l-l

SI ISTAlNED WQPFRATIONS: PRO1X3NGEDDUTY OVmIGHT

RoyalAir ForceSchoolof Aviation Medicine

and
DefenceEvaluationandResearchAgency
Farnborough,HampshireGU146S2, United Kingdom

sustainedair operationsimply round-the-clockscenariosand,inevitably,prolongedduty ovcmight

The ability of Creweto copewith suchwork-rest pm dependsto a largeextenton obtaining sufficient

sleepduringcritical restperiods.Hypnoticsmay be essentialto ensuresleepa.sthe restperiodsthemselves

arelimited in numberbndduration, and occur at all timesof the day andnight. However,evenif good

sleepis attainedduringal1the availablerestperiods,theremay still bemuchdifficulty in sustaining

aleme-ssduringduty ovemigh~ pticularly if the duty periodsthemselvw are prolonged. This paperdeals

with the WeOf variouspotentialinterventionsto swain alertnessduring inrensiveair operations.

Paper presented at the AGARD AMP Symposium on Aeromedical Support Issues in

Contingency Operations. held in Rotter&m, The Netherlands. 29 September -
I October 1997, and published in CP-599.
1-2

fRREG1TI.iQJTY OF WORK A-

During the late 19705the RAF Iustitutcof Aviation Medicinewu concernedwith thecapabilityof

aircrewepemtingin the interdictorrole, andlab&arorystudiesweredirectedtowarddefAng the issues

involvedin copingwith the inevitableirrcgultity of rest(Nicholsonet al, 1984). Suchinformationwas

essentialto ensurethatmeanscouldbe developedto assistsquadronsiu maintaininground-the-clockoperations

in which the high workloadwould be sharedbetweenail aircrew,and in which crewscouldbe deployedin a

flexible mannerbenveclinight andday operations.Thework andrestof suchprojectid scenarioswere

simulatedoverperiodsof 9 days,which wasthe periodoverwhich aircrewwould havebeznexpectedto

Opsrate at maximumOutput

The simulated9 day scheduleof work andrestinvolved24,6-h periodsof work and 12,6-h periodsof
restc Fig.1). The schedulewasprecededandfollowed by 2 daysof normaldaytimedury andnocturnalrest,add

theseperiodsprovidedcontrol data. It wasthe primary intentionof the studyto ensurethat sleepdeprivation

wasminimised,andsothe scheduleprovidedanaverageof 8 h resteach24h - thoughin unequalpartsfrom the

6h periodsof rest. The rest periods werearrangedoverthe 9 daysso that the numberof ni&t anddaytime

sleepswereequalandrunsof consecutivenight or da$me sleeps avoided.The synchronising effect of sleep

wasalsominimisedby avoiding,asfar aspossible,consecutiveperiodsof restaroundthe usuainnctu~~altime

of sleep.

Over the periodof 9 days,restperiodsbeganat eachof four times(0300,0900, 1500or 2100 h). There

were threesinglework periodsof 6 II, six doublework periodsof 12h andthreetiple work periodsof 18h

duration. The completefindingsof rhe simulationhavebeenpublishedelsewhere(Nicholsanet al 19841,and

for the purposeof the presentpapertht informationabtainedrelevantto prolongedduty overnightis reported.

 1-3

Day 1 Day 2 Day 3 Day 4

09 15 21 1 03 09

Day 5 Day 6 Day 7 Day 8 Day 9

Night Sleep Day Sleep Perfomanm

cl

Fig I. Scheduleof work and rcsf

The quaI.@of sleepduring eachrest period was measuredby electroencephalography.Total sleep

times (An) for eachrest period are given in Figure 2. In the rest periodsbeforeand after the schedule the rotal

sleeptime almost accountedfar the 6h period, but during the scheduletotal sleeptime varied considerabIy.

