoidnocturnalhypoglycaemia.
Itshouldbestressedthatthereisnoplacefortraditionalsliding-scaleinsulinregimens,thatprovide
afixeddoseofshort-actinginsulinadministeredforacertainlevelofhyperglycaemia,withoutbasal
insulinandwithoutanyindividualisation.This‘reactive’approachtreatshyperglycaemiaafterithas
already occurred, instead of preventing it. It results in a saw-toothcurve, exacerbating both hyper-
glycaemiaandhypoglycaemia.20,21
Transitionscheme
Beforethepatientisdischarged,theinsulinregimenmayneedtobesimplified,oralanti-diabetics
(re-)introduced or glycaemic treatment tapered off and stopped. One should proceed with this
Fig.2. Glucosesummarychartforpatientsonscinsulinand/ororalantidiabeticagents(O.L.V.hospitalAalst,Belgium,2009).
466 F.Nobelsetal./BestPractice&ResearchClinicalAnaesthesiology23(2009)461–472
planningasearlyaspossible,whenthepatienteatsdiscretemeals.Waitingtoolongcancauselonger
hospitalstayordischargewithaninappropriatetreatmentregimen.22,23 Thetimelyinvolvementof
a hospital diabetes nurse educator can facilitate this transition and reduce length of stay in the
hospital.22,23 Discharge planning demands assessment of the patient’s history of diabetes, previous
treatmentandmetaboliccontrol(HbA1c),emergenceofcontraindicationsfororalanti-diabeticagents,
adaptationsoftreatmentgoalsrelatedtoprognosisandneedforextraeducation.Everymemberofthe
treatment team – physicians, nurses and dieticians – should contribute information to aid the
responsiblephysicianinhisdecision-makingprocess.Thedesignofthepatientglucosesummarychart
orcomputerfile(exampleinFig.2)shouldfacilitatethisinformationgatheringandprovidehelpfor
structuringthedecisionprocess.
Atdischarge,oneshouldanticipatethatbloodglucoselevelscandecreaseathomewhenphysical
activityincreases,medicationwithhyperglycaemiceffect(e.g.,corticosteroids)istaperedandmedical
stress abates. A recent hospitalisation is a strong predictor of subsequent serious outpatient hypo-
glycaemia.24Thisshouldleadtocautioninthedosingofglycaemictherapyatdischargeandtocareful
communicationwithgeneralpractitioners.
Fastingprotocol
Therearenofixedguidelinestoadaptscinsulinororalanti-diabeticdrugswhenapatientskips
mealsforaninvestigationortreatment.Onemustresorttosomesimplerulesofthumbandcommon
sense.3,15Inourinstitutionweusethefollowinggeneralrules:
- Holdoralanti-diabeticdrugswhenfasting.
- Inpatientsonaregimenincludingalong-actinginsulinanalogue:Continuethisanalogue,andgive
theusualdoseofshort-actinginsulinbeforeameal.
- Inpatientsonotherinsulinregimens:Providebasalinsulinusinganintermediate-actinginsulin,
such as NPH. When breakfast is omitted, give half of the breakfast plus lunch insulin dose as
asinglemorninginjectionofNPH(e.g.,when18Uofa30/70premixedinsulinwouldhavebeen
givenbeforebreakfast,replace it by9UofNPH). Providesome extraregular insulin beforethe
lunch.Whenthepatientreceivesbreakfastbutskipslunch,givetwo-thirdsofthebreakfastplus
lunchinsulindoseastwo-thirdsregularandone-thirdsNPHinsulinbeforebreakfast(e.g.,when
18Uofa30/70premixedinsulinwouldhavebeengivenbeforebreakfast,replaceitby8Uregular
and4UNPHbeforebreakfast).
- Adapt the dose for hypo- and hyperglycaemia using an algorithm that takes into account the
insulinsensitivityofthepatientbyprovidinglargercorrectionsathigherinsulindoses.
- Inpatientsreceivingabolusdoseofcorticosteroids(e.g.,aspreventionofcontrastreaction)use
higherinsulindosesandamoreaggressivecorrectionalgorithm.
Sincetheserulesarerathercomplex,wecasttheminfourvisualdiagramstogivethenursesmore
guidance.Weusetwodiagramsforpatientsonaregimenincludingalong-actinginsulinanalogue,one
forskippingbreakfastandoneforskippinglunch,andsimilarly,twodiagramsforotherinsulinregi-
mens(seeexampleinFig.3).