Sleepat 21OObd 030Ohwas always restil, but sleepduring the day, ie at 0900h and particularly I500h, was

short unlessit followed an inb~al of significant sleepdeprivation. Furtheranalyseswererelatedtti the time of

day as well asto the precedingpattetll of work and rest. Theseindicated&at the greatestdifficulty in falling

whereas the longmt sleepwould havebeenmost,likely when tht sleepbegan

asleepwould accur aroundZOOOh,

around0 lOOh. Won of sleepduring the day would be particularly short exceptwhen precededby periods

of wakefulnessapproaching20 h, The most resti sleepwould be obtainedwhen it commencedbetween

2 1OOhand 0300b andthe quality would be improvedat any time by a reasonableperiod of preceding

wakefulness.
l-4

Total deep
time (min)

Control

I Night sleep
U Day sleep

Fig 2. Total sleeptime determinedelecwencephalographicaHyfor eachrest period

As far as performanceduring the work periods was coneaned, the usual circadian pat&m with lowest

scoresin the early morning, was maintainedthroughoutthe schedule. Performancedecreasedwith increasing

time awakeand for sometasksthere was a trend rewardsimpairedperformanceas the 9 day period proceeded,

However, merewas evidenceof improvementin sometasks over the scheduIc.thoughcaution musr be

eKerciscdin the interpretationof &is firtding. The intensity of testing during rbe scheduleclearly led to

continuedlearning on sometasks. and so the effect of cumulative sleeploss was probably masked. OveraIl, the

analysiscstablishcdthat impaired performancewas relatedto circadii rhythmicily, duration on task, and

cumulative sleep loss, andthesethree factors are fimdamentalto understandinghow performanceduring duty

againsta backgroundof irregularity of work and rest can be predicted.

Adaptarionof the individual to irregularity of rest and acrivity was rhc primary concernof the study. It

suggestedthat during a Ieng&y period of iq-+lar work and rest the efZciency of individuals is likely to be
 1-5

increasinglyimpaired,andthat their performancewill be influencedby their circadianrhyrhmicity and by the

length of time on task, aswell as by any cumulative loss of sleep. AbOw all, it was clear that, ia cemplex

schedules,a significant difficulty in maintaining capability would occur during prolongedperiods of work

extendingto the la%tcrpart of the night. It must also be bornein mind that behaviouralchangessuch as

impairedinterpersonalrelations.which arc difficult u) measure,may be equally, if not mere, impor&antdeem

decrementsindicatedby teatsof performance.

The I-day study confirmed the overriding importanceof avoiding cumulative sleeploss and of poor

performanceduring long periodsof work overnight to sustainingintensiveties of work. In this canroxtit must

be appreciatedthat in sustainingsuch operationsit ia highly unlikely that any significant changeto the work-rest

patternsthemselveswill be possible. The scheduleis determinedby a host of, ofien immutable,operational

constraints. Nevertheless,the study indicatedthat crewswere likely to be ableto copewith 12 x 6-h missions

over a 9 day period, and that u&g the longer periodsof wakefulnessfor doublemissions, I5 missionsmay bc

possible. Ensuringsleepis essentialandthis can be assistedby hypnotics,but overnight duty is the vulnerable

component in operationsdesignedto provide sustainedcapability.

PERF0RMANCE OVER 24 HOURS

Our studieson irregularity of work and rest also provided datanecessaryto quantify the characteristics

of periodsof work which determinepe&rmance.( Minors et al 1986,Spencer,1987) Severalfactors were

shown to influence performanceduring a particular duty period; thesewere : the interval betweenthe end of

the previoussleepand the commencementof duty (time since sleep):duration of duty (time on task); and the

clock time of duty @meof day). It was evidentthat the adversejuxtaposition of a long duty period and tie

timing of duty during the 24 hour period prejudiced the ability to sustainvigilance.