Safetyconsiderations
Any attempt at better in-hospital glycaemic control should focus on prevention, immediate
recognition and appropriate treatmentofhypoglycaemia. Hypoglycaemia is a majorsafetyconcern,
especiallyinpatientswithimpairedmentalstatus.Ontheotherhand,fearofhypoglycaemiaisamajor
barriertoachievetightcontrol.9,25InthetrialsoftightglycaemiccontrolintheICU,highratesofmajor
hypoglycaemia((cid:2)40mgdl(cid:3)1)werereported.1,2,4–6Theensuingdiscussionwasmainlyfocussedonthe
safety of the strict blood glucose target of 80–110mgdl(cid:3)1. Although hypoglycaemia occurs less
frequently with less strict targets of for example,140–180mgdl(cid:3)1, this offers no guarantee. More
importantthanthechoiceofthebloodglucosetargetistheorganisationofthebloodglucosecontrol,
withattentiontothefollowingessentialcomponents26–28:
F.Nobelsetal./BestPractice&ResearchClinicalAnaesthesiology23(2009)461–472 467
Fig.3. Fastingprotocolforskippingbreakfastforpatientsonaninsulinregimennotincludingglargineordetemir(O.L.V.hospital
Aalst,Belgium,2009).(exampleused:patienton24Uofa30/70insulinmixtureand1tabletofgliclazidebeforebreakfast,receiving
abolusdoseofcorticosteroidsforpreventionofcontrastreaction).
1. Accuratebedsidemeasurementofbloodglucose,
2. Thechoiceofaneffectiveandsafesystemofinsulinadministration,
3. Matchingofthetherapytocarbohydratedeliverance,
4. Earlydetectionandtreatmentofhypoglycaemiaand
5. Experthandlingoferrors.
With a policy that pays attention to these elements, severe hypoglycaemia can be drastically
reduced.
Accuratebedsidemeasurementofbloodglucose
Inadditiontothesourceofthebloodsample,thechoiceofthemeterandtheexpertiseoftheperson
whoperformsthemeasurementareimportant.7,8Incircumstancesoftightglycaemiccontrol,usingIV
insulininfusion,measurementsonbloodobtainedthroughanarterialcatheterarepreferable.Capillary
samplingislessreliableinseverelyillpatientswithperipheralvasoconstrictionoroedema.Inthecase
of venous sampling, admixture of dextrose from an infusion can give erroneous results. In the ICU
setting, measurement with an arterial blood gas analyser is preferred, as its reliability closely
approximates that of a central laboratory, and it also provides potassium levels, facilitating the
prevention of hypokalaemia induced by glucose–insulin administration.29 Classical glucose meters,
similartodevicesforhomeself-monitoringofbloodglucose,arelesspreciseandmoreinfluencedby
haematocritandoxygentension.7,8TheyarewellsuitedforuseoutsidetheICUandequivalentsettings,
providedthatqualityassuranceisorganised.30
Thechoiceofaneffectiveandsafesystemofinsulinadministration
Theabove-explainedcharacteristicsofeffectiveIVorscinsulinadministrationareimportantnot
onlytoreachthebloodglucosetargetsbutalsotominimisetheriskofhypoglycaemia.Thisincludes
theproactiveuseofagoodIVinsulininfusionprotocol,theuseofscheduledscinsulinwithprovisionof
basalcoverageinsteadofaslidingscaleandtheuseofalgorithmsforadaptationofIVandscinsulin
468 F.Nobelsetal./BestPractice&ResearchClinicalAnaesthesiology23(2009)461–472
thattakeintoaccounttheinsulinsensitivityofthepatient.Thephysicianwhoschedulesthedosesofsc
insulinandoralanti-diabeticagentsshouldtakecaretoproactivelydiminishthemincaseofrecovery
ofanacutemedicalsituation,deteriorationofkidneyfunctionortaperingofcorticosteroids.Thiscan
befacilitatedbyincludinginformationonkidneyfunctionanduseofcorticosteroidsonthepatient
glucosesummarychart(Fig.2).