Furtheranalysisof thesedatashowedthar as far 85Yime an Wk is concerned,performancerises

during the first few hours,falls to its initial value after around5 hours, andrhenlevels off around 12to 16hours

a&r commencementof duty (Fig 3). As far as Cmeof day is concernedperformancerises during the day and

falls during the late eveningand overnight reachingits nadir around0500 h in the morning (Fig 3). Hence,very

low levels of performancearereachedif the laser part of a prolongedduty period coincideswith the circadian

trough in performance. For example.if a 16hour duty period commencesaround0200 h it is IikeIy that
l-6

performancewill be main&&i as the fall durhg the latter half of the work period coincideswith the rising

phaseduring the day. On the other hand, if duty commence around 1400h the drop in performanceduring the

latter parr of the duty period would coincide with the circadian fall during the night, andso low levels may be

reached(Fig 4).

This knowledge is crucial to the managementof aircrew involved in high workload, round-the-clock

operations. Assuming sleeploss is avoidedit can be assumedthat long periodsof duayduring the day leadto

only limited deteriorationsin perfotmance,whereassimilar periodsovernight lead to seriousdecrementswhich

may be accompaniedby microsleeps. Further, it ir possibleto determinethe duration of night and day periods

of duty which would equateto the samedegreeof impairment. It is, therefore,evident that if prolongedduty

overnight is e&set&I. to maintain a sWained operation,then somemeansmust be developedto avoid the lowest

levels of impairment. Various approacheshavebeenstudied including sleepprior to duty overnight, naps,

during the period of duty itself if feasible , breaksduring the duty period, and StimuIants.

Time on task

Fig 3. A model of changein performancewith time on task(lefi) and time of day(right).

 l-7

Fig 4. Time on task andtime of day relatedto duty commetichg at 0200 and 1400h (left) andresultant of the
time on task and time of day relatedto duty commencingat 0200 aad 1400h. {right). If a 16 h duty
period commencedaround0200 h it is likely that performancewould bc maintainedasthe fall in
performanceduring the latter half af the work period would coincide with the rising phaseduring the
day. On the other hand if the duty period commencc~around 1400h the fall in p~otmance during the
latter part of the dury period would coincide with the lowest level during the night relatedto the
circadianrhythrnicity of the individual, and 50 very low levels of performancemay be reached. Such
adversejwtqositlons of time on task and time of day shouldbe avoidedIf crews are expectedto
remain cxmtinuouslyon their task. For this reasoncareful attentionmust be given to the length of duty
periods in suchoperations,and the length shouldbe determinedin relation to the time of day. These
considerationsassumethat the aircrew arc fully asted at the commencementof their duty.

SLEEPSTO MINiMISE PERFORMANCE DECREMENTS

m eveninesleenq

prolongedduty overnight is the &in~quanou of sustained and intensiveair operations,andthat such

duty periodshave very low levels of performancedue to the adversejuxtaposition of prolongedduty and the

nadir of circadian activity. Bowever. it is possiblerhat an eveningsleepperiod before a period of overnight

duty may amelioratethe deteriorationto someextent. tt is in this contextthat the possiblebeneficial effect of

an early eveningsleepprior to overnight duty was studied(Nicholson et al 1985)

The subjectswere six healthy male volunteersaged between 20 and21 years.In the study subjects

completednine different schedulesof work and rest eachseparatedby a week, though in this accountwe deal

only with thoseschedule concernedwith the effect of an eveningsleepon overnight performance. Schedules

were of 48 hours duration and began with an overnight sleepfrom 2300hto 07OOh.During the day after the

initial overnight sleep performance was measuredover threehour periodsfrom 0900h, 14OOh, 1900h,and then

through the night at OWOhandQ4OOh,

and similarly through the next day. In the schedulewith an evening
I-8
sleepthe period of skep was tirn 0800h to 2200h, and so replacedone of the performanceperiods preceding

rhe overnight period of duty (Fig.5).

Fig.5 Scheduleof wti and rest relatedto anticipatory sleep(left) with pcrfonaanceon Digit Symbol

Substitutionwiti andwi&ut sleep (right)

Performanceovernight (0000 - 03OOhandOS00- 07OOh) witfi early evening sleep(1800 - 22OOh)

showed improved digit symbol substirution, symbol wpyiq, mental arithmetic and cancebtion ia all subjects.