The way inwhich insulin is administered IV is also important. When using an IV pump, insulin
shouldbeadministeredthroughacentralvenousline,usinganaccuratesyringe-driveninfusionpump,
avoidingthevariabilityinducedbyperipheralvenousinfusionandvolumetricpumps.Inourinstitu-
tion,weonlyuseIVpumpsinunitswithintensivenursingsupervision,thatis,ICU,medium-careunits,
operating theatre and recovery unit. Outside these units, we give insulin only through a graduated
burette,connectedbetweentheinfusionandthepatient.Thisisfilledeveryhourwith100ccglucose
andtheamountofinsulinthatisneededforthenexthour.Thisprovidesthesafetythattheinsulinis
always administeredtogether with the glucose, contrary toa separateinsulinpumpthat continues
givinginsulinwhentheglucoseadministrationishampered.
Matchingofthetherapytocarbohydratedelivery
Failuretoadjustanti-hyperglycaemicmedicationappropriatelyforsuddenlossofcaloricexposure
is a major cause of iatrogenic hypoglycaemia.27,28 Nurses should receive repeated education on the
synchronisationofnutritionandbloodglucoseregulation.WithIVinsulin,therateshouldimmediately
be reduced when enteral or parenteral nutrition is interrupted. In the ICU, we use a computerised
expert system that requires input of information on nutritional intake before an IV insulin rate is
suggested. With sc insulin and/or oral anti-diabetic agents, nausea and emesis should immediately
launchmorefrequentmonitoringofglucoseandappropriateadaptationofthetreatment.Nutrition
should be included as a parameteron the patient glucose summarychart, reminding the nurses to
considerthenutritionalintakeeachtimetheyadministerinsulinororalanti-diabeticagents.
Earlydetectionandtreatmentofhypoglycaemia
Earlydetectionrequiresincreasedalertnessinthepresenceofriskfactorsforhypoglycaemia.Major
risk factors are decreased carbohydrate intake, reduction of corticosteroids and prior hypo-
glycaemia.27,28Aprevioushypoglycaemiagreatlyincreasesthechanceofanewone.28Inthesesitu-
ationsmorefrequentbloodglucosemonitoringisneeded.
Clearguidelinesshouldbeprovidedonhowtotreatahypoglycaemicevent.Adherencetothese
guidelinesshouldbepromotedregularlyandmonitored.Asinmostpatientswithdiabetes,nursesalso
tend to overcorrect hypoglycaemia, giving excess carbohydrates and withholding insulin. The
correction of the ensuing hyperglycaemia can again induce hypoglycaemia, leading to a saw-tooth
curve.
Experthandlingoferrors
Errorsintheadministrationofinsulinororalanti-diabeticdrugsshouldimmediatelybereportedto
the supervising physician or diabetes nurse, who should take on a non-reprimanding, positive and
supportingattitude.Forsomefrequentlyoccurringerrorsguidancecanbeofferedintheprotocols.For
example,patientswhomustpresentfastedforsmallsurgeryoranexaminationoftenmistakenlytake
theirdiabetesmedicationsathomebeforepresentinginthehospital.Recommendationsonhowthis
canbecompensatedwithadditionalIVglucoseinfusioncanbeincludedintheprotocols.
Systemdesign
Bycreativelycombiningtheabove-presentedbuildingblocks,andtakingintoaccountthesafety
considerations, any situation that a patient with diabetes or transient hyperglycaemia can be con-
frontedwithduringhospitalisation,canbehandled.Theuseofthesameelementsindifferentsitua-
tions,facilitatestheeducationoftheusersandminimisestheriskoferrors.Allwelltrainednursescan
F.Nobelsetal./BestPractice&ResearchClinicalAnaesthesiology23(2009)461–472 469
runtheseprotocols,reducingtheneedforexpertsupervision.Weillustratethiswiththeperioperative
bloodglucosepolicyusedinourinstitution.
Surgeryoffersacomplexsituationwithmanyvariables.31–33Thepatientcanpresentwithstress
hyperglycaemia,unknownorknowndiabetes,eitherwellorbadlycontrolled.Thepreoperativeblood
glucose treatment can consist of many different oral anti-diabetic drugs and/or insulin types and
schemes.Onthedayofsurgery,thepatientmayneedtoremainfastingfromthemorningonormaybe
allowedtotakebreakfast.Eatingcanberesumedimmediatelyaftertheoperationorbepostponedfor
several days. The surgerycanvary froma minimal procedure, only requiring a few hours hospital-
isation,tomajorsurgeryrequiringtransfertoICU.The effectonthebloodglucose control canvary
considerably, depending on the endogenous insulin reserve of the patient, on the ‘stress response’
induced by the procedure and on the use of medications with hyperglycaemic effect.31 For some
operations, verystrict glycaemic control is desirable14,34,35, whereas forothers a more conservative
approach can be justified.32 Many services are involved, some with standard, some with intensive
staffing.Asystematicapproach,withasmuchsimplificationaspossible,isessentialtogetagripon
suchacomplexsituation.
Westartthedayofsurgerywithourfastingprotocol.Thisallowsusto‘jump’tothesurgeryusingsc
insulin,evenwhenthesurgicalprocedurestartslateintheafternoon.Itisasimpleschemewithminor
workload. Blood glucose is only measured every 4–6h with a classical blood glucose meter on
acapillarybloodsample.IfaswitchtoIVinsulinisneeded,theinsulinthatisalreadyonboardfromthe
fastingprotocolposesnoproblem,sinceourIVinsulinprotocolautomaticallyadaptstothissituation,
searchingthemostsuitablecolumnforinsulinadministration.Ourfastingprotocolallowscorrectionof
hyperglycaemia with sc insulin up to a value of 350mgdl(cid:3)1. For blood glucose levels above
350mgdl(cid:3)1,weswitchtoIVinsulinonthesurgicalward,safelyadministeredwithagraduatedburette.
Intheoperatingtheatrethepolicydependsonthebloodglucosegoal.Inhigh-risksurgeryweaim
atbloodglucoselevelsbetween85and110mgdl(cid:3)1.Wedefinehigh-risksurgeryasanyintervention
that routinely leads to postoperative transfer to ICU (e.g., cardiac surgery, brain surgery and major
gastrointestinal surgery). In these patients we immediatelyswitch to IV insulin upon arrival in the
operatingtheatre,andcontinuethisintheICU(results,seesectiononIVinsulininfusionalgorithm).
Theinsulinisadministeredwithasyringe-driveninfusionpump,andbloodglucoseismeasuredin
arterialbloodwithanon-sitebloodgasanalyser.
Inallotherinterventionsweaimatbloodglucoselevelsbelow200mgdl(cid:3)1.Inthesepatients,blood
glucoseismeasuredhourlyduringlongoperations,usingaclassicalbloodglucosemeteronacapillary
bloodsample.Ifthebloodglucoseexceeds200mgdl(cid:3)1anIVinsulinpumpisstarted.Ifnot,thefasting
protocol is continued. We use the same insulin infusionprotocol as for high-risk surgery, but with
ahighertargetzoneof90–140mgdl(cid:3)1(greenzoneinFig.1).Whenthepatientreturnstothesurgical
ward,weswitchtoanIVinsulinburette.Onthewardmostpatientscanbeconvertedtoscinsulin,
usingabasal–bolusscheme,theeveningaftertheoperation.Patientswhoreturnhomethedayofthe
operation immediately switch to their home treatment. An audit of all 917 low- and medium-risk
interventionsinDecember2008,inpatientswithdiabetesand/orstresshyperglycaemia,showedthat
with this approach 89.7% of the measurements ranged between 70 and 200mgdl(cid:3)1, with only
0.1%<50mgdl(cid:3)1and1.2%>300mgdl(cid:3)1.