Therewas also someevidencethat visuo-motor coordination,critical flicker fusion, reactiontimes, bxcicingand

auditory vigilance were improved, It, therefore,appcand that relatively shart periodsof sleephad a b&cficial

effect on subsequentperformance even in the absenceof a precedingsleepdebt, and that sleep in the early

eveningcould attenuatethe circadian fall in performanceovernight.

Although impairedefficiency overnight may be amelioratedby an eveningsleep of about4h duration

QJichoisonet al I985 ), shortersleepswithin the periodbf duty may be beneficial in certain circumstances.

However, any advantage that may be gainedfrom short sleepsdependson severalf&ton, including tht length

of the precedingperiod of sleepIoss,the durationof the sleep,the phaseof the circadianrhythm when the sleep

was taken,and the relatively poor ability of individuals to perform soon after awakening,otherwisehown as

Weep inertia. The circadian time of performancetesting and the narureof the perfixmancetestsmust also be

taken into consideration. It was to establishthe effectiveness of a short period of sleepin reducingthe
 l-9

progressiveimpairmentof performancespecific to overnight duty that the effect of a 1h nap taken at WOOhwas

investigated(Rogerset al. 1989).

Fig.6 Scheduleof work and rest relatedto napsdurin,- a duty period (left) and performanceon a visual

vigilance task with and without a nap (right).

The subjectswere sti healthy females,agedbetween20 and 32 years(mean25 ye-). Each schedule

WBSof 17.Shduration andwag precededby a 4h rest period during which the subjectswere awakebut resuicted

to passiveactivities (reading, watching television etc.). Performancewas measuredduring eight sessions,each

lasting 1.75h,which beganat 1700h,1915h,2130h,2345h, 02OOh,O415h,0630h and 0845h, with a 30 min

break betweensessionswhen the subjectswere supervisedto ensurerhat they remainedawake. In the schectulc

which includeda nap the subjects retired to bed at 02OOh(Fig 6) Sleepwas measuredby

electroencephalography.Subject were awoken 1h after sleeponset,(indicacLdby latency m the firs unbroken

5 ruin of stage2), and pcrfammce testing was resumed at 04 15h.

Overnight, io the absenceof the nap, performanceon all tasksexceptshort-termmemory deteriorated.

Sustaintd tiention, digit symbol substitution,auditory and visual vigilance were imp&cd aroundmidnight

comparedwith the levels attained during the ptrly evening, and the lowest scoresappeared&round0630h,

though decrementsin complex vigilance, NO-letter cancellation,and logic, did not appearuntil Cl41Sh or

0630h. Within-run deteriorarionsin visual and auditory vigilance were evident aroundmidnight. and were

presentwithin 12 min of commencingeachtmzk.

l-10

The nap was characterisedby short late&es to slow wave sleep. As expected,the long period of prior

wakefihess influenced the propensity for slow wave sleepin the nap. which occupiedaroundhalf the sleep

rime. However, this was insufftiient to influencethe requirementfor slow wave sleepduring the recovery

sleep,unlike otherstudiesof post nap sleepat nigbrwhereB2h nap in the late afternoonreducedslow wave

activity Tom 10%to 5% oft& total sleeptime, The lh nap taken at 02OOhhad only a limited effect on the

usual decrementsin performanceobservedovernight An improvementwas discernedon auditory vigilance at

04I5h and 0630h and on digit symbol subsciturionat 063Oh,but impairmentsremainedon all other rash.

?revious studieshaveshownthat a 1h or 2h naptaken aroundthe nadir of the circadiancycle during a

single overnight period of work may &enuate the expecteddecrementsin performance,but it would appearthat

a sleepof much longer duration than 1h, which is unlikely to be practical is neededto h8vea persistenteffect

when t&g sessionsare particularly demanding.There is also the possibility that sleepinertia may have

persistedbeyondthe nap and counteractedany possibleimprovementin subsequentperfotmance.

In somecircumstancesit may be possibleto breakup periods of continuouswork with short breaks, andthis has

beenstudied in work periods of 12 hours (Rogers, 1997). Eit performancesessianr which lastedsevenly five

minutes were separatedby 15 minute breaks.Within eachsessiontherewere six runs of tie tracking component

of the task studied( Multi-atibute task battety). Ptirrnance overnight after an eveningsIeep,and during the

day wasmeasuredundertwa workload conditions.The decrementsassociatedwith time on task andwith

working at night were more severeunderc.onditioIlsof high workload. Tracking performancedeterioratedover

the six runs within a session, parricularly at night when significant decrementsoccurredby the third run

comparedwitb the Mh run during the day. Following a 15minutebreaktherewas a generalimprovementin

tracking performanceat the be&inningof a sessioncomparedwith the last run of the previous session,

particularly overnight (Fig. 7). As well as sroppingwork during thesebreaks,subjectswere also ableto eat,

drink, walk aroundand interact with others.It is not yet clear what aspectof the break was responsiblefor be
 l-11

recuperative effects.

mean RMS High workload condith 1

300

250

200

150

100

50

0
213Oh 003Oh Mooh 033Oh 0500h
Session
- . -. . I

Fig. 7 Mean RMS for the tracking componentof the MAT battery during tic high warkload condition

Theseand studiesfrom elsewhereOIIthe useof sleeppreceding and a nap during duty show that the

deteriorationin performanceovernight may be meliorslted undercertain circtur~stances,

but it must be bcme in

mind that in the managementof intensiveandsustainedepe-rationssuch techniqueshave pracrical difficulties,

and they are unlikely to be appropriateunlessthey provide an overriding improvementin capability. During a

sustainedopemtioneachovernight perid of work will fallow an inegular partemof wotk and rest over several

days and that good sleep precedingduty is WIessenrlaIelementto avoid cumulative sleepioss. It is unlikely to

be possibleto provide an extra sleepanticipating a duty period. Indeed,it is c~nsidereclthat impaired

perfomance overnight against a backgroundof intensiveratesof work and irregularity of rest is llnlikely ro be

amelioratedby such techniiuek It is in this way that we have soughtto establish w@her sti~ul~ts would be

more useful,

CONSIDERATIONSOF STIMULANTS
1-12

0ur approachto this problem hasbeento identify a stimulant free of adverseeffects on affkcti~e

behaviour(Nicholson & Pascoe1991, 1992,Nicholson et al 1989). In this context we do not supportthe useof

amphetaminesin military operationsbecauseof the well known euphoric effect. However, it is difkult ta

estabhshfrom experimentalstudies which stimulantr would be free of suchcentral effkck We have taken the

approachthat u stimulant free of effects on the noradrcuergicand serotonergicsystems,i.e. likely to be

predominatrtly dopamincrgic,is mere likely to be acceptable.In the searchfor such a drug we have usedthe

fact that noradreuergicand serotonergicdtugs dkcctty suppressREM sleep addltlonally to any decreasein

REM activiq which may be associatedwith increasedwukefulnms during skp. We havt observedthat both

caffeine and pemoliie are fire of such xtivity, andtheseobservutionsare consistentwith the drugs being free

of aclrenergicand surotinergicactivity. It hasbeento pemoiinethat we have given most of our attention.

Pemolineis bn indirect dopaminergicagentand is relatively ftec of sympethomimeticactivity. It is

lcnownto incrf%seaiernxss and improve perfannanceduring the day, and, for sleepdeprived subjects,

overnight However, it is envisagedthat a drug Iike pemolinewould only be used overnight and, therefore,

information is needed on the effects that such B drug would have on performanceand alerhess of duty periods

which commenceduring the latter part of the day, and which are,iddeed,likely to leadto very low levels of

performance. There is no doubt that pemolinc increasesakrinuss, but thereis uncertaintyconcerningthe

appropriatedosethat shouldbe usedin a situation which may weI1prove to have critical time constraints. It

was in this ooutextthat we have caked out studieson the effect of pemoline and on its potential use in a

simulatedovernight operation.