Qualityassurance
Itisimportanttoregularlyevaluatewhethertheprotocolsareeffectiveenough,andwhetherthey
are systematically and properly used. One would especially like to know whether blood glucose is
measuredsufficientlyfrequently,howmanymeasurementsfallwithinthesettargets,howlargethe
bloodglucosevariationsareandhowoftenhypoglycaemiaandsignificanthyperglycaemiaoccur.For
comparison between different centres, the Yale group proposes to organise this in a standardised
manner,andmakesguidelinesandacomputerprogramavailableontheWebunderthename‘glu-
cometrics’.36,37Thesedatacanbelookedathospital-wide,forageneraloverviewoftheinpatientblood
glucosecontrol,andalsopersituation(e.g.,perioperative)orperunit.Regularfeedbackoftheresultsis
important to motivate staff to continue to follow the protocols. In case of unsatisfying results, the
processshouldbeanalysedtodeterminewhatcanbeimproved.Besidesparametersofbloodglucose
470 F.Nobelsetal./BestPractice&ResearchClinicalAnaesthesiology23(2009)461–472
control,itisalsousefultofollowoutcomeparameters.Sinceweintroducedtightglycaemiccontrolin
ourICUsweseeareductionofmortalityandofcardiac,renalandinfectiouscomplications.14
Inbrief,effectiveandsafeinpatientglycaemiccontrolisfeasible,withthecreativeuseofafewwell-
chosenbuildingblocks,andcarefulattentiontosafety.Anessentialelementisawell-designedalgo-
rithm for IV insulin administration that graduallyadapts the insulin infusion rate to the individual
insulin sensitivity of the patient. Forsmooth transition to post-acutecare a sc basal–bolus scheme,
preferably using regular and NPH insulin, offers flexibility, but avoids the fluctuations induced by
atraditionalsliding-scaleregimen.Itshouldbeusedwithacorrectionalgorithmthatprovideslarger
corrections at higher insulin doses and vice versa. Other essential elements are fasting rules, and
agreementsontransitioningtodischargecare.Hypoglycaemiacanbereducedtoastrictminimumby
proactiveadaptationofthetreatmentinsituationsofdiminishinginsulinneeds,especiallyincaseof
suddenreductionofcalorieintake.Thesedifferentbuildingblocksandsafetyconsiderationscanbe
integratedineffectiveandsafeprotocolsfortheorganisationoftightglycaemiccontrolindifferent
hospitalenvironmentsandsituations.Moreresearchisneeded,however,onseveralcomponentsof
thispracticalapproach.Itissurprisinghowlittlepracticalguidanceisavailableinthecurrentliterature.
Hospitalsshouldbeencouragedtomaketheirprotocolsandresultsmorereadilyavailable.
Practicepoints
(cid:7) Whendevelopingprotocolsforinpatientbloodglucosecontrol,useasfewbuildingblocksas
possible.WiththecreativeuseofanIVinsulinalgorithm,ascbasal–bolussystem,fasting
ordersandagreementsontransitioningtodischargecare,mostsituationscanbehandled.
(cid:7) UseanIVinsulinalgorithmthatgraduallyadaptstheinsulininfusionratetotheindividual
insulinsensitivityofthepatient.
(cid:7) Forscuse,avoidslidingscales,butusescheduledinsulinwithprovisionofbasalcoverage.Use
acorrectionalgorithmthatprovideslargercorrectionsathigherinsulindoses.
(cid:7) To avoid hypoglycaemia, pay major attention to immediate reduction of blood glucose
medicationincaseofsuddenreductionofcalorieintakeordoseofcorticosteroids.
(cid:7) Thetighterthebloodglucosegoals,themoreaccuratetheglucosemonitoringandIVinsulin
administrationshouldbe.
(cid:7) Goodcommunicationisanimportantelementofinpatientbloodglucosetreatment.Include
informationonnutritionintake,corticosteroiduseandkidneyfunctionontheglucosecharts.
Researchagenda
(cid:7) More research is needed to optimise the effectiveness and safety of insulin infusion
algorithms.
(cid:7) TheaccuracyofbedsidemonitoringintheICUsettingshouldbeimproved.
(cid:7) Accurateandsafecontinuousglucosemonitoringshouldbedeveloped,thatcanbeusedto
createclosed-loopglycaemiccontrol.
(cid:7) Expert systems should be created that integrate information on blood glucose evolution,
insulinpharmacologyand determinants of insulin needs (such as carbohydratesupply) to
driveanIVinsulinpump,creatinga‘smartclosed-loopsystem’.
(cid:7) Researchisneededontheessentialcomponentsofscinsulindelivery,suchastransitioning
fromIVtoscdelivery,useofcorrectiondosesandfastingrules.
(cid:7) Nursingpoliciesforeffectivehypoglycaemiapreventionshouldbeexamined.
(cid:7) OutcomeresearchoftightglycaemiccontrolintheICUandperioperativesettingsshouldbe
repeatedwithalgorithmsthatprovideaccuratebloodglucosecontrolwithalowincidenceof
severehypoglycaemia.