An initial study highlighted the persistenteffect of pemoiine. Indeed, 30,60 and 90mg pemoline

increaseddaytime alertnessover periods of 8 hours, while 40mg pemolinc preventedthe deteriorationof

performanceover 8 and 12hour Periodsovernight, but also disturbedsleepwhich commenced up to I2 i~ours

after ingestiau. It was, therefore,decidedthat the studiespertinentto an operationalscenario would be carried

out with maximum doseof 40tng pcmoline.

Two further studies(Turner and Mills, I 996, Nicholson and Turner,1998) have shownthut a 2Omg

dosewould be approprhtc for maintaining performadceovernight without disturbing recovery steepthe next
 l-13

morning.In a high workloadtaskall measuresof perfbrmancewereimprovedwith 20mgpernolineassoonas

perfofr~~ance
wasimpairedwhencomparedto controllevels(Fig.).

Although modulationof the work-restpatterncanamelioratEimpairedperfarmanccduring prolonged

duty overnight,it is unlikelythat suchan approachwould bepracticalin intensiveandsustaiucdoperations.If

is in this contextthat stimulantsmay proveto bethe mosteffectivesolutionunderoptrationatconditions.

Visual vigilance
(% err*r rate)
70 -I

60

50

40

30

20

10

Time (h)
Fii. 8 Performanceat a visual vigilancetaskshowingimprovements
in perfomancewith pemctline.
1-14

REFERENCES

Minors, D.S.. Nicholson. A.N., Spencer,M.B., Stone,B.M. ad Waterhouse,J-M., IrreguIarity of rest

and activity : Studieson circddian rhythmicity in man, 1. Physial.,381, 1986,279-295.

PI Nicholson, A.N., Btlyavin, AL, Pascoe,P.A.. Modulatian of rapid eye movementsleepin humansby

drugsthat modi@ monoaminergicandpurinergic transmission,Neuropsychopharmacology,2, 1989,

pp 131-143.

PI Nicholson, A-N., Pa~coe,P-A., Monaaminergictransmissionend sleepin man. in: Serotonin,sleep

and meutal disorder,Ed. C. Idzikowrki and P.J. Cowen. Petersfield:Wright Biomedical Publishing

Limited, 1991,pp 215-226.

PI Nicholson, A.N., Pascoe,P.A., Roehrs,T., Roth, T., Spencer.M. B., Stane,B. M. and Zorich, F.,

Sustainedperformancewith short eveningand morning sleeps,Aviat. SpaceEnviron. Med.,56,1985,

105-l 14.

PI Nicholson, A.N., Stone,B-M,, Borland, R.G., Spencer,M.B., Adaptation IO irregularity of rest and

activity, Aviat SpaceEnviron. Med., 55,1984, pp 102-112.

I61 Rogers,A., Workluad and fatigue in long-rangeoperations- a laboratorystudy, Journalof Defence

Science,2, 1997,191-199.

PI Rogers,A.,Speucer,M.B., Stone,B.M. andNicholsan, A.N., The influence of a I h nap an perfonnauce

overnight, Ergonomics,32,1989,1193-1205.

PI Spencer,M.B., The influence of irregularity of rest and activity on performance:a model basedon

time sincesleepandtime of day, Ergonomics,30,1987,1275-1286.

E93 Nicholson, A.N. and Turner&., tntensive andsustainedair operations;Potential useof the stimulant,

pemoline,Aviat SpaceEatiron. Med., 1998,In press.

WI Turner, C. and Mills, S., Ef%cts of pemoline on avernight performanceinvolving high workload,

DERA Report,PLSD/CHS5/CR96/088,1996